The Development of an International Instrument to … Version 5.3 May 12, 2010. The Development of...

44
Revised Version 5.3 May 12, 2010. The Development of an International Instrument to Measure Public Service Motivation: A Research Note Sangmook Kim (Seoul National University of Science and Technology) Wouter Vandenabeele (Utrecht University & Katholieke Universiteit Leuven) Lotte B. Andersen (Aarhus University) Francesco Paolo Cerase (Federico II University of Naples) Robert K. Christensen (University of Georgia) Céline Desmarais (Université de Savoie) Maria Koumenta (Oxford Brookes University) Peter Leisink (Utrecht University) Bangcheng Liu (Shanghai Jiao Tong University) Jolanta Palidauskaite (Kaunas University of Technology) Lene H. Pedersen (Danish Institute for Governmental Research) James L. Perry (Indiana University) Adrian Ritz (University of Bern) Jeannette Taylor (University of Western Australia) Paola De Vivo (Federico II University of Naples) Bradley E. Wright (University of North Carolina at Charlotte) Contact Addresses: Professor Sangmook Kim Dr. Wouter Vandenabeele Department of Public Administration Utrecht School of Governance Seoul National University of Technology Utrecht University Seoul, 139-743 Republic of Korea Bijlhouwerstraat 6 Fax: +82 2 971 4647 3511 ZC Utrecht, the Netherlands E-mail: [email protected] E-mail: [email protected] Prepared for presentation at the 11th National Public Management Research Conference at Syracuse University, Syracuse, NY, June 2-4, 2011. Its earlier version was delivered at the annual conference of the European Group for Public Administration, Toulouse, France, September 8-10, 2010.

Transcript of The Development of an International Instrument to … Version 5.3 May 12, 2010. The Development of...

Revised Version 53

May 12 2010

The Development of an International Instrument to

Measure Public Service Motivation A Research Note

Sangmook Kim (Seoul National University of Science and Technology)

Wouter Vandenabeele (Utrecht University amp Katholieke Universiteit Leuven)

Lotte B Andersen (Aarhus University)

Francesco Paolo Cerase (Federico II University of Naples)

Robert K Christensen (University of Georgia)

Ceacuteline Desmarais (Universiteacute de Savoie)

Maria Koumenta (Oxford Brookes University)

Peter Leisink (Utrecht University)

Bangcheng Liu (Shanghai Jiao Tong University)

Jolanta Palidauskaite (Kaunas University of Technology)

Lene H Pedersen (Danish Institute for Governmental Research)

James L Perry (Indiana University)

Adrian Ritz (University of Bern)

Jeannette Taylor (University of Western Australia)

Paola De Vivo (Federico II University of Naples)

Bradley E Wright (University of North Carolina at Charlotte)

Contact Addresses

Professor Sangmook Kim Dr Wouter Vandenabeele

Department of Public Administration Utrecht School of Governance

Seoul National University of Technology Utrecht University

Seoul 139-743 Republic of Korea Bijlhouwerstraat 6

Fax +82 2 971 4647 3511 ZC Utrecht the Netherlands

E-mail smooksnutackr E-mail woutervandenabeelesockuleuvenbe

Prepared for presentation at the 11th National Public Management Research Conference at Syracuse

University Syracuse NY June 2-4 2011 Its earlier version was delivered at the annual conference of

the European Group for Public Administration Toulouse France September 8-10 2010

- 1 -

The Development of an International Instrument to

Measure Public Service Motivation A Research Note

ABSTRACT

As research on public service motivation (PSM) has grown and spread from the United States

to Asia Europe and Australia it is essential to elaborate PSM as both a construct and

measure in order to facilitate cross-national research and comparison Previous research has

sharpened the PSM construct and thoroughly discussed a PSM measurement model The

remaining issue is to develop an internationally robust PSM measurement instrument

Drawing upon possible items for measuring PSM which were suggested collectively by PSM

researchers and the data (n = 2868) collected in an international survey among public

servants at local governments in twelve countries the conceptual components of PSM are

empirically estimated through operational definition The survey results are analyzed and

discussed in order to confirm the dimensional structure and validate the items of PSM This

study provides a set of appropriate items for measuring PSM and confirms the dimensional

structure of PSM When all four dimensions of PSM are used in a survey cross-national

equivalence can be assured

- 2 -

The Development of an International Instrument to

Measure Public Service Motivation A Research Note

Introduction

Public service motivation (PSM) is thought of as ―an individualrsquos orientation to delivering

services to people with a purpose to do good for others and society (Perry and Hondgehem

2008a vii) The initial research on PSM began in the United States (Perry and Wise 1990

Rainey 1982) but scholars in many countries have contributed to the accumulation of

knowledge about its antecedents and consequences (Perry and Hondeghem 2008a Perry

Hondeghem and Wise 2010) As the scholarly research on PSM has grown and the

geographic scope of the research has expanded some scholars have questioned whether the

conceptual composition and dimensions of PSM are appropriate for explaining and predicting

public servicendashrelated behavior in different countries and internationally (Coursey and

Pandey 2007a Wise 2000 Wright and Pandey 2008 Giauque et al 2011 Vandenabeele

2008b) and whether different PSM measures are equivalent or if there is a need to develop a

more appropriate measure of PSM that can be used consistently and confidently (Castaing

2006 Leisink and Steijn 2009 Liu Tang and Zhu 2008 Taylor 2007 Vandenabeele 2008a

Wright 2008 Wright and Pandey 2008) Thus in order to facilitate research internationally it

is necessary to elaborate the construct and measurement of PSM (Perry and Hondeghem

2008b)

The first step was to develop a more universal concept that can be used globally and

that is likely to generate cumulative knowledge The construct of PSM has been sharpened by

refining the conceptual components in Perry and Wisersquos (1990) study Kim and Vandenabeele

- 3 -

(2010) proposed that PSM should continue to be conceived as a four-dimensional construct

with self-sacrifice as the foundational concept representing the altruistic or pro-social origins

of PSM The three other concepts attraction to public participation commitment to public

values and compassion which represent instrumental value-based and affective motives

round out their proposed modifications to the construct The revisions to the construct of

PSM will enhance cross-national research and generate cumulative knowledge

The second step was to specify a measurement model of PSM This was necessary

because the measurement model specifies the relationship between constructs and measures

(Diamantopoulos and Siguaw 2006) Reflective measures are treated as outcomes of

constructs and formative measures are causes of constructs (Edwards 2010) A measurement

model can either include reflective or formative indicators exclusively or consist of both

depending on the observed construct (Fornell and Bookstein 1982) The decision to

operationalize formative andor reflective indicators should be based on theoretical

considerations Previous studies discussing the conceptual definition of PSM theoretically

and empirically have proposed that the dimensions of PSM are formative and the indicators

of each dimension are reflective (Kim 2010 Kim and Vandenabeele 2010)

The next step is to develop an internationally robust measurement instrument of PSM

for cross-national research and comparison A survey-based measure for PSM has been useful

for facilitating comparisons across disparate services and national settings and creating

foundations for the accumulation of results (Perry Hondeghem and Wise 2010) Perry (1996)

identified a multidimensional 24-item scale to measure PSM Since then many studies have

used the items of Perryrsquos (1996) measure and improvements have been made to the PSM

scale (Coursey and Pandey 2007a Kim 2009a Taylor 2007 Vandenabeele 2008b) However

research has shown that the original items do not fully represent unique and salient qualities

- 4 -

of the dimensions of the PSM construct (Castaing 2006 Coursey and Pandey 2007a Coursey

et al 2008 DeHart-Davis Marlowe and Pandey 2006 Kim 2009b Leisink and Steijn 2009

Liu Tang and Zhu 2008 Moynihan and Pandey 2007b Taylor 2007 Vandenabeele 2008a

Wright and Pandey 2008) Moreover as the construct and operational definition of PSM are

refined it is necessary to adapt items for the dimensions of the PSM construct

This study is focused on developing a revised measurement instrument for PSM for

international use After developing the possible items for measuring PSM the international

survey was conducted in 2010 and the data (n = 2868) were collected from twelve countries

Dimensions and items were tested cross-nationally as this can be a useful tool to preclude

value-specific items and wordings that are case-specific The next section of this paper will

briefly explain the processes for developing the international measurement instrument The

survey results are analyzed and discussed in order to confirm the dimensional structure and

validate the items of PSM Finally the items of PSM are proposed and discussions and

suggestions are given for future research

Developing the Measurement Instrument of PSM for International Use

Background

Perry and Wise (1990) proposed that PSM is associated with three types of motives affective

norm-based and rational Perry (1996) identified four empirical components of the PSM

construct attraction to public policy making commitment to the public interestcivic duty

compassion and self-sacrifice Since then many studies have used the dimensions and items

of Perryrsquos (1996) measure and improvements have been made in the PSM scale (Kim 2009a

Taylor 2007 Vandenabeele 2008b) However Perryrsquos (1996) measure has limitations (Kim

and Vandenabeele 2010) In particular the items of the dimension of attraction to policy

- 5 -

making are not appropriate for measuring personal attraction to public policy making

(Camilleri 2006 Coursey and Pandey 2007a Coursey et al 2008 Kim 2009b Ritz 2011)

Moreover the dimension of commitment to public interest needs to concentrate more on a

personal disposition to pursue public values (Castaing 2006 Leisink and Steijn 2009 Taylor

2007) and the items of compassion do not represent affective motives (DeHart-Davis

Marlowe and Pandey 2006 Moynihan and Pandey 2007a Wright 2008) Thus the

dimensions of the PSM construct are modified as attraction to public participation

commitment to public values compassion and self-sacrifice Consequently it is necessary to

develop appropriate items for the dimensions (Kim and Vandenabeele 2010)

A related issue concerns the measurement model of PSM There are two

measurement models using multiple dimensions of latent constructs reflective and formative

When variation in a construct leads to variation in its dimensions such dimensions are

termed ―reflective while dimensions are labeled ―formative when they are viewed as

causes of a construct (Edwards and Bagozzi 2000 Diamantopoulos and Siguaw 2006

Diamantopoulos and Winklhofer 2001) It is understandable that each dimension of PSM has

several reflective indicators that are essentially interchangeable A problem is the relationship

between PSM and its dimensions If we regard the dimensions as reflective it is possible to

analyze PSM with only two or three dimensions because omitting dimensions does not

change the essential nature of PSM However if we regard the dimensions as formative it is

not acceptable to omit a dimension as this may change the meaning of PSM Construct

conceptualization is regarded as being the process that determines the formative or reflective

nature (Diamantopoulos and Siguaw 2006) Recent research has demonstrated theoretical and

empirical evidence in support of a second-order formative approach to PSM (Kim 2010)

This means that the dimensions represent different conceptual components of PSM and so

- 6 -

they are not interchangeable Thus PSM is regarded as a combination of specific dimensions

If any one of the dimensions changes PSM would change Conversely if an individualrsquos

PSM increases this would not necessarily be accompanied by an increase in all dimensions

Therefore as the PSM construct is refined for international use (Kim and Vandenabeele

2010) scholars must be attentive to identifying and using appropriate measurement models

Developing Possible Items

The conceptual components of PSM need to be empirically estimated through operational

definition Kim and Vandenabeele (2010) proposed that PSM is based on self-sacrifice and

associated with instrumental value-based and identification motives and that the dimensions

of the PSM construct be refined along the lines of attraction to public participation (APP)

commitment to public values (CPV) compassion (COM) and self-sacrifice (SS) As Perry

(1996) identified a multidimensional scale to measure PSM it is conceived as an overall

latent variable with various latent dimensions and is referred to as a second-order factor

model Research has found that the measurement model specifies the causal relationship

between PSM and its dimensions (Edwards and Bagozzi 2000) The direction of the

relationship is either from the PSM construct to the dimensions (second-order reflective

measurement) or from the dimensions to the construct (second-order formative measurement)

A theoretical consideration of the relationships between PSM and its dimensions led

to the proposal that PSM be defined in terms of multiple dimensions (Kim 2010 Kim and

Vandenabeele 2010) The dimensions represent different aspects of PSM with each

dimension capturing a distinct and potentially unique form of PSM Moreover the

dimensions may have different antecedents and consequences as well as different

characteristics and theoretical backgrounds (Moynihan and Pandey 2007a Pandey and

- 7 -

Stazyk 2008 Perry 1996 1997 2000 Perry and Wise 1990 Perry and Vandenabeele 2008

Taylor 2007 Vandenabeele 2007 2008a 2009) An individualrsquos PSM is shaped by the

individualrsquos APP CPV COM and SS Deleting a dimension means omitting a certain part of

PSM Therefore it is reasonable to suggest that the dimensions of PSM are formative and for

PSM to be regarded as a composite of its dimensions

Each dimension of PSM can be measured using reflective indicators that satisfy the

following criteria (Jarvis MacKenzie and Podsakoff 2003) a relative homogeneity and

interchangeability of indicators pertaining to each dimension a high degree of covariation

among indicators of each dimension and the expectation that the indicators of each

dimension are likely to be affected by the same antecedents and have the same consequences

Reflective indicators are essentially interchangeable and so adding or removing indicators

does not change the essential nature of a dimension Thus we consider PSM to be a second-

order formative construct with several dimensions that each has several reflective indicators

(reflective first-order formative second-order)

Based upon the refinement and clarification of the operational dimensions of PSM

(Kim and Vandenabeele 2010) items were developed for the four dimensions The items

selected were a mix of pre-existing items that loaded highly on PSM dimensions in multiple

studies (Giauque et al 2011 Kim 2009a Perry 1996 Vandenabeele 2008b) Where

appropriate items were not available new items were drafted from other previous studies

creating an initial total of 35 proposed items For the dimension of APP which focuses on

individual disposition to work in the public sector to participate in the public policy process

and in activities for community and social development seven possible items were suggested

The CPV dimension focuses on personal disposition to pursue public interest and public

values Two sub-dimensions of CPV seemed plausible one focusing on public interest and

- 8 -

another on public values such as social equity equal opportunity and democracy Thus five

possible items for the public interest sub-dimension (CPV1) and six items for the public

values sub-dimension (CPV2) were developed For the dimension of COM which focuses on

affective bonding with the identified objects such as other members of a social category or of

a political system ten possible items were suggested For the SS dimension Perryrsquos original

items (1996) were the primary source of items The full list of items is provided in Appendix

1

Review Processes

A group of researchers who shared our interest in better measurement of PSM collaborated

using e-mail in November 20091 The objective was to develop a measurement instrument

for PSM with a robust comparative character in other words to create a valid instrument (on

a national basis) that can be used internationally The first round for reviewing the possible

items lasted from the 16th

of November until the 7th

of December 2009 and all the

participants were asked to review these proposed items For this review two criteria were

employed (1) Do these items refer to the dimensions they claim to refer to and (2) Are

these items meaningful across various national contexts and in particular to the situation of

civil servants in that country Participants were also asked to add new items that might better

reflect a particular dimension or to rephrase items that were already on the list

A second round of item refinement lasted from the 9th

of December until the 22nd

of

December 2009 During this time 11 newly suggested items from the first round were

considered (see Appendix 2) A summary of all comments and suggestions emerging from the

1 Emmanuel Camilleri David Giauque Gerhard Hammerschmid Isabel Egger-Peitler Renate Meyer and

Cristian Pliscoff also participated in the review processes

- 9 -

first process was distributed to the participants to stimulate comments in the second round

(see Appendix 3) Based on the comments received in the second round discussion the

possible items for an international survey were selected However some items required

further discussion and so a third round was conducted from the 15th

of January until the 23rd

of January 2010 One issue that remained unresolved in the discussions relates to the

dimension of CPV Some argued that it is reasonable to divide CPV into two sub-dimensions

because committing to public interests is different from pursuing public values and there are

many important public values in democratic countries while others contended that CPV can

cover all important public values including public interest and there is no real benefit to add

one more dimension These differences were left to be resolved through empirical research

This did not affect the final items however and 33 items were confirmed The resulting

survey questionnaire was distributed to all the collaborators on February 9 2010 in order to

conduct a survey in each country

Measures

PSM was measured using the newly developed 33-item index 7 items for the dimension of

APP 13 items for CPV (4 for CPV1 and 9 for CPV2) 6 items for COM and 7 for SS In the

survey questionnaire a five-point Likert type scale (1 = strongly disagree 5 = strongly agree)

was used This scale is among the most widely used response formats in many empirical

investigations within the social and behavioral science in which there are no ―correct or

―incorrect responses (Spector 1992) The items of PSM were randomly distributed in the

survey format Gender age education length of service and employment status were also

measured in this survey

It was necessary to translate the measurement items into the official (native) language

- 10 -

of each country As this raised the possibility of inaccurate translations the back-translation

of the items into native languages was validated by asking two researchers to independently

translate the English-language items The researchers were instructed to consult with each

other over differences between translated items and to test the survey items with some native

respondents

[Table 1 about here]

Samples

The international survey was fielded from March to September 2010 The data upon which

this study are based were collected from civil servants in local governments in Australia

Belgium China Denmark France Italy Korea Lithuania the Netherlands Switzerland the

United Kingdom and the USA The principal criterion was that the samples collected in the

various countries should be comparable to the largest possible extent The criterion ruled out

national and state government as the competencies of national governments and also of state

governments if they exist constitutionally are different among most of the countries included

in the research Therefore we opted for local government as our focus The structure and

functions of local governments can differ widely even within countries and so the survey in

each country was focused on town hall bureaucrats in city county or township government

The respondents were permanent employees from both managerial and non-managerial levels

and from both administrative departments (ie no direct contact with the public) and service

departments (ie likely to have direct contact with the public) Despite being employed by

local governments in many or all of the countries police officers firefighters school teachers

public transport workers artists and musicians as well as nurses and doctors were excluded

- 11 -

from the sample

We aimed to receive 250 respondents for each country as this would enable us to

carry out the analyses we wanted to perform on the data Although there is no ideal sample

size one study recommends a sample of 200 to provide a sound basis for estimation (Hair

Black Babin and Anderson 2010) The method of data collection was left open Obviously a

web-survey is the easiest way to collect the data because it is economical to collect and to

code however compared to paper-pencil surveys this method has some drawbacks First not

everybody in the sample may have access to the internet which may cause a bias in the

responses Second internet surveys are generally known to produce lower response rates

Web surveys were recommended if possible but paper-pencil surveys were deemed

acceptable in cases where it was not possible to conduct a web survey Researchers were

encouraged to contact just one municipality except in cases where it was not possible to

gather enough data In such instances responses from multiple municipalities were accepted

By the end of September 2010 a total of 2868 responses were obtained2

2 In Australia seven local councils each with between from 70 and 232 employees were surveyed Paper-pencil

surveys were used in two councils while the rest relied on a web survey The local governments limited

distribution of paper-pencil surveys to only one or two sections of the council eg administration The response

rate was approximately 30 However it should be noted that not everyone working for the council has direct

access to a computer In Belgium three municipalities were surveyed on a census base by means of a web

survey as all employees had access to a computer This rendered a response of 120 (226) 79 (75) and 16

(53) responses for the respective municipalities resulting in an overall response rate of 331 with 662

invitations being sent out In China two departments social work and administration were surveyed in one

municipality using paper-pencil surveys In the social work department 65 out of 110 surveys were returned

and 165 out of 211 responded from the administration department giving an overall response rate of 767 In

Denmark a total of 1282 respondents in a municipality were sent an initial e-mail invitation to a web survey

and up to two e-mail reminders and the response rate was 473 In France all the employees in two

municipalities were surveyed In one municipality where both web and paper-pencil surveys were utilized 128

out of 685 questionnaires were returned while in another city using only a paper-pencil survey 138 out of 652

were received The average response rate was 199 In Italy in principle the link to the questionnaire was

made available to all employees (about 4000) in one city but for some of them access may have been difficult

yielding a response rate of 45 In Korea 500 employees in a city received a printed questionnaire and 253

responses were returned yielding a response rate of 506 In Lithuania 236 responses from 54 municipalities

were gathered by web survey In the Netherlands three municipalities were surveyed on a census base by means

of a web survey In total 319 responses of 910 invitations were returned (35) mdash 90 (45) 115 (38) and 114

(28) for the respective municipalities In Switzerland 516 out of 1326 local employees with e-mail access in

a municipality completed the survey giving a response rate of 389 In the United Kingdom the survey was

posted online and 1360 employees both in administrative jobs and front-line services of one municipality

- 12 -

Municipalities were selected because of their convenient accessibility and proximity to the

researcher and a convenience sample was used in most countries Seven countries surveyed

one municipality but seven local councils were surveyed in Australia Web surveys were used

in four countries while three countries used both web and paper-pencil survey The number

of respondents varied among the countries depending on the numbers of questionnaires

distributed the survey method and the response rate In most countries we used

approximately 250 responses but only 162 from Italy3 The majority of the respondents were

women (566) while in terms of service length those in the largest group (393) had

worked for fewer than 10 years Table 2 shows the distribution of respondentsrsquo sex length of

service employment status and country

[Table 2 about here]

Analyses

The statistical analysis applied confirmatory factor analysis (CFA) CFA was used to assess

the fit of the data to the hypothesized measurement model because it is preferred where

measurement models have a well-developed underlying theory for hypothesized patterns of

loadings The data were analyzed using Amos 180 (Arbuckle 2009) with the maximum

likelihood (MLE) estimation method and LISREL 872 (Joumlreskog and Soumlrbom 2005) with the

diagonally weighted least squares (D-WLS) estimation method

The individual items are designed to measure a theoretically continuous construct

but the observed responses are a discrete realization of five categories The common method

received an email to complete the survey The response rate was 19 In the USA the total number of

employees surveyed was 626 in one city using both electronic and paper-pencil surveys resulting in a response

rate of 588 3 In countries where significantly more than 250 surveys were collected we randomly selected only 250 cases

for analysis

- 13 -

of estimation within CFA is MLE a technique which assumes that the observed variables are

continuous and normally distributed These assumptions are not met when the observed data

are discrete as occurs when using ordinal scales as in this study A simulation study (Flora

and Curran 2004) found that D-WLS performs well for ordinal measures across a variety of

conditions especially for smaller samples Coursey and Pandey (2007a 2007b) argued the

importance of using D-WLS and not standard MLE in structural equations for ordinal data4

In most cases however the differences between MLE and D-WLS are not substantive to the

conclusions5 No comparison between standard MLE and D-WLS is provided in this paper

We give results from both MLE and D-WLS and this will aid in confirming the stability and

robustness of the results across varying mathematical specifications

For gauging reliability Cronbachrsquos alpha and composite reliability were tested

Composite reliability should be equal to or greater than 07 (Hair et al 2010) Convergent

validity was assessed from the measurement model by determining the statistical significance

of each indicatorrsquos estimated pattern coefficient on its specified underlying construct

Discriminant validity was assessed by determining whether the confidence interval around

the correlation estimate between two factors included 100 (Anderson and Gerbing 1988)

For model fit assessment goodness-of-fit index (GFI) comparative fit index (CFI)

and the root mean square error of approximation (RMSEA) were used The model achieves

an acceptable fit to the data when GFI and CFI equals or exceeds 090 and RMSEA values

4 Applying standard MLE to ordinal data such as Likert-type items may produce significant estimation

problems such as inflation of chi-square fit statistics and biased underestimation of parameters and standard

errors (Flora and Curran 2004) Coursey and Pandey (2007a 2007b) used a modified MLE using the asymptotic

covariance matrix and D-WLS for adjusting the ordinal nature of the data They reported that although the

modified MLE findings are slightly different on a few model parameters the differences are not substantive to

the conclusions However model fit estimators were clearly better for the D-WLS 5 Given normally distributed categorical variables MLE can be used with confidence when a variable has four

or more categories If there is a large sample a five-point Likert scale and the skewness and kurtosis of variables

is within acceptable limits it suggests distribution symmetry ―In most cases where the hypothesized model is

well specified and the scaling based on more than three categories it seems unlikely that there will be much

difference between the findings (Byrne 2010 160)

- 14 -

fall below 008 (Hair et al 2010) In general the larger the value of GFI and CFI and the

smaller the value of RMSEA the better the fit of the model (Bollen 1989) We also present

the χ2 test statistic Lower values of χ

2 indicate a better fit and should be nonsignificant but

the χ2 value increases as sample size increases and thus for large sample sizes this statistic

may lead to rejection of a model with good fit (Hair et al 2010)

Results

Using SPSS 180 descriptive statistics were computed for individual items as shown in table

1 One item (APP3) in the APP dimension ―I like to discuss topics regarding public programs

and policies with others was mistakenly omitted in the survey conducted in the United

States and so it was deleted in the statistical analysis In the assessment of normality no

items showed a skew or kurtosis value greater than the cutoffs of |3| or |8| recommended by

Kline (2005) As data were distributed normally MLE as well as D-WLS estimation was used

Based on inspection of item-total correlations one item (COM4) was dropped from further

analysis because it was weakly correlated with the overall index (011) Missing data is an

issue in CFA as it can cause estimation errors All variables in the survey have missing data

rates of approximately 016 to 143 Mean substitution was used to calculate replacement

values for the missing values in each countryrsquos dataset (Hair et al 2010) The deletion of the

two items left 31 items for the next stage of analysis

The five-correlated factor model was tested using CFA The CFA model in the initial

analysis hypothesized a priori that (1) responses to the 30-item PSM measurement

instrument could be explained by five factors such as APP CPV1 CPV2 COM SS (2) each

item would have a nonzero loading on the PSM factor that it was designed to measure and

- 15 -

zero loadings on all other factors (3) the five factors would be correlated and (4)

measurement error terms would be uncorrelated The resulting CFA with MLE suggested that

the initial five-correlated factor model was not a good fit to the data χsup2 (df = 424) = 43699 p

lt 001 GFI = 0899 CFI = 835 RMSEA = 0057 [055 059] At this stage APP was highly

correlated with CPV1 (r = 996 in MLE and r = 988 in D-WLS) suggesting the two

dimensions were not unique and lacked discriminant validity

[Table 3 about here]

Given the disconfirmation of the initial model an effort was made to estimate an

alternative model The two dimensions APP and CPV1 were combined to form a single

dimension Attraction to Public Service (APS) The items of CPV1 may be regarded as

action-oriented by the respondents because dedication to public service community and

common good is underlined in these items Consequently it is reasonable to combine APP

with CPV1 to form APS a dimension that focuses more on disposition to serve the public to

work for the common good and to participate in public policy processes After the overall

number of dimensions was reduced to four using standardized factor loadings and

modification indices as criteria the items with the lowest factor loading in each dimension

were deleted and the items whose modification index indicated association with multiple

dimensions were eliminated A CFA model with the remaining items was iteratively tested

until it achieved a good fit to the data The outcome of this refinement process was a 16-item

index of four factors Overall model fit by every measure is quite strong for the D-WLS

estimation technique The estimation results for the four-dimension model are presented in

table 4

- 16 -

[Table 4 about here]

Cronbachrsquos coefficient alpha for the 16-item PSM index was 860 and the coefficient

alphas for the four dimensions ranged from 645 to 784 The composite reliability of the set

of reflective indicators for each dimension of PSM ranged from 713 to 826 and indicated

adequate internal consistency The resulting factor structure showed a four-factor structure

with all items loading significantly onto their a priori dimension (p lt 001) and the

standardized factor loadings ranging from 0490 to 0766 in MLE and from 515 to 807 in

D-WLS The results provide support for convergent validity The correlation estimates

between the two factors ranged from 0462 to 0848 in MLE and from 505 to 859 in D-

WLS The confidence intervals (plusmn2 standard errors) around the correlation estimates between

the two factors did not include 100 The result provides support for discriminant validity To

test whether the four-dimension model is superior to other designs such as the two or three-

dimension models a series of CFA were conducted and the results show that the four-

dimension model is a significantly better fit than any of the other designs

To test whether the factorial structure of the PSM instrument is equivalent across the

samples from the twelve countries four nested models were tested as part of multi-group

analysis with MLE (Byrne 2010) First the unconstrained model was the baseline model

which relaxed all equality constraints This model hypothesized the same four-factor structure

of the instrument across twelve countries The fit indices for this baseline model were χsup2 (df =

1176) = 24977 p lt 001 GFI = 0901 CFI = 900 RMSEA = 0020 [019 021] These

indices indicate that the four-factor model of the PSM instrument exhibits acceptable fit

across the samples from the twelve countries Second the measurement weights model tested

- 17 -

the invariance of factor loadings across countries by placing equality constraints on these

parameters The fit indices for this model were χsup2 (df = 1308) = 28088 p lt 001 GFI = 0890

CFI = 887 RMSEA = 0020 [019 021] Since the measurement weights model was nested

within the unconstrained model the chi-square difference text ∆χsup2 (132) = 3111 p lt 001

indicated that some equality constraints of factor loadings did not hold across the twelve

countries Further test results suggested that the equality constraints of structural covariances

and measurement residuals were not upheld across the twelve countries It needs to be

stressed that despite the differences in factor loadings and factor covariances the results

support the main assumption that the same four-factor structure holds across different

countries We may assume that people in different cultures understand PSM in the same

fashion

To find differences in the factor structure of PSM among countries we conducted a

series of CFA with each countryrsquos dataset and followed the same procedure as when

analyzing the sample as a whole As table 5 shows although there are great similarities

between the appropriate items in the countries there are also some differences Even though

the items vary according to national differences the four-factor structure had a good fit to the

data Thus the results confirm that the four-factor structure of PSM can be generalized

[Table 5 about here]

The APS dimension covers both an individualrsquos attraction to public participation and

commitment to public interest but all its items refer to some notion of public interest such as

improving public service aiding community tackling social problems meaningful public

service and contributing to the common good The APS dimension captures an individualrsquos

- 18 -

disposition to serve the public interest through public participation When developing the

items some assumed that the items of committing to public interests can be separated from

those of pursuing public values in the CPV dimension The empirical test showed that

respondents tended to understand committing to public interests as behavior-oriented and so

the items for public interest are not differentiated from participation-oriented items in their

perceptions but are different from those for public values The respondents seem to

understand that pursuing public values is more related to personal beliefs or creeds Thus the

CPV dimension is solely focused on public values and the items of pursing public interests

are combined with those of the APP dimension to make the APS dimension

Public values and their relative importance are neither impervious to change over

time nor the same among all countries (Charles de Jong and Ryan 2011 Davis and West

2009) Moreover some values may be in conflict as Martinsen and Beck Joslashrgensen (2010)

discussed in regard to the tensions between accountability and efficiency It is therefore not

possible to include all public values in one dimension (Beck Joslashrgensen and Bozeman 2007)

Accordingly the items of the CPV dimension are developed for estimating individual

commitment to seek public values which there are general consensus about and which are

generally formulated

When all four dimensions of PSM are included in a survey cross-national

equivalence of the survey can be assured The items represent reflections or manifestations of

each dimension and hence are relatively homogeneous and interchangeable Adding or

deleting items does not change the essential nature of the dimension Thus any combination

of items in a dimension can be regarded as equivalent When using structural equation

modeling (SEM) each dimension needs to have at least three reflective indicators in order to

achieve identification or it needs to have at least two reflective indicators and be allowed to

- 19 -

covary with another dimension that has at least two reflective measures With one indicator

the dimension is identified if the loading and the variance of the uniqueness are both fixed

(Bollen 1989 Edwards 2010) Thus the number of items in each dimension is not restricted

when measuring PSM However we recommend using multiple items for each dimension

because more research is needed to investigate the appropriateness of this PSM instrument

with varying samples

The test results showed that model fit estimators were clearly better for the D-WLS

It means that in this study MLE was more rigorous than D-WLS in assessing the fit of the

model The empirical evidence indicated that MLE may be an acceptable technique for

ordinal data when there is a large sample a five-point Likert scale and the skewness and

kurtosis of variables is within acceptable limits as suggested by Byrne (2010)

Discussion

The analysis reaffirmed the four-dimensional structure of PSM but the results were not

entirely consistent with the original CFA we tested We were ultimately able to converge on a

good-fit operational model for PSM that was also consistent with underlying theory The

analysis is not able to resolve the debate about measurement models however for reasons

discussed below

Choosing the correct measurement model (ie reflective vs formative) is important

as measurement model misspecification results not only in misrepresentation of the epistemic

relationships between constructs and their measures but also in biased parameter estimates of

structural relationships (Jarvis et al 2003 Law and Wong 1999 MacKenzie 2003

MacKenzie Podsakoff and Jarvis 2005) A formative measurement model in isolation is

- 20 -

under-identified and cannot be estimated in the case of covariance-based SEM (CB-SEM)

(Bollen 1989) A construct with formative indicators must emit paths to at least two unrelated

reflective indicators two unrelated latent constructs or some combination of the two in order

to solve the problem of under-identification (Jarvis MacKenzie and Podsakoff 2003

MacCallum and Browne 1993 MacKenzie Podsakoff and Jarvis 2005) Thus ―when

dependent constructs change the empirical nature of the formatively measured construct

changes (Wilcox Howell and Breivik 2008 1223) and ―when specifying formative

measurement models what might seem like an innocuous choice of outcomes to achieve

identification can have fundamental ramifications for the meaning of the formative construct

and the inferences based on it (Edwards 2010 7)

Earlier research found that the association between a construct and its measures

should remain stable regardless of the larger causal model in which the constructs and

measures are embedded (Edwards and Bagozzi 2000) The association between a construct

and its measures is generally stable for reflective measures but associations of formative

measures with their construct are unstable The susceptibility of formative measurement to

both interpretational confounding and external inconsistency recommends that formative

measurement is not an equally attractive alternative to reflective measurement and that

whenever possible researchers should opt for reflective measurement (Howell Breivik and

Wilcox 2007a 2007b Kim Shin and Grover 2010) In the case of second-order models

researchers are limited to only a second-order reflective model in CB-SEM (Chin 2010)

Regardless of the measurement debate and the tradeoffs between reflective or

formative specification this study makes a theoretical argument that PSM is a second-order

formative construct This second-order formative factor model (reflective first-order

formative second-order) is also suggested as an alternative model that captures the

- 21 -

multidimensionality of complex concepts and avoids the shortcomings of first-order

formative measurement (Edwards 2010) However this model itself cannot achieve

identification One possible strategy is to develop direct reflective measures of PSM based on

the view that PSM is a general concept that is not only influenced by its specific dimensions

but also can be assessed using general measures (Edwards 2010 Javis MacKenzie and

Podsakoff 2003)6 However these general measures have not yet been developed for the

PSM construct

Another way around the problem of underidentification of formative models is to

apply the component-based approach known as partial least squares path modeling (PLS-PM)

which can model formative indicators (Fornell and Bookstein 1982 Goumltz Liehr-Gobbers

and Krafft 2010) The CB-SEM and the PLS-PM methods have distinctive statistical

characteristics and selecting an approach to SEM depends on the particular research situation

CB-SEM is the method of choice for theory testing while PLS-PM is appropriate for

prediction-oriented applications (Ringle Goumltz Wetzels and Wilson 2009) One advantage of

the PLS method is that it demands significantly fewer requirements compared to that of

covariance structure analyses but nevertheless delivers consistent estimation results Another

asset of the PLS approach is its ability to deal with formative as well as reflective indicators

even within one structural equation model This indicates that the PLS approach can be

appropriate for explorative analysis of structural equation models (Chin 2010 Goumltz Liehr-

Gobbers and Krafft 2010) A Monte Carlo simulation study to compare CB-SEM and PLS-

PM showed that PLS-PM provides a viable approximation of model parameters when the

prerequisites for CB-SEM are not met (Ringle Goumltz Wetzels and Wilson 2009) A formative

6 For example job satisfaction can be assessed with general reflective measures such as ―in general I am

satisfied with my job and the formative dimensions can be assessed with measures that describe satisfaction

with each job facet (Edwards 2010 15)

- 22 -

composite in the form of indexes needs to be tested to know the relative weights (γ-

parameters) of the dimensions of PSM In PLS-PM a second-order formative factor is

directly measured by observed variables for all the first order factors While this approach

repeats the number of manifest variables used the model can be identified without adding

outcome constructs that would be considered inappropriate or irrelevant from a conceptual

standpoint (Chin 2010 Edwards 2010) The test result of PLS-PM using Smart PLS 20

(Ringle Wende and Will 2005) produced the function PSM = 31APS + 36CPV + 28COM

+ 31SS However one consideration is whether and how construct error is modeled for the

formative construct (Diamantopoulos 2006) When using PLS-PM the researcher is making

the assumption that there is no error at the construct level as all constructs in PLS-PM are

modeled without error When using PLS-PM the absence of an error term will tend to

increase the weights therefore inflating their importance (Bollen and Lennox 1991) The

primary statistic for assessing a formative indicator is its weight the partialized effect of the

indicator on its intended construct controlling for the effects of all other indicators of that

construct (Cenfetelli and Bassellier 2009) A formative measure is essentially a multiple

regression with the construct representing the dependent variable and the indicators as the

predictors (Diamantopoulos and Winklhofer 2001) PSM may be regarded as a weighted

linear composite of its dimensions The weights used to create the composites are different

across CE-SEM and PLS-PM The causal dimensional weights generated in CB-SEM are

estimated in a model in which a construct-level error term is specified (eg PSM = γ1APM +

γ2CPI + γ3COM + γ4SS + ζ) whereas the weights in PLS-PM are not (eg PSM = γ1APM +

γ2CPI + γ3COM + γ4SS) This leads to two different conceptualizations of the PSM construct

one in which the causal indicators do not capture the construct in its entirety and another in

which they do (Hardin et al 2010) These differences need to be discussed in future research

- 23 -

in this area

One practical suggestion to measure PSM is to apply unit weights to its dimensions if

we assume the effects of the four dimensions are equivalent A literature review indicates that

unit weights have substantial predictive validity when compared with regression weights The

literature shows that the empirical correlations between scores generated from differential and

unit-weighting approaches are quite high In creating composite scores unit weights can be

an appropriate approach for weighting under many circumstances (Bobko Roth and Buster

2007) Although the uses of unit-weight composites will likely result in the explanation of

less variance than empirical estimations of dimensional weights they will remain consistently

defined across studies and hence suitable for theory testing Alternatively more thorough

discussions on the dimensions of PSM may suggest fixed-weight composites (Hardin et al

2010) A previous study (Kim 2010 22) also suggested ―Overall indexing PSM by

measuring all the dimensions of PSM and simply summing the averages of each dimension

can make it easier to measure PSM in practice as PSM = APS + CPV + COM + SS

An advantage of regarding PSM as a second-order formative construct is that the

dimensions can be analyzed both individually and collectively This is important as the

dimensions are expected to have different antecedents and consequences For instance Perry

(1997) found some significant differences in the influence of independent variables on the

four dimensions of PSM Taylor (2007) confirmed that certain dimensions are more important

than others in influencing work outcomes Moynihan and Pandey (2007a) reported that red

tape is negatively related to one dimension but reform orientation and hierarchical authority

is positively related to another dimension A recent study found that the SS dimension has a

positive relationship with occupational intention but the other dimensions do not (Liu et al

2011) Thus it is valuable to analyze the causes and effects of individual dimensions of PSM

- 24 -

as well as of the PSM construct

Conclusion

The purpose of this study is to develop a measurement instrument for PSM for cross-national

research and comparison On the basis of the sharpened construct of PSM (Kim and

Vandenabeele 2010) using data from an international survey conducted in 12 countries we

found that the four-factor structure of PSM is generalizable across country cases This study

provides the measurement index of PSM with the four dimensions of APS CPV COM and

SS The APS dimension measures the instrumental motives of PSM the CPV dimension

determines the value-based motives of PSM while the COM and SS dimensions gauge the

identification motives and self-sacrifice respectively When all four dimensions of PSM are

used in a survey cross-national equivalence of survey items can be assured The items for

measuring PSM in future research are suggested in Appendix 4

This study has several limitations First looking for samples which were comparable

to the largest possible extent we opted for surveys in local government The results found

and discussed in this study are based on PSM among local government employees The

results might be different when surveying central government employees or other specific

groups in the public sector Thus the potential impact of the sample bias needs to be

considered in understanding this study It also means that the survey items for measuring

PSM should be examined with different samples in various areas and at different government

levels

Second the test results showed that there is a high degree of correlation among the

dimension such as APS CPV and COM Future research should not only investigate whether

- 25 -

the items reflect the core content of the dimensions but also continue to develop more

appropriate items for better discriminant validity of each dimension Third the content of

APP dimension was assumed to be distinct from the CPV dimension that focuses on personal

disposition to pursue public interest and public values However the public interest sub-

dimension was separated from the public values sub-dimension in the CPV dimension and

combined with the APP dimension to form the APS dimension Further research should

analyze the CPV dimension and investigate the appropriateness of the APS dimension

In order to facilitate research internationally the PSM construct needs to be

improved conceptually and operationally Kim and Vandenabeele (2010) have dealt with the

conceptual composition and operational dimension of PSM as well as the measurement

model of PSM The next step is to develop an internationally robust measurement instrument

of PSM This study proposes that scholars use all four of the dimensions in measuring the

conceptual components of PSM such as self-sacrifice instrumental value-based and affective

motives because they collectively represent PSM It provides a set of appropriate items for

measuring PSM and confirms the dimensional structure of PSM The survey results suggest

that people in different cultures understand PSM in the same fashion We know that this study

is simply a step along the way to developing a more robust measure but we hope it

contributes to enhanced cross-national research and comparison along with the generation of

cumulative knowledge internationally

- 26 -

REFERENCES

Anderson James C and David D Gerbing 1988 Structural Equation Modeling in Practice

A Review and Recommended Two-Step Approach Psychological Bulletin 103

411-423

Arbuckle James L 2009 Amos 180 Userrsquos Guide Crawfordville FL Amos Development

Co

Beck Joslashrgensen Torben and Barry Bozeman 2007 Public Values An Inventory

Administration amp Society 39(3) 354-381

Beutel Ann M and Margaret Mooney Marini 1995 Gender and values American

Sociological Review 60 436 - 448

Bobko Philip Philip L Roth and Maury A Buster 2007 The Usefulness of Unit Weights in

Creating Composite Scores A Literature Review Application to Content Validity and

Meta-Analysis Organizational Research Methods 10(4) 689-709

Bollen Kenneth A 1989 Structural Equations with Latent Variables New York Wiley

Bollen K A and R Lennox 1991 Conventional Wisdom on Measurement A Structural

Equation Perspective Psychological Bulletin 110(2) 305-314

Byrne Barbara M 2010 Structural Equation Modeling with Amos Basic Concepts

Applications and Programming 2nd

ed New York Routledge

Camilleri Emanuel 2006 Towards Developing an Organizational Commitment ― Public

Service Motivation Model for Maltese Public Service Employees Public Policy

and Administration 21 63-83

Castaing Seacutebastien 2006 The Effects of Psychological Contract Fulfillment and Public

Service Motivation on Organizational Commitment in the French Civil Service

Public Policy and Administration 21 84-98

Cenfetelli Ronald T and Geneviegraveve Bassellier 2009 Interpretation of Formative

Measurement in Information Systems Research MIS Quarterly 33(4) 689-707

Charles Michael B W Martin de Jong and Neal Ryan 2011 Public Values in Western

- 27 -

Europe A Temporal Perspective American Review of Public Administration 41

75-91

Chin Wynne W 2010 How to Write Up and Report PLS Analysis In Handbook of Partical

Least Squares edited by V Esposito Vinzi et al 655-690 Berlin Springer-Verlag

Coursey David H and Sanjay K Pandey 2007a Public Service Motivation Measurement

Testing an Abridged Version of Perryrsquos Proposed Scale Administration amp Society

39 547-568

Coursey David H and Sanjay K Pandey 2007b Content Domain Measurement and

Validity of the Red Tape Concept American Review of Public Administration 37

342-361

Coursey David H James L Perry Jeffrey L Brudney and Laura Littlepage 2008

Psychometric Verification of Perryrsquos Public Service Motivation Instrument Results

for Volunteer Exemplars Review of Public Personnel Administration 28 79-90

Davis Paul and Karen West 2009 What Do Public Values Mean for Public Action Putting

Public Values in Their Plural Place American Review of Public Administration 39

602-618

DeHart-Davis Leisha Justin Marlowe and Sanjay K Pandey 2006 Gender Dimensions of

Public Service Motivation Public Administration Review 66 873-887

Diamantopoulos Adamantios 2006 The error term in formative measurement models

Interpretation and modeling implications Journal of Modeling in Management 1(1)

7-17

Diamantopoulos Adamantios and Judy A Siguaw 2006 Formative versus Reflective

Indicators in Organizational Measure Development A Comparison and Empirical

Illustration British Journal of Management 17 263-282

Diamantopoulos Adamantios and Heidi M Winklhofer 2001 Index Construction with

Formative Indicators An Alternative to Scale Development Journal of Marketing

Research 38 269-277

Edwards Jeffrey R 2010 The Fallacy of Formative Measurement Organizational Research

- 28 -

Methods Online First published on August 19 2010 as

doi1011771094428110378369

Edwards Jeffrey R and Richard P Bagozzi 2000 On the Nature and Direction of

Relationships between Constructs and Measures Psychological Methods 5 155-

174

Flora David B and Patrick J Curran 2004 An Empirical Evaluation of Alternative

Methods of Estimation for Confirmatory Factor Analysis with Ordinal Data

Psychological Methods 9 466-491

Fornell Claes and Fred L Bookstein 1982 Two structural equations models LISREL and

PLS applied to customer exit-voice theory Journal of Marketing Research 19 440-

452

Giauque David Adrian Ritz Freacutedeacuteric Varone Simon Anderfuhren-Biget and Christian

Waldner 2011 Putting Public Service Motivation into Context A Balance between

Universalism and Particularism Forthcoming in International Review of

Administrative Sciences 77(2)

Goumltz Oliver Kerstin Liehr-Gobbers and Manfred Krafft 2010 Evaluation of Structural

Equation Models Using the Partial Least Squares (PLS) Approach In V Esposito

Vinzi et al (eds) Handbook of Partial Least Squares 691-711 DOI 101007978-3-

540-32827-8_30 Springer-Verlag Berlin Heidelberg

Guinot Christiane Julie Latreille and Michel Tenenhaus 2001 PLS Path Modelling and

Multiple Table Analysis Application to the Cosmetic Habits of Women in Ile-de-

France Chemometrics and Intelligent Laboratory Systems 58 247-259

Hair Joseph F Jr William C Black Barry J Babin and Rolph E Anderson 2010

Multivariate Data Analysis 7th

ed Upper Saddle River NJ Pearson Education

Hardin Andrew M Jerry Cha-Jan Chang Mark A Fuller and Gholamreza Torkzadeh 2010

Formative Measurement and Academic Research In Search of Measurement

Theory Educational and Psychological Measurement Online First published on

June 11 2010 as doi101177001316440370208

- 29 -

Howell Roy D Einar Breivik and James B Wilcox 2007a Reconsidering Formative

Measurement Psychological Methods 12 205-218

Howell Roy D Einar Breivik and James B Wilcox 2007b Is Formative Measurement

Really Measurement Reply to Bollen (2007) and Bagozzi (2007) Psychological

Methods 12 238-245

Jarvis Cheryl Burke Scott B MacKenzie and Philip M Podsakoff 2003 A Critical Review

of Construct Indicators and Measurement Model Misspecification in Marketing and

Consumer Research Journal of Consumer Research 30 199-218

Joumlreskog Karl G and Dag Soumlrbom 2005 LISREL 872 (Computer Software) Lincolnwood

IL Scientific Software International

Kim Gimun Bongsik Shin and Varun Grover 2010 Investigating Two Contradictory Views

of Formative Measurement in Information Systems Research MIS Quarterly 34

345-365

Kim Sangmook 2009a Revising Perryrsquos Measurement Scale of Public Service Motivation

American Review of Public Administration 39 149-163

Kim Sangmook 2009b Testing the Structure of Public Service Motivation in Korea A

Research Note Journal of Public Administration Research and Theory 19 839-

851

Kim Sangmook 2010 Testing a Revised Measurement of Public Service Motivation

Reflective versus Formative Specification Journal of Public Administration

Research and Theory Doi101093jopartmuq048

Kim Sangmook and Wouter Vandenabeele 2010 A strategy for building public service

motivation research internationally Public Administration Review 70 701-709

Kline Rex B 2005 Principles and Practice of Structural Equation Modeling 2nd

ed New

York Guilford Press

Law Kenneth and Chi-Sum Wong 1999 Multidimensional Constructs in Structural

Equation Analysis An Illustration using the Job Perception and Job Satisfaction

Constructs Journal of Management 25 143-160

- 30 -

Leisink Peter and Bram Steijn 2009 Public Service Motivation and Job Performance of

Public Sector Employees in the Netherlands International Review of

Administrative Sciences 75 35-52

Liu Bangcheng Chun Hui Jin Hu Wensheng Yang and Xinli Yu 2011 How well can

Public Service Motivation Connect with Occupational Intention International

Review of Administrative Sciences 77 191-211

Liu Bangcheng Ningyu Tang and Xiaomei Zhu 2008 Public Service Motivation and Job

Satisfaction in China An Investigation of Generalisability and Instrumentality

International Journal of Manpower 29 684-699

MacCallum Robert C and Michael W Browne 1993 The Use of Causal Indicators in

Covariance Structure Models Some Practical Issues Psychological Bulletin 114

533-541

MacKenzie Scott B 2003 The dangers of poor construct conceptualization Journal of the

Academy of Marketing Science 31(3) 323-326

MacKenzie Scott B Philip M Podsakoff and Cheryl Burke Jarvis 2005 The Problem of

Measurement Model Misspecification in Behavioral and Organizational Research

and Some Recommended Solutions Journal of Applied Psychology 90 710-730

Martinsen Dorte Sindbjerg and Torben Beck Joslashrgensen 2010 Accountability as a

Differentiated Value in Supranational Governance American Review of Public

Administration 40 742-760

Moynihan Donald P and Sanjay K Pandey 2007a The Role of Organizations in Fostering

Public Service Motivation Public Administration Review 67 40-53

Moynihan Donald P and Sanjay K Pandey 2007b Finding Workable Levers over Work

Motivation Comparing Job Satisfaction Job Involvement and Organizational

Commitment Administration amp Society 39 803-832

Pandey Sanjay K and Edmund C Stazyk 2008 Antecedents and Correlates of Public

Service Motivation In Motivation in Public Management edited by James L Perry

and Annie Hondeghem 101-117 Oxford Oxford University Press

- 31 -

Perry James L 1996 Measuring Public Service Motivation An Assessment of Construct

Reliability and Validity Journal of Public Administration Research and Theory 6

5-22

Perry James L 1997 Antecedents of Public Service Motivation Journal of Public

Administration Research and Theory 7 181-197

Perry James L 2000 Bringing Society in Toward a Theory of Public Service Motivation

Journal of Public Administration Research and Theory 10 471-488

Perry James L and Annie Hondeghem 2008a Motivation in Public Management Oxford

Oxford University Press

Perry James L and Annie Hondeghem 2008b Directions for Future Theory and Research

In Motivation in Public Management edited by James L Perry and Annie

Hondeghem 294-313 Oxford Oxford University Press

Perry James L Annie Hondeghem and Lois Recascino Wise 2010 Revisiting the

Motivational Bases of Public Service Twenty Years of Research and an Agenda for

the Future Public Administration Review 70(5) 681-690

Perry James L and Wouter Vandenabeele 2008 Behavioral dynamics Institutions

Identities and Self-Regulation In Motivation in Public Management edited by

James L Perry and Annie Hondeghem 56-79 Oxford Oxford University Press

Perry James L and Lois Recascino Wise 1990 The Motivational Bases of Public Service

Public Administration Review 50 367-373

Rainey Hal G 1982 Reward Preference among Public and Private Managers In Search of

the Service Ethic American Review of Public Administration 16 288-302

Ringle Christian M Oliver Goumltz Martin Wetzels and Bradley Wilson 2009 On the Use of

Formative Measurement Specifications in Structural Equation Modeling A Monte

Carlo Simulation Study to Compare Covariance-Based and Partial Least Squares

Model Estimation Methodologies Faculty of Economics and Business Administration

Maastricht Univeristy

Ringle C M S Wende and S Will 2005 Smart PLS 20 (M3) Beta

- 32 -

httpwwwsmartplsde

Ritz Adrian 2011 Attraction to Public Policy Making A Qualitative Inquiry into

Improvements in PSM Measurement Forthcoming in Public Administration

Spector Paul E 1992 Summated Rating Scales Construction An Introduction Newbury

Park CA Sage

Steenbergen Marco R 1996 Compassion and American public opinion An analysis of the

NES Humanitarianism Scale Report to the Board of Overseers of the National

Election Studies

Steffen Patrick R and Kevin S Masters 2005 Does compassion mediate the intrinsic

religion-health relationship Annals of Behavioral Medicine 30 217-223

Taylor Jeannette 2007 The Impact of Public Service Motives on Work Outcomes in

Australia A Comparative Multi-Dimensional Analysis Public Administration 85

931-959

Vandenabeele Wouter 2007 Towards a Theory of Public Service Motivation An

Institutional Approach Public Management Review 9 545-556

Vandenabeele Wouter 2008a Government Calling Public Service Motivation as an Element

in Selecting Government as an Employer of Choice Public Administration 86

1089-1105

Vandenabeele Wouter 2008b Development of a Public Service Motivation Measurement

Scale Corroborating and Extending Perryrsquos Measurement Instrument International

Public Management Journal 11 143-167

Vandenabeele Wouter 2009 The Mediating Effect of Job Satisfaction and Organizational

Commitment on Self-Reported Performance More Robust Evidence of the PSM-

Performance Relationship International Review of Administrative Sciences 75 11-

34

Wilcox James B Roy D Howell and Einar Breivik 2008 Questions about Formative

Measurement Journal of Business Research 61 1219-1228

Wise Lois Recascino 2000 The Public Service Culture In Public Administration Concepts

- 33 -

and Cases 7th ed edited by Richard J Stillman II 342-353 Boston Houghton

Mifflin

Wright Bradley E 2008 Methodological Challenges associated with Public Service

Motivation Research In Motivation in Public Management edited by James L

Perry and Annie Hondeghem 80-98 Oxford Oxford University Press

Wright Bradley E and Sanjay K Pandey 2008 Public Service Motivation and the

Assumption of Person-Organization Fit Testing the Mediating Effect of Value

Congruence Administration amp Society 40 502-521

- 34 -

Table 1

Possible Indicators of PSM and Descriptive Statistics (n = 2868)

Dimensions and items Mean Standard

Deviation

Item-Total

Correlation

Attraction to Public Participation (APP)

APP1 I am interested in helping to improve public service

APP2 I am satisfied when I see people benefiting from the public programs I was

involved in

APP3 I like to discuss topics regarding public programs and policies with others

APP4 I believe that public sector activities contribute to our general welfare

APP5 I admire people who initiate or are involved in activities to aid my

community

APP6 Contributing to public programs and policies helps me realize myself

APP7 It is important to contribute to activities that tackle social problems

412

441

355

393

423

351

409

696

662

917

753

672

881

734

524

421

482

412

493

478

552

Commitment to Public Values (CPV)

Sub-dimension for Public Interests (CPV1)

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

CPI3 I would prefer seeing public officials do what is best for the whole

community even if it harmed my interests

CPI4 Serving the public interest is more important than helping a single individual

Sub-dimension for Public Values (CPV2)

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of public

services

CPV3 It is fundamental that public services respond to the needs of the citizens

CPV4 Decisions regarding public services should be democratic despite the time

and effort it takes

CPV5 Everybody is entitled to a good service even if it costs a lot of money

CPV6 It is fundamental that the interests of future generations are taken into

account when developing public policies

CPV7 To act ethically is essential for public servants

CPV8 I believe that public employees must always be aware of the legitimacy of

their activities

CPV9 I personally identify with the aim of protecting individual liberties and rights

418

409

348

337

428

416

433

394

367

426

441

434

414

695

662

899

995

730

704

710

852

932

689

681

692

763

525

596

480

251

484

437

356

325

376

439

462

402

392

Compassion (COM) COM1 It is difficult for me to contain my feelings when I see people in distress

COM2 I feel sympathetic to the plight of the underprivileged

COM3 I empathize with other people who face difficulties

COM4 I have little compassion for people in need who are unwilling to take the

first step to help themselves

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

354

387

388

299

412

391

951

825

707

1083

755

765

399

511

389

011

487

502

Self-Sacrifice (SS)

SS1 Making a difference to society means more to me than personal achievements

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS5 People should give back to society more than they get from it

SS6 Serving other citizens would give me a good feeling even if no one paid me for

it

SS7 I would agree to a good plan to make a better life for the poor even if it costs

me money

357

329

339

302

345

363

350

904

872

949

982

888

1049

895

508

584

454

531

412

368

546

- 35 -

Table 2

Background of Respondents (n = 2868)

Variables Characteristics Respondents

Sex

Male

Female

NA

1209

1622

37

422

566

13

Length of service

(years)

0 ~ 10

10 ~ 20

20 ~ 30

30+

NA

1126

723

484

491

44

393

252

169

171

15

Organizational status

―Do you supervise employees

No

Yes

NA

1472

1076

320

513

375

112

Country

Australia

Belgium

China

Denmark

France

Italy

Korea

Lithuania

The Netherlands

Switzerland

United Kingdom

USA

249

214

230

249

266

162

253

236

249

250

260

250

87

75

80

87

93

56

88

82

87

87

91

87 Note NA = no answer

- 36 -

Table 3

Standardized factor loadings and correlations for the 31-item PSM index (n = 2868)

Dimensions and items SFL (MLE) SFL (D-WLS) Alpha

APP

APP1

APP2

APP4

APP5

APP6

APP7

597

486

455

558

490

600

657

552

514

625

549

665

699

CPV1

CPI1

CPI2

CPI3

CPI4

588

660

488

289

653

732

535

294

542

CPV2

CPV1

CPV2

CPV3

CPV4

CPV5

CPV6

CPV7

CPV8

CPV9

611

546

436

393

418

534

598

496

460

690

631

505

457

486

601

669

573

530

745

COM

COM1

COM2

COM3

COM5

COM6

496

569

498

597

607

516

651

538

670

684

687

SS

SS1

SS2

SS3

SS4

SS5

SS6

SS7

567

736

558

728

467

409

663

660

762

591

695

515

494

709

782

Measures of fit (MLE) χsup2 (df = 424) = 43699 p lt 001 CFI = 835 GFI = 899

RMSEA = 057 [055 059]

Measure of fit (D-WLS) χsup2 (df = 424) = 39247 p lt 001 CFI = 971 GFI = 976

RMSEA = 054 [052 055]

Inter-factor correlations MLE (D-WLS)

1 2 3 4

1 APS

2 CPV1 996 (988)

3 CPV2 902 (890) 798 (798)

4 COM 787 (783) 758 (762) 779 (783)

5 SS 700 (742) 800 (859) 509 (537) 720 (729)

Note SFL = standardized factor loading All factor loadings and correlations are significant at p lt 001

- 37 -

Table 4

Standardized factor loadings and correlations for the 16-item PSM index (n = 2868)

Dimensions and items SFL

(MLE)

SFL

(D-WLS)

APS

APP5 I admire people who initiate or are involved in activities to aid my

community

APP7 It is important to contribute to activities that tackle social problems

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

611

592

637

666

665

669

690

751

CPV

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of

public services

CPV6 It is fundamental that the interests of future generations are taken

into account when developing public policies

CPV7 To act ethically is essential for public servants

612

546

551

590

698

631

618

658

COM

COM1 It is difficult for me to contain my feelings when I see people in

distress

COM2 I feel sympathetic to the plight of the underprivileged

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

490

546

596

604

515

608

669

679

SS

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS7 I would agree to a good plan to make a better life for the poor even if

it costs me money

763

571

766

678

807

629

747

755

Measures of fit (MLE)

χsup2 (df = 98) = 8490 p lt 001

CFI = 939 GFI = 965

RMSEA = 052 [049 055]

Measure of fit (D-WLS)

χsup2 (df = 98) = 6069 p lt 001

CFI = 988 GFI = 992

RMSEA = 043 [039 046]

Alpha CR (D-WLS) Inter-factor correlations MLE (D-WLS)

1 APS 717 788 1 APS 2 CPV 3 COM

2 CPV 663 747 848 (859)

3 COM 645 713 781 (791) 789 (795)

4 SS 784 826 565 (596) 462 (505) 715 (726)

Note SFL = standardized factor loading Alpha = Cronbachrsquos alpha CR = composite reliability

All factor loadings and correlations are significant at p lt 001

- 38 -

Table 5

Results of CFA in the individual countries Standardized factor loadings and measures of fit

Factors

amp Items

Australia Belgium China Denmark France Italy

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

672

756

799

785

804

943

710

703

597

790

799

730

584

675

685

681

731

742

578

748

674

643

765

739

645

704

678

708

812

779

632

604

765

766

672

853

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

552

593

560

699

727

632

624

605

560

713

709

645

702

602

616

724

686

735

567

560

715

654

605

811

608

658

587

696

755

698

673

620

523

782

697

654

COM

COM1

COM2

COM3

COM5

COM6

623

584

662

688

701

739

732

608

475

808

666

581

578

616

605

649

670

673

699

751

737

679

816

776

568

546

801

626

673

850

637

546

741

715

601

822

SS

SS2

SS3

SS4

SS7

654

794

806

892

767

813

727

693

799

793

725

838

857

820

773

896

855

842

806

749

792

829

753

884

847

731

753

889

816

753

815

692

664

795

782

734

Inter-factor

correlations

093

~

752

163

~

810

320

~

773

323

~

825

452

~

827

497

~

855

323

~

850

321

~

847

498

~

899

536

~

902

425

~

829

476

~

865

Sample size 249 214 230 249 266 162

Measures of Fit

χsup2 638 444 881 732 765 554 817 625 735 565 750 564

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 980 100 939 987 969 997 967 994 975 997 949 995

GFI 961 986 941 986 948 991 949 992 955 992 926 986

RMSEA 036 00 063 050 051 026 053 035 045 026 059 033

- 39 -

Table 5 (Continued)

Factors

amp Items

Korea Lithuania The Netherlands Switzerland UK USA

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

619

701

667

706

781

717

666

512

682

751

605

779

557

629

630

620

719

729

535

621

702

643

694

735

655

725

738

710

804

822

535

621

702

643

694

735

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

594

607

722

722

644

722

674

535

533

790

589

711

568

554

538

655

628

545

627

580

615

711

666

704

800

512

555

873

581

646

627

580

615

711

666

704

COM

COM1

COM2

COM3

COM5

COM6

694

677

550

763

736

627

557

639

428

550

696

534

737

623

508

766

616

608

659

592

683

703

661

769

682

583

729

739

644

803

659

592

683

703

661

769

SS

SS2

SS3

SS4

SS7

776

752

679

779

770

811

707

721

635

725

734

706

814

656

743

820

767

764

745

632

689

788

662

744

671

818

832

830

864

802

745

632

689

788

662

744

Inter-factor

correlations

432

~

726

466

~

746

376

~

788

406

~

902

377

~

852

455

~

836

637

~

818

660

~

828

317

~

852

360

~

852

637

~

818

660

~

828

Sample size 253 236 249 250 260 250

Measures of Fit

χsup2 896 573 612 462 773 616 815 715 769 542 815 715

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 951 996 975 100 958 993 957 991 972 998 957 991

GFI 942 988 960 989 951 987 948 990 955 990 948 990

RMSEA 059 028 034 00 050 034 053 044 048 022 053 044

Note All factor loadings and correlations are significant at p lt 001

- 40 -

Appendix 1 Dimensions and possible items of PSM (the first round)

1 Attraction to Public Participation

(1) I am interested in developing public programs helping my country or community (Kim 2009a)

(2) Sharing my views on public policies with others is attractive to me (Kim 2009a)

(3) I am satisfied when seeing people get benefits from the public programs I was involved in (Kim 2009a)

(4) I like to discuss political subjects with others (Giauque et al 2011)

(5) I think the governmental activities contribute to our welfare

(6) I like to initiate actions to help out my community

(7) I do not care much about politicians (Perry 1996)

2 Commitment to Public Values

Possible items for public interest

(1A) Meaningful public service is very important to me (Perry 1996)

(2A) It is important for me to contribute for the common good (Giauque et al 2011)

(3A) I recognize myself with the mission of protection of liberties and rights endorsed by the nation-state

(Giauque et al 2011)

(4A) I would prefer seeing public officials do what is best for the whole community even if it harmed my

interests (Perry 1996)

(5A) To me serving the public interest is more important than helping other people (Vandenabeele 2008b)

Possible items for public values

(1B) I recognize myself with the promotion of the equality of chances endorsed by the nation-state (Giauque et

al 2011)

(2B) It is fundamental that the public service provided to the citizen is regular and continuous (Giauque et al

2011)

(3B) It is fundamental that public service innovates to respond to the new needs of the citizensclients (Giauque

et al 2011)

(4B) The dignity and well-being of all should be the most important concerns in any society (Steenbergen

1996)

(5B) It is important that public servants account for all the costsexpenses they make (Vandenabeele 2008b)

(6B) We have to do everything in our power to pursue the goal of democracy

3 Compassion

(1) It is difficult for me to contain my feelings when I see people in distress (Perry 1996)

(2) I feel sympathetic to the plight of the underprivileged (Kim 2009a)

(3) I am often reminded by daily events how dependent we are on one another (Perry 1996)

(4) I feel compassion for others in the difficulties they are facing (Steffen and Masters 2005)

(5) It is not really my problem if others are in trouble and need help (Beutel and Marini 1995)

(6) I find it hard to be sympathetic toward starving people in foreign lands when there is so much trouble in our

own country (Beutel and Marini 1995)

(7) I get very upset when I see other people treated unfairly (Beutel and Marini 1995)

(8) I would agree to a good plan to make a better life for the poor even if it cost me money (Beutel and Marini

1995)

(9) To me patriotism includes seeing the welfare of others (Perry 1996)

(10) I care very much about other people

4 Self-Sacrifice

(1) Making a difference in society means more to me than personal achievements (Perry 1996)

(2) I am prepared to make enormous sacrifices for the good of society (Perry 1996)

(3) I believe in putting duty before self (Perry 1996)

(4) I am one of those rare people who would risk personal loss to help someone else (Perry 1996)

(5) I think people should give back to society more than they get from it (Perry 1996)

(6) Serving other citizens would give me a good feeling even if no one paid me for it (Perry 1996)

(7) Much of what I do is for a cause bigger than myself (Perry 1996)

- 41 -

Appendix 2 Newly suggested items through the first round

1 Items for the Attraction to Public Participation Dimension

(1) In the public domain every person should be counted as equal

(2) I enjoy to think and argue in a political way

(3) I am satisfied when seeing my ideas being integrated in public policies

(4) I can easily relate to the way of thinking and arguing of politicians

(5) Working in the public sector I can better realize myself

2 Items for the Public Values Sub-Dimension

(1) Public service should give value for money

(2) Cost considerations should play an important role in decisions about public service provision

(3) It is fundamental that the interests of future generations are taken into account when developing public

policies

(4) To act with correct ethics is for public servants as important as their professional competence

(5) Primarily public servants should be accountable to the public and not to their superiors

3 Items for the Self-Sacrifice Dimension

(1) I would agree to a good plan to make a better life for the poor even if it costs me money

- 42 -

Appendix 3 An example of a summary of comments and suggestions (the second round)

On the item ―I am interested in developing public programs helping my country or community

A It is a good sentence but is it possible for everyone to ―develop public programs Even public officials are

limited to do this

B OK

C or the special group

D Double barrel and difficult to translate to my country ―I am interested in helping to improve public services

E Good but I feel that wording requires revision If we draw sample from local government then country is

redundant ―I am interested in developing public policy programs that aim in helping my community

F 4 (General appreciation from1 to 4) Not clear whether it refers to ones work or to ones concern as a citizen

―(In my work) I am interested in contributing to the development of public etchellip

G Instead of development I would use implementing

H Delete public which is redundant because of the characteristic helping my country or community but add

or implementing ―I am interested in developing or implementing programs helping my country or

community

I Referring to dimension Wording Similar to item 6

Note Names have been changed to alphabets in order to ensure the anonymity of participants

- 43 -

Appendix 4 Dimensions and Items of Public Service Motivation for Future Research

Dimensions and items

Attraction to Public Service (APS)

1 I am interested in helping to improve public service

2 I admire people who initiate or are involved in activities to aid my community

3 It is important to contribute to activities that tackle social problems

4 Meaningful public service is very important to me

5 It is important for me to contribute to the common good

Commitment to Public Values (CPV)

1 I think equal opportunities for citizens are very important

2 It is important that citizens can rely on the continuous provision of public services

3 It is fundamental that the interests of future generations are taken into account when

developing public policies

4 To act ethically is essential for public servants

5 I believe that public employees must always be aware of the legitimacy of their activities

6 I personally identify with the aim of protecting individual liberties and rights

Compassion (COM)

1 It is difficult for me to contain my feelings when I see people in distress

2 I feel sympathetic to the plight of the underprivileged

3 I empathize with other people who face difficulties

4 I get very upset when I see other people being treated unfairly

5 Considering the welfare of others is very important

Self-Sacrifice (SS)

1 I am prepared to make sacrifices for the good of society

2 I believe in putting civic duty before self

3 I am willing to risk personal loss to help society

4 I would agree to a good plan to make a better life for the poor even if it costs me money

- 1 -

The Development of an International Instrument to

Measure Public Service Motivation A Research Note

ABSTRACT

As research on public service motivation (PSM) has grown and spread from the United States

to Asia Europe and Australia it is essential to elaborate PSM as both a construct and

measure in order to facilitate cross-national research and comparison Previous research has

sharpened the PSM construct and thoroughly discussed a PSM measurement model The

remaining issue is to develop an internationally robust PSM measurement instrument

Drawing upon possible items for measuring PSM which were suggested collectively by PSM

researchers and the data (n = 2868) collected in an international survey among public

servants at local governments in twelve countries the conceptual components of PSM are

empirically estimated through operational definition The survey results are analyzed and

discussed in order to confirm the dimensional structure and validate the items of PSM This

study provides a set of appropriate items for measuring PSM and confirms the dimensional

structure of PSM When all four dimensions of PSM are used in a survey cross-national

equivalence can be assured

- 2 -

The Development of an International Instrument to

Measure Public Service Motivation A Research Note

Introduction

Public service motivation (PSM) is thought of as ―an individualrsquos orientation to delivering

services to people with a purpose to do good for others and society (Perry and Hondgehem

2008a vii) The initial research on PSM began in the United States (Perry and Wise 1990

Rainey 1982) but scholars in many countries have contributed to the accumulation of

knowledge about its antecedents and consequences (Perry and Hondeghem 2008a Perry

Hondeghem and Wise 2010) As the scholarly research on PSM has grown and the

geographic scope of the research has expanded some scholars have questioned whether the

conceptual composition and dimensions of PSM are appropriate for explaining and predicting

public servicendashrelated behavior in different countries and internationally (Coursey and

Pandey 2007a Wise 2000 Wright and Pandey 2008 Giauque et al 2011 Vandenabeele

2008b) and whether different PSM measures are equivalent or if there is a need to develop a

more appropriate measure of PSM that can be used consistently and confidently (Castaing

2006 Leisink and Steijn 2009 Liu Tang and Zhu 2008 Taylor 2007 Vandenabeele 2008a

Wright 2008 Wright and Pandey 2008) Thus in order to facilitate research internationally it

is necessary to elaborate the construct and measurement of PSM (Perry and Hondeghem

2008b)

The first step was to develop a more universal concept that can be used globally and

that is likely to generate cumulative knowledge The construct of PSM has been sharpened by

refining the conceptual components in Perry and Wisersquos (1990) study Kim and Vandenabeele

- 3 -

(2010) proposed that PSM should continue to be conceived as a four-dimensional construct

with self-sacrifice as the foundational concept representing the altruistic or pro-social origins

of PSM The three other concepts attraction to public participation commitment to public

values and compassion which represent instrumental value-based and affective motives

round out their proposed modifications to the construct The revisions to the construct of

PSM will enhance cross-national research and generate cumulative knowledge

The second step was to specify a measurement model of PSM This was necessary

because the measurement model specifies the relationship between constructs and measures

(Diamantopoulos and Siguaw 2006) Reflective measures are treated as outcomes of

constructs and formative measures are causes of constructs (Edwards 2010) A measurement

model can either include reflective or formative indicators exclusively or consist of both

depending on the observed construct (Fornell and Bookstein 1982) The decision to

operationalize formative andor reflective indicators should be based on theoretical

considerations Previous studies discussing the conceptual definition of PSM theoretically

and empirically have proposed that the dimensions of PSM are formative and the indicators

of each dimension are reflective (Kim 2010 Kim and Vandenabeele 2010)

The next step is to develop an internationally robust measurement instrument of PSM

for cross-national research and comparison A survey-based measure for PSM has been useful

for facilitating comparisons across disparate services and national settings and creating

foundations for the accumulation of results (Perry Hondeghem and Wise 2010) Perry (1996)

identified a multidimensional 24-item scale to measure PSM Since then many studies have

used the items of Perryrsquos (1996) measure and improvements have been made to the PSM

scale (Coursey and Pandey 2007a Kim 2009a Taylor 2007 Vandenabeele 2008b) However

research has shown that the original items do not fully represent unique and salient qualities

- 4 -

of the dimensions of the PSM construct (Castaing 2006 Coursey and Pandey 2007a Coursey

et al 2008 DeHart-Davis Marlowe and Pandey 2006 Kim 2009b Leisink and Steijn 2009

Liu Tang and Zhu 2008 Moynihan and Pandey 2007b Taylor 2007 Vandenabeele 2008a

Wright and Pandey 2008) Moreover as the construct and operational definition of PSM are

refined it is necessary to adapt items for the dimensions of the PSM construct

This study is focused on developing a revised measurement instrument for PSM for

international use After developing the possible items for measuring PSM the international

survey was conducted in 2010 and the data (n = 2868) were collected from twelve countries

Dimensions and items were tested cross-nationally as this can be a useful tool to preclude

value-specific items and wordings that are case-specific The next section of this paper will

briefly explain the processes for developing the international measurement instrument The

survey results are analyzed and discussed in order to confirm the dimensional structure and

validate the items of PSM Finally the items of PSM are proposed and discussions and

suggestions are given for future research

Developing the Measurement Instrument of PSM for International Use

Background

Perry and Wise (1990) proposed that PSM is associated with three types of motives affective

norm-based and rational Perry (1996) identified four empirical components of the PSM

construct attraction to public policy making commitment to the public interestcivic duty

compassion and self-sacrifice Since then many studies have used the dimensions and items

of Perryrsquos (1996) measure and improvements have been made in the PSM scale (Kim 2009a

Taylor 2007 Vandenabeele 2008b) However Perryrsquos (1996) measure has limitations (Kim

and Vandenabeele 2010) In particular the items of the dimension of attraction to policy

- 5 -

making are not appropriate for measuring personal attraction to public policy making

(Camilleri 2006 Coursey and Pandey 2007a Coursey et al 2008 Kim 2009b Ritz 2011)

Moreover the dimension of commitment to public interest needs to concentrate more on a

personal disposition to pursue public values (Castaing 2006 Leisink and Steijn 2009 Taylor

2007) and the items of compassion do not represent affective motives (DeHart-Davis

Marlowe and Pandey 2006 Moynihan and Pandey 2007a Wright 2008) Thus the

dimensions of the PSM construct are modified as attraction to public participation

commitment to public values compassion and self-sacrifice Consequently it is necessary to

develop appropriate items for the dimensions (Kim and Vandenabeele 2010)

A related issue concerns the measurement model of PSM There are two

measurement models using multiple dimensions of latent constructs reflective and formative

When variation in a construct leads to variation in its dimensions such dimensions are

termed ―reflective while dimensions are labeled ―formative when they are viewed as

causes of a construct (Edwards and Bagozzi 2000 Diamantopoulos and Siguaw 2006

Diamantopoulos and Winklhofer 2001) It is understandable that each dimension of PSM has

several reflective indicators that are essentially interchangeable A problem is the relationship

between PSM and its dimensions If we regard the dimensions as reflective it is possible to

analyze PSM with only two or three dimensions because omitting dimensions does not

change the essential nature of PSM However if we regard the dimensions as formative it is

not acceptable to omit a dimension as this may change the meaning of PSM Construct

conceptualization is regarded as being the process that determines the formative or reflective

nature (Diamantopoulos and Siguaw 2006) Recent research has demonstrated theoretical and

empirical evidence in support of a second-order formative approach to PSM (Kim 2010)

This means that the dimensions represent different conceptual components of PSM and so

- 6 -

they are not interchangeable Thus PSM is regarded as a combination of specific dimensions

If any one of the dimensions changes PSM would change Conversely if an individualrsquos

PSM increases this would not necessarily be accompanied by an increase in all dimensions

Therefore as the PSM construct is refined for international use (Kim and Vandenabeele

2010) scholars must be attentive to identifying and using appropriate measurement models

Developing Possible Items

The conceptual components of PSM need to be empirically estimated through operational

definition Kim and Vandenabeele (2010) proposed that PSM is based on self-sacrifice and

associated with instrumental value-based and identification motives and that the dimensions

of the PSM construct be refined along the lines of attraction to public participation (APP)

commitment to public values (CPV) compassion (COM) and self-sacrifice (SS) As Perry

(1996) identified a multidimensional scale to measure PSM it is conceived as an overall

latent variable with various latent dimensions and is referred to as a second-order factor

model Research has found that the measurement model specifies the causal relationship

between PSM and its dimensions (Edwards and Bagozzi 2000) The direction of the

relationship is either from the PSM construct to the dimensions (second-order reflective

measurement) or from the dimensions to the construct (second-order formative measurement)

A theoretical consideration of the relationships between PSM and its dimensions led

to the proposal that PSM be defined in terms of multiple dimensions (Kim 2010 Kim and

Vandenabeele 2010) The dimensions represent different aspects of PSM with each

dimension capturing a distinct and potentially unique form of PSM Moreover the

dimensions may have different antecedents and consequences as well as different

characteristics and theoretical backgrounds (Moynihan and Pandey 2007a Pandey and

- 7 -

Stazyk 2008 Perry 1996 1997 2000 Perry and Wise 1990 Perry and Vandenabeele 2008

Taylor 2007 Vandenabeele 2007 2008a 2009) An individualrsquos PSM is shaped by the

individualrsquos APP CPV COM and SS Deleting a dimension means omitting a certain part of

PSM Therefore it is reasonable to suggest that the dimensions of PSM are formative and for

PSM to be regarded as a composite of its dimensions

Each dimension of PSM can be measured using reflective indicators that satisfy the

following criteria (Jarvis MacKenzie and Podsakoff 2003) a relative homogeneity and

interchangeability of indicators pertaining to each dimension a high degree of covariation

among indicators of each dimension and the expectation that the indicators of each

dimension are likely to be affected by the same antecedents and have the same consequences

Reflective indicators are essentially interchangeable and so adding or removing indicators

does not change the essential nature of a dimension Thus we consider PSM to be a second-

order formative construct with several dimensions that each has several reflective indicators

(reflective first-order formative second-order)

Based upon the refinement and clarification of the operational dimensions of PSM

(Kim and Vandenabeele 2010) items were developed for the four dimensions The items

selected were a mix of pre-existing items that loaded highly on PSM dimensions in multiple

studies (Giauque et al 2011 Kim 2009a Perry 1996 Vandenabeele 2008b) Where

appropriate items were not available new items were drafted from other previous studies

creating an initial total of 35 proposed items For the dimension of APP which focuses on

individual disposition to work in the public sector to participate in the public policy process

and in activities for community and social development seven possible items were suggested

The CPV dimension focuses on personal disposition to pursue public interest and public

values Two sub-dimensions of CPV seemed plausible one focusing on public interest and

- 8 -

another on public values such as social equity equal opportunity and democracy Thus five

possible items for the public interest sub-dimension (CPV1) and six items for the public

values sub-dimension (CPV2) were developed For the dimension of COM which focuses on

affective bonding with the identified objects such as other members of a social category or of

a political system ten possible items were suggested For the SS dimension Perryrsquos original

items (1996) were the primary source of items The full list of items is provided in Appendix

1

Review Processes

A group of researchers who shared our interest in better measurement of PSM collaborated

using e-mail in November 20091 The objective was to develop a measurement instrument

for PSM with a robust comparative character in other words to create a valid instrument (on

a national basis) that can be used internationally The first round for reviewing the possible

items lasted from the 16th

of November until the 7th

of December 2009 and all the

participants were asked to review these proposed items For this review two criteria were

employed (1) Do these items refer to the dimensions they claim to refer to and (2) Are

these items meaningful across various national contexts and in particular to the situation of

civil servants in that country Participants were also asked to add new items that might better

reflect a particular dimension or to rephrase items that were already on the list

A second round of item refinement lasted from the 9th

of December until the 22nd

of

December 2009 During this time 11 newly suggested items from the first round were

considered (see Appendix 2) A summary of all comments and suggestions emerging from the

1 Emmanuel Camilleri David Giauque Gerhard Hammerschmid Isabel Egger-Peitler Renate Meyer and

Cristian Pliscoff also participated in the review processes

- 9 -

first process was distributed to the participants to stimulate comments in the second round

(see Appendix 3) Based on the comments received in the second round discussion the

possible items for an international survey were selected However some items required

further discussion and so a third round was conducted from the 15th

of January until the 23rd

of January 2010 One issue that remained unresolved in the discussions relates to the

dimension of CPV Some argued that it is reasonable to divide CPV into two sub-dimensions

because committing to public interests is different from pursuing public values and there are

many important public values in democratic countries while others contended that CPV can

cover all important public values including public interest and there is no real benefit to add

one more dimension These differences were left to be resolved through empirical research

This did not affect the final items however and 33 items were confirmed The resulting

survey questionnaire was distributed to all the collaborators on February 9 2010 in order to

conduct a survey in each country

Measures

PSM was measured using the newly developed 33-item index 7 items for the dimension of

APP 13 items for CPV (4 for CPV1 and 9 for CPV2) 6 items for COM and 7 for SS In the

survey questionnaire a five-point Likert type scale (1 = strongly disagree 5 = strongly agree)

was used This scale is among the most widely used response formats in many empirical

investigations within the social and behavioral science in which there are no ―correct or

―incorrect responses (Spector 1992) The items of PSM were randomly distributed in the

survey format Gender age education length of service and employment status were also

measured in this survey

It was necessary to translate the measurement items into the official (native) language

- 10 -

of each country As this raised the possibility of inaccurate translations the back-translation

of the items into native languages was validated by asking two researchers to independently

translate the English-language items The researchers were instructed to consult with each

other over differences between translated items and to test the survey items with some native

respondents

[Table 1 about here]

Samples

The international survey was fielded from March to September 2010 The data upon which

this study are based were collected from civil servants in local governments in Australia

Belgium China Denmark France Italy Korea Lithuania the Netherlands Switzerland the

United Kingdom and the USA The principal criterion was that the samples collected in the

various countries should be comparable to the largest possible extent The criterion ruled out

national and state government as the competencies of national governments and also of state

governments if they exist constitutionally are different among most of the countries included

in the research Therefore we opted for local government as our focus The structure and

functions of local governments can differ widely even within countries and so the survey in

each country was focused on town hall bureaucrats in city county or township government

The respondents were permanent employees from both managerial and non-managerial levels

and from both administrative departments (ie no direct contact with the public) and service

departments (ie likely to have direct contact with the public) Despite being employed by

local governments in many or all of the countries police officers firefighters school teachers

public transport workers artists and musicians as well as nurses and doctors were excluded

- 11 -

from the sample

We aimed to receive 250 respondents for each country as this would enable us to

carry out the analyses we wanted to perform on the data Although there is no ideal sample

size one study recommends a sample of 200 to provide a sound basis for estimation (Hair

Black Babin and Anderson 2010) The method of data collection was left open Obviously a

web-survey is the easiest way to collect the data because it is economical to collect and to

code however compared to paper-pencil surveys this method has some drawbacks First not

everybody in the sample may have access to the internet which may cause a bias in the

responses Second internet surveys are generally known to produce lower response rates

Web surveys were recommended if possible but paper-pencil surveys were deemed

acceptable in cases where it was not possible to conduct a web survey Researchers were

encouraged to contact just one municipality except in cases where it was not possible to

gather enough data In such instances responses from multiple municipalities were accepted

By the end of September 2010 a total of 2868 responses were obtained2

2 In Australia seven local councils each with between from 70 and 232 employees were surveyed Paper-pencil

surveys were used in two councils while the rest relied on a web survey The local governments limited

distribution of paper-pencil surveys to only one or two sections of the council eg administration The response

rate was approximately 30 However it should be noted that not everyone working for the council has direct

access to a computer In Belgium three municipalities were surveyed on a census base by means of a web

survey as all employees had access to a computer This rendered a response of 120 (226) 79 (75) and 16

(53) responses for the respective municipalities resulting in an overall response rate of 331 with 662

invitations being sent out In China two departments social work and administration were surveyed in one

municipality using paper-pencil surveys In the social work department 65 out of 110 surveys were returned

and 165 out of 211 responded from the administration department giving an overall response rate of 767 In

Denmark a total of 1282 respondents in a municipality were sent an initial e-mail invitation to a web survey

and up to two e-mail reminders and the response rate was 473 In France all the employees in two

municipalities were surveyed In one municipality where both web and paper-pencil surveys were utilized 128

out of 685 questionnaires were returned while in another city using only a paper-pencil survey 138 out of 652

were received The average response rate was 199 In Italy in principle the link to the questionnaire was

made available to all employees (about 4000) in one city but for some of them access may have been difficult

yielding a response rate of 45 In Korea 500 employees in a city received a printed questionnaire and 253

responses were returned yielding a response rate of 506 In Lithuania 236 responses from 54 municipalities

were gathered by web survey In the Netherlands three municipalities were surveyed on a census base by means

of a web survey In total 319 responses of 910 invitations were returned (35) mdash 90 (45) 115 (38) and 114

(28) for the respective municipalities In Switzerland 516 out of 1326 local employees with e-mail access in

a municipality completed the survey giving a response rate of 389 In the United Kingdom the survey was

posted online and 1360 employees both in administrative jobs and front-line services of one municipality

- 12 -

Municipalities were selected because of their convenient accessibility and proximity to the

researcher and a convenience sample was used in most countries Seven countries surveyed

one municipality but seven local councils were surveyed in Australia Web surveys were used

in four countries while three countries used both web and paper-pencil survey The number

of respondents varied among the countries depending on the numbers of questionnaires

distributed the survey method and the response rate In most countries we used

approximately 250 responses but only 162 from Italy3 The majority of the respondents were

women (566) while in terms of service length those in the largest group (393) had

worked for fewer than 10 years Table 2 shows the distribution of respondentsrsquo sex length of

service employment status and country

[Table 2 about here]

Analyses

The statistical analysis applied confirmatory factor analysis (CFA) CFA was used to assess

the fit of the data to the hypothesized measurement model because it is preferred where

measurement models have a well-developed underlying theory for hypothesized patterns of

loadings The data were analyzed using Amos 180 (Arbuckle 2009) with the maximum

likelihood (MLE) estimation method and LISREL 872 (Joumlreskog and Soumlrbom 2005) with the

diagonally weighted least squares (D-WLS) estimation method

The individual items are designed to measure a theoretically continuous construct

but the observed responses are a discrete realization of five categories The common method

received an email to complete the survey The response rate was 19 In the USA the total number of

employees surveyed was 626 in one city using both electronic and paper-pencil surveys resulting in a response

rate of 588 3 In countries where significantly more than 250 surveys were collected we randomly selected only 250 cases

for analysis

- 13 -

of estimation within CFA is MLE a technique which assumes that the observed variables are

continuous and normally distributed These assumptions are not met when the observed data

are discrete as occurs when using ordinal scales as in this study A simulation study (Flora

and Curran 2004) found that D-WLS performs well for ordinal measures across a variety of

conditions especially for smaller samples Coursey and Pandey (2007a 2007b) argued the

importance of using D-WLS and not standard MLE in structural equations for ordinal data4

In most cases however the differences between MLE and D-WLS are not substantive to the

conclusions5 No comparison between standard MLE and D-WLS is provided in this paper

We give results from both MLE and D-WLS and this will aid in confirming the stability and

robustness of the results across varying mathematical specifications

For gauging reliability Cronbachrsquos alpha and composite reliability were tested

Composite reliability should be equal to or greater than 07 (Hair et al 2010) Convergent

validity was assessed from the measurement model by determining the statistical significance

of each indicatorrsquos estimated pattern coefficient on its specified underlying construct

Discriminant validity was assessed by determining whether the confidence interval around

the correlation estimate between two factors included 100 (Anderson and Gerbing 1988)

For model fit assessment goodness-of-fit index (GFI) comparative fit index (CFI)

and the root mean square error of approximation (RMSEA) were used The model achieves

an acceptable fit to the data when GFI and CFI equals or exceeds 090 and RMSEA values

4 Applying standard MLE to ordinal data such as Likert-type items may produce significant estimation

problems such as inflation of chi-square fit statistics and biased underestimation of parameters and standard

errors (Flora and Curran 2004) Coursey and Pandey (2007a 2007b) used a modified MLE using the asymptotic

covariance matrix and D-WLS for adjusting the ordinal nature of the data They reported that although the

modified MLE findings are slightly different on a few model parameters the differences are not substantive to

the conclusions However model fit estimators were clearly better for the D-WLS 5 Given normally distributed categorical variables MLE can be used with confidence when a variable has four

or more categories If there is a large sample a five-point Likert scale and the skewness and kurtosis of variables

is within acceptable limits it suggests distribution symmetry ―In most cases where the hypothesized model is

well specified and the scaling based on more than three categories it seems unlikely that there will be much

difference between the findings (Byrne 2010 160)

- 14 -

fall below 008 (Hair et al 2010) In general the larger the value of GFI and CFI and the

smaller the value of RMSEA the better the fit of the model (Bollen 1989) We also present

the χ2 test statistic Lower values of χ

2 indicate a better fit and should be nonsignificant but

the χ2 value increases as sample size increases and thus for large sample sizes this statistic

may lead to rejection of a model with good fit (Hair et al 2010)

Results

Using SPSS 180 descriptive statistics were computed for individual items as shown in table

1 One item (APP3) in the APP dimension ―I like to discuss topics regarding public programs

and policies with others was mistakenly omitted in the survey conducted in the United

States and so it was deleted in the statistical analysis In the assessment of normality no

items showed a skew or kurtosis value greater than the cutoffs of |3| or |8| recommended by

Kline (2005) As data were distributed normally MLE as well as D-WLS estimation was used

Based on inspection of item-total correlations one item (COM4) was dropped from further

analysis because it was weakly correlated with the overall index (011) Missing data is an

issue in CFA as it can cause estimation errors All variables in the survey have missing data

rates of approximately 016 to 143 Mean substitution was used to calculate replacement

values for the missing values in each countryrsquos dataset (Hair et al 2010) The deletion of the

two items left 31 items for the next stage of analysis

The five-correlated factor model was tested using CFA The CFA model in the initial

analysis hypothesized a priori that (1) responses to the 30-item PSM measurement

instrument could be explained by five factors such as APP CPV1 CPV2 COM SS (2) each

item would have a nonzero loading on the PSM factor that it was designed to measure and

- 15 -

zero loadings on all other factors (3) the five factors would be correlated and (4)

measurement error terms would be uncorrelated The resulting CFA with MLE suggested that

the initial five-correlated factor model was not a good fit to the data χsup2 (df = 424) = 43699 p

lt 001 GFI = 0899 CFI = 835 RMSEA = 0057 [055 059] At this stage APP was highly

correlated with CPV1 (r = 996 in MLE and r = 988 in D-WLS) suggesting the two

dimensions were not unique and lacked discriminant validity

[Table 3 about here]

Given the disconfirmation of the initial model an effort was made to estimate an

alternative model The two dimensions APP and CPV1 were combined to form a single

dimension Attraction to Public Service (APS) The items of CPV1 may be regarded as

action-oriented by the respondents because dedication to public service community and

common good is underlined in these items Consequently it is reasonable to combine APP

with CPV1 to form APS a dimension that focuses more on disposition to serve the public to

work for the common good and to participate in public policy processes After the overall

number of dimensions was reduced to four using standardized factor loadings and

modification indices as criteria the items with the lowest factor loading in each dimension

were deleted and the items whose modification index indicated association with multiple

dimensions were eliminated A CFA model with the remaining items was iteratively tested

until it achieved a good fit to the data The outcome of this refinement process was a 16-item

index of four factors Overall model fit by every measure is quite strong for the D-WLS

estimation technique The estimation results for the four-dimension model are presented in

table 4

- 16 -

[Table 4 about here]

Cronbachrsquos coefficient alpha for the 16-item PSM index was 860 and the coefficient

alphas for the four dimensions ranged from 645 to 784 The composite reliability of the set

of reflective indicators for each dimension of PSM ranged from 713 to 826 and indicated

adequate internal consistency The resulting factor structure showed a four-factor structure

with all items loading significantly onto their a priori dimension (p lt 001) and the

standardized factor loadings ranging from 0490 to 0766 in MLE and from 515 to 807 in

D-WLS The results provide support for convergent validity The correlation estimates

between the two factors ranged from 0462 to 0848 in MLE and from 505 to 859 in D-

WLS The confidence intervals (plusmn2 standard errors) around the correlation estimates between

the two factors did not include 100 The result provides support for discriminant validity To

test whether the four-dimension model is superior to other designs such as the two or three-

dimension models a series of CFA were conducted and the results show that the four-

dimension model is a significantly better fit than any of the other designs

To test whether the factorial structure of the PSM instrument is equivalent across the

samples from the twelve countries four nested models were tested as part of multi-group

analysis with MLE (Byrne 2010) First the unconstrained model was the baseline model

which relaxed all equality constraints This model hypothesized the same four-factor structure

of the instrument across twelve countries The fit indices for this baseline model were χsup2 (df =

1176) = 24977 p lt 001 GFI = 0901 CFI = 900 RMSEA = 0020 [019 021] These

indices indicate that the four-factor model of the PSM instrument exhibits acceptable fit

across the samples from the twelve countries Second the measurement weights model tested

- 17 -

the invariance of factor loadings across countries by placing equality constraints on these

parameters The fit indices for this model were χsup2 (df = 1308) = 28088 p lt 001 GFI = 0890

CFI = 887 RMSEA = 0020 [019 021] Since the measurement weights model was nested

within the unconstrained model the chi-square difference text ∆χsup2 (132) = 3111 p lt 001

indicated that some equality constraints of factor loadings did not hold across the twelve

countries Further test results suggested that the equality constraints of structural covariances

and measurement residuals were not upheld across the twelve countries It needs to be

stressed that despite the differences in factor loadings and factor covariances the results

support the main assumption that the same four-factor structure holds across different

countries We may assume that people in different cultures understand PSM in the same

fashion

To find differences in the factor structure of PSM among countries we conducted a

series of CFA with each countryrsquos dataset and followed the same procedure as when

analyzing the sample as a whole As table 5 shows although there are great similarities

between the appropriate items in the countries there are also some differences Even though

the items vary according to national differences the four-factor structure had a good fit to the

data Thus the results confirm that the four-factor structure of PSM can be generalized

[Table 5 about here]

The APS dimension covers both an individualrsquos attraction to public participation and

commitment to public interest but all its items refer to some notion of public interest such as

improving public service aiding community tackling social problems meaningful public

service and contributing to the common good The APS dimension captures an individualrsquos

- 18 -

disposition to serve the public interest through public participation When developing the

items some assumed that the items of committing to public interests can be separated from

those of pursuing public values in the CPV dimension The empirical test showed that

respondents tended to understand committing to public interests as behavior-oriented and so

the items for public interest are not differentiated from participation-oriented items in their

perceptions but are different from those for public values The respondents seem to

understand that pursuing public values is more related to personal beliefs or creeds Thus the

CPV dimension is solely focused on public values and the items of pursing public interests

are combined with those of the APP dimension to make the APS dimension

Public values and their relative importance are neither impervious to change over

time nor the same among all countries (Charles de Jong and Ryan 2011 Davis and West

2009) Moreover some values may be in conflict as Martinsen and Beck Joslashrgensen (2010)

discussed in regard to the tensions between accountability and efficiency It is therefore not

possible to include all public values in one dimension (Beck Joslashrgensen and Bozeman 2007)

Accordingly the items of the CPV dimension are developed for estimating individual

commitment to seek public values which there are general consensus about and which are

generally formulated

When all four dimensions of PSM are included in a survey cross-national

equivalence of the survey can be assured The items represent reflections or manifestations of

each dimension and hence are relatively homogeneous and interchangeable Adding or

deleting items does not change the essential nature of the dimension Thus any combination

of items in a dimension can be regarded as equivalent When using structural equation

modeling (SEM) each dimension needs to have at least three reflective indicators in order to

achieve identification or it needs to have at least two reflective indicators and be allowed to

- 19 -

covary with another dimension that has at least two reflective measures With one indicator

the dimension is identified if the loading and the variance of the uniqueness are both fixed

(Bollen 1989 Edwards 2010) Thus the number of items in each dimension is not restricted

when measuring PSM However we recommend using multiple items for each dimension

because more research is needed to investigate the appropriateness of this PSM instrument

with varying samples

The test results showed that model fit estimators were clearly better for the D-WLS

It means that in this study MLE was more rigorous than D-WLS in assessing the fit of the

model The empirical evidence indicated that MLE may be an acceptable technique for

ordinal data when there is a large sample a five-point Likert scale and the skewness and

kurtosis of variables is within acceptable limits as suggested by Byrne (2010)

Discussion

The analysis reaffirmed the four-dimensional structure of PSM but the results were not

entirely consistent with the original CFA we tested We were ultimately able to converge on a

good-fit operational model for PSM that was also consistent with underlying theory The

analysis is not able to resolve the debate about measurement models however for reasons

discussed below

Choosing the correct measurement model (ie reflective vs formative) is important

as measurement model misspecification results not only in misrepresentation of the epistemic

relationships between constructs and their measures but also in biased parameter estimates of

structural relationships (Jarvis et al 2003 Law and Wong 1999 MacKenzie 2003

MacKenzie Podsakoff and Jarvis 2005) A formative measurement model in isolation is

- 20 -

under-identified and cannot be estimated in the case of covariance-based SEM (CB-SEM)

(Bollen 1989) A construct with formative indicators must emit paths to at least two unrelated

reflective indicators two unrelated latent constructs or some combination of the two in order

to solve the problem of under-identification (Jarvis MacKenzie and Podsakoff 2003

MacCallum and Browne 1993 MacKenzie Podsakoff and Jarvis 2005) Thus ―when

dependent constructs change the empirical nature of the formatively measured construct

changes (Wilcox Howell and Breivik 2008 1223) and ―when specifying formative

measurement models what might seem like an innocuous choice of outcomes to achieve

identification can have fundamental ramifications for the meaning of the formative construct

and the inferences based on it (Edwards 2010 7)

Earlier research found that the association between a construct and its measures

should remain stable regardless of the larger causal model in which the constructs and

measures are embedded (Edwards and Bagozzi 2000) The association between a construct

and its measures is generally stable for reflective measures but associations of formative

measures with their construct are unstable The susceptibility of formative measurement to

both interpretational confounding and external inconsistency recommends that formative

measurement is not an equally attractive alternative to reflective measurement and that

whenever possible researchers should opt for reflective measurement (Howell Breivik and

Wilcox 2007a 2007b Kim Shin and Grover 2010) In the case of second-order models

researchers are limited to only a second-order reflective model in CB-SEM (Chin 2010)

Regardless of the measurement debate and the tradeoffs between reflective or

formative specification this study makes a theoretical argument that PSM is a second-order

formative construct This second-order formative factor model (reflective first-order

formative second-order) is also suggested as an alternative model that captures the

- 21 -

multidimensionality of complex concepts and avoids the shortcomings of first-order

formative measurement (Edwards 2010) However this model itself cannot achieve

identification One possible strategy is to develop direct reflective measures of PSM based on

the view that PSM is a general concept that is not only influenced by its specific dimensions

but also can be assessed using general measures (Edwards 2010 Javis MacKenzie and

Podsakoff 2003)6 However these general measures have not yet been developed for the

PSM construct

Another way around the problem of underidentification of formative models is to

apply the component-based approach known as partial least squares path modeling (PLS-PM)

which can model formative indicators (Fornell and Bookstein 1982 Goumltz Liehr-Gobbers

and Krafft 2010) The CB-SEM and the PLS-PM methods have distinctive statistical

characteristics and selecting an approach to SEM depends on the particular research situation

CB-SEM is the method of choice for theory testing while PLS-PM is appropriate for

prediction-oriented applications (Ringle Goumltz Wetzels and Wilson 2009) One advantage of

the PLS method is that it demands significantly fewer requirements compared to that of

covariance structure analyses but nevertheless delivers consistent estimation results Another

asset of the PLS approach is its ability to deal with formative as well as reflective indicators

even within one structural equation model This indicates that the PLS approach can be

appropriate for explorative analysis of structural equation models (Chin 2010 Goumltz Liehr-

Gobbers and Krafft 2010) A Monte Carlo simulation study to compare CB-SEM and PLS-

PM showed that PLS-PM provides a viable approximation of model parameters when the

prerequisites for CB-SEM are not met (Ringle Goumltz Wetzels and Wilson 2009) A formative

6 For example job satisfaction can be assessed with general reflective measures such as ―in general I am

satisfied with my job and the formative dimensions can be assessed with measures that describe satisfaction

with each job facet (Edwards 2010 15)

- 22 -

composite in the form of indexes needs to be tested to know the relative weights (γ-

parameters) of the dimensions of PSM In PLS-PM a second-order formative factor is

directly measured by observed variables for all the first order factors While this approach

repeats the number of manifest variables used the model can be identified without adding

outcome constructs that would be considered inappropriate or irrelevant from a conceptual

standpoint (Chin 2010 Edwards 2010) The test result of PLS-PM using Smart PLS 20

(Ringle Wende and Will 2005) produced the function PSM = 31APS + 36CPV + 28COM

+ 31SS However one consideration is whether and how construct error is modeled for the

formative construct (Diamantopoulos 2006) When using PLS-PM the researcher is making

the assumption that there is no error at the construct level as all constructs in PLS-PM are

modeled without error When using PLS-PM the absence of an error term will tend to

increase the weights therefore inflating their importance (Bollen and Lennox 1991) The

primary statistic for assessing a formative indicator is its weight the partialized effect of the

indicator on its intended construct controlling for the effects of all other indicators of that

construct (Cenfetelli and Bassellier 2009) A formative measure is essentially a multiple

regression with the construct representing the dependent variable and the indicators as the

predictors (Diamantopoulos and Winklhofer 2001) PSM may be regarded as a weighted

linear composite of its dimensions The weights used to create the composites are different

across CE-SEM and PLS-PM The causal dimensional weights generated in CB-SEM are

estimated in a model in which a construct-level error term is specified (eg PSM = γ1APM +

γ2CPI + γ3COM + γ4SS + ζ) whereas the weights in PLS-PM are not (eg PSM = γ1APM +

γ2CPI + γ3COM + γ4SS) This leads to two different conceptualizations of the PSM construct

one in which the causal indicators do not capture the construct in its entirety and another in

which they do (Hardin et al 2010) These differences need to be discussed in future research

- 23 -

in this area

One practical suggestion to measure PSM is to apply unit weights to its dimensions if

we assume the effects of the four dimensions are equivalent A literature review indicates that

unit weights have substantial predictive validity when compared with regression weights The

literature shows that the empirical correlations between scores generated from differential and

unit-weighting approaches are quite high In creating composite scores unit weights can be

an appropriate approach for weighting under many circumstances (Bobko Roth and Buster

2007) Although the uses of unit-weight composites will likely result in the explanation of

less variance than empirical estimations of dimensional weights they will remain consistently

defined across studies and hence suitable for theory testing Alternatively more thorough

discussions on the dimensions of PSM may suggest fixed-weight composites (Hardin et al

2010) A previous study (Kim 2010 22) also suggested ―Overall indexing PSM by

measuring all the dimensions of PSM and simply summing the averages of each dimension

can make it easier to measure PSM in practice as PSM = APS + CPV + COM + SS

An advantage of regarding PSM as a second-order formative construct is that the

dimensions can be analyzed both individually and collectively This is important as the

dimensions are expected to have different antecedents and consequences For instance Perry

(1997) found some significant differences in the influence of independent variables on the

four dimensions of PSM Taylor (2007) confirmed that certain dimensions are more important

than others in influencing work outcomes Moynihan and Pandey (2007a) reported that red

tape is negatively related to one dimension but reform orientation and hierarchical authority

is positively related to another dimension A recent study found that the SS dimension has a

positive relationship with occupational intention but the other dimensions do not (Liu et al

2011) Thus it is valuable to analyze the causes and effects of individual dimensions of PSM

- 24 -

as well as of the PSM construct

Conclusion

The purpose of this study is to develop a measurement instrument for PSM for cross-national

research and comparison On the basis of the sharpened construct of PSM (Kim and

Vandenabeele 2010) using data from an international survey conducted in 12 countries we

found that the four-factor structure of PSM is generalizable across country cases This study

provides the measurement index of PSM with the four dimensions of APS CPV COM and

SS The APS dimension measures the instrumental motives of PSM the CPV dimension

determines the value-based motives of PSM while the COM and SS dimensions gauge the

identification motives and self-sacrifice respectively When all four dimensions of PSM are

used in a survey cross-national equivalence of survey items can be assured The items for

measuring PSM in future research are suggested in Appendix 4

This study has several limitations First looking for samples which were comparable

to the largest possible extent we opted for surveys in local government The results found

and discussed in this study are based on PSM among local government employees The

results might be different when surveying central government employees or other specific

groups in the public sector Thus the potential impact of the sample bias needs to be

considered in understanding this study It also means that the survey items for measuring

PSM should be examined with different samples in various areas and at different government

levels

Second the test results showed that there is a high degree of correlation among the

dimension such as APS CPV and COM Future research should not only investigate whether

- 25 -

the items reflect the core content of the dimensions but also continue to develop more

appropriate items for better discriminant validity of each dimension Third the content of

APP dimension was assumed to be distinct from the CPV dimension that focuses on personal

disposition to pursue public interest and public values However the public interest sub-

dimension was separated from the public values sub-dimension in the CPV dimension and

combined with the APP dimension to form the APS dimension Further research should

analyze the CPV dimension and investigate the appropriateness of the APS dimension

In order to facilitate research internationally the PSM construct needs to be

improved conceptually and operationally Kim and Vandenabeele (2010) have dealt with the

conceptual composition and operational dimension of PSM as well as the measurement

model of PSM The next step is to develop an internationally robust measurement instrument

of PSM This study proposes that scholars use all four of the dimensions in measuring the

conceptual components of PSM such as self-sacrifice instrumental value-based and affective

motives because they collectively represent PSM It provides a set of appropriate items for

measuring PSM and confirms the dimensional structure of PSM The survey results suggest

that people in different cultures understand PSM in the same fashion We know that this study

is simply a step along the way to developing a more robust measure but we hope it

contributes to enhanced cross-national research and comparison along with the generation of

cumulative knowledge internationally

- 26 -

REFERENCES

Anderson James C and David D Gerbing 1988 Structural Equation Modeling in Practice

A Review and Recommended Two-Step Approach Psychological Bulletin 103

411-423

Arbuckle James L 2009 Amos 180 Userrsquos Guide Crawfordville FL Amos Development

Co

Beck Joslashrgensen Torben and Barry Bozeman 2007 Public Values An Inventory

Administration amp Society 39(3) 354-381

Beutel Ann M and Margaret Mooney Marini 1995 Gender and values American

Sociological Review 60 436 - 448

Bobko Philip Philip L Roth and Maury A Buster 2007 The Usefulness of Unit Weights in

Creating Composite Scores A Literature Review Application to Content Validity and

Meta-Analysis Organizational Research Methods 10(4) 689-709

Bollen Kenneth A 1989 Structural Equations with Latent Variables New York Wiley

Bollen K A and R Lennox 1991 Conventional Wisdom on Measurement A Structural

Equation Perspective Psychological Bulletin 110(2) 305-314

Byrne Barbara M 2010 Structural Equation Modeling with Amos Basic Concepts

Applications and Programming 2nd

ed New York Routledge

Camilleri Emanuel 2006 Towards Developing an Organizational Commitment ― Public

Service Motivation Model for Maltese Public Service Employees Public Policy

and Administration 21 63-83

Castaing Seacutebastien 2006 The Effects of Psychological Contract Fulfillment and Public

Service Motivation on Organizational Commitment in the French Civil Service

Public Policy and Administration 21 84-98

Cenfetelli Ronald T and Geneviegraveve Bassellier 2009 Interpretation of Formative

Measurement in Information Systems Research MIS Quarterly 33(4) 689-707

Charles Michael B W Martin de Jong and Neal Ryan 2011 Public Values in Western

- 27 -

Europe A Temporal Perspective American Review of Public Administration 41

75-91

Chin Wynne W 2010 How to Write Up and Report PLS Analysis In Handbook of Partical

Least Squares edited by V Esposito Vinzi et al 655-690 Berlin Springer-Verlag

Coursey David H and Sanjay K Pandey 2007a Public Service Motivation Measurement

Testing an Abridged Version of Perryrsquos Proposed Scale Administration amp Society

39 547-568

Coursey David H and Sanjay K Pandey 2007b Content Domain Measurement and

Validity of the Red Tape Concept American Review of Public Administration 37

342-361

Coursey David H James L Perry Jeffrey L Brudney and Laura Littlepage 2008

Psychometric Verification of Perryrsquos Public Service Motivation Instrument Results

for Volunteer Exemplars Review of Public Personnel Administration 28 79-90

Davis Paul and Karen West 2009 What Do Public Values Mean for Public Action Putting

Public Values in Their Plural Place American Review of Public Administration 39

602-618

DeHart-Davis Leisha Justin Marlowe and Sanjay K Pandey 2006 Gender Dimensions of

Public Service Motivation Public Administration Review 66 873-887

Diamantopoulos Adamantios 2006 The error term in formative measurement models

Interpretation and modeling implications Journal of Modeling in Management 1(1)

7-17

Diamantopoulos Adamantios and Judy A Siguaw 2006 Formative versus Reflective

Indicators in Organizational Measure Development A Comparison and Empirical

Illustration British Journal of Management 17 263-282

Diamantopoulos Adamantios and Heidi M Winklhofer 2001 Index Construction with

Formative Indicators An Alternative to Scale Development Journal of Marketing

Research 38 269-277

Edwards Jeffrey R 2010 The Fallacy of Formative Measurement Organizational Research

- 28 -

Methods Online First published on August 19 2010 as

doi1011771094428110378369

Edwards Jeffrey R and Richard P Bagozzi 2000 On the Nature and Direction of

Relationships between Constructs and Measures Psychological Methods 5 155-

174

Flora David B and Patrick J Curran 2004 An Empirical Evaluation of Alternative

Methods of Estimation for Confirmatory Factor Analysis with Ordinal Data

Psychological Methods 9 466-491

Fornell Claes and Fred L Bookstein 1982 Two structural equations models LISREL and

PLS applied to customer exit-voice theory Journal of Marketing Research 19 440-

452

Giauque David Adrian Ritz Freacutedeacuteric Varone Simon Anderfuhren-Biget and Christian

Waldner 2011 Putting Public Service Motivation into Context A Balance between

Universalism and Particularism Forthcoming in International Review of

Administrative Sciences 77(2)

Goumltz Oliver Kerstin Liehr-Gobbers and Manfred Krafft 2010 Evaluation of Structural

Equation Models Using the Partial Least Squares (PLS) Approach In V Esposito

Vinzi et al (eds) Handbook of Partial Least Squares 691-711 DOI 101007978-3-

540-32827-8_30 Springer-Verlag Berlin Heidelberg

Guinot Christiane Julie Latreille and Michel Tenenhaus 2001 PLS Path Modelling and

Multiple Table Analysis Application to the Cosmetic Habits of Women in Ile-de-

France Chemometrics and Intelligent Laboratory Systems 58 247-259

Hair Joseph F Jr William C Black Barry J Babin and Rolph E Anderson 2010

Multivariate Data Analysis 7th

ed Upper Saddle River NJ Pearson Education

Hardin Andrew M Jerry Cha-Jan Chang Mark A Fuller and Gholamreza Torkzadeh 2010

Formative Measurement and Academic Research In Search of Measurement

Theory Educational and Psychological Measurement Online First published on

June 11 2010 as doi101177001316440370208

- 29 -

Howell Roy D Einar Breivik and James B Wilcox 2007a Reconsidering Formative

Measurement Psychological Methods 12 205-218

Howell Roy D Einar Breivik and James B Wilcox 2007b Is Formative Measurement

Really Measurement Reply to Bollen (2007) and Bagozzi (2007) Psychological

Methods 12 238-245

Jarvis Cheryl Burke Scott B MacKenzie and Philip M Podsakoff 2003 A Critical Review

of Construct Indicators and Measurement Model Misspecification in Marketing and

Consumer Research Journal of Consumer Research 30 199-218

Joumlreskog Karl G and Dag Soumlrbom 2005 LISREL 872 (Computer Software) Lincolnwood

IL Scientific Software International

Kim Gimun Bongsik Shin and Varun Grover 2010 Investigating Two Contradictory Views

of Formative Measurement in Information Systems Research MIS Quarterly 34

345-365

Kim Sangmook 2009a Revising Perryrsquos Measurement Scale of Public Service Motivation

American Review of Public Administration 39 149-163

Kim Sangmook 2009b Testing the Structure of Public Service Motivation in Korea A

Research Note Journal of Public Administration Research and Theory 19 839-

851

Kim Sangmook 2010 Testing a Revised Measurement of Public Service Motivation

Reflective versus Formative Specification Journal of Public Administration

Research and Theory Doi101093jopartmuq048

Kim Sangmook and Wouter Vandenabeele 2010 A strategy for building public service

motivation research internationally Public Administration Review 70 701-709

Kline Rex B 2005 Principles and Practice of Structural Equation Modeling 2nd

ed New

York Guilford Press

Law Kenneth and Chi-Sum Wong 1999 Multidimensional Constructs in Structural

Equation Analysis An Illustration using the Job Perception and Job Satisfaction

Constructs Journal of Management 25 143-160

- 30 -

Leisink Peter and Bram Steijn 2009 Public Service Motivation and Job Performance of

Public Sector Employees in the Netherlands International Review of

Administrative Sciences 75 35-52

Liu Bangcheng Chun Hui Jin Hu Wensheng Yang and Xinli Yu 2011 How well can

Public Service Motivation Connect with Occupational Intention International

Review of Administrative Sciences 77 191-211

Liu Bangcheng Ningyu Tang and Xiaomei Zhu 2008 Public Service Motivation and Job

Satisfaction in China An Investigation of Generalisability and Instrumentality

International Journal of Manpower 29 684-699

MacCallum Robert C and Michael W Browne 1993 The Use of Causal Indicators in

Covariance Structure Models Some Practical Issues Psychological Bulletin 114

533-541

MacKenzie Scott B 2003 The dangers of poor construct conceptualization Journal of the

Academy of Marketing Science 31(3) 323-326

MacKenzie Scott B Philip M Podsakoff and Cheryl Burke Jarvis 2005 The Problem of

Measurement Model Misspecification in Behavioral and Organizational Research

and Some Recommended Solutions Journal of Applied Psychology 90 710-730

Martinsen Dorte Sindbjerg and Torben Beck Joslashrgensen 2010 Accountability as a

Differentiated Value in Supranational Governance American Review of Public

Administration 40 742-760

Moynihan Donald P and Sanjay K Pandey 2007a The Role of Organizations in Fostering

Public Service Motivation Public Administration Review 67 40-53

Moynihan Donald P and Sanjay K Pandey 2007b Finding Workable Levers over Work

Motivation Comparing Job Satisfaction Job Involvement and Organizational

Commitment Administration amp Society 39 803-832

Pandey Sanjay K and Edmund C Stazyk 2008 Antecedents and Correlates of Public

Service Motivation In Motivation in Public Management edited by James L Perry

and Annie Hondeghem 101-117 Oxford Oxford University Press

- 31 -

Perry James L 1996 Measuring Public Service Motivation An Assessment of Construct

Reliability and Validity Journal of Public Administration Research and Theory 6

5-22

Perry James L 1997 Antecedents of Public Service Motivation Journal of Public

Administration Research and Theory 7 181-197

Perry James L 2000 Bringing Society in Toward a Theory of Public Service Motivation

Journal of Public Administration Research and Theory 10 471-488

Perry James L and Annie Hondeghem 2008a Motivation in Public Management Oxford

Oxford University Press

Perry James L and Annie Hondeghem 2008b Directions for Future Theory and Research

In Motivation in Public Management edited by James L Perry and Annie

Hondeghem 294-313 Oxford Oxford University Press

Perry James L Annie Hondeghem and Lois Recascino Wise 2010 Revisiting the

Motivational Bases of Public Service Twenty Years of Research and an Agenda for

the Future Public Administration Review 70(5) 681-690

Perry James L and Wouter Vandenabeele 2008 Behavioral dynamics Institutions

Identities and Self-Regulation In Motivation in Public Management edited by

James L Perry and Annie Hondeghem 56-79 Oxford Oxford University Press

Perry James L and Lois Recascino Wise 1990 The Motivational Bases of Public Service

Public Administration Review 50 367-373

Rainey Hal G 1982 Reward Preference among Public and Private Managers In Search of

the Service Ethic American Review of Public Administration 16 288-302

Ringle Christian M Oliver Goumltz Martin Wetzels and Bradley Wilson 2009 On the Use of

Formative Measurement Specifications in Structural Equation Modeling A Monte

Carlo Simulation Study to Compare Covariance-Based and Partial Least Squares

Model Estimation Methodologies Faculty of Economics and Business Administration

Maastricht Univeristy

Ringle C M S Wende and S Will 2005 Smart PLS 20 (M3) Beta

- 32 -

httpwwwsmartplsde

Ritz Adrian 2011 Attraction to Public Policy Making A Qualitative Inquiry into

Improvements in PSM Measurement Forthcoming in Public Administration

Spector Paul E 1992 Summated Rating Scales Construction An Introduction Newbury

Park CA Sage

Steenbergen Marco R 1996 Compassion and American public opinion An analysis of the

NES Humanitarianism Scale Report to the Board of Overseers of the National

Election Studies

Steffen Patrick R and Kevin S Masters 2005 Does compassion mediate the intrinsic

religion-health relationship Annals of Behavioral Medicine 30 217-223

Taylor Jeannette 2007 The Impact of Public Service Motives on Work Outcomes in

Australia A Comparative Multi-Dimensional Analysis Public Administration 85

931-959

Vandenabeele Wouter 2007 Towards a Theory of Public Service Motivation An

Institutional Approach Public Management Review 9 545-556

Vandenabeele Wouter 2008a Government Calling Public Service Motivation as an Element

in Selecting Government as an Employer of Choice Public Administration 86

1089-1105

Vandenabeele Wouter 2008b Development of a Public Service Motivation Measurement

Scale Corroborating and Extending Perryrsquos Measurement Instrument International

Public Management Journal 11 143-167

Vandenabeele Wouter 2009 The Mediating Effect of Job Satisfaction and Organizational

Commitment on Self-Reported Performance More Robust Evidence of the PSM-

Performance Relationship International Review of Administrative Sciences 75 11-

34

Wilcox James B Roy D Howell and Einar Breivik 2008 Questions about Formative

Measurement Journal of Business Research 61 1219-1228

Wise Lois Recascino 2000 The Public Service Culture In Public Administration Concepts

- 33 -

and Cases 7th ed edited by Richard J Stillman II 342-353 Boston Houghton

Mifflin

Wright Bradley E 2008 Methodological Challenges associated with Public Service

Motivation Research In Motivation in Public Management edited by James L

Perry and Annie Hondeghem 80-98 Oxford Oxford University Press

Wright Bradley E and Sanjay K Pandey 2008 Public Service Motivation and the

Assumption of Person-Organization Fit Testing the Mediating Effect of Value

Congruence Administration amp Society 40 502-521

- 34 -

Table 1

Possible Indicators of PSM and Descriptive Statistics (n = 2868)

Dimensions and items Mean Standard

Deviation

Item-Total

Correlation

Attraction to Public Participation (APP)

APP1 I am interested in helping to improve public service

APP2 I am satisfied when I see people benefiting from the public programs I was

involved in

APP3 I like to discuss topics regarding public programs and policies with others

APP4 I believe that public sector activities contribute to our general welfare

APP5 I admire people who initiate or are involved in activities to aid my

community

APP6 Contributing to public programs and policies helps me realize myself

APP7 It is important to contribute to activities that tackle social problems

412

441

355

393

423

351

409

696

662

917

753

672

881

734

524

421

482

412

493

478

552

Commitment to Public Values (CPV)

Sub-dimension for Public Interests (CPV1)

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

CPI3 I would prefer seeing public officials do what is best for the whole

community even if it harmed my interests

CPI4 Serving the public interest is more important than helping a single individual

Sub-dimension for Public Values (CPV2)

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of public

services

CPV3 It is fundamental that public services respond to the needs of the citizens

CPV4 Decisions regarding public services should be democratic despite the time

and effort it takes

CPV5 Everybody is entitled to a good service even if it costs a lot of money

CPV6 It is fundamental that the interests of future generations are taken into

account when developing public policies

CPV7 To act ethically is essential for public servants

CPV8 I believe that public employees must always be aware of the legitimacy of

their activities

CPV9 I personally identify with the aim of protecting individual liberties and rights

418

409

348

337

428

416

433

394

367

426

441

434

414

695

662

899

995

730

704

710

852

932

689

681

692

763

525

596

480

251

484

437

356

325

376

439

462

402

392

Compassion (COM) COM1 It is difficult for me to contain my feelings when I see people in distress

COM2 I feel sympathetic to the plight of the underprivileged

COM3 I empathize with other people who face difficulties

COM4 I have little compassion for people in need who are unwilling to take the

first step to help themselves

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

354

387

388

299

412

391

951

825

707

1083

755

765

399

511

389

011

487

502

Self-Sacrifice (SS)

SS1 Making a difference to society means more to me than personal achievements

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS5 People should give back to society more than they get from it

SS6 Serving other citizens would give me a good feeling even if no one paid me for

it

SS7 I would agree to a good plan to make a better life for the poor even if it costs

me money

357

329

339

302

345

363

350

904

872

949

982

888

1049

895

508

584

454

531

412

368

546

- 35 -

Table 2

Background of Respondents (n = 2868)

Variables Characteristics Respondents

Sex

Male

Female

NA

1209

1622

37

422

566

13

Length of service

(years)

0 ~ 10

10 ~ 20

20 ~ 30

30+

NA

1126

723

484

491

44

393

252

169

171

15

Organizational status

―Do you supervise employees

No

Yes

NA

1472

1076

320

513

375

112

Country

Australia

Belgium

China

Denmark

France

Italy

Korea

Lithuania

The Netherlands

Switzerland

United Kingdom

USA

249

214

230

249

266

162

253

236

249

250

260

250

87

75

80

87

93

56

88

82

87

87

91

87 Note NA = no answer

- 36 -

Table 3

Standardized factor loadings and correlations for the 31-item PSM index (n = 2868)

Dimensions and items SFL (MLE) SFL (D-WLS) Alpha

APP

APP1

APP2

APP4

APP5

APP6

APP7

597

486

455

558

490

600

657

552

514

625

549

665

699

CPV1

CPI1

CPI2

CPI3

CPI4

588

660

488

289

653

732

535

294

542

CPV2

CPV1

CPV2

CPV3

CPV4

CPV5

CPV6

CPV7

CPV8

CPV9

611

546

436

393

418

534

598

496

460

690

631

505

457

486

601

669

573

530

745

COM

COM1

COM2

COM3

COM5

COM6

496

569

498

597

607

516

651

538

670

684

687

SS

SS1

SS2

SS3

SS4

SS5

SS6

SS7

567

736

558

728

467

409

663

660

762

591

695

515

494

709

782

Measures of fit (MLE) χsup2 (df = 424) = 43699 p lt 001 CFI = 835 GFI = 899

RMSEA = 057 [055 059]

Measure of fit (D-WLS) χsup2 (df = 424) = 39247 p lt 001 CFI = 971 GFI = 976

RMSEA = 054 [052 055]

Inter-factor correlations MLE (D-WLS)

1 2 3 4

1 APS

2 CPV1 996 (988)

3 CPV2 902 (890) 798 (798)

4 COM 787 (783) 758 (762) 779 (783)

5 SS 700 (742) 800 (859) 509 (537) 720 (729)

Note SFL = standardized factor loading All factor loadings and correlations are significant at p lt 001

- 37 -

Table 4

Standardized factor loadings and correlations for the 16-item PSM index (n = 2868)

Dimensions and items SFL

(MLE)

SFL

(D-WLS)

APS

APP5 I admire people who initiate or are involved in activities to aid my

community

APP7 It is important to contribute to activities that tackle social problems

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

611

592

637

666

665

669

690

751

CPV

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of

public services

CPV6 It is fundamental that the interests of future generations are taken

into account when developing public policies

CPV7 To act ethically is essential for public servants

612

546

551

590

698

631

618

658

COM

COM1 It is difficult for me to contain my feelings when I see people in

distress

COM2 I feel sympathetic to the plight of the underprivileged

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

490

546

596

604

515

608

669

679

SS

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS7 I would agree to a good plan to make a better life for the poor even if

it costs me money

763

571

766

678

807

629

747

755

Measures of fit (MLE)

χsup2 (df = 98) = 8490 p lt 001

CFI = 939 GFI = 965

RMSEA = 052 [049 055]

Measure of fit (D-WLS)

χsup2 (df = 98) = 6069 p lt 001

CFI = 988 GFI = 992

RMSEA = 043 [039 046]

Alpha CR (D-WLS) Inter-factor correlations MLE (D-WLS)

1 APS 717 788 1 APS 2 CPV 3 COM

2 CPV 663 747 848 (859)

3 COM 645 713 781 (791) 789 (795)

4 SS 784 826 565 (596) 462 (505) 715 (726)

Note SFL = standardized factor loading Alpha = Cronbachrsquos alpha CR = composite reliability

All factor loadings and correlations are significant at p lt 001

- 38 -

Table 5

Results of CFA in the individual countries Standardized factor loadings and measures of fit

Factors

amp Items

Australia Belgium China Denmark France Italy

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

672

756

799

785

804

943

710

703

597

790

799

730

584

675

685

681

731

742

578

748

674

643

765

739

645

704

678

708

812

779

632

604

765

766

672

853

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

552

593

560

699

727

632

624

605

560

713

709

645

702

602

616

724

686

735

567

560

715

654

605

811

608

658

587

696

755

698

673

620

523

782

697

654

COM

COM1

COM2

COM3

COM5

COM6

623

584

662

688

701

739

732

608

475

808

666

581

578

616

605

649

670

673

699

751

737

679

816

776

568

546

801

626

673

850

637

546

741

715

601

822

SS

SS2

SS3

SS4

SS7

654

794

806

892

767

813

727

693

799

793

725

838

857

820

773

896

855

842

806

749

792

829

753

884

847

731

753

889

816

753

815

692

664

795

782

734

Inter-factor

correlations

093

~

752

163

~

810

320

~

773

323

~

825

452

~

827

497

~

855

323

~

850

321

~

847

498

~

899

536

~

902

425

~

829

476

~

865

Sample size 249 214 230 249 266 162

Measures of Fit

χsup2 638 444 881 732 765 554 817 625 735 565 750 564

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 980 100 939 987 969 997 967 994 975 997 949 995

GFI 961 986 941 986 948 991 949 992 955 992 926 986

RMSEA 036 00 063 050 051 026 053 035 045 026 059 033

- 39 -

Table 5 (Continued)

Factors

amp Items

Korea Lithuania The Netherlands Switzerland UK USA

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

619

701

667

706

781

717

666

512

682

751

605

779

557

629

630

620

719

729

535

621

702

643

694

735

655

725

738

710

804

822

535

621

702

643

694

735

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

594

607

722

722

644

722

674

535

533

790

589

711

568

554

538

655

628

545

627

580

615

711

666

704

800

512

555

873

581

646

627

580

615

711

666

704

COM

COM1

COM2

COM3

COM5

COM6

694

677

550

763

736

627

557

639

428

550

696

534

737

623

508

766

616

608

659

592

683

703

661

769

682

583

729

739

644

803

659

592

683

703

661

769

SS

SS2

SS3

SS4

SS7

776

752

679

779

770

811

707

721

635

725

734

706

814

656

743

820

767

764

745

632

689

788

662

744

671

818

832

830

864

802

745

632

689

788

662

744

Inter-factor

correlations

432

~

726

466

~

746

376

~

788

406

~

902

377

~

852

455

~

836

637

~

818

660

~

828

317

~

852

360

~

852

637

~

818

660

~

828

Sample size 253 236 249 250 260 250

Measures of Fit

χsup2 896 573 612 462 773 616 815 715 769 542 815 715

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 951 996 975 100 958 993 957 991 972 998 957 991

GFI 942 988 960 989 951 987 948 990 955 990 948 990

RMSEA 059 028 034 00 050 034 053 044 048 022 053 044

Note All factor loadings and correlations are significant at p lt 001

- 40 -

Appendix 1 Dimensions and possible items of PSM (the first round)

1 Attraction to Public Participation

(1) I am interested in developing public programs helping my country or community (Kim 2009a)

(2) Sharing my views on public policies with others is attractive to me (Kim 2009a)

(3) I am satisfied when seeing people get benefits from the public programs I was involved in (Kim 2009a)

(4) I like to discuss political subjects with others (Giauque et al 2011)

(5) I think the governmental activities contribute to our welfare

(6) I like to initiate actions to help out my community

(7) I do not care much about politicians (Perry 1996)

2 Commitment to Public Values

Possible items for public interest

(1A) Meaningful public service is very important to me (Perry 1996)

(2A) It is important for me to contribute for the common good (Giauque et al 2011)

(3A) I recognize myself with the mission of protection of liberties and rights endorsed by the nation-state

(Giauque et al 2011)

(4A) I would prefer seeing public officials do what is best for the whole community even if it harmed my

interests (Perry 1996)

(5A) To me serving the public interest is more important than helping other people (Vandenabeele 2008b)

Possible items for public values

(1B) I recognize myself with the promotion of the equality of chances endorsed by the nation-state (Giauque et

al 2011)

(2B) It is fundamental that the public service provided to the citizen is regular and continuous (Giauque et al

2011)

(3B) It is fundamental that public service innovates to respond to the new needs of the citizensclients (Giauque

et al 2011)

(4B) The dignity and well-being of all should be the most important concerns in any society (Steenbergen

1996)

(5B) It is important that public servants account for all the costsexpenses they make (Vandenabeele 2008b)

(6B) We have to do everything in our power to pursue the goal of democracy

3 Compassion

(1) It is difficult for me to contain my feelings when I see people in distress (Perry 1996)

(2) I feel sympathetic to the plight of the underprivileged (Kim 2009a)

(3) I am often reminded by daily events how dependent we are on one another (Perry 1996)

(4) I feel compassion for others in the difficulties they are facing (Steffen and Masters 2005)

(5) It is not really my problem if others are in trouble and need help (Beutel and Marini 1995)

(6) I find it hard to be sympathetic toward starving people in foreign lands when there is so much trouble in our

own country (Beutel and Marini 1995)

(7) I get very upset when I see other people treated unfairly (Beutel and Marini 1995)

(8) I would agree to a good plan to make a better life for the poor even if it cost me money (Beutel and Marini

1995)

(9) To me patriotism includes seeing the welfare of others (Perry 1996)

(10) I care very much about other people

4 Self-Sacrifice

(1) Making a difference in society means more to me than personal achievements (Perry 1996)

(2) I am prepared to make enormous sacrifices for the good of society (Perry 1996)

(3) I believe in putting duty before self (Perry 1996)

(4) I am one of those rare people who would risk personal loss to help someone else (Perry 1996)

(5) I think people should give back to society more than they get from it (Perry 1996)

(6) Serving other citizens would give me a good feeling even if no one paid me for it (Perry 1996)

(7) Much of what I do is for a cause bigger than myself (Perry 1996)

- 41 -

Appendix 2 Newly suggested items through the first round

1 Items for the Attraction to Public Participation Dimension

(1) In the public domain every person should be counted as equal

(2) I enjoy to think and argue in a political way

(3) I am satisfied when seeing my ideas being integrated in public policies

(4) I can easily relate to the way of thinking and arguing of politicians

(5) Working in the public sector I can better realize myself

2 Items for the Public Values Sub-Dimension

(1) Public service should give value for money

(2) Cost considerations should play an important role in decisions about public service provision

(3) It is fundamental that the interests of future generations are taken into account when developing public

policies

(4) To act with correct ethics is for public servants as important as their professional competence

(5) Primarily public servants should be accountable to the public and not to their superiors

3 Items for the Self-Sacrifice Dimension

(1) I would agree to a good plan to make a better life for the poor even if it costs me money

- 42 -

Appendix 3 An example of a summary of comments and suggestions (the second round)

On the item ―I am interested in developing public programs helping my country or community

A It is a good sentence but is it possible for everyone to ―develop public programs Even public officials are

limited to do this

B OK

C or the special group

D Double barrel and difficult to translate to my country ―I am interested in helping to improve public services

E Good but I feel that wording requires revision If we draw sample from local government then country is

redundant ―I am interested in developing public policy programs that aim in helping my community

F 4 (General appreciation from1 to 4) Not clear whether it refers to ones work or to ones concern as a citizen

―(In my work) I am interested in contributing to the development of public etchellip

G Instead of development I would use implementing

H Delete public which is redundant because of the characteristic helping my country or community but add

or implementing ―I am interested in developing or implementing programs helping my country or

community

I Referring to dimension Wording Similar to item 6

Note Names have been changed to alphabets in order to ensure the anonymity of participants

- 43 -

Appendix 4 Dimensions and Items of Public Service Motivation for Future Research

Dimensions and items

Attraction to Public Service (APS)

1 I am interested in helping to improve public service

2 I admire people who initiate or are involved in activities to aid my community

3 It is important to contribute to activities that tackle social problems

4 Meaningful public service is very important to me

5 It is important for me to contribute to the common good

Commitment to Public Values (CPV)

1 I think equal opportunities for citizens are very important

2 It is important that citizens can rely on the continuous provision of public services

3 It is fundamental that the interests of future generations are taken into account when

developing public policies

4 To act ethically is essential for public servants

5 I believe that public employees must always be aware of the legitimacy of their activities

6 I personally identify with the aim of protecting individual liberties and rights

Compassion (COM)

1 It is difficult for me to contain my feelings when I see people in distress

2 I feel sympathetic to the plight of the underprivileged

3 I empathize with other people who face difficulties

4 I get very upset when I see other people being treated unfairly

5 Considering the welfare of others is very important

Self-Sacrifice (SS)

1 I am prepared to make sacrifices for the good of society

2 I believe in putting civic duty before self

3 I am willing to risk personal loss to help society

4 I would agree to a good plan to make a better life for the poor even if it costs me money

- 2 -

The Development of an International Instrument to

Measure Public Service Motivation A Research Note

Introduction

Public service motivation (PSM) is thought of as ―an individualrsquos orientation to delivering

services to people with a purpose to do good for others and society (Perry and Hondgehem

2008a vii) The initial research on PSM began in the United States (Perry and Wise 1990

Rainey 1982) but scholars in many countries have contributed to the accumulation of

knowledge about its antecedents and consequences (Perry and Hondeghem 2008a Perry

Hondeghem and Wise 2010) As the scholarly research on PSM has grown and the

geographic scope of the research has expanded some scholars have questioned whether the

conceptual composition and dimensions of PSM are appropriate for explaining and predicting

public servicendashrelated behavior in different countries and internationally (Coursey and

Pandey 2007a Wise 2000 Wright and Pandey 2008 Giauque et al 2011 Vandenabeele

2008b) and whether different PSM measures are equivalent or if there is a need to develop a

more appropriate measure of PSM that can be used consistently and confidently (Castaing

2006 Leisink and Steijn 2009 Liu Tang and Zhu 2008 Taylor 2007 Vandenabeele 2008a

Wright 2008 Wright and Pandey 2008) Thus in order to facilitate research internationally it

is necessary to elaborate the construct and measurement of PSM (Perry and Hondeghem

2008b)

The first step was to develop a more universal concept that can be used globally and

that is likely to generate cumulative knowledge The construct of PSM has been sharpened by

refining the conceptual components in Perry and Wisersquos (1990) study Kim and Vandenabeele

- 3 -

(2010) proposed that PSM should continue to be conceived as a four-dimensional construct

with self-sacrifice as the foundational concept representing the altruistic or pro-social origins

of PSM The three other concepts attraction to public participation commitment to public

values and compassion which represent instrumental value-based and affective motives

round out their proposed modifications to the construct The revisions to the construct of

PSM will enhance cross-national research and generate cumulative knowledge

The second step was to specify a measurement model of PSM This was necessary

because the measurement model specifies the relationship between constructs and measures

(Diamantopoulos and Siguaw 2006) Reflective measures are treated as outcomes of

constructs and formative measures are causes of constructs (Edwards 2010) A measurement

model can either include reflective or formative indicators exclusively or consist of both

depending on the observed construct (Fornell and Bookstein 1982) The decision to

operationalize formative andor reflective indicators should be based on theoretical

considerations Previous studies discussing the conceptual definition of PSM theoretically

and empirically have proposed that the dimensions of PSM are formative and the indicators

of each dimension are reflective (Kim 2010 Kim and Vandenabeele 2010)

The next step is to develop an internationally robust measurement instrument of PSM

for cross-national research and comparison A survey-based measure for PSM has been useful

for facilitating comparisons across disparate services and national settings and creating

foundations for the accumulation of results (Perry Hondeghem and Wise 2010) Perry (1996)

identified a multidimensional 24-item scale to measure PSM Since then many studies have

used the items of Perryrsquos (1996) measure and improvements have been made to the PSM

scale (Coursey and Pandey 2007a Kim 2009a Taylor 2007 Vandenabeele 2008b) However

research has shown that the original items do not fully represent unique and salient qualities

- 4 -

of the dimensions of the PSM construct (Castaing 2006 Coursey and Pandey 2007a Coursey

et al 2008 DeHart-Davis Marlowe and Pandey 2006 Kim 2009b Leisink and Steijn 2009

Liu Tang and Zhu 2008 Moynihan and Pandey 2007b Taylor 2007 Vandenabeele 2008a

Wright and Pandey 2008) Moreover as the construct and operational definition of PSM are

refined it is necessary to adapt items for the dimensions of the PSM construct

This study is focused on developing a revised measurement instrument for PSM for

international use After developing the possible items for measuring PSM the international

survey was conducted in 2010 and the data (n = 2868) were collected from twelve countries

Dimensions and items were tested cross-nationally as this can be a useful tool to preclude

value-specific items and wordings that are case-specific The next section of this paper will

briefly explain the processes for developing the international measurement instrument The

survey results are analyzed and discussed in order to confirm the dimensional structure and

validate the items of PSM Finally the items of PSM are proposed and discussions and

suggestions are given for future research

Developing the Measurement Instrument of PSM for International Use

Background

Perry and Wise (1990) proposed that PSM is associated with three types of motives affective

norm-based and rational Perry (1996) identified four empirical components of the PSM

construct attraction to public policy making commitment to the public interestcivic duty

compassion and self-sacrifice Since then many studies have used the dimensions and items

of Perryrsquos (1996) measure and improvements have been made in the PSM scale (Kim 2009a

Taylor 2007 Vandenabeele 2008b) However Perryrsquos (1996) measure has limitations (Kim

and Vandenabeele 2010) In particular the items of the dimension of attraction to policy

- 5 -

making are not appropriate for measuring personal attraction to public policy making

(Camilleri 2006 Coursey and Pandey 2007a Coursey et al 2008 Kim 2009b Ritz 2011)

Moreover the dimension of commitment to public interest needs to concentrate more on a

personal disposition to pursue public values (Castaing 2006 Leisink and Steijn 2009 Taylor

2007) and the items of compassion do not represent affective motives (DeHart-Davis

Marlowe and Pandey 2006 Moynihan and Pandey 2007a Wright 2008) Thus the

dimensions of the PSM construct are modified as attraction to public participation

commitment to public values compassion and self-sacrifice Consequently it is necessary to

develop appropriate items for the dimensions (Kim and Vandenabeele 2010)

A related issue concerns the measurement model of PSM There are two

measurement models using multiple dimensions of latent constructs reflective and formative

When variation in a construct leads to variation in its dimensions such dimensions are

termed ―reflective while dimensions are labeled ―formative when they are viewed as

causes of a construct (Edwards and Bagozzi 2000 Diamantopoulos and Siguaw 2006

Diamantopoulos and Winklhofer 2001) It is understandable that each dimension of PSM has

several reflective indicators that are essentially interchangeable A problem is the relationship

between PSM and its dimensions If we regard the dimensions as reflective it is possible to

analyze PSM with only two or three dimensions because omitting dimensions does not

change the essential nature of PSM However if we regard the dimensions as formative it is

not acceptable to omit a dimension as this may change the meaning of PSM Construct

conceptualization is regarded as being the process that determines the formative or reflective

nature (Diamantopoulos and Siguaw 2006) Recent research has demonstrated theoretical and

empirical evidence in support of a second-order formative approach to PSM (Kim 2010)

This means that the dimensions represent different conceptual components of PSM and so

- 6 -

they are not interchangeable Thus PSM is regarded as a combination of specific dimensions

If any one of the dimensions changes PSM would change Conversely if an individualrsquos

PSM increases this would not necessarily be accompanied by an increase in all dimensions

Therefore as the PSM construct is refined for international use (Kim and Vandenabeele

2010) scholars must be attentive to identifying and using appropriate measurement models

Developing Possible Items

The conceptual components of PSM need to be empirically estimated through operational

definition Kim and Vandenabeele (2010) proposed that PSM is based on self-sacrifice and

associated with instrumental value-based and identification motives and that the dimensions

of the PSM construct be refined along the lines of attraction to public participation (APP)

commitment to public values (CPV) compassion (COM) and self-sacrifice (SS) As Perry

(1996) identified a multidimensional scale to measure PSM it is conceived as an overall

latent variable with various latent dimensions and is referred to as a second-order factor

model Research has found that the measurement model specifies the causal relationship

between PSM and its dimensions (Edwards and Bagozzi 2000) The direction of the

relationship is either from the PSM construct to the dimensions (second-order reflective

measurement) or from the dimensions to the construct (second-order formative measurement)

A theoretical consideration of the relationships between PSM and its dimensions led

to the proposal that PSM be defined in terms of multiple dimensions (Kim 2010 Kim and

Vandenabeele 2010) The dimensions represent different aspects of PSM with each

dimension capturing a distinct and potentially unique form of PSM Moreover the

dimensions may have different antecedents and consequences as well as different

characteristics and theoretical backgrounds (Moynihan and Pandey 2007a Pandey and

- 7 -

Stazyk 2008 Perry 1996 1997 2000 Perry and Wise 1990 Perry and Vandenabeele 2008

Taylor 2007 Vandenabeele 2007 2008a 2009) An individualrsquos PSM is shaped by the

individualrsquos APP CPV COM and SS Deleting a dimension means omitting a certain part of

PSM Therefore it is reasonable to suggest that the dimensions of PSM are formative and for

PSM to be regarded as a composite of its dimensions

Each dimension of PSM can be measured using reflective indicators that satisfy the

following criteria (Jarvis MacKenzie and Podsakoff 2003) a relative homogeneity and

interchangeability of indicators pertaining to each dimension a high degree of covariation

among indicators of each dimension and the expectation that the indicators of each

dimension are likely to be affected by the same antecedents and have the same consequences

Reflective indicators are essentially interchangeable and so adding or removing indicators

does not change the essential nature of a dimension Thus we consider PSM to be a second-

order formative construct with several dimensions that each has several reflective indicators

(reflective first-order formative second-order)

Based upon the refinement and clarification of the operational dimensions of PSM

(Kim and Vandenabeele 2010) items were developed for the four dimensions The items

selected were a mix of pre-existing items that loaded highly on PSM dimensions in multiple

studies (Giauque et al 2011 Kim 2009a Perry 1996 Vandenabeele 2008b) Where

appropriate items were not available new items were drafted from other previous studies

creating an initial total of 35 proposed items For the dimension of APP which focuses on

individual disposition to work in the public sector to participate in the public policy process

and in activities for community and social development seven possible items were suggested

The CPV dimension focuses on personal disposition to pursue public interest and public

values Two sub-dimensions of CPV seemed plausible one focusing on public interest and

- 8 -

another on public values such as social equity equal opportunity and democracy Thus five

possible items for the public interest sub-dimension (CPV1) and six items for the public

values sub-dimension (CPV2) were developed For the dimension of COM which focuses on

affective bonding with the identified objects such as other members of a social category or of

a political system ten possible items were suggested For the SS dimension Perryrsquos original

items (1996) were the primary source of items The full list of items is provided in Appendix

1

Review Processes

A group of researchers who shared our interest in better measurement of PSM collaborated

using e-mail in November 20091 The objective was to develop a measurement instrument

for PSM with a robust comparative character in other words to create a valid instrument (on

a national basis) that can be used internationally The first round for reviewing the possible

items lasted from the 16th

of November until the 7th

of December 2009 and all the

participants were asked to review these proposed items For this review two criteria were

employed (1) Do these items refer to the dimensions they claim to refer to and (2) Are

these items meaningful across various national contexts and in particular to the situation of

civil servants in that country Participants were also asked to add new items that might better

reflect a particular dimension or to rephrase items that were already on the list

A second round of item refinement lasted from the 9th

of December until the 22nd

of

December 2009 During this time 11 newly suggested items from the first round were

considered (see Appendix 2) A summary of all comments and suggestions emerging from the

1 Emmanuel Camilleri David Giauque Gerhard Hammerschmid Isabel Egger-Peitler Renate Meyer and

Cristian Pliscoff also participated in the review processes

- 9 -

first process was distributed to the participants to stimulate comments in the second round

(see Appendix 3) Based on the comments received in the second round discussion the

possible items for an international survey were selected However some items required

further discussion and so a third round was conducted from the 15th

of January until the 23rd

of January 2010 One issue that remained unresolved in the discussions relates to the

dimension of CPV Some argued that it is reasonable to divide CPV into two sub-dimensions

because committing to public interests is different from pursuing public values and there are

many important public values in democratic countries while others contended that CPV can

cover all important public values including public interest and there is no real benefit to add

one more dimension These differences were left to be resolved through empirical research

This did not affect the final items however and 33 items were confirmed The resulting

survey questionnaire was distributed to all the collaborators on February 9 2010 in order to

conduct a survey in each country

Measures

PSM was measured using the newly developed 33-item index 7 items for the dimension of

APP 13 items for CPV (4 for CPV1 and 9 for CPV2) 6 items for COM and 7 for SS In the

survey questionnaire a five-point Likert type scale (1 = strongly disagree 5 = strongly agree)

was used This scale is among the most widely used response formats in many empirical

investigations within the social and behavioral science in which there are no ―correct or

―incorrect responses (Spector 1992) The items of PSM were randomly distributed in the

survey format Gender age education length of service and employment status were also

measured in this survey

It was necessary to translate the measurement items into the official (native) language

- 10 -

of each country As this raised the possibility of inaccurate translations the back-translation

of the items into native languages was validated by asking two researchers to independently

translate the English-language items The researchers were instructed to consult with each

other over differences between translated items and to test the survey items with some native

respondents

[Table 1 about here]

Samples

The international survey was fielded from March to September 2010 The data upon which

this study are based were collected from civil servants in local governments in Australia

Belgium China Denmark France Italy Korea Lithuania the Netherlands Switzerland the

United Kingdom and the USA The principal criterion was that the samples collected in the

various countries should be comparable to the largest possible extent The criterion ruled out

national and state government as the competencies of national governments and also of state

governments if they exist constitutionally are different among most of the countries included

in the research Therefore we opted for local government as our focus The structure and

functions of local governments can differ widely even within countries and so the survey in

each country was focused on town hall bureaucrats in city county or township government

The respondents were permanent employees from both managerial and non-managerial levels

and from both administrative departments (ie no direct contact with the public) and service

departments (ie likely to have direct contact with the public) Despite being employed by

local governments in many or all of the countries police officers firefighters school teachers

public transport workers artists and musicians as well as nurses and doctors were excluded

- 11 -

from the sample

We aimed to receive 250 respondents for each country as this would enable us to

carry out the analyses we wanted to perform on the data Although there is no ideal sample

size one study recommends a sample of 200 to provide a sound basis for estimation (Hair

Black Babin and Anderson 2010) The method of data collection was left open Obviously a

web-survey is the easiest way to collect the data because it is economical to collect and to

code however compared to paper-pencil surveys this method has some drawbacks First not

everybody in the sample may have access to the internet which may cause a bias in the

responses Second internet surveys are generally known to produce lower response rates

Web surveys were recommended if possible but paper-pencil surveys were deemed

acceptable in cases where it was not possible to conduct a web survey Researchers were

encouraged to contact just one municipality except in cases where it was not possible to

gather enough data In such instances responses from multiple municipalities were accepted

By the end of September 2010 a total of 2868 responses were obtained2

2 In Australia seven local councils each with between from 70 and 232 employees were surveyed Paper-pencil

surveys were used in two councils while the rest relied on a web survey The local governments limited

distribution of paper-pencil surveys to only one or two sections of the council eg administration The response

rate was approximately 30 However it should be noted that not everyone working for the council has direct

access to a computer In Belgium three municipalities were surveyed on a census base by means of a web

survey as all employees had access to a computer This rendered a response of 120 (226) 79 (75) and 16

(53) responses for the respective municipalities resulting in an overall response rate of 331 with 662

invitations being sent out In China two departments social work and administration were surveyed in one

municipality using paper-pencil surveys In the social work department 65 out of 110 surveys were returned

and 165 out of 211 responded from the administration department giving an overall response rate of 767 In

Denmark a total of 1282 respondents in a municipality were sent an initial e-mail invitation to a web survey

and up to two e-mail reminders and the response rate was 473 In France all the employees in two

municipalities were surveyed In one municipality where both web and paper-pencil surveys were utilized 128

out of 685 questionnaires were returned while in another city using only a paper-pencil survey 138 out of 652

were received The average response rate was 199 In Italy in principle the link to the questionnaire was

made available to all employees (about 4000) in one city but for some of them access may have been difficult

yielding a response rate of 45 In Korea 500 employees in a city received a printed questionnaire and 253

responses were returned yielding a response rate of 506 In Lithuania 236 responses from 54 municipalities

were gathered by web survey In the Netherlands three municipalities were surveyed on a census base by means

of a web survey In total 319 responses of 910 invitations were returned (35) mdash 90 (45) 115 (38) and 114

(28) for the respective municipalities In Switzerland 516 out of 1326 local employees with e-mail access in

a municipality completed the survey giving a response rate of 389 In the United Kingdom the survey was

posted online and 1360 employees both in administrative jobs and front-line services of one municipality

- 12 -

Municipalities were selected because of their convenient accessibility and proximity to the

researcher and a convenience sample was used in most countries Seven countries surveyed

one municipality but seven local councils were surveyed in Australia Web surveys were used

in four countries while three countries used both web and paper-pencil survey The number

of respondents varied among the countries depending on the numbers of questionnaires

distributed the survey method and the response rate In most countries we used

approximately 250 responses but only 162 from Italy3 The majority of the respondents were

women (566) while in terms of service length those in the largest group (393) had

worked for fewer than 10 years Table 2 shows the distribution of respondentsrsquo sex length of

service employment status and country

[Table 2 about here]

Analyses

The statistical analysis applied confirmatory factor analysis (CFA) CFA was used to assess

the fit of the data to the hypothesized measurement model because it is preferred where

measurement models have a well-developed underlying theory for hypothesized patterns of

loadings The data were analyzed using Amos 180 (Arbuckle 2009) with the maximum

likelihood (MLE) estimation method and LISREL 872 (Joumlreskog and Soumlrbom 2005) with the

diagonally weighted least squares (D-WLS) estimation method

The individual items are designed to measure a theoretically continuous construct

but the observed responses are a discrete realization of five categories The common method

received an email to complete the survey The response rate was 19 In the USA the total number of

employees surveyed was 626 in one city using both electronic and paper-pencil surveys resulting in a response

rate of 588 3 In countries where significantly more than 250 surveys were collected we randomly selected only 250 cases

for analysis

- 13 -

of estimation within CFA is MLE a technique which assumes that the observed variables are

continuous and normally distributed These assumptions are not met when the observed data

are discrete as occurs when using ordinal scales as in this study A simulation study (Flora

and Curran 2004) found that D-WLS performs well for ordinal measures across a variety of

conditions especially for smaller samples Coursey and Pandey (2007a 2007b) argued the

importance of using D-WLS and not standard MLE in structural equations for ordinal data4

In most cases however the differences between MLE and D-WLS are not substantive to the

conclusions5 No comparison between standard MLE and D-WLS is provided in this paper

We give results from both MLE and D-WLS and this will aid in confirming the stability and

robustness of the results across varying mathematical specifications

For gauging reliability Cronbachrsquos alpha and composite reliability were tested

Composite reliability should be equal to or greater than 07 (Hair et al 2010) Convergent

validity was assessed from the measurement model by determining the statistical significance

of each indicatorrsquos estimated pattern coefficient on its specified underlying construct

Discriminant validity was assessed by determining whether the confidence interval around

the correlation estimate between two factors included 100 (Anderson and Gerbing 1988)

For model fit assessment goodness-of-fit index (GFI) comparative fit index (CFI)

and the root mean square error of approximation (RMSEA) were used The model achieves

an acceptable fit to the data when GFI and CFI equals or exceeds 090 and RMSEA values

4 Applying standard MLE to ordinal data such as Likert-type items may produce significant estimation

problems such as inflation of chi-square fit statistics and biased underestimation of parameters and standard

errors (Flora and Curran 2004) Coursey and Pandey (2007a 2007b) used a modified MLE using the asymptotic

covariance matrix and D-WLS for adjusting the ordinal nature of the data They reported that although the

modified MLE findings are slightly different on a few model parameters the differences are not substantive to

the conclusions However model fit estimators were clearly better for the D-WLS 5 Given normally distributed categorical variables MLE can be used with confidence when a variable has four

or more categories If there is a large sample a five-point Likert scale and the skewness and kurtosis of variables

is within acceptable limits it suggests distribution symmetry ―In most cases where the hypothesized model is

well specified and the scaling based on more than three categories it seems unlikely that there will be much

difference between the findings (Byrne 2010 160)

- 14 -

fall below 008 (Hair et al 2010) In general the larger the value of GFI and CFI and the

smaller the value of RMSEA the better the fit of the model (Bollen 1989) We also present

the χ2 test statistic Lower values of χ

2 indicate a better fit and should be nonsignificant but

the χ2 value increases as sample size increases and thus for large sample sizes this statistic

may lead to rejection of a model with good fit (Hair et al 2010)

Results

Using SPSS 180 descriptive statistics were computed for individual items as shown in table

1 One item (APP3) in the APP dimension ―I like to discuss topics regarding public programs

and policies with others was mistakenly omitted in the survey conducted in the United

States and so it was deleted in the statistical analysis In the assessment of normality no

items showed a skew or kurtosis value greater than the cutoffs of |3| or |8| recommended by

Kline (2005) As data were distributed normally MLE as well as D-WLS estimation was used

Based on inspection of item-total correlations one item (COM4) was dropped from further

analysis because it was weakly correlated with the overall index (011) Missing data is an

issue in CFA as it can cause estimation errors All variables in the survey have missing data

rates of approximately 016 to 143 Mean substitution was used to calculate replacement

values for the missing values in each countryrsquos dataset (Hair et al 2010) The deletion of the

two items left 31 items for the next stage of analysis

The five-correlated factor model was tested using CFA The CFA model in the initial

analysis hypothesized a priori that (1) responses to the 30-item PSM measurement

instrument could be explained by five factors such as APP CPV1 CPV2 COM SS (2) each

item would have a nonzero loading on the PSM factor that it was designed to measure and

- 15 -

zero loadings on all other factors (3) the five factors would be correlated and (4)

measurement error terms would be uncorrelated The resulting CFA with MLE suggested that

the initial five-correlated factor model was not a good fit to the data χsup2 (df = 424) = 43699 p

lt 001 GFI = 0899 CFI = 835 RMSEA = 0057 [055 059] At this stage APP was highly

correlated with CPV1 (r = 996 in MLE and r = 988 in D-WLS) suggesting the two

dimensions were not unique and lacked discriminant validity

[Table 3 about here]

Given the disconfirmation of the initial model an effort was made to estimate an

alternative model The two dimensions APP and CPV1 were combined to form a single

dimension Attraction to Public Service (APS) The items of CPV1 may be regarded as

action-oriented by the respondents because dedication to public service community and

common good is underlined in these items Consequently it is reasonable to combine APP

with CPV1 to form APS a dimension that focuses more on disposition to serve the public to

work for the common good and to participate in public policy processes After the overall

number of dimensions was reduced to four using standardized factor loadings and

modification indices as criteria the items with the lowest factor loading in each dimension

were deleted and the items whose modification index indicated association with multiple

dimensions were eliminated A CFA model with the remaining items was iteratively tested

until it achieved a good fit to the data The outcome of this refinement process was a 16-item

index of four factors Overall model fit by every measure is quite strong for the D-WLS

estimation technique The estimation results for the four-dimension model are presented in

table 4

- 16 -

[Table 4 about here]

Cronbachrsquos coefficient alpha for the 16-item PSM index was 860 and the coefficient

alphas for the four dimensions ranged from 645 to 784 The composite reliability of the set

of reflective indicators for each dimension of PSM ranged from 713 to 826 and indicated

adequate internal consistency The resulting factor structure showed a four-factor structure

with all items loading significantly onto their a priori dimension (p lt 001) and the

standardized factor loadings ranging from 0490 to 0766 in MLE and from 515 to 807 in

D-WLS The results provide support for convergent validity The correlation estimates

between the two factors ranged from 0462 to 0848 in MLE and from 505 to 859 in D-

WLS The confidence intervals (plusmn2 standard errors) around the correlation estimates between

the two factors did not include 100 The result provides support for discriminant validity To

test whether the four-dimension model is superior to other designs such as the two or three-

dimension models a series of CFA were conducted and the results show that the four-

dimension model is a significantly better fit than any of the other designs

To test whether the factorial structure of the PSM instrument is equivalent across the

samples from the twelve countries four nested models were tested as part of multi-group

analysis with MLE (Byrne 2010) First the unconstrained model was the baseline model

which relaxed all equality constraints This model hypothesized the same four-factor structure

of the instrument across twelve countries The fit indices for this baseline model were χsup2 (df =

1176) = 24977 p lt 001 GFI = 0901 CFI = 900 RMSEA = 0020 [019 021] These

indices indicate that the four-factor model of the PSM instrument exhibits acceptable fit

across the samples from the twelve countries Second the measurement weights model tested

- 17 -

the invariance of factor loadings across countries by placing equality constraints on these

parameters The fit indices for this model were χsup2 (df = 1308) = 28088 p lt 001 GFI = 0890

CFI = 887 RMSEA = 0020 [019 021] Since the measurement weights model was nested

within the unconstrained model the chi-square difference text ∆χsup2 (132) = 3111 p lt 001

indicated that some equality constraints of factor loadings did not hold across the twelve

countries Further test results suggested that the equality constraints of structural covariances

and measurement residuals were not upheld across the twelve countries It needs to be

stressed that despite the differences in factor loadings and factor covariances the results

support the main assumption that the same four-factor structure holds across different

countries We may assume that people in different cultures understand PSM in the same

fashion

To find differences in the factor structure of PSM among countries we conducted a

series of CFA with each countryrsquos dataset and followed the same procedure as when

analyzing the sample as a whole As table 5 shows although there are great similarities

between the appropriate items in the countries there are also some differences Even though

the items vary according to national differences the four-factor structure had a good fit to the

data Thus the results confirm that the four-factor structure of PSM can be generalized

[Table 5 about here]

The APS dimension covers both an individualrsquos attraction to public participation and

commitment to public interest but all its items refer to some notion of public interest such as

improving public service aiding community tackling social problems meaningful public

service and contributing to the common good The APS dimension captures an individualrsquos

- 18 -

disposition to serve the public interest through public participation When developing the

items some assumed that the items of committing to public interests can be separated from

those of pursuing public values in the CPV dimension The empirical test showed that

respondents tended to understand committing to public interests as behavior-oriented and so

the items for public interest are not differentiated from participation-oriented items in their

perceptions but are different from those for public values The respondents seem to

understand that pursuing public values is more related to personal beliefs or creeds Thus the

CPV dimension is solely focused on public values and the items of pursing public interests

are combined with those of the APP dimension to make the APS dimension

Public values and their relative importance are neither impervious to change over

time nor the same among all countries (Charles de Jong and Ryan 2011 Davis and West

2009) Moreover some values may be in conflict as Martinsen and Beck Joslashrgensen (2010)

discussed in regard to the tensions between accountability and efficiency It is therefore not

possible to include all public values in one dimension (Beck Joslashrgensen and Bozeman 2007)

Accordingly the items of the CPV dimension are developed for estimating individual

commitment to seek public values which there are general consensus about and which are

generally formulated

When all four dimensions of PSM are included in a survey cross-national

equivalence of the survey can be assured The items represent reflections or manifestations of

each dimension and hence are relatively homogeneous and interchangeable Adding or

deleting items does not change the essential nature of the dimension Thus any combination

of items in a dimension can be regarded as equivalent When using structural equation

modeling (SEM) each dimension needs to have at least three reflective indicators in order to

achieve identification or it needs to have at least two reflective indicators and be allowed to

- 19 -

covary with another dimension that has at least two reflective measures With one indicator

the dimension is identified if the loading and the variance of the uniqueness are both fixed

(Bollen 1989 Edwards 2010) Thus the number of items in each dimension is not restricted

when measuring PSM However we recommend using multiple items for each dimension

because more research is needed to investigate the appropriateness of this PSM instrument

with varying samples

The test results showed that model fit estimators were clearly better for the D-WLS

It means that in this study MLE was more rigorous than D-WLS in assessing the fit of the

model The empirical evidence indicated that MLE may be an acceptable technique for

ordinal data when there is a large sample a five-point Likert scale and the skewness and

kurtosis of variables is within acceptable limits as suggested by Byrne (2010)

Discussion

The analysis reaffirmed the four-dimensional structure of PSM but the results were not

entirely consistent with the original CFA we tested We were ultimately able to converge on a

good-fit operational model for PSM that was also consistent with underlying theory The

analysis is not able to resolve the debate about measurement models however for reasons

discussed below

Choosing the correct measurement model (ie reflective vs formative) is important

as measurement model misspecification results not only in misrepresentation of the epistemic

relationships between constructs and their measures but also in biased parameter estimates of

structural relationships (Jarvis et al 2003 Law and Wong 1999 MacKenzie 2003

MacKenzie Podsakoff and Jarvis 2005) A formative measurement model in isolation is

- 20 -

under-identified and cannot be estimated in the case of covariance-based SEM (CB-SEM)

(Bollen 1989) A construct with formative indicators must emit paths to at least two unrelated

reflective indicators two unrelated latent constructs or some combination of the two in order

to solve the problem of under-identification (Jarvis MacKenzie and Podsakoff 2003

MacCallum and Browne 1993 MacKenzie Podsakoff and Jarvis 2005) Thus ―when

dependent constructs change the empirical nature of the formatively measured construct

changes (Wilcox Howell and Breivik 2008 1223) and ―when specifying formative

measurement models what might seem like an innocuous choice of outcomes to achieve

identification can have fundamental ramifications for the meaning of the formative construct

and the inferences based on it (Edwards 2010 7)

Earlier research found that the association between a construct and its measures

should remain stable regardless of the larger causal model in which the constructs and

measures are embedded (Edwards and Bagozzi 2000) The association between a construct

and its measures is generally stable for reflective measures but associations of formative

measures with their construct are unstable The susceptibility of formative measurement to

both interpretational confounding and external inconsistency recommends that formative

measurement is not an equally attractive alternative to reflective measurement and that

whenever possible researchers should opt for reflective measurement (Howell Breivik and

Wilcox 2007a 2007b Kim Shin and Grover 2010) In the case of second-order models

researchers are limited to only a second-order reflective model in CB-SEM (Chin 2010)

Regardless of the measurement debate and the tradeoffs between reflective or

formative specification this study makes a theoretical argument that PSM is a second-order

formative construct This second-order formative factor model (reflective first-order

formative second-order) is also suggested as an alternative model that captures the

- 21 -

multidimensionality of complex concepts and avoids the shortcomings of first-order

formative measurement (Edwards 2010) However this model itself cannot achieve

identification One possible strategy is to develop direct reflective measures of PSM based on

the view that PSM is a general concept that is not only influenced by its specific dimensions

but also can be assessed using general measures (Edwards 2010 Javis MacKenzie and

Podsakoff 2003)6 However these general measures have not yet been developed for the

PSM construct

Another way around the problem of underidentification of formative models is to

apply the component-based approach known as partial least squares path modeling (PLS-PM)

which can model formative indicators (Fornell and Bookstein 1982 Goumltz Liehr-Gobbers

and Krafft 2010) The CB-SEM and the PLS-PM methods have distinctive statistical

characteristics and selecting an approach to SEM depends on the particular research situation

CB-SEM is the method of choice for theory testing while PLS-PM is appropriate for

prediction-oriented applications (Ringle Goumltz Wetzels and Wilson 2009) One advantage of

the PLS method is that it demands significantly fewer requirements compared to that of

covariance structure analyses but nevertheless delivers consistent estimation results Another

asset of the PLS approach is its ability to deal with formative as well as reflective indicators

even within one structural equation model This indicates that the PLS approach can be

appropriate for explorative analysis of structural equation models (Chin 2010 Goumltz Liehr-

Gobbers and Krafft 2010) A Monte Carlo simulation study to compare CB-SEM and PLS-

PM showed that PLS-PM provides a viable approximation of model parameters when the

prerequisites for CB-SEM are not met (Ringle Goumltz Wetzels and Wilson 2009) A formative

6 For example job satisfaction can be assessed with general reflective measures such as ―in general I am

satisfied with my job and the formative dimensions can be assessed with measures that describe satisfaction

with each job facet (Edwards 2010 15)

- 22 -

composite in the form of indexes needs to be tested to know the relative weights (γ-

parameters) of the dimensions of PSM In PLS-PM a second-order formative factor is

directly measured by observed variables for all the first order factors While this approach

repeats the number of manifest variables used the model can be identified without adding

outcome constructs that would be considered inappropriate or irrelevant from a conceptual

standpoint (Chin 2010 Edwards 2010) The test result of PLS-PM using Smart PLS 20

(Ringle Wende and Will 2005) produced the function PSM = 31APS + 36CPV + 28COM

+ 31SS However one consideration is whether and how construct error is modeled for the

formative construct (Diamantopoulos 2006) When using PLS-PM the researcher is making

the assumption that there is no error at the construct level as all constructs in PLS-PM are

modeled without error When using PLS-PM the absence of an error term will tend to

increase the weights therefore inflating their importance (Bollen and Lennox 1991) The

primary statistic for assessing a formative indicator is its weight the partialized effect of the

indicator on its intended construct controlling for the effects of all other indicators of that

construct (Cenfetelli and Bassellier 2009) A formative measure is essentially a multiple

regression with the construct representing the dependent variable and the indicators as the

predictors (Diamantopoulos and Winklhofer 2001) PSM may be regarded as a weighted

linear composite of its dimensions The weights used to create the composites are different

across CE-SEM and PLS-PM The causal dimensional weights generated in CB-SEM are

estimated in a model in which a construct-level error term is specified (eg PSM = γ1APM +

γ2CPI + γ3COM + γ4SS + ζ) whereas the weights in PLS-PM are not (eg PSM = γ1APM +

γ2CPI + γ3COM + γ4SS) This leads to two different conceptualizations of the PSM construct

one in which the causal indicators do not capture the construct in its entirety and another in

which they do (Hardin et al 2010) These differences need to be discussed in future research

- 23 -

in this area

One practical suggestion to measure PSM is to apply unit weights to its dimensions if

we assume the effects of the four dimensions are equivalent A literature review indicates that

unit weights have substantial predictive validity when compared with regression weights The

literature shows that the empirical correlations between scores generated from differential and

unit-weighting approaches are quite high In creating composite scores unit weights can be

an appropriate approach for weighting under many circumstances (Bobko Roth and Buster

2007) Although the uses of unit-weight composites will likely result in the explanation of

less variance than empirical estimations of dimensional weights they will remain consistently

defined across studies and hence suitable for theory testing Alternatively more thorough

discussions on the dimensions of PSM may suggest fixed-weight composites (Hardin et al

2010) A previous study (Kim 2010 22) also suggested ―Overall indexing PSM by

measuring all the dimensions of PSM and simply summing the averages of each dimension

can make it easier to measure PSM in practice as PSM = APS + CPV + COM + SS

An advantage of regarding PSM as a second-order formative construct is that the

dimensions can be analyzed both individually and collectively This is important as the

dimensions are expected to have different antecedents and consequences For instance Perry

(1997) found some significant differences in the influence of independent variables on the

four dimensions of PSM Taylor (2007) confirmed that certain dimensions are more important

than others in influencing work outcomes Moynihan and Pandey (2007a) reported that red

tape is negatively related to one dimension but reform orientation and hierarchical authority

is positively related to another dimension A recent study found that the SS dimension has a

positive relationship with occupational intention but the other dimensions do not (Liu et al

2011) Thus it is valuable to analyze the causes and effects of individual dimensions of PSM

- 24 -

as well as of the PSM construct

Conclusion

The purpose of this study is to develop a measurement instrument for PSM for cross-national

research and comparison On the basis of the sharpened construct of PSM (Kim and

Vandenabeele 2010) using data from an international survey conducted in 12 countries we

found that the four-factor structure of PSM is generalizable across country cases This study

provides the measurement index of PSM with the four dimensions of APS CPV COM and

SS The APS dimension measures the instrumental motives of PSM the CPV dimension

determines the value-based motives of PSM while the COM and SS dimensions gauge the

identification motives and self-sacrifice respectively When all four dimensions of PSM are

used in a survey cross-national equivalence of survey items can be assured The items for

measuring PSM in future research are suggested in Appendix 4

This study has several limitations First looking for samples which were comparable

to the largest possible extent we opted for surveys in local government The results found

and discussed in this study are based on PSM among local government employees The

results might be different when surveying central government employees or other specific

groups in the public sector Thus the potential impact of the sample bias needs to be

considered in understanding this study It also means that the survey items for measuring

PSM should be examined with different samples in various areas and at different government

levels

Second the test results showed that there is a high degree of correlation among the

dimension such as APS CPV and COM Future research should not only investigate whether

- 25 -

the items reflect the core content of the dimensions but also continue to develop more

appropriate items for better discriminant validity of each dimension Third the content of

APP dimension was assumed to be distinct from the CPV dimension that focuses on personal

disposition to pursue public interest and public values However the public interest sub-

dimension was separated from the public values sub-dimension in the CPV dimension and

combined with the APP dimension to form the APS dimension Further research should

analyze the CPV dimension and investigate the appropriateness of the APS dimension

In order to facilitate research internationally the PSM construct needs to be

improved conceptually and operationally Kim and Vandenabeele (2010) have dealt with the

conceptual composition and operational dimension of PSM as well as the measurement

model of PSM The next step is to develop an internationally robust measurement instrument

of PSM This study proposes that scholars use all four of the dimensions in measuring the

conceptual components of PSM such as self-sacrifice instrumental value-based and affective

motives because they collectively represent PSM It provides a set of appropriate items for

measuring PSM and confirms the dimensional structure of PSM The survey results suggest

that people in different cultures understand PSM in the same fashion We know that this study

is simply a step along the way to developing a more robust measure but we hope it

contributes to enhanced cross-national research and comparison along with the generation of

cumulative knowledge internationally

- 26 -

REFERENCES

Anderson James C and David D Gerbing 1988 Structural Equation Modeling in Practice

A Review and Recommended Two-Step Approach Psychological Bulletin 103

411-423

Arbuckle James L 2009 Amos 180 Userrsquos Guide Crawfordville FL Amos Development

Co

Beck Joslashrgensen Torben and Barry Bozeman 2007 Public Values An Inventory

Administration amp Society 39(3) 354-381

Beutel Ann M and Margaret Mooney Marini 1995 Gender and values American

Sociological Review 60 436 - 448

Bobko Philip Philip L Roth and Maury A Buster 2007 The Usefulness of Unit Weights in

Creating Composite Scores A Literature Review Application to Content Validity and

Meta-Analysis Organizational Research Methods 10(4) 689-709

Bollen Kenneth A 1989 Structural Equations with Latent Variables New York Wiley

Bollen K A and R Lennox 1991 Conventional Wisdom on Measurement A Structural

Equation Perspective Psychological Bulletin 110(2) 305-314

Byrne Barbara M 2010 Structural Equation Modeling with Amos Basic Concepts

Applications and Programming 2nd

ed New York Routledge

Camilleri Emanuel 2006 Towards Developing an Organizational Commitment ― Public

Service Motivation Model for Maltese Public Service Employees Public Policy

and Administration 21 63-83

Castaing Seacutebastien 2006 The Effects of Psychological Contract Fulfillment and Public

Service Motivation on Organizational Commitment in the French Civil Service

Public Policy and Administration 21 84-98

Cenfetelli Ronald T and Geneviegraveve Bassellier 2009 Interpretation of Formative

Measurement in Information Systems Research MIS Quarterly 33(4) 689-707

Charles Michael B W Martin de Jong and Neal Ryan 2011 Public Values in Western

- 27 -

Europe A Temporal Perspective American Review of Public Administration 41

75-91

Chin Wynne W 2010 How to Write Up and Report PLS Analysis In Handbook of Partical

Least Squares edited by V Esposito Vinzi et al 655-690 Berlin Springer-Verlag

Coursey David H and Sanjay K Pandey 2007a Public Service Motivation Measurement

Testing an Abridged Version of Perryrsquos Proposed Scale Administration amp Society

39 547-568

Coursey David H and Sanjay K Pandey 2007b Content Domain Measurement and

Validity of the Red Tape Concept American Review of Public Administration 37

342-361

Coursey David H James L Perry Jeffrey L Brudney and Laura Littlepage 2008

Psychometric Verification of Perryrsquos Public Service Motivation Instrument Results

for Volunteer Exemplars Review of Public Personnel Administration 28 79-90

Davis Paul and Karen West 2009 What Do Public Values Mean for Public Action Putting

Public Values in Their Plural Place American Review of Public Administration 39

602-618

DeHart-Davis Leisha Justin Marlowe and Sanjay K Pandey 2006 Gender Dimensions of

Public Service Motivation Public Administration Review 66 873-887

Diamantopoulos Adamantios 2006 The error term in formative measurement models

Interpretation and modeling implications Journal of Modeling in Management 1(1)

7-17

Diamantopoulos Adamantios and Judy A Siguaw 2006 Formative versus Reflective

Indicators in Organizational Measure Development A Comparison and Empirical

Illustration British Journal of Management 17 263-282

Diamantopoulos Adamantios and Heidi M Winklhofer 2001 Index Construction with

Formative Indicators An Alternative to Scale Development Journal of Marketing

Research 38 269-277

Edwards Jeffrey R 2010 The Fallacy of Formative Measurement Organizational Research

- 28 -

Methods Online First published on August 19 2010 as

doi1011771094428110378369

Edwards Jeffrey R and Richard P Bagozzi 2000 On the Nature and Direction of

Relationships between Constructs and Measures Psychological Methods 5 155-

174

Flora David B and Patrick J Curran 2004 An Empirical Evaluation of Alternative

Methods of Estimation for Confirmatory Factor Analysis with Ordinal Data

Psychological Methods 9 466-491

Fornell Claes and Fred L Bookstein 1982 Two structural equations models LISREL and

PLS applied to customer exit-voice theory Journal of Marketing Research 19 440-

452

Giauque David Adrian Ritz Freacutedeacuteric Varone Simon Anderfuhren-Biget and Christian

Waldner 2011 Putting Public Service Motivation into Context A Balance between

Universalism and Particularism Forthcoming in International Review of

Administrative Sciences 77(2)

Goumltz Oliver Kerstin Liehr-Gobbers and Manfred Krafft 2010 Evaluation of Structural

Equation Models Using the Partial Least Squares (PLS) Approach In V Esposito

Vinzi et al (eds) Handbook of Partial Least Squares 691-711 DOI 101007978-3-

540-32827-8_30 Springer-Verlag Berlin Heidelberg

Guinot Christiane Julie Latreille and Michel Tenenhaus 2001 PLS Path Modelling and

Multiple Table Analysis Application to the Cosmetic Habits of Women in Ile-de-

France Chemometrics and Intelligent Laboratory Systems 58 247-259

Hair Joseph F Jr William C Black Barry J Babin and Rolph E Anderson 2010

Multivariate Data Analysis 7th

ed Upper Saddle River NJ Pearson Education

Hardin Andrew M Jerry Cha-Jan Chang Mark A Fuller and Gholamreza Torkzadeh 2010

Formative Measurement and Academic Research In Search of Measurement

Theory Educational and Psychological Measurement Online First published on

June 11 2010 as doi101177001316440370208

- 29 -

Howell Roy D Einar Breivik and James B Wilcox 2007a Reconsidering Formative

Measurement Psychological Methods 12 205-218

Howell Roy D Einar Breivik and James B Wilcox 2007b Is Formative Measurement

Really Measurement Reply to Bollen (2007) and Bagozzi (2007) Psychological

Methods 12 238-245

Jarvis Cheryl Burke Scott B MacKenzie and Philip M Podsakoff 2003 A Critical Review

of Construct Indicators and Measurement Model Misspecification in Marketing and

Consumer Research Journal of Consumer Research 30 199-218

Joumlreskog Karl G and Dag Soumlrbom 2005 LISREL 872 (Computer Software) Lincolnwood

IL Scientific Software International

Kim Gimun Bongsik Shin and Varun Grover 2010 Investigating Two Contradictory Views

of Formative Measurement in Information Systems Research MIS Quarterly 34

345-365

Kim Sangmook 2009a Revising Perryrsquos Measurement Scale of Public Service Motivation

American Review of Public Administration 39 149-163

Kim Sangmook 2009b Testing the Structure of Public Service Motivation in Korea A

Research Note Journal of Public Administration Research and Theory 19 839-

851

Kim Sangmook 2010 Testing a Revised Measurement of Public Service Motivation

Reflective versus Formative Specification Journal of Public Administration

Research and Theory Doi101093jopartmuq048

Kim Sangmook and Wouter Vandenabeele 2010 A strategy for building public service

motivation research internationally Public Administration Review 70 701-709

Kline Rex B 2005 Principles and Practice of Structural Equation Modeling 2nd

ed New

York Guilford Press

Law Kenneth and Chi-Sum Wong 1999 Multidimensional Constructs in Structural

Equation Analysis An Illustration using the Job Perception and Job Satisfaction

Constructs Journal of Management 25 143-160

- 30 -

Leisink Peter and Bram Steijn 2009 Public Service Motivation and Job Performance of

Public Sector Employees in the Netherlands International Review of

Administrative Sciences 75 35-52

Liu Bangcheng Chun Hui Jin Hu Wensheng Yang and Xinli Yu 2011 How well can

Public Service Motivation Connect with Occupational Intention International

Review of Administrative Sciences 77 191-211

Liu Bangcheng Ningyu Tang and Xiaomei Zhu 2008 Public Service Motivation and Job

Satisfaction in China An Investigation of Generalisability and Instrumentality

International Journal of Manpower 29 684-699

MacCallum Robert C and Michael W Browne 1993 The Use of Causal Indicators in

Covariance Structure Models Some Practical Issues Psychological Bulletin 114

533-541

MacKenzie Scott B 2003 The dangers of poor construct conceptualization Journal of the

Academy of Marketing Science 31(3) 323-326

MacKenzie Scott B Philip M Podsakoff and Cheryl Burke Jarvis 2005 The Problem of

Measurement Model Misspecification in Behavioral and Organizational Research

and Some Recommended Solutions Journal of Applied Psychology 90 710-730

Martinsen Dorte Sindbjerg and Torben Beck Joslashrgensen 2010 Accountability as a

Differentiated Value in Supranational Governance American Review of Public

Administration 40 742-760

Moynihan Donald P and Sanjay K Pandey 2007a The Role of Organizations in Fostering

Public Service Motivation Public Administration Review 67 40-53

Moynihan Donald P and Sanjay K Pandey 2007b Finding Workable Levers over Work

Motivation Comparing Job Satisfaction Job Involvement and Organizational

Commitment Administration amp Society 39 803-832

Pandey Sanjay K and Edmund C Stazyk 2008 Antecedents and Correlates of Public

Service Motivation In Motivation in Public Management edited by James L Perry

and Annie Hondeghem 101-117 Oxford Oxford University Press

- 31 -

Perry James L 1996 Measuring Public Service Motivation An Assessment of Construct

Reliability and Validity Journal of Public Administration Research and Theory 6

5-22

Perry James L 1997 Antecedents of Public Service Motivation Journal of Public

Administration Research and Theory 7 181-197

Perry James L 2000 Bringing Society in Toward a Theory of Public Service Motivation

Journal of Public Administration Research and Theory 10 471-488

Perry James L and Annie Hondeghem 2008a Motivation in Public Management Oxford

Oxford University Press

Perry James L and Annie Hondeghem 2008b Directions for Future Theory and Research

In Motivation in Public Management edited by James L Perry and Annie

Hondeghem 294-313 Oxford Oxford University Press

Perry James L Annie Hondeghem and Lois Recascino Wise 2010 Revisiting the

Motivational Bases of Public Service Twenty Years of Research and an Agenda for

the Future Public Administration Review 70(5) 681-690

Perry James L and Wouter Vandenabeele 2008 Behavioral dynamics Institutions

Identities and Self-Regulation In Motivation in Public Management edited by

James L Perry and Annie Hondeghem 56-79 Oxford Oxford University Press

Perry James L and Lois Recascino Wise 1990 The Motivational Bases of Public Service

Public Administration Review 50 367-373

Rainey Hal G 1982 Reward Preference among Public and Private Managers In Search of

the Service Ethic American Review of Public Administration 16 288-302

Ringle Christian M Oliver Goumltz Martin Wetzels and Bradley Wilson 2009 On the Use of

Formative Measurement Specifications in Structural Equation Modeling A Monte

Carlo Simulation Study to Compare Covariance-Based and Partial Least Squares

Model Estimation Methodologies Faculty of Economics and Business Administration

Maastricht Univeristy

Ringle C M S Wende and S Will 2005 Smart PLS 20 (M3) Beta

- 32 -

httpwwwsmartplsde

Ritz Adrian 2011 Attraction to Public Policy Making A Qualitative Inquiry into

Improvements in PSM Measurement Forthcoming in Public Administration

Spector Paul E 1992 Summated Rating Scales Construction An Introduction Newbury

Park CA Sage

Steenbergen Marco R 1996 Compassion and American public opinion An analysis of the

NES Humanitarianism Scale Report to the Board of Overseers of the National

Election Studies

Steffen Patrick R and Kevin S Masters 2005 Does compassion mediate the intrinsic

religion-health relationship Annals of Behavioral Medicine 30 217-223

Taylor Jeannette 2007 The Impact of Public Service Motives on Work Outcomes in

Australia A Comparative Multi-Dimensional Analysis Public Administration 85

931-959

Vandenabeele Wouter 2007 Towards a Theory of Public Service Motivation An

Institutional Approach Public Management Review 9 545-556

Vandenabeele Wouter 2008a Government Calling Public Service Motivation as an Element

in Selecting Government as an Employer of Choice Public Administration 86

1089-1105

Vandenabeele Wouter 2008b Development of a Public Service Motivation Measurement

Scale Corroborating and Extending Perryrsquos Measurement Instrument International

Public Management Journal 11 143-167

Vandenabeele Wouter 2009 The Mediating Effect of Job Satisfaction and Organizational

Commitment on Self-Reported Performance More Robust Evidence of the PSM-

Performance Relationship International Review of Administrative Sciences 75 11-

34

Wilcox James B Roy D Howell and Einar Breivik 2008 Questions about Formative

Measurement Journal of Business Research 61 1219-1228

Wise Lois Recascino 2000 The Public Service Culture In Public Administration Concepts

- 33 -

and Cases 7th ed edited by Richard J Stillman II 342-353 Boston Houghton

Mifflin

Wright Bradley E 2008 Methodological Challenges associated with Public Service

Motivation Research In Motivation in Public Management edited by James L

Perry and Annie Hondeghem 80-98 Oxford Oxford University Press

Wright Bradley E and Sanjay K Pandey 2008 Public Service Motivation and the

Assumption of Person-Organization Fit Testing the Mediating Effect of Value

Congruence Administration amp Society 40 502-521

- 34 -

Table 1

Possible Indicators of PSM and Descriptive Statistics (n = 2868)

Dimensions and items Mean Standard

Deviation

Item-Total

Correlation

Attraction to Public Participation (APP)

APP1 I am interested in helping to improve public service

APP2 I am satisfied when I see people benefiting from the public programs I was

involved in

APP3 I like to discuss topics regarding public programs and policies with others

APP4 I believe that public sector activities contribute to our general welfare

APP5 I admire people who initiate or are involved in activities to aid my

community

APP6 Contributing to public programs and policies helps me realize myself

APP7 It is important to contribute to activities that tackle social problems

412

441

355

393

423

351

409

696

662

917

753

672

881

734

524

421

482

412

493

478

552

Commitment to Public Values (CPV)

Sub-dimension for Public Interests (CPV1)

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

CPI3 I would prefer seeing public officials do what is best for the whole

community even if it harmed my interests

CPI4 Serving the public interest is more important than helping a single individual

Sub-dimension for Public Values (CPV2)

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of public

services

CPV3 It is fundamental that public services respond to the needs of the citizens

CPV4 Decisions regarding public services should be democratic despite the time

and effort it takes

CPV5 Everybody is entitled to a good service even if it costs a lot of money

CPV6 It is fundamental that the interests of future generations are taken into

account when developing public policies

CPV7 To act ethically is essential for public servants

CPV8 I believe that public employees must always be aware of the legitimacy of

their activities

CPV9 I personally identify with the aim of protecting individual liberties and rights

418

409

348

337

428

416

433

394

367

426

441

434

414

695

662

899

995

730

704

710

852

932

689

681

692

763

525

596

480

251

484

437

356

325

376

439

462

402

392

Compassion (COM) COM1 It is difficult for me to contain my feelings when I see people in distress

COM2 I feel sympathetic to the plight of the underprivileged

COM3 I empathize with other people who face difficulties

COM4 I have little compassion for people in need who are unwilling to take the

first step to help themselves

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

354

387

388

299

412

391

951

825

707

1083

755

765

399

511

389

011

487

502

Self-Sacrifice (SS)

SS1 Making a difference to society means more to me than personal achievements

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS5 People should give back to society more than they get from it

SS6 Serving other citizens would give me a good feeling even if no one paid me for

it

SS7 I would agree to a good plan to make a better life for the poor even if it costs

me money

357

329

339

302

345

363

350

904

872

949

982

888

1049

895

508

584

454

531

412

368

546

- 35 -

Table 2

Background of Respondents (n = 2868)

Variables Characteristics Respondents

Sex

Male

Female

NA

1209

1622

37

422

566

13

Length of service

(years)

0 ~ 10

10 ~ 20

20 ~ 30

30+

NA

1126

723

484

491

44

393

252

169

171

15

Organizational status

―Do you supervise employees

No

Yes

NA

1472

1076

320

513

375

112

Country

Australia

Belgium

China

Denmark

France

Italy

Korea

Lithuania

The Netherlands

Switzerland

United Kingdom

USA

249

214

230

249

266

162

253

236

249

250

260

250

87

75

80

87

93

56

88

82

87

87

91

87 Note NA = no answer

- 36 -

Table 3

Standardized factor loadings and correlations for the 31-item PSM index (n = 2868)

Dimensions and items SFL (MLE) SFL (D-WLS) Alpha

APP

APP1

APP2

APP4

APP5

APP6

APP7

597

486

455

558

490

600

657

552

514

625

549

665

699

CPV1

CPI1

CPI2

CPI3

CPI4

588

660

488

289

653

732

535

294

542

CPV2

CPV1

CPV2

CPV3

CPV4

CPV5

CPV6

CPV7

CPV8

CPV9

611

546

436

393

418

534

598

496

460

690

631

505

457

486

601

669

573

530

745

COM

COM1

COM2

COM3

COM5

COM6

496

569

498

597

607

516

651

538

670

684

687

SS

SS1

SS2

SS3

SS4

SS5

SS6

SS7

567

736

558

728

467

409

663

660

762

591

695

515

494

709

782

Measures of fit (MLE) χsup2 (df = 424) = 43699 p lt 001 CFI = 835 GFI = 899

RMSEA = 057 [055 059]

Measure of fit (D-WLS) χsup2 (df = 424) = 39247 p lt 001 CFI = 971 GFI = 976

RMSEA = 054 [052 055]

Inter-factor correlations MLE (D-WLS)

1 2 3 4

1 APS

2 CPV1 996 (988)

3 CPV2 902 (890) 798 (798)

4 COM 787 (783) 758 (762) 779 (783)

5 SS 700 (742) 800 (859) 509 (537) 720 (729)

Note SFL = standardized factor loading All factor loadings and correlations are significant at p lt 001

- 37 -

Table 4

Standardized factor loadings and correlations for the 16-item PSM index (n = 2868)

Dimensions and items SFL

(MLE)

SFL

(D-WLS)

APS

APP5 I admire people who initiate or are involved in activities to aid my

community

APP7 It is important to contribute to activities that tackle social problems

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

611

592

637

666

665

669

690

751

CPV

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of

public services

CPV6 It is fundamental that the interests of future generations are taken

into account when developing public policies

CPV7 To act ethically is essential for public servants

612

546

551

590

698

631

618

658

COM

COM1 It is difficult for me to contain my feelings when I see people in

distress

COM2 I feel sympathetic to the plight of the underprivileged

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

490

546

596

604

515

608

669

679

SS

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS7 I would agree to a good plan to make a better life for the poor even if

it costs me money

763

571

766

678

807

629

747

755

Measures of fit (MLE)

χsup2 (df = 98) = 8490 p lt 001

CFI = 939 GFI = 965

RMSEA = 052 [049 055]

Measure of fit (D-WLS)

χsup2 (df = 98) = 6069 p lt 001

CFI = 988 GFI = 992

RMSEA = 043 [039 046]

Alpha CR (D-WLS) Inter-factor correlations MLE (D-WLS)

1 APS 717 788 1 APS 2 CPV 3 COM

2 CPV 663 747 848 (859)

3 COM 645 713 781 (791) 789 (795)

4 SS 784 826 565 (596) 462 (505) 715 (726)

Note SFL = standardized factor loading Alpha = Cronbachrsquos alpha CR = composite reliability

All factor loadings and correlations are significant at p lt 001

- 38 -

Table 5

Results of CFA in the individual countries Standardized factor loadings and measures of fit

Factors

amp Items

Australia Belgium China Denmark France Italy

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

672

756

799

785

804

943

710

703

597

790

799

730

584

675

685

681

731

742

578

748

674

643

765

739

645

704

678

708

812

779

632

604

765

766

672

853

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

552

593

560

699

727

632

624

605

560

713

709

645

702

602

616

724

686

735

567

560

715

654

605

811

608

658

587

696

755

698

673

620

523

782

697

654

COM

COM1

COM2

COM3

COM5

COM6

623

584

662

688

701

739

732

608

475

808

666

581

578

616

605

649

670

673

699

751

737

679

816

776

568

546

801

626

673

850

637

546

741

715

601

822

SS

SS2

SS3

SS4

SS7

654

794

806

892

767

813

727

693

799

793

725

838

857

820

773

896

855

842

806

749

792

829

753

884

847

731

753

889

816

753

815

692

664

795

782

734

Inter-factor

correlations

093

~

752

163

~

810

320

~

773

323

~

825

452

~

827

497

~

855

323

~

850

321

~

847

498

~

899

536

~

902

425

~

829

476

~

865

Sample size 249 214 230 249 266 162

Measures of Fit

χsup2 638 444 881 732 765 554 817 625 735 565 750 564

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 980 100 939 987 969 997 967 994 975 997 949 995

GFI 961 986 941 986 948 991 949 992 955 992 926 986

RMSEA 036 00 063 050 051 026 053 035 045 026 059 033

- 39 -

Table 5 (Continued)

Factors

amp Items

Korea Lithuania The Netherlands Switzerland UK USA

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

619

701

667

706

781

717

666

512

682

751

605

779

557

629

630

620

719

729

535

621

702

643

694

735

655

725

738

710

804

822

535

621

702

643

694

735

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

594

607

722

722

644

722

674

535

533

790

589

711

568

554

538

655

628

545

627

580

615

711

666

704

800

512

555

873

581

646

627

580

615

711

666

704

COM

COM1

COM2

COM3

COM5

COM6

694

677

550

763

736

627

557

639

428

550

696

534

737

623

508

766

616

608

659

592

683

703

661

769

682

583

729

739

644

803

659

592

683

703

661

769

SS

SS2

SS3

SS4

SS7

776

752

679

779

770

811

707

721

635

725

734

706

814

656

743

820

767

764

745

632

689

788

662

744

671

818

832

830

864

802

745

632

689

788

662

744

Inter-factor

correlations

432

~

726

466

~

746

376

~

788

406

~

902

377

~

852

455

~

836

637

~

818

660

~

828

317

~

852

360

~

852

637

~

818

660

~

828

Sample size 253 236 249 250 260 250

Measures of Fit

χsup2 896 573 612 462 773 616 815 715 769 542 815 715

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 951 996 975 100 958 993 957 991 972 998 957 991

GFI 942 988 960 989 951 987 948 990 955 990 948 990

RMSEA 059 028 034 00 050 034 053 044 048 022 053 044

Note All factor loadings and correlations are significant at p lt 001

- 40 -

Appendix 1 Dimensions and possible items of PSM (the first round)

1 Attraction to Public Participation

(1) I am interested in developing public programs helping my country or community (Kim 2009a)

(2) Sharing my views on public policies with others is attractive to me (Kim 2009a)

(3) I am satisfied when seeing people get benefits from the public programs I was involved in (Kim 2009a)

(4) I like to discuss political subjects with others (Giauque et al 2011)

(5) I think the governmental activities contribute to our welfare

(6) I like to initiate actions to help out my community

(7) I do not care much about politicians (Perry 1996)

2 Commitment to Public Values

Possible items for public interest

(1A) Meaningful public service is very important to me (Perry 1996)

(2A) It is important for me to contribute for the common good (Giauque et al 2011)

(3A) I recognize myself with the mission of protection of liberties and rights endorsed by the nation-state

(Giauque et al 2011)

(4A) I would prefer seeing public officials do what is best for the whole community even if it harmed my

interests (Perry 1996)

(5A) To me serving the public interest is more important than helping other people (Vandenabeele 2008b)

Possible items for public values

(1B) I recognize myself with the promotion of the equality of chances endorsed by the nation-state (Giauque et

al 2011)

(2B) It is fundamental that the public service provided to the citizen is regular and continuous (Giauque et al

2011)

(3B) It is fundamental that public service innovates to respond to the new needs of the citizensclients (Giauque

et al 2011)

(4B) The dignity and well-being of all should be the most important concerns in any society (Steenbergen

1996)

(5B) It is important that public servants account for all the costsexpenses they make (Vandenabeele 2008b)

(6B) We have to do everything in our power to pursue the goal of democracy

3 Compassion

(1) It is difficult for me to contain my feelings when I see people in distress (Perry 1996)

(2) I feel sympathetic to the plight of the underprivileged (Kim 2009a)

(3) I am often reminded by daily events how dependent we are on one another (Perry 1996)

(4) I feel compassion for others in the difficulties they are facing (Steffen and Masters 2005)

(5) It is not really my problem if others are in trouble and need help (Beutel and Marini 1995)

(6) I find it hard to be sympathetic toward starving people in foreign lands when there is so much trouble in our

own country (Beutel and Marini 1995)

(7) I get very upset when I see other people treated unfairly (Beutel and Marini 1995)

(8) I would agree to a good plan to make a better life for the poor even if it cost me money (Beutel and Marini

1995)

(9) To me patriotism includes seeing the welfare of others (Perry 1996)

(10) I care very much about other people

4 Self-Sacrifice

(1) Making a difference in society means more to me than personal achievements (Perry 1996)

(2) I am prepared to make enormous sacrifices for the good of society (Perry 1996)

(3) I believe in putting duty before self (Perry 1996)

(4) I am one of those rare people who would risk personal loss to help someone else (Perry 1996)

(5) I think people should give back to society more than they get from it (Perry 1996)

(6) Serving other citizens would give me a good feeling even if no one paid me for it (Perry 1996)

(7) Much of what I do is for a cause bigger than myself (Perry 1996)

- 41 -

Appendix 2 Newly suggested items through the first round

1 Items for the Attraction to Public Participation Dimension

(1) In the public domain every person should be counted as equal

(2) I enjoy to think and argue in a political way

(3) I am satisfied when seeing my ideas being integrated in public policies

(4) I can easily relate to the way of thinking and arguing of politicians

(5) Working in the public sector I can better realize myself

2 Items for the Public Values Sub-Dimension

(1) Public service should give value for money

(2) Cost considerations should play an important role in decisions about public service provision

(3) It is fundamental that the interests of future generations are taken into account when developing public

policies

(4) To act with correct ethics is for public servants as important as their professional competence

(5) Primarily public servants should be accountable to the public and not to their superiors

3 Items for the Self-Sacrifice Dimension

(1) I would agree to a good plan to make a better life for the poor even if it costs me money

- 42 -

Appendix 3 An example of a summary of comments and suggestions (the second round)

On the item ―I am interested in developing public programs helping my country or community

A It is a good sentence but is it possible for everyone to ―develop public programs Even public officials are

limited to do this

B OK

C or the special group

D Double barrel and difficult to translate to my country ―I am interested in helping to improve public services

E Good but I feel that wording requires revision If we draw sample from local government then country is

redundant ―I am interested in developing public policy programs that aim in helping my community

F 4 (General appreciation from1 to 4) Not clear whether it refers to ones work or to ones concern as a citizen

―(In my work) I am interested in contributing to the development of public etchellip

G Instead of development I would use implementing

H Delete public which is redundant because of the characteristic helping my country or community but add

or implementing ―I am interested in developing or implementing programs helping my country or

community

I Referring to dimension Wording Similar to item 6

Note Names have been changed to alphabets in order to ensure the anonymity of participants

- 43 -

Appendix 4 Dimensions and Items of Public Service Motivation for Future Research

Dimensions and items

Attraction to Public Service (APS)

1 I am interested in helping to improve public service

2 I admire people who initiate or are involved in activities to aid my community

3 It is important to contribute to activities that tackle social problems

4 Meaningful public service is very important to me

5 It is important for me to contribute to the common good

Commitment to Public Values (CPV)

1 I think equal opportunities for citizens are very important

2 It is important that citizens can rely on the continuous provision of public services

3 It is fundamental that the interests of future generations are taken into account when

developing public policies

4 To act ethically is essential for public servants

5 I believe that public employees must always be aware of the legitimacy of their activities

6 I personally identify with the aim of protecting individual liberties and rights

Compassion (COM)

1 It is difficult for me to contain my feelings when I see people in distress

2 I feel sympathetic to the plight of the underprivileged

3 I empathize with other people who face difficulties

4 I get very upset when I see other people being treated unfairly

5 Considering the welfare of others is very important

Self-Sacrifice (SS)

1 I am prepared to make sacrifices for the good of society

2 I believe in putting civic duty before self

3 I am willing to risk personal loss to help society

4 I would agree to a good plan to make a better life for the poor even if it costs me money

- 3 -

(2010) proposed that PSM should continue to be conceived as a four-dimensional construct

with self-sacrifice as the foundational concept representing the altruistic or pro-social origins

of PSM The three other concepts attraction to public participation commitment to public

values and compassion which represent instrumental value-based and affective motives

round out their proposed modifications to the construct The revisions to the construct of

PSM will enhance cross-national research and generate cumulative knowledge

The second step was to specify a measurement model of PSM This was necessary

because the measurement model specifies the relationship between constructs and measures

(Diamantopoulos and Siguaw 2006) Reflective measures are treated as outcomes of

constructs and formative measures are causes of constructs (Edwards 2010) A measurement

model can either include reflective or formative indicators exclusively or consist of both

depending on the observed construct (Fornell and Bookstein 1982) The decision to

operationalize formative andor reflective indicators should be based on theoretical

considerations Previous studies discussing the conceptual definition of PSM theoretically

and empirically have proposed that the dimensions of PSM are formative and the indicators

of each dimension are reflective (Kim 2010 Kim and Vandenabeele 2010)

The next step is to develop an internationally robust measurement instrument of PSM

for cross-national research and comparison A survey-based measure for PSM has been useful

for facilitating comparisons across disparate services and national settings and creating

foundations for the accumulation of results (Perry Hondeghem and Wise 2010) Perry (1996)

identified a multidimensional 24-item scale to measure PSM Since then many studies have

used the items of Perryrsquos (1996) measure and improvements have been made to the PSM

scale (Coursey and Pandey 2007a Kim 2009a Taylor 2007 Vandenabeele 2008b) However

research has shown that the original items do not fully represent unique and salient qualities

- 4 -

of the dimensions of the PSM construct (Castaing 2006 Coursey and Pandey 2007a Coursey

et al 2008 DeHart-Davis Marlowe and Pandey 2006 Kim 2009b Leisink and Steijn 2009

Liu Tang and Zhu 2008 Moynihan and Pandey 2007b Taylor 2007 Vandenabeele 2008a

Wright and Pandey 2008) Moreover as the construct and operational definition of PSM are

refined it is necessary to adapt items for the dimensions of the PSM construct

This study is focused on developing a revised measurement instrument for PSM for

international use After developing the possible items for measuring PSM the international

survey was conducted in 2010 and the data (n = 2868) were collected from twelve countries

Dimensions and items were tested cross-nationally as this can be a useful tool to preclude

value-specific items and wordings that are case-specific The next section of this paper will

briefly explain the processes for developing the international measurement instrument The

survey results are analyzed and discussed in order to confirm the dimensional structure and

validate the items of PSM Finally the items of PSM are proposed and discussions and

suggestions are given for future research

Developing the Measurement Instrument of PSM for International Use

Background

Perry and Wise (1990) proposed that PSM is associated with three types of motives affective

norm-based and rational Perry (1996) identified four empirical components of the PSM

construct attraction to public policy making commitment to the public interestcivic duty

compassion and self-sacrifice Since then many studies have used the dimensions and items

of Perryrsquos (1996) measure and improvements have been made in the PSM scale (Kim 2009a

Taylor 2007 Vandenabeele 2008b) However Perryrsquos (1996) measure has limitations (Kim

and Vandenabeele 2010) In particular the items of the dimension of attraction to policy

- 5 -

making are not appropriate for measuring personal attraction to public policy making

(Camilleri 2006 Coursey and Pandey 2007a Coursey et al 2008 Kim 2009b Ritz 2011)

Moreover the dimension of commitment to public interest needs to concentrate more on a

personal disposition to pursue public values (Castaing 2006 Leisink and Steijn 2009 Taylor

2007) and the items of compassion do not represent affective motives (DeHart-Davis

Marlowe and Pandey 2006 Moynihan and Pandey 2007a Wright 2008) Thus the

dimensions of the PSM construct are modified as attraction to public participation

commitment to public values compassion and self-sacrifice Consequently it is necessary to

develop appropriate items for the dimensions (Kim and Vandenabeele 2010)

A related issue concerns the measurement model of PSM There are two

measurement models using multiple dimensions of latent constructs reflective and formative

When variation in a construct leads to variation in its dimensions such dimensions are

termed ―reflective while dimensions are labeled ―formative when they are viewed as

causes of a construct (Edwards and Bagozzi 2000 Diamantopoulos and Siguaw 2006

Diamantopoulos and Winklhofer 2001) It is understandable that each dimension of PSM has

several reflective indicators that are essentially interchangeable A problem is the relationship

between PSM and its dimensions If we regard the dimensions as reflective it is possible to

analyze PSM with only two or three dimensions because omitting dimensions does not

change the essential nature of PSM However if we regard the dimensions as formative it is

not acceptable to omit a dimension as this may change the meaning of PSM Construct

conceptualization is regarded as being the process that determines the formative or reflective

nature (Diamantopoulos and Siguaw 2006) Recent research has demonstrated theoretical and

empirical evidence in support of a second-order formative approach to PSM (Kim 2010)

This means that the dimensions represent different conceptual components of PSM and so

- 6 -

they are not interchangeable Thus PSM is regarded as a combination of specific dimensions

If any one of the dimensions changes PSM would change Conversely if an individualrsquos

PSM increases this would not necessarily be accompanied by an increase in all dimensions

Therefore as the PSM construct is refined for international use (Kim and Vandenabeele

2010) scholars must be attentive to identifying and using appropriate measurement models

Developing Possible Items

The conceptual components of PSM need to be empirically estimated through operational

definition Kim and Vandenabeele (2010) proposed that PSM is based on self-sacrifice and

associated with instrumental value-based and identification motives and that the dimensions

of the PSM construct be refined along the lines of attraction to public participation (APP)

commitment to public values (CPV) compassion (COM) and self-sacrifice (SS) As Perry

(1996) identified a multidimensional scale to measure PSM it is conceived as an overall

latent variable with various latent dimensions and is referred to as a second-order factor

model Research has found that the measurement model specifies the causal relationship

between PSM and its dimensions (Edwards and Bagozzi 2000) The direction of the

relationship is either from the PSM construct to the dimensions (second-order reflective

measurement) or from the dimensions to the construct (second-order formative measurement)

A theoretical consideration of the relationships between PSM and its dimensions led

to the proposal that PSM be defined in terms of multiple dimensions (Kim 2010 Kim and

Vandenabeele 2010) The dimensions represent different aspects of PSM with each

dimension capturing a distinct and potentially unique form of PSM Moreover the

dimensions may have different antecedents and consequences as well as different

characteristics and theoretical backgrounds (Moynihan and Pandey 2007a Pandey and

- 7 -

Stazyk 2008 Perry 1996 1997 2000 Perry and Wise 1990 Perry and Vandenabeele 2008

Taylor 2007 Vandenabeele 2007 2008a 2009) An individualrsquos PSM is shaped by the

individualrsquos APP CPV COM and SS Deleting a dimension means omitting a certain part of

PSM Therefore it is reasonable to suggest that the dimensions of PSM are formative and for

PSM to be regarded as a composite of its dimensions

Each dimension of PSM can be measured using reflective indicators that satisfy the

following criteria (Jarvis MacKenzie and Podsakoff 2003) a relative homogeneity and

interchangeability of indicators pertaining to each dimension a high degree of covariation

among indicators of each dimension and the expectation that the indicators of each

dimension are likely to be affected by the same antecedents and have the same consequences

Reflective indicators are essentially interchangeable and so adding or removing indicators

does not change the essential nature of a dimension Thus we consider PSM to be a second-

order formative construct with several dimensions that each has several reflective indicators

(reflective first-order formative second-order)

Based upon the refinement and clarification of the operational dimensions of PSM

(Kim and Vandenabeele 2010) items were developed for the four dimensions The items

selected were a mix of pre-existing items that loaded highly on PSM dimensions in multiple

studies (Giauque et al 2011 Kim 2009a Perry 1996 Vandenabeele 2008b) Where

appropriate items were not available new items were drafted from other previous studies

creating an initial total of 35 proposed items For the dimension of APP which focuses on

individual disposition to work in the public sector to participate in the public policy process

and in activities for community and social development seven possible items were suggested

The CPV dimension focuses on personal disposition to pursue public interest and public

values Two sub-dimensions of CPV seemed plausible one focusing on public interest and

- 8 -

another on public values such as social equity equal opportunity and democracy Thus five

possible items for the public interest sub-dimension (CPV1) and six items for the public

values sub-dimension (CPV2) were developed For the dimension of COM which focuses on

affective bonding with the identified objects such as other members of a social category or of

a political system ten possible items were suggested For the SS dimension Perryrsquos original

items (1996) were the primary source of items The full list of items is provided in Appendix

1

Review Processes

A group of researchers who shared our interest in better measurement of PSM collaborated

using e-mail in November 20091 The objective was to develop a measurement instrument

for PSM with a robust comparative character in other words to create a valid instrument (on

a national basis) that can be used internationally The first round for reviewing the possible

items lasted from the 16th

of November until the 7th

of December 2009 and all the

participants were asked to review these proposed items For this review two criteria were

employed (1) Do these items refer to the dimensions they claim to refer to and (2) Are

these items meaningful across various national contexts and in particular to the situation of

civil servants in that country Participants were also asked to add new items that might better

reflect a particular dimension or to rephrase items that were already on the list

A second round of item refinement lasted from the 9th

of December until the 22nd

of

December 2009 During this time 11 newly suggested items from the first round were

considered (see Appendix 2) A summary of all comments and suggestions emerging from the

1 Emmanuel Camilleri David Giauque Gerhard Hammerschmid Isabel Egger-Peitler Renate Meyer and

Cristian Pliscoff also participated in the review processes

- 9 -

first process was distributed to the participants to stimulate comments in the second round

(see Appendix 3) Based on the comments received in the second round discussion the

possible items for an international survey were selected However some items required

further discussion and so a third round was conducted from the 15th

of January until the 23rd

of January 2010 One issue that remained unresolved in the discussions relates to the

dimension of CPV Some argued that it is reasonable to divide CPV into two sub-dimensions

because committing to public interests is different from pursuing public values and there are

many important public values in democratic countries while others contended that CPV can

cover all important public values including public interest and there is no real benefit to add

one more dimension These differences were left to be resolved through empirical research

This did not affect the final items however and 33 items were confirmed The resulting

survey questionnaire was distributed to all the collaborators on February 9 2010 in order to

conduct a survey in each country

Measures

PSM was measured using the newly developed 33-item index 7 items for the dimension of

APP 13 items for CPV (4 for CPV1 and 9 for CPV2) 6 items for COM and 7 for SS In the

survey questionnaire a five-point Likert type scale (1 = strongly disagree 5 = strongly agree)

was used This scale is among the most widely used response formats in many empirical

investigations within the social and behavioral science in which there are no ―correct or

―incorrect responses (Spector 1992) The items of PSM were randomly distributed in the

survey format Gender age education length of service and employment status were also

measured in this survey

It was necessary to translate the measurement items into the official (native) language

- 10 -

of each country As this raised the possibility of inaccurate translations the back-translation

of the items into native languages was validated by asking two researchers to independently

translate the English-language items The researchers were instructed to consult with each

other over differences between translated items and to test the survey items with some native

respondents

[Table 1 about here]

Samples

The international survey was fielded from March to September 2010 The data upon which

this study are based were collected from civil servants in local governments in Australia

Belgium China Denmark France Italy Korea Lithuania the Netherlands Switzerland the

United Kingdom and the USA The principal criterion was that the samples collected in the

various countries should be comparable to the largest possible extent The criterion ruled out

national and state government as the competencies of national governments and also of state

governments if they exist constitutionally are different among most of the countries included

in the research Therefore we opted for local government as our focus The structure and

functions of local governments can differ widely even within countries and so the survey in

each country was focused on town hall bureaucrats in city county or township government

The respondents were permanent employees from both managerial and non-managerial levels

and from both administrative departments (ie no direct contact with the public) and service

departments (ie likely to have direct contact with the public) Despite being employed by

local governments in many or all of the countries police officers firefighters school teachers

public transport workers artists and musicians as well as nurses and doctors were excluded

- 11 -

from the sample

We aimed to receive 250 respondents for each country as this would enable us to

carry out the analyses we wanted to perform on the data Although there is no ideal sample

size one study recommends a sample of 200 to provide a sound basis for estimation (Hair

Black Babin and Anderson 2010) The method of data collection was left open Obviously a

web-survey is the easiest way to collect the data because it is economical to collect and to

code however compared to paper-pencil surveys this method has some drawbacks First not

everybody in the sample may have access to the internet which may cause a bias in the

responses Second internet surveys are generally known to produce lower response rates

Web surveys were recommended if possible but paper-pencil surveys were deemed

acceptable in cases where it was not possible to conduct a web survey Researchers were

encouraged to contact just one municipality except in cases where it was not possible to

gather enough data In such instances responses from multiple municipalities were accepted

By the end of September 2010 a total of 2868 responses were obtained2

2 In Australia seven local councils each with between from 70 and 232 employees were surveyed Paper-pencil

surveys were used in two councils while the rest relied on a web survey The local governments limited

distribution of paper-pencil surveys to only one or two sections of the council eg administration The response

rate was approximately 30 However it should be noted that not everyone working for the council has direct

access to a computer In Belgium three municipalities were surveyed on a census base by means of a web

survey as all employees had access to a computer This rendered a response of 120 (226) 79 (75) and 16

(53) responses for the respective municipalities resulting in an overall response rate of 331 with 662

invitations being sent out In China two departments social work and administration were surveyed in one

municipality using paper-pencil surveys In the social work department 65 out of 110 surveys were returned

and 165 out of 211 responded from the administration department giving an overall response rate of 767 In

Denmark a total of 1282 respondents in a municipality were sent an initial e-mail invitation to a web survey

and up to two e-mail reminders and the response rate was 473 In France all the employees in two

municipalities were surveyed In one municipality where both web and paper-pencil surveys were utilized 128

out of 685 questionnaires were returned while in another city using only a paper-pencil survey 138 out of 652

were received The average response rate was 199 In Italy in principle the link to the questionnaire was

made available to all employees (about 4000) in one city but for some of them access may have been difficult

yielding a response rate of 45 In Korea 500 employees in a city received a printed questionnaire and 253

responses were returned yielding a response rate of 506 In Lithuania 236 responses from 54 municipalities

were gathered by web survey In the Netherlands three municipalities were surveyed on a census base by means

of a web survey In total 319 responses of 910 invitations were returned (35) mdash 90 (45) 115 (38) and 114

(28) for the respective municipalities In Switzerland 516 out of 1326 local employees with e-mail access in

a municipality completed the survey giving a response rate of 389 In the United Kingdom the survey was

posted online and 1360 employees both in administrative jobs and front-line services of one municipality

- 12 -

Municipalities were selected because of their convenient accessibility and proximity to the

researcher and a convenience sample was used in most countries Seven countries surveyed

one municipality but seven local councils were surveyed in Australia Web surveys were used

in four countries while three countries used both web and paper-pencil survey The number

of respondents varied among the countries depending on the numbers of questionnaires

distributed the survey method and the response rate In most countries we used

approximately 250 responses but only 162 from Italy3 The majority of the respondents were

women (566) while in terms of service length those in the largest group (393) had

worked for fewer than 10 years Table 2 shows the distribution of respondentsrsquo sex length of

service employment status and country

[Table 2 about here]

Analyses

The statistical analysis applied confirmatory factor analysis (CFA) CFA was used to assess

the fit of the data to the hypothesized measurement model because it is preferred where

measurement models have a well-developed underlying theory for hypothesized patterns of

loadings The data were analyzed using Amos 180 (Arbuckle 2009) with the maximum

likelihood (MLE) estimation method and LISREL 872 (Joumlreskog and Soumlrbom 2005) with the

diagonally weighted least squares (D-WLS) estimation method

The individual items are designed to measure a theoretically continuous construct

but the observed responses are a discrete realization of five categories The common method

received an email to complete the survey The response rate was 19 In the USA the total number of

employees surveyed was 626 in one city using both electronic and paper-pencil surveys resulting in a response

rate of 588 3 In countries where significantly more than 250 surveys were collected we randomly selected only 250 cases

for analysis

- 13 -

of estimation within CFA is MLE a technique which assumes that the observed variables are

continuous and normally distributed These assumptions are not met when the observed data

are discrete as occurs when using ordinal scales as in this study A simulation study (Flora

and Curran 2004) found that D-WLS performs well for ordinal measures across a variety of

conditions especially for smaller samples Coursey and Pandey (2007a 2007b) argued the

importance of using D-WLS and not standard MLE in structural equations for ordinal data4

In most cases however the differences between MLE and D-WLS are not substantive to the

conclusions5 No comparison between standard MLE and D-WLS is provided in this paper

We give results from both MLE and D-WLS and this will aid in confirming the stability and

robustness of the results across varying mathematical specifications

For gauging reliability Cronbachrsquos alpha and composite reliability were tested

Composite reliability should be equal to or greater than 07 (Hair et al 2010) Convergent

validity was assessed from the measurement model by determining the statistical significance

of each indicatorrsquos estimated pattern coefficient on its specified underlying construct

Discriminant validity was assessed by determining whether the confidence interval around

the correlation estimate between two factors included 100 (Anderson and Gerbing 1988)

For model fit assessment goodness-of-fit index (GFI) comparative fit index (CFI)

and the root mean square error of approximation (RMSEA) were used The model achieves

an acceptable fit to the data when GFI and CFI equals or exceeds 090 and RMSEA values

4 Applying standard MLE to ordinal data such as Likert-type items may produce significant estimation

problems such as inflation of chi-square fit statistics and biased underestimation of parameters and standard

errors (Flora and Curran 2004) Coursey and Pandey (2007a 2007b) used a modified MLE using the asymptotic

covariance matrix and D-WLS for adjusting the ordinal nature of the data They reported that although the

modified MLE findings are slightly different on a few model parameters the differences are not substantive to

the conclusions However model fit estimators were clearly better for the D-WLS 5 Given normally distributed categorical variables MLE can be used with confidence when a variable has four

or more categories If there is a large sample a five-point Likert scale and the skewness and kurtosis of variables

is within acceptable limits it suggests distribution symmetry ―In most cases where the hypothesized model is

well specified and the scaling based on more than three categories it seems unlikely that there will be much

difference between the findings (Byrne 2010 160)

- 14 -

fall below 008 (Hair et al 2010) In general the larger the value of GFI and CFI and the

smaller the value of RMSEA the better the fit of the model (Bollen 1989) We also present

the χ2 test statistic Lower values of χ

2 indicate a better fit and should be nonsignificant but

the χ2 value increases as sample size increases and thus for large sample sizes this statistic

may lead to rejection of a model with good fit (Hair et al 2010)

Results

Using SPSS 180 descriptive statistics were computed for individual items as shown in table

1 One item (APP3) in the APP dimension ―I like to discuss topics regarding public programs

and policies with others was mistakenly omitted in the survey conducted in the United

States and so it was deleted in the statistical analysis In the assessment of normality no

items showed a skew or kurtosis value greater than the cutoffs of |3| or |8| recommended by

Kline (2005) As data were distributed normally MLE as well as D-WLS estimation was used

Based on inspection of item-total correlations one item (COM4) was dropped from further

analysis because it was weakly correlated with the overall index (011) Missing data is an

issue in CFA as it can cause estimation errors All variables in the survey have missing data

rates of approximately 016 to 143 Mean substitution was used to calculate replacement

values for the missing values in each countryrsquos dataset (Hair et al 2010) The deletion of the

two items left 31 items for the next stage of analysis

The five-correlated factor model was tested using CFA The CFA model in the initial

analysis hypothesized a priori that (1) responses to the 30-item PSM measurement

instrument could be explained by five factors such as APP CPV1 CPV2 COM SS (2) each

item would have a nonzero loading on the PSM factor that it was designed to measure and

- 15 -

zero loadings on all other factors (3) the five factors would be correlated and (4)

measurement error terms would be uncorrelated The resulting CFA with MLE suggested that

the initial five-correlated factor model was not a good fit to the data χsup2 (df = 424) = 43699 p

lt 001 GFI = 0899 CFI = 835 RMSEA = 0057 [055 059] At this stage APP was highly

correlated with CPV1 (r = 996 in MLE and r = 988 in D-WLS) suggesting the two

dimensions were not unique and lacked discriminant validity

[Table 3 about here]

Given the disconfirmation of the initial model an effort was made to estimate an

alternative model The two dimensions APP and CPV1 were combined to form a single

dimension Attraction to Public Service (APS) The items of CPV1 may be regarded as

action-oriented by the respondents because dedication to public service community and

common good is underlined in these items Consequently it is reasonable to combine APP

with CPV1 to form APS a dimension that focuses more on disposition to serve the public to

work for the common good and to participate in public policy processes After the overall

number of dimensions was reduced to four using standardized factor loadings and

modification indices as criteria the items with the lowest factor loading in each dimension

were deleted and the items whose modification index indicated association with multiple

dimensions were eliminated A CFA model with the remaining items was iteratively tested

until it achieved a good fit to the data The outcome of this refinement process was a 16-item

index of four factors Overall model fit by every measure is quite strong for the D-WLS

estimation technique The estimation results for the four-dimension model are presented in

table 4

- 16 -

[Table 4 about here]

Cronbachrsquos coefficient alpha for the 16-item PSM index was 860 and the coefficient

alphas for the four dimensions ranged from 645 to 784 The composite reliability of the set

of reflective indicators for each dimension of PSM ranged from 713 to 826 and indicated

adequate internal consistency The resulting factor structure showed a four-factor structure

with all items loading significantly onto their a priori dimension (p lt 001) and the

standardized factor loadings ranging from 0490 to 0766 in MLE and from 515 to 807 in

D-WLS The results provide support for convergent validity The correlation estimates

between the two factors ranged from 0462 to 0848 in MLE and from 505 to 859 in D-

WLS The confidence intervals (plusmn2 standard errors) around the correlation estimates between

the two factors did not include 100 The result provides support for discriminant validity To

test whether the four-dimension model is superior to other designs such as the two or three-

dimension models a series of CFA were conducted and the results show that the four-

dimension model is a significantly better fit than any of the other designs

To test whether the factorial structure of the PSM instrument is equivalent across the

samples from the twelve countries four nested models were tested as part of multi-group

analysis with MLE (Byrne 2010) First the unconstrained model was the baseline model

which relaxed all equality constraints This model hypothesized the same four-factor structure

of the instrument across twelve countries The fit indices for this baseline model were χsup2 (df =

1176) = 24977 p lt 001 GFI = 0901 CFI = 900 RMSEA = 0020 [019 021] These

indices indicate that the four-factor model of the PSM instrument exhibits acceptable fit

across the samples from the twelve countries Second the measurement weights model tested

- 17 -

the invariance of factor loadings across countries by placing equality constraints on these

parameters The fit indices for this model were χsup2 (df = 1308) = 28088 p lt 001 GFI = 0890

CFI = 887 RMSEA = 0020 [019 021] Since the measurement weights model was nested

within the unconstrained model the chi-square difference text ∆χsup2 (132) = 3111 p lt 001

indicated that some equality constraints of factor loadings did not hold across the twelve

countries Further test results suggested that the equality constraints of structural covariances

and measurement residuals were not upheld across the twelve countries It needs to be

stressed that despite the differences in factor loadings and factor covariances the results

support the main assumption that the same four-factor structure holds across different

countries We may assume that people in different cultures understand PSM in the same

fashion

To find differences in the factor structure of PSM among countries we conducted a

series of CFA with each countryrsquos dataset and followed the same procedure as when

analyzing the sample as a whole As table 5 shows although there are great similarities

between the appropriate items in the countries there are also some differences Even though

the items vary according to national differences the four-factor structure had a good fit to the

data Thus the results confirm that the four-factor structure of PSM can be generalized

[Table 5 about here]

The APS dimension covers both an individualrsquos attraction to public participation and

commitment to public interest but all its items refer to some notion of public interest such as

improving public service aiding community tackling social problems meaningful public

service and contributing to the common good The APS dimension captures an individualrsquos

- 18 -

disposition to serve the public interest through public participation When developing the

items some assumed that the items of committing to public interests can be separated from

those of pursuing public values in the CPV dimension The empirical test showed that

respondents tended to understand committing to public interests as behavior-oriented and so

the items for public interest are not differentiated from participation-oriented items in their

perceptions but are different from those for public values The respondents seem to

understand that pursuing public values is more related to personal beliefs or creeds Thus the

CPV dimension is solely focused on public values and the items of pursing public interests

are combined with those of the APP dimension to make the APS dimension

Public values and their relative importance are neither impervious to change over

time nor the same among all countries (Charles de Jong and Ryan 2011 Davis and West

2009) Moreover some values may be in conflict as Martinsen and Beck Joslashrgensen (2010)

discussed in regard to the tensions between accountability and efficiency It is therefore not

possible to include all public values in one dimension (Beck Joslashrgensen and Bozeman 2007)

Accordingly the items of the CPV dimension are developed for estimating individual

commitment to seek public values which there are general consensus about and which are

generally formulated

When all four dimensions of PSM are included in a survey cross-national

equivalence of the survey can be assured The items represent reflections or manifestations of

each dimension and hence are relatively homogeneous and interchangeable Adding or

deleting items does not change the essential nature of the dimension Thus any combination

of items in a dimension can be regarded as equivalent When using structural equation

modeling (SEM) each dimension needs to have at least three reflective indicators in order to

achieve identification or it needs to have at least two reflective indicators and be allowed to

- 19 -

covary with another dimension that has at least two reflective measures With one indicator

the dimension is identified if the loading and the variance of the uniqueness are both fixed

(Bollen 1989 Edwards 2010) Thus the number of items in each dimension is not restricted

when measuring PSM However we recommend using multiple items for each dimension

because more research is needed to investigate the appropriateness of this PSM instrument

with varying samples

The test results showed that model fit estimators were clearly better for the D-WLS

It means that in this study MLE was more rigorous than D-WLS in assessing the fit of the

model The empirical evidence indicated that MLE may be an acceptable technique for

ordinal data when there is a large sample a five-point Likert scale and the skewness and

kurtosis of variables is within acceptable limits as suggested by Byrne (2010)

Discussion

The analysis reaffirmed the four-dimensional structure of PSM but the results were not

entirely consistent with the original CFA we tested We were ultimately able to converge on a

good-fit operational model for PSM that was also consistent with underlying theory The

analysis is not able to resolve the debate about measurement models however for reasons

discussed below

Choosing the correct measurement model (ie reflective vs formative) is important

as measurement model misspecification results not only in misrepresentation of the epistemic

relationships between constructs and their measures but also in biased parameter estimates of

structural relationships (Jarvis et al 2003 Law and Wong 1999 MacKenzie 2003

MacKenzie Podsakoff and Jarvis 2005) A formative measurement model in isolation is

- 20 -

under-identified and cannot be estimated in the case of covariance-based SEM (CB-SEM)

(Bollen 1989) A construct with formative indicators must emit paths to at least two unrelated

reflective indicators two unrelated latent constructs or some combination of the two in order

to solve the problem of under-identification (Jarvis MacKenzie and Podsakoff 2003

MacCallum and Browne 1993 MacKenzie Podsakoff and Jarvis 2005) Thus ―when

dependent constructs change the empirical nature of the formatively measured construct

changes (Wilcox Howell and Breivik 2008 1223) and ―when specifying formative

measurement models what might seem like an innocuous choice of outcomes to achieve

identification can have fundamental ramifications for the meaning of the formative construct

and the inferences based on it (Edwards 2010 7)

Earlier research found that the association between a construct and its measures

should remain stable regardless of the larger causal model in which the constructs and

measures are embedded (Edwards and Bagozzi 2000) The association between a construct

and its measures is generally stable for reflective measures but associations of formative

measures with their construct are unstable The susceptibility of formative measurement to

both interpretational confounding and external inconsistency recommends that formative

measurement is not an equally attractive alternative to reflective measurement and that

whenever possible researchers should opt for reflective measurement (Howell Breivik and

Wilcox 2007a 2007b Kim Shin and Grover 2010) In the case of second-order models

researchers are limited to only a second-order reflective model in CB-SEM (Chin 2010)

Regardless of the measurement debate and the tradeoffs between reflective or

formative specification this study makes a theoretical argument that PSM is a second-order

formative construct This second-order formative factor model (reflective first-order

formative second-order) is also suggested as an alternative model that captures the

- 21 -

multidimensionality of complex concepts and avoids the shortcomings of first-order

formative measurement (Edwards 2010) However this model itself cannot achieve

identification One possible strategy is to develop direct reflective measures of PSM based on

the view that PSM is a general concept that is not only influenced by its specific dimensions

but also can be assessed using general measures (Edwards 2010 Javis MacKenzie and

Podsakoff 2003)6 However these general measures have not yet been developed for the

PSM construct

Another way around the problem of underidentification of formative models is to

apply the component-based approach known as partial least squares path modeling (PLS-PM)

which can model formative indicators (Fornell and Bookstein 1982 Goumltz Liehr-Gobbers

and Krafft 2010) The CB-SEM and the PLS-PM methods have distinctive statistical

characteristics and selecting an approach to SEM depends on the particular research situation

CB-SEM is the method of choice for theory testing while PLS-PM is appropriate for

prediction-oriented applications (Ringle Goumltz Wetzels and Wilson 2009) One advantage of

the PLS method is that it demands significantly fewer requirements compared to that of

covariance structure analyses but nevertheless delivers consistent estimation results Another

asset of the PLS approach is its ability to deal with formative as well as reflective indicators

even within one structural equation model This indicates that the PLS approach can be

appropriate for explorative analysis of structural equation models (Chin 2010 Goumltz Liehr-

Gobbers and Krafft 2010) A Monte Carlo simulation study to compare CB-SEM and PLS-

PM showed that PLS-PM provides a viable approximation of model parameters when the

prerequisites for CB-SEM are not met (Ringle Goumltz Wetzels and Wilson 2009) A formative

6 For example job satisfaction can be assessed with general reflective measures such as ―in general I am

satisfied with my job and the formative dimensions can be assessed with measures that describe satisfaction

with each job facet (Edwards 2010 15)

- 22 -

composite in the form of indexes needs to be tested to know the relative weights (γ-

parameters) of the dimensions of PSM In PLS-PM a second-order formative factor is

directly measured by observed variables for all the first order factors While this approach

repeats the number of manifest variables used the model can be identified without adding

outcome constructs that would be considered inappropriate or irrelevant from a conceptual

standpoint (Chin 2010 Edwards 2010) The test result of PLS-PM using Smart PLS 20

(Ringle Wende and Will 2005) produced the function PSM = 31APS + 36CPV + 28COM

+ 31SS However one consideration is whether and how construct error is modeled for the

formative construct (Diamantopoulos 2006) When using PLS-PM the researcher is making

the assumption that there is no error at the construct level as all constructs in PLS-PM are

modeled without error When using PLS-PM the absence of an error term will tend to

increase the weights therefore inflating their importance (Bollen and Lennox 1991) The

primary statistic for assessing a formative indicator is its weight the partialized effect of the

indicator on its intended construct controlling for the effects of all other indicators of that

construct (Cenfetelli and Bassellier 2009) A formative measure is essentially a multiple

regression with the construct representing the dependent variable and the indicators as the

predictors (Diamantopoulos and Winklhofer 2001) PSM may be regarded as a weighted

linear composite of its dimensions The weights used to create the composites are different

across CE-SEM and PLS-PM The causal dimensional weights generated in CB-SEM are

estimated in a model in which a construct-level error term is specified (eg PSM = γ1APM +

γ2CPI + γ3COM + γ4SS + ζ) whereas the weights in PLS-PM are not (eg PSM = γ1APM +

γ2CPI + γ3COM + γ4SS) This leads to two different conceptualizations of the PSM construct

one in which the causal indicators do not capture the construct in its entirety and another in

which they do (Hardin et al 2010) These differences need to be discussed in future research

- 23 -

in this area

One practical suggestion to measure PSM is to apply unit weights to its dimensions if

we assume the effects of the four dimensions are equivalent A literature review indicates that

unit weights have substantial predictive validity when compared with regression weights The

literature shows that the empirical correlations between scores generated from differential and

unit-weighting approaches are quite high In creating composite scores unit weights can be

an appropriate approach for weighting under many circumstances (Bobko Roth and Buster

2007) Although the uses of unit-weight composites will likely result in the explanation of

less variance than empirical estimations of dimensional weights they will remain consistently

defined across studies and hence suitable for theory testing Alternatively more thorough

discussions on the dimensions of PSM may suggest fixed-weight composites (Hardin et al

2010) A previous study (Kim 2010 22) also suggested ―Overall indexing PSM by

measuring all the dimensions of PSM and simply summing the averages of each dimension

can make it easier to measure PSM in practice as PSM = APS + CPV + COM + SS

An advantage of regarding PSM as a second-order formative construct is that the

dimensions can be analyzed both individually and collectively This is important as the

dimensions are expected to have different antecedents and consequences For instance Perry

(1997) found some significant differences in the influence of independent variables on the

four dimensions of PSM Taylor (2007) confirmed that certain dimensions are more important

than others in influencing work outcomes Moynihan and Pandey (2007a) reported that red

tape is negatively related to one dimension but reform orientation and hierarchical authority

is positively related to another dimension A recent study found that the SS dimension has a

positive relationship with occupational intention but the other dimensions do not (Liu et al

2011) Thus it is valuable to analyze the causes and effects of individual dimensions of PSM

- 24 -

as well as of the PSM construct

Conclusion

The purpose of this study is to develop a measurement instrument for PSM for cross-national

research and comparison On the basis of the sharpened construct of PSM (Kim and

Vandenabeele 2010) using data from an international survey conducted in 12 countries we

found that the four-factor structure of PSM is generalizable across country cases This study

provides the measurement index of PSM with the four dimensions of APS CPV COM and

SS The APS dimension measures the instrumental motives of PSM the CPV dimension

determines the value-based motives of PSM while the COM and SS dimensions gauge the

identification motives and self-sacrifice respectively When all four dimensions of PSM are

used in a survey cross-national equivalence of survey items can be assured The items for

measuring PSM in future research are suggested in Appendix 4

This study has several limitations First looking for samples which were comparable

to the largest possible extent we opted for surveys in local government The results found

and discussed in this study are based on PSM among local government employees The

results might be different when surveying central government employees or other specific

groups in the public sector Thus the potential impact of the sample bias needs to be

considered in understanding this study It also means that the survey items for measuring

PSM should be examined with different samples in various areas and at different government

levels

Second the test results showed that there is a high degree of correlation among the

dimension such as APS CPV and COM Future research should not only investigate whether

- 25 -

the items reflect the core content of the dimensions but also continue to develop more

appropriate items for better discriminant validity of each dimension Third the content of

APP dimension was assumed to be distinct from the CPV dimension that focuses on personal

disposition to pursue public interest and public values However the public interest sub-

dimension was separated from the public values sub-dimension in the CPV dimension and

combined with the APP dimension to form the APS dimension Further research should

analyze the CPV dimension and investigate the appropriateness of the APS dimension

In order to facilitate research internationally the PSM construct needs to be

improved conceptually and operationally Kim and Vandenabeele (2010) have dealt with the

conceptual composition and operational dimension of PSM as well as the measurement

model of PSM The next step is to develop an internationally robust measurement instrument

of PSM This study proposes that scholars use all four of the dimensions in measuring the

conceptual components of PSM such as self-sacrifice instrumental value-based and affective

motives because they collectively represent PSM It provides a set of appropriate items for

measuring PSM and confirms the dimensional structure of PSM The survey results suggest

that people in different cultures understand PSM in the same fashion We know that this study

is simply a step along the way to developing a more robust measure but we hope it

contributes to enhanced cross-national research and comparison along with the generation of

cumulative knowledge internationally

- 26 -

REFERENCES

Anderson James C and David D Gerbing 1988 Structural Equation Modeling in Practice

A Review and Recommended Two-Step Approach Psychological Bulletin 103

411-423

Arbuckle James L 2009 Amos 180 Userrsquos Guide Crawfordville FL Amos Development

Co

Beck Joslashrgensen Torben and Barry Bozeman 2007 Public Values An Inventory

Administration amp Society 39(3) 354-381

Beutel Ann M and Margaret Mooney Marini 1995 Gender and values American

Sociological Review 60 436 - 448

Bobko Philip Philip L Roth and Maury A Buster 2007 The Usefulness of Unit Weights in

Creating Composite Scores A Literature Review Application to Content Validity and

Meta-Analysis Organizational Research Methods 10(4) 689-709

Bollen Kenneth A 1989 Structural Equations with Latent Variables New York Wiley

Bollen K A and R Lennox 1991 Conventional Wisdom on Measurement A Structural

Equation Perspective Psychological Bulletin 110(2) 305-314

Byrne Barbara M 2010 Structural Equation Modeling with Amos Basic Concepts

Applications and Programming 2nd

ed New York Routledge

Camilleri Emanuel 2006 Towards Developing an Organizational Commitment ― Public

Service Motivation Model for Maltese Public Service Employees Public Policy

and Administration 21 63-83

Castaing Seacutebastien 2006 The Effects of Psychological Contract Fulfillment and Public

Service Motivation on Organizational Commitment in the French Civil Service

Public Policy and Administration 21 84-98

Cenfetelli Ronald T and Geneviegraveve Bassellier 2009 Interpretation of Formative

Measurement in Information Systems Research MIS Quarterly 33(4) 689-707

Charles Michael B W Martin de Jong and Neal Ryan 2011 Public Values in Western

- 27 -

Europe A Temporal Perspective American Review of Public Administration 41

75-91

Chin Wynne W 2010 How to Write Up and Report PLS Analysis In Handbook of Partical

Least Squares edited by V Esposito Vinzi et al 655-690 Berlin Springer-Verlag

Coursey David H and Sanjay K Pandey 2007a Public Service Motivation Measurement

Testing an Abridged Version of Perryrsquos Proposed Scale Administration amp Society

39 547-568

Coursey David H and Sanjay K Pandey 2007b Content Domain Measurement and

Validity of the Red Tape Concept American Review of Public Administration 37

342-361

Coursey David H James L Perry Jeffrey L Brudney and Laura Littlepage 2008

Psychometric Verification of Perryrsquos Public Service Motivation Instrument Results

for Volunteer Exemplars Review of Public Personnel Administration 28 79-90

Davis Paul and Karen West 2009 What Do Public Values Mean for Public Action Putting

Public Values in Their Plural Place American Review of Public Administration 39

602-618

DeHart-Davis Leisha Justin Marlowe and Sanjay K Pandey 2006 Gender Dimensions of

Public Service Motivation Public Administration Review 66 873-887

Diamantopoulos Adamantios 2006 The error term in formative measurement models

Interpretation and modeling implications Journal of Modeling in Management 1(1)

7-17

Diamantopoulos Adamantios and Judy A Siguaw 2006 Formative versus Reflective

Indicators in Organizational Measure Development A Comparison and Empirical

Illustration British Journal of Management 17 263-282

Diamantopoulos Adamantios and Heidi M Winklhofer 2001 Index Construction with

Formative Indicators An Alternative to Scale Development Journal of Marketing

Research 38 269-277

Edwards Jeffrey R 2010 The Fallacy of Formative Measurement Organizational Research

- 28 -

Methods Online First published on August 19 2010 as

doi1011771094428110378369

Edwards Jeffrey R and Richard P Bagozzi 2000 On the Nature and Direction of

Relationships between Constructs and Measures Psychological Methods 5 155-

174

Flora David B and Patrick J Curran 2004 An Empirical Evaluation of Alternative

Methods of Estimation for Confirmatory Factor Analysis with Ordinal Data

Psychological Methods 9 466-491

Fornell Claes and Fred L Bookstein 1982 Two structural equations models LISREL and

PLS applied to customer exit-voice theory Journal of Marketing Research 19 440-

452

Giauque David Adrian Ritz Freacutedeacuteric Varone Simon Anderfuhren-Biget and Christian

Waldner 2011 Putting Public Service Motivation into Context A Balance between

Universalism and Particularism Forthcoming in International Review of

Administrative Sciences 77(2)

Goumltz Oliver Kerstin Liehr-Gobbers and Manfred Krafft 2010 Evaluation of Structural

Equation Models Using the Partial Least Squares (PLS) Approach In V Esposito

Vinzi et al (eds) Handbook of Partial Least Squares 691-711 DOI 101007978-3-

540-32827-8_30 Springer-Verlag Berlin Heidelberg

Guinot Christiane Julie Latreille and Michel Tenenhaus 2001 PLS Path Modelling and

Multiple Table Analysis Application to the Cosmetic Habits of Women in Ile-de-

France Chemometrics and Intelligent Laboratory Systems 58 247-259

Hair Joseph F Jr William C Black Barry J Babin and Rolph E Anderson 2010

Multivariate Data Analysis 7th

ed Upper Saddle River NJ Pearson Education

Hardin Andrew M Jerry Cha-Jan Chang Mark A Fuller and Gholamreza Torkzadeh 2010

Formative Measurement and Academic Research In Search of Measurement

Theory Educational and Psychological Measurement Online First published on

June 11 2010 as doi101177001316440370208

- 29 -

Howell Roy D Einar Breivik and James B Wilcox 2007a Reconsidering Formative

Measurement Psychological Methods 12 205-218

Howell Roy D Einar Breivik and James B Wilcox 2007b Is Formative Measurement

Really Measurement Reply to Bollen (2007) and Bagozzi (2007) Psychological

Methods 12 238-245

Jarvis Cheryl Burke Scott B MacKenzie and Philip M Podsakoff 2003 A Critical Review

of Construct Indicators and Measurement Model Misspecification in Marketing and

Consumer Research Journal of Consumer Research 30 199-218

Joumlreskog Karl G and Dag Soumlrbom 2005 LISREL 872 (Computer Software) Lincolnwood

IL Scientific Software International

Kim Gimun Bongsik Shin and Varun Grover 2010 Investigating Two Contradictory Views

of Formative Measurement in Information Systems Research MIS Quarterly 34

345-365

Kim Sangmook 2009a Revising Perryrsquos Measurement Scale of Public Service Motivation

American Review of Public Administration 39 149-163

Kim Sangmook 2009b Testing the Structure of Public Service Motivation in Korea A

Research Note Journal of Public Administration Research and Theory 19 839-

851

Kim Sangmook 2010 Testing a Revised Measurement of Public Service Motivation

Reflective versus Formative Specification Journal of Public Administration

Research and Theory Doi101093jopartmuq048

Kim Sangmook and Wouter Vandenabeele 2010 A strategy for building public service

motivation research internationally Public Administration Review 70 701-709

Kline Rex B 2005 Principles and Practice of Structural Equation Modeling 2nd

ed New

York Guilford Press

Law Kenneth and Chi-Sum Wong 1999 Multidimensional Constructs in Structural

Equation Analysis An Illustration using the Job Perception and Job Satisfaction

Constructs Journal of Management 25 143-160

- 30 -

Leisink Peter and Bram Steijn 2009 Public Service Motivation and Job Performance of

Public Sector Employees in the Netherlands International Review of

Administrative Sciences 75 35-52

Liu Bangcheng Chun Hui Jin Hu Wensheng Yang and Xinli Yu 2011 How well can

Public Service Motivation Connect with Occupational Intention International

Review of Administrative Sciences 77 191-211

Liu Bangcheng Ningyu Tang and Xiaomei Zhu 2008 Public Service Motivation and Job

Satisfaction in China An Investigation of Generalisability and Instrumentality

International Journal of Manpower 29 684-699

MacCallum Robert C and Michael W Browne 1993 The Use of Causal Indicators in

Covariance Structure Models Some Practical Issues Psychological Bulletin 114

533-541

MacKenzie Scott B 2003 The dangers of poor construct conceptualization Journal of the

Academy of Marketing Science 31(3) 323-326

MacKenzie Scott B Philip M Podsakoff and Cheryl Burke Jarvis 2005 The Problem of

Measurement Model Misspecification in Behavioral and Organizational Research

and Some Recommended Solutions Journal of Applied Psychology 90 710-730

Martinsen Dorte Sindbjerg and Torben Beck Joslashrgensen 2010 Accountability as a

Differentiated Value in Supranational Governance American Review of Public

Administration 40 742-760

Moynihan Donald P and Sanjay K Pandey 2007a The Role of Organizations in Fostering

Public Service Motivation Public Administration Review 67 40-53

Moynihan Donald P and Sanjay K Pandey 2007b Finding Workable Levers over Work

Motivation Comparing Job Satisfaction Job Involvement and Organizational

Commitment Administration amp Society 39 803-832

Pandey Sanjay K and Edmund C Stazyk 2008 Antecedents and Correlates of Public

Service Motivation In Motivation in Public Management edited by James L Perry

and Annie Hondeghem 101-117 Oxford Oxford University Press

- 31 -

Perry James L 1996 Measuring Public Service Motivation An Assessment of Construct

Reliability and Validity Journal of Public Administration Research and Theory 6

5-22

Perry James L 1997 Antecedents of Public Service Motivation Journal of Public

Administration Research and Theory 7 181-197

Perry James L 2000 Bringing Society in Toward a Theory of Public Service Motivation

Journal of Public Administration Research and Theory 10 471-488

Perry James L and Annie Hondeghem 2008a Motivation in Public Management Oxford

Oxford University Press

Perry James L and Annie Hondeghem 2008b Directions for Future Theory and Research

In Motivation in Public Management edited by James L Perry and Annie

Hondeghem 294-313 Oxford Oxford University Press

Perry James L Annie Hondeghem and Lois Recascino Wise 2010 Revisiting the

Motivational Bases of Public Service Twenty Years of Research and an Agenda for

the Future Public Administration Review 70(5) 681-690

Perry James L and Wouter Vandenabeele 2008 Behavioral dynamics Institutions

Identities and Self-Regulation In Motivation in Public Management edited by

James L Perry and Annie Hondeghem 56-79 Oxford Oxford University Press

Perry James L and Lois Recascino Wise 1990 The Motivational Bases of Public Service

Public Administration Review 50 367-373

Rainey Hal G 1982 Reward Preference among Public and Private Managers In Search of

the Service Ethic American Review of Public Administration 16 288-302

Ringle Christian M Oliver Goumltz Martin Wetzels and Bradley Wilson 2009 On the Use of

Formative Measurement Specifications in Structural Equation Modeling A Monte

Carlo Simulation Study to Compare Covariance-Based and Partial Least Squares

Model Estimation Methodologies Faculty of Economics and Business Administration

Maastricht Univeristy

Ringle C M S Wende and S Will 2005 Smart PLS 20 (M3) Beta

- 32 -

httpwwwsmartplsde

Ritz Adrian 2011 Attraction to Public Policy Making A Qualitative Inquiry into

Improvements in PSM Measurement Forthcoming in Public Administration

Spector Paul E 1992 Summated Rating Scales Construction An Introduction Newbury

Park CA Sage

Steenbergen Marco R 1996 Compassion and American public opinion An analysis of the

NES Humanitarianism Scale Report to the Board of Overseers of the National

Election Studies

Steffen Patrick R and Kevin S Masters 2005 Does compassion mediate the intrinsic

religion-health relationship Annals of Behavioral Medicine 30 217-223

Taylor Jeannette 2007 The Impact of Public Service Motives on Work Outcomes in

Australia A Comparative Multi-Dimensional Analysis Public Administration 85

931-959

Vandenabeele Wouter 2007 Towards a Theory of Public Service Motivation An

Institutional Approach Public Management Review 9 545-556

Vandenabeele Wouter 2008a Government Calling Public Service Motivation as an Element

in Selecting Government as an Employer of Choice Public Administration 86

1089-1105

Vandenabeele Wouter 2008b Development of a Public Service Motivation Measurement

Scale Corroborating and Extending Perryrsquos Measurement Instrument International

Public Management Journal 11 143-167

Vandenabeele Wouter 2009 The Mediating Effect of Job Satisfaction and Organizational

Commitment on Self-Reported Performance More Robust Evidence of the PSM-

Performance Relationship International Review of Administrative Sciences 75 11-

34

Wilcox James B Roy D Howell and Einar Breivik 2008 Questions about Formative

Measurement Journal of Business Research 61 1219-1228

Wise Lois Recascino 2000 The Public Service Culture In Public Administration Concepts

- 33 -

and Cases 7th ed edited by Richard J Stillman II 342-353 Boston Houghton

Mifflin

Wright Bradley E 2008 Methodological Challenges associated with Public Service

Motivation Research In Motivation in Public Management edited by James L

Perry and Annie Hondeghem 80-98 Oxford Oxford University Press

Wright Bradley E and Sanjay K Pandey 2008 Public Service Motivation and the

Assumption of Person-Organization Fit Testing the Mediating Effect of Value

Congruence Administration amp Society 40 502-521

- 34 -

Table 1

Possible Indicators of PSM and Descriptive Statistics (n = 2868)

Dimensions and items Mean Standard

Deviation

Item-Total

Correlation

Attraction to Public Participation (APP)

APP1 I am interested in helping to improve public service

APP2 I am satisfied when I see people benefiting from the public programs I was

involved in

APP3 I like to discuss topics regarding public programs and policies with others

APP4 I believe that public sector activities contribute to our general welfare

APP5 I admire people who initiate or are involved in activities to aid my

community

APP6 Contributing to public programs and policies helps me realize myself

APP7 It is important to contribute to activities that tackle social problems

412

441

355

393

423

351

409

696

662

917

753

672

881

734

524

421

482

412

493

478

552

Commitment to Public Values (CPV)

Sub-dimension for Public Interests (CPV1)

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

CPI3 I would prefer seeing public officials do what is best for the whole

community even if it harmed my interests

CPI4 Serving the public interest is more important than helping a single individual

Sub-dimension for Public Values (CPV2)

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of public

services

CPV3 It is fundamental that public services respond to the needs of the citizens

CPV4 Decisions regarding public services should be democratic despite the time

and effort it takes

CPV5 Everybody is entitled to a good service even if it costs a lot of money

CPV6 It is fundamental that the interests of future generations are taken into

account when developing public policies

CPV7 To act ethically is essential for public servants

CPV8 I believe that public employees must always be aware of the legitimacy of

their activities

CPV9 I personally identify with the aim of protecting individual liberties and rights

418

409

348

337

428

416

433

394

367

426

441

434

414

695

662

899

995

730

704

710

852

932

689

681

692

763

525

596

480

251

484

437

356

325

376

439

462

402

392

Compassion (COM) COM1 It is difficult for me to contain my feelings when I see people in distress

COM2 I feel sympathetic to the plight of the underprivileged

COM3 I empathize with other people who face difficulties

COM4 I have little compassion for people in need who are unwilling to take the

first step to help themselves

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

354

387

388

299

412

391

951

825

707

1083

755

765

399

511

389

011

487

502

Self-Sacrifice (SS)

SS1 Making a difference to society means more to me than personal achievements

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS5 People should give back to society more than they get from it

SS6 Serving other citizens would give me a good feeling even if no one paid me for

it

SS7 I would agree to a good plan to make a better life for the poor even if it costs

me money

357

329

339

302

345

363

350

904

872

949

982

888

1049

895

508

584

454

531

412

368

546

- 35 -

Table 2

Background of Respondents (n = 2868)

Variables Characteristics Respondents

Sex

Male

Female

NA

1209

1622

37

422

566

13

Length of service

(years)

0 ~ 10

10 ~ 20

20 ~ 30

30+

NA

1126

723

484

491

44

393

252

169

171

15

Organizational status

―Do you supervise employees

No

Yes

NA

1472

1076

320

513

375

112

Country

Australia

Belgium

China

Denmark

France

Italy

Korea

Lithuania

The Netherlands

Switzerland

United Kingdom

USA

249

214

230

249

266

162

253

236

249

250

260

250

87

75

80

87

93

56

88

82

87

87

91

87 Note NA = no answer

- 36 -

Table 3

Standardized factor loadings and correlations for the 31-item PSM index (n = 2868)

Dimensions and items SFL (MLE) SFL (D-WLS) Alpha

APP

APP1

APP2

APP4

APP5

APP6

APP7

597

486

455

558

490

600

657

552

514

625

549

665

699

CPV1

CPI1

CPI2

CPI3

CPI4

588

660

488

289

653

732

535

294

542

CPV2

CPV1

CPV2

CPV3

CPV4

CPV5

CPV6

CPV7

CPV8

CPV9

611

546

436

393

418

534

598

496

460

690

631

505

457

486

601

669

573

530

745

COM

COM1

COM2

COM3

COM5

COM6

496

569

498

597

607

516

651

538

670

684

687

SS

SS1

SS2

SS3

SS4

SS5

SS6

SS7

567

736

558

728

467

409

663

660

762

591

695

515

494

709

782

Measures of fit (MLE) χsup2 (df = 424) = 43699 p lt 001 CFI = 835 GFI = 899

RMSEA = 057 [055 059]

Measure of fit (D-WLS) χsup2 (df = 424) = 39247 p lt 001 CFI = 971 GFI = 976

RMSEA = 054 [052 055]

Inter-factor correlations MLE (D-WLS)

1 2 3 4

1 APS

2 CPV1 996 (988)

3 CPV2 902 (890) 798 (798)

4 COM 787 (783) 758 (762) 779 (783)

5 SS 700 (742) 800 (859) 509 (537) 720 (729)

Note SFL = standardized factor loading All factor loadings and correlations are significant at p lt 001

- 37 -

Table 4

Standardized factor loadings and correlations for the 16-item PSM index (n = 2868)

Dimensions and items SFL

(MLE)

SFL

(D-WLS)

APS

APP5 I admire people who initiate or are involved in activities to aid my

community

APP7 It is important to contribute to activities that tackle social problems

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

611

592

637

666

665

669

690

751

CPV

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of

public services

CPV6 It is fundamental that the interests of future generations are taken

into account when developing public policies

CPV7 To act ethically is essential for public servants

612

546

551

590

698

631

618

658

COM

COM1 It is difficult for me to contain my feelings when I see people in

distress

COM2 I feel sympathetic to the plight of the underprivileged

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

490

546

596

604

515

608

669

679

SS

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS7 I would agree to a good plan to make a better life for the poor even if

it costs me money

763

571

766

678

807

629

747

755

Measures of fit (MLE)

χsup2 (df = 98) = 8490 p lt 001

CFI = 939 GFI = 965

RMSEA = 052 [049 055]

Measure of fit (D-WLS)

χsup2 (df = 98) = 6069 p lt 001

CFI = 988 GFI = 992

RMSEA = 043 [039 046]

Alpha CR (D-WLS) Inter-factor correlations MLE (D-WLS)

1 APS 717 788 1 APS 2 CPV 3 COM

2 CPV 663 747 848 (859)

3 COM 645 713 781 (791) 789 (795)

4 SS 784 826 565 (596) 462 (505) 715 (726)

Note SFL = standardized factor loading Alpha = Cronbachrsquos alpha CR = composite reliability

All factor loadings and correlations are significant at p lt 001

- 38 -

Table 5

Results of CFA in the individual countries Standardized factor loadings and measures of fit

Factors

amp Items

Australia Belgium China Denmark France Italy

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

672

756

799

785

804

943

710

703

597

790

799

730

584

675

685

681

731

742

578

748

674

643

765

739

645

704

678

708

812

779

632

604

765

766

672

853

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

552

593

560

699

727

632

624

605

560

713

709

645

702

602

616

724

686

735

567

560

715

654

605

811

608

658

587

696

755

698

673

620

523

782

697

654

COM

COM1

COM2

COM3

COM5

COM6

623

584

662

688

701

739

732

608

475

808

666

581

578

616

605

649

670

673

699

751

737

679

816

776

568

546

801

626

673

850

637

546

741

715

601

822

SS

SS2

SS3

SS4

SS7

654

794

806

892

767

813

727

693

799

793

725

838

857

820

773

896

855

842

806

749

792

829

753

884

847

731

753

889

816

753

815

692

664

795

782

734

Inter-factor

correlations

093

~

752

163

~

810

320

~

773

323

~

825

452

~

827

497

~

855

323

~

850

321

~

847

498

~

899

536

~

902

425

~

829

476

~

865

Sample size 249 214 230 249 266 162

Measures of Fit

χsup2 638 444 881 732 765 554 817 625 735 565 750 564

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 980 100 939 987 969 997 967 994 975 997 949 995

GFI 961 986 941 986 948 991 949 992 955 992 926 986

RMSEA 036 00 063 050 051 026 053 035 045 026 059 033

- 39 -

Table 5 (Continued)

Factors

amp Items

Korea Lithuania The Netherlands Switzerland UK USA

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

619

701

667

706

781

717

666

512

682

751

605

779

557

629

630

620

719

729

535

621

702

643

694

735

655

725

738

710

804

822

535

621

702

643

694

735

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

594

607

722

722

644

722

674

535

533

790

589

711

568

554

538

655

628

545

627

580

615

711

666

704

800

512

555

873

581

646

627

580

615

711

666

704

COM

COM1

COM2

COM3

COM5

COM6

694

677

550

763

736

627

557

639

428

550

696

534

737

623

508

766

616

608

659

592

683

703

661

769

682

583

729

739

644

803

659

592

683

703

661

769

SS

SS2

SS3

SS4

SS7

776

752

679

779

770

811

707

721

635

725

734

706

814

656

743

820

767

764

745

632

689

788

662

744

671

818

832

830

864

802

745

632

689

788

662

744

Inter-factor

correlations

432

~

726

466

~

746

376

~

788

406

~

902

377

~

852

455

~

836

637

~

818

660

~

828

317

~

852

360

~

852

637

~

818

660

~

828

Sample size 253 236 249 250 260 250

Measures of Fit

χsup2 896 573 612 462 773 616 815 715 769 542 815 715

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 951 996 975 100 958 993 957 991 972 998 957 991

GFI 942 988 960 989 951 987 948 990 955 990 948 990

RMSEA 059 028 034 00 050 034 053 044 048 022 053 044

Note All factor loadings and correlations are significant at p lt 001

- 40 -

Appendix 1 Dimensions and possible items of PSM (the first round)

1 Attraction to Public Participation

(1) I am interested in developing public programs helping my country or community (Kim 2009a)

(2) Sharing my views on public policies with others is attractive to me (Kim 2009a)

(3) I am satisfied when seeing people get benefits from the public programs I was involved in (Kim 2009a)

(4) I like to discuss political subjects with others (Giauque et al 2011)

(5) I think the governmental activities contribute to our welfare

(6) I like to initiate actions to help out my community

(7) I do not care much about politicians (Perry 1996)

2 Commitment to Public Values

Possible items for public interest

(1A) Meaningful public service is very important to me (Perry 1996)

(2A) It is important for me to contribute for the common good (Giauque et al 2011)

(3A) I recognize myself with the mission of protection of liberties and rights endorsed by the nation-state

(Giauque et al 2011)

(4A) I would prefer seeing public officials do what is best for the whole community even if it harmed my

interests (Perry 1996)

(5A) To me serving the public interest is more important than helping other people (Vandenabeele 2008b)

Possible items for public values

(1B) I recognize myself with the promotion of the equality of chances endorsed by the nation-state (Giauque et

al 2011)

(2B) It is fundamental that the public service provided to the citizen is regular and continuous (Giauque et al

2011)

(3B) It is fundamental that public service innovates to respond to the new needs of the citizensclients (Giauque

et al 2011)

(4B) The dignity and well-being of all should be the most important concerns in any society (Steenbergen

1996)

(5B) It is important that public servants account for all the costsexpenses they make (Vandenabeele 2008b)

(6B) We have to do everything in our power to pursue the goal of democracy

3 Compassion

(1) It is difficult for me to contain my feelings when I see people in distress (Perry 1996)

(2) I feel sympathetic to the plight of the underprivileged (Kim 2009a)

(3) I am often reminded by daily events how dependent we are on one another (Perry 1996)

(4) I feel compassion for others in the difficulties they are facing (Steffen and Masters 2005)

(5) It is not really my problem if others are in trouble and need help (Beutel and Marini 1995)

(6) I find it hard to be sympathetic toward starving people in foreign lands when there is so much trouble in our

own country (Beutel and Marini 1995)

(7) I get very upset when I see other people treated unfairly (Beutel and Marini 1995)

(8) I would agree to a good plan to make a better life for the poor even if it cost me money (Beutel and Marini

1995)

(9) To me patriotism includes seeing the welfare of others (Perry 1996)

(10) I care very much about other people

4 Self-Sacrifice

(1) Making a difference in society means more to me than personal achievements (Perry 1996)

(2) I am prepared to make enormous sacrifices for the good of society (Perry 1996)

(3) I believe in putting duty before self (Perry 1996)

(4) I am one of those rare people who would risk personal loss to help someone else (Perry 1996)

(5) I think people should give back to society more than they get from it (Perry 1996)

(6) Serving other citizens would give me a good feeling even if no one paid me for it (Perry 1996)

(7) Much of what I do is for a cause bigger than myself (Perry 1996)

- 41 -

Appendix 2 Newly suggested items through the first round

1 Items for the Attraction to Public Participation Dimension

(1) In the public domain every person should be counted as equal

(2) I enjoy to think and argue in a political way

(3) I am satisfied when seeing my ideas being integrated in public policies

(4) I can easily relate to the way of thinking and arguing of politicians

(5) Working in the public sector I can better realize myself

2 Items for the Public Values Sub-Dimension

(1) Public service should give value for money

(2) Cost considerations should play an important role in decisions about public service provision

(3) It is fundamental that the interests of future generations are taken into account when developing public

policies

(4) To act with correct ethics is for public servants as important as their professional competence

(5) Primarily public servants should be accountable to the public and not to their superiors

3 Items for the Self-Sacrifice Dimension

(1) I would agree to a good plan to make a better life for the poor even if it costs me money

- 42 -

Appendix 3 An example of a summary of comments and suggestions (the second round)

On the item ―I am interested in developing public programs helping my country or community

A It is a good sentence but is it possible for everyone to ―develop public programs Even public officials are

limited to do this

B OK

C or the special group

D Double barrel and difficult to translate to my country ―I am interested in helping to improve public services

E Good but I feel that wording requires revision If we draw sample from local government then country is

redundant ―I am interested in developing public policy programs that aim in helping my community

F 4 (General appreciation from1 to 4) Not clear whether it refers to ones work or to ones concern as a citizen

―(In my work) I am interested in contributing to the development of public etchellip

G Instead of development I would use implementing

H Delete public which is redundant because of the characteristic helping my country or community but add

or implementing ―I am interested in developing or implementing programs helping my country or

community

I Referring to dimension Wording Similar to item 6

Note Names have been changed to alphabets in order to ensure the anonymity of participants

- 43 -

Appendix 4 Dimensions and Items of Public Service Motivation for Future Research

Dimensions and items

Attraction to Public Service (APS)

1 I am interested in helping to improve public service

2 I admire people who initiate or are involved in activities to aid my community

3 It is important to contribute to activities that tackle social problems

4 Meaningful public service is very important to me

5 It is important for me to contribute to the common good

Commitment to Public Values (CPV)

1 I think equal opportunities for citizens are very important

2 It is important that citizens can rely on the continuous provision of public services

3 It is fundamental that the interests of future generations are taken into account when

developing public policies

4 To act ethically is essential for public servants

5 I believe that public employees must always be aware of the legitimacy of their activities

6 I personally identify with the aim of protecting individual liberties and rights

Compassion (COM)

1 It is difficult for me to contain my feelings when I see people in distress

2 I feel sympathetic to the plight of the underprivileged

3 I empathize with other people who face difficulties

4 I get very upset when I see other people being treated unfairly

5 Considering the welfare of others is very important

Self-Sacrifice (SS)

1 I am prepared to make sacrifices for the good of society

2 I believe in putting civic duty before self

3 I am willing to risk personal loss to help society

4 I would agree to a good plan to make a better life for the poor even if it costs me money

- 4 -

of the dimensions of the PSM construct (Castaing 2006 Coursey and Pandey 2007a Coursey

et al 2008 DeHart-Davis Marlowe and Pandey 2006 Kim 2009b Leisink and Steijn 2009

Liu Tang and Zhu 2008 Moynihan and Pandey 2007b Taylor 2007 Vandenabeele 2008a

Wright and Pandey 2008) Moreover as the construct and operational definition of PSM are

refined it is necessary to adapt items for the dimensions of the PSM construct

This study is focused on developing a revised measurement instrument for PSM for

international use After developing the possible items for measuring PSM the international

survey was conducted in 2010 and the data (n = 2868) were collected from twelve countries

Dimensions and items were tested cross-nationally as this can be a useful tool to preclude

value-specific items and wordings that are case-specific The next section of this paper will

briefly explain the processes for developing the international measurement instrument The

survey results are analyzed and discussed in order to confirm the dimensional structure and

validate the items of PSM Finally the items of PSM are proposed and discussions and

suggestions are given for future research

Developing the Measurement Instrument of PSM for International Use

Background

Perry and Wise (1990) proposed that PSM is associated with three types of motives affective

norm-based and rational Perry (1996) identified four empirical components of the PSM

construct attraction to public policy making commitment to the public interestcivic duty

compassion and self-sacrifice Since then many studies have used the dimensions and items

of Perryrsquos (1996) measure and improvements have been made in the PSM scale (Kim 2009a

Taylor 2007 Vandenabeele 2008b) However Perryrsquos (1996) measure has limitations (Kim

and Vandenabeele 2010) In particular the items of the dimension of attraction to policy

- 5 -

making are not appropriate for measuring personal attraction to public policy making

(Camilleri 2006 Coursey and Pandey 2007a Coursey et al 2008 Kim 2009b Ritz 2011)

Moreover the dimension of commitment to public interest needs to concentrate more on a

personal disposition to pursue public values (Castaing 2006 Leisink and Steijn 2009 Taylor

2007) and the items of compassion do not represent affective motives (DeHart-Davis

Marlowe and Pandey 2006 Moynihan and Pandey 2007a Wright 2008) Thus the

dimensions of the PSM construct are modified as attraction to public participation

commitment to public values compassion and self-sacrifice Consequently it is necessary to

develop appropriate items for the dimensions (Kim and Vandenabeele 2010)

A related issue concerns the measurement model of PSM There are two

measurement models using multiple dimensions of latent constructs reflective and formative

When variation in a construct leads to variation in its dimensions such dimensions are

termed ―reflective while dimensions are labeled ―formative when they are viewed as

causes of a construct (Edwards and Bagozzi 2000 Diamantopoulos and Siguaw 2006

Diamantopoulos and Winklhofer 2001) It is understandable that each dimension of PSM has

several reflective indicators that are essentially interchangeable A problem is the relationship

between PSM and its dimensions If we regard the dimensions as reflective it is possible to

analyze PSM with only two or three dimensions because omitting dimensions does not

change the essential nature of PSM However if we regard the dimensions as formative it is

not acceptable to omit a dimension as this may change the meaning of PSM Construct

conceptualization is regarded as being the process that determines the formative or reflective

nature (Diamantopoulos and Siguaw 2006) Recent research has demonstrated theoretical and

empirical evidence in support of a second-order formative approach to PSM (Kim 2010)

This means that the dimensions represent different conceptual components of PSM and so

- 6 -

they are not interchangeable Thus PSM is regarded as a combination of specific dimensions

If any one of the dimensions changes PSM would change Conversely if an individualrsquos

PSM increases this would not necessarily be accompanied by an increase in all dimensions

Therefore as the PSM construct is refined for international use (Kim and Vandenabeele

2010) scholars must be attentive to identifying and using appropriate measurement models

Developing Possible Items

The conceptual components of PSM need to be empirically estimated through operational

definition Kim and Vandenabeele (2010) proposed that PSM is based on self-sacrifice and

associated with instrumental value-based and identification motives and that the dimensions

of the PSM construct be refined along the lines of attraction to public participation (APP)

commitment to public values (CPV) compassion (COM) and self-sacrifice (SS) As Perry

(1996) identified a multidimensional scale to measure PSM it is conceived as an overall

latent variable with various latent dimensions and is referred to as a second-order factor

model Research has found that the measurement model specifies the causal relationship

between PSM and its dimensions (Edwards and Bagozzi 2000) The direction of the

relationship is either from the PSM construct to the dimensions (second-order reflective

measurement) or from the dimensions to the construct (second-order formative measurement)

A theoretical consideration of the relationships between PSM and its dimensions led

to the proposal that PSM be defined in terms of multiple dimensions (Kim 2010 Kim and

Vandenabeele 2010) The dimensions represent different aspects of PSM with each

dimension capturing a distinct and potentially unique form of PSM Moreover the

dimensions may have different antecedents and consequences as well as different

characteristics and theoretical backgrounds (Moynihan and Pandey 2007a Pandey and

- 7 -

Stazyk 2008 Perry 1996 1997 2000 Perry and Wise 1990 Perry and Vandenabeele 2008

Taylor 2007 Vandenabeele 2007 2008a 2009) An individualrsquos PSM is shaped by the

individualrsquos APP CPV COM and SS Deleting a dimension means omitting a certain part of

PSM Therefore it is reasonable to suggest that the dimensions of PSM are formative and for

PSM to be regarded as a composite of its dimensions

Each dimension of PSM can be measured using reflective indicators that satisfy the

following criteria (Jarvis MacKenzie and Podsakoff 2003) a relative homogeneity and

interchangeability of indicators pertaining to each dimension a high degree of covariation

among indicators of each dimension and the expectation that the indicators of each

dimension are likely to be affected by the same antecedents and have the same consequences

Reflective indicators are essentially interchangeable and so adding or removing indicators

does not change the essential nature of a dimension Thus we consider PSM to be a second-

order formative construct with several dimensions that each has several reflective indicators

(reflective first-order formative second-order)

Based upon the refinement and clarification of the operational dimensions of PSM

(Kim and Vandenabeele 2010) items were developed for the four dimensions The items

selected were a mix of pre-existing items that loaded highly on PSM dimensions in multiple

studies (Giauque et al 2011 Kim 2009a Perry 1996 Vandenabeele 2008b) Where

appropriate items were not available new items were drafted from other previous studies

creating an initial total of 35 proposed items For the dimension of APP which focuses on

individual disposition to work in the public sector to participate in the public policy process

and in activities for community and social development seven possible items were suggested

The CPV dimension focuses on personal disposition to pursue public interest and public

values Two sub-dimensions of CPV seemed plausible one focusing on public interest and

- 8 -

another on public values such as social equity equal opportunity and democracy Thus five

possible items for the public interest sub-dimension (CPV1) and six items for the public

values sub-dimension (CPV2) were developed For the dimension of COM which focuses on

affective bonding with the identified objects such as other members of a social category or of

a political system ten possible items were suggested For the SS dimension Perryrsquos original

items (1996) were the primary source of items The full list of items is provided in Appendix

1

Review Processes

A group of researchers who shared our interest in better measurement of PSM collaborated

using e-mail in November 20091 The objective was to develop a measurement instrument

for PSM with a robust comparative character in other words to create a valid instrument (on

a national basis) that can be used internationally The first round for reviewing the possible

items lasted from the 16th

of November until the 7th

of December 2009 and all the

participants were asked to review these proposed items For this review two criteria were

employed (1) Do these items refer to the dimensions they claim to refer to and (2) Are

these items meaningful across various national contexts and in particular to the situation of

civil servants in that country Participants were also asked to add new items that might better

reflect a particular dimension or to rephrase items that were already on the list

A second round of item refinement lasted from the 9th

of December until the 22nd

of

December 2009 During this time 11 newly suggested items from the first round were

considered (see Appendix 2) A summary of all comments and suggestions emerging from the

1 Emmanuel Camilleri David Giauque Gerhard Hammerschmid Isabel Egger-Peitler Renate Meyer and

Cristian Pliscoff also participated in the review processes

- 9 -

first process was distributed to the participants to stimulate comments in the second round

(see Appendix 3) Based on the comments received in the second round discussion the

possible items for an international survey were selected However some items required

further discussion and so a third round was conducted from the 15th

of January until the 23rd

of January 2010 One issue that remained unresolved in the discussions relates to the

dimension of CPV Some argued that it is reasonable to divide CPV into two sub-dimensions

because committing to public interests is different from pursuing public values and there are

many important public values in democratic countries while others contended that CPV can

cover all important public values including public interest and there is no real benefit to add

one more dimension These differences were left to be resolved through empirical research

This did not affect the final items however and 33 items were confirmed The resulting

survey questionnaire was distributed to all the collaborators on February 9 2010 in order to

conduct a survey in each country

Measures

PSM was measured using the newly developed 33-item index 7 items for the dimension of

APP 13 items for CPV (4 for CPV1 and 9 for CPV2) 6 items for COM and 7 for SS In the

survey questionnaire a five-point Likert type scale (1 = strongly disagree 5 = strongly agree)

was used This scale is among the most widely used response formats in many empirical

investigations within the social and behavioral science in which there are no ―correct or

―incorrect responses (Spector 1992) The items of PSM were randomly distributed in the

survey format Gender age education length of service and employment status were also

measured in this survey

It was necessary to translate the measurement items into the official (native) language

- 10 -

of each country As this raised the possibility of inaccurate translations the back-translation

of the items into native languages was validated by asking two researchers to independently

translate the English-language items The researchers were instructed to consult with each

other over differences between translated items and to test the survey items with some native

respondents

[Table 1 about here]

Samples

The international survey was fielded from March to September 2010 The data upon which

this study are based were collected from civil servants in local governments in Australia

Belgium China Denmark France Italy Korea Lithuania the Netherlands Switzerland the

United Kingdom and the USA The principal criterion was that the samples collected in the

various countries should be comparable to the largest possible extent The criterion ruled out

national and state government as the competencies of national governments and also of state

governments if they exist constitutionally are different among most of the countries included

in the research Therefore we opted for local government as our focus The structure and

functions of local governments can differ widely even within countries and so the survey in

each country was focused on town hall bureaucrats in city county or township government

The respondents were permanent employees from both managerial and non-managerial levels

and from both administrative departments (ie no direct contact with the public) and service

departments (ie likely to have direct contact with the public) Despite being employed by

local governments in many or all of the countries police officers firefighters school teachers

public transport workers artists and musicians as well as nurses and doctors were excluded

- 11 -

from the sample

We aimed to receive 250 respondents for each country as this would enable us to

carry out the analyses we wanted to perform on the data Although there is no ideal sample

size one study recommends a sample of 200 to provide a sound basis for estimation (Hair

Black Babin and Anderson 2010) The method of data collection was left open Obviously a

web-survey is the easiest way to collect the data because it is economical to collect and to

code however compared to paper-pencil surveys this method has some drawbacks First not

everybody in the sample may have access to the internet which may cause a bias in the

responses Second internet surveys are generally known to produce lower response rates

Web surveys were recommended if possible but paper-pencil surveys were deemed

acceptable in cases where it was not possible to conduct a web survey Researchers were

encouraged to contact just one municipality except in cases where it was not possible to

gather enough data In such instances responses from multiple municipalities were accepted

By the end of September 2010 a total of 2868 responses were obtained2

2 In Australia seven local councils each with between from 70 and 232 employees were surveyed Paper-pencil

surveys were used in two councils while the rest relied on a web survey The local governments limited

distribution of paper-pencil surveys to only one or two sections of the council eg administration The response

rate was approximately 30 However it should be noted that not everyone working for the council has direct

access to a computer In Belgium three municipalities were surveyed on a census base by means of a web

survey as all employees had access to a computer This rendered a response of 120 (226) 79 (75) and 16

(53) responses for the respective municipalities resulting in an overall response rate of 331 with 662

invitations being sent out In China two departments social work and administration were surveyed in one

municipality using paper-pencil surveys In the social work department 65 out of 110 surveys were returned

and 165 out of 211 responded from the administration department giving an overall response rate of 767 In

Denmark a total of 1282 respondents in a municipality were sent an initial e-mail invitation to a web survey

and up to two e-mail reminders and the response rate was 473 In France all the employees in two

municipalities were surveyed In one municipality where both web and paper-pencil surveys were utilized 128

out of 685 questionnaires were returned while in another city using only a paper-pencil survey 138 out of 652

were received The average response rate was 199 In Italy in principle the link to the questionnaire was

made available to all employees (about 4000) in one city but for some of them access may have been difficult

yielding a response rate of 45 In Korea 500 employees in a city received a printed questionnaire and 253

responses were returned yielding a response rate of 506 In Lithuania 236 responses from 54 municipalities

were gathered by web survey In the Netherlands three municipalities were surveyed on a census base by means

of a web survey In total 319 responses of 910 invitations were returned (35) mdash 90 (45) 115 (38) and 114

(28) for the respective municipalities In Switzerland 516 out of 1326 local employees with e-mail access in

a municipality completed the survey giving a response rate of 389 In the United Kingdom the survey was

posted online and 1360 employees both in administrative jobs and front-line services of one municipality

- 12 -

Municipalities were selected because of their convenient accessibility and proximity to the

researcher and a convenience sample was used in most countries Seven countries surveyed

one municipality but seven local councils were surveyed in Australia Web surveys were used

in four countries while three countries used both web and paper-pencil survey The number

of respondents varied among the countries depending on the numbers of questionnaires

distributed the survey method and the response rate In most countries we used

approximately 250 responses but only 162 from Italy3 The majority of the respondents were

women (566) while in terms of service length those in the largest group (393) had

worked for fewer than 10 years Table 2 shows the distribution of respondentsrsquo sex length of

service employment status and country

[Table 2 about here]

Analyses

The statistical analysis applied confirmatory factor analysis (CFA) CFA was used to assess

the fit of the data to the hypothesized measurement model because it is preferred where

measurement models have a well-developed underlying theory for hypothesized patterns of

loadings The data were analyzed using Amos 180 (Arbuckle 2009) with the maximum

likelihood (MLE) estimation method and LISREL 872 (Joumlreskog and Soumlrbom 2005) with the

diagonally weighted least squares (D-WLS) estimation method

The individual items are designed to measure a theoretically continuous construct

but the observed responses are a discrete realization of five categories The common method

received an email to complete the survey The response rate was 19 In the USA the total number of

employees surveyed was 626 in one city using both electronic and paper-pencil surveys resulting in a response

rate of 588 3 In countries where significantly more than 250 surveys were collected we randomly selected only 250 cases

for analysis

- 13 -

of estimation within CFA is MLE a technique which assumes that the observed variables are

continuous and normally distributed These assumptions are not met when the observed data

are discrete as occurs when using ordinal scales as in this study A simulation study (Flora

and Curran 2004) found that D-WLS performs well for ordinal measures across a variety of

conditions especially for smaller samples Coursey and Pandey (2007a 2007b) argued the

importance of using D-WLS and not standard MLE in structural equations for ordinal data4

In most cases however the differences between MLE and D-WLS are not substantive to the

conclusions5 No comparison between standard MLE and D-WLS is provided in this paper

We give results from both MLE and D-WLS and this will aid in confirming the stability and

robustness of the results across varying mathematical specifications

For gauging reliability Cronbachrsquos alpha and composite reliability were tested

Composite reliability should be equal to or greater than 07 (Hair et al 2010) Convergent

validity was assessed from the measurement model by determining the statistical significance

of each indicatorrsquos estimated pattern coefficient on its specified underlying construct

Discriminant validity was assessed by determining whether the confidence interval around

the correlation estimate between two factors included 100 (Anderson and Gerbing 1988)

For model fit assessment goodness-of-fit index (GFI) comparative fit index (CFI)

and the root mean square error of approximation (RMSEA) were used The model achieves

an acceptable fit to the data when GFI and CFI equals or exceeds 090 and RMSEA values

4 Applying standard MLE to ordinal data such as Likert-type items may produce significant estimation

problems such as inflation of chi-square fit statistics and biased underestimation of parameters and standard

errors (Flora and Curran 2004) Coursey and Pandey (2007a 2007b) used a modified MLE using the asymptotic

covariance matrix and D-WLS for adjusting the ordinal nature of the data They reported that although the

modified MLE findings are slightly different on a few model parameters the differences are not substantive to

the conclusions However model fit estimators were clearly better for the D-WLS 5 Given normally distributed categorical variables MLE can be used with confidence when a variable has four

or more categories If there is a large sample a five-point Likert scale and the skewness and kurtosis of variables

is within acceptable limits it suggests distribution symmetry ―In most cases where the hypothesized model is

well specified and the scaling based on more than three categories it seems unlikely that there will be much

difference between the findings (Byrne 2010 160)

- 14 -

fall below 008 (Hair et al 2010) In general the larger the value of GFI and CFI and the

smaller the value of RMSEA the better the fit of the model (Bollen 1989) We also present

the χ2 test statistic Lower values of χ

2 indicate a better fit and should be nonsignificant but

the χ2 value increases as sample size increases and thus for large sample sizes this statistic

may lead to rejection of a model with good fit (Hair et al 2010)

Results

Using SPSS 180 descriptive statistics were computed for individual items as shown in table

1 One item (APP3) in the APP dimension ―I like to discuss topics regarding public programs

and policies with others was mistakenly omitted in the survey conducted in the United

States and so it was deleted in the statistical analysis In the assessment of normality no

items showed a skew or kurtosis value greater than the cutoffs of |3| or |8| recommended by

Kline (2005) As data were distributed normally MLE as well as D-WLS estimation was used

Based on inspection of item-total correlations one item (COM4) was dropped from further

analysis because it was weakly correlated with the overall index (011) Missing data is an

issue in CFA as it can cause estimation errors All variables in the survey have missing data

rates of approximately 016 to 143 Mean substitution was used to calculate replacement

values for the missing values in each countryrsquos dataset (Hair et al 2010) The deletion of the

two items left 31 items for the next stage of analysis

The five-correlated factor model was tested using CFA The CFA model in the initial

analysis hypothesized a priori that (1) responses to the 30-item PSM measurement

instrument could be explained by five factors such as APP CPV1 CPV2 COM SS (2) each

item would have a nonzero loading on the PSM factor that it was designed to measure and

- 15 -

zero loadings on all other factors (3) the five factors would be correlated and (4)

measurement error terms would be uncorrelated The resulting CFA with MLE suggested that

the initial five-correlated factor model was not a good fit to the data χsup2 (df = 424) = 43699 p

lt 001 GFI = 0899 CFI = 835 RMSEA = 0057 [055 059] At this stage APP was highly

correlated with CPV1 (r = 996 in MLE and r = 988 in D-WLS) suggesting the two

dimensions were not unique and lacked discriminant validity

[Table 3 about here]

Given the disconfirmation of the initial model an effort was made to estimate an

alternative model The two dimensions APP and CPV1 were combined to form a single

dimension Attraction to Public Service (APS) The items of CPV1 may be regarded as

action-oriented by the respondents because dedication to public service community and

common good is underlined in these items Consequently it is reasonable to combine APP

with CPV1 to form APS a dimension that focuses more on disposition to serve the public to

work for the common good and to participate in public policy processes After the overall

number of dimensions was reduced to four using standardized factor loadings and

modification indices as criteria the items with the lowest factor loading in each dimension

were deleted and the items whose modification index indicated association with multiple

dimensions were eliminated A CFA model with the remaining items was iteratively tested

until it achieved a good fit to the data The outcome of this refinement process was a 16-item

index of four factors Overall model fit by every measure is quite strong for the D-WLS

estimation technique The estimation results for the four-dimension model are presented in

table 4

- 16 -

[Table 4 about here]

Cronbachrsquos coefficient alpha for the 16-item PSM index was 860 and the coefficient

alphas for the four dimensions ranged from 645 to 784 The composite reliability of the set

of reflective indicators for each dimension of PSM ranged from 713 to 826 and indicated

adequate internal consistency The resulting factor structure showed a four-factor structure

with all items loading significantly onto their a priori dimension (p lt 001) and the

standardized factor loadings ranging from 0490 to 0766 in MLE and from 515 to 807 in

D-WLS The results provide support for convergent validity The correlation estimates

between the two factors ranged from 0462 to 0848 in MLE and from 505 to 859 in D-

WLS The confidence intervals (plusmn2 standard errors) around the correlation estimates between

the two factors did not include 100 The result provides support for discriminant validity To

test whether the four-dimension model is superior to other designs such as the two or three-

dimension models a series of CFA were conducted and the results show that the four-

dimension model is a significantly better fit than any of the other designs

To test whether the factorial structure of the PSM instrument is equivalent across the

samples from the twelve countries four nested models were tested as part of multi-group

analysis with MLE (Byrne 2010) First the unconstrained model was the baseline model

which relaxed all equality constraints This model hypothesized the same four-factor structure

of the instrument across twelve countries The fit indices for this baseline model were χsup2 (df =

1176) = 24977 p lt 001 GFI = 0901 CFI = 900 RMSEA = 0020 [019 021] These

indices indicate that the four-factor model of the PSM instrument exhibits acceptable fit

across the samples from the twelve countries Second the measurement weights model tested

- 17 -

the invariance of factor loadings across countries by placing equality constraints on these

parameters The fit indices for this model were χsup2 (df = 1308) = 28088 p lt 001 GFI = 0890

CFI = 887 RMSEA = 0020 [019 021] Since the measurement weights model was nested

within the unconstrained model the chi-square difference text ∆χsup2 (132) = 3111 p lt 001

indicated that some equality constraints of factor loadings did not hold across the twelve

countries Further test results suggested that the equality constraints of structural covariances

and measurement residuals were not upheld across the twelve countries It needs to be

stressed that despite the differences in factor loadings and factor covariances the results

support the main assumption that the same four-factor structure holds across different

countries We may assume that people in different cultures understand PSM in the same

fashion

To find differences in the factor structure of PSM among countries we conducted a

series of CFA with each countryrsquos dataset and followed the same procedure as when

analyzing the sample as a whole As table 5 shows although there are great similarities

between the appropriate items in the countries there are also some differences Even though

the items vary according to national differences the four-factor structure had a good fit to the

data Thus the results confirm that the four-factor structure of PSM can be generalized

[Table 5 about here]

The APS dimension covers both an individualrsquos attraction to public participation and

commitment to public interest but all its items refer to some notion of public interest such as

improving public service aiding community tackling social problems meaningful public

service and contributing to the common good The APS dimension captures an individualrsquos

- 18 -

disposition to serve the public interest through public participation When developing the

items some assumed that the items of committing to public interests can be separated from

those of pursuing public values in the CPV dimension The empirical test showed that

respondents tended to understand committing to public interests as behavior-oriented and so

the items for public interest are not differentiated from participation-oriented items in their

perceptions but are different from those for public values The respondents seem to

understand that pursuing public values is more related to personal beliefs or creeds Thus the

CPV dimension is solely focused on public values and the items of pursing public interests

are combined with those of the APP dimension to make the APS dimension

Public values and their relative importance are neither impervious to change over

time nor the same among all countries (Charles de Jong and Ryan 2011 Davis and West

2009) Moreover some values may be in conflict as Martinsen and Beck Joslashrgensen (2010)

discussed in regard to the tensions between accountability and efficiency It is therefore not

possible to include all public values in one dimension (Beck Joslashrgensen and Bozeman 2007)

Accordingly the items of the CPV dimension are developed for estimating individual

commitment to seek public values which there are general consensus about and which are

generally formulated

When all four dimensions of PSM are included in a survey cross-national

equivalence of the survey can be assured The items represent reflections or manifestations of

each dimension and hence are relatively homogeneous and interchangeable Adding or

deleting items does not change the essential nature of the dimension Thus any combination

of items in a dimension can be regarded as equivalent When using structural equation

modeling (SEM) each dimension needs to have at least three reflective indicators in order to

achieve identification or it needs to have at least two reflective indicators and be allowed to

- 19 -

covary with another dimension that has at least two reflective measures With one indicator

the dimension is identified if the loading and the variance of the uniqueness are both fixed

(Bollen 1989 Edwards 2010) Thus the number of items in each dimension is not restricted

when measuring PSM However we recommend using multiple items for each dimension

because more research is needed to investigate the appropriateness of this PSM instrument

with varying samples

The test results showed that model fit estimators were clearly better for the D-WLS

It means that in this study MLE was more rigorous than D-WLS in assessing the fit of the

model The empirical evidence indicated that MLE may be an acceptable technique for

ordinal data when there is a large sample a five-point Likert scale and the skewness and

kurtosis of variables is within acceptable limits as suggested by Byrne (2010)

Discussion

The analysis reaffirmed the four-dimensional structure of PSM but the results were not

entirely consistent with the original CFA we tested We were ultimately able to converge on a

good-fit operational model for PSM that was also consistent with underlying theory The

analysis is not able to resolve the debate about measurement models however for reasons

discussed below

Choosing the correct measurement model (ie reflective vs formative) is important

as measurement model misspecification results not only in misrepresentation of the epistemic

relationships between constructs and their measures but also in biased parameter estimates of

structural relationships (Jarvis et al 2003 Law and Wong 1999 MacKenzie 2003

MacKenzie Podsakoff and Jarvis 2005) A formative measurement model in isolation is

- 20 -

under-identified and cannot be estimated in the case of covariance-based SEM (CB-SEM)

(Bollen 1989) A construct with formative indicators must emit paths to at least two unrelated

reflective indicators two unrelated latent constructs or some combination of the two in order

to solve the problem of under-identification (Jarvis MacKenzie and Podsakoff 2003

MacCallum and Browne 1993 MacKenzie Podsakoff and Jarvis 2005) Thus ―when

dependent constructs change the empirical nature of the formatively measured construct

changes (Wilcox Howell and Breivik 2008 1223) and ―when specifying formative

measurement models what might seem like an innocuous choice of outcomes to achieve

identification can have fundamental ramifications for the meaning of the formative construct

and the inferences based on it (Edwards 2010 7)

Earlier research found that the association between a construct and its measures

should remain stable regardless of the larger causal model in which the constructs and

measures are embedded (Edwards and Bagozzi 2000) The association between a construct

and its measures is generally stable for reflective measures but associations of formative

measures with their construct are unstable The susceptibility of formative measurement to

both interpretational confounding and external inconsistency recommends that formative

measurement is not an equally attractive alternative to reflective measurement and that

whenever possible researchers should opt for reflective measurement (Howell Breivik and

Wilcox 2007a 2007b Kim Shin and Grover 2010) In the case of second-order models

researchers are limited to only a second-order reflective model in CB-SEM (Chin 2010)

Regardless of the measurement debate and the tradeoffs between reflective or

formative specification this study makes a theoretical argument that PSM is a second-order

formative construct This second-order formative factor model (reflective first-order

formative second-order) is also suggested as an alternative model that captures the

- 21 -

multidimensionality of complex concepts and avoids the shortcomings of first-order

formative measurement (Edwards 2010) However this model itself cannot achieve

identification One possible strategy is to develop direct reflective measures of PSM based on

the view that PSM is a general concept that is not only influenced by its specific dimensions

but also can be assessed using general measures (Edwards 2010 Javis MacKenzie and

Podsakoff 2003)6 However these general measures have not yet been developed for the

PSM construct

Another way around the problem of underidentification of formative models is to

apply the component-based approach known as partial least squares path modeling (PLS-PM)

which can model formative indicators (Fornell and Bookstein 1982 Goumltz Liehr-Gobbers

and Krafft 2010) The CB-SEM and the PLS-PM methods have distinctive statistical

characteristics and selecting an approach to SEM depends on the particular research situation

CB-SEM is the method of choice for theory testing while PLS-PM is appropriate for

prediction-oriented applications (Ringle Goumltz Wetzels and Wilson 2009) One advantage of

the PLS method is that it demands significantly fewer requirements compared to that of

covariance structure analyses but nevertheless delivers consistent estimation results Another

asset of the PLS approach is its ability to deal with formative as well as reflective indicators

even within one structural equation model This indicates that the PLS approach can be

appropriate for explorative analysis of structural equation models (Chin 2010 Goumltz Liehr-

Gobbers and Krafft 2010) A Monte Carlo simulation study to compare CB-SEM and PLS-

PM showed that PLS-PM provides a viable approximation of model parameters when the

prerequisites for CB-SEM are not met (Ringle Goumltz Wetzels and Wilson 2009) A formative

6 For example job satisfaction can be assessed with general reflective measures such as ―in general I am

satisfied with my job and the formative dimensions can be assessed with measures that describe satisfaction

with each job facet (Edwards 2010 15)

- 22 -

composite in the form of indexes needs to be tested to know the relative weights (γ-

parameters) of the dimensions of PSM In PLS-PM a second-order formative factor is

directly measured by observed variables for all the first order factors While this approach

repeats the number of manifest variables used the model can be identified without adding

outcome constructs that would be considered inappropriate or irrelevant from a conceptual

standpoint (Chin 2010 Edwards 2010) The test result of PLS-PM using Smart PLS 20

(Ringle Wende and Will 2005) produced the function PSM = 31APS + 36CPV + 28COM

+ 31SS However one consideration is whether and how construct error is modeled for the

formative construct (Diamantopoulos 2006) When using PLS-PM the researcher is making

the assumption that there is no error at the construct level as all constructs in PLS-PM are

modeled without error When using PLS-PM the absence of an error term will tend to

increase the weights therefore inflating their importance (Bollen and Lennox 1991) The

primary statistic for assessing a formative indicator is its weight the partialized effect of the

indicator on its intended construct controlling for the effects of all other indicators of that

construct (Cenfetelli and Bassellier 2009) A formative measure is essentially a multiple

regression with the construct representing the dependent variable and the indicators as the

predictors (Diamantopoulos and Winklhofer 2001) PSM may be regarded as a weighted

linear composite of its dimensions The weights used to create the composites are different

across CE-SEM and PLS-PM The causal dimensional weights generated in CB-SEM are

estimated in a model in which a construct-level error term is specified (eg PSM = γ1APM +

γ2CPI + γ3COM + γ4SS + ζ) whereas the weights in PLS-PM are not (eg PSM = γ1APM +

γ2CPI + γ3COM + γ4SS) This leads to two different conceptualizations of the PSM construct

one in which the causal indicators do not capture the construct in its entirety and another in

which they do (Hardin et al 2010) These differences need to be discussed in future research

- 23 -

in this area

One practical suggestion to measure PSM is to apply unit weights to its dimensions if

we assume the effects of the four dimensions are equivalent A literature review indicates that

unit weights have substantial predictive validity when compared with regression weights The

literature shows that the empirical correlations between scores generated from differential and

unit-weighting approaches are quite high In creating composite scores unit weights can be

an appropriate approach for weighting under many circumstances (Bobko Roth and Buster

2007) Although the uses of unit-weight composites will likely result in the explanation of

less variance than empirical estimations of dimensional weights they will remain consistently

defined across studies and hence suitable for theory testing Alternatively more thorough

discussions on the dimensions of PSM may suggest fixed-weight composites (Hardin et al

2010) A previous study (Kim 2010 22) also suggested ―Overall indexing PSM by

measuring all the dimensions of PSM and simply summing the averages of each dimension

can make it easier to measure PSM in practice as PSM = APS + CPV + COM + SS

An advantage of regarding PSM as a second-order formative construct is that the

dimensions can be analyzed both individually and collectively This is important as the

dimensions are expected to have different antecedents and consequences For instance Perry

(1997) found some significant differences in the influence of independent variables on the

four dimensions of PSM Taylor (2007) confirmed that certain dimensions are more important

than others in influencing work outcomes Moynihan and Pandey (2007a) reported that red

tape is negatively related to one dimension but reform orientation and hierarchical authority

is positively related to another dimension A recent study found that the SS dimension has a

positive relationship with occupational intention but the other dimensions do not (Liu et al

2011) Thus it is valuable to analyze the causes and effects of individual dimensions of PSM

- 24 -

as well as of the PSM construct

Conclusion

The purpose of this study is to develop a measurement instrument for PSM for cross-national

research and comparison On the basis of the sharpened construct of PSM (Kim and

Vandenabeele 2010) using data from an international survey conducted in 12 countries we

found that the four-factor structure of PSM is generalizable across country cases This study

provides the measurement index of PSM with the four dimensions of APS CPV COM and

SS The APS dimension measures the instrumental motives of PSM the CPV dimension

determines the value-based motives of PSM while the COM and SS dimensions gauge the

identification motives and self-sacrifice respectively When all four dimensions of PSM are

used in a survey cross-national equivalence of survey items can be assured The items for

measuring PSM in future research are suggested in Appendix 4

This study has several limitations First looking for samples which were comparable

to the largest possible extent we opted for surveys in local government The results found

and discussed in this study are based on PSM among local government employees The

results might be different when surveying central government employees or other specific

groups in the public sector Thus the potential impact of the sample bias needs to be

considered in understanding this study It also means that the survey items for measuring

PSM should be examined with different samples in various areas and at different government

levels

Second the test results showed that there is a high degree of correlation among the

dimension such as APS CPV and COM Future research should not only investigate whether

- 25 -

the items reflect the core content of the dimensions but also continue to develop more

appropriate items for better discriminant validity of each dimension Third the content of

APP dimension was assumed to be distinct from the CPV dimension that focuses on personal

disposition to pursue public interest and public values However the public interest sub-

dimension was separated from the public values sub-dimension in the CPV dimension and

combined with the APP dimension to form the APS dimension Further research should

analyze the CPV dimension and investigate the appropriateness of the APS dimension

In order to facilitate research internationally the PSM construct needs to be

improved conceptually and operationally Kim and Vandenabeele (2010) have dealt with the

conceptual composition and operational dimension of PSM as well as the measurement

model of PSM The next step is to develop an internationally robust measurement instrument

of PSM This study proposes that scholars use all four of the dimensions in measuring the

conceptual components of PSM such as self-sacrifice instrumental value-based and affective

motives because they collectively represent PSM It provides a set of appropriate items for

measuring PSM and confirms the dimensional structure of PSM The survey results suggest

that people in different cultures understand PSM in the same fashion We know that this study

is simply a step along the way to developing a more robust measure but we hope it

contributes to enhanced cross-national research and comparison along with the generation of

cumulative knowledge internationally

- 26 -

REFERENCES

Anderson James C and David D Gerbing 1988 Structural Equation Modeling in Practice

A Review and Recommended Two-Step Approach Psychological Bulletin 103

411-423

Arbuckle James L 2009 Amos 180 Userrsquos Guide Crawfordville FL Amos Development

Co

Beck Joslashrgensen Torben and Barry Bozeman 2007 Public Values An Inventory

Administration amp Society 39(3) 354-381

Beutel Ann M and Margaret Mooney Marini 1995 Gender and values American

Sociological Review 60 436 - 448

Bobko Philip Philip L Roth and Maury A Buster 2007 The Usefulness of Unit Weights in

Creating Composite Scores A Literature Review Application to Content Validity and

Meta-Analysis Organizational Research Methods 10(4) 689-709

Bollen Kenneth A 1989 Structural Equations with Latent Variables New York Wiley

Bollen K A and R Lennox 1991 Conventional Wisdom on Measurement A Structural

Equation Perspective Psychological Bulletin 110(2) 305-314

Byrne Barbara M 2010 Structural Equation Modeling with Amos Basic Concepts

Applications and Programming 2nd

ed New York Routledge

Camilleri Emanuel 2006 Towards Developing an Organizational Commitment ― Public

Service Motivation Model for Maltese Public Service Employees Public Policy

and Administration 21 63-83

Castaing Seacutebastien 2006 The Effects of Psychological Contract Fulfillment and Public

Service Motivation on Organizational Commitment in the French Civil Service

Public Policy and Administration 21 84-98

Cenfetelli Ronald T and Geneviegraveve Bassellier 2009 Interpretation of Formative

Measurement in Information Systems Research MIS Quarterly 33(4) 689-707

Charles Michael B W Martin de Jong and Neal Ryan 2011 Public Values in Western

- 27 -

Europe A Temporal Perspective American Review of Public Administration 41

75-91

Chin Wynne W 2010 How to Write Up and Report PLS Analysis In Handbook of Partical

Least Squares edited by V Esposito Vinzi et al 655-690 Berlin Springer-Verlag

Coursey David H and Sanjay K Pandey 2007a Public Service Motivation Measurement

Testing an Abridged Version of Perryrsquos Proposed Scale Administration amp Society

39 547-568

Coursey David H and Sanjay K Pandey 2007b Content Domain Measurement and

Validity of the Red Tape Concept American Review of Public Administration 37

342-361

Coursey David H James L Perry Jeffrey L Brudney and Laura Littlepage 2008

Psychometric Verification of Perryrsquos Public Service Motivation Instrument Results

for Volunteer Exemplars Review of Public Personnel Administration 28 79-90

Davis Paul and Karen West 2009 What Do Public Values Mean for Public Action Putting

Public Values in Their Plural Place American Review of Public Administration 39

602-618

DeHart-Davis Leisha Justin Marlowe and Sanjay K Pandey 2006 Gender Dimensions of

Public Service Motivation Public Administration Review 66 873-887

Diamantopoulos Adamantios 2006 The error term in formative measurement models

Interpretation and modeling implications Journal of Modeling in Management 1(1)

7-17

Diamantopoulos Adamantios and Judy A Siguaw 2006 Formative versus Reflective

Indicators in Organizational Measure Development A Comparison and Empirical

Illustration British Journal of Management 17 263-282

Diamantopoulos Adamantios and Heidi M Winklhofer 2001 Index Construction with

Formative Indicators An Alternative to Scale Development Journal of Marketing

Research 38 269-277

Edwards Jeffrey R 2010 The Fallacy of Formative Measurement Organizational Research

- 28 -

Methods Online First published on August 19 2010 as

doi1011771094428110378369

Edwards Jeffrey R and Richard P Bagozzi 2000 On the Nature and Direction of

Relationships between Constructs and Measures Psychological Methods 5 155-

174

Flora David B and Patrick J Curran 2004 An Empirical Evaluation of Alternative

Methods of Estimation for Confirmatory Factor Analysis with Ordinal Data

Psychological Methods 9 466-491

Fornell Claes and Fred L Bookstein 1982 Two structural equations models LISREL and

PLS applied to customer exit-voice theory Journal of Marketing Research 19 440-

452

Giauque David Adrian Ritz Freacutedeacuteric Varone Simon Anderfuhren-Biget and Christian

Waldner 2011 Putting Public Service Motivation into Context A Balance between

Universalism and Particularism Forthcoming in International Review of

Administrative Sciences 77(2)

Goumltz Oliver Kerstin Liehr-Gobbers and Manfred Krafft 2010 Evaluation of Structural

Equation Models Using the Partial Least Squares (PLS) Approach In V Esposito

Vinzi et al (eds) Handbook of Partial Least Squares 691-711 DOI 101007978-3-

540-32827-8_30 Springer-Verlag Berlin Heidelberg

Guinot Christiane Julie Latreille and Michel Tenenhaus 2001 PLS Path Modelling and

Multiple Table Analysis Application to the Cosmetic Habits of Women in Ile-de-

France Chemometrics and Intelligent Laboratory Systems 58 247-259

Hair Joseph F Jr William C Black Barry J Babin and Rolph E Anderson 2010

Multivariate Data Analysis 7th

ed Upper Saddle River NJ Pearson Education

Hardin Andrew M Jerry Cha-Jan Chang Mark A Fuller and Gholamreza Torkzadeh 2010

Formative Measurement and Academic Research In Search of Measurement

Theory Educational and Psychological Measurement Online First published on

June 11 2010 as doi101177001316440370208

- 29 -

Howell Roy D Einar Breivik and James B Wilcox 2007a Reconsidering Formative

Measurement Psychological Methods 12 205-218

Howell Roy D Einar Breivik and James B Wilcox 2007b Is Formative Measurement

Really Measurement Reply to Bollen (2007) and Bagozzi (2007) Psychological

Methods 12 238-245

Jarvis Cheryl Burke Scott B MacKenzie and Philip M Podsakoff 2003 A Critical Review

of Construct Indicators and Measurement Model Misspecification in Marketing and

Consumer Research Journal of Consumer Research 30 199-218

Joumlreskog Karl G and Dag Soumlrbom 2005 LISREL 872 (Computer Software) Lincolnwood

IL Scientific Software International

Kim Gimun Bongsik Shin and Varun Grover 2010 Investigating Two Contradictory Views

of Formative Measurement in Information Systems Research MIS Quarterly 34

345-365

Kim Sangmook 2009a Revising Perryrsquos Measurement Scale of Public Service Motivation

American Review of Public Administration 39 149-163

Kim Sangmook 2009b Testing the Structure of Public Service Motivation in Korea A

Research Note Journal of Public Administration Research and Theory 19 839-

851

Kim Sangmook 2010 Testing a Revised Measurement of Public Service Motivation

Reflective versus Formative Specification Journal of Public Administration

Research and Theory Doi101093jopartmuq048

Kim Sangmook and Wouter Vandenabeele 2010 A strategy for building public service

motivation research internationally Public Administration Review 70 701-709

Kline Rex B 2005 Principles and Practice of Structural Equation Modeling 2nd

ed New

York Guilford Press

Law Kenneth and Chi-Sum Wong 1999 Multidimensional Constructs in Structural

Equation Analysis An Illustration using the Job Perception and Job Satisfaction

Constructs Journal of Management 25 143-160

- 30 -

Leisink Peter and Bram Steijn 2009 Public Service Motivation and Job Performance of

Public Sector Employees in the Netherlands International Review of

Administrative Sciences 75 35-52

Liu Bangcheng Chun Hui Jin Hu Wensheng Yang and Xinli Yu 2011 How well can

Public Service Motivation Connect with Occupational Intention International

Review of Administrative Sciences 77 191-211

Liu Bangcheng Ningyu Tang and Xiaomei Zhu 2008 Public Service Motivation and Job

Satisfaction in China An Investigation of Generalisability and Instrumentality

International Journal of Manpower 29 684-699

MacCallum Robert C and Michael W Browne 1993 The Use of Causal Indicators in

Covariance Structure Models Some Practical Issues Psychological Bulletin 114

533-541

MacKenzie Scott B 2003 The dangers of poor construct conceptualization Journal of the

Academy of Marketing Science 31(3) 323-326

MacKenzie Scott B Philip M Podsakoff and Cheryl Burke Jarvis 2005 The Problem of

Measurement Model Misspecification in Behavioral and Organizational Research

and Some Recommended Solutions Journal of Applied Psychology 90 710-730

Martinsen Dorte Sindbjerg and Torben Beck Joslashrgensen 2010 Accountability as a

Differentiated Value in Supranational Governance American Review of Public

Administration 40 742-760

Moynihan Donald P and Sanjay K Pandey 2007a The Role of Organizations in Fostering

Public Service Motivation Public Administration Review 67 40-53

Moynihan Donald P and Sanjay K Pandey 2007b Finding Workable Levers over Work

Motivation Comparing Job Satisfaction Job Involvement and Organizational

Commitment Administration amp Society 39 803-832

Pandey Sanjay K and Edmund C Stazyk 2008 Antecedents and Correlates of Public

Service Motivation In Motivation in Public Management edited by James L Perry

and Annie Hondeghem 101-117 Oxford Oxford University Press

- 31 -

Perry James L 1996 Measuring Public Service Motivation An Assessment of Construct

Reliability and Validity Journal of Public Administration Research and Theory 6

5-22

Perry James L 1997 Antecedents of Public Service Motivation Journal of Public

Administration Research and Theory 7 181-197

Perry James L 2000 Bringing Society in Toward a Theory of Public Service Motivation

Journal of Public Administration Research and Theory 10 471-488

Perry James L and Annie Hondeghem 2008a Motivation in Public Management Oxford

Oxford University Press

Perry James L and Annie Hondeghem 2008b Directions for Future Theory and Research

In Motivation in Public Management edited by James L Perry and Annie

Hondeghem 294-313 Oxford Oxford University Press

Perry James L Annie Hondeghem and Lois Recascino Wise 2010 Revisiting the

Motivational Bases of Public Service Twenty Years of Research and an Agenda for

the Future Public Administration Review 70(5) 681-690

Perry James L and Wouter Vandenabeele 2008 Behavioral dynamics Institutions

Identities and Self-Regulation In Motivation in Public Management edited by

James L Perry and Annie Hondeghem 56-79 Oxford Oxford University Press

Perry James L and Lois Recascino Wise 1990 The Motivational Bases of Public Service

Public Administration Review 50 367-373

Rainey Hal G 1982 Reward Preference among Public and Private Managers In Search of

the Service Ethic American Review of Public Administration 16 288-302

Ringle Christian M Oliver Goumltz Martin Wetzels and Bradley Wilson 2009 On the Use of

Formative Measurement Specifications in Structural Equation Modeling A Monte

Carlo Simulation Study to Compare Covariance-Based and Partial Least Squares

Model Estimation Methodologies Faculty of Economics and Business Administration

Maastricht Univeristy

Ringle C M S Wende and S Will 2005 Smart PLS 20 (M3) Beta

- 32 -

httpwwwsmartplsde

Ritz Adrian 2011 Attraction to Public Policy Making A Qualitative Inquiry into

Improvements in PSM Measurement Forthcoming in Public Administration

Spector Paul E 1992 Summated Rating Scales Construction An Introduction Newbury

Park CA Sage

Steenbergen Marco R 1996 Compassion and American public opinion An analysis of the

NES Humanitarianism Scale Report to the Board of Overseers of the National

Election Studies

Steffen Patrick R and Kevin S Masters 2005 Does compassion mediate the intrinsic

religion-health relationship Annals of Behavioral Medicine 30 217-223

Taylor Jeannette 2007 The Impact of Public Service Motives on Work Outcomes in

Australia A Comparative Multi-Dimensional Analysis Public Administration 85

931-959

Vandenabeele Wouter 2007 Towards a Theory of Public Service Motivation An

Institutional Approach Public Management Review 9 545-556

Vandenabeele Wouter 2008a Government Calling Public Service Motivation as an Element

in Selecting Government as an Employer of Choice Public Administration 86

1089-1105

Vandenabeele Wouter 2008b Development of a Public Service Motivation Measurement

Scale Corroborating and Extending Perryrsquos Measurement Instrument International

Public Management Journal 11 143-167

Vandenabeele Wouter 2009 The Mediating Effect of Job Satisfaction and Organizational

Commitment on Self-Reported Performance More Robust Evidence of the PSM-

Performance Relationship International Review of Administrative Sciences 75 11-

34

Wilcox James B Roy D Howell and Einar Breivik 2008 Questions about Formative

Measurement Journal of Business Research 61 1219-1228

Wise Lois Recascino 2000 The Public Service Culture In Public Administration Concepts

- 33 -

and Cases 7th ed edited by Richard J Stillman II 342-353 Boston Houghton

Mifflin

Wright Bradley E 2008 Methodological Challenges associated with Public Service

Motivation Research In Motivation in Public Management edited by James L

Perry and Annie Hondeghem 80-98 Oxford Oxford University Press

Wright Bradley E and Sanjay K Pandey 2008 Public Service Motivation and the

Assumption of Person-Organization Fit Testing the Mediating Effect of Value

Congruence Administration amp Society 40 502-521

- 34 -

Table 1

Possible Indicators of PSM and Descriptive Statistics (n = 2868)

Dimensions and items Mean Standard

Deviation

Item-Total

Correlation

Attraction to Public Participation (APP)

APP1 I am interested in helping to improve public service

APP2 I am satisfied when I see people benefiting from the public programs I was

involved in

APP3 I like to discuss topics regarding public programs and policies with others

APP4 I believe that public sector activities contribute to our general welfare

APP5 I admire people who initiate or are involved in activities to aid my

community

APP6 Contributing to public programs and policies helps me realize myself

APP7 It is important to contribute to activities that tackle social problems

412

441

355

393

423

351

409

696

662

917

753

672

881

734

524

421

482

412

493

478

552

Commitment to Public Values (CPV)

Sub-dimension for Public Interests (CPV1)

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

CPI3 I would prefer seeing public officials do what is best for the whole

community even if it harmed my interests

CPI4 Serving the public interest is more important than helping a single individual

Sub-dimension for Public Values (CPV2)

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of public

services

CPV3 It is fundamental that public services respond to the needs of the citizens

CPV4 Decisions regarding public services should be democratic despite the time

and effort it takes

CPV5 Everybody is entitled to a good service even if it costs a lot of money

CPV6 It is fundamental that the interests of future generations are taken into

account when developing public policies

CPV7 To act ethically is essential for public servants

CPV8 I believe that public employees must always be aware of the legitimacy of

their activities

CPV9 I personally identify with the aim of protecting individual liberties and rights

418

409

348

337

428

416

433

394

367

426

441

434

414

695

662

899

995

730

704

710

852

932

689

681

692

763

525

596

480

251

484

437

356

325

376

439

462

402

392

Compassion (COM) COM1 It is difficult for me to contain my feelings when I see people in distress

COM2 I feel sympathetic to the plight of the underprivileged

COM3 I empathize with other people who face difficulties

COM4 I have little compassion for people in need who are unwilling to take the

first step to help themselves

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

354

387

388

299

412

391

951

825

707

1083

755

765

399

511

389

011

487

502

Self-Sacrifice (SS)

SS1 Making a difference to society means more to me than personal achievements

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS5 People should give back to society more than they get from it

SS6 Serving other citizens would give me a good feeling even if no one paid me for

it

SS7 I would agree to a good plan to make a better life for the poor even if it costs

me money

357

329

339

302

345

363

350

904

872

949

982

888

1049

895

508

584

454

531

412

368

546

- 35 -

Table 2

Background of Respondents (n = 2868)

Variables Characteristics Respondents

Sex

Male

Female

NA

1209

1622

37

422

566

13

Length of service

(years)

0 ~ 10

10 ~ 20

20 ~ 30

30+

NA

1126

723

484

491

44

393

252

169

171

15

Organizational status

―Do you supervise employees

No

Yes

NA

1472

1076

320

513

375

112

Country

Australia

Belgium

China

Denmark

France

Italy

Korea

Lithuania

The Netherlands

Switzerland

United Kingdom

USA

249

214

230

249

266

162

253

236

249

250

260

250

87

75

80

87

93

56

88

82

87

87

91

87 Note NA = no answer

- 36 -

Table 3

Standardized factor loadings and correlations for the 31-item PSM index (n = 2868)

Dimensions and items SFL (MLE) SFL (D-WLS) Alpha

APP

APP1

APP2

APP4

APP5

APP6

APP7

597

486

455

558

490

600

657

552

514

625

549

665

699

CPV1

CPI1

CPI2

CPI3

CPI4

588

660

488

289

653

732

535

294

542

CPV2

CPV1

CPV2

CPV3

CPV4

CPV5

CPV6

CPV7

CPV8

CPV9

611

546

436

393

418

534

598

496

460

690

631

505

457

486

601

669

573

530

745

COM

COM1

COM2

COM3

COM5

COM6

496

569

498

597

607

516

651

538

670

684

687

SS

SS1

SS2

SS3

SS4

SS5

SS6

SS7

567

736

558

728

467

409

663

660

762

591

695

515

494

709

782

Measures of fit (MLE) χsup2 (df = 424) = 43699 p lt 001 CFI = 835 GFI = 899

RMSEA = 057 [055 059]

Measure of fit (D-WLS) χsup2 (df = 424) = 39247 p lt 001 CFI = 971 GFI = 976

RMSEA = 054 [052 055]

Inter-factor correlations MLE (D-WLS)

1 2 3 4

1 APS

2 CPV1 996 (988)

3 CPV2 902 (890) 798 (798)

4 COM 787 (783) 758 (762) 779 (783)

5 SS 700 (742) 800 (859) 509 (537) 720 (729)

Note SFL = standardized factor loading All factor loadings and correlations are significant at p lt 001

- 37 -

Table 4

Standardized factor loadings and correlations for the 16-item PSM index (n = 2868)

Dimensions and items SFL

(MLE)

SFL

(D-WLS)

APS

APP5 I admire people who initiate or are involved in activities to aid my

community

APP7 It is important to contribute to activities that tackle social problems

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

611

592

637

666

665

669

690

751

CPV

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of

public services

CPV6 It is fundamental that the interests of future generations are taken

into account when developing public policies

CPV7 To act ethically is essential for public servants

612

546

551

590

698

631

618

658

COM

COM1 It is difficult for me to contain my feelings when I see people in

distress

COM2 I feel sympathetic to the plight of the underprivileged

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

490

546

596

604

515

608

669

679

SS

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS7 I would agree to a good plan to make a better life for the poor even if

it costs me money

763

571

766

678

807

629

747

755

Measures of fit (MLE)

χsup2 (df = 98) = 8490 p lt 001

CFI = 939 GFI = 965

RMSEA = 052 [049 055]

Measure of fit (D-WLS)

χsup2 (df = 98) = 6069 p lt 001

CFI = 988 GFI = 992

RMSEA = 043 [039 046]

Alpha CR (D-WLS) Inter-factor correlations MLE (D-WLS)

1 APS 717 788 1 APS 2 CPV 3 COM

2 CPV 663 747 848 (859)

3 COM 645 713 781 (791) 789 (795)

4 SS 784 826 565 (596) 462 (505) 715 (726)

Note SFL = standardized factor loading Alpha = Cronbachrsquos alpha CR = composite reliability

All factor loadings and correlations are significant at p lt 001

- 38 -

Table 5

Results of CFA in the individual countries Standardized factor loadings and measures of fit

Factors

amp Items

Australia Belgium China Denmark France Italy

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

672

756

799

785

804

943

710

703

597

790

799

730

584

675

685

681

731

742

578

748

674

643

765

739

645

704

678

708

812

779

632

604

765

766

672

853

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

552

593

560

699

727

632

624

605

560

713

709

645

702

602

616

724

686

735

567

560

715

654

605

811

608

658

587

696

755

698

673

620

523

782

697

654

COM

COM1

COM2

COM3

COM5

COM6

623

584

662

688

701

739

732

608

475

808

666

581

578

616

605

649

670

673

699

751

737

679

816

776

568

546

801

626

673

850

637

546

741

715

601

822

SS

SS2

SS3

SS4

SS7

654

794

806

892

767

813

727

693

799

793

725

838

857

820

773

896

855

842

806

749

792

829

753

884

847

731

753

889

816

753

815

692

664

795

782

734

Inter-factor

correlations

093

~

752

163

~

810

320

~

773

323

~

825

452

~

827

497

~

855

323

~

850

321

~

847

498

~

899

536

~

902

425

~

829

476

~

865

Sample size 249 214 230 249 266 162

Measures of Fit

χsup2 638 444 881 732 765 554 817 625 735 565 750 564

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 980 100 939 987 969 997 967 994 975 997 949 995

GFI 961 986 941 986 948 991 949 992 955 992 926 986

RMSEA 036 00 063 050 051 026 053 035 045 026 059 033

- 39 -

Table 5 (Continued)

Factors

amp Items

Korea Lithuania The Netherlands Switzerland UK USA

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

619

701

667

706

781

717

666

512

682

751

605

779

557

629

630

620

719

729

535

621

702

643

694

735

655

725

738

710

804

822

535

621

702

643

694

735

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

594

607

722

722

644

722

674

535

533

790

589

711

568

554

538

655

628

545

627

580

615

711

666

704

800

512

555

873

581

646

627

580

615

711

666

704

COM

COM1

COM2

COM3

COM5

COM6

694

677

550

763

736

627

557

639

428

550

696

534

737

623

508

766

616

608

659

592

683

703

661

769

682

583

729

739

644

803

659

592

683

703

661

769

SS

SS2

SS3

SS4

SS7

776

752

679

779

770

811

707

721

635

725

734

706

814

656

743

820

767

764

745

632

689

788

662

744

671

818

832

830

864

802

745

632

689

788

662

744

Inter-factor

correlations

432

~

726

466

~

746

376

~

788

406

~

902

377

~

852

455

~

836

637

~

818

660

~

828

317

~

852

360

~

852

637

~

818

660

~

828

Sample size 253 236 249 250 260 250

Measures of Fit

χsup2 896 573 612 462 773 616 815 715 769 542 815 715

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 951 996 975 100 958 993 957 991 972 998 957 991

GFI 942 988 960 989 951 987 948 990 955 990 948 990

RMSEA 059 028 034 00 050 034 053 044 048 022 053 044

Note All factor loadings and correlations are significant at p lt 001

- 40 -

Appendix 1 Dimensions and possible items of PSM (the first round)

1 Attraction to Public Participation

(1) I am interested in developing public programs helping my country or community (Kim 2009a)

(2) Sharing my views on public policies with others is attractive to me (Kim 2009a)

(3) I am satisfied when seeing people get benefits from the public programs I was involved in (Kim 2009a)

(4) I like to discuss political subjects with others (Giauque et al 2011)

(5) I think the governmental activities contribute to our welfare

(6) I like to initiate actions to help out my community

(7) I do not care much about politicians (Perry 1996)

2 Commitment to Public Values

Possible items for public interest

(1A) Meaningful public service is very important to me (Perry 1996)

(2A) It is important for me to contribute for the common good (Giauque et al 2011)

(3A) I recognize myself with the mission of protection of liberties and rights endorsed by the nation-state

(Giauque et al 2011)

(4A) I would prefer seeing public officials do what is best for the whole community even if it harmed my

interests (Perry 1996)

(5A) To me serving the public interest is more important than helping other people (Vandenabeele 2008b)

Possible items for public values

(1B) I recognize myself with the promotion of the equality of chances endorsed by the nation-state (Giauque et

al 2011)

(2B) It is fundamental that the public service provided to the citizen is regular and continuous (Giauque et al

2011)

(3B) It is fundamental that public service innovates to respond to the new needs of the citizensclients (Giauque

et al 2011)

(4B) The dignity and well-being of all should be the most important concerns in any society (Steenbergen

1996)

(5B) It is important that public servants account for all the costsexpenses they make (Vandenabeele 2008b)

(6B) We have to do everything in our power to pursue the goal of democracy

3 Compassion

(1) It is difficult for me to contain my feelings when I see people in distress (Perry 1996)

(2) I feel sympathetic to the plight of the underprivileged (Kim 2009a)

(3) I am often reminded by daily events how dependent we are on one another (Perry 1996)

(4) I feel compassion for others in the difficulties they are facing (Steffen and Masters 2005)

(5) It is not really my problem if others are in trouble and need help (Beutel and Marini 1995)

(6) I find it hard to be sympathetic toward starving people in foreign lands when there is so much trouble in our

own country (Beutel and Marini 1995)

(7) I get very upset when I see other people treated unfairly (Beutel and Marini 1995)

(8) I would agree to a good plan to make a better life for the poor even if it cost me money (Beutel and Marini

1995)

(9) To me patriotism includes seeing the welfare of others (Perry 1996)

(10) I care very much about other people

4 Self-Sacrifice

(1) Making a difference in society means more to me than personal achievements (Perry 1996)

(2) I am prepared to make enormous sacrifices for the good of society (Perry 1996)

(3) I believe in putting duty before self (Perry 1996)

(4) I am one of those rare people who would risk personal loss to help someone else (Perry 1996)

(5) I think people should give back to society more than they get from it (Perry 1996)

(6) Serving other citizens would give me a good feeling even if no one paid me for it (Perry 1996)

(7) Much of what I do is for a cause bigger than myself (Perry 1996)

- 41 -

Appendix 2 Newly suggested items through the first round

1 Items for the Attraction to Public Participation Dimension

(1) In the public domain every person should be counted as equal

(2) I enjoy to think and argue in a political way

(3) I am satisfied when seeing my ideas being integrated in public policies

(4) I can easily relate to the way of thinking and arguing of politicians

(5) Working in the public sector I can better realize myself

2 Items for the Public Values Sub-Dimension

(1) Public service should give value for money

(2) Cost considerations should play an important role in decisions about public service provision

(3) It is fundamental that the interests of future generations are taken into account when developing public

policies

(4) To act with correct ethics is for public servants as important as their professional competence

(5) Primarily public servants should be accountable to the public and not to their superiors

3 Items for the Self-Sacrifice Dimension

(1) I would agree to a good plan to make a better life for the poor even if it costs me money

- 42 -

Appendix 3 An example of a summary of comments and suggestions (the second round)

On the item ―I am interested in developing public programs helping my country or community

A It is a good sentence but is it possible for everyone to ―develop public programs Even public officials are

limited to do this

B OK

C or the special group

D Double barrel and difficult to translate to my country ―I am interested in helping to improve public services

E Good but I feel that wording requires revision If we draw sample from local government then country is

redundant ―I am interested in developing public policy programs that aim in helping my community

F 4 (General appreciation from1 to 4) Not clear whether it refers to ones work or to ones concern as a citizen

―(In my work) I am interested in contributing to the development of public etchellip

G Instead of development I would use implementing

H Delete public which is redundant because of the characteristic helping my country or community but add

or implementing ―I am interested in developing or implementing programs helping my country or

community

I Referring to dimension Wording Similar to item 6

Note Names have been changed to alphabets in order to ensure the anonymity of participants

- 43 -

Appendix 4 Dimensions and Items of Public Service Motivation for Future Research

Dimensions and items

Attraction to Public Service (APS)

1 I am interested in helping to improve public service

2 I admire people who initiate or are involved in activities to aid my community

3 It is important to contribute to activities that tackle social problems

4 Meaningful public service is very important to me

5 It is important for me to contribute to the common good

Commitment to Public Values (CPV)

1 I think equal opportunities for citizens are very important

2 It is important that citizens can rely on the continuous provision of public services

3 It is fundamental that the interests of future generations are taken into account when

developing public policies

4 To act ethically is essential for public servants

5 I believe that public employees must always be aware of the legitimacy of their activities

6 I personally identify with the aim of protecting individual liberties and rights

Compassion (COM)

1 It is difficult for me to contain my feelings when I see people in distress

2 I feel sympathetic to the plight of the underprivileged

3 I empathize with other people who face difficulties

4 I get very upset when I see other people being treated unfairly

5 Considering the welfare of others is very important

Self-Sacrifice (SS)

1 I am prepared to make sacrifices for the good of society

2 I believe in putting civic duty before self

3 I am willing to risk personal loss to help society

4 I would agree to a good plan to make a better life for the poor even if it costs me money

- 5 -

making are not appropriate for measuring personal attraction to public policy making

(Camilleri 2006 Coursey and Pandey 2007a Coursey et al 2008 Kim 2009b Ritz 2011)

Moreover the dimension of commitment to public interest needs to concentrate more on a

personal disposition to pursue public values (Castaing 2006 Leisink and Steijn 2009 Taylor

2007) and the items of compassion do not represent affective motives (DeHart-Davis

Marlowe and Pandey 2006 Moynihan and Pandey 2007a Wright 2008) Thus the

dimensions of the PSM construct are modified as attraction to public participation

commitment to public values compassion and self-sacrifice Consequently it is necessary to

develop appropriate items for the dimensions (Kim and Vandenabeele 2010)

A related issue concerns the measurement model of PSM There are two

measurement models using multiple dimensions of latent constructs reflective and formative

When variation in a construct leads to variation in its dimensions such dimensions are

termed ―reflective while dimensions are labeled ―formative when they are viewed as

causes of a construct (Edwards and Bagozzi 2000 Diamantopoulos and Siguaw 2006

Diamantopoulos and Winklhofer 2001) It is understandable that each dimension of PSM has

several reflective indicators that are essentially interchangeable A problem is the relationship

between PSM and its dimensions If we regard the dimensions as reflective it is possible to

analyze PSM with only two or three dimensions because omitting dimensions does not

change the essential nature of PSM However if we regard the dimensions as formative it is

not acceptable to omit a dimension as this may change the meaning of PSM Construct

conceptualization is regarded as being the process that determines the formative or reflective

nature (Diamantopoulos and Siguaw 2006) Recent research has demonstrated theoretical and

empirical evidence in support of a second-order formative approach to PSM (Kim 2010)

This means that the dimensions represent different conceptual components of PSM and so

- 6 -

they are not interchangeable Thus PSM is regarded as a combination of specific dimensions

If any one of the dimensions changes PSM would change Conversely if an individualrsquos

PSM increases this would not necessarily be accompanied by an increase in all dimensions

Therefore as the PSM construct is refined for international use (Kim and Vandenabeele

2010) scholars must be attentive to identifying and using appropriate measurement models

Developing Possible Items

The conceptual components of PSM need to be empirically estimated through operational

definition Kim and Vandenabeele (2010) proposed that PSM is based on self-sacrifice and

associated with instrumental value-based and identification motives and that the dimensions

of the PSM construct be refined along the lines of attraction to public participation (APP)

commitment to public values (CPV) compassion (COM) and self-sacrifice (SS) As Perry

(1996) identified a multidimensional scale to measure PSM it is conceived as an overall

latent variable with various latent dimensions and is referred to as a second-order factor

model Research has found that the measurement model specifies the causal relationship

between PSM and its dimensions (Edwards and Bagozzi 2000) The direction of the

relationship is either from the PSM construct to the dimensions (second-order reflective

measurement) or from the dimensions to the construct (second-order formative measurement)

A theoretical consideration of the relationships between PSM and its dimensions led

to the proposal that PSM be defined in terms of multiple dimensions (Kim 2010 Kim and

Vandenabeele 2010) The dimensions represent different aspects of PSM with each

dimension capturing a distinct and potentially unique form of PSM Moreover the

dimensions may have different antecedents and consequences as well as different

characteristics and theoretical backgrounds (Moynihan and Pandey 2007a Pandey and

- 7 -

Stazyk 2008 Perry 1996 1997 2000 Perry and Wise 1990 Perry and Vandenabeele 2008

Taylor 2007 Vandenabeele 2007 2008a 2009) An individualrsquos PSM is shaped by the

individualrsquos APP CPV COM and SS Deleting a dimension means omitting a certain part of

PSM Therefore it is reasonable to suggest that the dimensions of PSM are formative and for

PSM to be regarded as a composite of its dimensions

Each dimension of PSM can be measured using reflective indicators that satisfy the

following criteria (Jarvis MacKenzie and Podsakoff 2003) a relative homogeneity and

interchangeability of indicators pertaining to each dimension a high degree of covariation

among indicators of each dimension and the expectation that the indicators of each

dimension are likely to be affected by the same antecedents and have the same consequences

Reflective indicators are essentially interchangeable and so adding or removing indicators

does not change the essential nature of a dimension Thus we consider PSM to be a second-

order formative construct with several dimensions that each has several reflective indicators

(reflective first-order formative second-order)

Based upon the refinement and clarification of the operational dimensions of PSM

(Kim and Vandenabeele 2010) items were developed for the four dimensions The items

selected were a mix of pre-existing items that loaded highly on PSM dimensions in multiple

studies (Giauque et al 2011 Kim 2009a Perry 1996 Vandenabeele 2008b) Where

appropriate items were not available new items were drafted from other previous studies

creating an initial total of 35 proposed items For the dimension of APP which focuses on

individual disposition to work in the public sector to participate in the public policy process

and in activities for community and social development seven possible items were suggested

The CPV dimension focuses on personal disposition to pursue public interest and public

values Two sub-dimensions of CPV seemed plausible one focusing on public interest and

- 8 -

another on public values such as social equity equal opportunity and democracy Thus five

possible items for the public interest sub-dimension (CPV1) and six items for the public

values sub-dimension (CPV2) were developed For the dimension of COM which focuses on

affective bonding with the identified objects such as other members of a social category or of

a political system ten possible items were suggested For the SS dimension Perryrsquos original

items (1996) were the primary source of items The full list of items is provided in Appendix

1

Review Processes

A group of researchers who shared our interest in better measurement of PSM collaborated

using e-mail in November 20091 The objective was to develop a measurement instrument

for PSM with a robust comparative character in other words to create a valid instrument (on

a national basis) that can be used internationally The first round for reviewing the possible

items lasted from the 16th

of November until the 7th

of December 2009 and all the

participants were asked to review these proposed items For this review two criteria were

employed (1) Do these items refer to the dimensions they claim to refer to and (2) Are

these items meaningful across various national contexts and in particular to the situation of

civil servants in that country Participants were also asked to add new items that might better

reflect a particular dimension or to rephrase items that were already on the list

A second round of item refinement lasted from the 9th

of December until the 22nd

of

December 2009 During this time 11 newly suggested items from the first round were

considered (see Appendix 2) A summary of all comments and suggestions emerging from the

1 Emmanuel Camilleri David Giauque Gerhard Hammerschmid Isabel Egger-Peitler Renate Meyer and

Cristian Pliscoff also participated in the review processes

- 9 -

first process was distributed to the participants to stimulate comments in the second round

(see Appendix 3) Based on the comments received in the second round discussion the

possible items for an international survey were selected However some items required

further discussion and so a third round was conducted from the 15th

of January until the 23rd

of January 2010 One issue that remained unresolved in the discussions relates to the

dimension of CPV Some argued that it is reasonable to divide CPV into two sub-dimensions

because committing to public interests is different from pursuing public values and there are

many important public values in democratic countries while others contended that CPV can

cover all important public values including public interest and there is no real benefit to add

one more dimension These differences were left to be resolved through empirical research

This did not affect the final items however and 33 items were confirmed The resulting

survey questionnaire was distributed to all the collaborators on February 9 2010 in order to

conduct a survey in each country

Measures

PSM was measured using the newly developed 33-item index 7 items for the dimension of

APP 13 items for CPV (4 for CPV1 and 9 for CPV2) 6 items for COM and 7 for SS In the

survey questionnaire a five-point Likert type scale (1 = strongly disagree 5 = strongly agree)

was used This scale is among the most widely used response formats in many empirical

investigations within the social and behavioral science in which there are no ―correct or

―incorrect responses (Spector 1992) The items of PSM were randomly distributed in the

survey format Gender age education length of service and employment status were also

measured in this survey

It was necessary to translate the measurement items into the official (native) language

- 10 -

of each country As this raised the possibility of inaccurate translations the back-translation

of the items into native languages was validated by asking two researchers to independently

translate the English-language items The researchers were instructed to consult with each

other over differences between translated items and to test the survey items with some native

respondents

[Table 1 about here]

Samples

The international survey was fielded from March to September 2010 The data upon which

this study are based were collected from civil servants in local governments in Australia

Belgium China Denmark France Italy Korea Lithuania the Netherlands Switzerland the

United Kingdom and the USA The principal criterion was that the samples collected in the

various countries should be comparable to the largest possible extent The criterion ruled out

national and state government as the competencies of national governments and also of state

governments if they exist constitutionally are different among most of the countries included

in the research Therefore we opted for local government as our focus The structure and

functions of local governments can differ widely even within countries and so the survey in

each country was focused on town hall bureaucrats in city county or township government

The respondents were permanent employees from both managerial and non-managerial levels

and from both administrative departments (ie no direct contact with the public) and service

departments (ie likely to have direct contact with the public) Despite being employed by

local governments in many or all of the countries police officers firefighters school teachers

public transport workers artists and musicians as well as nurses and doctors were excluded

- 11 -

from the sample

We aimed to receive 250 respondents for each country as this would enable us to

carry out the analyses we wanted to perform on the data Although there is no ideal sample

size one study recommends a sample of 200 to provide a sound basis for estimation (Hair

Black Babin and Anderson 2010) The method of data collection was left open Obviously a

web-survey is the easiest way to collect the data because it is economical to collect and to

code however compared to paper-pencil surveys this method has some drawbacks First not

everybody in the sample may have access to the internet which may cause a bias in the

responses Second internet surveys are generally known to produce lower response rates

Web surveys were recommended if possible but paper-pencil surveys were deemed

acceptable in cases where it was not possible to conduct a web survey Researchers were

encouraged to contact just one municipality except in cases where it was not possible to

gather enough data In such instances responses from multiple municipalities were accepted

By the end of September 2010 a total of 2868 responses were obtained2

2 In Australia seven local councils each with between from 70 and 232 employees were surveyed Paper-pencil

surveys were used in two councils while the rest relied on a web survey The local governments limited

distribution of paper-pencil surveys to only one or two sections of the council eg administration The response

rate was approximately 30 However it should be noted that not everyone working for the council has direct

access to a computer In Belgium three municipalities were surveyed on a census base by means of a web

survey as all employees had access to a computer This rendered a response of 120 (226) 79 (75) and 16

(53) responses for the respective municipalities resulting in an overall response rate of 331 with 662

invitations being sent out In China two departments social work and administration were surveyed in one

municipality using paper-pencil surveys In the social work department 65 out of 110 surveys were returned

and 165 out of 211 responded from the administration department giving an overall response rate of 767 In

Denmark a total of 1282 respondents in a municipality were sent an initial e-mail invitation to a web survey

and up to two e-mail reminders and the response rate was 473 In France all the employees in two

municipalities were surveyed In one municipality where both web and paper-pencil surveys were utilized 128

out of 685 questionnaires were returned while in another city using only a paper-pencil survey 138 out of 652

were received The average response rate was 199 In Italy in principle the link to the questionnaire was

made available to all employees (about 4000) in one city but for some of them access may have been difficult

yielding a response rate of 45 In Korea 500 employees in a city received a printed questionnaire and 253

responses were returned yielding a response rate of 506 In Lithuania 236 responses from 54 municipalities

were gathered by web survey In the Netherlands three municipalities were surveyed on a census base by means

of a web survey In total 319 responses of 910 invitations were returned (35) mdash 90 (45) 115 (38) and 114

(28) for the respective municipalities In Switzerland 516 out of 1326 local employees with e-mail access in

a municipality completed the survey giving a response rate of 389 In the United Kingdom the survey was

posted online and 1360 employees both in administrative jobs and front-line services of one municipality

- 12 -

Municipalities were selected because of their convenient accessibility and proximity to the

researcher and a convenience sample was used in most countries Seven countries surveyed

one municipality but seven local councils were surveyed in Australia Web surveys were used

in four countries while three countries used both web and paper-pencil survey The number

of respondents varied among the countries depending on the numbers of questionnaires

distributed the survey method and the response rate In most countries we used

approximately 250 responses but only 162 from Italy3 The majority of the respondents were

women (566) while in terms of service length those in the largest group (393) had

worked for fewer than 10 years Table 2 shows the distribution of respondentsrsquo sex length of

service employment status and country

[Table 2 about here]

Analyses

The statistical analysis applied confirmatory factor analysis (CFA) CFA was used to assess

the fit of the data to the hypothesized measurement model because it is preferred where

measurement models have a well-developed underlying theory for hypothesized patterns of

loadings The data were analyzed using Amos 180 (Arbuckle 2009) with the maximum

likelihood (MLE) estimation method and LISREL 872 (Joumlreskog and Soumlrbom 2005) with the

diagonally weighted least squares (D-WLS) estimation method

The individual items are designed to measure a theoretically continuous construct

but the observed responses are a discrete realization of five categories The common method

received an email to complete the survey The response rate was 19 In the USA the total number of

employees surveyed was 626 in one city using both electronic and paper-pencil surveys resulting in a response

rate of 588 3 In countries where significantly more than 250 surveys were collected we randomly selected only 250 cases

for analysis

- 13 -

of estimation within CFA is MLE a technique which assumes that the observed variables are

continuous and normally distributed These assumptions are not met when the observed data

are discrete as occurs when using ordinal scales as in this study A simulation study (Flora

and Curran 2004) found that D-WLS performs well for ordinal measures across a variety of

conditions especially for smaller samples Coursey and Pandey (2007a 2007b) argued the

importance of using D-WLS and not standard MLE in structural equations for ordinal data4

In most cases however the differences between MLE and D-WLS are not substantive to the

conclusions5 No comparison between standard MLE and D-WLS is provided in this paper

We give results from both MLE and D-WLS and this will aid in confirming the stability and

robustness of the results across varying mathematical specifications

For gauging reliability Cronbachrsquos alpha and composite reliability were tested

Composite reliability should be equal to or greater than 07 (Hair et al 2010) Convergent

validity was assessed from the measurement model by determining the statistical significance

of each indicatorrsquos estimated pattern coefficient on its specified underlying construct

Discriminant validity was assessed by determining whether the confidence interval around

the correlation estimate between two factors included 100 (Anderson and Gerbing 1988)

For model fit assessment goodness-of-fit index (GFI) comparative fit index (CFI)

and the root mean square error of approximation (RMSEA) were used The model achieves

an acceptable fit to the data when GFI and CFI equals or exceeds 090 and RMSEA values

4 Applying standard MLE to ordinal data such as Likert-type items may produce significant estimation

problems such as inflation of chi-square fit statistics and biased underestimation of parameters and standard

errors (Flora and Curran 2004) Coursey and Pandey (2007a 2007b) used a modified MLE using the asymptotic

covariance matrix and D-WLS for adjusting the ordinal nature of the data They reported that although the

modified MLE findings are slightly different on a few model parameters the differences are not substantive to

the conclusions However model fit estimators were clearly better for the D-WLS 5 Given normally distributed categorical variables MLE can be used with confidence when a variable has four

or more categories If there is a large sample a five-point Likert scale and the skewness and kurtosis of variables

is within acceptable limits it suggests distribution symmetry ―In most cases where the hypothesized model is

well specified and the scaling based on more than three categories it seems unlikely that there will be much

difference between the findings (Byrne 2010 160)

- 14 -

fall below 008 (Hair et al 2010) In general the larger the value of GFI and CFI and the

smaller the value of RMSEA the better the fit of the model (Bollen 1989) We also present

the χ2 test statistic Lower values of χ

2 indicate a better fit and should be nonsignificant but

the χ2 value increases as sample size increases and thus for large sample sizes this statistic

may lead to rejection of a model with good fit (Hair et al 2010)

Results

Using SPSS 180 descriptive statistics were computed for individual items as shown in table

1 One item (APP3) in the APP dimension ―I like to discuss topics regarding public programs

and policies with others was mistakenly omitted in the survey conducted in the United

States and so it was deleted in the statistical analysis In the assessment of normality no

items showed a skew or kurtosis value greater than the cutoffs of |3| or |8| recommended by

Kline (2005) As data were distributed normally MLE as well as D-WLS estimation was used

Based on inspection of item-total correlations one item (COM4) was dropped from further

analysis because it was weakly correlated with the overall index (011) Missing data is an

issue in CFA as it can cause estimation errors All variables in the survey have missing data

rates of approximately 016 to 143 Mean substitution was used to calculate replacement

values for the missing values in each countryrsquos dataset (Hair et al 2010) The deletion of the

two items left 31 items for the next stage of analysis

The five-correlated factor model was tested using CFA The CFA model in the initial

analysis hypothesized a priori that (1) responses to the 30-item PSM measurement

instrument could be explained by five factors such as APP CPV1 CPV2 COM SS (2) each

item would have a nonzero loading on the PSM factor that it was designed to measure and

- 15 -

zero loadings on all other factors (3) the five factors would be correlated and (4)

measurement error terms would be uncorrelated The resulting CFA with MLE suggested that

the initial five-correlated factor model was not a good fit to the data χsup2 (df = 424) = 43699 p

lt 001 GFI = 0899 CFI = 835 RMSEA = 0057 [055 059] At this stage APP was highly

correlated with CPV1 (r = 996 in MLE and r = 988 in D-WLS) suggesting the two

dimensions were not unique and lacked discriminant validity

[Table 3 about here]

Given the disconfirmation of the initial model an effort was made to estimate an

alternative model The two dimensions APP and CPV1 were combined to form a single

dimension Attraction to Public Service (APS) The items of CPV1 may be regarded as

action-oriented by the respondents because dedication to public service community and

common good is underlined in these items Consequently it is reasonable to combine APP

with CPV1 to form APS a dimension that focuses more on disposition to serve the public to

work for the common good and to participate in public policy processes After the overall

number of dimensions was reduced to four using standardized factor loadings and

modification indices as criteria the items with the lowest factor loading in each dimension

were deleted and the items whose modification index indicated association with multiple

dimensions were eliminated A CFA model with the remaining items was iteratively tested

until it achieved a good fit to the data The outcome of this refinement process was a 16-item

index of four factors Overall model fit by every measure is quite strong for the D-WLS

estimation technique The estimation results for the four-dimension model are presented in

table 4

- 16 -

[Table 4 about here]

Cronbachrsquos coefficient alpha for the 16-item PSM index was 860 and the coefficient

alphas for the four dimensions ranged from 645 to 784 The composite reliability of the set

of reflective indicators for each dimension of PSM ranged from 713 to 826 and indicated

adequate internal consistency The resulting factor structure showed a four-factor structure

with all items loading significantly onto their a priori dimension (p lt 001) and the

standardized factor loadings ranging from 0490 to 0766 in MLE and from 515 to 807 in

D-WLS The results provide support for convergent validity The correlation estimates

between the two factors ranged from 0462 to 0848 in MLE and from 505 to 859 in D-

WLS The confidence intervals (plusmn2 standard errors) around the correlation estimates between

the two factors did not include 100 The result provides support for discriminant validity To

test whether the four-dimension model is superior to other designs such as the two or three-

dimension models a series of CFA were conducted and the results show that the four-

dimension model is a significantly better fit than any of the other designs

To test whether the factorial structure of the PSM instrument is equivalent across the

samples from the twelve countries four nested models were tested as part of multi-group

analysis with MLE (Byrne 2010) First the unconstrained model was the baseline model

which relaxed all equality constraints This model hypothesized the same four-factor structure

of the instrument across twelve countries The fit indices for this baseline model were χsup2 (df =

1176) = 24977 p lt 001 GFI = 0901 CFI = 900 RMSEA = 0020 [019 021] These

indices indicate that the four-factor model of the PSM instrument exhibits acceptable fit

across the samples from the twelve countries Second the measurement weights model tested

- 17 -

the invariance of factor loadings across countries by placing equality constraints on these

parameters The fit indices for this model were χsup2 (df = 1308) = 28088 p lt 001 GFI = 0890

CFI = 887 RMSEA = 0020 [019 021] Since the measurement weights model was nested

within the unconstrained model the chi-square difference text ∆χsup2 (132) = 3111 p lt 001

indicated that some equality constraints of factor loadings did not hold across the twelve

countries Further test results suggested that the equality constraints of structural covariances

and measurement residuals were not upheld across the twelve countries It needs to be

stressed that despite the differences in factor loadings and factor covariances the results

support the main assumption that the same four-factor structure holds across different

countries We may assume that people in different cultures understand PSM in the same

fashion

To find differences in the factor structure of PSM among countries we conducted a

series of CFA with each countryrsquos dataset and followed the same procedure as when

analyzing the sample as a whole As table 5 shows although there are great similarities

between the appropriate items in the countries there are also some differences Even though

the items vary according to national differences the four-factor structure had a good fit to the

data Thus the results confirm that the four-factor structure of PSM can be generalized

[Table 5 about here]

The APS dimension covers both an individualrsquos attraction to public participation and

commitment to public interest but all its items refer to some notion of public interest such as

improving public service aiding community tackling social problems meaningful public

service and contributing to the common good The APS dimension captures an individualrsquos

- 18 -

disposition to serve the public interest through public participation When developing the

items some assumed that the items of committing to public interests can be separated from

those of pursuing public values in the CPV dimension The empirical test showed that

respondents tended to understand committing to public interests as behavior-oriented and so

the items for public interest are not differentiated from participation-oriented items in their

perceptions but are different from those for public values The respondents seem to

understand that pursuing public values is more related to personal beliefs or creeds Thus the

CPV dimension is solely focused on public values and the items of pursing public interests

are combined with those of the APP dimension to make the APS dimension

Public values and their relative importance are neither impervious to change over

time nor the same among all countries (Charles de Jong and Ryan 2011 Davis and West

2009) Moreover some values may be in conflict as Martinsen and Beck Joslashrgensen (2010)

discussed in regard to the tensions between accountability and efficiency It is therefore not

possible to include all public values in one dimension (Beck Joslashrgensen and Bozeman 2007)

Accordingly the items of the CPV dimension are developed for estimating individual

commitment to seek public values which there are general consensus about and which are

generally formulated

When all four dimensions of PSM are included in a survey cross-national

equivalence of the survey can be assured The items represent reflections or manifestations of

each dimension and hence are relatively homogeneous and interchangeable Adding or

deleting items does not change the essential nature of the dimension Thus any combination

of items in a dimension can be regarded as equivalent When using structural equation

modeling (SEM) each dimension needs to have at least three reflective indicators in order to

achieve identification or it needs to have at least two reflective indicators and be allowed to

- 19 -

covary with another dimension that has at least two reflective measures With one indicator

the dimension is identified if the loading and the variance of the uniqueness are both fixed

(Bollen 1989 Edwards 2010) Thus the number of items in each dimension is not restricted

when measuring PSM However we recommend using multiple items for each dimension

because more research is needed to investigate the appropriateness of this PSM instrument

with varying samples

The test results showed that model fit estimators were clearly better for the D-WLS

It means that in this study MLE was more rigorous than D-WLS in assessing the fit of the

model The empirical evidence indicated that MLE may be an acceptable technique for

ordinal data when there is a large sample a five-point Likert scale and the skewness and

kurtosis of variables is within acceptable limits as suggested by Byrne (2010)

Discussion

The analysis reaffirmed the four-dimensional structure of PSM but the results were not

entirely consistent with the original CFA we tested We were ultimately able to converge on a

good-fit operational model for PSM that was also consistent with underlying theory The

analysis is not able to resolve the debate about measurement models however for reasons

discussed below

Choosing the correct measurement model (ie reflective vs formative) is important

as measurement model misspecification results not only in misrepresentation of the epistemic

relationships between constructs and their measures but also in biased parameter estimates of

structural relationships (Jarvis et al 2003 Law and Wong 1999 MacKenzie 2003

MacKenzie Podsakoff and Jarvis 2005) A formative measurement model in isolation is

- 20 -

under-identified and cannot be estimated in the case of covariance-based SEM (CB-SEM)

(Bollen 1989) A construct with formative indicators must emit paths to at least two unrelated

reflective indicators two unrelated latent constructs or some combination of the two in order

to solve the problem of under-identification (Jarvis MacKenzie and Podsakoff 2003

MacCallum and Browne 1993 MacKenzie Podsakoff and Jarvis 2005) Thus ―when

dependent constructs change the empirical nature of the formatively measured construct

changes (Wilcox Howell and Breivik 2008 1223) and ―when specifying formative

measurement models what might seem like an innocuous choice of outcomes to achieve

identification can have fundamental ramifications for the meaning of the formative construct

and the inferences based on it (Edwards 2010 7)

Earlier research found that the association between a construct and its measures

should remain stable regardless of the larger causal model in which the constructs and

measures are embedded (Edwards and Bagozzi 2000) The association between a construct

and its measures is generally stable for reflective measures but associations of formative

measures with their construct are unstable The susceptibility of formative measurement to

both interpretational confounding and external inconsistency recommends that formative

measurement is not an equally attractive alternative to reflective measurement and that

whenever possible researchers should opt for reflective measurement (Howell Breivik and

Wilcox 2007a 2007b Kim Shin and Grover 2010) In the case of second-order models

researchers are limited to only a second-order reflective model in CB-SEM (Chin 2010)

Regardless of the measurement debate and the tradeoffs between reflective or

formative specification this study makes a theoretical argument that PSM is a second-order

formative construct This second-order formative factor model (reflective first-order

formative second-order) is also suggested as an alternative model that captures the

- 21 -

multidimensionality of complex concepts and avoids the shortcomings of first-order

formative measurement (Edwards 2010) However this model itself cannot achieve

identification One possible strategy is to develop direct reflective measures of PSM based on

the view that PSM is a general concept that is not only influenced by its specific dimensions

but also can be assessed using general measures (Edwards 2010 Javis MacKenzie and

Podsakoff 2003)6 However these general measures have not yet been developed for the

PSM construct

Another way around the problem of underidentification of formative models is to

apply the component-based approach known as partial least squares path modeling (PLS-PM)

which can model formative indicators (Fornell and Bookstein 1982 Goumltz Liehr-Gobbers

and Krafft 2010) The CB-SEM and the PLS-PM methods have distinctive statistical

characteristics and selecting an approach to SEM depends on the particular research situation

CB-SEM is the method of choice for theory testing while PLS-PM is appropriate for

prediction-oriented applications (Ringle Goumltz Wetzels and Wilson 2009) One advantage of

the PLS method is that it demands significantly fewer requirements compared to that of

covariance structure analyses but nevertheless delivers consistent estimation results Another

asset of the PLS approach is its ability to deal with formative as well as reflective indicators

even within one structural equation model This indicates that the PLS approach can be

appropriate for explorative analysis of structural equation models (Chin 2010 Goumltz Liehr-

Gobbers and Krafft 2010) A Monte Carlo simulation study to compare CB-SEM and PLS-

PM showed that PLS-PM provides a viable approximation of model parameters when the

prerequisites for CB-SEM are not met (Ringle Goumltz Wetzels and Wilson 2009) A formative

6 For example job satisfaction can be assessed with general reflective measures such as ―in general I am

satisfied with my job and the formative dimensions can be assessed with measures that describe satisfaction

with each job facet (Edwards 2010 15)

- 22 -

composite in the form of indexes needs to be tested to know the relative weights (γ-

parameters) of the dimensions of PSM In PLS-PM a second-order formative factor is

directly measured by observed variables for all the first order factors While this approach

repeats the number of manifest variables used the model can be identified without adding

outcome constructs that would be considered inappropriate or irrelevant from a conceptual

standpoint (Chin 2010 Edwards 2010) The test result of PLS-PM using Smart PLS 20

(Ringle Wende and Will 2005) produced the function PSM = 31APS + 36CPV + 28COM

+ 31SS However one consideration is whether and how construct error is modeled for the

formative construct (Diamantopoulos 2006) When using PLS-PM the researcher is making

the assumption that there is no error at the construct level as all constructs in PLS-PM are

modeled without error When using PLS-PM the absence of an error term will tend to

increase the weights therefore inflating their importance (Bollen and Lennox 1991) The

primary statistic for assessing a formative indicator is its weight the partialized effect of the

indicator on its intended construct controlling for the effects of all other indicators of that

construct (Cenfetelli and Bassellier 2009) A formative measure is essentially a multiple

regression with the construct representing the dependent variable and the indicators as the

predictors (Diamantopoulos and Winklhofer 2001) PSM may be regarded as a weighted

linear composite of its dimensions The weights used to create the composites are different

across CE-SEM and PLS-PM The causal dimensional weights generated in CB-SEM are

estimated in a model in which a construct-level error term is specified (eg PSM = γ1APM +

γ2CPI + γ3COM + γ4SS + ζ) whereas the weights in PLS-PM are not (eg PSM = γ1APM +

γ2CPI + γ3COM + γ4SS) This leads to two different conceptualizations of the PSM construct

one in which the causal indicators do not capture the construct in its entirety and another in

which they do (Hardin et al 2010) These differences need to be discussed in future research

- 23 -

in this area

One practical suggestion to measure PSM is to apply unit weights to its dimensions if

we assume the effects of the four dimensions are equivalent A literature review indicates that

unit weights have substantial predictive validity when compared with regression weights The

literature shows that the empirical correlations between scores generated from differential and

unit-weighting approaches are quite high In creating composite scores unit weights can be

an appropriate approach for weighting under many circumstances (Bobko Roth and Buster

2007) Although the uses of unit-weight composites will likely result in the explanation of

less variance than empirical estimations of dimensional weights they will remain consistently

defined across studies and hence suitable for theory testing Alternatively more thorough

discussions on the dimensions of PSM may suggest fixed-weight composites (Hardin et al

2010) A previous study (Kim 2010 22) also suggested ―Overall indexing PSM by

measuring all the dimensions of PSM and simply summing the averages of each dimension

can make it easier to measure PSM in practice as PSM = APS + CPV + COM + SS

An advantage of regarding PSM as a second-order formative construct is that the

dimensions can be analyzed both individually and collectively This is important as the

dimensions are expected to have different antecedents and consequences For instance Perry

(1997) found some significant differences in the influence of independent variables on the

four dimensions of PSM Taylor (2007) confirmed that certain dimensions are more important

than others in influencing work outcomes Moynihan and Pandey (2007a) reported that red

tape is negatively related to one dimension but reform orientation and hierarchical authority

is positively related to another dimension A recent study found that the SS dimension has a

positive relationship with occupational intention but the other dimensions do not (Liu et al

2011) Thus it is valuable to analyze the causes and effects of individual dimensions of PSM

- 24 -

as well as of the PSM construct

Conclusion

The purpose of this study is to develop a measurement instrument for PSM for cross-national

research and comparison On the basis of the sharpened construct of PSM (Kim and

Vandenabeele 2010) using data from an international survey conducted in 12 countries we

found that the four-factor structure of PSM is generalizable across country cases This study

provides the measurement index of PSM with the four dimensions of APS CPV COM and

SS The APS dimension measures the instrumental motives of PSM the CPV dimension

determines the value-based motives of PSM while the COM and SS dimensions gauge the

identification motives and self-sacrifice respectively When all four dimensions of PSM are

used in a survey cross-national equivalence of survey items can be assured The items for

measuring PSM in future research are suggested in Appendix 4

This study has several limitations First looking for samples which were comparable

to the largest possible extent we opted for surveys in local government The results found

and discussed in this study are based on PSM among local government employees The

results might be different when surveying central government employees or other specific

groups in the public sector Thus the potential impact of the sample bias needs to be

considered in understanding this study It also means that the survey items for measuring

PSM should be examined with different samples in various areas and at different government

levels

Second the test results showed that there is a high degree of correlation among the

dimension such as APS CPV and COM Future research should not only investigate whether

- 25 -

the items reflect the core content of the dimensions but also continue to develop more

appropriate items for better discriminant validity of each dimension Third the content of

APP dimension was assumed to be distinct from the CPV dimension that focuses on personal

disposition to pursue public interest and public values However the public interest sub-

dimension was separated from the public values sub-dimension in the CPV dimension and

combined with the APP dimension to form the APS dimension Further research should

analyze the CPV dimension and investigate the appropriateness of the APS dimension

In order to facilitate research internationally the PSM construct needs to be

improved conceptually and operationally Kim and Vandenabeele (2010) have dealt with the

conceptual composition and operational dimension of PSM as well as the measurement

model of PSM The next step is to develop an internationally robust measurement instrument

of PSM This study proposes that scholars use all four of the dimensions in measuring the

conceptual components of PSM such as self-sacrifice instrumental value-based and affective

motives because they collectively represent PSM It provides a set of appropriate items for

measuring PSM and confirms the dimensional structure of PSM The survey results suggest

that people in different cultures understand PSM in the same fashion We know that this study

is simply a step along the way to developing a more robust measure but we hope it

contributes to enhanced cross-national research and comparison along with the generation of

cumulative knowledge internationally

- 26 -

REFERENCES

Anderson James C and David D Gerbing 1988 Structural Equation Modeling in Practice

A Review and Recommended Two-Step Approach Psychological Bulletin 103

411-423

Arbuckle James L 2009 Amos 180 Userrsquos Guide Crawfordville FL Amos Development

Co

Beck Joslashrgensen Torben and Barry Bozeman 2007 Public Values An Inventory

Administration amp Society 39(3) 354-381

Beutel Ann M and Margaret Mooney Marini 1995 Gender and values American

Sociological Review 60 436 - 448

Bobko Philip Philip L Roth and Maury A Buster 2007 The Usefulness of Unit Weights in

Creating Composite Scores A Literature Review Application to Content Validity and

Meta-Analysis Organizational Research Methods 10(4) 689-709

Bollen Kenneth A 1989 Structural Equations with Latent Variables New York Wiley

Bollen K A and R Lennox 1991 Conventional Wisdom on Measurement A Structural

Equation Perspective Psychological Bulletin 110(2) 305-314

Byrne Barbara M 2010 Structural Equation Modeling with Amos Basic Concepts

Applications and Programming 2nd

ed New York Routledge

Camilleri Emanuel 2006 Towards Developing an Organizational Commitment ― Public

Service Motivation Model for Maltese Public Service Employees Public Policy

and Administration 21 63-83

Castaing Seacutebastien 2006 The Effects of Psychological Contract Fulfillment and Public

Service Motivation on Organizational Commitment in the French Civil Service

Public Policy and Administration 21 84-98

Cenfetelli Ronald T and Geneviegraveve Bassellier 2009 Interpretation of Formative

Measurement in Information Systems Research MIS Quarterly 33(4) 689-707

Charles Michael B W Martin de Jong and Neal Ryan 2011 Public Values in Western

- 27 -

Europe A Temporal Perspective American Review of Public Administration 41

75-91

Chin Wynne W 2010 How to Write Up and Report PLS Analysis In Handbook of Partical

Least Squares edited by V Esposito Vinzi et al 655-690 Berlin Springer-Verlag

Coursey David H and Sanjay K Pandey 2007a Public Service Motivation Measurement

Testing an Abridged Version of Perryrsquos Proposed Scale Administration amp Society

39 547-568

Coursey David H and Sanjay K Pandey 2007b Content Domain Measurement and

Validity of the Red Tape Concept American Review of Public Administration 37

342-361

Coursey David H James L Perry Jeffrey L Brudney and Laura Littlepage 2008

Psychometric Verification of Perryrsquos Public Service Motivation Instrument Results

for Volunteer Exemplars Review of Public Personnel Administration 28 79-90

Davis Paul and Karen West 2009 What Do Public Values Mean for Public Action Putting

Public Values in Their Plural Place American Review of Public Administration 39

602-618

DeHart-Davis Leisha Justin Marlowe and Sanjay K Pandey 2006 Gender Dimensions of

Public Service Motivation Public Administration Review 66 873-887

Diamantopoulos Adamantios 2006 The error term in formative measurement models

Interpretation and modeling implications Journal of Modeling in Management 1(1)

7-17

Diamantopoulos Adamantios and Judy A Siguaw 2006 Formative versus Reflective

Indicators in Organizational Measure Development A Comparison and Empirical

Illustration British Journal of Management 17 263-282

Diamantopoulos Adamantios and Heidi M Winklhofer 2001 Index Construction with

Formative Indicators An Alternative to Scale Development Journal of Marketing

Research 38 269-277

Edwards Jeffrey R 2010 The Fallacy of Formative Measurement Organizational Research

- 28 -

Methods Online First published on August 19 2010 as

doi1011771094428110378369

Edwards Jeffrey R and Richard P Bagozzi 2000 On the Nature and Direction of

Relationships between Constructs and Measures Psychological Methods 5 155-

174

Flora David B and Patrick J Curran 2004 An Empirical Evaluation of Alternative

Methods of Estimation for Confirmatory Factor Analysis with Ordinal Data

Psychological Methods 9 466-491

Fornell Claes and Fred L Bookstein 1982 Two structural equations models LISREL and

PLS applied to customer exit-voice theory Journal of Marketing Research 19 440-

452

Giauque David Adrian Ritz Freacutedeacuteric Varone Simon Anderfuhren-Biget and Christian

Waldner 2011 Putting Public Service Motivation into Context A Balance between

Universalism and Particularism Forthcoming in International Review of

Administrative Sciences 77(2)

Goumltz Oliver Kerstin Liehr-Gobbers and Manfred Krafft 2010 Evaluation of Structural

Equation Models Using the Partial Least Squares (PLS) Approach In V Esposito

Vinzi et al (eds) Handbook of Partial Least Squares 691-711 DOI 101007978-3-

540-32827-8_30 Springer-Verlag Berlin Heidelberg

Guinot Christiane Julie Latreille and Michel Tenenhaus 2001 PLS Path Modelling and

Multiple Table Analysis Application to the Cosmetic Habits of Women in Ile-de-

France Chemometrics and Intelligent Laboratory Systems 58 247-259

Hair Joseph F Jr William C Black Barry J Babin and Rolph E Anderson 2010

Multivariate Data Analysis 7th

ed Upper Saddle River NJ Pearson Education

Hardin Andrew M Jerry Cha-Jan Chang Mark A Fuller and Gholamreza Torkzadeh 2010

Formative Measurement and Academic Research In Search of Measurement

Theory Educational and Psychological Measurement Online First published on

June 11 2010 as doi101177001316440370208

- 29 -

Howell Roy D Einar Breivik and James B Wilcox 2007a Reconsidering Formative

Measurement Psychological Methods 12 205-218

Howell Roy D Einar Breivik and James B Wilcox 2007b Is Formative Measurement

Really Measurement Reply to Bollen (2007) and Bagozzi (2007) Psychological

Methods 12 238-245

Jarvis Cheryl Burke Scott B MacKenzie and Philip M Podsakoff 2003 A Critical Review

of Construct Indicators and Measurement Model Misspecification in Marketing and

Consumer Research Journal of Consumer Research 30 199-218

Joumlreskog Karl G and Dag Soumlrbom 2005 LISREL 872 (Computer Software) Lincolnwood

IL Scientific Software International

Kim Gimun Bongsik Shin and Varun Grover 2010 Investigating Two Contradictory Views

of Formative Measurement in Information Systems Research MIS Quarterly 34

345-365

Kim Sangmook 2009a Revising Perryrsquos Measurement Scale of Public Service Motivation

American Review of Public Administration 39 149-163

Kim Sangmook 2009b Testing the Structure of Public Service Motivation in Korea A

Research Note Journal of Public Administration Research and Theory 19 839-

851

Kim Sangmook 2010 Testing a Revised Measurement of Public Service Motivation

Reflective versus Formative Specification Journal of Public Administration

Research and Theory Doi101093jopartmuq048

Kim Sangmook and Wouter Vandenabeele 2010 A strategy for building public service

motivation research internationally Public Administration Review 70 701-709

Kline Rex B 2005 Principles and Practice of Structural Equation Modeling 2nd

ed New

York Guilford Press

Law Kenneth and Chi-Sum Wong 1999 Multidimensional Constructs in Structural

Equation Analysis An Illustration using the Job Perception and Job Satisfaction

Constructs Journal of Management 25 143-160

- 30 -

Leisink Peter and Bram Steijn 2009 Public Service Motivation and Job Performance of

Public Sector Employees in the Netherlands International Review of

Administrative Sciences 75 35-52

Liu Bangcheng Chun Hui Jin Hu Wensheng Yang and Xinli Yu 2011 How well can

Public Service Motivation Connect with Occupational Intention International

Review of Administrative Sciences 77 191-211

Liu Bangcheng Ningyu Tang and Xiaomei Zhu 2008 Public Service Motivation and Job

Satisfaction in China An Investigation of Generalisability and Instrumentality

International Journal of Manpower 29 684-699

MacCallum Robert C and Michael W Browne 1993 The Use of Causal Indicators in

Covariance Structure Models Some Practical Issues Psychological Bulletin 114

533-541

MacKenzie Scott B 2003 The dangers of poor construct conceptualization Journal of the

Academy of Marketing Science 31(3) 323-326

MacKenzie Scott B Philip M Podsakoff and Cheryl Burke Jarvis 2005 The Problem of

Measurement Model Misspecification in Behavioral and Organizational Research

and Some Recommended Solutions Journal of Applied Psychology 90 710-730

Martinsen Dorte Sindbjerg and Torben Beck Joslashrgensen 2010 Accountability as a

Differentiated Value in Supranational Governance American Review of Public

Administration 40 742-760

Moynihan Donald P and Sanjay K Pandey 2007a The Role of Organizations in Fostering

Public Service Motivation Public Administration Review 67 40-53

Moynihan Donald P and Sanjay K Pandey 2007b Finding Workable Levers over Work

Motivation Comparing Job Satisfaction Job Involvement and Organizational

Commitment Administration amp Society 39 803-832

Pandey Sanjay K and Edmund C Stazyk 2008 Antecedents and Correlates of Public

Service Motivation In Motivation in Public Management edited by James L Perry

and Annie Hondeghem 101-117 Oxford Oxford University Press

- 31 -

Perry James L 1996 Measuring Public Service Motivation An Assessment of Construct

Reliability and Validity Journal of Public Administration Research and Theory 6

5-22

Perry James L 1997 Antecedents of Public Service Motivation Journal of Public

Administration Research and Theory 7 181-197

Perry James L 2000 Bringing Society in Toward a Theory of Public Service Motivation

Journal of Public Administration Research and Theory 10 471-488

Perry James L and Annie Hondeghem 2008a Motivation in Public Management Oxford

Oxford University Press

Perry James L and Annie Hondeghem 2008b Directions for Future Theory and Research

In Motivation in Public Management edited by James L Perry and Annie

Hondeghem 294-313 Oxford Oxford University Press

Perry James L Annie Hondeghem and Lois Recascino Wise 2010 Revisiting the

Motivational Bases of Public Service Twenty Years of Research and an Agenda for

the Future Public Administration Review 70(5) 681-690

Perry James L and Wouter Vandenabeele 2008 Behavioral dynamics Institutions

Identities and Self-Regulation In Motivation in Public Management edited by

James L Perry and Annie Hondeghem 56-79 Oxford Oxford University Press

Perry James L and Lois Recascino Wise 1990 The Motivational Bases of Public Service

Public Administration Review 50 367-373

Rainey Hal G 1982 Reward Preference among Public and Private Managers In Search of

the Service Ethic American Review of Public Administration 16 288-302

Ringle Christian M Oliver Goumltz Martin Wetzels and Bradley Wilson 2009 On the Use of

Formative Measurement Specifications in Structural Equation Modeling A Monte

Carlo Simulation Study to Compare Covariance-Based and Partial Least Squares

Model Estimation Methodologies Faculty of Economics and Business Administration

Maastricht Univeristy

Ringle C M S Wende and S Will 2005 Smart PLS 20 (M3) Beta

- 32 -

httpwwwsmartplsde

Ritz Adrian 2011 Attraction to Public Policy Making A Qualitative Inquiry into

Improvements in PSM Measurement Forthcoming in Public Administration

Spector Paul E 1992 Summated Rating Scales Construction An Introduction Newbury

Park CA Sage

Steenbergen Marco R 1996 Compassion and American public opinion An analysis of the

NES Humanitarianism Scale Report to the Board of Overseers of the National

Election Studies

Steffen Patrick R and Kevin S Masters 2005 Does compassion mediate the intrinsic

religion-health relationship Annals of Behavioral Medicine 30 217-223

Taylor Jeannette 2007 The Impact of Public Service Motives on Work Outcomes in

Australia A Comparative Multi-Dimensional Analysis Public Administration 85

931-959

Vandenabeele Wouter 2007 Towards a Theory of Public Service Motivation An

Institutional Approach Public Management Review 9 545-556

Vandenabeele Wouter 2008a Government Calling Public Service Motivation as an Element

in Selecting Government as an Employer of Choice Public Administration 86

1089-1105

Vandenabeele Wouter 2008b Development of a Public Service Motivation Measurement

Scale Corroborating and Extending Perryrsquos Measurement Instrument International

Public Management Journal 11 143-167

Vandenabeele Wouter 2009 The Mediating Effect of Job Satisfaction and Organizational

Commitment on Self-Reported Performance More Robust Evidence of the PSM-

Performance Relationship International Review of Administrative Sciences 75 11-

34

Wilcox James B Roy D Howell and Einar Breivik 2008 Questions about Formative

Measurement Journal of Business Research 61 1219-1228

Wise Lois Recascino 2000 The Public Service Culture In Public Administration Concepts

- 33 -

and Cases 7th ed edited by Richard J Stillman II 342-353 Boston Houghton

Mifflin

Wright Bradley E 2008 Methodological Challenges associated with Public Service

Motivation Research In Motivation in Public Management edited by James L

Perry and Annie Hondeghem 80-98 Oxford Oxford University Press

Wright Bradley E and Sanjay K Pandey 2008 Public Service Motivation and the

Assumption of Person-Organization Fit Testing the Mediating Effect of Value

Congruence Administration amp Society 40 502-521

- 34 -

Table 1

Possible Indicators of PSM and Descriptive Statistics (n = 2868)

Dimensions and items Mean Standard

Deviation

Item-Total

Correlation

Attraction to Public Participation (APP)

APP1 I am interested in helping to improve public service

APP2 I am satisfied when I see people benefiting from the public programs I was

involved in

APP3 I like to discuss topics regarding public programs and policies with others

APP4 I believe that public sector activities contribute to our general welfare

APP5 I admire people who initiate or are involved in activities to aid my

community

APP6 Contributing to public programs and policies helps me realize myself

APP7 It is important to contribute to activities that tackle social problems

412

441

355

393

423

351

409

696

662

917

753

672

881

734

524

421

482

412

493

478

552

Commitment to Public Values (CPV)

Sub-dimension for Public Interests (CPV1)

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

CPI3 I would prefer seeing public officials do what is best for the whole

community even if it harmed my interests

CPI4 Serving the public interest is more important than helping a single individual

Sub-dimension for Public Values (CPV2)

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of public

services

CPV3 It is fundamental that public services respond to the needs of the citizens

CPV4 Decisions regarding public services should be democratic despite the time

and effort it takes

CPV5 Everybody is entitled to a good service even if it costs a lot of money

CPV6 It is fundamental that the interests of future generations are taken into

account when developing public policies

CPV7 To act ethically is essential for public servants

CPV8 I believe that public employees must always be aware of the legitimacy of

their activities

CPV9 I personally identify with the aim of protecting individual liberties and rights

418

409

348

337

428

416

433

394

367

426

441

434

414

695

662

899

995

730

704

710

852

932

689

681

692

763

525

596

480

251

484

437

356

325

376

439

462

402

392

Compassion (COM) COM1 It is difficult for me to contain my feelings when I see people in distress

COM2 I feel sympathetic to the plight of the underprivileged

COM3 I empathize with other people who face difficulties

COM4 I have little compassion for people in need who are unwilling to take the

first step to help themselves

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

354

387

388

299

412

391

951

825

707

1083

755

765

399

511

389

011

487

502

Self-Sacrifice (SS)

SS1 Making a difference to society means more to me than personal achievements

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS5 People should give back to society more than they get from it

SS6 Serving other citizens would give me a good feeling even if no one paid me for

it

SS7 I would agree to a good plan to make a better life for the poor even if it costs

me money

357

329

339

302

345

363

350

904

872

949

982

888

1049

895

508

584

454

531

412

368

546

- 35 -

Table 2

Background of Respondents (n = 2868)

Variables Characteristics Respondents

Sex

Male

Female

NA

1209

1622

37

422

566

13

Length of service

(years)

0 ~ 10

10 ~ 20

20 ~ 30

30+

NA

1126

723

484

491

44

393

252

169

171

15

Organizational status

―Do you supervise employees

No

Yes

NA

1472

1076

320

513

375

112

Country

Australia

Belgium

China

Denmark

France

Italy

Korea

Lithuania

The Netherlands

Switzerland

United Kingdom

USA

249

214

230

249

266

162

253

236

249

250

260

250

87

75

80

87

93

56

88

82

87

87

91

87 Note NA = no answer

- 36 -

Table 3

Standardized factor loadings and correlations for the 31-item PSM index (n = 2868)

Dimensions and items SFL (MLE) SFL (D-WLS) Alpha

APP

APP1

APP2

APP4

APP5

APP6

APP7

597

486

455

558

490

600

657

552

514

625

549

665

699

CPV1

CPI1

CPI2

CPI3

CPI4

588

660

488

289

653

732

535

294

542

CPV2

CPV1

CPV2

CPV3

CPV4

CPV5

CPV6

CPV7

CPV8

CPV9

611

546

436

393

418

534

598

496

460

690

631

505

457

486

601

669

573

530

745

COM

COM1

COM2

COM3

COM5

COM6

496

569

498

597

607

516

651

538

670

684

687

SS

SS1

SS2

SS3

SS4

SS5

SS6

SS7

567

736

558

728

467

409

663

660

762

591

695

515

494

709

782

Measures of fit (MLE) χsup2 (df = 424) = 43699 p lt 001 CFI = 835 GFI = 899

RMSEA = 057 [055 059]

Measure of fit (D-WLS) χsup2 (df = 424) = 39247 p lt 001 CFI = 971 GFI = 976

RMSEA = 054 [052 055]

Inter-factor correlations MLE (D-WLS)

1 2 3 4

1 APS

2 CPV1 996 (988)

3 CPV2 902 (890) 798 (798)

4 COM 787 (783) 758 (762) 779 (783)

5 SS 700 (742) 800 (859) 509 (537) 720 (729)

Note SFL = standardized factor loading All factor loadings and correlations are significant at p lt 001

- 37 -

Table 4

Standardized factor loadings and correlations for the 16-item PSM index (n = 2868)

Dimensions and items SFL

(MLE)

SFL

(D-WLS)

APS

APP5 I admire people who initiate or are involved in activities to aid my

community

APP7 It is important to contribute to activities that tackle social problems

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

611

592

637

666

665

669

690

751

CPV

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of

public services

CPV6 It is fundamental that the interests of future generations are taken

into account when developing public policies

CPV7 To act ethically is essential for public servants

612

546

551

590

698

631

618

658

COM

COM1 It is difficult for me to contain my feelings when I see people in

distress

COM2 I feel sympathetic to the plight of the underprivileged

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

490

546

596

604

515

608

669

679

SS

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS7 I would agree to a good plan to make a better life for the poor even if

it costs me money

763

571

766

678

807

629

747

755

Measures of fit (MLE)

χsup2 (df = 98) = 8490 p lt 001

CFI = 939 GFI = 965

RMSEA = 052 [049 055]

Measure of fit (D-WLS)

χsup2 (df = 98) = 6069 p lt 001

CFI = 988 GFI = 992

RMSEA = 043 [039 046]

Alpha CR (D-WLS) Inter-factor correlations MLE (D-WLS)

1 APS 717 788 1 APS 2 CPV 3 COM

2 CPV 663 747 848 (859)

3 COM 645 713 781 (791) 789 (795)

4 SS 784 826 565 (596) 462 (505) 715 (726)

Note SFL = standardized factor loading Alpha = Cronbachrsquos alpha CR = composite reliability

All factor loadings and correlations are significant at p lt 001

- 38 -

Table 5

Results of CFA in the individual countries Standardized factor loadings and measures of fit

Factors

amp Items

Australia Belgium China Denmark France Italy

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

672

756

799

785

804

943

710

703

597

790

799

730

584

675

685

681

731

742

578

748

674

643

765

739

645

704

678

708

812

779

632

604

765

766

672

853

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

552

593

560

699

727

632

624

605

560

713

709

645

702

602

616

724

686

735

567

560

715

654

605

811

608

658

587

696

755

698

673

620

523

782

697

654

COM

COM1

COM2

COM3

COM5

COM6

623

584

662

688

701

739

732

608

475

808

666

581

578

616

605

649

670

673

699

751

737

679

816

776

568

546

801

626

673

850

637

546

741

715

601

822

SS

SS2

SS3

SS4

SS7

654

794

806

892

767

813

727

693

799

793

725

838

857

820

773

896

855

842

806

749

792

829

753

884

847

731

753

889

816

753

815

692

664

795

782

734

Inter-factor

correlations

093

~

752

163

~

810

320

~

773

323

~

825

452

~

827

497

~

855

323

~

850

321

~

847

498

~

899

536

~

902

425

~

829

476

~

865

Sample size 249 214 230 249 266 162

Measures of Fit

χsup2 638 444 881 732 765 554 817 625 735 565 750 564

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 980 100 939 987 969 997 967 994 975 997 949 995

GFI 961 986 941 986 948 991 949 992 955 992 926 986

RMSEA 036 00 063 050 051 026 053 035 045 026 059 033

- 39 -

Table 5 (Continued)

Factors

amp Items

Korea Lithuania The Netherlands Switzerland UK USA

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

619

701

667

706

781

717

666

512

682

751

605

779

557

629

630

620

719

729

535

621

702

643

694

735

655

725

738

710

804

822

535

621

702

643

694

735

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

594

607

722

722

644

722

674

535

533

790

589

711

568

554

538

655

628

545

627

580

615

711

666

704

800

512

555

873

581

646

627

580

615

711

666

704

COM

COM1

COM2

COM3

COM5

COM6

694

677

550

763

736

627

557

639

428

550

696

534

737

623

508

766

616

608

659

592

683

703

661

769

682

583

729

739

644

803

659

592

683

703

661

769

SS

SS2

SS3

SS4

SS7

776

752

679

779

770

811

707

721

635

725

734

706

814

656

743

820

767

764

745

632

689

788

662

744

671

818

832

830

864

802

745

632

689

788

662

744

Inter-factor

correlations

432

~

726

466

~

746

376

~

788

406

~

902

377

~

852

455

~

836

637

~

818

660

~

828

317

~

852

360

~

852

637

~

818

660

~

828

Sample size 253 236 249 250 260 250

Measures of Fit

χsup2 896 573 612 462 773 616 815 715 769 542 815 715

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 951 996 975 100 958 993 957 991 972 998 957 991

GFI 942 988 960 989 951 987 948 990 955 990 948 990

RMSEA 059 028 034 00 050 034 053 044 048 022 053 044

Note All factor loadings and correlations are significant at p lt 001

- 40 -

Appendix 1 Dimensions and possible items of PSM (the first round)

1 Attraction to Public Participation

(1) I am interested in developing public programs helping my country or community (Kim 2009a)

(2) Sharing my views on public policies with others is attractive to me (Kim 2009a)

(3) I am satisfied when seeing people get benefits from the public programs I was involved in (Kim 2009a)

(4) I like to discuss political subjects with others (Giauque et al 2011)

(5) I think the governmental activities contribute to our welfare

(6) I like to initiate actions to help out my community

(7) I do not care much about politicians (Perry 1996)

2 Commitment to Public Values

Possible items for public interest

(1A) Meaningful public service is very important to me (Perry 1996)

(2A) It is important for me to contribute for the common good (Giauque et al 2011)

(3A) I recognize myself with the mission of protection of liberties and rights endorsed by the nation-state

(Giauque et al 2011)

(4A) I would prefer seeing public officials do what is best for the whole community even if it harmed my

interests (Perry 1996)

(5A) To me serving the public interest is more important than helping other people (Vandenabeele 2008b)

Possible items for public values

(1B) I recognize myself with the promotion of the equality of chances endorsed by the nation-state (Giauque et

al 2011)

(2B) It is fundamental that the public service provided to the citizen is regular and continuous (Giauque et al

2011)

(3B) It is fundamental that public service innovates to respond to the new needs of the citizensclients (Giauque

et al 2011)

(4B) The dignity and well-being of all should be the most important concerns in any society (Steenbergen

1996)

(5B) It is important that public servants account for all the costsexpenses they make (Vandenabeele 2008b)

(6B) We have to do everything in our power to pursue the goal of democracy

3 Compassion

(1) It is difficult for me to contain my feelings when I see people in distress (Perry 1996)

(2) I feel sympathetic to the plight of the underprivileged (Kim 2009a)

(3) I am often reminded by daily events how dependent we are on one another (Perry 1996)

(4) I feel compassion for others in the difficulties they are facing (Steffen and Masters 2005)

(5) It is not really my problem if others are in trouble and need help (Beutel and Marini 1995)

(6) I find it hard to be sympathetic toward starving people in foreign lands when there is so much trouble in our

own country (Beutel and Marini 1995)

(7) I get very upset when I see other people treated unfairly (Beutel and Marini 1995)

(8) I would agree to a good plan to make a better life for the poor even if it cost me money (Beutel and Marini

1995)

(9) To me patriotism includes seeing the welfare of others (Perry 1996)

(10) I care very much about other people

4 Self-Sacrifice

(1) Making a difference in society means more to me than personal achievements (Perry 1996)

(2) I am prepared to make enormous sacrifices for the good of society (Perry 1996)

(3) I believe in putting duty before self (Perry 1996)

(4) I am one of those rare people who would risk personal loss to help someone else (Perry 1996)

(5) I think people should give back to society more than they get from it (Perry 1996)

(6) Serving other citizens would give me a good feeling even if no one paid me for it (Perry 1996)

(7) Much of what I do is for a cause bigger than myself (Perry 1996)

- 41 -

Appendix 2 Newly suggested items through the first round

1 Items for the Attraction to Public Participation Dimension

(1) In the public domain every person should be counted as equal

(2) I enjoy to think and argue in a political way

(3) I am satisfied when seeing my ideas being integrated in public policies

(4) I can easily relate to the way of thinking and arguing of politicians

(5) Working in the public sector I can better realize myself

2 Items for the Public Values Sub-Dimension

(1) Public service should give value for money

(2) Cost considerations should play an important role in decisions about public service provision

(3) It is fundamental that the interests of future generations are taken into account when developing public

policies

(4) To act with correct ethics is for public servants as important as their professional competence

(5) Primarily public servants should be accountable to the public and not to their superiors

3 Items for the Self-Sacrifice Dimension

(1) I would agree to a good plan to make a better life for the poor even if it costs me money

- 42 -

Appendix 3 An example of a summary of comments and suggestions (the second round)

On the item ―I am interested in developing public programs helping my country or community

A It is a good sentence but is it possible for everyone to ―develop public programs Even public officials are

limited to do this

B OK

C or the special group

D Double barrel and difficult to translate to my country ―I am interested in helping to improve public services

E Good but I feel that wording requires revision If we draw sample from local government then country is

redundant ―I am interested in developing public policy programs that aim in helping my community

F 4 (General appreciation from1 to 4) Not clear whether it refers to ones work or to ones concern as a citizen

―(In my work) I am interested in contributing to the development of public etchellip

G Instead of development I would use implementing

H Delete public which is redundant because of the characteristic helping my country or community but add

or implementing ―I am interested in developing or implementing programs helping my country or

community

I Referring to dimension Wording Similar to item 6

Note Names have been changed to alphabets in order to ensure the anonymity of participants

- 43 -

Appendix 4 Dimensions and Items of Public Service Motivation for Future Research

Dimensions and items

Attraction to Public Service (APS)

1 I am interested in helping to improve public service

2 I admire people who initiate or are involved in activities to aid my community

3 It is important to contribute to activities that tackle social problems

4 Meaningful public service is very important to me

5 It is important for me to contribute to the common good

Commitment to Public Values (CPV)

1 I think equal opportunities for citizens are very important

2 It is important that citizens can rely on the continuous provision of public services

3 It is fundamental that the interests of future generations are taken into account when

developing public policies

4 To act ethically is essential for public servants

5 I believe that public employees must always be aware of the legitimacy of their activities

6 I personally identify with the aim of protecting individual liberties and rights

Compassion (COM)

1 It is difficult for me to contain my feelings when I see people in distress

2 I feel sympathetic to the plight of the underprivileged

3 I empathize with other people who face difficulties

4 I get very upset when I see other people being treated unfairly

5 Considering the welfare of others is very important

Self-Sacrifice (SS)

1 I am prepared to make sacrifices for the good of society

2 I believe in putting civic duty before self

3 I am willing to risk personal loss to help society

4 I would agree to a good plan to make a better life for the poor even if it costs me money

- 6 -

they are not interchangeable Thus PSM is regarded as a combination of specific dimensions

If any one of the dimensions changes PSM would change Conversely if an individualrsquos

PSM increases this would not necessarily be accompanied by an increase in all dimensions

Therefore as the PSM construct is refined for international use (Kim and Vandenabeele

2010) scholars must be attentive to identifying and using appropriate measurement models

Developing Possible Items

The conceptual components of PSM need to be empirically estimated through operational

definition Kim and Vandenabeele (2010) proposed that PSM is based on self-sacrifice and

associated with instrumental value-based and identification motives and that the dimensions

of the PSM construct be refined along the lines of attraction to public participation (APP)

commitment to public values (CPV) compassion (COM) and self-sacrifice (SS) As Perry

(1996) identified a multidimensional scale to measure PSM it is conceived as an overall

latent variable with various latent dimensions and is referred to as a second-order factor

model Research has found that the measurement model specifies the causal relationship

between PSM and its dimensions (Edwards and Bagozzi 2000) The direction of the

relationship is either from the PSM construct to the dimensions (second-order reflective

measurement) or from the dimensions to the construct (second-order formative measurement)

A theoretical consideration of the relationships between PSM and its dimensions led

to the proposal that PSM be defined in terms of multiple dimensions (Kim 2010 Kim and

Vandenabeele 2010) The dimensions represent different aspects of PSM with each

dimension capturing a distinct and potentially unique form of PSM Moreover the

dimensions may have different antecedents and consequences as well as different

characteristics and theoretical backgrounds (Moynihan and Pandey 2007a Pandey and

- 7 -

Stazyk 2008 Perry 1996 1997 2000 Perry and Wise 1990 Perry and Vandenabeele 2008

Taylor 2007 Vandenabeele 2007 2008a 2009) An individualrsquos PSM is shaped by the

individualrsquos APP CPV COM and SS Deleting a dimension means omitting a certain part of

PSM Therefore it is reasonable to suggest that the dimensions of PSM are formative and for

PSM to be regarded as a composite of its dimensions

Each dimension of PSM can be measured using reflective indicators that satisfy the

following criteria (Jarvis MacKenzie and Podsakoff 2003) a relative homogeneity and

interchangeability of indicators pertaining to each dimension a high degree of covariation

among indicators of each dimension and the expectation that the indicators of each

dimension are likely to be affected by the same antecedents and have the same consequences

Reflective indicators are essentially interchangeable and so adding or removing indicators

does not change the essential nature of a dimension Thus we consider PSM to be a second-

order formative construct with several dimensions that each has several reflective indicators

(reflective first-order formative second-order)

Based upon the refinement and clarification of the operational dimensions of PSM

(Kim and Vandenabeele 2010) items were developed for the four dimensions The items

selected were a mix of pre-existing items that loaded highly on PSM dimensions in multiple

studies (Giauque et al 2011 Kim 2009a Perry 1996 Vandenabeele 2008b) Where

appropriate items were not available new items were drafted from other previous studies

creating an initial total of 35 proposed items For the dimension of APP which focuses on

individual disposition to work in the public sector to participate in the public policy process

and in activities for community and social development seven possible items were suggested

The CPV dimension focuses on personal disposition to pursue public interest and public

values Two sub-dimensions of CPV seemed plausible one focusing on public interest and

- 8 -

another on public values such as social equity equal opportunity and democracy Thus five

possible items for the public interest sub-dimension (CPV1) and six items for the public

values sub-dimension (CPV2) were developed For the dimension of COM which focuses on

affective bonding with the identified objects such as other members of a social category or of

a political system ten possible items were suggested For the SS dimension Perryrsquos original

items (1996) were the primary source of items The full list of items is provided in Appendix

1

Review Processes

A group of researchers who shared our interest in better measurement of PSM collaborated

using e-mail in November 20091 The objective was to develop a measurement instrument

for PSM with a robust comparative character in other words to create a valid instrument (on

a national basis) that can be used internationally The first round for reviewing the possible

items lasted from the 16th

of November until the 7th

of December 2009 and all the

participants were asked to review these proposed items For this review two criteria were

employed (1) Do these items refer to the dimensions they claim to refer to and (2) Are

these items meaningful across various national contexts and in particular to the situation of

civil servants in that country Participants were also asked to add new items that might better

reflect a particular dimension or to rephrase items that were already on the list

A second round of item refinement lasted from the 9th

of December until the 22nd

of

December 2009 During this time 11 newly suggested items from the first round were

considered (see Appendix 2) A summary of all comments and suggestions emerging from the

1 Emmanuel Camilleri David Giauque Gerhard Hammerschmid Isabel Egger-Peitler Renate Meyer and

Cristian Pliscoff also participated in the review processes

- 9 -

first process was distributed to the participants to stimulate comments in the second round

(see Appendix 3) Based on the comments received in the second round discussion the

possible items for an international survey were selected However some items required

further discussion and so a third round was conducted from the 15th

of January until the 23rd

of January 2010 One issue that remained unresolved in the discussions relates to the

dimension of CPV Some argued that it is reasonable to divide CPV into two sub-dimensions

because committing to public interests is different from pursuing public values and there are

many important public values in democratic countries while others contended that CPV can

cover all important public values including public interest and there is no real benefit to add

one more dimension These differences were left to be resolved through empirical research

This did not affect the final items however and 33 items were confirmed The resulting

survey questionnaire was distributed to all the collaborators on February 9 2010 in order to

conduct a survey in each country

Measures

PSM was measured using the newly developed 33-item index 7 items for the dimension of

APP 13 items for CPV (4 for CPV1 and 9 for CPV2) 6 items for COM and 7 for SS In the

survey questionnaire a five-point Likert type scale (1 = strongly disagree 5 = strongly agree)

was used This scale is among the most widely used response formats in many empirical

investigations within the social and behavioral science in which there are no ―correct or

―incorrect responses (Spector 1992) The items of PSM were randomly distributed in the

survey format Gender age education length of service and employment status were also

measured in this survey

It was necessary to translate the measurement items into the official (native) language

- 10 -

of each country As this raised the possibility of inaccurate translations the back-translation

of the items into native languages was validated by asking two researchers to independently

translate the English-language items The researchers were instructed to consult with each

other over differences between translated items and to test the survey items with some native

respondents

[Table 1 about here]

Samples

The international survey was fielded from March to September 2010 The data upon which

this study are based were collected from civil servants in local governments in Australia

Belgium China Denmark France Italy Korea Lithuania the Netherlands Switzerland the

United Kingdom and the USA The principal criterion was that the samples collected in the

various countries should be comparable to the largest possible extent The criterion ruled out

national and state government as the competencies of national governments and also of state

governments if they exist constitutionally are different among most of the countries included

in the research Therefore we opted for local government as our focus The structure and

functions of local governments can differ widely even within countries and so the survey in

each country was focused on town hall bureaucrats in city county or township government

The respondents were permanent employees from both managerial and non-managerial levels

and from both administrative departments (ie no direct contact with the public) and service

departments (ie likely to have direct contact with the public) Despite being employed by

local governments in many or all of the countries police officers firefighters school teachers

public transport workers artists and musicians as well as nurses and doctors were excluded

- 11 -

from the sample

We aimed to receive 250 respondents for each country as this would enable us to

carry out the analyses we wanted to perform on the data Although there is no ideal sample

size one study recommends a sample of 200 to provide a sound basis for estimation (Hair

Black Babin and Anderson 2010) The method of data collection was left open Obviously a

web-survey is the easiest way to collect the data because it is economical to collect and to

code however compared to paper-pencil surveys this method has some drawbacks First not

everybody in the sample may have access to the internet which may cause a bias in the

responses Second internet surveys are generally known to produce lower response rates

Web surveys were recommended if possible but paper-pencil surveys were deemed

acceptable in cases where it was not possible to conduct a web survey Researchers were

encouraged to contact just one municipality except in cases where it was not possible to

gather enough data In such instances responses from multiple municipalities were accepted

By the end of September 2010 a total of 2868 responses were obtained2

2 In Australia seven local councils each with between from 70 and 232 employees were surveyed Paper-pencil

surveys were used in two councils while the rest relied on a web survey The local governments limited

distribution of paper-pencil surveys to only one or two sections of the council eg administration The response

rate was approximately 30 However it should be noted that not everyone working for the council has direct

access to a computer In Belgium three municipalities were surveyed on a census base by means of a web

survey as all employees had access to a computer This rendered a response of 120 (226) 79 (75) and 16

(53) responses for the respective municipalities resulting in an overall response rate of 331 with 662

invitations being sent out In China two departments social work and administration were surveyed in one

municipality using paper-pencil surveys In the social work department 65 out of 110 surveys were returned

and 165 out of 211 responded from the administration department giving an overall response rate of 767 In

Denmark a total of 1282 respondents in a municipality were sent an initial e-mail invitation to a web survey

and up to two e-mail reminders and the response rate was 473 In France all the employees in two

municipalities were surveyed In one municipality where both web and paper-pencil surveys were utilized 128

out of 685 questionnaires were returned while in another city using only a paper-pencil survey 138 out of 652

were received The average response rate was 199 In Italy in principle the link to the questionnaire was

made available to all employees (about 4000) in one city but for some of them access may have been difficult

yielding a response rate of 45 In Korea 500 employees in a city received a printed questionnaire and 253

responses were returned yielding a response rate of 506 In Lithuania 236 responses from 54 municipalities

were gathered by web survey In the Netherlands three municipalities were surveyed on a census base by means

of a web survey In total 319 responses of 910 invitations were returned (35) mdash 90 (45) 115 (38) and 114

(28) for the respective municipalities In Switzerland 516 out of 1326 local employees with e-mail access in

a municipality completed the survey giving a response rate of 389 In the United Kingdom the survey was

posted online and 1360 employees both in administrative jobs and front-line services of one municipality

- 12 -

Municipalities were selected because of their convenient accessibility and proximity to the

researcher and a convenience sample was used in most countries Seven countries surveyed

one municipality but seven local councils were surveyed in Australia Web surveys were used

in four countries while three countries used both web and paper-pencil survey The number

of respondents varied among the countries depending on the numbers of questionnaires

distributed the survey method and the response rate In most countries we used

approximately 250 responses but only 162 from Italy3 The majority of the respondents were

women (566) while in terms of service length those in the largest group (393) had

worked for fewer than 10 years Table 2 shows the distribution of respondentsrsquo sex length of

service employment status and country

[Table 2 about here]

Analyses

The statistical analysis applied confirmatory factor analysis (CFA) CFA was used to assess

the fit of the data to the hypothesized measurement model because it is preferred where

measurement models have a well-developed underlying theory for hypothesized patterns of

loadings The data were analyzed using Amos 180 (Arbuckle 2009) with the maximum

likelihood (MLE) estimation method and LISREL 872 (Joumlreskog and Soumlrbom 2005) with the

diagonally weighted least squares (D-WLS) estimation method

The individual items are designed to measure a theoretically continuous construct

but the observed responses are a discrete realization of five categories The common method

received an email to complete the survey The response rate was 19 In the USA the total number of

employees surveyed was 626 in one city using both electronic and paper-pencil surveys resulting in a response

rate of 588 3 In countries where significantly more than 250 surveys were collected we randomly selected only 250 cases

for analysis

- 13 -

of estimation within CFA is MLE a technique which assumes that the observed variables are

continuous and normally distributed These assumptions are not met when the observed data

are discrete as occurs when using ordinal scales as in this study A simulation study (Flora

and Curran 2004) found that D-WLS performs well for ordinal measures across a variety of

conditions especially for smaller samples Coursey and Pandey (2007a 2007b) argued the

importance of using D-WLS and not standard MLE in structural equations for ordinal data4

In most cases however the differences between MLE and D-WLS are not substantive to the

conclusions5 No comparison between standard MLE and D-WLS is provided in this paper

We give results from both MLE and D-WLS and this will aid in confirming the stability and

robustness of the results across varying mathematical specifications

For gauging reliability Cronbachrsquos alpha and composite reliability were tested

Composite reliability should be equal to or greater than 07 (Hair et al 2010) Convergent

validity was assessed from the measurement model by determining the statistical significance

of each indicatorrsquos estimated pattern coefficient on its specified underlying construct

Discriminant validity was assessed by determining whether the confidence interval around

the correlation estimate between two factors included 100 (Anderson and Gerbing 1988)

For model fit assessment goodness-of-fit index (GFI) comparative fit index (CFI)

and the root mean square error of approximation (RMSEA) were used The model achieves

an acceptable fit to the data when GFI and CFI equals or exceeds 090 and RMSEA values

4 Applying standard MLE to ordinal data such as Likert-type items may produce significant estimation

problems such as inflation of chi-square fit statistics and biased underestimation of parameters and standard

errors (Flora and Curran 2004) Coursey and Pandey (2007a 2007b) used a modified MLE using the asymptotic

covariance matrix and D-WLS for adjusting the ordinal nature of the data They reported that although the

modified MLE findings are slightly different on a few model parameters the differences are not substantive to

the conclusions However model fit estimators were clearly better for the D-WLS 5 Given normally distributed categorical variables MLE can be used with confidence when a variable has four

or more categories If there is a large sample a five-point Likert scale and the skewness and kurtosis of variables

is within acceptable limits it suggests distribution symmetry ―In most cases where the hypothesized model is

well specified and the scaling based on more than three categories it seems unlikely that there will be much

difference between the findings (Byrne 2010 160)

- 14 -

fall below 008 (Hair et al 2010) In general the larger the value of GFI and CFI and the

smaller the value of RMSEA the better the fit of the model (Bollen 1989) We also present

the χ2 test statistic Lower values of χ

2 indicate a better fit and should be nonsignificant but

the χ2 value increases as sample size increases and thus for large sample sizes this statistic

may lead to rejection of a model with good fit (Hair et al 2010)

Results

Using SPSS 180 descriptive statistics were computed for individual items as shown in table

1 One item (APP3) in the APP dimension ―I like to discuss topics regarding public programs

and policies with others was mistakenly omitted in the survey conducted in the United

States and so it was deleted in the statistical analysis In the assessment of normality no

items showed a skew or kurtosis value greater than the cutoffs of |3| or |8| recommended by

Kline (2005) As data were distributed normally MLE as well as D-WLS estimation was used

Based on inspection of item-total correlations one item (COM4) was dropped from further

analysis because it was weakly correlated with the overall index (011) Missing data is an

issue in CFA as it can cause estimation errors All variables in the survey have missing data

rates of approximately 016 to 143 Mean substitution was used to calculate replacement

values for the missing values in each countryrsquos dataset (Hair et al 2010) The deletion of the

two items left 31 items for the next stage of analysis

The five-correlated factor model was tested using CFA The CFA model in the initial

analysis hypothesized a priori that (1) responses to the 30-item PSM measurement

instrument could be explained by five factors such as APP CPV1 CPV2 COM SS (2) each

item would have a nonzero loading on the PSM factor that it was designed to measure and

- 15 -

zero loadings on all other factors (3) the five factors would be correlated and (4)

measurement error terms would be uncorrelated The resulting CFA with MLE suggested that

the initial five-correlated factor model was not a good fit to the data χsup2 (df = 424) = 43699 p

lt 001 GFI = 0899 CFI = 835 RMSEA = 0057 [055 059] At this stage APP was highly

correlated with CPV1 (r = 996 in MLE and r = 988 in D-WLS) suggesting the two

dimensions were not unique and lacked discriminant validity

[Table 3 about here]

Given the disconfirmation of the initial model an effort was made to estimate an

alternative model The two dimensions APP and CPV1 were combined to form a single

dimension Attraction to Public Service (APS) The items of CPV1 may be regarded as

action-oriented by the respondents because dedication to public service community and

common good is underlined in these items Consequently it is reasonable to combine APP

with CPV1 to form APS a dimension that focuses more on disposition to serve the public to

work for the common good and to participate in public policy processes After the overall

number of dimensions was reduced to four using standardized factor loadings and

modification indices as criteria the items with the lowest factor loading in each dimension

were deleted and the items whose modification index indicated association with multiple

dimensions were eliminated A CFA model with the remaining items was iteratively tested

until it achieved a good fit to the data The outcome of this refinement process was a 16-item

index of four factors Overall model fit by every measure is quite strong for the D-WLS

estimation technique The estimation results for the four-dimension model are presented in

table 4

- 16 -

[Table 4 about here]

Cronbachrsquos coefficient alpha for the 16-item PSM index was 860 and the coefficient

alphas for the four dimensions ranged from 645 to 784 The composite reliability of the set

of reflective indicators for each dimension of PSM ranged from 713 to 826 and indicated

adequate internal consistency The resulting factor structure showed a four-factor structure

with all items loading significantly onto their a priori dimension (p lt 001) and the

standardized factor loadings ranging from 0490 to 0766 in MLE and from 515 to 807 in

D-WLS The results provide support for convergent validity The correlation estimates

between the two factors ranged from 0462 to 0848 in MLE and from 505 to 859 in D-

WLS The confidence intervals (plusmn2 standard errors) around the correlation estimates between

the two factors did not include 100 The result provides support for discriminant validity To

test whether the four-dimension model is superior to other designs such as the two or three-

dimension models a series of CFA were conducted and the results show that the four-

dimension model is a significantly better fit than any of the other designs

To test whether the factorial structure of the PSM instrument is equivalent across the

samples from the twelve countries four nested models were tested as part of multi-group

analysis with MLE (Byrne 2010) First the unconstrained model was the baseline model

which relaxed all equality constraints This model hypothesized the same four-factor structure

of the instrument across twelve countries The fit indices for this baseline model were χsup2 (df =

1176) = 24977 p lt 001 GFI = 0901 CFI = 900 RMSEA = 0020 [019 021] These

indices indicate that the four-factor model of the PSM instrument exhibits acceptable fit

across the samples from the twelve countries Second the measurement weights model tested

- 17 -

the invariance of factor loadings across countries by placing equality constraints on these

parameters The fit indices for this model were χsup2 (df = 1308) = 28088 p lt 001 GFI = 0890

CFI = 887 RMSEA = 0020 [019 021] Since the measurement weights model was nested

within the unconstrained model the chi-square difference text ∆χsup2 (132) = 3111 p lt 001

indicated that some equality constraints of factor loadings did not hold across the twelve

countries Further test results suggested that the equality constraints of structural covariances

and measurement residuals were not upheld across the twelve countries It needs to be

stressed that despite the differences in factor loadings and factor covariances the results

support the main assumption that the same four-factor structure holds across different

countries We may assume that people in different cultures understand PSM in the same

fashion

To find differences in the factor structure of PSM among countries we conducted a

series of CFA with each countryrsquos dataset and followed the same procedure as when

analyzing the sample as a whole As table 5 shows although there are great similarities

between the appropriate items in the countries there are also some differences Even though

the items vary according to national differences the four-factor structure had a good fit to the

data Thus the results confirm that the four-factor structure of PSM can be generalized

[Table 5 about here]

The APS dimension covers both an individualrsquos attraction to public participation and

commitment to public interest but all its items refer to some notion of public interest such as

improving public service aiding community tackling social problems meaningful public

service and contributing to the common good The APS dimension captures an individualrsquos

- 18 -

disposition to serve the public interest through public participation When developing the

items some assumed that the items of committing to public interests can be separated from

those of pursuing public values in the CPV dimension The empirical test showed that

respondents tended to understand committing to public interests as behavior-oriented and so

the items for public interest are not differentiated from participation-oriented items in their

perceptions but are different from those for public values The respondents seem to

understand that pursuing public values is more related to personal beliefs or creeds Thus the

CPV dimension is solely focused on public values and the items of pursing public interests

are combined with those of the APP dimension to make the APS dimension

Public values and their relative importance are neither impervious to change over

time nor the same among all countries (Charles de Jong and Ryan 2011 Davis and West

2009) Moreover some values may be in conflict as Martinsen and Beck Joslashrgensen (2010)

discussed in regard to the tensions between accountability and efficiency It is therefore not

possible to include all public values in one dimension (Beck Joslashrgensen and Bozeman 2007)

Accordingly the items of the CPV dimension are developed for estimating individual

commitment to seek public values which there are general consensus about and which are

generally formulated

When all four dimensions of PSM are included in a survey cross-national

equivalence of the survey can be assured The items represent reflections or manifestations of

each dimension and hence are relatively homogeneous and interchangeable Adding or

deleting items does not change the essential nature of the dimension Thus any combination

of items in a dimension can be regarded as equivalent When using structural equation

modeling (SEM) each dimension needs to have at least three reflective indicators in order to

achieve identification or it needs to have at least two reflective indicators and be allowed to

- 19 -

covary with another dimension that has at least two reflective measures With one indicator

the dimension is identified if the loading and the variance of the uniqueness are both fixed

(Bollen 1989 Edwards 2010) Thus the number of items in each dimension is not restricted

when measuring PSM However we recommend using multiple items for each dimension

because more research is needed to investigate the appropriateness of this PSM instrument

with varying samples

The test results showed that model fit estimators were clearly better for the D-WLS

It means that in this study MLE was more rigorous than D-WLS in assessing the fit of the

model The empirical evidence indicated that MLE may be an acceptable technique for

ordinal data when there is a large sample a five-point Likert scale and the skewness and

kurtosis of variables is within acceptable limits as suggested by Byrne (2010)

Discussion

The analysis reaffirmed the four-dimensional structure of PSM but the results were not

entirely consistent with the original CFA we tested We were ultimately able to converge on a

good-fit operational model for PSM that was also consistent with underlying theory The

analysis is not able to resolve the debate about measurement models however for reasons

discussed below

Choosing the correct measurement model (ie reflective vs formative) is important

as measurement model misspecification results not only in misrepresentation of the epistemic

relationships between constructs and their measures but also in biased parameter estimates of

structural relationships (Jarvis et al 2003 Law and Wong 1999 MacKenzie 2003

MacKenzie Podsakoff and Jarvis 2005) A formative measurement model in isolation is

- 20 -

under-identified and cannot be estimated in the case of covariance-based SEM (CB-SEM)

(Bollen 1989) A construct with formative indicators must emit paths to at least two unrelated

reflective indicators two unrelated latent constructs or some combination of the two in order

to solve the problem of under-identification (Jarvis MacKenzie and Podsakoff 2003

MacCallum and Browne 1993 MacKenzie Podsakoff and Jarvis 2005) Thus ―when

dependent constructs change the empirical nature of the formatively measured construct

changes (Wilcox Howell and Breivik 2008 1223) and ―when specifying formative

measurement models what might seem like an innocuous choice of outcomes to achieve

identification can have fundamental ramifications for the meaning of the formative construct

and the inferences based on it (Edwards 2010 7)

Earlier research found that the association between a construct and its measures

should remain stable regardless of the larger causal model in which the constructs and

measures are embedded (Edwards and Bagozzi 2000) The association between a construct

and its measures is generally stable for reflective measures but associations of formative

measures with their construct are unstable The susceptibility of formative measurement to

both interpretational confounding and external inconsistency recommends that formative

measurement is not an equally attractive alternative to reflective measurement and that

whenever possible researchers should opt for reflective measurement (Howell Breivik and

Wilcox 2007a 2007b Kim Shin and Grover 2010) In the case of second-order models

researchers are limited to only a second-order reflective model in CB-SEM (Chin 2010)

Regardless of the measurement debate and the tradeoffs between reflective or

formative specification this study makes a theoretical argument that PSM is a second-order

formative construct This second-order formative factor model (reflective first-order

formative second-order) is also suggested as an alternative model that captures the

- 21 -

multidimensionality of complex concepts and avoids the shortcomings of first-order

formative measurement (Edwards 2010) However this model itself cannot achieve

identification One possible strategy is to develop direct reflective measures of PSM based on

the view that PSM is a general concept that is not only influenced by its specific dimensions

but also can be assessed using general measures (Edwards 2010 Javis MacKenzie and

Podsakoff 2003)6 However these general measures have not yet been developed for the

PSM construct

Another way around the problem of underidentification of formative models is to

apply the component-based approach known as partial least squares path modeling (PLS-PM)

which can model formative indicators (Fornell and Bookstein 1982 Goumltz Liehr-Gobbers

and Krafft 2010) The CB-SEM and the PLS-PM methods have distinctive statistical

characteristics and selecting an approach to SEM depends on the particular research situation

CB-SEM is the method of choice for theory testing while PLS-PM is appropriate for

prediction-oriented applications (Ringle Goumltz Wetzels and Wilson 2009) One advantage of

the PLS method is that it demands significantly fewer requirements compared to that of

covariance structure analyses but nevertheless delivers consistent estimation results Another

asset of the PLS approach is its ability to deal with formative as well as reflective indicators

even within one structural equation model This indicates that the PLS approach can be

appropriate for explorative analysis of structural equation models (Chin 2010 Goumltz Liehr-

Gobbers and Krafft 2010) A Monte Carlo simulation study to compare CB-SEM and PLS-

PM showed that PLS-PM provides a viable approximation of model parameters when the

prerequisites for CB-SEM are not met (Ringle Goumltz Wetzels and Wilson 2009) A formative

6 For example job satisfaction can be assessed with general reflective measures such as ―in general I am

satisfied with my job and the formative dimensions can be assessed with measures that describe satisfaction

with each job facet (Edwards 2010 15)

- 22 -

composite in the form of indexes needs to be tested to know the relative weights (γ-

parameters) of the dimensions of PSM In PLS-PM a second-order formative factor is

directly measured by observed variables for all the first order factors While this approach

repeats the number of manifest variables used the model can be identified without adding

outcome constructs that would be considered inappropriate or irrelevant from a conceptual

standpoint (Chin 2010 Edwards 2010) The test result of PLS-PM using Smart PLS 20

(Ringle Wende and Will 2005) produced the function PSM = 31APS + 36CPV + 28COM

+ 31SS However one consideration is whether and how construct error is modeled for the

formative construct (Diamantopoulos 2006) When using PLS-PM the researcher is making

the assumption that there is no error at the construct level as all constructs in PLS-PM are

modeled without error When using PLS-PM the absence of an error term will tend to

increase the weights therefore inflating their importance (Bollen and Lennox 1991) The

primary statistic for assessing a formative indicator is its weight the partialized effect of the

indicator on its intended construct controlling for the effects of all other indicators of that

construct (Cenfetelli and Bassellier 2009) A formative measure is essentially a multiple

regression with the construct representing the dependent variable and the indicators as the

predictors (Diamantopoulos and Winklhofer 2001) PSM may be regarded as a weighted

linear composite of its dimensions The weights used to create the composites are different

across CE-SEM and PLS-PM The causal dimensional weights generated in CB-SEM are

estimated in a model in which a construct-level error term is specified (eg PSM = γ1APM +

γ2CPI + γ3COM + γ4SS + ζ) whereas the weights in PLS-PM are not (eg PSM = γ1APM +

γ2CPI + γ3COM + γ4SS) This leads to two different conceptualizations of the PSM construct

one in which the causal indicators do not capture the construct in its entirety and another in

which they do (Hardin et al 2010) These differences need to be discussed in future research

- 23 -

in this area

One practical suggestion to measure PSM is to apply unit weights to its dimensions if

we assume the effects of the four dimensions are equivalent A literature review indicates that

unit weights have substantial predictive validity when compared with regression weights The

literature shows that the empirical correlations between scores generated from differential and

unit-weighting approaches are quite high In creating composite scores unit weights can be

an appropriate approach for weighting under many circumstances (Bobko Roth and Buster

2007) Although the uses of unit-weight composites will likely result in the explanation of

less variance than empirical estimations of dimensional weights they will remain consistently

defined across studies and hence suitable for theory testing Alternatively more thorough

discussions on the dimensions of PSM may suggest fixed-weight composites (Hardin et al

2010) A previous study (Kim 2010 22) also suggested ―Overall indexing PSM by

measuring all the dimensions of PSM and simply summing the averages of each dimension

can make it easier to measure PSM in practice as PSM = APS + CPV + COM + SS

An advantage of regarding PSM as a second-order formative construct is that the

dimensions can be analyzed both individually and collectively This is important as the

dimensions are expected to have different antecedents and consequences For instance Perry

(1997) found some significant differences in the influence of independent variables on the

four dimensions of PSM Taylor (2007) confirmed that certain dimensions are more important

than others in influencing work outcomes Moynihan and Pandey (2007a) reported that red

tape is negatively related to one dimension but reform orientation and hierarchical authority

is positively related to another dimension A recent study found that the SS dimension has a

positive relationship with occupational intention but the other dimensions do not (Liu et al

2011) Thus it is valuable to analyze the causes and effects of individual dimensions of PSM

- 24 -

as well as of the PSM construct

Conclusion

The purpose of this study is to develop a measurement instrument for PSM for cross-national

research and comparison On the basis of the sharpened construct of PSM (Kim and

Vandenabeele 2010) using data from an international survey conducted in 12 countries we

found that the four-factor structure of PSM is generalizable across country cases This study

provides the measurement index of PSM with the four dimensions of APS CPV COM and

SS The APS dimension measures the instrumental motives of PSM the CPV dimension

determines the value-based motives of PSM while the COM and SS dimensions gauge the

identification motives and self-sacrifice respectively When all four dimensions of PSM are

used in a survey cross-national equivalence of survey items can be assured The items for

measuring PSM in future research are suggested in Appendix 4

This study has several limitations First looking for samples which were comparable

to the largest possible extent we opted for surveys in local government The results found

and discussed in this study are based on PSM among local government employees The

results might be different when surveying central government employees or other specific

groups in the public sector Thus the potential impact of the sample bias needs to be

considered in understanding this study It also means that the survey items for measuring

PSM should be examined with different samples in various areas and at different government

levels

Second the test results showed that there is a high degree of correlation among the

dimension such as APS CPV and COM Future research should not only investigate whether

- 25 -

the items reflect the core content of the dimensions but also continue to develop more

appropriate items for better discriminant validity of each dimension Third the content of

APP dimension was assumed to be distinct from the CPV dimension that focuses on personal

disposition to pursue public interest and public values However the public interest sub-

dimension was separated from the public values sub-dimension in the CPV dimension and

combined with the APP dimension to form the APS dimension Further research should

analyze the CPV dimension and investigate the appropriateness of the APS dimension

In order to facilitate research internationally the PSM construct needs to be

improved conceptually and operationally Kim and Vandenabeele (2010) have dealt with the

conceptual composition and operational dimension of PSM as well as the measurement

model of PSM The next step is to develop an internationally robust measurement instrument

of PSM This study proposes that scholars use all four of the dimensions in measuring the

conceptual components of PSM such as self-sacrifice instrumental value-based and affective

motives because they collectively represent PSM It provides a set of appropriate items for

measuring PSM and confirms the dimensional structure of PSM The survey results suggest

that people in different cultures understand PSM in the same fashion We know that this study

is simply a step along the way to developing a more robust measure but we hope it

contributes to enhanced cross-national research and comparison along with the generation of

cumulative knowledge internationally

- 26 -

REFERENCES

Anderson James C and David D Gerbing 1988 Structural Equation Modeling in Practice

A Review and Recommended Two-Step Approach Psychological Bulletin 103

411-423

Arbuckle James L 2009 Amos 180 Userrsquos Guide Crawfordville FL Amos Development

Co

Beck Joslashrgensen Torben and Barry Bozeman 2007 Public Values An Inventory

Administration amp Society 39(3) 354-381

Beutel Ann M and Margaret Mooney Marini 1995 Gender and values American

Sociological Review 60 436 - 448

Bobko Philip Philip L Roth and Maury A Buster 2007 The Usefulness of Unit Weights in

Creating Composite Scores A Literature Review Application to Content Validity and

Meta-Analysis Organizational Research Methods 10(4) 689-709

Bollen Kenneth A 1989 Structural Equations with Latent Variables New York Wiley

Bollen K A and R Lennox 1991 Conventional Wisdom on Measurement A Structural

Equation Perspective Psychological Bulletin 110(2) 305-314

Byrne Barbara M 2010 Structural Equation Modeling with Amos Basic Concepts

Applications and Programming 2nd

ed New York Routledge

Camilleri Emanuel 2006 Towards Developing an Organizational Commitment ― Public

Service Motivation Model for Maltese Public Service Employees Public Policy

and Administration 21 63-83

Castaing Seacutebastien 2006 The Effects of Psychological Contract Fulfillment and Public

Service Motivation on Organizational Commitment in the French Civil Service

Public Policy and Administration 21 84-98

Cenfetelli Ronald T and Geneviegraveve Bassellier 2009 Interpretation of Formative

Measurement in Information Systems Research MIS Quarterly 33(4) 689-707

Charles Michael B W Martin de Jong and Neal Ryan 2011 Public Values in Western

- 27 -

Europe A Temporal Perspective American Review of Public Administration 41

75-91

Chin Wynne W 2010 How to Write Up and Report PLS Analysis In Handbook of Partical

Least Squares edited by V Esposito Vinzi et al 655-690 Berlin Springer-Verlag

Coursey David H and Sanjay K Pandey 2007a Public Service Motivation Measurement

Testing an Abridged Version of Perryrsquos Proposed Scale Administration amp Society

39 547-568

Coursey David H and Sanjay K Pandey 2007b Content Domain Measurement and

Validity of the Red Tape Concept American Review of Public Administration 37

342-361

Coursey David H James L Perry Jeffrey L Brudney and Laura Littlepage 2008

Psychometric Verification of Perryrsquos Public Service Motivation Instrument Results

for Volunteer Exemplars Review of Public Personnel Administration 28 79-90

Davis Paul and Karen West 2009 What Do Public Values Mean for Public Action Putting

Public Values in Their Plural Place American Review of Public Administration 39

602-618

DeHart-Davis Leisha Justin Marlowe and Sanjay K Pandey 2006 Gender Dimensions of

Public Service Motivation Public Administration Review 66 873-887

Diamantopoulos Adamantios 2006 The error term in formative measurement models

Interpretation and modeling implications Journal of Modeling in Management 1(1)

7-17

Diamantopoulos Adamantios and Judy A Siguaw 2006 Formative versus Reflective

Indicators in Organizational Measure Development A Comparison and Empirical

Illustration British Journal of Management 17 263-282

Diamantopoulos Adamantios and Heidi M Winklhofer 2001 Index Construction with

Formative Indicators An Alternative to Scale Development Journal of Marketing

Research 38 269-277

Edwards Jeffrey R 2010 The Fallacy of Formative Measurement Organizational Research

- 28 -

Methods Online First published on August 19 2010 as

doi1011771094428110378369

Edwards Jeffrey R and Richard P Bagozzi 2000 On the Nature and Direction of

Relationships between Constructs and Measures Psychological Methods 5 155-

174

Flora David B and Patrick J Curran 2004 An Empirical Evaluation of Alternative

Methods of Estimation for Confirmatory Factor Analysis with Ordinal Data

Psychological Methods 9 466-491

Fornell Claes and Fred L Bookstein 1982 Two structural equations models LISREL and

PLS applied to customer exit-voice theory Journal of Marketing Research 19 440-

452

Giauque David Adrian Ritz Freacutedeacuteric Varone Simon Anderfuhren-Biget and Christian

Waldner 2011 Putting Public Service Motivation into Context A Balance between

Universalism and Particularism Forthcoming in International Review of

Administrative Sciences 77(2)

Goumltz Oliver Kerstin Liehr-Gobbers and Manfred Krafft 2010 Evaluation of Structural

Equation Models Using the Partial Least Squares (PLS) Approach In V Esposito

Vinzi et al (eds) Handbook of Partial Least Squares 691-711 DOI 101007978-3-

540-32827-8_30 Springer-Verlag Berlin Heidelberg

Guinot Christiane Julie Latreille and Michel Tenenhaus 2001 PLS Path Modelling and

Multiple Table Analysis Application to the Cosmetic Habits of Women in Ile-de-

France Chemometrics and Intelligent Laboratory Systems 58 247-259

Hair Joseph F Jr William C Black Barry J Babin and Rolph E Anderson 2010

Multivariate Data Analysis 7th

ed Upper Saddle River NJ Pearson Education

Hardin Andrew M Jerry Cha-Jan Chang Mark A Fuller and Gholamreza Torkzadeh 2010

Formative Measurement and Academic Research In Search of Measurement

Theory Educational and Psychological Measurement Online First published on

June 11 2010 as doi101177001316440370208

- 29 -

Howell Roy D Einar Breivik and James B Wilcox 2007a Reconsidering Formative

Measurement Psychological Methods 12 205-218

Howell Roy D Einar Breivik and James B Wilcox 2007b Is Formative Measurement

Really Measurement Reply to Bollen (2007) and Bagozzi (2007) Psychological

Methods 12 238-245

Jarvis Cheryl Burke Scott B MacKenzie and Philip M Podsakoff 2003 A Critical Review

of Construct Indicators and Measurement Model Misspecification in Marketing and

Consumer Research Journal of Consumer Research 30 199-218

Joumlreskog Karl G and Dag Soumlrbom 2005 LISREL 872 (Computer Software) Lincolnwood

IL Scientific Software International

Kim Gimun Bongsik Shin and Varun Grover 2010 Investigating Two Contradictory Views

of Formative Measurement in Information Systems Research MIS Quarterly 34

345-365

Kim Sangmook 2009a Revising Perryrsquos Measurement Scale of Public Service Motivation

American Review of Public Administration 39 149-163

Kim Sangmook 2009b Testing the Structure of Public Service Motivation in Korea A

Research Note Journal of Public Administration Research and Theory 19 839-

851

Kim Sangmook 2010 Testing a Revised Measurement of Public Service Motivation

Reflective versus Formative Specification Journal of Public Administration

Research and Theory Doi101093jopartmuq048

Kim Sangmook and Wouter Vandenabeele 2010 A strategy for building public service

motivation research internationally Public Administration Review 70 701-709

Kline Rex B 2005 Principles and Practice of Structural Equation Modeling 2nd

ed New

York Guilford Press

Law Kenneth and Chi-Sum Wong 1999 Multidimensional Constructs in Structural

Equation Analysis An Illustration using the Job Perception and Job Satisfaction

Constructs Journal of Management 25 143-160

- 30 -

Leisink Peter and Bram Steijn 2009 Public Service Motivation and Job Performance of

Public Sector Employees in the Netherlands International Review of

Administrative Sciences 75 35-52

Liu Bangcheng Chun Hui Jin Hu Wensheng Yang and Xinli Yu 2011 How well can

Public Service Motivation Connect with Occupational Intention International

Review of Administrative Sciences 77 191-211

Liu Bangcheng Ningyu Tang and Xiaomei Zhu 2008 Public Service Motivation and Job

Satisfaction in China An Investigation of Generalisability and Instrumentality

International Journal of Manpower 29 684-699

MacCallum Robert C and Michael W Browne 1993 The Use of Causal Indicators in

Covariance Structure Models Some Practical Issues Psychological Bulletin 114

533-541

MacKenzie Scott B 2003 The dangers of poor construct conceptualization Journal of the

Academy of Marketing Science 31(3) 323-326

MacKenzie Scott B Philip M Podsakoff and Cheryl Burke Jarvis 2005 The Problem of

Measurement Model Misspecification in Behavioral and Organizational Research

and Some Recommended Solutions Journal of Applied Psychology 90 710-730

Martinsen Dorte Sindbjerg and Torben Beck Joslashrgensen 2010 Accountability as a

Differentiated Value in Supranational Governance American Review of Public

Administration 40 742-760

Moynihan Donald P and Sanjay K Pandey 2007a The Role of Organizations in Fostering

Public Service Motivation Public Administration Review 67 40-53

Moynihan Donald P and Sanjay K Pandey 2007b Finding Workable Levers over Work

Motivation Comparing Job Satisfaction Job Involvement and Organizational

Commitment Administration amp Society 39 803-832

Pandey Sanjay K and Edmund C Stazyk 2008 Antecedents and Correlates of Public

Service Motivation In Motivation in Public Management edited by James L Perry

and Annie Hondeghem 101-117 Oxford Oxford University Press

- 31 -

Perry James L 1996 Measuring Public Service Motivation An Assessment of Construct

Reliability and Validity Journal of Public Administration Research and Theory 6

5-22

Perry James L 1997 Antecedents of Public Service Motivation Journal of Public

Administration Research and Theory 7 181-197

Perry James L 2000 Bringing Society in Toward a Theory of Public Service Motivation

Journal of Public Administration Research and Theory 10 471-488

Perry James L and Annie Hondeghem 2008a Motivation in Public Management Oxford

Oxford University Press

Perry James L and Annie Hondeghem 2008b Directions for Future Theory and Research

In Motivation in Public Management edited by James L Perry and Annie

Hondeghem 294-313 Oxford Oxford University Press

Perry James L Annie Hondeghem and Lois Recascino Wise 2010 Revisiting the

Motivational Bases of Public Service Twenty Years of Research and an Agenda for

the Future Public Administration Review 70(5) 681-690

Perry James L and Wouter Vandenabeele 2008 Behavioral dynamics Institutions

Identities and Self-Regulation In Motivation in Public Management edited by

James L Perry and Annie Hondeghem 56-79 Oxford Oxford University Press

Perry James L and Lois Recascino Wise 1990 The Motivational Bases of Public Service

Public Administration Review 50 367-373

Rainey Hal G 1982 Reward Preference among Public and Private Managers In Search of

the Service Ethic American Review of Public Administration 16 288-302

Ringle Christian M Oliver Goumltz Martin Wetzels and Bradley Wilson 2009 On the Use of

Formative Measurement Specifications in Structural Equation Modeling A Monte

Carlo Simulation Study to Compare Covariance-Based and Partial Least Squares

Model Estimation Methodologies Faculty of Economics and Business Administration

Maastricht Univeristy

Ringle C M S Wende and S Will 2005 Smart PLS 20 (M3) Beta

- 32 -

httpwwwsmartplsde

Ritz Adrian 2011 Attraction to Public Policy Making A Qualitative Inquiry into

Improvements in PSM Measurement Forthcoming in Public Administration

Spector Paul E 1992 Summated Rating Scales Construction An Introduction Newbury

Park CA Sage

Steenbergen Marco R 1996 Compassion and American public opinion An analysis of the

NES Humanitarianism Scale Report to the Board of Overseers of the National

Election Studies

Steffen Patrick R and Kevin S Masters 2005 Does compassion mediate the intrinsic

religion-health relationship Annals of Behavioral Medicine 30 217-223

Taylor Jeannette 2007 The Impact of Public Service Motives on Work Outcomes in

Australia A Comparative Multi-Dimensional Analysis Public Administration 85

931-959

Vandenabeele Wouter 2007 Towards a Theory of Public Service Motivation An

Institutional Approach Public Management Review 9 545-556

Vandenabeele Wouter 2008a Government Calling Public Service Motivation as an Element

in Selecting Government as an Employer of Choice Public Administration 86

1089-1105

Vandenabeele Wouter 2008b Development of a Public Service Motivation Measurement

Scale Corroborating and Extending Perryrsquos Measurement Instrument International

Public Management Journal 11 143-167

Vandenabeele Wouter 2009 The Mediating Effect of Job Satisfaction and Organizational

Commitment on Self-Reported Performance More Robust Evidence of the PSM-

Performance Relationship International Review of Administrative Sciences 75 11-

34

Wilcox James B Roy D Howell and Einar Breivik 2008 Questions about Formative

Measurement Journal of Business Research 61 1219-1228

Wise Lois Recascino 2000 The Public Service Culture In Public Administration Concepts

- 33 -

and Cases 7th ed edited by Richard J Stillman II 342-353 Boston Houghton

Mifflin

Wright Bradley E 2008 Methodological Challenges associated with Public Service

Motivation Research In Motivation in Public Management edited by James L

Perry and Annie Hondeghem 80-98 Oxford Oxford University Press

Wright Bradley E and Sanjay K Pandey 2008 Public Service Motivation and the

Assumption of Person-Organization Fit Testing the Mediating Effect of Value

Congruence Administration amp Society 40 502-521

- 34 -

Table 1

Possible Indicators of PSM and Descriptive Statistics (n = 2868)

Dimensions and items Mean Standard

Deviation

Item-Total

Correlation

Attraction to Public Participation (APP)

APP1 I am interested in helping to improve public service

APP2 I am satisfied when I see people benefiting from the public programs I was

involved in

APP3 I like to discuss topics regarding public programs and policies with others

APP4 I believe that public sector activities contribute to our general welfare

APP5 I admire people who initiate or are involved in activities to aid my

community

APP6 Contributing to public programs and policies helps me realize myself

APP7 It is important to contribute to activities that tackle social problems

412

441

355

393

423

351

409

696

662

917

753

672

881

734

524

421

482

412

493

478

552

Commitment to Public Values (CPV)

Sub-dimension for Public Interests (CPV1)

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

CPI3 I would prefer seeing public officials do what is best for the whole

community even if it harmed my interests

CPI4 Serving the public interest is more important than helping a single individual

Sub-dimension for Public Values (CPV2)

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of public

services

CPV3 It is fundamental that public services respond to the needs of the citizens

CPV4 Decisions regarding public services should be democratic despite the time

and effort it takes

CPV5 Everybody is entitled to a good service even if it costs a lot of money

CPV6 It is fundamental that the interests of future generations are taken into

account when developing public policies

CPV7 To act ethically is essential for public servants

CPV8 I believe that public employees must always be aware of the legitimacy of

their activities

CPV9 I personally identify with the aim of protecting individual liberties and rights

418

409

348

337

428

416

433

394

367

426

441

434

414

695

662

899

995

730

704

710

852

932

689

681

692

763

525

596

480

251

484

437

356

325

376

439

462

402

392

Compassion (COM) COM1 It is difficult for me to contain my feelings when I see people in distress

COM2 I feel sympathetic to the plight of the underprivileged

COM3 I empathize with other people who face difficulties

COM4 I have little compassion for people in need who are unwilling to take the

first step to help themselves

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

354

387

388

299

412

391

951

825

707

1083

755

765

399

511

389

011

487

502

Self-Sacrifice (SS)

SS1 Making a difference to society means more to me than personal achievements

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS5 People should give back to society more than they get from it

SS6 Serving other citizens would give me a good feeling even if no one paid me for

it

SS7 I would agree to a good plan to make a better life for the poor even if it costs

me money

357

329

339

302

345

363

350

904

872

949

982

888

1049

895

508

584

454

531

412

368

546

- 35 -

Table 2

Background of Respondents (n = 2868)

Variables Characteristics Respondents

Sex

Male

Female

NA

1209

1622

37

422

566

13

Length of service

(years)

0 ~ 10

10 ~ 20

20 ~ 30

30+

NA

1126

723

484

491

44

393

252

169

171

15

Organizational status

―Do you supervise employees

No

Yes

NA

1472

1076

320

513

375

112

Country

Australia

Belgium

China

Denmark

France

Italy

Korea

Lithuania

The Netherlands

Switzerland

United Kingdom

USA

249

214

230

249

266

162

253

236

249

250

260

250

87

75

80

87

93

56

88

82

87

87

91

87 Note NA = no answer

- 36 -

Table 3

Standardized factor loadings and correlations for the 31-item PSM index (n = 2868)

Dimensions and items SFL (MLE) SFL (D-WLS) Alpha

APP

APP1

APP2

APP4

APP5

APP6

APP7

597

486

455

558

490

600

657

552

514

625

549

665

699

CPV1

CPI1

CPI2

CPI3

CPI4

588

660

488

289

653

732

535

294

542

CPV2

CPV1

CPV2

CPV3

CPV4

CPV5

CPV6

CPV7

CPV8

CPV9

611

546

436

393

418

534

598

496

460

690

631

505

457

486

601

669

573

530

745

COM

COM1

COM2

COM3

COM5

COM6

496

569

498

597

607

516

651

538

670

684

687

SS

SS1

SS2

SS3

SS4

SS5

SS6

SS7

567

736

558

728

467

409

663

660

762

591

695

515

494

709

782

Measures of fit (MLE) χsup2 (df = 424) = 43699 p lt 001 CFI = 835 GFI = 899

RMSEA = 057 [055 059]

Measure of fit (D-WLS) χsup2 (df = 424) = 39247 p lt 001 CFI = 971 GFI = 976

RMSEA = 054 [052 055]

Inter-factor correlations MLE (D-WLS)

1 2 3 4

1 APS

2 CPV1 996 (988)

3 CPV2 902 (890) 798 (798)

4 COM 787 (783) 758 (762) 779 (783)

5 SS 700 (742) 800 (859) 509 (537) 720 (729)

Note SFL = standardized factor loading All factor loadings and correlations are significant at p lt 001

- 37 -

Table 4

Standardized factor loadings and correlations for the 16-item PSM index (n = 2868)

Dimensions and items SFL

(MLE)

SFL

(D-WLS)

APS

APP5 I admire people who initiate or are involved in activities to aid my

community

APP7 It is important to contribute to activities that tackle social problems

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

611

592

637

666

665

669

690

751

CPV

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of

public services

CPV6 It is fundamental that the interests of future generations are taken

into account when developing public policies

CPV7 To act ethically is essential for public servants

612

546

551

590

698

631

618

658

COM

COM1 It is difficult for me to contain my feelings when I see people in

distress

COM2 I feel sympathetic to the plight of the underprivileged

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

490

546

596

604

515

608

669

679

SS

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS7 I would agree to a good plan to make a better life for the poor even if

it costs me money

763

571

766

678

807

629

747

755

Measures of fit (MLE)

χsup2 (df = 98) = 8490 p lt 001

CFI = 939 GFI = 965

RMSEA = 052 [049 055]

Measure of fit (D-WLS)

χsup2 (df = 98) = 6069 p lt 001

CFI = 988 GFI = 992

RMSEA = 043 [039 046]

Alpha CR (D-WLS) Inter-factor correlations MLE (D-WLS)

1 APS 717 788 1 APS 2 CPV 3 COM

2 CPV 663 747 848 (859)

3 COM 645 713 781 (791) 789 (795)

4 SS 784 826 565 (596) 462 (505) 715 (726)

Note SFL = standardized factor loading Alpha = Cronbachrsquos alpha CR = composite reliability

All factor loadings and correlations are significant at p lt 001

- 38 -

Table 5

Results of CFA in the individual countries Standardized factor loadings and measures of fit

Factors

amp Items

Australia Belgium China Denmark France Italy

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

672

756

799

785

804

943

710

703

597

790

799

730

584

675

685

681

731

742

578

748

674

643

765

739

645

704

678

708

812

779

632

604

765

766

672

853

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

552

593

560

699

727

632

624

605

560

713

709

645

702

602

616

724

686

735

567

560

715

654

605

811

608

658

587

696

755

698

673

620

523

782

697

654

COM

COM1

COM2

COM3

COM5

COM6

623

584

662

688

701

739

732

608

475

808

666

581

578

616

605

649

670

673

699

751

737

679

816

776

568

546

801

626

673

850

637

546

741

715

601

822

SS

SS2

SS3

SS4

SS7

654

794

806

892

767

813

727

693

799

793

725

838

857

820

773

896

855

842

806

749

792

829

753

884

847

731

753

889

816

753

815

692

664

795

782

734

Inter-factor

correlations

093

~

752

163

~

810

320

~

773

323

~

825

452

~

827

497

~

855

323

~

850

321

~

847

498

~

899

536

~

902

425

~

829

476

~

865

Sample size 249 214 230 249 266 162

Measures of Fit

χsup2 638 444 881 732 765 554 817 625 735 565 750 564

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 980 100 939 987 969 997 967 994 975 997 949 995

GFI 961 986 941 986 948 991 949 992 955 992 926 986

RMSEA 036 00 063 050 051 026 053 035 045 026 059 033

- 39 -

Table 5 (Continued)

Factors

amp Items

Korea Lithuania The Netherlands Switzerland UK USA

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

619

701

667

706

781

717

666

512

682

751

605

779

557

629

630

620

719

729

535

621

702

643

694

735

655

725

738

710

804

822

535

621

702

643

694

735

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

594

607

722

722

644

722

674

535

533

790

589

711

568

554

538

655

628

545

627

580

615

711

666

704

800

512

555

873

581

646

627

580

615

711

666

704

COM

COM1

COM2

COM3

COM5

COM6

694

677

550

763

736

627

557

639

428

550

696

534

737

623

508

766

616

608

659

592

683

703

661

769

682

583

729

739

644

803

659

592

683

703

661

769

SS

SS2

SS3

SS4

SS7

776

752

679

779

770

811

707

721

635

725

734

706

814

656

743

820

767

764

745

632

689

788

662

744

671

818

832

830

864

802

745

632

689

788

662

744

Inter-factor

correlations

432

~

726

466

~

746

376

~

788

406

~

902

377

~

852

455

~

836

637

~

818

660

~

828

317

~

852

360

~

852

637

~

818

660

~

828

Sample size 253 236 249 250 260 250

Measures of Fit

χsup2 896 573 612 462 773 616 815 715 769 542 815 715

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 951 996 975 100 958 993 957 991 972 998 957 991

GFI 942 988 960 989 951 987 948 990 955 990 948 990

RMSEA 059 028 034 00 050 034 053 044 048 022 053 044

Note All factor loadings and correlations are significant at p lt 001

- 40 -

Appendix 1 Dimensions and possible items of PSM (the first round)

1 Attraction to Public Participation

(1) I am interested in developing public programs helping my country or community (Kim 2009a)

(2) Sharing my views on public policies with others is attractive to me (Kim 2009a)

(3) I am satisfied when seeing people get benefits from the public programs I was involved in (Kim 2009a)

(4) I like to discuss political subjects with others (Giauque et al 2011)

(5) I think the governmental activities contribute to our welfare

(6) I like to initiate actions to help out my community

(7) I do not care much about politicians (Perry 1996)

2 Commitment to Public Values

Possible items for public interest

(1A) Meaningful public service is very important to me (Perry 1996)

(2A) It is important for me to contribute for the common good (Giauque et al 2011)

(3A) I recognize myself with the mission of protection of liberties and rights endorsed by the nation-state

(Giauque et al 2011)

(4A) I would prefer seeing public officials do what is best for the whole community even if it harmed my

interests (Perry 1996)

(5A) To me serving the public interest is more important than helping other people (Vandenabeele 2008b)

Possible items for public values

(1B) I recognize myself with the promotion of the equality of chances endorsed by the nation-state (Giauque et

al 2011)

(2B) It is fundamental that the public service provided to the citizen is regular and continuous (Giauque et al

2011)

(3B) It is fundamental that public service innovates to respond to the new needs of the citizensclients (Giauque

et al 2011)

(4B) The dignity and well-being of all should be the most important concerns in any society (Steenbergen

1996)

(5B) It is important that public servants account for all the costsexpenses they make (Vandenabeele 2008b)

(6B) We have to do everything in our power to pursue the goal of democracy

3 Compassion

(1) It is difficult for me to contain my feelings when I see people in distress (Perry 1996)

(2) I feel sympathetic to the plight of the underprivileged (Kim 2009a)

(3) I am often reminded by daily events how dependent we are on one another (Perry 1996)

(4) I feel compassion for others in the difficulties they are facing (Steffen and Masters 2005)

(5) It is not really my problem if others are in trouble and need help (Beutel and Marini 1995)

(6) I find it hard to be sympathetic toward starving people in foreign lands when there is so much trouble in our

own country (Beutel and Marini 1995)

(7) I get very upset when I see other people treated unfairly (Beutel and Marini 1995)

(8) I would agree to a good plan to make a better life for the poor even if it cost me money (Beutel and Marini

1995)

(9) To me patriotism includes seeing the welfare of others (Perry 1996)

(10) I care very much about other people

4 Self-Sacrifice

(1) Making a difference in society means more to me than personal achievements (Perry 1996)

(2) I am prepared to make enormous sacrifices for the good of society (Perry 1996)

(3) I believe in putting duty before self (Perry 1996)

(4) I am one of those rare people who would risk personal loss to help someone else (Perry 1996)

(5) I think people should give back to society more than they get from it (Perry 1996)

(6) Serving other citizens would give me a good feeling even if no one paid me for it (Perry 1996)

(7) Much of what I do is for a cause bigger than myself (Perry 1996)

- 41 -

Appendix 2 Newly suggested items through the first round

1 Items for the Attraction to Public Participation Dimension

(1) In the public domain every person should be counted as equal

(2) I enjoy to think and argue in a political way

(3) I am satisfied when seeing my ideas being integrated in public policies

(4) I can easily relate to the way of thinking and arguing of politicians

(5) Working in the public sector I can better realize myself

2 Items for the Public Values Sub-Dimension

(1) Public service should give value for money

(2) Cost considerations should play an important role in decisions about public service provision

(3) It is fundamental that the interests of future generations are taken into account when developing public

policies

(4) To act with correct ethics is for public servants as important as their professional competence

(5) Primarily public servants should be accountable to the public and not to their superiors

3 Items for the Self-Sacrifice Dimension

(1) I would agree to a good plan to make a better life for the poor even if it costs me money

- 42 -

Appendix 3 An example of a summary of comments and suggestions (the second round)

On the item ―I am interested in developing public programs helping my country or community

A It is a good sentence but is it possible for everyone to ―develop public programs Even public officials are

limited to do this

B OK

C or the special group

D Double barrel and difficult to translate to my country ―I am interested in helping to improve public services

E Good but I feel that wording requires revision If we draw sample from local government then country is

redundant ―I am interested in developing public policy programs that aim in helping my community

F 4 (General appreciation from1 to 4) Not clear whether it refers to ones work or to ones concern as a citizen

―(In my work) I am interested in contributing to the development of public etchellip

G Instead of development I would use implementing

H Delete public which is redundant because of the characteristic helping my country or community but add

or implementing ―I am interested in developing or implementing programs helping my country or

community

I Referring to dimension Wording Similar to item 6

Note Names have been changed to alphabets in order to ensure the anonymity of participants

- 43 -

Appendix 4 Dimensions and Items of Public Service Motivation for Future Research

Dimensions and items

Attraction to Public Service (APS)

1 I am interested in helping to improve public service

2 I admire people who initiate or are involved in activities to aid my community

3 It is important to contribute to activities that tackle social problems

4 Meaningful public service is very important to me

5 It is important for me to contribute to the common good

Commitment to Public Values (CPV)

1 I think equal opportunities for citizens are very important

2 It is important that citizens can rely on the continuous provision of public services

3 It is fundamental that the interests of future generations are taken into account when

developing public policies

4 To act ethically is essential for public servants

5 I believe that public employees must always be aware of the legitimacy of their activities

6 I personally identify with the aim of protecting individual liberties and rights

Compassion (COM)

1 It is difficult for me to contain my feelings when I see people in distress

2 I feel sympathetic to the plight of the underprivileged

3 I empathize with other people who face difficulties

4 I get very upset when I see other people being treated unfairly

5 Considering the welfare of others is very important

Self-Sacrifice (SS)

1 I am prepared to make sacrifices for the good of society

2 I believe in putting civic duty before self

3 I am willing to risk personal loss to help society

4 I would agree to a good plan to make a better life for the poor even if it costs me money

- 7 -

Stazyk 2008 Perry 1996 1997 2000 Perry and Wise 1990 Perry and Vandenabeele 2008

Taylor 2007 Vandenabeele 2007 2008a 2009) An individualrsquos PSM is shaped by the

individualrsquos APP CPV COM and SS Deleting a dimension means omitting a certain part of

PSM Therefore it is reasonable to suggest that the dimensions of PSM are formative and for

PSM to be regarded as a composite of its dimensions

Each dimension of PSM can be measured using reflective indicators that satisfy the

following criteria (Jarvis MacKenzie and Podsakoff 2003) a relative homogeneity and

interchangeability of indicators pertaining to each dimension a high degree of covariation

among indicators of each dimension and the expectation that the indicators of each

dimension are likely to be affected by the same antecedents and have the same consequences

Reflective indicators are essentially interchangeable and so adding or removing indicators

does not change the essential nature of a dimension Thus we consider PSM to be a second-

order formative construct with several dimensions that each has several reflective indicators

(reflective first-order formative second-order)

Based upon the refinement and clarification of the operational dimensions of PSM

(Kim and Vandenabeele 2010) items were developed for the four dimensions The items

selected were a mix of pre-existing items that loaded highly on PSM dimensions in multiple

studies (Giauque et al 2011 Kim 2009a Perry 1996 Vandenabeele 2008b) Where

appropriate items were not available new items were drafted from other previous studies

creating an initial total of 35 proposed items For the dimension of APP which focuses on

individual disposition to work in the public sector to participate in the public policy process

and in activities for community and social development seven possible items were suggested

The CPV dimension focuses on personal disposition to pursue public interest and public

values Two sub-dimensions of CPV seemed plausible one focusing on public interest and

- 8 -

another on public values such as social equity equal opportunity and democracy Thus five

possible items for the public interest sub-dimension (CPV1) and six items for the public

values sub-dimension (CPV2) were developed For the dimension of COM which focuses on

affective bonding with the identified objects such as other members of a social category or of

a political system ten possible items were suggested For the SS dimension Perryrsquos original

items (1996) were the primary source of items The full list of items is provided in Appendix

1

Review Processes

A group of researchers who shared our interest in better measurement of PSM collaborated

using e-mail in November 20091 The objective was to develop a measurement instrument

for PSM with a robust comparative character in other words to create a valid instrument (on

a national basis) that can be used internationally The first round for reviewing the possible

items lasted from the 16th

of November until the 7th

of December 2009 and all the

participants were asked to review these proposed items For this review two criteria were

employed (1) Do these items refer to the dimensions they claim to refer to and (2) Are

these items meaningful across various national contexts and in particular to the situation of

civil servants in that country Participants were also asked to add new items that might better

reflect a particular dimension or to rephrase items that were already on the list

A second round of item refinement lasted from the 9th

of December until the 22nd

of

December 2009 During this time 11 newly suggested items from the first round were

considered (see Appendix 2) A summary of all comments and suggestions emerging from the

1 Emmanuel Camilleri David Giauque Gerhard Hammerschmid Isabel Egger-Peitler Renate Meyer and

Cristian Pliscoff also participated in the review processes

- 9 -

first process was distributed to the participants to stimulate comments in the second round

(see Appendix 3) Based on the comments received in the second round discussion the

possible items for an international survey were selected However some items required

further discussion and so a third round was conducted from the 15th

of January until the 23rd

of January 2010 One issue that remained unresolved in the discussions relates to the

dimension of CPV Some argued that it is reasonable to divide CPV into two sub-dimensions

because committing to public interests is different from pursuing public values and there are

many important public values in democratic countries while others contended that CPV can

cover all important public values including public interest and there is no real benefit to add

one more dimension These differences were left to be resolved through empirical research

This did not affect the final items however and 33 items were confirmed The resulting

survey questionnaire was distributed to all the collaborators on February 9 2010 in order to

conduct a survey in each country

Measures

PSM was measured using the newly developed 33-item index 7 items for the dimension of

APP 13 items for CPV (4 for CPV1 and 9 for CPV2) 6 items for COM and 7 for SS In the

survey questionnaire a five-point Likert type scale (1 = strongly disagree 5 = strongly agree)

was used This scale is among the most widely used response formats in many empirical

investigations within the social and behavioral science in which there are no ―correct or

―incorrect responses (Spector 1992) The items of PSM were randomly distributed in the

survey format Gender age education length of service and employment status were also

measured in this survey

It was necessary to translate the measurement items into the official (native) language

- 10 -

of each country As this raised the possibility of inaccurate translations the back-translation

of the items into native languages was validated by asking two researchers to independently

translate the English-language items The researchers were instructed to consult with each

other over differences between translated items and to test the survey items with some native

respondents

[Table 1 about here]

Samples

The international survey was fielded from March to September 2010 The data upon which

this study are based were collected from civil servants in local governments in Australia

Belgium China Denmark France Italy Korea Lithuania the Netherlands Switzerland the

United Kingdom and the USA The principal criterion was that the samples collected in the

various countries should be comparable to the largest possible extent The criterion ruled out

national and state government as the competencies of national governments and also of state

governments if they exist constitutionally are different among most of the countries included

in the research Therefore we opted for local government as our focus The structure and

functions of local governments can differ widely even within countries and so the survey in

each country was focused on town hall bureaucrats in city county or township government

The respondents were permanent employees from both managerial and non-managerial levels

and from both administrative departments (ie no direct contact with the public) and service

departments (ie likely to have direct contact with the public) Despite being employed by

local governments in many or all of the countries police officers firefighters school teachers

public transport workers artists and musicians as well as nurses and doctors were excluded

- 11 -

from the sample

We aimed to receive 250 respondents for each country as this would enable us to

carry out the analyses we wanted to perform on the data Although there is no ideal sample

size one study recommends a sample of 200 to provide a sound basis for estimation (Hair

Black Babin and Anderson 2010) The method of data collection was left open Obviously a

web-survey is the easiest way to collect the data because it is economical to collect and to

code however compared to paper-pencil surveys this method has some drawbacks First not

everybody in the sample may have access to the internet which may cause a bias in the

responses Second internet surveys are generally known to produce lower response rates

Web surveys were recommended if possible but paper-pencil surveys were deemed

acceptable in cases where it was not possible to conduct a web survey Researchers were

encouraged to contact just one municipality except in cases where it was not possible to

gather enough data In such instances responses from multiple municipalities were accepted

By the end of September 2010 a total of 2868 responses were obtained2

2 In Australia seven local councils each with between from 70 and 232 employees were surveyed Paper-pencil

surveys were used in two councils while the rest relied on a web survey The local governments limited

distribution of paper-pencil surveys to only one or two sections of the council eg administration The response

rate was approximately 30 However it should be noted that not everyone working for the council has direct

access to a computer In Belgium three municipalities were surveyed on a census base by means of a web

survey as all employees had access to a computer This rendered a response of 120 (226) 79 (75) and 16

(53) responses for the respective municipalities resulting in an overall response rate of 331 with 662

invitations being sent out In China two departments social work and administration were surveyed in one

municipality using paper-pencil surveys In the social work department 65 out of 110 surveys were returned

and 165 out of 211 responded from the administration department giving an overall response rate of 767 In

Denmark a total of 1282 respondents in a municipality were sent an initial e-mail invitation to a web survey

and up to two e-mail reminders and the response rate was 473 In France all the employees in two

municipalities were surveyed In one municipality where both web and paper-pencil surveys were utilized 128

out of 685 questionnaires were returned while in another city using only a paper-pencil survey 138 out of 652

were received The average response rate was 199 In Italy in principle the link to the questionnaire was

made available to all employees (about 4000) in one city but for some of them access may have been difficult

yielding a response rate of 45 In Korea 500 employees in a city received a printed questionnaire and 253

responses were returned yielding a response rate of 506 In Lithuania 236 responses from 54 municipalities

were gathered by web survey In the Netherlands three municipalities were surveyed on a census base by means

of a web survey In total 319 responses of 910 invitations were returned (35) mdash 90 (45) 115 (38) and 114

(28) for the respective municipalities In Switzerland 516 out of 1326 local employees with e-mail access in

a municipality completed the survey giving a response rate of 389 In the United Kingdom the survey was

posted online and 1360 employees both in administrative jobs and front-line services of one municipality

- 12 -

Municipalities were selected because of their convenient accessibility and proximity to the

researcher and a convenience sample was used in most countries Seven countries surveyed

one municipality but seven local councils were surveyed in Australia Web surveys were used

in four countries while three countries used both web and paper-pencil survey The number

of respondents varied among the countries depending on the numbers of questionnaires

distributed the survey method and the response rate In most countries we used

approximately 250 responses but only 162 from Italy3 The majority of the respondents were

women (566) while in terms of service length those in the largest group (393) had

worked for fewer than 10 years Table 2 shows the distribution of respondentsrsquo sex length of

service employment status and country

[Table 2 about here]

Analyses

The statistical analysis applied confirmatory factor analysis (CFA) CFA was used to assess

the fit of the data to the hypothesized measurement model because it is preferred where

measurement models have a well-developed underlying theory for hypothesized patterns of

loadings The data were analyzed using Amos 180 (Arbuckle 2009) with the maximum

likelihood (MLE) estimation method and LISREL 872 (Joumlreskog and Soumlrbom 2005) with the

diagonally weighted least squares (D-WLS) estimation method

The individual items are designed to measure a theoretically continuous construct

but the observed responses are a discrete realization of five categories The common method

received an email to complete the survey The response rate was 19 In the USA the total number of

employees surveyed was 626 in one city using both electronic and paper-pencil surveys resulting in a response

rate of 588 3 In countries where significantly more than 250 surveys were collected we randomly selected only 250 cases

for analysis

- 13 -

of estimation within CFA is MLE a technique which assumes that the observed variables are

continuous and normally distributed These assumptions are not met when the observed data

are discrete as occurs when using ordinal scales as in this study A simulation study (Flora

and Curran 2004) found that D-WLS performs well for ordinal measures across a variety of

conditions especially for smaller samples Coursey and Pandey (2007a 2007b) argued the

importance of using D-WLS and not standard MLE in structural equations for ordinal data4

In most cases however the differences between MLE and D-WLS are not substantive to the

conclusions5 No comparison between standard MLE and D-WLS is provided in this paper

We give results from both MLE and D-WLS and this will aid in confirming the stability and

robustness of the results across varying mathematical specifications

For gauging reliability Cronbachrsquos alpha and composite reliability were tested

Composite reliability should be equal to or greater than 07 (Hair et al 2010) Convergent

validity was assessed from the measurement model by determining the statistical significance

of each indicatorrsquos estimated pattern coefficient on its specified underlying construct

Discriminant validity was assessed by determining whether the confidence interval around

the correlation estimate between two factors included 100 (Anderson and Gerbing 1988)

For model fit assessment goodness-of-fit index (GFI) comparative fit index (CFI)

and the root mean square error of approximation (RMSEA) were used The model achieves

an acceptable fit to the data when GFI and CFI equals or exceeds 090 and RMSEA values

4 Applying standard MLE to ordinal data such as Likert-type items may produce significant estimation

problems such as inflation of chi-square fit statistics and biased underestimation of parameters and standard

errors (Flora and Curran 2004) Coursey and Pandey (2007a 2007b) used a modified MLE using the asymptotic

covariance matrix and D-WLS for adjusting the ordinal nature of the data They reported that although the

modified MLE findings are slightly different on a few model parameters the differences are not substantive to

the conclusions However model fit estimators were clearly better for the D-WLS 5 Given normally distributed categorical variables MLE can be used with confidence when a variable has four

or more categories If there is a large sample a five-point Likert scale and the skewness and kurtosis of variables

is within acceptable limits it suggests distribution symmetry ―In most cases where the hypothesized model is

well specified and the scaling based on more than three categories it seems unlikely that there will be much

difference between the findings (Byrne 2010 160)

- 14 -

fall below 008 (Hair et al 2010) In general the larger the value of GFI and CFI and the

smaller the value of RMSEA the better the fit of the model (Bollen 1989) We also present

the χ2 test statistic Lower values of χ

2 indicate a better fit and should be nonsignificant but

the χ2 value increases as sample size increases and thus for large sample sizes this statistic

may lead to rejection of a model with good fit (Hair et al 2010)

Results

Using SPSS 180 descriptive statistics were computed for individual items as shown in table

1 One item (APP3) in the APP dimension ―I like to discuss topics regarding public programs

and policies with others was mistakenly omitted in the survey conducted in the United

States and so it was deleted in the statistical analysis In the assessment of normality no

items showed a skew or kurtosis value greater than the cutoffs of |3| or |8| recommended by

Kline (2005) As data were distributed normally MLE as well as D-WLS estimation was used

Based on inspection of item-total correlations one item (COM4) was dropped from further

analysis because it was weakly correlated with the overall index (011) Missing data is an

issue in CFA as it can cause estimation errors All variables in the survey have missing data

rates of approximately 016 to 143 Mean substitution was used to calculate replacement

values for the missing values in each countryrsquos dataset (Hair et al 2010) The deletion of the

two items left 31 items for the next stage of analysis

The five-correlated factor model was tested using CFA The CFA model in the initial

analysis hypothesized a priori that (1) responses to the 30-item PSM measurement

instrument could be explained by five factors such as APP CPV1 CPV2 COM SS (2) each

item would have a nonzero loading on the PSM factor that it was designed to measure and

- 15 -

zero loadings on all other factors (3) the five factors would be correlated and (4)

measurement error terms would be uncorrelated The resulting CFA with MLE suggested that

the initial five-correlated factor model was not a good fit to the data χsup2 (df = 424) = 43699 p

lt 001 GFI = 0899 CFI = 835 RMSEA = 0057 [055 059] At this stage APP was highly

correlated with CPV1 (r = 996 in MLE and r = 988 in D-WLS) suggesting the two

dimensions were not unique and lacked discriminant validity

[Table 3 about here]

Given the disconfirmation of the initial model an effort was made to estimate an

alternative model The two dimensions APP and CPV1 were combined to form a single

dimension Attraction to Public Service (APS) The items of CPV1 may be regarded as

action-oriented by the respondents because dedication to public service community and

common good is underlined in these items Consequently it is reasonable to combine APP

with CPV1 to form APS a dimension that focuses more on disposition to serve the public to

work for the common good and to participate in public policy processes After the overall

number of dimensions was reduced to four using standardized factor loadings and

modification indices as criteria the items with the lowest factor loading in each dimension

were deleted and the items whose modification index indicated association with multiple

dimensions were eliminated A CFA model with the remaining items was iteratively tested

until it achieved a good fit to the data The outcome of this refinement process was a 16-item

index of four factors Overall model fit by every measure is quite strong for the D-WLS

estimation technique The estimation results for the four-dimension model are presented in

table 4

- 16 -

[Table 4 about here]

Cronbachrsquos coefficient alpha for the 16-item PSM index was 860 and the coefficient

alphas for the four dimensions ranged from 645 to 784 The composite reliability of the set

of reflective indicators for each dimension of PSM ranged from 713 to 826 and indicated

adequate internal consistency The resulting factor structure showed a four-factor structure

with all items loading significantly onto their a priori dimension (p lt 001) and the

standardized factor loadings ranging from 0490 to 0766 in MLE and from 515 to 807 in

D-WLS The results provide support for convergent validity The correlation estimates

between the two factors ranged from 0462 to 0848 in MLE and from 505 to 859 in D-

WLS The confidence intervals (plusmn2 standard errors) around the correlation estimates between

the two factors did not include 100 The result provides support for discriminant validity To

test whether the four-dimension model is superior to other designs such as the two or three-

dimension models a series of CFA were conducted and the results show that the four-

dimension model is a significantly better fit than any of the other designs

To test whether the factorial structure of the PSM instrument is equivalent across the

samples from the twelve countries four nested models were tested as part of multi-group

analysis with MLE (Byrne 2010) First the unconstrained model was the baseline model

which relaxed all equality constraints This model hypothesized the same four-factor structure

of the instrument across twelve countries The fit indices for this baseline model were χsup2 (df =

1176) = 24977 p lt 001 GFI = 0901 CFI = 900 RMSEA = 0020 [019 021] These

indices indicate that the four-factor model of the PSM instrument exhibits acceptable fit

across the samples from the twelve countries Second the measurement weights model tested

- 17 -

the invariance of factor loadings across countries by placing equality constraints on these

parameters The fit indices for this model were χsup2 (df = 1308) = 28088 p lt 001 GFI = 0890

CFI = 887 RMSEA = 0020 [019 021] Since the measurement weights model was nested

within the unconstrained model the chi-square difference text ∆χsup2 (132) = 3111 p lt 001

indicated that some equality constraints of factor loadings did not hold across the twelve

countries Further test results suggested that the equality constraints of structural covariances

and measurement residuals were not upheld across the twelve countries It needs to be

stressed that despite the differences in factor loadings and factor covariances the results

support the main assumption that the same four-factor structure holds across different

countries We may assume that people in different cultures understand PSM in the same

fashion

To find differences in the factor structure of PSM among countries we conducted a

series of CFA with each countryrsquos dataset and followed the same procedure as when

analyzing the sample as a whole As table 5 shows although there are great similarities

between the appropriate items in the countries there are also some differences Even though

the items vary according to national differences the four-factor structure had a good fit to the

data Thus the results confirm that the four-factor structure of PSM can be generalized

[Table 5 about here]

The APS dimension covers both an individualrsquos attraction to public participation and

commitment to public interest but all its items refer to some notion of public interest such as

improving public service aiding community tackling social problems meaningful public

service and contributing to the common good The APS dimension captures an individualrsquos

- 18 -

disposition to serve the public interest through public participation When developing the

items some assumed that the items of committing to public interests can be separated from

those of pursuing public values in the CPV dimension The empirical test showed that

respondents tended to understand committing to public interests as behavior-oriented and so

the items for public interest are not differentiated from participation-oriented items in their

perceptions but are different from those for public values The respondents seem to

understand that pursuing public values is more related to personal beliefs or creeds Thus the

CPV dimension is solely focused on public values and the items of pursing public interests

are combined with those of the APP dimension to make the APS dimension

Public values and their relative importance are neither impervious to change over

time nor the same among all countries (Charles de Jong and Ryan 2011 Davis and West

2009) Moreover some values may be in conflict as Martinsen and Beck Joslashrgensen (2010)

discussed in regard to the tensions between accountability and efficiency It is therefore not

possible to include all public values in one dimension (Beck Joslashrgensen and Bozeman 2007)

Accordingly the items of the CPV dimension are developed for estimating individual

commitment to seek public values which there are general consensus about and which are

generally formulated

When all four dimensions of PSM are included in a survey cross-national

equivalence of the survey can be assured The items represent reflections or manifestations of

each dimension and hence are relatively homogeneous and interchangeable Adding or

deleting items does not change the essential nature of the dimension Thus any combination

of items in a dimension can be regarded as equivalent When using structural equation

modeling (SEM) each dimension needs to have at least three reflective indicators in order to

achieve identification or it needs to have at least two reflective indicators and be allowed to

- 19 -

covary with another dimension that has at least two reflective measures With one indicator

the dimension is identified if the loading and the variance of the uniqueness are both fixed

(Bollen 1989 Edwards 2010) Thus the number of items in each dimension is not restricted

when measuring PSM However we recommend using multiple items for each dimension

because more research is needed to investigate the appropriateness of this PSM instrument

with varying samples

The test results showed that model fit estimators were clearly better for the D-WLS

It means that in this study MLE was more rigorous than D-WLS in assessing the fit of the

model The empirical evidence indicated that MLE may be an acceptable technique for

ordinal data when there is a large sample a five-point Likert scale and the skewness and

kurtosis of variables is within acceptable limits as suggested by Byrne (2010)

Discussion

The analysis reaffirmed the four-dimensional structure of PSM but the results were not

entirely consistent with the original CFA we tested We were ultimately able to converge on a

good-fit operational model for PSM that was also consistent with underlying theory The

analysis is not able to resolve the debate about measurement models however for reasons

discussed below

Choosing the correct measurement model (ie reflective vs formative) is important

as measurement model misspecification results not only in misrepresentation of the epistemic

relationships between constructs and their measures but also in biased parameter estimates of

structural relationships (Jarvis et al 2003 Law and Wong 1999 MacKenzie 2003

MacKenzie Podsakoff and Jarvis 2005) A formative measurement model in isolation is

- 20 -

under-identified and cannot be estimated in the case of covariance-based SEM (CB-SEM)

(Bollen 1989) A construct with formative indicators must emit paths to at least two unrelated

reflective indicators two unrelated latent constructs or some combination of the two in order

to solve the problem of under-identification (Jarvis MacKenzie and Podsakoff 2003

MacCallum and Browne 1993 MacKenzie Podsakoff and Jarvis 2005) Thus ―when

dependent constructs change the empirical nature of the formatively measured construct

changes (Wilcox Howell and Breivik 2008 1223) and ―when specifying formative

measurement models what might seem like an innocuous choice of outcomes to achieve

identification can have fundamental ramifications for the meaning of the formative construct

and the inferences based on it (Edwards 2010 7)

Earlier research found that the association between a construct and its measures

should remain stable regardless of the larger causal model in which the constructs and

measures are embedded (Edwards and Bagozzi 2000) The association between a construct

and its measures is generally stable for reflective measures but associations of formative

measures with their construct are unstable The susceptibility of formative measurement to

both interpretational confounding and external inconsistency recommends that formative

measurement is not an equally attractive alternative to reflective measurement and that

whenever possible researchers should opt for reflective measurement (Howell Breivik and

Wilcox 2007a 2007b Kim Shin and Grover 2010) In the case of second-order models

researchers are limited to only a second-order reflective model in CB-SEM (Chin 2010)

Regardless of the measurement debate and the tradeoffs between reflective or

formative specification this study makes a theoretical argument that PSM is a second-order

formative construct This second-order formative factor model (reflective first-order

formative second-order) is also suggested as an alternative model that captures the

- 21 -

multidimensionality of complex concepts and avoids the shortcomings of first-order

formative measurement (Edwards 2010) However this model itself cannot achieve

identification One possible strategy is to develop direct reflective measures of PSM based on

the view that PSM is a general concept that is not only influenced by its specific dimensions

but also can be assessed using general measures (Edwards 2010 Javis MacKenzie and

Podsakoff 2003)6 However these general measures have not yet been developed for the

PSM construct

Another way around the problem of underidentification of formative models is to

apply the component-based approach known as partial least squares path modeling (PLS-PM)

which can model formative indicators (Fornell and Bookstein 1982 Goumltz Liehr-Gobbers

and Krafft 2010) The CB-SEM and the PLS-PM methods have distinctive statistical

characteristics and selecting an approach to SEM depends on the particular research situation

CB-SEM is the method of choice for theory testing while PLS-PM is appropriate for

prediction-oriented applications (Ringle Goumltz Wetzels and Wilson 2009) One advantage of

the PLS method is that it demands significantly fewer requirements compared to that of

covariance structure analyses but nevertheless delivers consistent estimation results Another

asset of the PLS approach is its ability to deal with formative as well as reflective indicators

even within one structural equation model This indicates that the PLS approach can be

appropriate for explorative analysis of structural equation models (Chin 2010 Goumltz Liehr-

Gobbers and Krafft 2010) A Monte Carlo simulation study to compare CB-SEM and PLS-

PM showed that PLS-PM provides a viable approximation of model parameters when the

prerequisites for CB-SEM are not met (Ringle Goumltz Wetzels and Wilson 2009) A formative

6 For example job satisfaction can be assessed with general reflective measures such as ―in general I am

satisfied with my job and the formative dimensions can be assessed with measures that describe satisfaction

with each job facet (Edwards 2010 15)

- 22 -

composite in the form of indexes needs to be tested to know the relative weights (γ-

parameters) of the dimensions of PSM In PLS-PM a second-order formative factor is

directly measured by observed variables for all the first order factors While this approach

repeats the number of manifest variables used the model can be identified without adding

outcome constructs that would be considered inappropriate or irrelevant from a conceptual

standpoint (Chin 2010 Edwards 2010) The test result of PLS-PM using Smart PLS 20

(Ringle Wende and Will 2005) produced the function PSM = 31APS + 36CPV + 28COM

+ 31SS However one consideration is whether and how construct error is modeled for the

formative construct (Diamantopoulos 2006) When using PLS-PM the researcher is making

the assumption that there is no error at the construct level as all constructs in PLS-PM are

modeled without error When using PLS-PM the absence of an error term will tend to

increase the weights therefore inflating their importance (Bollen and Lennox 1991) The

primary statistic for assessing a formative indicator is its weight the partialized effect of the

indicator on its intended construct controlling for the effects of all other indicators of that

construct (Cenfetelli and Bassellier 2009) A formative measure is essentially a multiple

regression with the construct representing the dependent variable and the indicators as the

predictors (Diamantopoulos and Winklhofer 2001) PSM may be regarded as a weighted

linear composite of its dimensions The weights used to create the composites are different

across CE-SEM and PLS-PM The causal dimensional weights generated in CB-SEM are

estimated in a model in which a construct-level error term is specified (eg PSM = γ1APM +

γ2CPI + γ3COM + γ4SS + ζ) whereas the weights in PLS-PM are not (eg PSM = γ1APM +

γ2CPI + γ3COM + γ4SS) This leads to two different conceptualizations of the PSM construct

one in which the causal indicators do not capture the construct in its entirety and another in

which they do (Hardin et al 2010) These differences need to be discussed in future research

- 23 -

in this area

One practical suggestion to measure PSM is to apply unit weights to its dimensions if

we assume the effects of the four dimensions are equivalent A literature review indicates that

unit weights have substantial predictive validity when compared with regression weights The

literature shows that the empirical correlations between scores generated from differential and

unit-weighting approaches are quite high In creating composite scores unit weights can be

an appropriate approach for weighting under many circumstances (Bobko Roth and Buster

2007) Although the uses of unit-weight composites will likely result in the explanation of

less variance than empirical estimations of dimensional weights they will remain consistently

defined across studies and hence suitable for theory testing Alternatively more thorough

discussions on the dimensions of PSM may suggest fixed-weight composites (Hardin et al

2010) A previous study (Kim 2010 22) also suggested ―Overall indexing PSM by

measuring all the dimensions of PSM and simply summing the averages of each dimension

can make it easier to measure PSM in practice as PSM = APS + CPV + COM + SS

An advantage of regarding PSM as a second-order formative construct is that the

dimensions can be analyzed both individually and collectively This is important as the

dimensions are expected to have different antecedents and consequences For instance Perry

(1997) found some significant differences in the influence of independent variables on the

four dimensions of PSM Taylor (2007) confirmed that certain dimensions are more important

than others in influencing work outcomes Moynihan and Pandey (2007a) reported that red

tape is negatively related to one dimension but reform orientation and hierarchical authority

is positively related to another dimension A recent study found that the SS dimension has a

positive relationship with occupational intention but the other dimensions do not (Liu et al

2011) Thus it is valuable to analyze the causes and effects of individual dimensions of PSM

- 24 -

as well as of the PSM construct

Conclusion

The purpose of this study is to develop a measurement instrument for PSM for cross-national

research and comparison On the basis of the sharpened construct of PSM (Kim and

Vandenabeele 2010) using data from an international survey conducted in 12 countries we

found that the four-factor structure of PSM is generalizable across country cases This study

provides the measurement index of PSM with the four dimensions of APS CPV COM and

SS The APS dimension measures the instrumental motives of PSM the CPV dimension

determines the value-based motives of PSM while the COM and SS dimensions gauge the

identification motives and self-sacrifice respectively When all four dimensions of PSM are

used in a survey cross-national equivalence of survey items can be assured The items for

measuring PSM in future research are suggested in Appendix 4

This study has several limitations First looking for samples which were comparable

to the largest possible extent we opted for surveys in local government The results found

and discussed in this study are based on PSM among local government employees The

results might be different when surveying central government employees or other specific

groups in the public sector Thus the potential impact of the sample bias needs to be

considered in understanding this study It also means that the survey items for measuring

PSM should be examined with different samples in various areas and at different government

levels

Second the test results showed that there is a high degree of correlation among the

dimension such as APS CPV and COM Future research should not only investigate whether

- 25 -

the items reflect the core content of the dimensions but also continue to develop more

appropriate items for better discriminant validity of each dimension Third the content of

APP dimension was assumed to be distinct from the CPV dimension that focuses on personal

disposition to pursue public interest and public values However the public interest sub-

dimension was separated from the public values sub-dimension in the CPV dimension and

combined with the APP dimension to form the APS dimension Further research should

analyze the CPV dimension and investigate the appropriateness of the APS dimension

In order to facilitate research internationally the PSM construct needs to be

improved conceptually and operationally Kim and Vandenabeele (2010) have dealt with the

conceptual composition and operational dimension of PSM as well as the measurement

model of PSM The next step is to develop an internationally robust measurement instrument

of PSM This study proposes that scholars use all four of the dimensions in measuring the

conceptual components of PSM such as self-sacrifice instrumental value-based and affective

motives because they collectively represent PSM It provides a set of appropriate items for

measuring PSM and confirms the dimensional structure of PSM The survey results suggest

that people in different cultures understand PSM in the same fashion We know that this study

is simply a step along the way to developing a more robust measure but we hope it

contributes to enhanced cross-national research and comparison along with the generation of

cumulative knowledge internationally

- 26 -

REFERENCES

Anderson James C and David D Gerbing 1988 Structural Equation Modeling in Practice

A Review and Recommended Two-Step Approach Psychological Bulletin 103

411-423

Arbuckle James L 2009 Amos 180 Userrsquos Guide Crawfordville FL Amos Development

Co

Beck Joslashrgensen Torben and Barry Bozeman 2007 Public Values An Inventory

Administration amp Society 39(3) 354-381

Beutel Ann M and Margaret Mooney Marini 1995 Gender and values American

Sociological Review 60 436 - 448

Bobko Philip Philip L Roth and Maury A Buster 2007 The Usefulness of Unit Weights in

Creating Composite Scores A Literature Review Application to Content Validity and

Meta-Analysis Organizational Research Methods 10(4) 689-709

Bollen Kenneth A 1989 Structural Equations with Latent Variables New York Wiley

Bollen K A and R Lennox 1991 Conventional Wisdom on Measurement A Structural

Equation Perspective Psychological Bulletin 110(2) 305-314

Byrne Barbara M 2010 Structural Equation Modeling with Amos Basic Concepts

Applications and Programming 2nd

ed New York Routledge

Camilleri Emanuel 2006 Towards Developing an Organizational Commitment ― Public

Service Motivation Model for Maltese Public Service Employees Public Policy

and Administration 21 63-83

Castaing Seacutebastien 2006 The Effects of Psychological Contract Fulfillment and Public

Service Motivation on Organizational Commitment in the French Civil Service

Public Policy and Administration 21 84-98

Cenfetelli Ronald T and Geneviegraveve Bassellier 2009 Interpretation of Formative

Measurement in Information Systems Research MIS Quarterly 33(4) 689-707

Charles Michael B W Martin de Jong and Neal Ryan 2011 Public Values in Western

- 27 -

Europe A Temporal Perspective American Review of Public Administration 41

75-91

Chin Wynne W 2010 How to Write Up and Report PLS Analysis In Handbook of Partical

Least Squares edited by V Esposito Vinzi et al 655-690 Berlin Springer-Verlag

Coursey David H and Sanjay K Pandey 2007a Public Service Motivation Measurement

Testing an Abridged Version of Perryrsquos Proposed Scale Administration amp Society

39 547-568

Coursey David H and Sanjay K Pandey 2007b Content Domain Measurement and

Validity of the Red Tape Concept American Review of Public Administration 37

342-361

Coursey David H James L Perry Jeffrey L Brudney and Laura Littlepage 2008

Psychometric Verification of Perryrsquos Public Service Motivation Instrument Results

for Volunteer Exemplars Review of Public Personnel Administration 28 79-90

Davis Paul and Karen West 2009 What Do Public Values Mean for Public Action Putting

Public Values in Their Plural Place American Review of Public Administration 39

602-618

DeHart-Davis Leisha Justin Marlowe and Sanjay K Pandey 2006 Gender Dimensions of

Public Service Motivation Public Administration Review 66 873-887

Diamantopoulos Adamantios 2006 The error term in formative measurement models

Interpretation and modeling implications Journal of Modeling in Management 1(1)

7-17

Diamantopoulos Adamantios and Judy A Siguaw 2006 Formative versus Reflective

Indicators in Organizational Measure Development A Comparison and Empirical

Illustration British Journal of Management 17 263-282

Diamantopoulos Adamantios and Heidi M Winklhofer 2001 Index Construction with

Formative Indicators An Alternative to Scale Development Journal of Marketing

Research 38 269-277

Edwards Jeffrey R 2010 The Fallacy of Formative Measurement Organizational Research

- 28 -

Methods Online First published on August 19 2010 as

doi1011771094428110378369

Edwards Jeffrey R and Richard P Bagozzi 2000 On the Nature and Direction of

Relationships between Constructs and Measures Psychological Methods 5 155-

174

Flora David B and Patrick J Curran 2004 An Empirical Evaluation of Alternative

Methods of Estimation for Confirmatory Factor Analysis with Ordinal Data

Psychological Methods 9 466-491

Fornell Claes and Fred L Bookstein 1982 Two structural equations models LISREL and

PLS applied to customer exit-voice theory Journal of Marketing Research 19 440-

452

Giauque David Adrian Ritz Freacutedeacuteric Varone Simon Anderfuhren-Biget and Christian

Waldner 2011 Putting Public Service Motivation into Context A Balance between

Universalism and Particularism Forthcoming in International Review of

Administrative Sciences 77(2)

Goumltz Oliver Kerstin Liehr-Gobbers and Manfred Krafft 2010 Evaluation of Structural

Equation Models Using the Partial Least Squares (PLS) Approach In V Esposito

Vinzi et al (eds) Handbook of Partial Least Squares 691-711 DOI 101007978-3-

540-32827-8_30 Springer-Verlag Berlin Heidelberg

Guinot Christiane Julie Latreille and Michel Tenenhaus 2001 PLS Path Modelling and

Multiple Table Analysis Application to the Cosmetic Habits of Women in Ile-de-

France Chemometrics and Intelligent Laboratory Systems 58 247-259

Hair Joseph F Jr William C Black Barry J Babin and Rolph E Anderson 2010

Multivariate Data Analysis 7th

ed Upper Saddle River NJ Pearson Education

Hardin Andrew M Jerry Cha-Jan Chang Mark A Fuller and Gholamreza Torkzadeh 2010

Formative Measurement and Academic Research In Search of Measurement

Theory Educational and Psychological Measurement Online First published on

June 11 2010 as doi101177001316440370208

- 29 -

Howell Roy D Einar Breivik and James B Wilcox 2007a Reconsidering Formative

Measurement Psychological Methods 12 205-218

Howell Roy D Einar Breivik and James B Wilcox 2007b Is Formative Measurement

Really Measurement Reply to Bollen (2007) and Bagozzi (2007) Psychological

Methods 12 238-245

Jarvis Cheryl Burke Scott B MacKenzie and Philip M Podsakoff 2003 A Critical Review

of Construct Indicators and Measurement Model Misspecification in Marketing and

Consumer Research Journal of Consumer Research 30 199-218

Joumlreskog Karl G and Dag Soumlrbom 2005 LISREL 872 (Computer Software) Lincolnwood

IL Scientific Software International

Kim Gimun Bongsik Shin and Varun Grover 2010 Investigating Two Contradictory Views

of Formative Measurement in Information Systems Research MIS Quarterly 34

345-365

Kim Sangmook 2009a Revising Perryrsquos Measurement Scale of Public Service Motivation

American Review of Public Administration 39 149-163

Kim Sangmook 2009b Testing the Structure of Public Service Motivation in Korea A

Research Note Journal of Public Administration Research and Theory 19 839-

851

Kim Sangmook 2010 Testing a Revised Measurement of Public Service Motivation

Reflective versus Formative Specification Journal of Public Administration

Research and Theory Doi101093jopartmuq048

Kim Sangmook and Wouter Vandenabeele 2010 A strategy for building public service

motivation research internationally Public Administration Review 70 701-709

Kline Rex B 2005 Principles and Practice of Structural Equation Modeling 2nd

ed New

York Guilford Press

Law Kenneth and Chi-Sum Wong 1999 Multidimensional Constructs in Structural

Equation Analysis An Illustration using the Job Perception and Job Satisfaction

Constructs Journal of Management 25 143-160

- 30 -

Leisink Peter and Bram Steijn 2009 Public Service Motivation and Job Performance of

Public Sector Employees in the Netherlands International Review of

Administrative Sciences 75 35-52

Liu Bangcheng Chun Hui Jin Hu Wensheng Yang and Xinli Yu 2011 How well can

Public Service Motivation Connect with Occupational Intention International

Review of Administrative Sciences 77 191-211

Liu Bangcheng Ningyu Tang and Xiaomei Zhu 2008 Public Service Motivation and Job

Satisfaction in China An Investigation of Generalisability and Instrumentality

International Journal of Manpower 29 684-699

MacCallum Robert C and Michael W Browne 1993 The Use of Causal Indicators in

Covariance Structure Models Some Practical Issues Psychological Bulletin 114

533-541

MacKenzie Scott B 2003 The dangers of poor construct conceptualization Journal of the

Academy of Marketing Science 31(3) 323-326

MacKenzie Scott B Philip M Podsakoff and Cheryl Burke Jarvis 2005 The Problem of

Measurement Model Misspecification in Behavioral and Organizational Research

and Some Recommended Solutions Journal of Applied Psychology 90 710-730

Martinsen Dorte Sindbjerg and Torben Beck Joslashrgensen 2010 Accountability as a

Differentiated Value in Supranational Governance American Review of Public

Administration 40 742-760

Moynihan Donald P and Sanjay K Pandey 2007a The Role of Organizations in Fostering

Public Service Motivation Public Administration Review 67 40-53

Moynihan Donald P and Sanjay K Pandey 2007b Finding Workable Levers over Work

Motivation Comparing Job Satisfaction Job Involvement and Organizational

Commitment Administration amp Society 39 803-832

Pandey Sanjay K and Edmund C Stazyk 2008 Antecedents and Correlates of Public

Service Motivation In Motivation in Public Management edited by James L Perry

and Annie Hondeghem 101-117 Oxford Oxford University Press

- 31 -

Perry James L 1996 Measuring Public Service Motivation An Assessment of Construct

Reliability and Validity Journal of Public Administration Research and Theory 6

5-22

Perry James L 1997 Antecedents of Public Service Motivation Journal of Public

Administration Research and Theory 7 181-197

Perry James L 2000 Bringing Society in Toward a Theory of Public Service Motivation

Journal of Public Administration Research and Theory 10 471-488

Perry James L and Annie Hondeghem 2008a Motivation in Public Management Oxford

Oxford University Press

Perry James L and Annie Hondeghem 2008b Directions for Future Theory and Research

In Motivation in Public Management edited by James L Perry and Annie

Hondeghem 294-313 Oxford Oxford University Press

Perry James L Annie Hondeghem and Lois Recascino Wise 2010 Revisiting the

Motivational Bases of Public Service Twenty Years of Research and an Agenda for

the Future Public Administration Review 70(5) 681-690

Perry James L and Wouter Vandenabeele 2008 Behavioral dynamics Institutions

Identities and Self-Regulation In Motivation in Public Management edited by

James L Perry and Annie Hondeghem 56-79 Oxford Oxford University Press

Perry James L and Lois Recascino Wise 1990 The Motivational Bases of Public Service

Public Administration Review 50 367-373

Rainey Hal G 1982 Reward Preference among Public and Private Managers In Search of

the Service Ethic American Review of Public Administration 16 288-302

Ringle Christian M Oliver Goumltz Martin Wetzels and Bradley Wilson 2009 On the Use of

Formative Measurement Specifications in Structural Equation Modeling A Monte

Carlo Simulation Study to Compare Covariance-Based and Partial Least Squares

Model Estimation Methodologies Faculty of Economics and Business Administration

Maastricht Univeristy

Ringle C M S Wende and S Will 2005 Smart PLS 20 (M3) Beta

- 32 -

httpwwwsmartplsde

Ritz Adrian 2011 Attraction to Public Policy Making A Qualitative Inquiry into

Improvements in PSM Measurement Forthcoming in Public Administration

Spector Paul E 1992 Summated Rating Scales Construction An Introduction Newbury

Park CA Sage

Steenbergen Marco R 1996 Compassion and American public opinion An analysis of the

NES Humanitarianism Scale Report to the Board of Overseers of the National

Election Studies

Steffen Patrick R and Kevin S Masters 2005 Does compassion mediate the intrinsic

religion-health relationship Annals of Behavioral Medicine 30 217-223

Taylor Jeannette 2007 The Impact of Public Service Motives on Work Outcomes in

Australia A Comparative Multi-Dimensional Analysis Public Administration 85

931-959

Vandenabeele Wouter 2007 Towards a Theory of Public Service Motivation An

Institutional Approach Public Management Review 9 545-556

Vandenabeele Wouter 2008a Government Calling Public Service Motivation as an Element

in Selecting Government as an Employer of Choice Public Administration 86

1089-1105

Vandenabeele Wouter 2008b Development of a Public Service Motivation Measurement

Scale Corroborating and Extending Perryrsquos Measurement Instrument International

Public Management Journal 11 143-167

Vandenabeele Wouter 2009 The Mediating Effect of Job Satisfaction and Organizational

Commitment on Self-Reported Performance More Robust Evidence of the PSM-

Performance Relationship International Review of Administrative Sciences 75 11-

34

Wilcox James B Roy D Howell and Einar Breivik 2008 Questions about Formative

Measurement Journal of Business Research 61 1219-1228

Wise Lois Recascino 2000 The Public Service Culture In Public Administration Concepts

- 33 -

and Cases 7th ed edited by Richard J Stillman II 342-353 Boston Houghton

Mifflin

Wright Bradley E 2008 Methodological Challenges associated with Public Service

Motivation Research In Motivation in Public Management edited by James L

Perry and Annie Hondeghem 80-98 Oxford Oxford University Press

Wright Bradley E and Sanjay K Pandey 2008 Public Service Motivation and the

Assumption of Person-Organization Fit Testing the Mediating Effect of Value

Congruence Administration amp Society 40 502-521

- 34 -

Table 1

Possible Indicators of PSM and Descriptive Statistics (n = 2868)

Dimensions and items Mean Standard

Deviation

Item-Total

Correlation

Attraction to Public Participation (APP)

APP1 I am interested in helping to improve public service

APP2 I am satisfied when I see people benefiting from the public programs I was

involved in

APP3 I like to discuss topics regarding public programs and policies with others

APP4 I believe that public sector activities contribute to our general welfare

APP5 I admire people who initiate or are involved in activities to aid my

community

APP6 Contributing to public programs and policies helps me realize myself

APP7 It is important to contribute to activities that tackle social problems

412

441

355

393

423

351

409

696

662

917

753

672

881

734

524

421

482

412

493

478

552

Commitment to Public Values (CPV)

Sub-dimension for Public Interests (CPV1)

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

CPI3 I would prefer seeing public officials do what is best for the whole

community even if it harmed my interests

CPI4 Serving the public interest is more important than helping a single individual

Sub-dimension for Public Values (CPV2)

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of public

services

CPV3 It is fundamental that public services respond to the needs of the citizens

CPV4 Decisions regarding public services should be democratic despite the time

and effort it takes

CPV5 Everybody is entitled to a good service even if it costs a lot of money

CPV6 It is fundamental that the interests of future generations are taken into

account when developing public policies

CPV7 To act ethically is essential for public servants

CPV8 I believe that public employees must always be aware of the legitimacy of

their activities

CPV9 I personally identify with the aim of protecting individual liberties and rights

418

409

348

337

428

416

433

394

367

426

441

434

414

695

662

899

995

730

704

710

852

932

689

681

692

763

525

596

480

251

484

437

356

325

376

439

462

402

392

Compassion (COM) COM1 It is difficult for me to contain my feelings when I see people in distress

COM2 I feel sympathetic to the plight of the underprivileged

COM3 I empathize with other people who face difficulties

COM4 I have little compassion for people in need who are unwilling to take the

first step to help themselves

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

354

387

388

299

412

391

951

825

707

1083

755

765

399

511

389

011

487

502

Self-Sacrifice (SS)

SS1 Making a difference to society means more to me than personal achievements

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS5 People should give back to society more than they get from it

SS6 Serving other citizens would give me a good feeling even if no one paid me for

it

SS7 I would agree to a good plan to make a better life for the poor even if it costs

me money

357

329

339

302

345

363

350

904

872

949

982

888

1049

895

508

584

454

531

412

368

546

- 35 -

Table 2

Background of Respondents (n = 2868)

Variables Characteristics Respondents

Sex

Male

Female

NA

1209

1622

37

422

566

13

Length of service

(years)

0 ~ 10

10 ~ 20

20 ~ 30

30+

NA

1126

723

484

491

44

393

252

169

171

15

Organizational status

―Do you supervise employees

No

Yes

NA

1472

1076

320

513

375

112

Country

Australia

Belgium

China

Denmark

France

Italy

Korea

Lithuania

The Netherlands

Switzerland

United Kingdom

USA

249

214

230

249

266

162

253

236

249

250

260

250

87

75

80

87

93

56

88

82

87

87

91

87 Note NA = no answer

- 36 -

Table 3

Standardized factor loadings and correlations for the 31-item PSM index (n = 2868)

Dimensions and items SFL (MLE) SFL (D-WLS) Alpha

APP

APP1

APP2

APP4

APP5

APP6

APP7

597

486

455

558

490

600

657

552

514

625

549

665

699

CPV1

CPI1

CPI2

CPI3

CPI4

588

660

488

289

653

732

535

294

542

CPV2

CPV1

CPV2

CPV3

CPV4

CPV5

CPV6

CPV7

CPV8

CPV9

611

546

436

393

418

534

598

496

460

690

631

505

457

486

601

669

573

530

745

COM

COM1

COM2

COM3

COM5

COM6

496

569

498

597

607

516

651

538

670

684

687

SS

SS1

SS2

SS3

SS4

SS5

SS6

SS7

567

736

558

728

467

409

663

660

762

591

695

515

494

709

782

Measures of fit (MLE) χsup2 (df = 424) = 43699 p lt 001 CFI = 835 GFI = 899

RMSEA = 057 [055 059]

Measure of fit (D-WLS) χsup2 (df = 424) = 39247 p lt 001 CFI = 971 GFI = 976

RMSEA = 054 [052 055]

Inter-factor correlations MLE (D-WLS)

1 2 3 4

1 APS

2 CPV1 996 (988)

3 CPV2 902 (890) 798 (798)

4 COM 787 (783) 758 (762) 779 (783)

5 SS 700 (742) 800 (859) 509 (537) 720 (729)

Note SFL = standardized factor loading All factor loadings and correlations are significant at p lt 001

- 37 -

Table 4

Standardized factor loadings and correlations for the 16-item PSM index (n = 2868)

Dimensions and items SFL

(MLE)

SFL

(D-WLS)

APS

APP5 I admire people who initiate or are involved in activities to aid my

community

APP7 It is important to contribute to activities that tackle social problems

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

611

592

637

666

665

669

690

751

CPV

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of

public services

CPV6 It is fundamental that the interests of future generations are taken

into account when developing public policies

CPV7 To act ethically is essential for public servants

612

546

551

590

698

631

618

658

COM

COM1 It is difficult for me to contain my feelings when I see people in

distress

COM2 I feel sympathetic to the plight of the underprivileged

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

490

546

596

604

515

608

669

679

SS

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS7 I would agree to a good plan to make a better life for the poor even if

it costs me money

763

571

766

678

807

629

747

755

Measures of fit (MLE)

χsup2 (df = 98) = 8490 p lt 001

CFI = 939 GFI = 965

RMSEA = 052 [049 055]

Measure of fit (D-WLS)

χsup2 (df = 98) = 6069 p lt 001

CFI = 988 GFI = 992

RMSEA = 043 [039 046]

Alpha CR (D-WLS) Inter-factor correlations MLE (D-WLS)

1 APS 717 788 1 APS 2 CPV 3 COM

2 CPV 663 747 848 (859)

3 COM 645 713 781 (791) 789 (795)

4 SS 784 826 565 (596) 462 (505) 715 (726)

Note SFL = standardized factor loading Alpha = Cronbachrsquos alpha CR = composite reliability

All factor loadings and correlations are significant at p lt 001

- 38 -

Table 5

Results of CFA in the individual countries Standardized factor loadings and measures of fit

Factors

amp Items

Australia Belgium China Denmark France Italy

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

672

756

799

785

804

943

710

703

597

790

799

730

584

675

685

681

731

742

578

748

674

643

765

739

645

704

678

708

812

779

632

604

765

766

672

853

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

552

593

560

699

727

632

624

605

560

713

709

645

702

602

616

724

686

735

567

560

715

654

605

811

608

658

587

696

755

698

673

620

523

782

697

654

COM

COM1

COM2

COM3

COM5

COM6

623

584

662

688

701

739

732

608

475

808

666

581

578

616

605

649

670

673

699

751

737

679

816

776

568

546

801

626

673

850

637

546

741

715

601

822

SS

SS2

SS3

SS4

SS7

654

794

806

892

767

813

727

693

799

793

725

838

857

820

773

896

855

842

806

749

792

829

753

884

847

731

753

889

816

753

815

692

664

795

782

734

Inter-factor

correlations

093

~

752

163

~

810

320

~

773

323

~

825

452

~

827

497

~

855

323

~

850

321

~

847

498

~

899

536

~

902

425

~

829

476

~

865

Sample size 249 214 230 249 266 162

Measures of Fit

χsup2 638 444 881 732 765 554 817 625 735 565 750 564

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 980 100 939 987 969 997 967 994 975 997 949 995

GFI 961 986 941 986 948 991 949 992 955 992 926 986

RMSEA 036 00 063 050 051 026 053 035 045 026 059 033

- 39 -

Table 5 (Continued)

Factors

amp Items

Korea Lithuania The Netherlands Switzerland UK USA

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

619

701

667

706

781

717

666

512

682

751

605

779

557

629

630

620

719

729

535

621

702

643

694

735

655

725

738

710

804

822

535

621

702

643

694

735

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

594

607

722

722

644

722

674

535

533

790

589

711

568

554

538

655

628

545

627

580

615

711

666

704

800

512

555

873

581

646

627

580

615

711

666

704

COM

COM1

COM2

COM3

COM5

COM6

694

677

550

763

736

627

557

639

428

550

696

534

737

623

508

766

616

608

659

592

683

703

661

769

682

583

729

739

644

803

659

592

683

703

661

769

SS

SS2

SS3

SS4

SS7

776

752

679

779

770

811

707

721

635

725

734

706

814

656

743

820

767

764

745

632

689

788

662

744

671

818

832

830

864

802

745

632

689

788

662

744

Inter-factor

correlations

432

~

726

466

~

746

376

~

788

406

~

902

377

~

852

455

~

836

637

~

818

660

~

828

317

~

852

360

~

852

637

~

818

660

~

828

Sample size 253 236 249 250 260 250

Measures of Fit

χsup2 896 573 612 462 773 616 815 715 769 542 815 715

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 951 996 975 100 958 993 957 991 972 998 957 991

GFI 942 988 960 989 951 987 948 990 955 990 948 990

RMSEA 059 028 034 00 050 034 053 044 048 022 053 044

Note All factor loadings and correlations are significant at p lt 001

- 40 -

Appendix 1 Dimensions and possible items of PSM (the first round)

1 Attraction to Public Participation

(1) I am interested in developing public programs helping my country or community (Kim 2009a)

(2) Sharing my views on public policies with others is attractive to me (Kim 2009a)

(3) I am satisfied when seeing people get benefits from the public programs I was involved in (Kim 2009a)

(4) I like to discuss political subjects with others (Giauque et al 2011)

(5) I think the governmental activities contribute to our welfare

(6) I like to initiate actions to help out my community

(7) I do not care much about politicians (Perry 1996)

2 Commitment to Public Values

Possible items for public interest

(1A) Meaningful public service is very important to me (Perry 1996)

(2A) It is important for me to contribute for the common good (Giauque et al 2011)

(3A) I recognize myself with the mission of protection of liberties and rights endorsed by the nation-state

(Giauque et al 2011)

(4A) I would prefer seeing public officials do what is best for the whole community even if it harmed my

interests (Perry 1996)

(5A) To me serving the public interest is more important than helping other people (Vandenabeele 2008b)

Possible items for public values

(1B) I recognize myself with the promotion of the equality of chances endorsed by the nation-state (Giauque et

al 2011)

(2B) It is fundamental that the public service provided to the citizen is regular and continuous (Giauque et al

2011)

(3B) It is fundamental that public service innovates to respond to the new needs of the citizensclients (Giauque

et al 2011)

(4B) The dignity and well-being of all should be the most important concerns in any society (Steenbergen

1996)

(5B) It is important that public servants account for all the costsexpenses they make (Vandenabeele 2008b)

(6B) We have to do everything in our power to pursue the goal of democracy

3 Compassion

(1) It is difficult for me to contain my feelings when I see people in distress (Perry 1996)

(2) I feel sympathetic to the plight of the underprivileged (Kim 2009a)

(3) I am often reminded by daily events how dependent we are on one another (Perry 1996)

(4) I feel compassion for others in the difficulties they are facing (Steffen and Masters 2005)

(5) It is not really my problem if others are in trouble and need help (Beutel and Marini 1995)

(6) I find it hard to be sympathetic toward starving people in foreign lands when there is so much trouble in our

own country (Beutel and Marini 1995)

(7) I get very upset when I see other people treated unfairly (Beutel and Marini 1995)

(8) I would agree to a good plan to make a better life for the poor even if it cost me money (Beutel and Marini

1995)

(9) To me patriotism includes seeing the welfare of others (Perry 1996)

(10) I care very much about other people

4 Self-Sacrifice

(1) Making a difference in society means more to me than personal achievements (Perry 1996)

(2) I am prepared to make enormous sacrifices for the good of society (Perry 1996)

(3) I believe in putting duty before self (Perry 1996)

(4) I am one of those rare people who would risk personal loss to help someone else (Perry 1996)

(5) I think people should give back to society more than they get from it (Perry 1996)

(6) Serving other citizens would give me a good feeling even if no one paid me for it (Perry 1996)

(7) Much of what I do is for a cause bigger than myself (Perry 1996)

- 41 -

Appendix 2 Newly suggested items through the first round

1 Items for the Attraction to Public Participation Dimension

(1) In the public domain every person should be counted as equal

(2) I enjoy to think and argue in a political way

(3) I am satisfied when seeing my ideas being integrated in public policies

(4) I can easily relate to the way of thinking and arguing of politicians

(5) Working in the public sector I can better realize myself

2 Items for the Public Values Sub-Dimension

(1) Public service should give value for money

(2) Cost considerations should play an important role in decisions about public service provision

(3) It is fundamental that the interests of future generations are taken into account when developing public

policies

(4) To act with correct ethics is for public servants as important as their professional competence

(5) Primarily public servants should be accountable to the public and not to their superiors

3 Items for the Self-Sacrifice Dimension

(1) I would agree to a good plan to make a better life for the poor even if it costs me money

- 42 -

Appendix 3 An example of a summary of comments and suggestions (the second round)

On the item ―I am interested in developing public programs helping my country or community

A It is a good sentence but is it possible for everyone to ―develop public programs Even public officials are

limited to do this

B OK

C or the special group

D Double barrel and difficult to translate to my country ―I am interested in helping to improve public services

E Good but I feel that wording requires revision If we draw sample from local government then country is

redundant ―I am interested in developing public policy programs that aim in helping my community

F 4 (General appreciation from1 to 4) Not clear whether it refers to ones work or to ones concern as a citizen

―(In my work) I am interested in contributing to the development of public etchellip

G Instead of development I would use implementing

H Delete public which is redundant because of the characteristic helping my country or community but add

or implementing ―I am interested in developing or implementing programs helping my country or

community

I Referring to dimension Wording Similar to item 6

Note Names have been changed to alphabets in order to ensure the anonymity of participants

- 43 -

Appendix 4 Dimensions and Items of Public Service Motivation for Future Research

Dimensions and items

Attraction to Public Service (APS)

1 I am interested in helping to improve public service

2 I admire people who initiate or are involved in activities to aid my community

3 It is important to contribute to activities that tackle social problems

4 Meaningful public service is very important to me

5 It is important for me to contribute to the common good

Commitment to Public Values (CPV)

1 I think equal opportunities for citizens are very important

2 It is important that citizens can rely on the continuous provision of public services

3 It is fundamental that the interests of future generations are taken into account when

developing public policies

4 To act ethically is essential for public servants

5 I believe that public employees must always be aware of the legitimacy of their activities

6 I personally identify with the aim of protecting individual liberties and rights

Compassion (COM)

1 It is difficult for me to contain my feelings when I see people in distress

2 I feel sympathetic to the plight of the underprivileged

3 I empathize with other people who face difficulties

4 I get very upset when I see other people being treated unfairly

5 Considering the welfare of others is very important

Self-Sacrifice (SS)

1 I am prepared to make sacrifices for the good of society

2 I believe in putting civic duty before self

3 I am willing to risk personal loss to help society

4 I would agree to a good plan to make a better life for the poor even if it costs me money

- 8 -

another on public values such as social equity equal opportunity and democracy Thus five

possible items for the public interest sub-dimension (CPV1) and six items for the public

values sub-dimension (CPV2) were developed For the dimension of COM which focuses on

affective bonding with the identified objects such as other members of a social category or of

a political system ten possible items were suggested For the SS dimension Perryrsquos original

items (1996) were the primary source of items The full list of items is provided in Appendix

1

Review Processes

A group of researchers who shared our interest in better measurement of PSM collaborated

using e-mail in November 20091 The objective was to develop a measurement instrument

for PSM with a robust comparative character in other words to create a valid instrument (on

a national basis) that can be used internationally The first round for reviewing the possible

items lasted from the 16th

of November until the 7th

of December 2009 and all the

participants were asked to review these proposed items For this review two criteria were

employed (1) Do these items refer to the dimensions they claim to refer to and (2) Are

these items meaningful across various national contexts and in particular to the situation of

civil servants in that country Participants were also asked to add new items that might better

reflect a particular dimension or to rephrase items that were already on the list

A second round of item refinement lasted from the 9th

of December until the 22nd

of

December 2009 During this time 11 newly suggested items from the first round were

considered (see Appendix 2) A summary of all comments and suggestions emerging from the

1 Emmanuel Camilleri David Giauque Gerhard Hammerschmid Isabel Egger-Peitler Renate Meyer and

Cristian Pliscoff also participated in the review processes

- 9 -

first process was distributed to the participants to stimulate comments in the second round

(see Appendix 3) Based on the comments received in the second round discussion the

possible items for an international survey were selected However some items required

further discussion and so a third round was conducted from the 15th

of January until the 23rd

of January 2010 One issue that remained unresolved in the discussions relates to the

dimension of CPV Some argued that it is reasonable to divide CPV into two sub-dimensions

because committing to public interests is different from pursuing public values and there are

many important public values in democratic countries while others contended that CPV can

cover all important public values including public interest and there is no real benefit to add

one more dimension These differences were left to be resolved through empirical research

This did not affect the final items however and 33 items were confirmed The resulting

survey questionnaire was distributed to all the collaborators on February 9 2010 in order to

conduct a survey in each country

Measures

PSM was measured using the newly developed 33-item index 7 items for the dimension of

APP 13 items for CPV (4 for CPV1 and 9 for CPV2) 6 items for COM and 7 for SS In the

survey questionnaire a five-point Likert type scale (1 = strongly disagree 5 = strongly agree)

was used This scale is among the most widely used response formats in many empirical

investigations within the social and behavioral science in which there are no ―correct or

―incorrect responses (Spector 1992) The items of PSM were randomly distributed in the

survey format Gender age education length of service and employment status were also

measured in this survey

It was necessary to translate the measurement items into the official (native) language

- 10 -

of each country As this raised the possibility of inaccurate translations the back-translation

of the items into native languages was validated by asking two researchers to independently

translate the English-language items The researchers were instructed to consult with each

other over differences between translated items and to test the survey items with some native

respondents

[Table 1 about here]

Samples

The international survey was fielded from March to September 2010 The data upon which

this study are based were collected from civil servants in local governments in Australia

Belgium China Denmark France Italy Korea Lithuania the Netherlands Switzerland the

United Kingdom and the USA The principal criterion was that the samples collected in the

various countries should be comparable to the largest possible extent The criterion ruled out

national and state government as the competencies of national governments and also of state

governments if they exist constitutionally are different among most of the countries included

in the research Therefore we opted for local government as our focus The structure and

functions of local governments can differ widely even within countries and so the survey in

each country was focused on town hall bureaucrats in city county or township government

The respondents were permanent employees from both managerial and non-managerial levels

and from both administrative departments (ie no direct contact with the public) and service

departments (ie likely to have direct contact with the public) Despite being employed by

local governments in many or all of the countries police officers firefighters school teachers

public transport workers artists and musicians as well as nurses and doctors were excluded

- 11 -

from the sample

We aimed to receive 250 respondents for each country as this would enable us to

carry out the analyses we wanted to perform on the data Although there is no ideal sample

size one study recommends a sample of 200 to provide a sound basis for estimation (Hair

Black Babin and Anderson 2010) The method of data collection was left open Obviously a

web-survey is the easiest way to collect the data because it is economical to collect and to

code however compared to paper-pencil surveys this method has some drawbacks First not

everybody in the sample may have access to the internet which may cause a bias in the

responses Second internet surveys are generally known to produce lower response rates

Web surveys were recommended if possible but paper-pencil surveys were deemed

acceptable in cases where it was not possible to conduct a web survey Researchers were

encouraged to contact just one municipality except in cases where it was not possible to

gather enough data In such instances responses from multiple municipalities were accepted

By the end of September 2010 a total of 2868 responses were obtained2

2 In Australia seven local councils each with between from 70 and 232 employees were surveyed Paper-pencil

surveys were used in two councils while the rest relied on a web survey The local governments limited

distribution of paper-pencil surveys to only one or two sections of the council eg administration The response

rate was approximately 30 However it should be noted that not everyone working for the council has direct

access to a computer In Belgium three municipalities were surveyed on a census base by means of a web

survey as all employees had access to a computer This rendered a response of 120 (226) 79 (75) and 16

(53) responses for the respective municipalities resulting in an overall response rate of 331 with 662

invitations being sent out In China two departments social work and administration were surveyed in one

municipality using paper-pencil surveys In the social work department 65 out of 110 surveys were returned

and 165 out of 211 responded from the administration department giving an overall response rate of 767 In

Denmark a total of 1282 respondents in a municipality were sent an initial e-mail invitation to a web survey

and up to two e-mail reminders and the response rate was 473 In France all the employees in two

municipalities were surveyed In one municipality where both web and paper-pencil surveys were utilized 128

out of 685 questionnaires were returned while in another city using only a paper-pencil survey 138 out of 652

were received The average response rate was 199 In Italy in principle the link to the questionnaire was

made available to all employees (about 4000) in one city but for some of them access may have been difficult

yielding a response rate of 45 In Korea 500 employees in a city received a printed questionnaire and 253

responses were returned yielding a response rate of 506 In Lithuania 236 responses from 54 municipalities

were gathered by web survey In the Netherlands three municipalities were surveyed on a census base by means

of a web survey In total 319 responses of 910 invitations were returned (35) mdash 90 (45) 115 (38) and 114

(28) for the respective municipalities In Switzerland 516 out of 1326 local employees with e-mail access in

a municipality completed the survey giving a response rate of 389 In the United Kingdom the survey was

posted online and 1360 employees both in administrative jobs and front-line services of one municipality

- 12 -

Municipalities were selected because of their convenient accessibility and proximity to the

researcher and a convenience sample was used in most countries Seven countries surveyed

one municipality but seven local councils were surveyed in Australia Web surveys were used

in four countries while three countries used both web and paper-pencil survey The number

of respondents varied among the countries depending on the numbers of questionnaires

distributed the survey method and the response rate In most countries we used

approximately 250 responses but only 162 from Italy3 The majority of the respondents were

women (566) while in terms of service length those in the largest group (393) had

worked for fewer than 10 years Table 2 shows the distribution of respondentsrsquo sex length of

service employment status and country

[Table 2 about here]

Analyses

The statistical analysis applied confirmatory factor analysis (CFA) CFA was used to assess

the fit of the data to the hypothesized measurement model because it is preferred where

measurement models have a well-developed underlying theory for hypothesized patterns of

loadings The data were analyzed using Amos 180 (Arbuckle 2009) with the maximum

likelihood (MLE) estimation method and LISREL 872 (Joumlreskog and Soumlrbom 2005) with the

diagonally weighted least squares (D-WLS) estimation method

The individual items are designed to measure a theoretically continuous construct

but the observed responses are a discrete realization of five categories The common method

received an email to complete the survey The response rate was 19 In the USA the total number of

employees surveyed was 626 in one city using both electronic and paper-pencil surveys resulting in a response

rate of 588 3 In countries where significantly more than 250 surveys were collected we randomly selected only 250 cases

for analysis

- 13 -

of estimation within CFA is MLE a technique which assumes that the observed variables are

continuous and normally distributed These assumptions are not met when the observed data

are discrete as occurs when using ordinal scales as in this study A simulation study (Flora

and Curran 2004) found that D-WLS performs well for ordinal measures across a variety of

conditions especially for smaller samples Coursey and Pandey (2007a 2007b) argued the

importance of using D-WLS and not standard MLE in structural equations for ordinal data4

In most cases however the differences between MLE and D-WLS are not substantive to the

conclusions5 No comparison between standard MLE and D-WLS is provided in this paper

We give results from both MLE and D-WLS and this will aid in confirming the stability and

robustness of the results across varying mathematical specifications

For gauging reliability Cronbachrsquos alpha and composite reliability were tested

Composite reliability should be equal to or greater than 07 (Hair et al 2010) Convergent

validity was assessed from the measurement model by determining the statistical significance

of each indicatorrsquos estimated pattern coefficient on its specified underlying construct

Discriminant validity was assessed by determining whether the confidence interval around

the correlation estimate between two factors included 100 (Anderson and Gerbing 1988)

For model fit assessment goodness-of-fit index (GFI) comparative fit index (CFI)

and the root mean square error of approximation (RMSEA) were used The model achieves

an acceptable fit to the data when GFI and CFI equals or exceeds 090 and RMSEA values

4 Applying standard MLE to ordinal data such as Likert-type items may produce significant estimation

problems such as inflation of chi-square fit statistics and biased underestimation of parameters and standard

errors (Flora and Curran 2004) Coursey and Pandey (2007a 2007b) used a modified MLE using the asymptotic

covariance matrix and D-WLS for adjusting the ordinal nature of the data They reported that although the

modified MLE findings are slightly different on a few model parameters the differences are not substantive to

the conclusions However model fit estimators were clearly better for the D-WLS 5 Given normally distributed categorical variables MLE can be used with confidence when a variable has four

or more categories If there is a large sample a five-point Likert scale and the skewness and kurtosis of variables

is within acceptable limits it suggests distribution symmetry ―In most cases where the hypothesized model is

well specified and the scaling based on more than three categories it seems unlikely that there will be much

difference between the findings (Byrne 2010 160)

- 14 -

fall below 008 (Hair et al 2010) In general the larger the value of GFI and CFI and the

smaller the value of RMSEA the better the fit of the model (Bollen 1989) We also present

the χ2 test statistic Lower values of χ

2 indicate a better fit and should be nonsignificant but

the χ2 value increases as sample size increases and thus for large sample sizes this statistic

may lead to rejection of a model with good fit (Hair et al 2010)

Results

Using SPSS 180 descriptive statistics were computed for individual items as shown in table

1 One item (APP3) in the APP dimension ―I like to discuss topics regarding public programs

and policies with others was mistakenly omitted in the survey conducted in the United

States and so it was deleted in the statistical analysis In the assessment of normality no

items showed a skew or kurtosis value greater than the cutoffs of |3| or |8| recommended by

Kline (2005) As data were distributed normally MLE as well as D-WLS estimation was used

Based on inspection of item-total correlations one item (COM4) was dropped from further

analysis because it was weakly correlated with the overall index (011) Missing data is an

issue in CFA as it can cause estimation errors All variables in the survey have missing data

rates of approximately 016 to 143 Mean substitution was used to calculate replacement

values for the missing values in each countryrsquos dataset (Hair et al 2010) The deletion of the

two items left 31 items for the next stage of analysis

The five-correlated factor model was tested using CFA The CFA model in the initial

analysis hypothesized a priori that (1) responses to the 30-item PSM measurement

instrument could be explained by five factors such as APP CPV1 CPV2 COM SS (2) each

item would have a nonzero loading on the PSM factor that it was designed to measure and

- 15 -

zero loadings on all other factors (3) the five factors would be correlated and (4)

measurement error terms would be uncorrelated The resulting CFA with MLE suggested that

the initial five-correlated factor model was not a good fit to the data χsup2 (df = 424) = 43699 p

lt 001 GFI = 0899 CFI = 835 RMSEA = 0057 [055 059] At this stage APP was highly

correlated with CPV1 (r = 996 in MLE and r = 988 in D-WLS) suggesting the two

dimensions were not unique and lacked discriminant validity

[Table 3 about here]

Given the disconfirmation of the initial model an effort was made to estimate an

alternative model The two dimensions APP and CPV1 were combined to form a single

dimension Attraction to Public Service (APS) The items of CPV1 may be regarded as

action-oriented by the respondents because dedication to public service community and

common good is underlined in these items Consequently it is reasonable to combine APP

with CPV1 to form APS a dimension that focuses more on disposition to serve the public to

work for the common good and to participate in public policy processes After the overall

number of dimensions was reduced to four using standardized factor loadings and

modification indices as criteria the items with the lowest factor loading in each dimension

were deleted and the items whose modification index indicated association with multiple

dimensions were eliminated A CFA model with the remaining items was iteratively tested

until it achieved a good fit to the data The outcome of this refinement process was a 16-item

index of four factors Overall model fit by every measure is quite strong for the D-WLS

estimation technique The estimation results for the four-dimension model are presented in

table 4

- 16 -

[Table 4 about here]

Cronbachrsquos coefficient alpha for the 16-item PSM index was 860 and the coefficient

alphas for the four dimensions ranged from 645 to 784 The composite reliability of the set

of reflective indicators for each dimension of PSM ranged from 713 to 826 and indicated

adequate internal consistency The resulting factor structure showed a four-factor structure

with all items loading significantly onto their a priori dimension (p lt 001) and the

standardized factor loadings ranging from 0490 to 0766 in MLE and from 515 to 807 in

D-WLS The results provide support for convergent validity The correlation estimates

between the two factors ranged from 0462 to 0848 in MLE and from 505 to 859 in D-

WLS The confidence intervals (plusmn2 standard errors) around the correlation estimates between

the two factors did not include 100 The result provides support for discriminant validity To

test whether the four-dimension model is superior to other designs such as the two or three-

dimension models a series of CFA were conducted and the results show that the four-

dimension model is a significantly better fit than any of the other designs

To test whether the factorial structure of the PSM instrument is equivalent across the

samples from the twelve countries four nested models were tested as part of multi-group

analysis with MLE (Byrne 2010) First the unconstrained model was the baseline model

which relaxed all equality constraints This model hypothesized the same four-factor structure

of the instrument across twelve countries The fit indices for this baseline model were χsup2 (df =

1176) = 24977 p lt 001 GFI = 0901 CFI = 900 RMSEA = 0020 [019 021] These

indices indicate that the four-factor model of the PSM instrument exhibits acceptable fit

across the samples from the twelve countries Second the measurement weights model tested

- 17 -

the invariance of factor loadings across countries by placing equality constraints on these

parameters The fit indices for this model were χsup2 (df = 1308) = 28088 p lt 001 GFI = 0890

CFI = 887 RMSEA = 0020 [019 021] Since the measurement weights model was nested

within the unconstrained model the chi-square difference text ∆χsup2 (132) = 3111 p lt 001

indicated that some equality constraints of factor loadings did not hold across the twelve

countries Further test results suggested that the equality constraints of structural covariances

and measurement residuals were not upheld across the twelve countries It needs to be

stressed that despite the differences in factor loadings and factor covariances the results

support the main assumption that the same four-factor structure holds across different

countries We may assume that people in different cultures understand PSM in the same

fashion

To find differences in the factor structure of PSM among countries we conducted a

series of CFA with each countryrsquos dataset and followed the same procedure as when

analyzing the sample as a whole As table 5 shows although there are great similarities

between the appropriate items in the countries there are also some differences Even though

the items vary according to national differences the four-factor structure had a good fit to the

data Thus the results confirm that the four-factor structure of PSM can be generalized

[Table 5 about here]

The APS dimension covers both an individualrsquos attraction to public participation and

commitment to public interest but all its items refer to some notion of public interest such as

improving public service aiding community tackling social problems meaningful public

service and contributing to the common good The APS dimension captures an individualrsquos

- 18 -

disposition to serve the public interest through public participation When developing the

items some assumed that the items of committing to public interests can be separated from

those of pursuing public values in the CPV dimension The empirical test showed that

respondents tended to understand committing to public interests as behavior-oriented and so

the items for public interest are not differentiated from participation-oriented items in their

perceptions but are different from those for public values The respondents seem to

understand that pursuing public values is more related to personal beliefs or creeds Thus the

CPV dimension is solely focused on public values and the items of pursing public interests

are combined with those of the APP dimension to make the APS dimension

Public values and their relative importance are neither impervious to change over

time nor the same among all countries (Charles de Jong and Ryan 2011 Davis and West

2009) Moreover some values may be in conflict as Martinsen and Beck Joslashrgensen (2010)

discussed in regard to the tensions between accountability and efficiency It is therefore not

possible to include all public values in one dimension (Beck Joslashrgensen and Bozeman 2007)

Accordingly the items of the CPV dimension are developed for estimating individual

commitment to seek public values which there are general consensus about and which are

generally formulated

When all four dimensions of PSM are included in a survey cross-national

equivalence of the survey can be assured The items represent reflections or manifestations of

each dimension and hence are relatively homogeneous and interchangeable Adding or

deleting items does not change the essential nature of the dimension Thus any combination

of items in a dimension can be regarded as equivalent When using structural equation

modeling (SEM) each dimension needs to have at least three reflective indicators in order to

achieve identification or it needs to have at least two reflective indicators and be allowed to

- 19 -

covary with another dimension that has at least two reflective measures With one indicator

the dimension is identified if the loading and the variance of the uniqueness are both fixed

(Bollen 1989 Edwards 2010) Thus the number of items in each dimension is not restricted

when measuring PSM However we recommend using multiple items for each dimension

because more research is needed to investigate the appropriateness of this PSM instrument

with varying samples

The test results showed that model fit estimators were clearly better for the D-WLS

It means that in this study MLE was more rigorous than D-WLS in assessing the fit of the

model The empirical evidence indicated that MLE may be an acceptable technique for

ordinal data when there is a large sample a five-point Likert scale and the skewness and

kurtosis of variables is within acceptable limits as suggested by Byrne (2010)

Discussion

The analysis reaffirmed the four-dimensional structure of PSM but the results were not

entirely consistent with the original CFA we tested We were ultimately able to converge on a

good-fit operational model for PSM that was also consistent with underlying theory The

analysis is not able to resolve the debate about measurement models however for reasons

discussed below

Choosing the correct measurement model (ie reflective vs formative) is important

as measurement model misspecification results not only in misrepresentation of the epistemic

relationships between constructs and their measures but also in biased parameter estimates of

structural relationships (Jarvis et al 2003 Law and Wong 1999 MacKenzie 2003

MacKenzie Podsakoff and Jarvis 2005) A formative measurement model in isolation is

- 20 -

under-identified and cannot be estimated in the case of covariance-based SEM (CB-SEM)

(Bollen 1989) A construct with formative indicators must emit paths to at least two unrelated

reflective indicators two unrelated latent constructs or some combination of the two in order

to solve the problem of under-identification (Jarvis MacKenzie and Podsakoff 2003

MacCallum and Browne 1993 MacKenzie Podsakoff and Jarvis 2005) Thus ―when

dependent constructs change the empirical nature of the formatively measured construct

changes (Wilcox Howell and Breivik 2008 1223) and ―when specifying formative

measurement models what might seem like an innocuous choice of outcomes to achieve

identification can have fundamental ramifications for the meaning of the formative construct

and the inferences based on it (Edwards 2010 7)

Earlier research found that the association between a construct and its measures

should remain stable regardless of the larger causal model in which the constructs and

measures are embedded (Edwards and Bagozzi 2000) The association between a construct

and its measures is generally stable for reflective measures but associations of formative

measures with their construct are unstable The susceptibility of formative measurement to

both interpretational confounding and external inconsistency recommends that formative

measurement is not an equally attractive alternative to reflective measurement and that

whenever possible researchers should opt for reflective measurement (Howell Breivik and

Wilcox 2007a 2007b Kim Shin and Grover 2010) In the case of second-order models

researchers are limited to only a second-order reflective model in CB-SEM (Chin 2010)

Regardless of the measurement debate and the tradeoffs between reflective or

formative specification this study makes a theoretical argument that PSM is a second-order

formative construct This second-order formative factor model (reflective first-order

formative second-order) is also suggested as an alternative model that captures the

- 21 -

multidimensionality of complex concepts and avoids the shortcomings of first-order

formative measurement (Edwards 2010) However this model itself cannot achieve

identification One possible strategy is to develop direct reflective measures of PSM based on

the view that PSM is a general concept that is not only influenced by its specific dimensions

but also can be assessed using general measures (Edwards 2010 Javis MacKenzie and

Podsakoff 2003)6 However these general measures have not yet been developed for the

PSM construct

Another way around the problem of underidentification of formative models is to

apply the component-based approach known as partial least squares path modeling (PLS-PM)

which can model formative indicators (Fornell and Bookstein 1982 Goumltz Liehr-Gobbers

and Krafft 2010) The CB-SEM and the PLS-PM methods have distinctive statistical

characteristics and selecting an approach to SEM depends on the particular research situation

CB-SEM is the method of choice for theory testing while PLS-PM is appropriate for

prediction-oriented applications (Ringle Goumltz Wetzels and Wilson 2009) One advantage of

the PLS method is that it demands significantly fewer requirements compared to that of

covariance structure analyses but nevertheless delivers consistent estimation results Another

asset of the PLS approach is its ability to deal with formative as well as reflective indicators

even within one structural equation model This indicates that the PLS approach can be

appropriate for explorative analysis of structural equation models (Chin 2010 Goumltz Liehr-

Gobbers and Krafft 2010) A Monte Carlo simulation study to compare CB-SEM and PLS-

PM showed that PLS-PM provides a viable approximation of model parameters when the

prerequisites for CB-SEM are not met (Ringle Goumltz Wetzels and Wilson 2009) A formative

6 For example job satisfaction can be assessed with general reflective measures such as ―in general I am

satisfied with my job and the formative dimensions can be assessed with measures that describe satisfaction

with each job facet (Edwards 2010 15)

- 22 -

composite in the form of indexes needs to be tested to know the relative weights (γ-

parameters) of the dimensions of PSM In PLS-PM a second-order formative factor is

directly measured by observed variables for all the first order factors While this approach

repeats the number of manifest variables used the model can be identified without adding

outcome constructs that would be considered inappropriate or irrelevant from a conceptual

standpoint (Chin 2010 Edwards 2010) The test result of PLS-PM using Smart PLS 20

(Ringle Wende and Will 2005) produced the function PSM = 31APS + 36CPV + 28COM

+ 31SS However one consideration is whether and how construct error is modeled for the

formative construct (Diamantopoulos 2006) When using PLS-PM the researcher is making

the assumption that there is no error at the construct level as all constructs in PLS-PM are

modeled without error When using PLS-PM the absence of an error term will tend to

increase the weights therefore inflating their importance (Bollen and Lennox 1991) The

primary statistic for assessing a formative indicator is its weight the partialized effect of the

indicator on its intended construct controlling for the effects of all other indicators of that

construct (Cenfetelli and Bassellier 2009) A formative measure is essentially a multiple

regression with the construct representing the dependent variable and the indicators as the

predictors (Diamantopoulos and Winklhofer 2001) PSM may be regarded as a weighted

linear composite of its dimensions The weights used to create the composites are different

across CE-SEM and PLS-PM The causal dimensional weights generated in CB-SEM are

estimated in a model in which a construct-level error term is specified (eg PSM = γ1APM +

γ2CPI + γ3COM + γ4SS + ζ) whereas the weights in PLS-PM are not (eg PSM = γ1APM +

γ2CPI + γ3COM + γ4SS) This leads to two different conceptualizations of the PSM construct

one in which the causal indicators do not capture the construct in its entirety and another in

which they do (Hardin et al 2010) These differences need to be discussed in future research

- 23 -

in this area

One practical suggestion to measure PSM is to apply unit weights to its dimensions if

we assume the effects of the four dimensions are equivalent A literature review indicates that

unit weights have substantial predictive validity when compared with regression weights The

literature shows that the empirical correlations between scores generated from differential and

unit-weighting approaches are quite high In creating composite scores unit weights can be

an appropriate approach for weighting under many circumstances (Bobko Roth and Buster

2007) Although the uses of unit-weight composites will likely result in the explanation of

less variance than empirical estimations of dimensional weights they will remain consistently

defined across studies and hence suitable for theory testing Alternatively more thorough

discussions on the dimensions of PSM may suggest fixed-weight composites (Hardin et al

2010) A previous study (Kim 2010 22) also suggested ―Overall indexing PSM by

measuring all the dimensions of PSM and simply summing the averages of each dimension

can make it easier to measure PSM in practice as PSM = APS + CPV + COM + SS

An advantage of regarding PSM as a second-order formative construct is that the

dimensions can be analyzed both individually and collectively This is important as the

dimensions are expected to have different antecedents and consequences For instance Perry

(1997) found some significant differences in the influence of independent variables on the

four dimensions of PSM Taylor (2007) confirmed that certain dimensions are more important

than others in influencing work outcomes Moynihan and Pandey (2007a) reported that red

tape is negatively related to one dimension but reform orientation and hierarchical authority

is positively related to another dimension A recent study found that the SS dimension has a

positive relationship with occupational intention but the other dimensions do not (Liu et al

2011) Thus it is valuable to analyze the causes and effects of individual dimensions of PSM

- 24 -

as well as of the PSM construct

Conclusion

The purpose of this study is to develop a measurement instrument for PSM for cross-national

research and comparison On the basis of the sharpened construct of PSM (Kim and

Vandenabeele 2010) using data from an international survey conducted in 12 countries we

found that the four-factor structure of PSM is generalizable across country cases This study

provides the measurement index of PSM with the four dimensions of APS CPV COM and

SS The APS dimension measures the instrumental motives of PSM the CPV dimension

determines the value-based motives of PSM while the COM and SS dimensions gauge the

identification motives and self-sacrifice respectively When all four dimensions of PSM are

used in a survey cross-national equivalence of survey items can be assured The items for

measuring PSM in future research are suggested in Appendix 4

This study has several limitations First looking for samples which were comparable

to the largest possible extent we opted for surveys in local government The results found

and discussed in this study are based on PSM among local government employees The

results might be different when surveying central government employees or other specific

groups in the public sector Thus the potential impact of the sample bias needs to be

considered in understanding this study It also means that the survey items for measuring

PSM should be examined with different samples in various areas and at different government

levels

Second the test results showed that there is a high degree of correlation among the

dimension such as APS CPV and COM Future research should not only investigate whether

- 25 -

the items reflect the core content of the dimensions but also continue to develop more

appropriate items for better discriminant validity of each dimension Third the content of

APP dimension was assumed to be distinct from the CPV dimension that focuses on personal

disposition to pursue public interest and public values However the public interest sub-

dimension was separated from the public values sub-dimension in the CPV dimension and

combined with the APP dimension to form the APS dimension Further research should

analyze the CPV dimension and investigate the appropriateness of the APS dimension

In order to facilitate research internationally the PSM construct needs to be

improved conceptually and operationally Kim and Vandenabeele (2010) have dealt with the

conceptual composition and operational dimension of PSM as well as the measurement

model of PSM The next step is to develop an internationally robust measurement instrument

of PSM This study proposes that scholars use all four of the dimensions in measuring the

conceptual components of PSM such as self-sacrifice instrumental value-based and affective

motives because they collectively represent PSM It provides a set of appropriate items for

measuring PSM and confirms the dimensional structure of PSM The survey results suggest

that people in different cultures understand PSM in the same fashion We know that this study

is simply a step along the way to developing a more robust measure but we hope it

contributes to enhanced cross-national research and comparison along with the generation of

cumulative knowledge internationally

- 26 -

REFERENCES

Anderson James C and David D Gerbing 1988 Structural Equation Modeling in Practice

A Review and Recommended Two-Step Approach Psychological Bulletin 103

411-423

Arbuckle James L 2009 Amos 180 Userrsquos Guide Crawfordville FL Amos Development

Co

Beck Joslashrgensen Torben and Barry Bozeman 2007 Public Values An Inventory

Administration amp Society 39(3) 354-381

Beutel Ann M and Margaret Mooney Marini 1995 Gender and values American

Sociological Review 60 436 - 448

Bobko Philip Philip L Roth and Maury A Buster 2007 The Usefulness of Unit Weights in

Creating Composite Scores A Literature Review Application to Content Validity and

Meta-Analysis Organizational Research Methods 10(4) 689-709

Bollen Kenneth A 1989 Structural Equations with Latent Variables New York Wiley

Bollen K A and R Lennox 1991 Conventional Wisdom on Measurement A Structural

Equation Perspective Psychological Bulletin 110(2) 305-314

Byrne Barbara M 2010 Structural Equation Modeling with Amos Basic Concepts

Applications and Programming 2nd

ed New York Routledge

Camilleri Emanuel 2006 Towards Developing an Organizational Commitment ― Public

Service Motivation Model for Maltese Public Service Employees Public Policy

and Administration 21 63-83

Castaing Seacutebastien 2006 The Effects of Psychological Contract Fulfillment and Public

Service Motivation on Organizational Commitment in the French Civil Service

Public Policy and Administration 21 84-98

Cenfetelli Ronald T and Geneviegraveve Bassellier 2009 Interpretation of Formative

Measurement in Information Systems Research MIS Quarterly 33(4) 689-707

Charles Michael B W Martin de Jong and Neal Ryan 2011 Public Values in Western

- 27 -

Europe A Temporal Perspective American Review of Public Administration 41

75-91

Chin Wynne W 2010 How to Write Up and Report PLS Analysis In Handbook of Partical

Least Squares edited by V Esposito Vinzi et al 655-690 Berlin Springer-Verlag

Coursey David H and Sanjay K Pandey 2007a Public Service Motivation Measurement

Testing an Abridged Version of Perryrsquos Proposed Scale Administration amp Society

39 547-568

Coursey David H and Sanjay K Pandey 2007b Content Domain Measurement and

Validity of the Red Tape Concept American Review of Public Administration 37

342-361

Coursey David H James L Perry Jeffrey L Brudney and Laura Littlepage 2008

Psychometric Verification of Perryrsquos Public Service Motivation Instrument Results

for Volunteer Exemplars Review of Public Personnel Administration 28 79-90

Davis Paul and Karen West 2009 What Do Public Values Mean for Public Action Putting

Public Values in Their Plural Place American Review of Public Administration 39

602-618

DeHart-Davis Leisha Justin Marlowe and Sanjay K Pandey 2006 Gender Dimensions of

Public Service Motivation Public Administration Review 66 873-887

Diamantopoulos Adamantios 2006 The error term in formative measurement models

Interpretation and modeling implications Journal of Modeling in Management 1(1)

7-17

Diamantopoulos Adamantios and Judy A Siguaw 2006 Formative versus Reflective

Indicators in Organizational Measure Development A Comparison and Empirical

Illustration British Journal of Management 17 263-282

Diamantopoulos Adamantios and Heidi M Winklhofer 2001 Index Construction with

Formative Indicators An Alternative to Scale Development Journal of Marketing

Research 38 269-277

Edwards Jeffrey R 2010 The Fallacy of Formative Measurement Organizational Research

- 28 -

Methods Online First published on August 19 2010 as

doi1011771094428110378369

Edwards Jeffrey R and Richard P Bagozzi 2000 On the Nature and Direction of

Relationships between Constructs and Measures Psychological Methods 5 155-

174

Flora David B and Patrick J Curran 2004 An Empirical Evaluation of Alternative

Methods of Estimation for Confirmatory Factor Analysis with Ordinal Data

Psychological Methods 9 466-491

Fornell Claes and Fred L Bookstein 1982 Two structural equations models LISREL and

PLS applied to customer exit-voice theory Journal of Marketing Research 19 440-

452

Giauque David Adrian Ritz Freacutedeacuteric Varone Simon Anderfuhren-Biget and Christian

Waldner 2011 Putting Public Service Motivation into Context A Balance between

Universalism and Particularism Forthcoming in International Review of

Administrative Sciences 77(2)

Goumltz Oliver Kerstin Liehr-Gobbers and Manfred Krafft 2010 Evaluation of Structural

Equation Models Using the Partial Least Squares (PLS) Approach In V Esposito

Vinzi et al (eds) Handbook of Partial Least Squares 691-711 DOI 101007978-3-

540-32827-8_30 Springer-Verlag Berlin Heidelberg

Guinot Christiane Julie Latreille and Michel Tenenhaus 2001 PLS Path Modelling and

Multiple Table Analysis Application to the Cosmetic Habits of Women in Ile-de-

France Chemometrics and Intelligent Laboratory Systems 58 247-259

Hair Joseph F Jr William C Black Barry J Babin and Rolph E Anderson 2010

Multivariate Data Analysis 7th

ed Upper Saddle River NJ Pearson Education

Hardin Andrew M Jerry Cha-Jan Chang Mark A Fuller and Gholamreza Torkzadeh 2010

Formative Measurement and Academic Research In Search of Measurement

Theory Educational and Psychological Measurement Online First published on

June 11 2010 as doi101177001316440370208

- 29 -

Howell Roy D Einar Breivik and James B Wilcox 2007a Reconsidering Formative

Measurement Psychological Methods 12 205-218

Howell Roy D Einar Breivik and James B Wilcox 2007b Is Formative Measurement

Really Measurement Reply to Bollen (2007) and Bagozzi (2007) Psychological

Methods 12 238-245

Jarvis Cheryl Burke Scott B MacKenzie and Philip M Podsakoff 2003 A Critical Review

of Construct Indicators and Measurement Model Misspecification in Marketing and

Consumer Research Journal of Consumer Research 30 199-218

Joumlreskog Karl G and Dag Soumlrbom 2005 LISREL 872 (Computer Software) Lincolnwood

IL Scientific Software International

Kim Gimun Bongsik Shin and Varun Grover 2010 Investigating Two Contradictory Views

of Formative Measurement in Information Systems Research MIS Quarterly 34

345-365

Kim Sangmook 2009a Revising Perryrsquos Measurement Scale of Public Service Motivation

American Review of Public Administration 39 149-163

Kim Sangmook 2009b Testing the Structure of Public Service Motivation in Korea A

Research Note Journal of Public Administration Research and Theory 19 839-

851

Kim Sangmook 2010 Testing a Revised Measurement of Public Service Motivation

Reflective versus Formative Specification Journal of Public Administration

Research and Theory Doi101093jopartmuq048

Kim Sangmook and Wouter Vandenabeele 2010 A strategy for building public service

motivation research internationally Public Administration Review 70 701-709

Kline Rex B 2005 Principles and Practice of Structural Equation Modeling 2nd

ed New

York Guilford Press

Law Kenneth and Chi-Sum Wong 1999 Multidimensional Constructs in Structural

Equation Analysis An Illustration using the Job Perception and Job Satisfaction

Constructs Journal of Management 25 143-160

- 30 -

Leisink Peter and Bram Steijn 2009 Public Service Motivation and Job Performance of

Public Sector Employees in the Netherlands International Review of

Administrative Sciences 75 35-52

Liu Bangcheng Chun Hui Jin Hu Wensheng Yang and Xinli Yu 2011 How well can

Public Service Motivation Connect with Occupational Intention International

Review of Administrative Sciences 77 191-211

Liu Bangcheng Ningyu Tang and Xiaomei Zhu 2008 Public Service Motivation and Job

Satisfaction in China An Investigation of Generalisability and Instrumentality

International Journal of Manpower 29 684-699

MacCallum Robert C and Michael W Browne 1993 The Use of Causal Indicators in

Covariance Structure Models Some Practical Issues Psychological Bulletin 114

533-541

MacKenzie Scott B 2003 The dangers of poor construct conceptualization Journal of the

Academy of Marketing Science 31(3) 323-326

MacKenzie Scott B Philip M Podsakoff and Cheryl Burke Jarvis 2005 The Problem of

Measurement Model Misspecification in Behavioral and Organizational Research

and Some Recommended Solutions Journal of Applied Psychology 90 710-730

Martinsen Dorte Sindbjerg and Torben Beck Joslashrgensen 2010 Accountability as a

Differentiated Value in Supranational Governance American Review of Public

Administration 40 742-760

Moynihan Donald P and Sanjay K Pandey 2007a The Role of Organizations in Fostering

Public Service Motivation Public Administration Review 67 40-53

Moynihan Donald P and Sanjay K Pandey 2007b Finding Workable Levers over Work

Motivation Comparing Job Satisfaction Job Involvement and Organizational

Commitment Administration amp Society 39 803-832

Pandey Sanjay K and Edmund C Stazyk 2008 Antecedents and Correlates of Public

Service Motivation In Motivation in Public Management edited by James L Perry

and Annie Hondeghem 101-117 Oxford Oxford University Press

- 31 -

Perry James L 1996 Measuring Public Service Motivation An Assessment of Construct

Reliability and Validity Journal of Public Administration Research and Theory 6

5-22

Perry James L 1997 Antecedents of Public Service Motivation Journal of Public

Administration Research and Theory 7 181-197

Perry James L 2000 Bringing Society in Toward a Theory of Public Service Motivation

Journal of Public Administration Research and Theory 10 471-488

Perry James L and Annie Hondeghem 2008a Motivation in Public Management Oxford

Oxford University Press

Perry James L and Annie Hondeghem 2008b Directions for Future Theory and Research

In Motivation in Public Management edited by James L Perry and Annie

Hondeghem 294-313 Oxford Oxford University Press

Perry James L Annie Hondeghem and Lois Recascino Wise 2010 Revisiting the

Motivational Bases of Public Service Twenty Years of Research and an Agenda for

the Future Public Administration Review 70(5) 681-690

Perry James L and Wouter Vandenabeele 2008 Behavioral dynamics Institutions

Identities and Self-Regulation In Motivation in Public Management edited by

James L Perry and Annie Hondeghem 56-79 Oxford Oxford University Press

Perry James L and Lois Recascino Wise 1990 The Motivational Bases of Public Service

Public Administration Review 50 367-373

Rainey Hal G 1982 Reward Preference among Public and Private Managers In Search of

the Service Ethic American Review of Public Administration 16 288-302

Ringle Christian M Oliver Goumltz Martin Wetzels and Bradley Wilson 2009 On the Use of

Formative Measurement Specifications in Structural Equation Modeling A Monte

Carlo Simulation Study to Compare Covariance-Based and Partial Least Squares

Model Estimation Methodologies Faculty of Economics and Business Administration

Maastricht Univeristy

Ringle C M S Wende and S Will 2005 Smart PLS 20 (M3) Beta

- 32 -

httpwwwsmartplsde

Ritz Adrian 2011 Attraction to Public Policy Making A Qualitative Inquiry into

Improvements in PSM Measurement Forthcoming in Public Administration

Spector Paul E 1992 Summated Rating Scales Construction An Introduction Newbury

Park CA Sage

Steenbergen Marco R 1996 Compassion and American public opinion An analysis of the

NES Humanitarianism Scale Report to the Board of Overseers of the National

Election Studies

Steffen Patrick R and Kevin S Masters 2005 Does compassion mediate the intrinsic

religion-health relationship Annals of Behavioral Medicine 30 217-223

Taylor Jeannette 2007 The Impact of Public Service Motives on Work Outcomes in

Australia A Comparative Multi-Dimensional Analysis Public Administration 85

931-959

Vandenabeele Wouter 2007 Towards a Theory of Public Service Motivation An

Institutional Approach Public Management Review 9 545-556

Vandenabeele Wouter 2008a Government Calling Public Service Motivation as an Element

in Selecting Government as an Employer of Choice Public Administration 86

1089-1105

Vandenabeele Wouter 2008b Development of a Public Service Motivation Measurement

Scale Corroborating and Extending Perryrsquos Measurement Instrument International

Public Management Journal 11 143-167

Vandenabeele Wouter 2009 The Mediating Effect of Job Satisfaction and Organizational

Commitment on Self-Reported Performance More Robust Evidence of the PSM-

Performance Relationship International Review of Administrative Sciences 75 11-

34

Wilcox James B Roy D Howell and Einar Breivik 2008 Questions about Formative

Measurement Journal of Business Research 61 1219-1228

Wise Lois Recascino 2000 The Public Service Culture In Public Administration Concepts

- 33 -

and Cases 7th ed edited by Richard J Stillman II 342-353 Boston Houghton

Mifflin

Wright Bradley E 2008 Methodological Challenges associated with Public Service

Motivation Research In Motivation in Public Management edited by James L

Perry and Annie Hondeghem 80-98 Oxford Oxford University Press

Wright Bradley E and Sanjay K Pandey 2008 Public Service Motivation and the

Assumption of Person-Organization Fit Testing the Mediating Effect of Value

Congruence Administration amp Society 40 502-521

- 34 -

Table 1

Possible Indicators of PSM and Descriptive Statistics (n = 2868)

Dimensions and items Mean Standard

Deviation

Item-Total

Correlation

Attraction to Public Participation (APP)

APP1 I am interested in helping to improve public service

APP2 I am satisfied when I see people benefiting from the public programs I was

involved in

APP3 I like to discuss topics regarding public programs and policies with others

APP4 I believe that public sector activities contribute to our general welfare

APP5 I admire people who initiate or are involved in activities to aid my

community

APP6 Contributing to public programs and policies helps me realize myself

APP7 It is important to contribute to activities that tackle social problems

412

441

355

393

423

351

409

696

662

917

753

672

881

734

524

421

482

412

493

478

552

Commitment to Public Values (CPV)

Sub-dimension for Public Interests (CPV1)

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

CPI3 I would prefer seeing public officials do what is best for the whole

community even if it harmed my interests

CPI4 Serving the public interest is more important than helping a single individual

Sub-dimension for Public Values (CPV2)

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of public

services

CPV3 It is fundamental that public services respond to the needs of the citizens

CPV4 Decisions regarding public services should be democratic despite the time

and effort it takes

CPV5 Everybody is entitled to a good service even if it costs a lot of money

CPV6 It is fundamental that the interests of future generations are taken into

account when developing public policies

CPV7 To act ethically is essential for public servants

CPV8 I believe that public employees must always be aware of the legitimacy of

their activities

CPV9 I personally identify with the aim of protecting individual liberties and rights

418

409

348

337

428

416

433

394

367

426

441

434

414

695

662

899

995

730

704

710

852

932

689

681

692

763

525

596

480

251

484

437

356

325

376

439

462

402

392

Compassion (COM) COM1 It is difficult for me to contain my feelings when I see people in distress

COM2 I feel sympathetic to the plight of the underprivileged

COM3 I empathize with other people who face difficulties

COM4 I have little compassion for people in need who are unwilling to take the

first step to help themselves

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

354

387

388

299

412

391

951

825

707

1083

755

765

399

511

389

011

487

502

Self-Sacrifice (SS)

SS1 Making a difference to society means more to me than personal achievements

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS5 People should give back to society more than they get from it

SS6 Serving other citizens would give me a good feeling even if no one paid me for

it

SS7 I would agree to a good plan to make a better life for the poor even if it costs

me money

357

329

339

302

345

363

350

904

872

949

982

888

1049

895

508

584

454

531

412

368

546

- 35 -

Table 2

Background of Respondents (n = 2868)

Variables Characteristics Respondents

Sex

Male

Female

NA

1209

1622

37

422

566

13

Length of service

(years)

0 ~ 10

10 ~ 20

20 ~ 30

30+

NA

1126

723

484

491

44

393

252

169

171

15

Organizational status

―Do you supervise employees

No

Yes

NA

1472

1076

320

513

375

112

Country

Australia

Belgium

China

Denmark

France

Italy

Korea

Lithuania

The Netherlands

Switzerland

United Kingdom

USA

249

214

230

249

266

162

253

236

249

250

260

250

87

75

80

87

93

56

88

82

87

87

91

87 Note NA = no answer

- 36 -

Table 3

Standardized factor loadings and correlations for the 31-item PSM index (n = 2868)

Dimensions and items SFL (MLE) SFL (D-WLS) Alpha

APP

APP1

APP2

APP4

APP5

APP6

APP7

597

486

455

558

490

600

657

552

514

625

549

665

699

CPV1

CPI1

CPI2

CPI3

CPI4

588

660

488

289

653

732

535

294

542

CPV2

CPV1

CPV2

CPV3

CPV4

CPV5

CPV6

CPV7

CPV8

CPV9

611

546

436

393

418

534

598

496

460

690

631

505

457

486

601

669

573

530

745

COM

COM1

COM2

COM3

COM5

COM6

496

569

498

597

607

516

651

538

670

684

687

SS

SS1

SS2

SS3

SS4

SS5

SS6

SS7

567

736

558

728

467

409

663

660

762

591

695

515

494

709

782

Measures of fit (MLE) χsup2 (df = 424) = 43699 p lt 001 CFI = 835 GFI = 899

RMSEA = 057 [055 059]

Measure of fit (D-WLS) χsup2 (df = 424) = 39247 p lt 001 CFI = 971 GFI = 976

RMSEA = 054 [052 055]

Inter-factor correlations MLE (D-WLS)

1 2 3 4

1 APS

2 CPV1 996 (988)

3 CPV2 902 (890) 798 (798)

4 COM 787 (783) 758 (762) 779 (783)

5 SS 700 (742) 800 (859) 509 (537) 720 (729)

Note SFL = standardized factor loading All factor loadings and correlations are significant at p lt 001

- 37 -

Table 4

Standardized factor loadings and correlations for the 16-item PSM index (n = 2868)

Dimensions and items SFL

(MLE)

SFL

(D-WLS)

APS

APP5 I admire people who initiate or are involved in activities to aid my

community

APP7 It is important to contribute to activities that tackle social problems

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

611

592

637

666

665

669

690

751

CPV

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of

public services

CPV6 It is fundamental that the interests of future generations are taken

into account when developing public policies

CPV7 To act ethically is essential for public servants

612

546

551

590

698

631

618

658

COM

COM1 It is difficult for me to contain my feelings when I see people in

distress

COM2 I feel sympathetic to the plight of the underprivileged

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

490

546

596

604

515

608

669

679

SS

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS7 I would agree to a good plan to make a better life for the poor even if

it costs me money

763

571

766

678

807

629

747

755

Measures of fit (MLE)

χsup2 (df = 98) = 8490 p lt 001

CFI = 939 GFI = 965

RMSEA = 052 [049 055]

Measure of fit (D-WLS)

χsup2 (df = 98) = 6069 p lt 001

CFI = 988 GFI = 992

RMSEA = 043 [039 046]

Alpha CR (D-WLS) Inter-factor correlations MLE (D-WLS)

1 APS 717 788 1 APS 2 CPV 3 COM

2 CPV 663 747 848 (859)

3 COM 645 713 781 (791) 789 (795)

4 SS 784 826 565 (596) 462 (505) 715 (726)

Note SFL = standardized factor loading Alpha = Cronbachrsquos alpha CR = composite reliability

All factor loadings and correlations are significant at p lt 001

- 38 -

Table 5

Results of CFA in the individual countries Standardized factor loadings and measures of fit

Factors

amp Items

Australia Belgium China Denmark France Italy

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

672

756

799

785

804

943

710

703

597

790

799

730

584

675

685

681

731

742

578

748

674

643

765

739

645

704

678

708

812

779

632

604

765

766

672

853

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

552

593

560

699

727

632

624

605

560

713

709

645

702

602

616

724

686

735

567

560

715

654

605

811

608

658

587

696

755

698

673

620

523

782

697

654

COM

COM1

COM2

COM3

COM5

COM6

623

584

662

688

701

739

732

608

475

808

666

581

578

616

605

649

670

673

699

751

737

679

816

776

568

546

801

626

673

850

637

546

741

715

601

822

SS

SS2

SS3

SS4

SS7

654

794

806

892

767

813

727

693

799

793

725

838

857

820

773

896

855

842

806

749

792

829

753

884

847

731

753

889

816

753

815

692

664

795

782

734

Inter-factor

correlations

093

~

752

163

~

810

320

~

773

323

~

825

452

~

827

497

~

855

323

~

850

321

~

847

498

~

899

536

~

902

425

~

829

476

~

865

Sample size 249 214 230 249 266 162

Measures of Fit

χsup2 638 444 881 732 765 554 817 625 735 565 750 564

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 980 100 939 987 969 997 967 994 975 997 949 995

GFI 961 986 941 986 948 991 949 992 955 992 926 986

RMSEA 036 00 063 050 051 026 053 035 045 026 059 033

- 39 -

Table 5 (Continued)

Factors

amp Items

Korea Lithuania The Netherlands Switzerland UK USA

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

619

701

667

706

781

717

666

512

682

751

605

779

557

629

630

620

719

729

535

621

702

643

694

735

655

725

738

710

804

822

535

621

702

643

694

735

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

594

607

722

722

644

722

674

535

533

790

589

711

568

554

538

655

628

545

627

580

615

711

666

704

800

512

555

873

581

646

627

580

615

711

666

704

COM

COM1

COM2

COM3

COM5

COM6

694

677

550

763

736

627

557

639

428

550

696

534

737

623

508

766

616

608

659

592

683

703

661

769

682

583

729

739

644

803

659

592

683

703

661

769

SS

SS2

SS3

SS4

SS7

776

752

679

779

770

811

707

721

635

725

734

706

814

656

743

820

767

764

745

632

689

788

662

744

671

818

832

830

864

802

745

632

689

788

662

744

Inter-factor

correlations

432

~

726

466

~

746

376

~

788

406

~

902

377

~

852

455

~

836

637

~

818

660

~

828

317

~

852

360

~

852

637

~

818

660

~

828

Sample size 253 236 249 250 260 250

Measures of Fit

χsup2 896 573 612 462 773 616 815 715 769 542 815 715

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 951 996 975 100 958 993 957 991 972 998 957 991

GFI 942 988 960 989 951 987 948 990 955 990 948 990

RMSEA 059 028 034 00 050 034 053 044 048 022 053 044

Note All factor loadings and correlations are significant at p lt 001

- 40 -

Appendix 1 Dimensions and possible items of PSM (the first round)

1 Attraction to Public Participation

(1) I am interested in developing public programs helping my country or community (Kim 2009a)

(2) Sharing my views on public policies with others is attractive to me (Kim 2009a)

(3) I am satisfied when seeing people get benefits from the public programs I was involved in (Kim 2009a)

(4) I like to discuss political subjects with others (Giauque et al 2011)

(5) I think the governmental activities contribute to our welfare

(6) I like to initiate actions to help out my community

(7) I do not care much about politicians (Perry 1996)

2 Commitment to Public Values

Possible items for public interest

(1A) Meaningful public service is very important to me (Perry 1996)

(2A) It is important for me to contribute for the common good (Giauque et al 2011)

(3A) I recognize myself with the mission of protection of liberties and rights endorsed by the nation-state

(Giauque et al 2011)

(4A) I would prefer seeing public officials do what is best for the whole community even if it harmed my

interests (Perry 1996)

(5A) To me serving the public interest is more important than helping other people (Vandenabeele 2008b)

Possible items for public values

(1B) I recognize myself with the promotion of the equality of chances endorsed by the nation-state (Giauque et

al 2011)

(2B) It is fundamental that the public service provided to the citizen is regular and continuous (Giauque et al

2011)

(3B) It is fundamental that public service innovates to respond to the new needs of the citizensclients (Giauque

et al 2011)

(4B) The dignity and well-being of all should be the most important concerns in any society (Steenbergen

1996)

(5B) It is important that public servants account for all the costsexpenses they make (Vandenabeele 2008b)

(6B) We have to do everything in our power to pursue the goal of democracy

3 Compassion

(1) It is difficult for me to contain my feelings when I see people in distress (Perry 1996)

(2) I feel sympathetic to the plight of the underprivileged (Kim 2009a)

(3) I am often reminded by daily events how dependent we are on one another (Perry 1996)

(4) I feel compassion for others in the difficulties they are facing (Steffen and Masters 2005)

(5) It is not really my problem if others are in trouble and need help (Beutel and Marini 1995)

(6) I find it hard to be sympathetic toward starving people in foreign lands when there is so much trouble in our

own country (Beutel and Marini 1995)

(7) I get very upset when I see other people treated unfairly (Beutel and Marini 1995)

(8) I would agree to a good plan to make a better life for the poor even if it cost me money (Beutel and Marini

1995)

(9) To me patriotism includes seeing the welfare of others (Perry 1996)

(10) I care very much about other people

4 Self-Sacrifice

(1) Making a difference in society means more to me than personal achievements (Perry 1996)

(2) I am prepared to make enormous sacrifices for the good of society (Perry 1996)

(3) I believe in putting duty before self (Perry 1996)

(4) I am one of those rare people who would risk personal loss to help someone else (Perry 1996)

(5) I think people should give back to society more than they get from it (Perry 1996)

(6) Serving other citizens would give me a good feeling even if no one paid me for it (Perry 1996)

(7) Much of what I do is for a cause bigger than myself (Perry 1996)

- 41 -

Appendix 2 Newly suggested items through the first round

1 Items for the Attraction to Public Participation Dimension

(1) In the public domain every person should be counted as equal

(2) I enjoy to think and argue in a political way

(3) I am satisfied when seeing my ideas being integrated in public policies

(4) I can easily relate to the way of thinking and arguing of politicians

(5) Working in the public sector I can better realize myself

2 Items for the Public Values Sub-Dimension

(1) Public service should give value for money

(2) Cost considerations should play an important role in decisions about public service provision

(3) It is fundamental that the interests of future generations are taken into account when developing public

policies

(4) To act with correct ethics is for public servants as important as their professional competence

(5) Primarily public servants should be accountable to the public and not to their superiors

3 Items for the Self-Sacrifice Dimension

(1) I would agree to a good plan to make a better life for the poor even if it costs me money

- 42 -

Appendix 3 An example of a summary of comments and suggestions (the second round)

On the item ―I am interested in developing public programs helping my country or community

A It is a good sentence but is it possible for everyone to ―develop public programs Even public officials are

limited to do this

B OK

C or the special group

D Double barrel and difficult to translate to my country ―I am interested in helping to improve public services

E Good but I feel that wording requires revision If we draw sample from local government then country is

redundant ―I am interested in developing public policy programs that aim in helping my community

F 4 (General appreciation from1 to 4) Not clear whether it refers to ones work or to ones concern as a citizen

―(In my work) I am interested in contributing to the development of public etchellip

G Instead of development I would use implementing

H Delete public which is redundant because of the characteristic helping my country or community but add

or implementing ―I am interested in developing or implementing programs helping my country or

community

I Referring to dimension Wording Similar to item 6

Note Names have been changed to alphabets in order to ensure the anonymity of participants

- 43 -

Appendix 4 Dimensions and Items of Public Service Motivation for Future Research

Dimensions and items

Attraction to Public Service (APS)

1 I am interested in helping to improve public service

2 I admire people who initiate or are involved in activities to aid my community

3 It is important to contribute to activities that tackle social problems

4 Meaningful public service is very important to me

5 It is important for me to contribute to the common good

Commitment to Public Values (CPV)

1 I think equal opportunities for citizens are very important

2 It is important that citizens can rely on the continuous provision of public services

3 It is fundamental that the interests of future generations are taken into account when

developing public policies

4 To act ethically is essential for public servants

5 I believe that public employees must always be aware of the legitimacy of their activities

6 I personally identify with the aim of protecting individual liberties and rights

Compassion (COM)

1 It is difficult for me to contain my feelings when I see people in distress

2 I feel sympathetic to the plight of the underprivileged

3 I empathize with other people who face difficulties

4 I get very upset when I see other people being treated unfairly

5 Considering the welfare of others is very important

Self-Sacrifice (SS)

1 I am prepared to make sacrifices for the good of society

2 I believe in putting civic duty before self

3 I am willing to risk personal loss to help society

4 I would agree to a good plan to make a better life for the poor even if it costs me money

- 9 -

first process was distributed to the participants to stimulate comments in the second round

(see Appendix 3) Based on the comments received in the second round discussion the

possible items for an international survey were selected However some items required

further discussion and so a third round was conducted from the 15th

of January until the 23rd

of January 2010 One issue that remained unresolved in the discussions relates to the

dimension of CPV Some argued that it is reasonable to divide CPV into two sub-dimensions

because committing to public interests is different from pursuing public values and there are

many important public values in democratic countries while others contended that CPV can

cover all important public values including public interest and there is no real benefit to add

one more dimension These differences were left to be resolved through empirical research

This did not affect the final items however and 33 items were confirmed The resulting

survey questionnaire was distributed to all the collaborators on February 9 2010 in order to

conduct a survey in each country

Measures

PSM was measured using the newly developed 33-item index 7 items for the dimension of

APP 13 items for CPV (4 for CPV1 and 9 for CPV2) 6 items for COM and 7 for SS In the

survey questionnaire a five-point Likert type scale (1 = strongly disagree 5 = strongly agree)

was used This scale is among the most widely used response formats in many empirical

investigations within the social and behavioral science in which there are no ―correct or

―incorrect responses (Spector 1992) The items of PSM were randomly distributed in the

survey format Gender age education length of service and employment status were also

measured in this survey

It was necessary to translate the measurement items into the official (native) language

- 10 -

of each country As this raised the possibility of inaccurate translations the back-translation

of the items into native languages was validated by asking two researchers to independently

translate the English-language items The researchers were instructed to consult with each

other over differences between translated items and to test the survey items with some native

respondents

[Table 1 about here]

Samples

The international survey was fielded from March to September 2010 The data upon which

this study are based were collected from civil servants in local governments in Australia

Belgium China Denmark France Italy Korea Lithuania the Netherlands Switzerland the

United Kingdom and the USA The principal criterion was that the samples collected in the

various countries should be comparable to the largest possible extent The criterion ruled out

national and state government as the competencies of national governments and also of state

governments if they exist constitutionally are different among most of the countries included

in the research Therefore we opted for local government as our focus The structure and

functions of local governments can differ widely even within countries and so the survey in

each country was focused on town hall bureaucrats in city county or township government

The respondents were permanent employees from both managerial and non-managerial levels

and from both administrative departments (ie no direct contact with the public) and service

departments (ie likely to have direct contact with the public) Despite being employed by

local governments in many or all of the countries police officers firefighters school teachers

public transport workers artists and musicians as well as nurses and doctors were excluded

- 11 -

from the sample

We aimed to receive 250 respondents for each country as this would enable us to

carry out the analyses we wanted to perform on the data Although there is no ideal sample

size one study recommends a sample of 200 to provide a sound basis for estimation (Hair

Black Babin and Anderson 2010) The method of data collection was left open Obviously a

web-survey is the easiest way to collect the data because it is economical to collect and to

code however compared to paper-pencil surveys this method has some drawbacks First not

everybody in the sample may have access to the internet which may cause a bias in the

responses Second internet surveys are generally known to produce lower response rates

Web surveys were recommended if possible but paper-pencil surveys were deemed

acceptable in cases where it was not possible to conduct a web survey Researchers were

encouraged to contact just one municipality except in cases where it was not possible to

gather enough data In such instances responses from multiple municipalities were accepted

By the end of September 2010 a total of 2868 responses were obtained2

2 In Australia seven local councils each with between from 70 and 232 employees were surveyed Paper-pencil

surveys were used in two councils while the rest relied on a web survey The local governments limited

distribution of paper-pencil surveys to only one or two sections of the council eg administration The response

rate was approximately 30 However it should be noted that not everyone working for the council has direct

access to a computer In Belgium three municipalities were surveyed on a census base by means of a web

survey as all employees had access to a computer This rendered a response of 120 (226) 79 (75) and 16

(53) responses for the respective municipalities resulting in an overall response rate of 331 with 662

invitations being sent out In China two departments social work and administration were surveyed in one

municipality using paper-pencil surveys In the social work department 65 out of 110 surveys were returned

and 165 out of 211 responded from the administration department giving an overall response rate of 767 In

Denmark a total of 1282 respondents in a municipality were sent an initial e-mail invitation to a web survey

and up to two e-mail reminders and the response rate was 473 In France all the employees in two

municipalities were surveyed In one municipality where both web and paper-pencil surveys were utilized 128

out of 685 questionnaires were returned while in another city using only a paper-pencil survey 138 out of 652

were received The average response rate was 199 In Italy in principle the link to the questionnaire was

made available to all employees (about 4000) in one city but for some of them access may have been difficult

yielding a response rate of 45 In Korea 500 employees in a city received a printed questionnaire and 253

responses were returned yielding a response rate of 506 In Lithuania 236 responses from 54 municipalities

were gathered by web survey In the Netherlands three municipalities were surveyed on a census base by means

of a web survey In total 319 responses of 910 invitations were returned (35) mdash 90 (45) 115 (38) and 114

(28) for the respective municipalities In Switzerland 516 out of 1326 local employees with e-mail access in

a municipality completed the survey giving a response rate of 389 In the United Kingdom the survey was

posted online and 1360 employees both in administrative jobs and front-line services of one municipality

- 12 -

Municipalities were selected because of their convenient accessibility and proximity to the

researcher and a convenience sample was used in most countries Seven countries surveyed

one municipality but seven local councils were surveyed in Australia Web surveys were used

in four countries while three countries used both web and paper-pencil survey The number

of respondents varied among the countries depending on the numbers of questionnaires

distributed the survey method and the response rate In most countries we used

approximately 250 responses but only 162 from Italy3 The majority of the respondents were

women (566) while in terms of service length those in the largest group (393) had

worked for fewer than 10 years Table 2 shows the distribution of respondentsrsquo sex length of

service employment status and country

[Table 2 about here]

Analyses

The statistical analysis applied confirmatory factor analysis (CFA) CFA was used to assess

the fit of the data to the hypothesized measurement model because it is preferred where

measurement models have a well-developed underlying theory for hypothesized patterns of

loadings The data were analyzed using Amos 180 (Arbuckle 2009) with the maximum

likelihood (MLE) estimation method and LISREL 872 (Joumlreskog and Soumlrbom 2005) with the

diagonally weighted least squares (D-WLS) estimation method

The individual items are designed to measure a theoretically continuous construct

but the observed responses are a discrete realization of five categories The common method

received an email to complete the survey The response rate was 19 In the USA the total number of

employees surveyed was 626 in one city using both electronic and paper-pencil surveys resulting in a response

rate of 588 3 In countries where significantly more than 250 surveys were collected we randomly selected only 250 cases

for analysis

- 13 -

of estimation within CFA is MLE a technique which assumes that the observed variables are

continuous and normally distributed These assumptions are not met when the observed data

are discrete as occurs when using ordinal scales as in this study A simulation study (Flora

and Curran 2004) found that D-WLS performs well for ordinal measures across a variety of

conditions especially for smaller samples Coursey and Pandey (2007a 2007b) argued the

importance of using D-WLS and not standard MLE in structural equations for ordinal data4

In most cases however the differences between MLE and D-WLS are not substantive to the

conclusions5 No comparison between standard MLE and D-WLS is provided in this paper

We give results from both MLE and D-WLS and this will aid in confirming the stability and

robustness of the results across varying mathematical specifications

For gauging reliability Cronbachrsquos alpha and composite reliability were tested

Composite reliability should be equal to or greater than 07 (Hair et al 2010) Convergent

validity was assessed from the measurement model by determining the statistical significance

of each indicatorrsquos estimated pattern coefficient on its specified underlying construct

Discriminant validity was assessed by determining whether the confidence interval around

the correlation estimate between two factors included 100 (Anderson and Gerbing 1988)

For model fit assessment goodness-of-fit index (GFI) comparative fit index (CFI)

and the root mean square error of approximation (RMSEA) were used The model achieves

an acceptable fit to the data when GFI and CFI equals or exceeds 090 and RMSEA values

4 Applying standard MLE to ordinal data such as Likert-type items may produce significant estimation

problems such as inflation of chi-square fit statistics and biased underestimation of parameters and standard

errors (Flora and Curran 2004) Coursey and Pandey (2007a 2007b) used a modified MLE using the asymptotic

covariance matrix and D-WLS for adjusting the ordinal nature of the data They reported that although the

modified MLE findings are slightly different on a few model parameters the differences are not substantive to

the conclusions However model fit estimators were clearly better for the D-WLS 5 Given normally distributed categorical variables MLE can be used with confidence when a variable has four

or more categories If there is a large sample a five-point Likert scale and the skewness and kurtosis of variables

is within acceptable limits it suggests distribution symmetry ―In most cases where the hypothesized model is

well specified and the scaling based on more than three categories it seems unlikely that there will be much

difference between the findings (Byrne 2010 160)

- 14 -

fall below 008 (Hair et al 2010) In general the larger the value of GFI and CFI and the

smaller the value of RMSEA the better the fit of the model (Bollen 1989) We also present

the χ2 test statistic Lower values of χ

2 indicate a better fit and should be nonsignificant but

the χ2 value increases as sample size increases and thus for large sample sizes this statistic

may lead to rejection of a model with good fit (Hair et al 2010)

Results

Using SPSS 180 descriptive statistics were computed for individual items as shown in table

1 One item (APP3) in the APP dimension ―I like to discuss topics regarding public programs

and policies with others was mistakenly omitted in the survey conducted in the United

States and so it was deleted in the statistical analysis In the assessment of normality no

items showed a skew or kurtosis value greater than the cutoffs of |3| or |8| recommended by

Kline (2005) As data were distributed normally MLE as well as D-WLS estimation was used

Based on inspection of item-total correlations one item (COM4) was dropped from further

analysis because it was weakly correlated with the overall index (011) Missing data is an

issue in CFA as it can cause estimation errors All variables in the survey have missing data

rates of approximately 016 to 143 Mean substitution was used to calculate replacement

values for the missing values in each countryrsquos dataset (Hair et al 2010) The deletion of the

two items left 31 items for the next stage of analysis

The five-correlated factor model was tested using CFA The CFA model in the initial

analysis hypothesized a priori that (1) responses to the 30-item PSM measurement

instrument could be explained by five factors such as APP CPV1 CPV2 COM SS (2) each

item would have a nonzero loading on the PSM factor that it was designed to measure and

- 15 -

zero loadings on all other factors (3) the five factors would be correlated and (4)

measurement error terms would be uncorrelated The resulting CFA with MLE suggested that

the initial five-correlated factor model was not a good fit to the data χsup2 (df = 424) = 43699 p

lt 001 GFI = 0899 CFI = 835 RMSEA = 0057 [055 059] At this stage APP was highly

correlated with CPV1 (r = 996 in MLE and r = 988 in D-WLS) suggesting the two

dimensions were not unique and lacked discriminant validity

[Table 3 about here]

Given the disconfirmation of the initial model an effort was made to estimate an

alternative model The two dimensions APP and CPV1 were combined to form a single

dimension Attraction to Public Service (APS) The items of CPV1 may be regarded as

action-oriented by the respondents because dedication to public service community and

common good is underlined in these items Consequently it is reasonable to combine APP

with CPV1 to form APS a dimension that focuses more on disposition to serve the public to

work for the common good and to participate in public policy processes After the overall

number of dimensions was reduced to four using standardized factor loadings and

modification indices as criteria the items with the lowest factor loading in each dimension

were deleted and the items whose modification index indicated association with multiple

dimensions were eliminated A CFA model with the remaining items was iteratively tested

until it achieved a good fit to the data The outcome of this refinement process was a 16-item

index of four factors Overall model fit by every measure is quite strong for the D-WLS

estimation technique The estimation results for the four-dimension model are presented in

table 4

- 16 -

[Table 4 about here]

Cronbachrsquos coefficient alpha for the 16-item PSM index was 860 and the coefficient

alphas for the four dimensions ranged from 645 to 784 The composite reliability of the set

of reflective indicators for each dimension of PSM ranged from 713 to 826 and indicated

adequate internal consistency The resulting factor structure showed a four-factor structure

with all items loading significantly onto their a priori dimension (p lt 001) and the

standardized factor loadings ranging from 0490 to 0766 in MLE and from 515 to 807 in

D-WLS The results provide support for convergent validity The correlation estimates

between the two factors ranged from 0462 to 0848 in MLE and from 505 to 859 in D-

WLS The confidence intervals (plusmn2 standard errors) around the correlation estimates between

the two factors did not include 100 The result provides support for discriminant validity To

test whether the four-dimension model is superior to other designs such as the two or three-

dimension models a series of CFA were conducted and the results show that the four-

dimension model is a significantly better fit than any of the other designs

To test whether the factorial structure of the PSM instrument is equivalent across the

samples from the twelve countries four nested models were tested as part of multi-group

analysis with MLE (Byrne 2010) First the unconstrained model was the baseline model

which relaxed all equality constraints This model hypothesized the same four-factor structure

of the instrument across twelve countries The fit indices for this baseline model were χsup2 (df =

1176) = 24977 p lt 001 GFI = 0901 CFI = 900 RMSEA = 0020 [019 021] These

indices indicate that the four-factor model of the PSM instrument exhibits acceptable fit

across the samples from the twelve countries Second the measurement weights model tested

- 17 -

the invariance of factor loadings across countries by placing equality constraints on these

parameters The fit indices for this model were χsup2 (df = 1308) = 28088 p lt 001 GFI = 0890

CFI = 887 RMSEA = 0020 [019 021] Since the measurement weights model was nested

within the unconstrained model the chi-square difference text ∆χsup2 (132) = 3111 p lt 001

indicated that some equality constraints of factor loadings did not hold across the twelve

countries Further test results suggested that the equality constraints of structural covariances

and measurement residuals were not upheld across the twelve countries It needs to be

stressed that despite the differences in factor loadings and factor covariances the results

support the main assumption that the same four-factor structure holds across different

countries We may assume that people in different cultures understand PSM in the same

fashion

To find differences in the factor structure of PSM among countries we conducted a

series of CFA with each countryrsquos dataset and followed the same procedure as when

analyzing the sample as a whole As table 5 shows although there are great similarities

between the appropriate items in the countries there are also some differences Even though

the items vary according to national differences the four-factor structure had a good fit to the

data Thus the results confirm that the four-factor structure of PSM can be generalized

[Table 5 about here]

The APS dimension covers both an individualrsquos attraction to public participation and

commitment to public interest but all its items refer to some notion of public interest such as

improving public service aiding community tackling social problems meaningful public

service and contributing to the common good The APS dimension captures an individualrsquos

- 18 -

disposition to serve the public interest through public participation When developing the

items some assumed that the items of committing to public interests can be separated from

those of pursuing public values in the CPV dimension The empirical test showed that

respondents tended to understand committing to public interests as behavior-oriented and so

the items for public interest are not differentiated from participation-oriented items in their

perceptions but are different from those for public values The respondents seem to

understand that pursuing public values is more related to personal beliefs or creeds Thus the

CPV dimension is solely focused on public values and the items of pursing public interests

are combined with those of the APP dimension to make the APS dimension

Public values and their relative importance are neither impervious to change over

time nor the same among all countries (Charles de Jong and Ryan 2011 Davis and West

2009) Moreover some values may be in conflict as Martinsen and Beck Joslashrgensen (2010)

discussed in regard to the tensions between accountability and efficiency It is therefore not

possible to include all public values in one dimension (Beck Joslashrgensen and Bozeman 2007)

Accordingly the items of the CPV dimension are developed for estimating individual

commitment to seek public values which there are general consensus about and which are

generally formulated

When all four dimensions of PSM are included in a survey cross-national

equivalence of the survey can be assured The items represent reflections or manifestations of

each dimension and hence are relatively homogeneous and interchangeable Adding or

deleting items does not change the essential nature of the dimension Thus any combination

of items in a dimension can be regarded as equivalent When using structural equation

modeling (SEM) each dimension needs to have at least three reflective indicators in order to

achieve identification or it needs to have at least two reflective indicators and be allowed to

- 19 -

covary with another dimension that has at least two reflective measures With one indicator

the dimension is identified if the loading and the variance of the uniqueness are both fixed

(Bollen 1989 Edwards 2010) Thus the number of items in each dimension is not restricted

when measuring PSM However we recommend using multiple items for each dimension

because more research is needed to investigate the appropriateness of this PSM instrument

with varying samples

The test results showed that model fit estimators were clearly better for the D-WLS

It means that in this study MLE was more rigorous than D-WLS in assessing the fit of the

model The empirical evidence indicated that MLE may be an acceptable technique for

ordinal data when there is a large sample a five-point Likert scale and the skewness and

kurtosis of variables is within acceptable limits as suggested by Byrne (2010)

Discussion

The analysis reaffirmed the four-dimensional structure of PSM but the results were not

entirely consistent with the original CFA we tested We were ultimately able to converge on a

good-fit operational model for PSM that was also consistent with underlying theory The

analysis is not able to resolve the debate about measurement models however for reasons

discussed below

Choosing the correct measurement model (ie reflective vs formative) is important

as measurement model misspecification results not only in misrepresentation of the epistemic

relationships between constructs and their measures but also in biased parameter estimates of

structural relationships (Jarvis et al 2003 Law and Wong 1999 MacKenzie 2003

MacKenzie Podsakoff and Jarvis 2005) A formative measurement model in isolation is

- 20 -

under-identified and cannot be estimated in the case of covariance-based SEM (CB-SEM)

(Bollen 1989) A construct with formative indicators must emit paths to at least two unrelated

reflective indicators two unrelated latent constructs or some combination of the two in order

to solve the problem of under-identification (Jarvis MacKenzie and Podsakoff 2003

MacCallum and Browne 1993 MacKenzie Podsakoff and Jarvis 2005) Thus ―when

dependent constructs change the empirical nature of the formatively measured construct

changes (Wilcox Howell and Breivik 2008 1223) and ―when specifying formative

measurement models what might seem like an innocuous choice of outcomes to achieve

identification can have fundamental ramifications for the meaning of the formative construct

and the inferences based on it (Edwards 2010 7)

Earlier research found that the association between a construct and its measures

should remain stable regardless of the larger causal model in which the constructs and

measures are embedded (Edwards and Bagozzi 2000) The association between a construct

and its measures is generally stable for reflective measures but associations of formative

measures with their construct are unstable The susceptibility of formative measurement to

both interpretational confounding and external inconsistency recommends that formative

measurement is not an equally attractive alternative to reflective measurement and that

whenever possible researchers should opt for reflective measurement (Howell Breivik and

Wilcox 2007a 2007b Kim Shin and Grover 2010) In the case of second-order models

researchers are limited to only a second-order reflective model in CB-SEM (Chin 2010)

Regardless of the measurement debate and the tradeoffs between reflective or

formative specification this study makes a theoretical argument that PSM is a second-order

formative construct This second-order formative factor model (reflective first-order

formative second-order) is also suggested as an alternative model that captures the

- 21 -

multidimensionality of complex concepts and avoids the shortcomings of first-order

formative measurement (Edwards 2010) However this model itself cannot achieve

identification One possible strategy is to develop direct reflective measures of PSM based on

the view that PSM is a general concept that is not only influenced by its specific dimensions

but also can be assessed using general measures (Edwards 2010 Javis MacKenzie and

Podsakoff 2003)6 However these general measures have not yet been developed for the

PSM construct

Another way around the problem of underidentification of formative models is to

apply the component-based approach known as partial least squares path modeling (PLS-PM)

which can model formative indicators (Fornell and Bookstein 1982 Goumltz Liehr-Gobbers

and Krafft 2010) The CB-SEM and the PLS-PM methods have distinctive statistical

characteristics and selecting an approach to SEM depends on the particular research situation

CB-SEM is the method of choice for theory testing while PLS-PM is appropriate for

prediction-oriented applications (Ringle Goumltz Wetzels and Wilson 2009) One advantage of

the PLS method is that it demands significantly fewer requirements compared to that of

covariance structure analyses but nevertheless delivers consistent estimation results Another

asset of the PLS approach is its ability to deal with formative as well as reflective indicators

even within one structural equation model This indicates that the PLS approach can be

appropriate for explorative analysis of structural equation models (Chin 2010 Goumltz Liehr-

Gobbers and Krafft 2010) A Monte Carlo simulation study to compare CB-SEM and PLS-

PM showed that PLS-PM provides a viable approximation of model parameters when the

prerequisites for CB-SEM are not met (Ringle Goumltz Wetzels and Wilson 2009) A formative

6 For example job satisfaction can be assessed with general reflective measures such as ―in general I am

satisfied with my job and the formative dimensions can be assessed with measures that describe satisfaction

with each job facet (Edwards 2010 15)

- 22 -

composite in the form of indexes needs to be tested to know the relative weights (γ-

parameters) of the dimensions of PSM In PLS-PM a second-order formative factor is

directly measured by observed variables for all the first order factors While this approach

repeats the number of manifest variables used the model can be identified without adding

outcome constructs that would be considered inappropriate or irrelevant from a conceptual

standpoint (Chin 2010 Edwards 2010) The test result of PLS-PM using Smart PLS 20

(Ringle Wende and Will 2005) produced the function PSM = 31APS + 36CPV + 28COM

+ 31SS However one consideration is whether and how construct error is modeled for the

formative construct (Diamantopoulos 2006) When using PLS-PM the researcher is making

the assumption that there is no error at the construct level as all constructs in PLS-PM are

modeled without error When using PLS-PM the absence of an error term will tend to

increase the weights therefore inflating their importance (Bollen and Lennox 1991) The

primary statistic for assessing a formative indicator is its weight the partialized effect of the

indicator on its intended construct controlling for the effects of all other indicators of that

construct (Cenfetelli and Bassellier 2009) A formative measure is essentially a multiple

regression with the construct representing the dependent variable and the indicators as the

predictors (Diamantopoulos and Winklhofer 2001) PSM may be regarded as a weighted

linear composite of its dimensions The weights used to create the composites are different

across CE-SEM and PLS-PM The causal dimensional weights generated in CB-SEM are

estimated in a model in which a construct-level error term is specified (eg PSM = γ1APM +

γ2CPI + γ3COM + γ4SS + ζ) whereas the weights in PLS-PM are not (eg PSM = γ1APM +

γ2CPI + γ3COM + γ4SS) This leads to two different conceptualizations of the PSM construct

one in which the causal indicators do not capture the construct in its entirety and another in

which they do (Hardin et al 2010) These differences need to be discussed in future research

- 23 -

in this area

One practical suggestion to measure PSM is to apply unit weights to its dimensions if

we assume the effects of the four dimensions are equivalent A literature review indicates that

unit weights have substantial predictive validity when compared with regression weights The

literature shows that the empirical correlations between scores generated from differential and

unit-weighting approaches are quite high In creating composite scores unit weights can be

an appropriate approach for weighting under many circumstances (Bobko Roth and Buster

2007) Although the uses of unit-weight composites will likely result in the explanation of

less variance than empirical estimations of dimensional weights they will remain consistently

defined across studies and hence suitable for theory testing Alternatively more thorough

discussions on the dimensions of PSM may suggest fixed-weight composites (Hardin et al

2010) A previous study (Kim 2010 22) also suggested ―Overall indexing PSM by

measuring all the dimensions of PSM and simply summing the averages of each dimension

can make it easier to measure PSM in practice as PSM = APS + CPV + COM + SS

An advantage of regarding PSM as a second-order formative construct is that the

dimensions can be analyzed both individually and collectively This is important as the

dimensions are expected to have different antecedents and consequences For instance Perry

(1997) found some significant differences in the influence of independent variables on the

four dimensions of PSM Taylor (2007) confirmed that certain dimensions are more important

than others in influencing work outcomes Moynihan and Pandey (2007a) reported that red

tape is negatively related to one dimension but reform orientation and hierarchical authority

is positively related to another dimension A recent study found that the SS dimension has a

positive relationship with occupational intention but the other dimensions do not (Liu et al

2011) Thus it is valuable to analyze the causes and effects of individual dimensions of PSM

- 24 -

as well as of the PSM construct

Conclusion

The purpose of this study is to develop a measurement instrument for PSM for cross-national

research and comparison On the basis of the sharpened construct of PSM (Kim and

Vandenabeele 2010) using data from an international survey conducted in 12 countries we

found that the four-factor structure of PSM is generalizable across country cases This study

provides the measurement index of PSM with the four dimensions of APS CPV COM and

SS The APS dimension measures the instrumental motives of PSM the CPV dimension

determines the value-based motives of PSM while the COM and SS dimensions gauge the

identification motives and self-sacrifice respectively When all four dimensions of PSM are

used in a survey cross-national equivalence of survey items can be assured The items for

measuring PSM in future research are suggested in Appendix 4

This study has several limitations First looking for samples which were comparable

to the largest possible extent we opted for surveys in local government The results found

and discussed in this study are based on PSM among local government employees The

results might be different when surveying central government employees or other specific

groups in the public sector Thus the potential impact of the sample bias needs to be

considered in understanding this study It also means that the survey items for measuring

PSM should be examined with different samples in various areas and at different government

levels

Second the test results showed that there is a high degree of correlation among the

dimension such as APS CPV and COM Future research should not only investigate whether

- 25 -

the items reflect the core content of the dimensions but also continue to develop more

appropriate items for better discriminant validity of each dimension Third the content of

APP dimension was assumed to be distinct from the CPV dimension that focuses on personal

disposition to pursue public interest and public values However the public interest sub-

dimension was separated from the public values sub-dimension in the CPV dimension and

combined with the APP dimension to form the APS dimension Further research should

analyze the CPV dimension and investigate the appropriateness of the APS dimension

In order to facilitate research internationally the PSM construct needs to be

improved conceptually and operationally Kim and Vandenabeele (2010) have dealt with the

conceptual composition and operational dimension of PSM as well as the measurement

model of PSM The next step is to develop an internationally robust measurement instrument

of PSM This study proposes that scholars use all four of the dimensions in measuring the

conceptual components of PSM such as self-sacrifice instrumental value-based and affective

motives because they collectively represent PSM It provides a set of appropriate items for

measuring PSM and confirms the dimensional structure of PSM The survey results suggest

that people in different cultures understand PSM in the same fashion We know that this study

is simply a step along the way to developing a more robust measure but we hope it

contributes to enhanced cross-national research and comparison along with the generation of

cumulative knowledge internationally

- 26 -

REFERENCES

Anderson James C and David D Gerbing 1988 Structural Equation Modeling in Practice

A Review and Recommended Two-Step Approach Psychological Bulletin 103

411-423

Arbuckle James L 2009 Amos 180 Userrsquos Guide Crawfordville FL Amos Development

Co

Beck Joslashrgensen Torben and Barry Bozeman 2007 Public Values An Inventory

Administration amp Society 39(3) 354-381

Beutel Ann M and Margaret Mooney Marini 1995 Gender and values American

Sociological Review 60 436 - 448

Bobko Philip Philip L Roth and Maury A Buster 2007 The Usefulness of Unit Weights in

Creating Composite Scores A Literature Review Application to Content Validity and

Meta-Analysis Organizational Research Methods 10(4) 689-709

Bollen Kenneth A 1989 Structural Equations with Latent Variables New York Wiley

Bollen K A and R Lennox 1991 Conventional Wisdom on Measurement A Structural

Equation Perspective Psychological Bulletin 110(2) 305-314

Byrne Barbara M 2010 Structural Equation Modeling with Amos Basic Concepts

Applications and Programming 2nd

ed New York Routledge

Camilleri Emanuel 2006 Towards Developing an Organizational Commitment ― Public

Service Motivation Model for Maltese Public Service Employees Public Policy

and Administration 21 63-83

Castaing Seacutebastien 2006 The Effects of Psychological Contract Fulfillment and Public

Service Motivation on Organizational Commitment in the French Civil Service

Public Policy and Administration 21 84-98

Cenfetelli Ronald T and Geneviegraveve Bassellier 2009 Interpretation of Formative

Measurement in Information Systems Research MIS Quarterly 33(4) 689-707

Charles Michael B W Martin de Jong and Neal Ryan 2011 Public Values in Western

- 27 -

Europe A Temporal Perspective American Review of Public Administration 41

75-91

Chin Wynne W 2010 How to Write Up and Report PLS Analysis In Handbook of Partical

Least Squares edited by V Esposito Vinzi et al 655-690 Berlin Springer-Verlag

Coursey David H and Sanjay K Pandey 2007a Public Service Motivation Measurement

Testing an Abridged Version of Perryrsquos Proposed Scale Administration amp Society

39 547-568

Coursey David H and Sanjay K Pandey 2007b Content Domain Measurement and

Validity of the Red Tape Concept American Review of Public Administration 37

342-361

Coursey David H James L Perry Jeffrey L Brudney and Laura Littlepage 2008

Psychometric Verification of Perryrsquos Public Service Motivation Instrument Results

for Volunteer Exemplars Review of Public Personnel Administration 28 79-90

Davis Paul and Karen West 2009 What Do Public Values Mean for Public Action Putting

Public Values in Their Plural Place American Review of Public Administration 39

602-618

DeHart-Davis Leisha Justin Marlowe and Sanjay K Pandey 2006 Gender Dimensions of

Public Service Motivation Public Administration Review 66 873-887

Diamantopoulos Adamantios 2006 The error term in formative measurement models

Interpretation and modeling implications Journal of Modeling in Management 1(1)

7-17

Diamantopoulos Adamantios and Judy A Siguaw 2006 Formative versus Reflective

Indicators in Organizational Measure Development A Comparison and Empirical

Illustration British Journal of Management 17 263-282

Diamantopoulos Adamantios and Heidi M Winklhofer 2001 Index Construction with

Formative Indicators An Alternative to Scale Development Journal of Marketing

Research 38 269-277

Edwards Jeffrey R 2010 The Fallacy of Formative Measurement Organizational Research

- 28 -

Methods Online First published on August 19 2010 as

doi1011771094428110378369

Edwards Jeffrey R and Richard P Bagozzi 2000 On the Nature and Direction of

Relationships between Constructs and Measures Psychological Methods 5 155-

174

Flora David B and Patrick J Curran 2004 An Empirical Evaluation of Alternative

Methods of Estimation for Confirmatory Factor Analysis with Ordinal Data

Psychological Methods 9 466-491

Fornell Claes and Fred L Bookstein 1982 Two structural equations models LISREL and

PLS applied to customer exit-voice theory Journal of Marketing Research 19 440-

452

Giauque David Adrian Ritz Freacutedeacuteric Varone Simon Anderfuhren-Biget and Christian

Waldner 2011 Putting Public Service Motivation into Context A Balance between

Universalism and Particularism Forthcoming in International Review of

Administrative Sciences 77(2)

Goumltz Oliver Kerstin Liehr-Gobbers and Manfred Krafft 2010 Evaluation of Structural

Equation Models Using the Partial Least Squares (PLS) Approach In V Esposito

Vinzi et al (eds) Handbook of Partial Least Squares 691-711 DOI 101007978-3-

540-32827-8_30 Springer-Verlag Berlin Heidelberg

Guinot Christiane Julie Latreille and Michel Tenenhaus 2001 PLS Path Modelling and

Multiple Table Analysis Application to the Cosmetic Habits of Women in Ile-de-

France Chemometrics and Intelligent Laboratory Systems 58 247-259

Hair Joseph F Jr William C Black Barry J Babin and Rolph E Anderson 2010

Multivariate Data Analysis 7th

ed Upper Saddle River NJ Pearson Education

Hardin Andrew M Jerry Cha-Jan Chang Mark A Fuller and Gholamreza Torkzadeh 2010

Formative Measurement and Academic Research In Search of Measurement

Theory Educational and Psychological Measurement Online First published on

June 11 2010 as doi101177001316440370208

- 29 -

Howell Roy D Einar Breivik and James B Wilcox 2007a Reconsidering Formative

Measurement Psychological Methods 12 205-218

Howell Roy D Einar Breivik and James B Wilcox 2007b Is Formative Measurement

Really Measurement Reply to Bollen (2007) and Bagozzi (2007) Psychological

Methods 12 238-245

Jarvis Cheryl Burke Scott B MacKenzie and Philip M Podsakoff 2003 A Critical Review

of Construct Indicators and Measurement Model Misspecification in Marketing and

Consumer Research Journal of Consumer Research 30 199-218

Joumlreskog Karl G and Dag Soumlrbom 2005 LISREL 872 (Computer Software) Lincolnwood

IL Scientific Software International

Kim Gimun Bongsik Shin and Varun Grover 2010 Investigating Two Contradictory Views

of Formative Measurement in Information Systems Research MIS Quarterly 34

345-365

Kim Sangmook 2009a Revising Perryrsquos Measurement Scale of Public Service Motivation

American Review of Public Administration 39 149-163

Kim Sangmook 2009b Testing the Structure of Public Service Motivation in Korea A

Research Note Journal of Public Administration Research and Theory 19 839-

851

Kim Sangmook 2010 Testing a Revised Measurement of Public Service Motivation

Reflective versus Formative Specification Journal of Public Administration

Research and Theory Doi101093jopartmuq048

Kim Sangmook and Wouter Vandenabeele 2010 A strategy for building public service

motivation research internationally Public Administration Review 70 701-709

Kline Rex B 2005 Principles and Practice of Structural Equation Modeling 2nd

ed New

York Guilford Press

Law Kenneth and Chi-Sum Wong 1999 Multidimensional Constructs in Structural

Equation Analysis An Illustration using the Job Perception and Job Satisfaction

Constructs Journal of Management 25 143-160

- 30 -

Leisink Peter and Bram Steijn 2009 Public Service Motivation and Job Performance of

Public Sector Employees in the Netherlands International Review of

Administrative Sciences 75 35-52

Liu Bangcheng Chun Hui Jin Hu Wensheng Yang and Xinli Yu 2011 How well can

Public Service Motivation Connect with Occupational Intention International

Review of Administrative Sciences 77 191-211

Liu Bangcheng Ningyu Tang and Xiaomei Zhu 2008 Public Service Motivation and Job

Satisfaction in China An Investigation of Generalisability and Instrumentality

International Journal of Manpower 29 684-699

MacCallum Robert C and Michael W Browne 1993 The Use of Causal Indicators in

Covariance Structure Models Some Practical Issues Psychological Bulletin 114

533-541

MacKenzie Scott B 2003 The dangers of poor construct conceptualization Journal of the

Academy of Marketing Science 31(3) 323-326

MacKenzie Scott B Philip M Podsakoff and Cheryl Burke Jarvis 2005 The Problem of

Measurement Model Misspecification in Behavioral and Organizational Research

and Some Recommended Solutions Journal of Applied Psychology 90 710-730

Martinsen Dorte Sindbjerg and Torben Beck Joslashrgensen 2010 Accountability as a

Differentiated Value in Supranational Governance American Review of Public

Administration 40 742-760

Moynihan Donald P and Sanjay K Pandey 2007a The Role of Organizations in Fostering

Public Service Motivation Public Administration Review 67 40-53

Moynihan Donald P and Sanjay K Pandey 2007b Finding Workable Levers over Work

Motivation Comparing Job Satisfaction Job Involvement and Organizational

Commitment Administration amp Society 39 803-832

Pandey Sanjay K and Edmund C Stazyk 2008 Antecedents and Correlates of Public

Service Motivation In Motivation in Public Management edited by James L Perry

and Annie Hondeghem 101-117 Oxford Oxford University Press

- 31 -

Perry James L 1996 Measuring Public Service Motivation An Assessment of Construct

Reliability and Validity Journal of Public Administration Research and Theory 6

5-22

Perry James L 1997 Antecedents of Public Service Motivation Journal of Public

Administration Research and Theory 7 181-197

Perry James L 2000 Bringing Society in Toward a Theory of Public Service Motivation

Journal of Public Administration Research and Theory 10 471-488

Perry James L and Annie Hondeghem 2008a Motivation in Public Management Oxford

Oxford University Press

Perry James L and Annie Hondeghem 2008b Directions for Future Theory and Research

In Motivation in Public Management edited by James L Perry and Annie

Hondeghem 294-313 Oxford Oxford University Press

Perry James L Annie Hondeghem and Lois Recascino Wise 2010 Revisiting the

Motivational Bases of Public Service Twenty Years of Research and an Agenda for

the Future Public Administration Review 70(5) 681-690

Perry James L and Wouter Vandenabeele 2008 Behavioral dynamics Institutions

Identities and Self-Regulation In Motivation in Public Management edited by

James L Perry and Annie Hondeghem 56-79 Oxford Oxford University Press

Perry James L and Lois Recascino Wise 1990 The Motivational Bases of Public Service

Public Administration Review 50 367-373

Rainey Hal G 1982 Reward Preference among Public and Private Managers In Search of

the Service Ethic American Review of Public Administration 16 288-302

Ringle Christian M Oliver Goumltz Martin Wetzels and Bradley Wilson 2009 On the Use of

Formative Measurement Specifications in Structural Equation Modeling A Monte

Carlo Simulation Study to Compare Covariance-Based and Partial Least Squares

Model Estimation Methodologies Faculty of Economics and Business Administration

Maastricht Univeristy

Ringle C M S Wende and S Will 2005 Smart PLS 20 (M3) Beta

- 32 -

httpwwwsmartplsde

Ritz Adrian 2011 Attraction to Public Policy Making A Qualitative Inquiry into

Improvements in PSM Measurement Forthcoming in Public Administration

Spector Paul E 1992 Summated Rating Scales Construction An Introduction Newbury

Park CA Sage

Steenbergen Marco R 1996 Compassion and American public opinion An analysis of the

NES Humanitarianism Scale Report to the Board of Overseers of the National

Election Studies

Steffen Patrick R and Kevin S Masters 2005 Does compassion mediate the intrinsic

religion-health relationship Annals of Behavioral Medicine 30 217-223

Taylor Jeannette 2007 The Impact of Public Service Motives on Work Outcomes in

Australia A Comparative Multi-Dimensional Analysis Public Administration 85

931-959

Vandenabeele Wouter 2007 Towards a Theory of Public Service Motivation An

Institutional Approach Public Management Review 9 545-556

Vandenabeele Wouter 2008a Government Calling Public Service Motivation as an Element

in Selecting Government as an Employer of Choice Public Administration 86

1089-1105

Vandenabeele Wouter 2008b Development of a Public Service Motivation Measurement

Scale Corroborating and Extending Perryrsquos Measurement Instrument International

Public Management Journal 11 143-167

Vandenabeele Wouter 2009 The Mediating Effect of Job Satisfaction and Organizational

Commitment on Self-Reported Performance More Robust Evidence of the PSM-

Performance Relationship International Review of Administrative Sciences 75 11-

34

Wilcox James B Roy D Howell and Einar Breivik 2008 Questions about Formative

Measurement Journal of Business Research 61 1219-1228

Wise Lois Recascino 2000 The Public Service Culture In Public Administration Concepts

- 33 -

and Cases 7th ed edited by Richard J Stillman II 342-353 Boston Houghton

Mifflin

Wright Bradley E 2008 Methodological Challenges associated with Public Service

Motivation Research In Motivation in Public Management edited by James L

Perry and Annie Hondeghem 80-98 Oxford Oxford University Press

Wright Bradley E and Sanjay K Pandey 2008 Public Service Motivation and the

Assumption of Person-Organization Fit Testing the Mediating Effect of Value

Congruence Administration amp Society 40 502-521

- 34 -

Table 1

Possible Indicators of PSM and Descriptive Statistics (n = 2868)

Dimensions and items Mean Standard

Deviation

Item-Total

Correlation

Attraction to Public Participation (APP)

APP1 I am interested in helping to improve public service

APP2 I am satisfied when I see people benefiting from the public programs I was

involved in

APP3 I like to discuss topics regarding public programs and policies with others

APP4 I believe that public sector activities contribute to our general welfare

APP5 I admire people who initiate or are involved in activities to aid my

community

APP6 Contributing to public programs and policies helps me realize myself

APP7 It is important to contribute to activities that tackle social problems

412

441

355

393

423

351

409

696

662

917

753

672

881

734

524

421

482

412

493

478

552

Commitment to Public Values (CPV)

Sub-dimension for Public Interests (CPV1)

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

CPI3 I would prefer seeing public officials do what is best for the whole

community even if it harmed my interests

CPI4 Serving the public interest is more important than helping a single individual

Sub-dimension for Public Values (CPV2)

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of public

services

CPV3 It is fundamental that public services respond to the needs of the citizens

CPV4 Decisions regarding public services should be democratic despite the time

and effort it takes

CPV5 Everybody is entitled to a good service even if it costs a lot of money

CPV6 It is fundamental that the interests of future generations are taken into

account when developing public policies

CPV7 To act ethically is essential for public servants

CPV8 I believe that public employees must always be aware of the legitimacy of

their activities

CPV9 I personally identify with the aim of protecting individual liberties and rights

418

409

348

337

428

416

433

394

367

426

441

434

414

695

662

899

995

730

704

710

852

932

689

681

692

763

525

596

480

251

484

437

356

325

376

439

462

402

392

Compassion (COM) COM1 It is difficult for me to contain my feelings when I see people in distress

COM2 I feel sympathetic to the plight of the underprivileged

COM3 I empathize with other people who face difficulties

COM4 I have little compassion for people in need who are unwilling to take the

first step to help themselves

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

354

387

388

299

412

391

951

825

707

1083

755

765

399

511

389

011

487

502

Self-Sacrifice (SS)

SS1 Making a difference to society means more to me than personal achievements

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS5 People should give back to society more than they get from it

SS6 Serving other citizens would give me a good feeling even if no one paid me for

it

SS7 I would agree to a good plan to make a better life for the poor even if it costs

me money

357

329

339

302

345

363

350

904

872

949

982

888

1049

895

508

584

454

531

412

368

546

- 35 -

Table 2

Background of Respondents (n = 2868)

Variables Characteristics Respondents

Sex

Male

Female

NA

1209

1622

37

422

566

13

Length of service

(years)

0 ~ 10

10 ~ 20

20 ~ 30

30+

NA

1126

723

484

491

44

393

252

169

171

15

Organizational status

―Do you supervise employees

No

Yes

NA

1472

1076

320

513

375

112

Country

Australia

Belgium

China

Denmark

France

Italy

Korea

Lithuania

The Netherlands

Switzerland

United Kingdom

USA

249

214

230

249

266

162

253

236

249

250

260

250

87

75

80

87

93

56

88

82

87

87

91

87 Note NA = no answer

- 36 -

Table 3

Standardized factor loadings and correlations for the 31-item PSM index (n = 2868)

Dimensions and items SFL (MLE) SFL (D-WLS) Alpha

APP

APP1

APP2

APP4

APP5

APP6

APP7

597

486

455

558

490

600

657

552

514

625

549

665

699

CPV1

CPI1

CPI2

CPI3

CPI4

588

660

488

289

653

732

535

294

542

CPV2

CPV1

CPV2

CPV3

CPV4

CPV5

CPV6

CPV7

CPV8

CPV9

611

546

436

393

418

534

598

496

460

690

631

505

457

486

601

669

573

530

745

COM

COM1

COM2

COM3

COM5

COM6

496

569

498

597

607

516

651

538

670

684

687

SS

SS1

SS2

SS3

SS4

SS5

SS6

SS7

567

736

558

728

467

409

663

660

762

591

695

515

494

709

782

Measures of fit (MLE) χsup2 (df = 424) = 43699 p lt 001 CFI = 835 GFI = 899

RMSEA = 057 [055 059]

Measure of fit (D-WLS) χsup2 (df = 424) = 39247 p lt 001 CFI = 971 GFI = 976

RMSEA = 054 [052 055]

Inter-factor correlations MLE (D-WLS)

1 2 3 4

1 APS

2 CPV1 996 (988)

3 CPV2 902 (890) 798 (798)

4 COM 787 (783) 758 (762) 779 (783)

5 SS 700 (742) 800 (859) 509 (537) 720 (729)

Note SFL = standardized factor loading All factor loadings and correlations are significant at p lt 001

- 37 -

Table 4

Standardized factor loadings and correlations for the 16-item PSM index (n = 2868)

Dimensions and items SFL

(MLE)

SFL

(D-WLS)

APS

APP5 I admire people who initiate or are involved in activities to aid my

community

APP7 It is important to contribute to activities that tackle social problems

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

611

592

637

666

665

669

690

751

CPV

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of

public services

CPV6 It is fundamental that the interests of future generations are taken

into account when developing public policies

CPV7 To act ethically is essential for public servants

612

546

551

590

698

631

618

658

COM

COM1 It is difficult for me to contain my feelings when I see people in

distress

COM2 I feel sympathetic to the plight of the underprivileged

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

490

546

596

604

515

608

669

679

SS

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS7 I would agree to a good plan to make a better life for the poor even if

it costs me money

763

571

766

678

807

629

747

755

Measures of fit (MLE)

χsup2 (df = 98) = 8490 p lt 001

CFI = 939 GFI = 965

RMSEA = 052 [049 055]

Measure of fit (D-WLS)

χsup2 (df = 98) = 6069 p lt 001

CFI = 988 GFI = 992

RMSEA = 043 [039 046]

Alpha CR (D-WLS) Inter-factor correlations MLE (D-WLS)

1 APS 717 788 1 APS 2 CPV 3 COM

2 CPV 663 747 848 (859)

3 COM 645 713 781 (791) 789 (795)

4 SS 784 826 565 (596) 462 (505) 715 (726)

Note SFL = standardized factor loading Alpha = Cronbachrsquos alpha CR = composite reliability

All factor loadings and correlations are significant at p lt 001

- 38 -

Table 5

Results of CFA in the individual countries Standardized factor loadings and measures of fit

Factors

amp Items

Australia Belgium China Denmark France Italy

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

672

756

799

785

804

943

710

703

597

790

799

730

584

675

685

681

731

742

578

748

674

643

765

739

645

704

678

708

812

779

632

604

765

766

672

853

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

552

593

560

699

727

632

624

605

560

713

709

645

702

602

616

724

686

735

567

560

715

654

605

811

608

658

587

696

755

698

673

620

523

782

697

654

COM

COM1

COM2

COM3

COM5

COM6

623

584

662

688

701

739

732

608

475

808

666

581

578

616

605

649

670

673

699

751

737

679

816

776

568

546

801

626

673

850

637

546

741

715

601

822

SS

SS2

SS3

SS4

SS7

654

794

806

892

767

813

727

693

799

793

725

838

857

820

773

896

855

842

806

749

792

829

753

884

847

731

753

889

816

753

815

692

664

795

782

734

Inter-factor

correlations

093

~

752

163

~

810

320

~

773

323

~

825

452

~

827

497

~

855

323

~

850

321

~

847

498

~

899

536

~

902

425

~

829

476

~

865

Sample size 249 214 230 249 266 162

Measures of Fit

χsup2 638 444 881 732 765 554 817 625 735 565 750 564

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 980 100 939 987 969 997 967 994 975 997 949 995

GFI 961 986 941 986 948 991 949 992 955 992 926 986

RMSEA 036 00 063 050 051 026 053 035 045 026 059 033

- 39 -

Table 5 (Continued)

Factors

amp Items

Korea Lithuania The Netherlands Switzerland UK USA

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

619

701

667

706

781

717

666

512

682

751

605

779

557

629

630

620

719

729

535

621

702

643

694

735

655

725

738

710

804

822

535

621

702

643

694

735

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

594

607

722

722

644

722

674

535

533

790

589

711

568

554

538

655

628

545

627

580

615

711

666

704

800

512

555

873

581

646

627

580

615

711

666

704

COM

COM1

COM2

COM3

COM5

COM6

694

677

550

763

736

627

557

639

428

550

696

534

737

623

508

766

616

608

659

592

683

703

661

769

682

583

729

739

644

803

659

592

683

703

661

769

SS

SS2

SS3

SS4

SS7

776

752

679

779

770

811

707

721

635

725

734

706

814

656

743

820

767

764

745

632

689

788

662

744

671

818

832

830

864

802

745

632

689

788

662

744

Inter-factor

correlations

432

~

726

466

~

746

376

~

788

406

~

902

377

~

852

455

~

836

637

~

818

660

~

828

317

~

852

360

~

852

637

~

818

660

~

828

Sample size 253 236 249 250 260 250

Measures of Fit

χsup2 896 573 612 462 773 616 815 715 769 542 815 715

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 951 996 975 100 958 993 957 991 972 998 957 991

GFI 942 988 960 989 951 987 948 990 955 990 948 990

RMSEA 059 028 034 00 050 034 053 044 048 022 053 044

Note All factor loadings and correlations are significant at p lt 001

- 40 -

Appendix 1 Dimensions and possible items of PSM (the first round)

1 Attraction to Public Participation

(1) I am interested in developing public programs helping my country or community (Kim 2009a)

(2) Sharing my views on public policies with others is attractive to me (Kim 2009a)

(3) I am satisfied when seeing people get benefits from the public programs I was involved in (Kim 2009a)

(4) I like to discuss political subjects with others (Giauque et al 2011)

(5) I think the governmental activities contribute to our welfare

(6) I like to initiate actions to help out my community

(7) I do not care much about politicians (Perry 1996)

2 Commitment to Public Values

Possible items for public interest

(1A) Meaningful public service is very important to me (Perry 1996)

(2A) It is important for me to contribute for the common good (Giauque et al 2011)

(3A) I recognize myself with the mission of protection of liberties and rights endorsed by the nation-state

(Giauque et al 2011)

(4A) I would prefer seeing public officials do what is best for the whole community even if it harmed my

interests (Perry 1996)

(5A) To me serving the public interest is more important than helping other people (Vandenabeele 2008b)

Possible items for public values

(1B) I recognize myself with the promotion of the equality of chances endorsed by the nation-state (Giauque et

al 2011)

(2B) It is fundamental that the public service provided to the citizen is regular and continuous (Giauque et al

2011)

(3B) It is fundamental that public service innovates to respond to the new needs of the citizensclients (Giauque

et al 2011)

(4B) The dignity and well-being of all should be the most important concerns in any society (Steenbergen

1996)

(5B) It is important that public servants account for all the costsexpenses they make (Vandenabeele 2008b)

(6B) We have to do everything in our power to pursue the goal of democracy

3 Compassion

(1) It is difficult for me to contain my feelings when I see people in distress (Perry 1996)

(2) I feel sympathetic to the plight of the underprivileged (Kim 2009a)

(3) I am often reminded by daily events how dependent we are on one another (Perry 1996)

(4) I feel compassion for others in the difficulties they are facing (Steffen and Masters 2005)

(5) It is not really my problem if others are in trouble and need help (Beutel and Marini 1995)

(6) I find it hard to be sympathetic toward starving people in foreign lands when there is so much trouble in our

own country (Beutel and Marini 1995)

(7) I get very upset when I see other people treated unfairly (Beutel and Marini 1995)

(8) I would agree to a good plan to make a better life for the poor even if it cost me money (Beutel and Marini

1995)

(9) To me patriotism includes seeing the welfare of others (Perry 1996)

(10) I care very much about other people

4 Self-Sacrifice

(1) Making a difference in society means more to me than personal achievements (Perry 1996)

(2) I am prepared to make enormous sacrifices for the good of society (Perry 1996)

(3) I believe in putting duty before self (Perry 1996)

(4) I am one of those rare people who would risk personal loss to help someone else (Perry 1996)

(5) I think people should give back to society more than they get from it (Perry 1996)

(6) Serving other citizens would give me a good feeling even if no one paid me for it (Perry 1996)

(7) Much of what I do is for a cause bigger than myself (Perry 1996)

- 41 -

Appendix 2 Newly suggested items through the first round

1 Items for the Attraction to Public Participation Dimension

(1) In the public domain every person should be counted as equal

(2) I enjoy to think and argue in a political way

(3) I am satisfied when seeing my ideas being integrated in public policies

(4) I can easily relate to the way of thinking and arguing of politicians

(5) Working in the public sector I can better realize myself

2 Items for the Public Values Sub-Dimension

(1) Public service should give value for money

(2) Cost considerations should play an important role in decisions about public service provision

(3) It is fundamental that the interests of future generations are taken into account when developing public

policies

(4) To act with correct ethics is for public servants as important as their professional competence

(5) Primarily public servants should be accountable to the public and not to their superiors

3 Items for the Self-Sacrifice Dimension

(1) I would agree to a good plan to make a better life for the poor even if it costs me money

- 42 -

Appendix 3 An example of a summary of comments and suggestions (the second round)

On the item ―I am interested in developing public programs helping my country or community

A It is a good sentence but is it possible for everyone to ―develop public programs Even public officials are

limited to do this

B OK

C or the special group

D Double barrel and difficult to translate to my country ―I am interested in helping to improve public services

E Good but I feel that wording requires revision If we draw sample from local government then country is

redundant ―I am interested in developing public policy programs that aim in helping my community

F 4 (General appreciation from1 to 4) Not clear whether it refers to ones work or to ones concern as a citizen

―(In my work) I am interested in contributing to the development of public etchellip

G Instead of development I would use implementing

H Delete public which is redundant because of the characteristic helping my country or community but add

or implementing ―I am interested in developing or implementing programs helping my country or

community

I Referring to dimension Wording Similar to item 6

Note Names have been changed to alphabets in order to ensure the anonymity of participants

- 43 -

Appendix 4 Dimensions and Items of Public Service Motivation for Future Research

Dimensions and items

Attraction to Public Service (APS)

1 I am interested in helping to improve public service

2 I admire people who initiate or are involved in activities to aid my community

3 It is important to contribute to activities that tackle social problems

4 Meaningful public service is very important to me

5 It is important for me to contribute to the common good

Commitment to Public Values (CPV)

1 I think equal opportunities for citizens are very important

2 It is important that citizens can rely on the continuous provision of public services

3 It is fundamental that the interests of future generations are taken into account when

developing public policies

4 To act ethically is essential for public servants

5 I believe that public employees must always be aware of the legitimacy of their activities

6 I personally identify with the aim of protecting individual liberties and rights

Compassion (COM)

1 It is difficult for me to contain my feelings when I see people in distress

2 I feel sympathetic to the plight of the underprivileged

3 I empathize with other people who face difficulties

4 I get very upset when I see other people being treated unfairly

5 Considering the welfare of others is very important

Self-Sacrifice (SS)

1 I am prepared to make sacrifices for the good of society

2 I believe in putting civic duty before self

3 I am willing to risk personal loss to help society

4 I would agree to a good plan to make a better life for the poor even if it costs me money

- 10 -

of each country As this raised the possibility of inaccurate translations the back-translation

of the items into native languages was validated by asking two researchers to independently

translate the English-language items The researchers were instructed to consult with each

other over differences between translated items and to test the survey items with some native

respondents

[Table 1 about here]

Samples

The international survey was fielded from March to September 2010 The data upon which

this study are based were collected from civil servants in local governments in Australia

Belgium China Denmark France Italy Korea Lithuania the Netherlands Switzerland the

United Kingdom and the USA The principal criterion was that the samples collected in the

various countries should be comparable to the largest possible extent The criterion ruled out

national and state government as the competencies of national governments and also of state

governments if they exist constitutionally are different among most of the countries included

in the research Therefore we opted for local government as our focus The structure and

functions of local governments can differ widely even within countries and so the survey in

each country was focused on town hall bureaucrats in city county or township government

The respondents were permanent employees from both managerial and non-managerial levels

and from both administrative departments (ie no direct contact with the public) and service

departments (ie likely to have direct contact with the public) Despite being employed by

local governments in many or all of the countries police officers firefighters school teachers

public transport workers artists and musicians as well as nurses and doctors were excluded

- 11 -

from the sample

We aimed to receive 250 respondents for each country as this would enable us to

carry out the analyses we wanted to perform on the data Although there is no ideal sample

size one study recommends a sample of 200 to provide a sound basis for estimation (Hair

Black Babin and Anderson 2010) The method of data collection was left open Obviously a

web-survey is the easiest way to collect the data because it is economical to collect and to

code however compared to paper-pencil surveys this method has some drawbacks First not

everybody in the sample may have access to the internet which may cause a bias in the

responses Second internet surveys are generally known to produce lower response rates

Web surveys were recommended if possible but paper-pencil surveys were deemed

acceptable in cases where it was not possible to conduct a web survey Researchers were

encouraged to contact just one municipality except in cases where it was not possible to

gather enough data In such instances responses from multiple municipalities were accepted

By the end of September 2010 a total of 2868 responses were obtained2

2 In Australia seven local councils each with between from 70 and 232 employees were surveyed Paper-pencil

surveys were used in two councils while the rest relied on a web survey The local governments limited

distribution of paper-pencil surveys to only one or two sections of the council eg administration The response

rate was approximately 30 However it should be noted that not everyone working for the council has direct

access to a computer In Belgium three municipalities were surveyed on a census base by means of a web

survey as all employees had access to a computer This rendered a response of 120 (226) 79 (75) and 16

(53) responses for the respective municipalities resulting in an overall response rate of 331 with 662

invitations being sent out In China two departments social work and administration were surveyed in one

municipality using paper-pencil surveys In the social work department 65 out of 110 surveys were returned

and 165 out of 211 responded from the administration department giving an overall response rate of 767 In

Denmark a total of 1282 respondents in a municipality were sent an initial e-mail invitation to a web survey

and up to two e-mail reminders and the response rate was 473 In France all the employees in two

municipalities were surveyed In one municipality where both web and paper-pencil surveys were utilized 128

out of 685 questionnaires were returned while in another city using only a paper-pencil survey 138 out of 652

were received The average response rate was 199 In Italy in principle the link to the questionnaire was

made available to all employees (about 4000) in one city but for some of them access may have been difficult

yielding a response rate of 45 In Korea 500 employees in a city received a printed questionnaire and 253

responses were returned yielding a response rate of 506 In Lithuania 236 responses from 54 municipalities

were gathered by web survey In the Netherlands three municipalities were surveyed on a census base by means

of a web survey In total 319 responses of 910 invitations were returned (35) mdash 90 (45) 115 (38) and 114

(28) for the respective municipalities In Switzerland 516 out of 1326 local employees with e-mail access in

a municipality completed the survey giving a response rate of 389 In the United Kingdom the survey was

posted online and 1360 employees both in administrative jobs and front-line services of one municipality

- 12 -

Municipalities were selected because of their convenient accessibility and proximity to the

researcher and a convenience sample was used in most countries Seven countries surveyed

one municipality but seven local councils were surveyed in Australia Web surveys were used

in four countries while three countries used both web and paper-pencil survey The number

of respondents varied among the countries depending on the numbers of questionnaires

distributed the survey method and the response rate In most countries we used

approximately 250 responses but only 162 from Italy3 The majority of the respondents were

women (566) while in terms of service length those in the largest group (393) had

worked for fewer than 10 years Table 2 shows the distribution of respondentsrsquo sex length of

service employment status and country

[Table 2 about here]

Analyses

The statistical analysis applied confirmatory factor analysis (CFA) CFA was used to assess

the fit of the data to the hypothesized measurement model because it is preferred where

measurement models have a well-developed underlying theory for hypothesized patterns of

loadings The data were analyzed using Amos 180 (Arbuckle 2009) with the maximum

likelihood (MLE) estimation method and LISREL 872 (Joumlreskog and Soumlrbom 2005) with the

diagonally weighted least squares (D-WLS) estimation method

The individual items are designed to measure a theoretically continuous construct

but the observed responses are a discrete realization of five categories The common method

received an email to complete the survey The response rate was 19 In the USA the total number of

employees surveyed was 626 in one city using both electronic and paper-pencil surveys resulting in a response

rate of 588 3 In countries where significantly more than 250 surveys were collected we randomly selected only 250 cases

for analysis

- 13 -

of estimation within CFA is MLE a technique which assumes that the observed variables are

continuous and normally distributed These assumptions are not met when the observed data

are discrete as occurs when using ordinal scales as in this study A simulation study (Flora

and Curran 2004) found that D-WLS performs well for ordinal measures across a variety of

conditions especially for smaller samples Coursey and Pandey (2007a 2007b) argued the

importance of using D-WLS and not standard MLE in structural equations for ordinal data4

In most cases however the differences between MLE and D-WLS are not substantive to the

conclusions5 No comparison between standard MLE and D-WLS is provided in this paper

We give results from both MLE and D-WLS and this will aid in confirming the stability and

robustness of the results across varying mathematical specifications

For gauging reliability Cronbachrsquos alpha and composite reliability were tested

Composite reliability should be equal to or greater than 07 (Hair et al 2010) Convergent

validity was assessed from the measurement model by determining the statistical significance

of each indicatorrsquos estimated pattern coefficient on its specified underlying construct

Discriminant validity was assessed by determining whether the confidence interval around

the correlation estimate between two factors included 100 (Anderson and Gerbing 1988)

For model fit assessment goodness-of-fit index (GFI) comparative fit index (CFI)

and the root mean square error of approximation (RMSEA) were used The model achieves

an acceptable fit to the data when GFI and CFI equals or exceeds 090 and RMSEA values

4 Applying standard MLE to ordinal data such as Likert-type items may produce significant estimation

problems such as inflation of chi-square fit statistics and biased underestimation of parameters and standard

errors (Flora and Curran 2004) Coursey and Pandey (2007a 2007b) used a modified MLE using the asymptotic

covariance matrix and D-WLS for adjusting the ordinal nature of the data They reported that although the

modified MLE findings are slightly different on a few model parameters the differences are not substantive to

the conclusions However model fit estimators were clearly better for the D-WLS 5 Given normally distributed categorical variables MLE can be used with confidence when a variable has four

or more categories If there is a large sample a five-point Likert scale and the skewness and kurtosis of variables

is within acceptable limits it suggests distribution symmetry ―In most cases where the hypothesized model is

well specified and the scaling based on more than three categories it seems unlikely that there will be much

difference between the findings (Byrne 2010 160)

- 14 -

fall below 008 (Hair et al 2010) In general the larger the value of GFI and CFI and the

smaller the value of RMSEA the better the fit of the model (Bollen 1989) We also present

the χ2 test statistic Lower values of χ

2 indicate a better fit and should be nonsignificant but

the χ2 value increases as sample size increases and thus for large sample sizes this statistic

may lead to rejection of a model with good fit (Hair et al 2010)

Results

Using SPSS 180 descriptive statistics were computed for individual items as shown in table

1 One item (APP3) in the APP dimension ―I like to discuss topics regarding public programs

and policies with others was mistakenly omitted in the survey conducted in the United

States and so it was deleted in the statistical analysis In the assessment of normality no

items showed a skew or kurtosis value greater than the cutoffs of |3| or |8| recommended by

Kline (2005) As data were distributed normally MLE as well as D-WLS estimation was used

Based on inspection of item-total correlations one item (COM4) was dropped from further

analysis because it was weakly correlated with the overall index (011) Missing data is an

issue in CFA as it can cause estimation errors All variables in the survey have missing data

rates of approximately 016 to 143 Mean substitution was used to calculate replacement

values for the missing values in each countryrsquos dataset (Hair et al 2010) The deletion of the

two items left 31 items for the next stage of analysis

The five-correlated factor model was tested using CFA The CFA model in the initial

analysis hypothesized a priori that (1) responses to the 30-item PSM measurement

instrument could be explained by five factors such as APP CPV1 CPV2 COM SS (2) each

item would have a nonzero loading on the PSM factor that it was designed to measure and

- 15 -

zero loadings on all other factors (3) the five factors would be correlated and (4)

measurement error terms would be uncorrelated The resulting CFA with MLE suggested that

the initial five-correlated factor model was not a good fit to the data χsup2 (df = 424) = 43699 p

lt 001 GFI = 0899 CFI = 835 RMSEA = 0057 [055 059] At this stage APP was highly

correlated with CPV1 (r = 996 in MLE and r = 988 in D-WLS) suggesting the two

dimensions were not unique and lacked discriminant validity

[Table 3 about here]

Given the disconfirmation of the initial model an effort was made to estimate an

alternative model The two dimensions APP and CPV1 were combined to form a single

dimension Attraction to Public Service (APS) The items of CPV1 may be regarded as

action-oriented by the respondents because dedication to public service community and

common good is underlined in these items Consequently it is reasonable to combine APP

with CPV1 to form APS a dimension that focuses more on disposition to serve the public to

work for the common good and to participate in public policy processes After the overall

number of dimensions was reduced to four using standardized factor loadings and

modification indices as criteria the items with the lowest factor loading in each dimension

were deleted and the items whose modification index indicated association with multiple

dimensions were eliminated A CFA model with the remaining items was iteratively tested

until it achieved a good fit to the data The outcome of this refinement process was a 16-item

index of four factors Overall model fit by every measure is quite strong for the D-WLS

estimation technique The estimation results for the four-dimension model are presented in

table 4

- 16 -

[Table 4 about here]

Cronbachrsquos coefficient alpha for the 16-item PSM index was 860 and the coefficient

alphas for the four dimensions ranged from 645 to 784 The composite reliability of the set

of reflective indicators for each dimension of PSM ranged from 713 to 826 and indicated

adequate internal consistency The resulting factor structure showed a four-factor structure

with all items loading significantly onto their a priori dimension (p lt 001) and the

standardized factor loadings ranging from 0490 to 0766 in MLE and from 515 to 807 in

D-WLS The results provide support for convergent validity The correlation estimates

between the two factors ranged from 0462 to 0848 in MLE and from 505 to 859 in D-

WLS The confidence intervals (plusmn2 standard errors) around the correlation estimates between

the two factors did not include 100 The result provides support for discriminant validity To

test whether the four-dimension model is superior to other designs such as the two or three-

dimension models a series of CFA were conducted and the results show that the four-

dimension model is a significantly better fit than any of the other designs

To test whether the factorial structure of the PSM instrument is equivalent across the

samples from the twelve countries four nested models were tested as part of multi-group

analysis with MLE (Byrne 2010) First the unconstrained model was the baseline model

which relaxed all equality constraints This model hypothesized the same four-factor structure

of the instrument across twelve countries The fit indices for this baseline model were χsup2 (df =

1176) = 24977 p lt 001 GFI = 0901 CFI = 900 RMSEA = 0020 [019 021] These

indices indicate that the four-factor model of the PSM instrument exhibits acceptable fit

across the samples from the twelve countries Second the measurement weights model tested

- 17 -

the invariance of factor loadings across countries by placing equality constraints on these

parameters The fit indices for this model were χsup2 (df = 1308) = 28088 p lt 001 GFI = 0890

CFI = 887 RMSEA = 0020 [019 021] Since the measurement weights model was nested

within the unconstrained model the chi-square difference text ∆χsup2 (132) = 3111 p lt 001

indicated that some equality constraints of factor loadings did not hold across the twelve

countries Further test results suggested that the equality constraints of structural covariances

and measurement residuals were not upheld across the twelve countries It needs to be

stressed that despite the differences in factor loadings and factor covariances the results

support the main assumption that the same four-factor structure holds across different

countries We may assume that people in different cultures understand PSM in the same

fashion

To find differences in the factor structure of PSM among countries we conducted a

series of CFA with each countryrsquos dataset and followed the same procedure as when

analyzing the sample as a whole As table 5 shows although there are great similarities

between the appropriate items in the countries there are also some differences Even though

the items vary according to national differences the four-factor structure had a good fit to the

data Thus the results confirm that the four-factor structure of PSM can be generalized

[Table 5 about here]

The APS dimension covers both an individualrsquos attraction to public participation and

commitment to public interest but all its items refer to some notion of public interest such as

improving public service aiding community tackling social problems meaningful public

service and contributing to the common good The APS dimension captures an individualrsquos

- 18 -

disposition to serve the public interest through public participation When developing the

items some assumed that the items of committing to public interests can be separated from

those of pursuing public values in the CPV dimension The empirical test showed that

respondents tended to understand committing to public interests as behavior-oriented and so

the items for public interest are not differentiated from participation-oriented items in their

perceptions but are different from those for public values The respondents seem to

understand that pursuing public values is more related to personal beliefs or creeds Thus the

CPV dimension is solely focused on public values and the items of pursing public interests

are combined with those of the APP dimension to make the APS dimension

Public values and their relative importance are neither impervious to change over

time nor the same among all countries (Charles de Jong and Ryan 2011 Davis and West

2009) Moreover some values may be in conflict as Martinsen and Beck Joslashrgensen (2010)

discussed in regard to the tensions between accountability and efficiency It is therefore not

possible to include all public values in one dimension (Beck Joslashrgensen and Bozeman 2007)

Accordingly the items of the CPV dimension are developed for estimating individual

commitment to seek public values which there are general consensus about and which are

generally formulated

When all four dimensions of PSM are included in a survey cross-national

equivalence of the survey can be assured The items represent reflections or manifestations of

each dimension and hence are relatively homogeneous and interchangeable Adding or

deleting items does not change the essential nature of the dimension Thus any combination

of items in a dimension can be regarded as equivalent When using structural equation

modeling (SEM) each dimension needs to have at least three reflective indicators in order to

achieve identification or it needs to have at least two reflective indicators and be allowed to

- 19 -

covary with another dimension that has at least two reflective measures With one indicator

the dimension is identified if the loading and the variance of the uniqueness are both fixed

(Bollen 1989 Edwards 2010) Thus the number of items in each dimension is not restricted

when measuring PSM However we recommend using multiple items for each dimension

because more research is needed to investigate the appropriateness of this PSM instrument

with varying samples

The test results showed that model fit estimators were clearly better for the D-WLS

It means that in this study MLE was more rigorous than D-WLS in assessing the fit of the

model The empirical evidence indicated that MLE may be an acceptable technique for

ordinal data when there is a large sample a five-point Likert scale and the skewness and

kurtosis of variables is within acceptable limits as suggested by Byrne (2010)

Discussion

The analysis reaffirmed the four-dimensional structure of PSM but the results were not

entirely consistent with the original CFA we tested We were ultimately able to converge on a

good-fit operational model for PSM that was also consistent with underlying theory The

analysis is not able to resolve the debate about measurement models however for reasons

discussed below

Choosing the correct measurement model (ie reflective vs formative) is important

as measurement model misspecification results not only in misrepresentation of the epistemic

relationships between constructs and their measures but also in biased parameter estimates of

structural relationships (Jarvis et al 2003 Law and Wong 1999 MacKenzie 2003

MacKenzie Podsakoff and Jarvis 2005) A formative measurement model in isolation is

- 20 -

under-identified and cannot be estimated in the case of covariance-based SEM (CB-SEM)

(Bollen 1989) A construct with formative indicators must emit paths to at least two unrelated

reflective indicators two unrelated latent constructs or some combination of the two in order

to solve the problem of under-identification (Jarvis MacKenzie and Podsakoff 2003

MacCallum and Browne 1993 MacKenzie Podsakoff and Jarvis 2005) Thus ―when

dependent constructs change the empirical nature of the formatively measured construct

changes (Wilcox Howell and Breivik 2008 1223) and ―when specifying formative

measurement models what might seem like an innocuous choice of outcomes to achieve

identification can have fundamental ramifications for the meaning of the formative construct

and the inferences based on it (Edwards 2010 7)

Earlier research found that the association between a construct and its measures

should remain stable regardless of the larger causal model in which the constructs and

measures are embedded (Edwards and Bagozzi 2000) The association between a construct

and its measures is generally stable for reflective measures but associations of formative

measures with their construct are unstable The susceptibility of formative measurement to

both interpretational confounding and external inconsistency recommends that formative

measurement is not an equally attractive alternative to reflective measurement and that

whenever possible researchers should opt for reflective measurement (Howell Breivik and

Wilcox 2007a 2007b Kim Shin and Grover 2010) In the case of second-order models

researchers are limited to only a second-order reflective model in CB-SEM (Chin 2010)

Regardless of the measurement debate and the tradeoffs between reflective or

formative specification this study makes a theoretical argument that PSM is a second-order

formative construct This second-order formative factor model (reflective first-order

formative second-order) is also suggested as an alternative model that captures the

- 21 -

multidimensionality of complex concepts and avoids the shortcomings of first-order

formative measurement (Edwards 2010) However this model itself cannot achieve

identification One possible strategy is to develop direct reflective measures of PSM based on

the view that PSM is a general concept that is not only influenced by its specific dimensions

but also can be assessed using general measures (Edwards 2010 Javis MacKenzie and

Podsakoff 2003)6 However these general measures have not yet been developed for the

PSM construct

Another way around the problem of underidentification of formative models is to

apply the component-based approach known as partial least squares path modeling (PLS-PM)

which can model formative indicators (Fornell and Bookstein 1982 Goumltz Liehr-Gobbers

and Krafft 2010) The CB-SEM and the PLS-PM methods have distinctive statistical

characteristics and selecting an approach to SEM depends on the particular research situation

CB-SEM is the method of choice for theory testing while PLS-PM is appropriate for

prediction-oriented applications (Ringle Goumltz Wetzels and Wilson 2009) One advantage of

the PLS method is that it demands significantly fewer requirements compared to that of

covariance structure analyses but nevertheless delivers consistent estimation results Another

asset of the PLS approach is its ability to deal with formative as well as reflective indicators

even within one structural equation model This indicates that the PLS approach can be

appropriate for explorative analysis of structural equation models (Chin 2010 Goumltz Liehr-

Gobbers and Krafft 2010) A Monte Carlo simulation study to compare CB-SEM and PLS-

PM showed that PLS-PM provides a viable approximation of model parameters when the

prerequisites for CB-SEM are not met (Ringle Goumltz Wetzels and Wilson 2009) A formative

6 For example job satisfaction can be assessed with general reflective measures such as ―in general I am

satisfied with my job and the formative dimensions can be assessed with measures that describe satisfaction

with each job facet (Edwards 2010 15)

- 22 -

composite in the form of indexes needs to be tested to know the relative weights (γ-

parameters) of the dimensions of PSM In PLS-PM a second-order formative factor is

directly measured by observed variables for all the first order factors While this approach

repeats the number of manifest variables used the model can be identified without adding

outcome constructs that would be considered inappropriate or irrelevant from a conceptual

standpoint (Chin 2010 Edwards 2010) The test result of PLS-PM using Smart PLS 20

(Ringle Wende and Will 2005) produced the function PSM = 31APS + 36CPV + 28COM

+ 31SS However one consideration is whether and how construct error is modeled for the

formative construct (Diamantopoulos 2006) When using PLS-PM the researcher is making

the assumption that there is no error at the construct level as all constructs in PLS-PM are

modeled without error When using PLS-PM the absence of an error term will tend to

increase the weights therefore inflating their importance (Bollen and Lennox 1991) The

primary statistic for assessing a formative indicator is its weight the partialized effect of the

indicator on its intended construct controlling for the effects of all other indicators of that

construct (Cenfetelli and Bassellier 2009) A formative measure is essentially a multiple

regression with the construct representing the dependent variable and the indicators as the

predictors (Diamantopoulos and Winklhofer 2001) PSM may be regarded as a weighted

linear composite of its dimensions The weights used to create the composites are different

across CE-SEM and PLS-PM The causal dimensional weights generated in CB-SEM are

estimated in a model in which a construct-level error term is specified (eg PSM = γ1APM +

γ2CPI + γ3COM + γ4SS + ζ) whereas the weights in PLS-PM are not (eg PSM = γ1APM +

γ2CPI + γ3COM + γ4SS) This leads to two different conceptualizations of the PSM construct

one in which the causal indicators do not capture the construct in its entirety and another in

which they do (Hardin et al 2010) These differences need to be discussed in future research

- 23 -

in this area

One practical suggestion to measure PSM is to apply unit weights to its dimensions if

we assume the effects of the four dimensions are equivalent A literature review indicates that

unit weights have substantial predictive validity when compared with regression weights The

literature shows that the empirical correlations between scores generated from differential and

unit-weighting approaches are quite high In creating composite scores unit weights can be

an appropriate approach for weighting under many circumstances (Bobko Roth and Buster

2007) Although the uses of unit-weight composites will likely result in the explanation of

less variance than empirical estimations of dimensional weights they will remain consistently

defined across studies and hence suitable for theory testing Alternatively more thorough

discussions on the dimensions of PSM may suggest fixed-weight composites (Hardin et al

2010) A previous study (Kim 2010 22) also suggested ―Overall indexing PSM by

measuring all the dimensions of PSM and simply summing the averages of each dimension

can make it easier to measure PSM in practice as PSM = APS + CPV + COM + SS

An advantage of regarding PSM as a second-order formative construct is that the

dimensions can be analyzed both individually and collectively This is important as the

dimensions are expected to have different antecedents and consequences For instance Perry

(1997) found some significant differences in the influence of independent variables on the

four dimensions of PSM Taylor (2007) confirmed that certain dimensions are more important

than others in influencing work outcomes Moynihan and Pandey (2007a) reported that red

tape is negatively related to one dimension but reform orientation and hierarchical authority

is positively related to another dimension A recent study found that the SS dimension has a

positive relationship with occupational intention but the other dimensions do not (Liu et al

2011) Thus it is valuable to analyze the causes and effects of individual dimensions of PSM

- 24 -

as well as of the PSM construct

Conclusion

The purpose of this study is to develop a measurement instrument for PSM for cross-national

research and comparison On the basis of the sharpened construct of PSM (Kim and

Vandenabeele 2010) using data from an international survey conducted in 12 countries we

found that the four-factor structure of PSM is generalizable across country cases This study

provides the measurement index of PSM with the four dimensions of APS CPV COM and

SS The APS dimension measures the instrumental motives of PSM the CPV dimension

determines the value-based motives of PSM while the COM and SS dimensions gauge the

identification motives and self-sacrifice respectively When all four dimensions of PSM are

used in a survey cross-national equivalence of survey items can be assured The items for

measuring PSM in future research are suggested in Appendix 4

This study has several limitations First looking for samples which were comparable

to the largest possible extent we opted for surveys in local government The results found

and discussed in this study are based on PSM among local government employees The

results might be different when surveying central government employees or other specific

groups in the public sector Thus the potential impact of the sample bias needs to be

considered in understanding this study It also means that the survey items for measuring

PSM should be examined with different samples in various areas and at different government

levels

Second the test results showed that there is a high degree of correlation among the

dimension such as APS CPV and COM Future research should not only investigate whether

- 25 -

the items reflect the core content of the dimensions but also continue to develop more

appropriate items for better discriminant validity of each dimension Third the content of

APP dimension was assumed to be distinct from the CPV dimension that focuses on personal

disposition to pursue public interest and public values However the public interest sub-

dimension was separated from the public values sub-dimension in the CPV dimension and

combined with the APP dimension to form the APS dimension Further research should

analyze the CPV dimension and investigate the appropriateness of the APS dimension

In order to facilitate research internationally the PSM construct needs to be

improved conceptually and operationally Kim and Vandenabeele (2010) have dealt with the

conceptual composition and operational dimension of PSM as well as the measurement

model of PSM The next step is to develop an internationally robust measurement instrument

of PSM This study proposes that scholars use all four of the dimensions in measuring the

conceptual components of PSM such as self-sacrifice instrumental value-based and affective

motives because they collectively represent PSM It provides a set of appropriate items for

measuring PSM and confirms the dimensional structure of PSM The survey results suggest

that people in different cultures understand PSM in the same fashion We know that this study

is simply a step along the way to developing a more robust measure but we hope it

contributes to enhanced cross-national research and comparison along with the generation of

cumulative knowledge internationally

- 26 -

REFERENCES

Anderson James C and David D Gerbing 1988 Structural Equation Modeling in Practice

A Review and Recommended Two-Step Approach Psychological Bulletin 103

411-423

Arbuckle James L 2009 Amos 180 Userrsquos Guide Crawfordville FL Amos Development

Co

Beck Joslashrgensen Torben and Barry Bozeman 2007 Public Values An Inventory

Administration amp Society 39(3) 354-381

Beutel Ann M and Margaret Mooney Marini 1995 Gender and values American

Sociological Review 60 436 - 448

Bobko Philip Philip L Roth and Maury A Buster 2007 The Usefulness of Unit Weights in

Creating Composite Scores A Literature Review Application to Content Validity and

Meta-Analysis Organizational Research Methods 10(4) 689-709

Bollen Kenneth A 1989 Structural Equations with Latent Variables New York Wiley

Bollen K A and R Lennox 1991 Conventional Wisdom on Measurement A Structural

Equation Perspective Psychological Bulletin 110(2) 305-314

Byrne Barbara M 2010 Structural Equation Modeling with Amos Basic Concepts

Applications and Programming 2nd

ed New York Routledge

Camilleri Emanuel 2006 Towards Developing an Organizational Commitment ― Public

Service Motivation Model for Maltese Public Service Employees Public Policy

and Administration 21 63-83

Castaing Seacutebastien 2006 The Effects of Psychological Contract Fulfillment and Public

Service Motivation on Organizational Commitment in the French Civil Service

Public Policy and Administration 21 84-98

Cenfetelli Ronald T and Geneviegraveve Bassellier 2009 Interpretation of Formative

Measurement in Information Systems Research MIS Quarterly 33(4) 689-707

Charles Michael B W Martin de Jong and Neal Ryan 2011 Public Values in Western

- 27 -

Europe A Temporal Perspective American Review of Public Administration 41

75-91

Chin Wynne W 2010 How to Write Up and Report PLS Analysis In Handbook of Partical

Least Squares edited by V Esposito Vinzi et al 655-690 Berlin Springer-Verlag

Coursey David H and Sanjay K Pandey 2007a Public Service Motivation Measurement

Testing an Abridged Version of Perryrsquos Proposed Scale Administration amp Society

39 547-568

Coursey David H and Sanjay K Pandey 2007b Content Domain Measurement and

Validity of the Red Tape Concept American Review of Public Administration 37

342-361

Coursey David H James L Perry Jeffrey L Brudney and Laura Littlepage 2008

Psychometric Verification of Perryrsquos Public Service Motivation Instrument Results

for Volunteer Exemplars Review of Public Personnel Administration 28 79-90

Davis Paul and Karen West 2009 What Do Public Values Mean for Public Action Putting

Public Values in Their Plural Place American Review of Public Administration 39

602-618

DeHart-Davis Leisha Justin Marlowe and Sanjay K Pandey 2006 Gender Dimensions of

Public Service Motivation Public Administration Review 66 873-887

Diamantopoulos Adamantios 2006 The error term in formative measurement models

Interpretation and modeling implications Journal of Modeling in Management 1(1)

7-17

Diamantopoulos Adamantios and Judy A Siguaw 2006 Formative versus Reflective

Indicators in Organizational Measure Development A Comparison and Empirical

Illustration British Journal of Management 17 263-282

Diamantopoulos Adamantios and Heidi M Winklhofer 2001 Index Construction with

Formative Indicators An Alternative to Scale Development Journal of Marketing

Research 38 269-277

Edwards Jeffrey R 2010 The Fallacy of Formative Measurement Organizational Research

- 28 -

Methods Online First published on August 19 2010 as

doi1011771094428110378369

Edwards Jeffrey R and Richard P Bagozzi 2000 On the Nature and Direction of

Relationships between Constructs and Measures Psychological Methods 5 155-

174

Flora David B and Patrick J Curran 2004 An Empirical Evaluation of Alternative

Methods of Estimation for Confirmatory Factor Analysis with Ordinal Data

Psychological Methods 9 466-491

Fornell Claes and Fred L Bookstein 1982 Two structural equations models LISREL and

PLS applied to customer exit-voice theory Journal of Marketing Research 19 440-

452

Giauque David Adrian Ritz Freacutedeacuteric Varone Simon Anderfuhren-Biget and Christian

Waldner 2011 Putting Public Service Motivation into Context A Balance between

Universalism and Particularism Forthcoming in International Review of

Administrative Sciences 77(2)

Goumltz Oliver Kerstin Liehr-Gobbers and Manfred Krafft 2010 Evaluation of Structural

Equation Models Using the Partial Least Squares (PLS) Approach In V Esposito

Vinzi et al (eds) Handbook of Partial Least Squares 691-711 DOI 101007978-3-

540-32827-8_30 Springer-Verlag Berlin Heidelberg

Guinot Christiane Julie Latreille and Michel Tenenhaus 2001 PLS Path Modelling and

Multiple Table Analysis Application to the Cosmetic Habits of Women in Ile-de-

France Chemometrics and Intelligent Laboratory Systems 58 247-259

Hair Joseph F Jr William C Black Barry J Babin and Rolph E Anderson 2010

Multivariate Data Analysis 7th

ed Upper Saddle River NJ Pearson Education

Hardin Andrew M Jerry Cha-Jan Chang Mark A Fuller and Gholamreza Torkzadeh 2010

Formative Measurement and Academic Research In Search of Measurement

Theory Educational and Psychological Measurement Online First published on

June 11 2010 as doi101177001316440370208

- 29 -

Howell Roy D Einar Breivik and James B Wilcox 2007a Reconsidering Formative

Measurement Psychological Methods 12 205-218

Howell Roy D Einar Breivik and James B Wilcox 2007b Is Formative Measurement

Really Measurement Reply to Bollen (2007) and Bagozzi (2007) Psychological

Methods 12 238-245

Jarvis Cheryl Burke Scott B MacKenzie and Philip M Podsakoff 2003 A Critical Review

of Construct Indicators and Measurement Model Misspecification in Marketing and

Consumer Research Journal of Consumer Research 30 199-218

Joumlreskog Karl G and Dag Soumlrbom 2005 LISREL 872 (Computer Software) Lincolnwood

IL Scientific Software International

Kim Gimun Bongsik Shin and Varun Grover 2010 Investigating Two Contradictory Views

of Formative Measurement in Information Systems Research MIS Quarterly 34

345-365

Kim Sangmook 2009a Revising Perryrsquos Measurement Scale of Public Service Motivation

American Review of Public Administration 39 149-163

Kim Sangmook 2009b Testing the Structure of Public Service Motivation in Korea A

Research Note Journal of Public Administration Research and Theory 19 839-

851

Kim Sangmook 2010 Testing a Revised Measurement of Public Service Motivation

Reflective versus Formative Specification Journal of Public Administration

Research and Theory Doi101093jopartmuq048

Kim Sangmook and Wouter Vandenabeele 2010 A strategy for building public service

motivation research internationally Public Administration Review 70 701-709

Kline Rex B 2005 Principles and Practice of Structural Equation Modeling 2nd

ed New

York Guilford Press

Law Kenneth and Chi-Sum Wong 1999 Multidimensional Constructs in Structural

Equation Analysis An Illustration using the Job Perception and Job Satisfaction

Constructs Journal of Management 25 143-160

- 30 -

Leisink Peter and Bram Steijn 2009 Public Service Motivation and Job Performance of

Public Sector Employees in the Netherlands International Review of

Administrative Sciences 75 35-52

Liu Bangcheng Chun Hui Jin Hu Wensheng Yang and Xinli Yu 2011 How well can

Public Service Motivation Connect with Occupational Intention International

Review of Administrative Sciences 77 191-211

Liu Bangcheng Ningyu Tang and Xiaomei Zhu 2008 Public Service Motivation and Job

Satisfaction in China An Investigation of Generalisability and Instrumentality

International Journal of Manpower 29 684-699

MacCallum Robert C and Michael W Browne 1993 The Use of Causal Indicators in

Covariance Structure Models Some Practical Issues Psychological Bulletin 114

533-541

MacKenzie Scott B 2003 The dangers of poor construct conceptualization Journal of the

Academy of Marketing Science 31(3) 323-326

MacKenzie Scott B Philip M Podsakoff and Cheryl Burke Jarvis 2005 The Problem of

Measurement Model Misspecification in Behavioral and Organizational Research

and Some Recommended Solutions Journal of Applied Psychology 90 710-730

Martinsen Dorte Sindbjerg and Torben Beck Joslashrgensen 2010 Accountability as a

Differentiated Value in Supranational Governance American Review of Public

Administration 40 742-760

Moynihan Donald P and Sanjay K Pandey 2007a The Role of Organizations in Fostering

Public Service Motivation Public Administration Review 67 40-53

Moynihan Donald P and Sanjay K Pandey 2007b Finding Workable Levers over Work

Motivation Comparing Job Satisfaction Job Involvement and Organizational

Commitment Administration amp Society 39 803-832

Pandey Sanjay K and Edmund C Stazyk 2008 Antecedents and Correlates of Public

Service Motivation In Motivation in Public Management edited by James L Perry

and Annie Hondeghem 101-117 Oxford Oxford University Press

- 31 -

Perry James L 1996 Measuring Public Service Motivation An Assessment of Construct

Reliability and Validity Journal of Public Administration Research and Theory 6

5-22

Perry James L 1997 Antecedents of Public Service Motivation Journal of Public

Administration Research and Theory 7 181-197

Perry James L 2000 Bringing Society in Toward a Theory of Public Service Motivation

Journal of Public Administration Research and Theory 10 471-488

Perry James L and Annie Hondeghem 2008a Motivation in Public Management Oxford

Oxford University Press

Perry James L and Annie Hondeghem 2008b Directions for Future Theory and Research

In Motivation in Public Management edited by James L Perry and Annie

Hondeghem 294-313 Oxford Oxford University Press

Perry James L Annie Hondeghem and Lois Recascino Wise 2010 Revisiting the

Motivational Bases of Public Service Twenty Years of Research and an Agenda for

the Future Public Administration Review 70(5) 681-690

Perry James L and Wouter Vandenabeele 2008 Behavioral dynamics Institutions

Identities and Self-Regulation In Motivation in Public Management edited by

James L Perry and Annie Hondeghem 56-79 Oxford Oxford University Press

Perry James L and Lois Recascino Wise 1990 The Motivational Bases of Public Service

Public Administration Review 50 367-373

Rainey Hal G 1982 Reward Preference among Public and Private Managers In Search of

the Service Ethic American Review of Public Administration 16 288-302

Ringle Christian M Oliver Goumltz Martin Wetzels and Bradley Wilson 2009 On the Use of

Formative Measurement Specifications in Structural Equation Modeling A Monte

Carlo Simulation Study to Compare Covariance-Based and Partial Least Squares

Model Estimation Methodologies Faculty of Economics and Business Administration

Maastricht Univeristy

Ringle C M S Wende and S Will 2005 Smart PLS 20 (M3) Beta

- 32 -

httpwwwsmartplsde

Ritz Adrian 2011 Attraction to Public Policy Making A Qualitative Inquiry into

Improvements in PSM Measurement Forthcoming in Public Administration

Spector Paul E 1992 Summated Rating Scales Construction An Introduction Newbury

Park CA Sage

Steenbergen Marco R 1996 Compassion and American public opinion An analysis of the

NES Humanitarianism Scale Report to the Board of Overseers of the National

Election Studies

Steffen Patrick R and Kevin S Masters 2005 Does compassion mediate the intrinsic

religion-health relationship Annals of Behavioral Medicine 30 217-223

Taylor Jeannette 2007 The Impact of Public Service Motives on Work Outcomes in

Australia A Comparative Multi-Dimensional Analysis Public Administration 85

931-959

Vandenabeele Wouter 2007 Towards a Theory of Public Service Motivation An

Institutional Approach Public Management Review 9 545-556

Vandenabeele Wouter 2008a Government Calling Public Service Motivation as an Element

in Selecting Government as an Employer of Choice Public Administration 86

1089-1105

Vandenabeele Wouter 2008b Development of a Public Service Motivation Measurement

Scale Corroborating and Extending Perryrsquos Measurement Instrument International

Public Management Journal 11 143-167

Vandenabeele Wouter 2009 The Mediating Effect of Job Satisfaction and Organizational

Commitment on Self-Reported Performance More Robust Evidence of the PSM-

Performance Relationship International Review of Administrative Sciences 75 11-

34

Wilcox James B Roy D Howell and Einar Breivik 2008 Questions about Formative

Measurement Journal of Business Research 61 1219-1228

Wise Lois Recascino 2000 The Public Service Culture In Public Administration Concepts

- 33 -

and Cases 7th ed edited by Richard J Stillman II 342-353 Boston Houghton

Mifflin

Wright Bradley E 2008 Methodological Challenges associated with Public Service

Motivation Research In Motivation in Public Management edited by James L

Perry and Annie Hondeghem 80-98 Oxford Oxford University Press

Wright Bradley E and Sanjay K Pandey 2008 Public Service Motivation and the

Assumption of Person-Organization Fit Testing the Mediating Effect of Value

Congruence Administration amp Society 40 502-521

- 34 -

Table 1

Possible Indicators of PSM and Descriptive Statistics (n = 2868)

Dimensions and items Mean Standard

Deviation

Item-Total

Correlation

Attraction to Public Participation (APP)

APP1 I am interested in helping to improve public service

APP2 I am satisfied when I see people benefiting from the public programs I was

involved in

APP3 I like to discuss topics regarding public programs and policies with others

APP4 I believe that public sector activities contribute to our general welfare

APP5 I admire people who initiate or are involved in activities to aid my

community

APP6 Contributing to public programs and policies helps me realize myself

APP7 It is important to contribute to activities that tackle social problems

412

441

355

393

423

351

409

696

662

917

753

672

881

734

524

421

482

412

493

478

552

Commitment to Public Values (CPV)

Sub-dimension for Public Interests (CPV1)

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

CPI3 I would prefer seeing public officials do what is best for the whole

community even if it harmed my interests

CPI4 Serving the public interest is more important than helping a single individual

Sub-dimension for Public Values (CPV2)

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of public

services

CPV3 It is fundamental that public services respond to the needs of the citizens

CPV4 Decisions regarding public services should be democratic despite the time

and effort it takes

CPV5 Everybody is entitled to a good service even if it costs a lot of money

CPV6 It is fundamental that the interests of future generations are taken into

account when developing public policies

CPV7 To act ethically is essential for public servants

CPV8 I believe that public employees must always be aware of the legitimacy of

their activities

CPV9 I personally identify with the aim of protecting individual liberties and rights

418

409

348

337

428

416

433

394

367

426

441

434

414

695

662

899

995

730

704

710

852

932

689

681

692

763

525

596

480

251

484

437

356

325

376

439

462

402

392

Compassion (COM) COM1 It is difficult for me to contain my feelings when I see people in distress

COM2 I feel sympathetic to the plight of the underprivileged

COM3 I empathize with other people who face difficulties

COM4 I have little compassion for people in need who are unwilling to take the

first step to help themselves

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

354

387

388

299

412

391

951

825

707

1083

755

765

399

511

389

011

487

502

Self-Sacrifice (SS)

SS1 Making a difference to society means more to me than personal achievements

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS5 People should give back to society more than they get from it

SS6 Serving other citizens would give me a good feeling even if no one paid me for

it

SS7 I would agree to a good plan to make a better life for the poor even if it costs

me money

357

329

339

302

345

363

350

904

872

949

982

888

1049

895

508

584

454

531

412

368

546

- 35 -

Table 2

Background of Respondents (n = 2868)

Variables Characteristics Respondents

Sex

Male

Female

NA

1209

1622

37

422

566

13

Length of service

(years)

0 ~ 10

10 ~ 20

20 ~ 30

30+

NA

1126

723

484

491

44

393

252

169

171

15

Organizational status

―Do you supervise employees

No

Yes

NA

1472

1076

320

513

375

112

Country

Australia

Belgium

China

Denmark

France

Italy

Korea

Lithuania

The Netherlands

Switzerland

United Kingdom

USA

249

214

230

249

266

162

253

236

249

250

260

250

87

75

80

87

93

56

88

82

87

87

91

87 Note NA = no answer

- 36 -

Table 3

Standardized factor loadings and correlations for the 31-item PSM index (n = 2868)

Dimensions and items SFL (MLE) SFL (D-WLS) Alpha

APP

APP1

APP2

APP4

APP5

APP6

APP7

597

486

455

558

490

600

657

552

514

625

549

665

699

CPV1

CPI1

CPI2

CPI3

CPI4

588

660

488

289

653

732

535

294

542

CPV2

CPV1

CPV2

CPV3

CPV4

CPV5

CPV6

CPV7

CPV8

CPV9

611

546

436

393

418

534

598

496

460

690

631

505

457

486

601

669

573

530

745

COM

COM1

COM2

COM3

COM5

COM6

496

569

498

597

607

516

651

538

670

684

687

SS

SS1

SS2

SS3

SS4

SS5

SS6

SS7

567

736

558

728

467

409

663

660

762

591

695

515

494

709

782

Measures of fit (MLE) χsup2 (df = 424) = 43699 p lt 001 CFI = 835 GFI = 899

RMSEA = 057 [055 059]

Measure of fit (D-WLS) χsup2 (df = 424) = 39247 p lt 001 CFI = 971 GFI = 976

RMSEA = 054 [052 055]

Inter-factor correlations MLE (D-WLS)

1 2 3 4

1 APS

2 CPV1 996 (988)

3 CPV2 902 (890) 798 (798)

4 COM 787 (783) 758 (762) 779 (783)

5 SS 700 (742) 800 (859) 509 (537) 720 (729)

Note SFL = standardized factor loading All factor loadings and correlations are significant at p lt 001

- 37 -

Table 4

Standardized factor loadings and correlations for the 16-item PSM index (n = 2868)

Dimensions and items SFL

(MLE)

SFL

(D-WLS)

APS

APP5 I admire people who initiate or are involved in activities to aid my

community

APP7 It is important to contribute to activities that tackle social problems

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

611

592

637

666

665

669

690

751

CPV

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of

public services

CPV6 It is fundamental that the interests of future generations are taken

into account when developing public policies

CPV7 To act ethically is essential for public servants

612

546

551

590

698

631

618

658

COM

COM1 It is difficult for me to contain my feelings when I see people in

distress

COM2 I feel sympathetic to the plight of the underprivileged

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

490

546

596

604

515

608

669

679

SS

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS7 I would agree to a good plan to make a better life for the poor even if

it costs me money

763

571

766

678

807

629

747

755

Measures of fit (MLE)

χsup2 (df = 98) = 8490 p lt 001

CFI = 939 GFI = 965

RMSEA = 052 [049 055]

Measure of fit (D-WLS)

χsup2 (df = 98) = 6069 p lt 001

CFI = 988 GFI = 992

RMSEA = 043 [039 046]

Alpha CR (D-WLS) Inter-factor correlations MLE (D-WLS)

1 APS 717 788 1 APS 2 CPV 3 COM

2 CPV 663 747 848 (859)

3 COM 645 713 781 (791) 789 (795)

4 SS 784 826 565 (596) 462 (505) 715 (726)

Note SFL = standardized factor loading Alpha = Cronbachrsquos alpha CR = composite reliability

All factor loadings and correlations are significant at p lt 001

- 38 -

Table 5

Results of CFA in the individual countries Standardized factor loadings and measures of fit

Factors

amp Items

Australia Belgium China Denmark France Italy

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

672

756

799

785

804

943

710

703

597

790

799

730

584

675

685

681

731

742

578

748

674

643

765

739

645

704

678

708

812

779

632

604

765

766

672

853

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

552

593

560

699

727

632

624

605

560

713

709

645

702

602

616

724

686

735

567

560

715

654

605

811

608

658

587

696

755

698

673

620

523

782

697

654

COM

COM1

COM2

COM3

COM5

COM6

623

584

662

688

701

739

732

608

475

808

666

581

578

616

605

649

670

673

699

751

737

679

816

776

568

546

801

626

673

850

637

546

741

715

601

822

SS

SS2

SS3

SS4

SS7

654

794

806

892

767

813

727

693

799

793

725

838

857

820

773

896

855

842

806

749

792

829

753

884

847

731

753

889

816

753

815

692

664

795

782

734

Inter-factor

correlations

093

~

752

163

~

810

320

~

773

323

~

825

452

~

827

497

~

855

323

~

850

321

~

847

498

~

899

536

~

902

425

~

829

476

~

865

Sample size 249 214 230 249 266 162

Measures of Fit

χsup2 638 444 881 732 765 554 817 625 735 565 750 564

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 980 100 939 987 969 997 967 994 975 997 949 995

GFI 961 986 941 986 948 991 949 992 955 992 926 986

RMSEA 036 00 063 050 051 026 053 035 045 026 059 033

- 39 -

Table 5 (Continued)

Factors

amp Items

Korea Lithuania The Netherlands Switzerland UK USA

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

619

701

667

706

781

717

666

512

682

751

605

779

557

629

630

620

719

729

535

621

702

643

694

735

655

725

738

710

804

822

535

621

702

643

694

735

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

594

607

722

722

644

722

674

535

533

790

589

711

568

554

538

655

628

545

627

580

615

711

666

704

800

512

555

873

581

646

627

580

615

711

666

704

COM

COM1

COM2

COM3

COM5

COM6

694

677

550

763

736

627

557

639

428

550

696

534

737

623

508

766

616

608

659

592

683

703

661

769

682

583

729

739

644

803

659

592

683

703

661

769

SS

SS2

SS3

SS4

SS7

776

752

679

779

770

811

707

721

635

725

734

706

814

656

743

820

767

764

745

632

689

788

662

744

671

818

832

830

864

802

745

632

689

788

662

744

Inter-factor

correlations

432

~

726

466

~

746

376

~

788

406

~

902

377

~

852

455

~

836

637

~

818

660

~

828

317

~

852

360

~

852

637

~

818

660

~

828

Sample size 253 236 249 250 260 250

Measures of Fit

χsup2 896 573 612 462 773 616 815 715 769 542 815 715

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 951 996 975 100 958 993 957 991 972 998 957 991

GFI 942 988 960 989 951 987 948 990 955 990 948 990

RMSEA 059 028 034 00 050 034 053 044 048 022 053 044

Note All factor loadings and correlations are significant at p lt 001

- 40 -

Appendix 1 Dimensions and possible items of PSM (the first round)

1 Attraction to Public Participation

(1) I am interested in developing public programs helping my country or community (Kim 2009a)

(2) Sharing my views on public policies with others is attractive to me (Kim 2009a)

(3) I am satisfied when seeing people get benefits from the public programs I was involved in (Kim 2009a)

(4) I like to discuss political subjects with others (Giauque et al 2011)

(5) I think the governmental activities contribute to our welfare

(6) I like to initiate actions to help out my community

(7) I do not care much about politicians (Perry 1996)

2 Commitment to Public Values

Possible items for public interest

(1A) Meaningful public service is very important to me (Perry 1996)

(2A) It is important for me to contribute for the common good (Giauque et al 2011)

(3A) I recognize myself with the mission of protection of liberties and rights endorsed by the nation-state

(Giauque et al 2011)

(4A) I would prefer seeing public officials do what is best for the whole community even if it harmed my

interests (Perry 1996)

(5A) To me serving the public interest is more important than helping other people (Vandenabeele 2008b)

Possible items for public values

(1B) I recognize myself with the promotion of the equality of chances endorsed by the nation-state (Giauque et

al 2011)

(2B) It is fundamental that the public service provided to the citizen is regular and continuous (Giauque et al

2011)

(3B) It is fundamental that public service innovates to respond to the new needs of the citizensclients (Giauque

et al 2011)

(4B) The dignity and well-being of all should be the most important concerns in any society (Steenbergen

1996)

(5B) It is important that public servants account for all the costsexpenses they make (Vandenabeele 2008b)

(6B) We have to do everything in our power to pursue the goal of democracy

3 Compassion

(1) It is difficult for me to contain my feelings when I see people in distress (Perry 1996)

(2) I feel sympathetic to the plight of the underprivileged (Kim 2009a)

(3) I am often reminded by daily events how dependent we are on one another (Perry 1996)

(4) I feel compassion for others in the difficulties they are facing (Steffen and Masters 2005)

(5) It is not really my problem if others are in trouble and need help (Beutel and Marini 1995)

(6) I find it hard to be sympathetic toward starving people in foreign lands when there is so much trouble in our

own country (Beutel and Marini 1995)

(7) I get very upset when I see other people treated unfairly (Beutel and Marini 1995)

(8) I would agree to a good plan to make a better life for the poor even if it cost me money (Beutel and Marini

1995)

(9) To me patriotism includes seeing the welfare of others (Perry 1996)

(10) I care very much about other people

4 Self-Sacrifice

(1) Making a difference in society means more to me than personal achievements (Perry 1996)

(2) I am prepared to make enormous sacrifices for the good of society (Perry 1996)

(3) I believe in putting duty before self (Perry 1996)

(4) I am one of those rare people who would risk personal loss to help someone else (Perry 1996)

(5) I think people should give back to society more than they get from it (Perry 1996)

(6) Serving other citizens would give me a good feeling even if no one paid me for it (Perry 1996)

(7) Much of what I do is for a cause bigger than myself (Perry 1996)

- 41 -

Appendix 2 Newly suggested items through the first round

1 Items for the Attraction to Public Participation Dimension

(1) In the public domain every person should be counted as equal

(2) I enjoy to think and argue in a political way

(3) I am satisfied when seeing my ideas being integrated in public policies

(4) I can easily relate to the way of thinking and arguing of politicians

(5) Working in the public sector I can better realize myself

2 Items for the Public Values Sub-Dimension

(1) Public service should give value for money

(2) Cost considerations should play an important role in decisions about public service provision

(3) It is fundamental that the interests of future generations are taken into account when developing public

policies

(4) To act with correct ethics is for public servants as important as their professional competence

(5) Primarily public servants should be accountable to the public and not to their superiors

3 Items for the Self-Sacrifice Dimension

(1) I would agree to a good plan to make a better life for the poor even if it costs me money

- 42 -

Appendix 3 An example of a summary of comments and suggestions (the second round)

On the item ―I am interested in developing public programs helping my country or community

A It is a good sentence but is it possible for everyone to ―develop public programs Even public officials are

limited to do this

B OK

C or the special group

D Double barrel and difficult to translate to my country ―I am interested in helping to improve public services

E Good but I feel that wording requires revision If we draw sample from local government then country is

redundant ―I am interested in developing public policy programs that aim in helping my community

F 4 (General appreciation from1 to 4) Not clear whether it refers to ones work or to ones concern as a citizen

―(In my work) I am interested in contributing to the development of public etchellip

G Instead of development I would use implementing

H Delete public which is redundant because of the characteristic helping my country or community but add

or implementing ―I am interested in developing or implementing programs helping my country or

community

I Referring to dimension Wording Similar to item 6

Note Names have been changed to alphabets in order to ensure the anonymity of participants

- 43 -

Appendix 4 Dimensions and Items of Public Service Motivation for Future Research

Dimensions and items

Attraction to Public Service (APS)

1 I am interested in helping to improve public service

2 I admire people who initiate or are involved in activities to aid my community

3 It is important to contribute to activities that tackle social problems

4 Meaningful public service is very important to me

5 It is important for me to contribute to the common good

Commitment to Public Values (CPV)

1 I think equal opportunities for citizens are very important

2 It is important that citizens can rely on the continuous provision of public services

3 It is fundamental that the interests of future generations are taken into account when

developing public policies

4 To act ethically is essential for public servants

5 I believe that public employees must always be aware of the legitimacy of their activities

6 I personally identify with the aim of protecting individual liberties and rights

Compassion (COM)

1 It is difficult for me to contain my feelings when I see people in distress

2 I feel sympathetic to the plight of the underprivileged

3 I empathize with other people who face difficulties

4 I get very upset when I see other people being treated unfairly

5 Considering the welfare of others is very important

Self-Sacrifice (SS)

1 I am prepared to make sacrifices for the good of society

2 I believe in putting civic duty before self

3 I am willing to risk personal loss to help society

4 I would agree to a good plan to make a better life for the poor even if it costs me money

- 11 -

from the sample

We aimed to receive 250 respondents for each country as this would enable us to

carry out the analyses we wanted to perform on the data Although there is no ideal sample

size one study recommends a sample of 200 to provide a sound basis for estimation (Hair

Black Babin and Anderson 2010) The method of data collection was left open Obviously a

web-survey is the easiest way to collect the data because it is economical to collect and to

code however compared to paper-pencil surveys this method has some drawbacks First not

everybody in the sample may have access to the internet which may cause a bias in the

responses Second internet surveys are generally known to produce lower response rates

Web surveys were recommended if possible but paper-pencil surveys were deemed

acceptable in cases where it was not possible to conduct a web survey Researchers were

encouraged to contact just one municipality except in cases where it was not possible to

gather enough data In such instances responses from multiple municipalities were accepted

By the end of September 2010 a total of 2868 responses were obtained2

2 In Australia seven local councils each with between from 70 and 232 employees were surveyed Paper-pencil

surveys were used in two councils while the rest relied on a web survey The local governments limited

distribution of paper-pencil surveys to only one or two sections of the council eg administration The response

rate was approximately 30 However it should be noted that not everyone working for the council has direct

access to a computer In Belgium three municipalities were surveyed on a census base by means of a web

survey as all employees had access to a computer This rendered a response of 120 (226) 79 (75) and 16

(53) responses for the respective municipalities resulting in an overall response rate of 331 with 662

invitations being sent out In China two departments social work and administration were surveyed in one

municipality using paper-pencil surveys In the social work department 65 out of 110 surveys were returned

and 165 out of 211 responded from the administration department giving an overall response rate of 767 In

Denmark a total of 1282 respondents in a municipality were sent an initial e-mail invitation to a web survey

and up to two e-mail reminders and the response rate was 473 In France all the employees in two

municipalities were surveyed In one municipality where both web and paper-pencil surveys were utilized 128

out of 685 questionnaires were returned while in another city using only a paper-pencil survey 138 out of 652

were received The average response rate was 199 In Italy in principle the link to the questionnaire was

made available to all employees (about 4000) in one city but for some of them access may have been difficult

yielding a response rate of 45 In Korea 500 employees in a city received a printed questionnaire and 253

responses were returned yielding a response rate of 506 In Lithuania 236 responses from 54 municipalities

were gathered by web survey In the Netherlands three municipalities were surveyed on a census base by means

of a web survey In total 319 responses of 910 invitations were returned (35) mdash 90 (45) 115 (38) and 114

(28) for the respective municipalities In Switzerland 516 out of 1326 local employees with e-mail access in

a municipality completed the survey giving a response rate of 389 In the United Kingdom the survey was

posted online and 1360 employees both in administrative jobs and front-line services of one municipality

- 12 -

Municipalities were selected because of their convenient accessibility and proximity to the

researcher and a convenience sample was used in most countries Seven countries surveyed

one municipality but seven local councils were surveyed in Australia Web surveys were used

in four countries while three countries used both web and paper-pencil survey The number

of respondents varied among the countries depending on the numbers of questionnaires

distributed the survey method and the response rate In most countries we used

approximately 250 responses but only 162 from Italy3 The majority of the respondents were

women (566) while in terms of service length those in the largest group (393) had

worked for fewer than 10 years Table 2 shows the distribution of respondentsrsquo sex length of

service employment status and country

[Table 2 about here]

Analyses

The statistical analysis applied confirmatory factor analysis (CFA) CFA was used to assess

the fit of the data to the hypothesized measurement model because it is preferred where

measurement models have a well-developed underlying theory for hypothesized patterns of

loadings The data were analyzed using Amos 180 (Arbuckle 2009) with the maximum

likelihood (MLE) estimation method and LISREL 872 (Joumlreskog and Soumlrbom 2005) with the

diagonally weighted least squares (D-WLS) estimation method

The individual items are designed to measure a theoretically continuous construct

but the observed responses are a discrete realization of five categories The common method

received an email to complete the survey The response rate was 19 In the USA the total number of

employees surveyed was 626 in one city using both electronic and paper-pencil surveys resulting in a response

rate of 588 3 In countries where significantly more than 250 surveys were collected we randomly selected only 250 cases

for analysis

- 13 -

of estimation within CFA is MLE a technique which assumes that the observed variables are

continuous and normally distributed These assumptions are not met when the observed data

are discrete as occurs when using ordinal scales as in this study A simulation study (Flora

and Curran 2004) found that D-WLS performs well for ordinal measures across a variety of

conditions especially for smaller samples Coursey and Pandey (2007a 2007b) argued the

importance of using D-WLS and not standard MLE in structural equations for ordinal data4

In most cases however the differences between MLE and D-WLS are not substantive to the

conclusions5 No comparison between standard MLE and D-WLS is provided in this paper

We give results from both MLE and D-WLS and this will aid in confirming the stability and

robustness of the results across varying mathematical specifications

For gauging reliability Cronbachrsquos alpha and composite reliability were tested

Composite reliability should be equal to or greater than 07 (Hair et al 2010) Convergent

validity was assessed from the measurement model by determining the statistical significance

of each indicatorrsquos estimated pattern coefficient on its specified underlying construct

Discriminant validity was assessed by determining whether the confidence interval around

the correlation estimate between two factors included 100 (Anderson and Gerbing 1988)

For model fit assessment goodness-of-fit index (GFI) comparative fit index (CFI)

and the root mean square error of approximation (RMSEA) were used The model achieves

an acceptable fit to the data when GFI and CFI equals or exceeds 090 and RMSEA values

4 Applying standard MLE to ordinal data such as Likert-type items may produce significant estimation

problems such as inflation of chi-square fit statistics and biased underestimation of parameters and standard

errors (Flora and Curran 2004) Coursey and Pandey (2007a 2007b) used a modified MLE using the asymptotic

covariance matrix and D-WLS for adjusting the ordinal nature of the data They reported that although the

modified MLE findings are slightly different on a few model parameters the differences are not substantive to

the conclusions However model fit estimators were clearly better for the D-WLS 5 Given normally distributed categorical variables MLE can be used with confidence when a variable has four

or more categories If there is a large sample a five-point Likert scale and the skewness and kurtosis of variables

is within acceptable limits it suggests distribution symmetry ―In most cases where the hypothesized model is

well specified and the scaling based on more than three categories it seems unlikely that there will be much

difference between the findings (Byrne 2010 160)

- 14 -

fall below 008 (Hair et al 2010) In general the larger the value of GFI and CFI and the

smaller the value of RMSEA the better the fit of the model (Bollen 1989) We also present

the χ2 test statistic Lower values of χ

2 indicate a better fit and should be nonsignificant but

the χ2 value increases as sample size increases and thus for large sample sizes this statistic

may lead to rejection of a model with good fit (Hair et al 2010)

Results

Using SPSS 180 descriptive statistics were computed for individual items as shown in table

1 One item (APP3) in the APP dimension ―I like to discuss topics regarding public programs

and policies with others was mistakenly omitted in the survey conducted in the United

States and so it was deleted in the statistical analysis In the assessment of normality no

items showed a skew or kurtosis value greater than the cutoffs of |3| or |8| recommended by

Kline (2005) As data were distributed normally MLE as well as D-WLS estimation was used

Based on inspection of item-total correlations one item (COM4) was dropped from further

analysis because it was weakly correlated with the overall index (011) Missing data is an

issue in CFA as it can cause estimation errors All variables in the survey have missing data

rates of approximately 016 to 143 Mean substitution was used to calculate replacement

values for the missing values in each countryrsquos dataset (Hair et al 2010) The deletion of the

two items left 31 items for the next stage of analysis

The five-correlated factor model was tested using CFA The CFA model in the initial

analysis hypothesized a priori that (1) responses to the 30-item PSM measurement

instrument could be explained by five factors such as APP CPV1 CPV2 COM SS (2) each

item would have a nonzero loading on the PSM factor that it was designed to measure and

- 15 -

zero loadings on all other factors (3) the five factors would be correlated and (4)

measurement error terms would be uncorrelated The resulting CFA with MLE suggested that

the initial five-correlated factor model was not a good fit to the data χsup2 (df = 424) = 43699 p

lt 001 GFI = 0899 CFI = 835 RMSEA = 0057 [055 059] At this stage APP was highly

correlated with CPV1 (r = 996 in MLE and r = 988 in D-WLS) suggesting the two

dimensions were not unique and lacked discriminant validity

[Table 3 about here]

Given the disconfirmation of the initial model an effort was made to estimate an

alternative model The two dimensions APP and CPV1 were combined to form a single

dimension Attraction to Public Service (APS) The items of CPV1 may be regarded as

action-oriented by the respondents because dedication to public service community and

common good is underlined in these items Consequently it is reasonable to combine APP

with CPV1 to form APS a dimension that focuses more on disposition to serve the public to

work for the common good and to participate in public policy processes After the overall

number of dimensions was reduced to four using standardized factor loadings and

modification indices as criteria the items with the lowest factor loading in each dimension

were deleted and the items whose modification index indicated association with multiple

dimensions were eliminated A CFA model with the remaining items was iteratively tested

until it achieved a good fit to the data The outcome of this refinement process was a 16-item

index of four factors Overall model fit by every measure is quite strong for the D-WLS

estimation technique The estimation results for the four-dimension model are presented in

table 4

- 16 -

[Table 4 about here]

Cronbachrsquos coefficient alpha for the 16-item PSM index was 860 and the coefficient

alphas for the four dimensions ranged from 645 to 784 The composite reliability of the set

of reflective indicators for each dimension of PSM ranged from 713 to 826 and indicated

adequate internal consistency The resulting factor structure showed a four-factor structure

with all items loading significantly onto their a priori dimension (p lt 001) and the

standardized factor loadings ranging from 0490 to 0766 in MLE and from 515 to 807 in

D-WLS The results provide support for convergent validity The correlation estimates

between the two factors ranged from 0462 to 0848 in MLE and from 505 to 859 in D-

WLS The confidence intervals (plusmn2 standard errors) around the correlation estimates between

the two factors did not include 100 The result provides support for discriminant validity To

test whether the four-dimension model is superior to other designs such as the two or three-

dimension models a series of CFA were conducted and the results show that the four-

dimension model is a significantly better fit than any of the other designs

To test whether the factorial structure of the PSM instrument is equivalent across the

samples from the twelve countries four nested models were tested as part of multi-group

analysis with MLE (Byrne 2010) First the unconstrained model was the baseline model

which relaxed all equality constraints This model hypothesized the same four-factor structure

of the instrument across twelve countries The fit indices for this baseline model were χsup2 (df =

1176) = 24977 p lt 001 GFI = 0901 CFI = 900 RMSEA = 0020 [019 021] These

indices indicate that the four-factor model of the PSM instrument exhibits acceptable fit

across the samples from the twelve countries Second the measurement weights model tested

- 17 -

the invariance of factor loadings across countries by placing equality constraints on these

parameters The fit indices for this model were χsup2 (df = 1308) = 28088 p lt 001 GFI = 0890

CFI = 887 RMSEA = 0020 [019 021] Since the measurement weights model was nested

within the unconstrained model the chi-square difference text ∆χsup2 (132) = 3111 p lt 001

indicated that some equality constraints of factor loadings did not hold across the twelve

countries Further test results suggested that the equality constraints of structural covariances

and measurement residuals were not upheld across the twelve countries It needs to be

stressed that despite the differences in factor loadings and factor covariances the results

support the main assumption that the same four-factor structure holds across different

countries We may assume that people in different cultures understand PSM in the same

fashion

To find differences in the factor structure of PSM among countries we conducted a

series of CFA with each countryrsquos dataset and followed the same procedure as when

analyzing the sample as a whole As table 5 shows although there are great similarities

between the appropriate items in the countries there are also some differences Even though

the items vary according to national differences the four-factor structure had a good fit to the

data Thus the results confirm that the four-factor structure of PSM can be generalized

[Table 5 about here]

The APS dimension covers both an individualrsquos attraction to public participation and

commitment to public interest but all its items refer to some notion of public interest such as

improving public service aiding community tackling social problems meaningful public

service and contributing to the common good The APS dimension captures an individualrsquos

- 18 -

disposition to serve the public interest through public participation When developing the

items some assumed that the items of committing to public interests can be separated from

those of pursuing public values in the CPV dimension The empirical test showed that

respondents tended to understand committing to public interests as behavior-oriented and so

the items for public interest are not differentiated from participation-oriented items in their

perceptions but are different from those for public values The respondents seem to

understand that pursuing public values is more related to personal beliefs or creeds Thus the

CPV dimension is solely focused on public values and the items of pursing public interests

are combined with those of the APP dimension to make the APS dimension

Public values and their relative importance are neither impervious to change over

time nor the same among all countries (Charles de Jong and Ryan 2011 Davis and West

2009) Moreover some values may be in conflict as Martinsen and Beck Joslashrgensen (2010)

discussed in regard to the tensions between accountability and efficiency It is therefore not

possible to include all public values in one dimension (Beck Joslashrgensen and Bozeman 2007)

Accordingly the items of the CPV dimension are developed for estimating individual

commitment to seek public values which there are general consensus about and which are

generally formulated

When all four dimensions of PSM are included in a survey cross-national

equivalence of the survey can be assured The items represent reflections or manifestations of

each dimension and hence are relatively homogeneous and interchangeable Adding or

deleting items does not change the essential nature of the dimension Thus any combination

of items in a dimension can be regarded as equivalent When using structural equation

modeling (SEM) each dimension needs to have at least three reflective indicators in order to

achieve identification or it needs to have at least two reflective indicators and be allowed to

- 19 -

covary with another dimension that has at least two reflective measures With one indicator

the dimension is identified if the loading and the variance of the uniqueness are both fixed

(Bollen 1989 Edwards 2010) Thus the number of items in each dimension is not restricted

when measuring PSM However we recommend using multiple items for each dimension

because more research is needed to investigate the appropriateness of this PSM instrument

with varying samples

The test results showed that model fit estimators were clearly better for the D-WLS

It means that in this study MLE was more rigorous than D-WLS in assessing the fit of the

model The empirical evidence indicated that MLE may be an acceptable technique for

ordinal data when there is a large sample a five-point Likert scale and the skewness and

kurtosis of variables is within acceptable limits as suggested by Byrne (2010)

Discussion

The analysis reaffirmed the four-dimensional structure of PSM but the results were not

entirely consistent with the original CFA we tested We were ultimately able to converge on a

good-fit operational model for PSM that was also consistent with underlying theory The

analysis is not able to resolve the debate about measurement models however for reasons

discussed below

Choosing the correct measurement model (ie reflective vs formative) is important

as measurement model misspecification results not only in misrepresentation of the epistemic

relationships between constructs and their measures but also in biased parameter estimates of

structural relationships (Jarvis et al 2003 Law and Wong 1999 MacKenzie 2003

MacKenzie Podsakoff and Jarvis 2005) A formative measurement model in isolation is

- 20 -

under-identified and cannot be estimated in the case of covariance-based SEM (CB-SEM)

(Bollen 1989) A construct with formative indicators must emit paths to at least two unrelated

reflective indicators two unrelated latent constructs or some combination of the two in order

to solve the problem of under-identification (Jarvis MacKenzie and Podsakoff 2003

MacCallum and Browne 1993 MacKenzie Podsakoff and Jarvis 2005) Thus ―when

dependent constructs change the empirical nature of the formatively measured construct

changes (Wilcox Howell and Breivik 2008 1223) and ―when specifying formative

measurement models what might seem like an innocuous choice of outcomes to achieve

identification can have fundamental ramifications for the meaning of the formative construct

and the inferences based on it (Edwards 2010 7)

Earlier research found that the association between a construct and its measures

should remain stable regardless of the larger causal model in which the constructs and

measures are embedded (Edwards and Bagozzi 2000) The association between a construct

and its measures is generally stable for reflective measures but associations of formative

measures with their construct are unstable The susceptibility of formative measurement to

both interpretational confounding and external inconsistency recommends that formative

measurement is not an equally attractive alternative to reflective measurement and that

whenever possible researchers should opt for reflective measurement (Howell Breivik and

Wilcox 2007a 2007b Kim Shin and Grover 2010) In the case of second-order models

researchers are limited to only a second-order reflective model in CB-SEM (Chin 2010)

Regardless of the measurement debate and the tradeoffs between reflective or

formative specification this study makes a theoretical argument that PSM is a second-order

formative construct This second-order formative factor model (reflective first-order

formative second-order) is also suggested as an alternative model that captures the

- 21 -

multidimensionality of complex concepts and avoids the shortcomings of first-order

formative measurement (Edwards 2010) However this model itself cannot achieve

identification One possible strategy is to develop direct reflective measures of PSM based on

the view that PSM is a general concept that is not only influenced by its specific dimensions

but also can be assessed using general measures (Edwards 2010 Javis MacKenzie and

Podsakoff 2003)6 However these general measures have not yet been developed for the

PSM construct

Another way around the problem of underidentification of formative models is to

apply the component-based approach known as partial least squares path modeling (PLS-PM)

which can model formative indicators (Fornell and Bookstein 1982 Goumltz Liehr-Gobbers

and Krafft 2010) The CB-SEM and the PLS-PM methods have distinctive statistical

characteristics and selecting an approach to SEM depends on the particular research situation

CB-SEM is the method of choice for theory testing while PLS-PM is appropriate for

prediction-oriented applications (Ringle Goumltz Wetzels and Wilson 2009) One advantage of

the PLS method is that it demands significantly fewer requirements compared to that of

covariance structure analyses but nevertheless delivers consistent estimation results Another

asset of the PLS approach is its ability to deal with formative as well as reflective indicators

even within one structural equation model This indicates that the PLS approach can be

appropriate for explorative analysis of structural equation models (Chin 2010 Goumltz Liehr-

Gobbers and Krafft 2010) A Monte Carlo simulation study to compare CB-SEM and PLS-

PM showed that PLS-PM provides a viable approximation of model parameters when the

prerequisites for CB-SEM are not met (Ringle Goumltz Wetzels and Wilson 2009) A formative

6 For example job satisfaction can be assessed with general reflective measures such as ―in general I am

satisfied with my job and the formative dimensions can be assessed with measures that describe satisfaction

with each job facet (Edwards 2010 15)

- 22 -

composite in the form of indexes needs to be tested to know the relative weights (γ-

parameters) of the dimensions of PSM In PLS-PM a second-order formative factor is

directly measured by observed variables for all the first order factors While this approach

repeats the number of manifest variables used the model can be identified without adding

outcome constructs that would be considered inappropriate or irrelevant from a conceptual

standpoint (Chin 2010 Edwards 2010) The test result of PLS-PM using Smart PLS 20

(Ringle Wende and Will 2005) produced the function PSM = 31APS + 36CPV + 28COM

+ 31SS However one consideration is whether and how construct error is modeled for the

formative construct (Diamantopoulos 2006) When using PLS-PM the researcher is making

the assumption that there is no error at the construct level as all constructs in PLS-PM are

modeled without error When using PLS-PM the absence of an error term will tend to

increase the weights therefore inflating their importance (Bollen and Lennox 1991) The

primary statistic for assessing a formative indicator is its weight the partialized effect of the

indicator on its intended construct controlling for the effects of all other indicators of that

construct (Cenfetelli and Bassellier 2009) A formative measure is essentially a multiple

regression with the construct representing the dependent variable and the indicators as the

predictors (Diamantopoulos and Winklhofer 2001) PSM may be regarded as a weighted

linear composite of its dimensions The weights used to create the composites are different

across CE-SEM and PLS-PM The causal dimensional weights generated in CB-SEM are

estimated in a model in which a construct-level error term is specified (eg PSM = γ1APM +

γ2CPI + γ3COM + γ4SS + ζ) whereas the weights in PLS-PM are not (eg PSM = γ1APM +

γ2CPI + γ3COM + γ4SS) This leads to two different conceptualizations of the PSM construct

one in which the causal indicators do not capture the construct in its entirety and another in

which they do (Hardin et al 2010) These differences need to be discussed in future research

- 23 -

in this area

One practical suggestion to measure PSM is to apply unit weights to its dimensions if

we assume the effects of the four dimensions are equivalent A literature review indicates that

unit weights have substantial predictive validity when compared with regression weights The

literature shows that the empirical correlations between scores generated from differential and

unit-weighting approaches are quite high In creating composite scores unit weights can be

an appropriate approach for weighting under many circumstances (Bobko Roth and Buster

2007) Although the uses of unit-weight composites will likely result in the explanation of

less variance than empirical estimations of dimensional weights they will remain consistently

defined across studies and hence suitable for theory testing Alternatively more thorough

discussions on the dimensions of PSM may suggest fixed-weight composites (Hardin et al

2010) A previous study (Kim 2010 22) also suggested ―Overall indexing PSM by

measuring all the dimensions of PSM and simply summing the averages of each dimension

can make it easier to measure PSM in practice as PSM = APS + CPV + COM + SS

An advantage of regarding PSM as a second-order formative construct is that the

dimensions can be analyzed both individually and collectively This is important as the

dimensions are expected to have different antecedents and consequences For instance Perry

(1997) found some significant differences in the influence of independent variables on the

four dimensions of PSM Taylor (2007) confirmed that certain dimensions are more important

than others in influencing work outcomes Moynihan and Pandey (2007a) reported that red

tape is negatively related to one dimension but reform orientation and hierarchical authority

is positively related to another dimension A recent study found that the SS dimension has a

positive relationship with occupational intention but the other dimensions do not (Liu et al

2011) Thus it is valuable to analyze the causes and effects of individual dimensions of PSM

- 24 -

as well as of the PSM construct

Conclusion

The purpose of this study is to develop a measurement instrument for PSM for cross-national

research and comparison On the basis of the sharpened construct of PSM (Kim and

Vandenabeele 2010) using data from an international survey conducted in 12 countries we

found that the four-factor structure of PSM is generalizable across country cases This study

provides the measurement index of PSM with the four dimensions of APS CPV COM and

SS The APS dimension measures the instrumental motives of PSM the CPV dimension

determines the value-based motives of PSM while the COM and SS dimensions gauge the

identification motives and self-sacrifice respectively When all four dimensions of PSM are

used in a survey cross-national equivalence of survey items can be assured The items for

measuring PSM in future research are suggested in Appendix 4

This study has several limitations First looking for samples which were comparable

to the largest possible extent we opted for surveys in local government The results found

and discussed in this study are based on PSM among local government employees The

results might be different when surveying central government employees or other specific

groups in the public sector Thus the potential impact of the sample bias needs to be

considered in understanding this study It also means that the survey items for measuring

PSM should be examined with different samples in various areas and at different government

levels

Second the test results showed that there is a high degree of correlation among the

dimension such as APS CPV and COM Future research should not only investigate whether

- 25 -

the items reflect the core content of the dimensions but also continue to develop more

appropriate items for better discriminant validity of each dimension Third the content of

APP dimension was assumed to be distinct from the CPV dimension that focuses on personal

disposition to pursue public interest and public values However the public interest sub-

dimension was separated from the public values sub-dimension in the CPV dimension and

combined with the APP dimension to form the APS dimension Further research should

analyze the CPV dimension and investigate the appropriateness of the APS dimension

In order to facilitate research internationally the PSM construct needs to be

improved conceptually and operationally Kim and Vandenabeele (2010) have dealt with the

conceptual composition and operational dimension of PSM as well as the measurement

model of PSM The next step is to develop an internationally robust measurement instrument

of PSM This study proposes that scholars use all four of the dimensions in measuring the

conceptual components of PSM such as self-sacrifice instrumental value-based and affective

motives because they collectively represent PSM It provides a set of appropriate items for

measuring PSM and confirms the dimensional structure of PSM The survey results suggest

that people in different cultures understand PSM in the same fashion We know that this study

is simply a step along the way to developing a more robust measure but we hope it

contributes to enhanced cross-national research and comparison along with the generation of

cumulative knowledge internationally

- 26 -

REFERENCES

Anderson James C and David D Gerbing 1988 Structural Equation Modeling in Practice

A Review and Recommended Two-Step Approach Psychological Bulletin 103

411-423

Arbuckle James L 2009 Amos 180 Userrsquos Guide Crawfordville FL Amos Development

Co

Beck Joslashrgensen Torben and Barry Bozeman 2007 Public Values An Inventory

Administration amp Society 39(3) 354-381

Beutel Ann M and Margaret Mooney Marini 1995 Gender and values American

Sociological Review 60 436 - 448

Bobko Philip Philip L Roth and Maury A Buster 2007 The Usefulness of Unit Weights in

Creating Composite Scores A Literature Review Application to Content Validity and

Meta-Analysis Organizational Research Methods 10(4) 689-709

Bollen Kenneth A 1989 Structural Equations with Latent Variables New York Wiley

Bollen K A and R Lennox 1991 Conventional Wisdom on Measurement A Structural

Equation Perspective Psychological Bulletin 110(2) 305-314

Byrne Barbara M 2010 Structural Equation Modeling with Amos Basic Concepts

Applications and Programming 2nd

ed New York Routledge

Camilleri Emanuel 2006 Towards Developing an Organizational Commitment ― Public

Service Motivation Model for Maltese Public Service Employees Public Policy

and Administration 21 63-83

Castaing Seacutebastien 2006 The Effects of Psychological Contract Fulfillment and Public

Service Motivation on Organizational Commitment in the French Civil Service

Public Policy and Administration 21 84-98

Cenfetelli Ronald T and Geneviegraveve Bassellier 2009 Interpretation of Formative

Measurement in Information Systems Research MIS Quarterly 33(4) 689-707

Charles Michael B W Martin de Jong and Neal Ryan 2011 Public Values in Western

- 27 -

Europe A Temporal Perspective American Review of Public Administration 41

75-91

Chin Wynne W 2010 How to Write Up and Report PLS Analysis In Handbook of Partical

Least Squares edited by V Esposito Vinzi et al 655-690 Berlin Springer-Verlag

Coursey David H and Sanjay K Pandey 2007a Public Service Motivation Measurement

Testing an Abridged Version of Perryrsquos Proposed Scale Administration amp Society

39 547-568

Coursey David H and Sanjay K Pandey 2007b Content Domain Measurement and

Validity of the Red Tape Concept American Review of Public Administration 37

342-361

Coursey David H James L Perry Jeffrey L Brudney and Laura Littlepage 2008

Psychometric Verification of Perryrsquos Public Service Motivation Instrument Results

for Volunteer Exemplars Review of Public Personnel Administration 28 79-90

Davis Paul and Karen West 2009 What Do Public Values Mean for Public Action Putting

Public Values in Their Plural Place American Review of Public Administration 39

602-618

DeHart-Davis Leisha Justin Marlowe and Sanjay K Pandey 2006 Gender Dimensions of

Public Service Motivation Public Administration Review 66 873-887

Diamantopoulos Adamantios 2006 The error term in formative measurement models

Interpretation and modeling implications Journal of Modeling in Management 1(1)

7-17

Diamantopoulos Adamantios and Judy A Siguaw 2006 Formative versus Reflective

Indicators in Organizational Measure Development A Comparison and Empirical

Illustration British Journal of Management 17 263-282

Diamantopoulos Adamantios and Heidi M Winklhofer 2001 Index Construction with

Formative Indicators An Alternative to Scale Development Journal of Marketing

Research 38 269-277

Edwards Jeffrey R 2010 The Fallacy of Formative Measurement Organizational Research

- 28 -

Methods Online First published on August 19 2010 as

doi1011771094428110378369

Edwards Jeffrey R and Richard P Bagozzi 2000 On the Nature and Direction of

Relationships between Constructs and Measures Psychological Methods 5 155-

174

Flora David B and Patrick J Curran 2004 An Empirical Evaluation of Alternative

Methods of Estimation for Confirmatory Factor Analysis with Ordinal Data

Psychological Methods 9 466-491

Fornell Claes and Fred L Bookstein 1982 Two structural equations models LISREL and

PLS applied to customer exit-voice theory Journal of Marketing Research 19 440-

452

Giauque David Adrian Ritz Freacutedeacuteric Varone Simon Anderfuhren-Biget and Christian

Waldner 2011 Putting Public Service Motivation into Context A Balance between

Universalism and Particularism Forthcoming in International Review of

Administrative Sciences 77(2)

Goumltz Oliver Kerstin Liehr-Gobbers and Manfred Krafft 2010 Evaluation of Structural

Equation Models Using the Partial Least Squares (PLS) Approach In V Esposito

Vinzi et al (eds) Handbook of Partial Least Squares 691-711 DOI 101007978-3-

540-32827-8_30 Springer-Verlag Berlin Heidelberg

Guinot Christiane Julie Latreille and Michel Tenenhaus 2001 PLS Path Modelling and

Multiple Table Analysis Application to the Cosmetic Habits of Women in Ile-de-

France Chemometrics and Intelligent Laboratory Systems 58 247-259

Hair Joseph F Jr William C Black Barry J Babin and Rolph E Anderson 2010

Multivariate Data Analysis 7th

ed Upper Saddle River NJ Pearson Education

Hardin Andrew M Jerry Cha-Jan Chang Mark A Fuller and Gholamreza Torkzadeh 2010

Formative Measurement and Academic Research In Search of Measurement

Theory Educational and Psychological Measurement Online First published on

June 11 2010 as doi101177001316440370208

- 29 -

Howell Roy D Einar Breivik and James B Wilcox 2007a Reconsidering Formative

Measurement Psychological Methods 12 205-218

Howell Roy D Einar Breivik and James B Wilcox 2007b Is Formative Measurement

Really Measurement Reply to Bollen (2007) and Bagozzi (2007) Psychological

Methods 12 238-245

Jarvis Cheryl Burke Scott B MacKenzie and Philip M Podsakoff 2003 A Critical Review

of Construct Indicators and Measurement Model Misspecification in Marketing and

Consumer Research Journal of Consumer Research 30 199-218

Joumlreskog Karl G and Dag Soumlrbom 2005 LISREL 872 (Computer Software) Lincolnwood

IL Scientific Software International

Kim Gimun Bongsik Shin and Varun Grover 2010 Investigating Two Contradictory Views

of Formative Measurement in Information Systems Research MIS Quarterly 34

345-365

Kim Sangmook 2009a Revising Perryrsquos Measurement Scale of Public Service Motivation

American Review of Public Administration 39 149-163

Kim Sangmook 2009b Testing the Structure of Public Service Motivation in Korea A

Research Note Journal of Public Administration Research and Theory 19 839-

851

Kim Sangmook 2010 Testing a Revised Measurement of Public Service Motivation

Reflective versus Formative Specification Journal of Public Administration

Research and Theory Doi101093jopartmuq048

Kim Sangmook and Wouter Vandenabeele 2010 A strategy for building public service

motivation research internationally Public Administration Review 70 701-709

Kline Rex B 2005 Principles and Practice of Structural Equation Modeling 2nd

ed New

York Guilford Press

Law Kenneth and Chi-Sum Wong 1999 Multidimensional Constructs in Structural

Equation Analysis An Illustration using the Job Perception and Job Satisfaction

Constructs Journal of Management 25 143-160

- 30 -

Leisink Peter and Bram Steijn 2009 Public Service Motivation and Job Performance of

Public Sector Employees in the Netherlands International Review of

Administrative Sciences 75 35-52

Liu Bangcheng Chun Hui Jin Hu Wensheng Yang and Xinli Yu 2011 How well can

Public Service Motivation Connect with Occupational Intention International

Review of Administrative Sciences 77 191-211

Liu Bangcheng Ningyu Tang and Xiaomei Zhu 2008 Public Service Motivation and Job

Satisfaction in China An Investigation of Generalisability and Instrumentality

International Journal of Manpower 29 684-699

MacCallum Robert C and Michael W Browne 1993 The Use of Causal Indicators in

Covariance Structure Models Some Practical Issues Psychological Bulletin 114

533-541

MacKenzie Scott B 2003 The dangers of poor construct conceptualization Journal of the

Academy of Marketing Science 31(3) 323-326

MacKenzie Scott B Philip M Podsakoff and Cheryl Burke Jarvis 2005 The Problem of

Measurement Model Misspecification in Behavioral and Organizational Research

and Some Recommended Solutions Journal of Applied Psychology 90 710-730

Martinsen Dorte Sindbjerg and Torben Beck Joslashrgensen 2010 Accountability as a

Differentiated Value in Supranational Governance American Review of Public

Administration 40 742-760

Moynihan Donald P and Sanjay K Pandey 2007a The Role of Organizations in Fostering

Public Service Motivation Public Administration Review 67 40-53

Moynihan Donald P and Sanjay K Pandey 2007b Finding Workable Levers over Work

Motivation Comparing Job Satisfaction Job Involvement and Organizational

Commitment Administration amp Society 39 803-832

Pandey Sanjay K and Edmund C Stazyk 2008 Antecedents and Correlates of Public

Service Motivation In Motivation in Public Management edited by James L Perry

and Annie Hondeghem 101-117 Oxford Oxford University Press

- 31 -

Perry James L 1996 Measuring Public Service Motivation An Assessment of Construct

Reliability and Validity Journal of Public Administration Research and Theory 6

5-22

Perry James L 1997 Antecedents of Public Service Motivation Journal of Public

Administration Research and Theory 7 181-197

Perry James L 2000 Bringing Society in Toward a Theory of Public Service Motivation

Journal of Public Administration Research and Theory 10 471-488

Perry James L and Annie Hondeghem 2008a Motivation in Public Management Oxford

Oxford University Press

Perry James L and Annie Hondeghem 2008b Directions for Future Theory and Research

In Motivation in Public Management edited by James L Perry and Annie

Hondeghem 294-313 Oxford Oxford University Press

Perry James L Annie Hondeghem and Lois Recascino Wise 2010 Revisiting the

Motivational Bases of Public Service Twenty Years of Research and an Agenda for

the Future Public Administration Review 70(5) 681-690

Perry James L and Wouter Vandenabeele 2008 Behavioral dynamics Institutions

Identities and Self-Regulation In Motivation in Public Management edited by

James L Perry and Annie Hondeghem 56-79 Oxford Oxford University Press

Perry James L and Lois Recascino Wise 1990 The Motivational Bases of Public Service

Public Administration Review 50 367-373

Rainey Hal G 1982 Reward Preference among Public and Private Managers In Search of

the Service Ethic American Review of Public Administration 16 288-302

Ringle Christian M Oliver Goumltz Martin Wetzels and Bradley Wilson 2009 On the Use of

Formative Measurement Specifications in Structural Equation Modeling A Monte

Carlo Simulation Study to Compare Covariance-Based and Partial Least Squares

Model Estimation Methodologies Faculty of Economics and Business Administration

Maastricht Univeristy

Ringle C M S Wende and S Will 2005 Smart PLS 20 (M3) Beta

- 32 -

httpwwwsmartplsde

Ritz Adrian 2011 Attraction to Public Policy Making A Qualitative Inquiry into

Improvements in PSM Measurement Forthcoming in Public Administration

Spector Paul E 1992 Summated Rating Scales Construction An Introduction Newbury

Park CA Sage

Steenbergen Marco R 1996 Compassion and American public opinion An analysis of the

NES Humanitarianism Scale Report to the Board of Overseers of the National

Election Studies

Steffen Patrick R and Kevin S Masters 2005 Does compassion mediate the intrinsic

religion-health relationship Annals of Behavioral Medicine 30 217-223

Taylor Jeannette 2007 The Impact of Public Service Motives on Work Outcomes in

Australia A Comparative Multi-Dimensional Analysis Public Administration 85

931-959

Vandenabeele Wouter 2007 Towards a Theory of Public Service Motivation An

Institutional Approach Public Management Review 9 545-556

Vandenabeele Wouter 2008a Government Calling Public Service Motivation as an Element

in Selecting Government as an Employer of Choice Public Administration 86

1089-1105

Vandenabeele Wouter 2008b Development of a Public Service Motivation Measurement

Scale Corroborating and Extending Perryrsquos Measurement Instrument International

Public Management Journal 11 143-167

Vandenabeele Wouter 2009 The Mediating Effect of Job Satisfaction and Organizational

Commitment on Self-Reported Performance More Robust Evidence of the PSM-

Performance Relationship International Review of Administrative Sciences 75 11-

34

Wilcox James B Roy D Howell and Einar Breivik 2008 Questions about Formative

Measurement Journal of Business Research 61 1219-1228

Wise Lois Recascino 2000 The Public Service Culture In Public Administration Concepts

- 33 -

and Cases 7th ed edited by Richard J Stillman II 342-353 Boston Houghton

Mifflin

Wright Bradley E 2008 Methodological Challenges associated with Public Service

Motivation Research In Motivation in Public Management edited by James L

Perry and Annie Hondeghem 80-98 Oxford Oxford University Press

Wright Bradley E and Sanjay K Pandey 2008 Public Service Motivation and the

Assumption of Person-Organization Fit Testing the Mediating Effect of Value

Congruence Administration amp Society 40 502-521

- 34 -

Table 1

Possible Indicators of PSM and Descriptive Statistics (n = 2868)

Dimensions and items Mean Standard

Deviation

Item-Total

Correlation

Attraction to Public Participation (APP)

APP1 I am interested in helping to improve public service

APP2 I am satisfied when I see people benefiting from the public programs I was

involved in

APP3 I like to discuss topics regarding public programs and policies with others

APP4 I believe that public sector activities contribute to our general welfare

APP5 I admire people who initiate or are involved in activities to aid my

community

APP6 Contributing to public programs and policies helps me realize myself

APP7 It is important to contribute to activities that tackle social problems

412

441

355

393

423

351

409

696

662

917

753

672

881

734

524

421

482

412

493

478

552

Commitment to Public Values (CPV)

Sub-dimension for Public Interests (CPV1)

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

CPI3 I would prefer seeing public officials do what is best for the whole

community even if it harmed my interests

CPI4 Serving the public interest is more important than helping a single individual

Sub-dimension for Public Values (CPV2)

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of public

services

CPV3 It is fundamental that public services respond to the needs of the citizens

CPV4 Decisions regarding public services should be democratic despite the time

and effort it takes

CPV5 Everybody is entitled to a good service even if it costs a lot of money

CPV6 It is fundamental that the interests of future generations are taken into

account when developing public policies

CPV7 To act ethically is essential for public servants

CPV8 I believe that public employees must always be aware of the legitimacy of

their activities

CPV9 I personally identify with the aim of protecting individual liberties and rights

418

409

348

337

428

416

433

394

367

426

441

434

414

695

662

899

995

730

704

710

852

932

689

681

692

763

525

596

480

251

484

437

356

325

376

439

462

402

392

Compassion (COM) COM1 It is difficult for me to contain my feelings when I see people in distress

COM2 I feel sympathetic to the plight of the underprivileged

COM3 I empathize with other people who face difficulties

COM4 I have little compassion for people in need who are unwilling to take the

first step to help themselves

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

354

387

388

299

412

391

951

825

707

1083

755

765

399

511

389

011

487

502

Self-Sacrifice (SS)

SS1 Making a difference to society means more to me than personal achievements

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS5 People should give back to society more than they get from it

SS6 Serving other citizens would give me a good feeling even if no one paid me for

it

SS7 I would agree to a good plan to make a better life for the poor even if it costs

me money

357

329

339

302

345

363

350

904

872

949

982

888

1049

895

508

584

454

531

412

368

546

- 35 -

Table 2

Background of Respondents (n = 2868)

Variables Characteristics Respondents

Sex

Male

Female

NA

1209

1622

37

422

566

13

Length of service

(years)

0 ~ 10

10 ~ 20

20 ~ 30

30+

NA

1126

723

484

491

44

393

252

169

171

15

Organizational status

―Do you supervise employees

No

Yes

NA

1472

1076

320

513

375

112

Country

Australia

Belgium

China

Denmark

France

Italy

Korea

Lithuania

The Netherlands

Switzerland

United Kingdom

USA

249

214

230

249

266

162

253

236

249

250

260

250

87

75

80

87

93

56

88

82

87

87

91

87 Note NA = no answer

- 36 -

Table 3

Standardized factor loadings and correlations for the 31-item PSM index (n = 2868)

Dimensions and items SFL (MLE) SFL (D-WLS) Alpha

APP

APP1

APP2

APP4

APP5

APP6

APP7

597

486

455

558

490

600

657

552

514

625

549

665

699

CPV1

CPI1

CPI2

CPI3

CPI4

588

660

488

289

653

732

535

294

542

CPV2

CPV1

CPV2

CPV3

CPV4

CPV5

CPV6

CPV7

CPV8

CPV9

611

546

436

393

418

534

598

496

460

690

631

505

457

486

601

669

573

530

745

COM

COM1

COM2

COM3

COM5

COM6

496

569

498

597

607

516

651

538

670

684

687

SS

SS1

SS2

SS3

SS4

SS5

SS6

SS7

567

736

558

728

467

409

663

660

762

591

695

515

494

709

782

Measures of fit (MLE) χsup2 (df = 424) = 43699 p lt 001 CFI = 835 GFI = 899

RMSEA = 057 [055 059]

Measure of fit (D-WLS) χsup2 (df = 424) = 39247 p lt 001 CFI = 971 GFI = 976

RMSEA = 054 [052 055]

Inter-factor correlations MLE (D-WLS)

1 2 3 4

1 APS

2 CPV1 996 (988)

3 CPV2 902 (890) 798 (798)

4 COM 787 (783) 758 (762) 779 (783)

5 SS 700 (742) 800 (859) 509 (537) 720 (729)

Note SFL = standardized factor loading All factor loadings and correlations are significant at p lt 001

- 37 -

Table 4

Standardized factor loadings and correlations for the 16-item PSM index (n = 2868)

Dimensions and items SFL

(MLE)

SFL

(D-WLS)

APS

APP5 I admire people who initiate or are involved in activities to aid my

community

APP7 It is important to contribute to activities that tackle social problems

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

611

592

637

666

665

669

690

751

CPV

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of

public services

CPV6 It is fundamental that the interests of future generations are taken

into account when developing public policies

CPV7 To act ethically is essential for public servants

612

546

551

590

698

631

618

658

COM

COM1 It is difficult for me to contain my feelings when I see people in

distress

COM2 I feel sympathetic to the plight of the underprivileged

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

490

546

596

604

515

608

669

679

SS

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS7 I would agree to a good plan to make a better life for the poor even if

it costs me money

763

571

766

678

807

629

747

755

Measures of fit (MLE)

χsup2 (df = 98) = 8490 p lt 001

CFI = 939 GFI = 965

RMSEA = 052 [049 055]

Measure of fit (D-WLS)

χsup2 (df = 98) = 6069 p lt 001

CFI = 988 GFI = 992

RMSEA = 043 [039 046]

Alpha CR (D-WLS) Inter-factor correlations MLE (D-WLS)

1 APS 717 788 1 APS 2 CPV 3 COM

2 CPV 663 747 848 (859)

3 COM 645 713 781 (791) 789 (795)

4 SS 784 826 565 (596) 462 (505) 715 (726)

Note SFL = standardized factor loading Alpha = Cronbachrsquos alpha CR = composite reliability

All factor loadings and correlations are significant at p lt 001

- 38 -

Table 5

Results of CFA in the individual countries Standardized factor loadings and measures of fit

Factors

amp Items

Australia Belgium China Denmark France Italy

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

672

756

799

785

804

943

710

703

597

790

799

730

584

675

685

681

731

742

578

748

674

643

765

739

645

704

678

708

812

779

632

604

765

766

672

853

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

552

593

560

699

727

632

624

605

560

713

709

645

702

602

616

724

686

735

567

560

715

654

605

811

608

658

587

696

755

698

673

620

523

782

697

654

COM

COM1

COM2

COM3

COM5

COM6

623

584

662

688

701

739

732

608

475

808

666

581

578

616

605

649

670

673

699

751

737

679

816

776

568

546

801

626

673

850

637

546

741

715

601

822

SS

SS2

SS3

SS4

SS7

654

794

806

892

767

813

727

693

799

793

725

838

857

820

773

896

855

842

806

749

792

829

753

884

847

731

753

889

816

753

815

692

664

795

782

734

Inter-factor

correlations

093

~

752

163

~

810

320

~

773

323

~

825

452

~

827

497

~

855

323

~

850

321

~

847

498

~

899

536

~

902

425

~

829

476

~

865

Sample size 249 214 230 249 266 162

Measures of Fit

χsup2 638 444 881 732 765 554 817 625 735 565 750 564

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 980 100 939 987 969 997 967 994 975 997 949 995

GFI 961 986 941 986 948 991 949 992 955 992 926 986

RMSEA 036 00 063 050 051 026 053 035 045 026 059 033

- 39 -

Table 5 (Continued)

Factors

amp Items

Korea Lithuania The Netherlands Switzerland UK USA

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

619

701

667

706

781

717

666

512

682

751

605

779

557

629

630

620

719

729

535

621

702

643

694

735

655

725

738

710

804

822

535

621

702

643

694

735

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

594

607

722

722

644

722

674

535

533

790

589

711

568

554

538

655

628

545

627

580

615

711

666

704

800

512

555

873

581

646

627

580

615

711

666

704

COM

COM1

COM2

COM3

COM5

COM6

694

677

550

763

736

627

557

639

428

550

696

534

737

623

508

766

616

608

659

592

683

703

661

769

682

583

729

739

644

803

659

592

683

703

661

769

SS

SS2

SS3

SS4

SS7

776

752

679

779

770

811

707

721

635

725

734

706

814

656

743

820

767

764

745

632

689

788

662

744

671

818

832

830

864

802

745

632

689

788

662

744

Inter-factor

correlations

432

~

726

466

~

746

376

~

788

406

~

902

377

~

852

455

~

836

637

~

818

660

~

828

317

~

852

360

~

852

637

~

818

660

~

828

Sample size 253 236 249 250 260 250

Measures of Fit

χsup2 896 573 612 462 773 616 815 715 769 542 815 715

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 951 996 975 100 958 993 957 991 972 998 957 991

GFI 942 988 960 989 951 987 948 990 955 990 948 990

RMSEA 059 028 034 00 050 034 053 044 048 022 053 044

Note All factor loadings and correlations are significant at p lt 001

- 40 -

Appendix 1 Dimensions and possible items of PSM (the first round)

1 Attraction to Public Participation

(1) I am interested in developing public programs helping my country or community (Kim 2009a)

(2) Sharing my views on public policies with others is attractive to me (Kim 2009a)

(3) I am satisfied when seeing people get benefits from the public programs I was involved in (Kim 2009a)

(4) I like to discuss political subjects with others (Giauque et al 2011)

(5) I think the governmental activities contribute to our welfare

(6) I like to initiate actions to help out my community

(7) I do not care much about politicians (Perry 1996)

2 Commitment to Public Values

Possible items for public interest

(1A) Meaningful public service is very important to me (Perry 1996)

(2A) It is important for me to contribute for the common good (Giauque et al 2011)

(3A) I recognize myself with the mission of protection of liberties and rights endorsed by the nation-state

(Giauque et al 2011)

(4A) I would prefer seeing public officials do what is best for the whole community even if it harmed my

interests (Perry 1996)

(5A) To me serving the public interest is more important than helping other people (Vandenabeele 2008b)

Possible items for public values

(1B) I recognize myself with the promotion of the equality of chances endorsed by the nation-state (Giauque et

al 2011)

(2B) It is fundamental that the public service provided to the citizen is regular and continuous (Giauque et al

2011)

(3B) It is fundamental that public service innovates to respond to the new needs of the citizensclients (Giauque

et al 2011)

(4B) The dignity and well-being of all should be the most important concerns in any society (Steenbergen

1996)

(5B) It is important that public servants account for all the costsexpenses they make (Vandenabeele 2008b)

(6B) We have to do everything in our power to pursue the goal of democracy

3 Compassion

(1) It is difficult for me to contain my feelings when I see people in distress (Perry 1996)

(2) I feel sympathetic to the plight of the underprivileged (Kim 2009a)

(3) I am often reminded by daily events how dependent we are on one another (Perry 1996)

(4) I feel compassion for others in the difficulties they are facing (Steffen and Masters 2005)

(5) It is not really my problem if others are in trouble and need help (Beutel and Marini 1995)

(6) I find it hard to be sympathetic toward starving people in foreign lands when there is so much trouble in our

own country (Beutel and Marini 1995)

(7) I get very upset when I see other people treated unfairly (Beutel and Marini 1995)

(8) I would agree to a good plan to make a better life for the poor even if it cost me money (Beutel and Marini

1995)

(9) To me patriotism includes seeing the welfare of others (Perry 1996)

(10) I care very much about other people

4 Self-Sacrifice

(1) Making a difference in society means more to me than personal achievements (Perry 1996)

(2) I am prepared to make enormous sacrifices for the good of society (Perry 1996)

(3) I believe in putting duty before self (Perry 1996)

(4) I am one of those rare people who would risk personal loss to help someone else (Perry 1996)

(5) I think people should give back to society more than they get from it (Perry 1996)

(6) Serving other citizens would give me a good feeling even if no one paid me for it (Perry 1996)

(7) Much of what I do is for a cause bigger than myself (Perry 1996)

- 41 -

Appendix 2 Newly suggested items through the first round

1 Items for the Attraction to Public Participation Dimension

(1) In the public domain every person should be counted as equal

(2) I enjoy to think and argue in a political way

(3) I am satisfied when seeing my ideas being integrated in public policies

(4) I can easily relate to the way of thinking and arguing of politicians

(5) Working in the public sector I can better realize myself

2 Items for the Public Values Sub-Dimension

(1) Public service should give value for money

(2) Cost considerations should play an important role in decisions about public service provision

(3) It is fundamental that the interests of future generations are taken into account when developing public

policies

(4) To act with correct ethics is for public servants as important as their professional competence

(5) Primarily public servants should be accountable to the public and not to their superiors

3 Items for the Self-Sacrifice Dimension

(1) I would agree to a good plan to make a better life for the poor even if it costs me money

- 42 -

Appendix 3 An example of a summary of comments and suggestions (the second round)

On the item ―I am interested in developing public programs helping my country or community

A It is a good sentence but is it possible for everyone to ―develop public programs Even public officials are

limited to do this

B OK

C or the special group

D Double barrel and difficult to translate to my country ―I am interested in helping to improve public services

E Good but I feel that wording requires revision If we draw sample from local government then country is

redundant ―I am interested in developing public policy programs that aim in helping my community

F 4 (General appreciation from1 to 4) Not clear whether it refers to ones work or to ones concern as a citizen

―(In my work) I am interested in contributing to the development of public etchellip

G Instead of development I would use implementing

H Delete public which is redundant because of the characteristic helping my country or community but add

or implementing ―I am interested in developing or implementing programs helping my country or

community

I Referring to dimension Wording Similar to item 6

Note Names have been changed to alphabets in order to ensure the anonymity of participants

- 43 -

Appendix 4 Dimensions and Items of Public Service Motivation for Future Research

Dimensions and items

Attraction to Public Service (APS)

1 I am interested in helping to improve public service

2 I admire people who initiate or are involved in activities to aid my community

3 It is important to contribute to activities that tackle social problems

4 Meaningful public service is very important to me

5 It is important for me to contribute to the common good

Commitment to Public Values (CPV)

1 I think equal opportunities for citizens are very important

2 It is important that citizens can rely on the continuous provision of public services

3 It is fundamental that the interests of future generations are taken into account when

developing public policies

4 To act ethically is essential for public servants

5 I believe that public employees must always be aware of the legitimacy of their activities

6 I personally identify with the aim of protecting individual liberties and rights

Compassion (COM)

1 It is difficult for me to contain my feelings when I see people in distress

2 I feel sympathetic to the plight of the underprivileged

3 I empathize with other people who face difficulties

4 I get very upset when I see other people being treated unfairly

5 Considering the welfare of others is very important

Self-Sacrifice (SS)

1 I am prepared to make sacrifices for the good of society

2 I believe in putting civic duty before self

3 I am willing to risk personal loss to help society

4 I would agree to a good plan to make a better life for the poor even if it costs me money

- 12 -

Municipalities were selected because of their convenient accessibility and proximity to the

researcher and a convenience sample was used in most countries Seven countries surveyed

one municipality but seven local councils were surveyed in Australia Web surveys were used

in four countries while three countries used both web and paper-pencil survey The number

of respondents varied among the countries depending on the numbers of questionnaires

distributed the survey method and the response rate In most countries we used

approximately 250 responses but only 162 from Italy3 The majority of the respondents were

women (566) while in terms of service length those in the largest group (393) had

worked for fewer than 10 years Table 2 shows the distribution of respondentsrsquo sex length of

service employment status and country

[Table 2 about here]

Analyses

The statistical analysis applied confirmatory factor analysis (CFA) CFA was used to assess

the fit of the data to the hypothesized measurement model because it is preferred where

measurement models have a well-developed underlying theory for hypothesized patterns of

loadings The data were analyzed using Amos 180 (Arbuckle 2009) with the maximum

likelihood (MLE) estimation method and LISREL 872 (Joumlreskog and Soumlrbom 2005) with the

diagonally weighted least squares (D-WLS) estimation method

The individual items are designed to measure a theoretically continuous construct

but the observed responses are a discrete realization of five categories The common method

received an email to complete the survey The response rate was 19 In the USA the total number of

employees surveyed was 626 in one city using both electronic and paper-pencil surveys resulting in a response

rate of 588 3 In countries where significantly more than 250 surveys were collected we randomly selected only 250 cases

for analysis

- 13 -

of estimation within CFA is MLE a technique which assumes that the observed variables are

continuous and normally distributed These assumptions are not met when the observed data

are discrete as occurs when using ordinal scales as in this study A simulation study (Flora

and Curran 2004) found that D-WLS performs well for ordinal measures across a variety of

conditions especially for smaller samples Coursey and Pandey (2007a 2007b) argued the

importance of using D-WLS and not standard MLE in structural equations for ordinal data4

In most cases however the differences between MLE and D-WLS are not substantive to the

conclusions5 No comparison between standard MLE and D-WLS is provided in this paper

We give results from both MLE and D-WLS and this will aid in confirming the stability and

robustness of the results across varying mathematical specifications

For gauging reliability Cronbachrsquos alpha and composite reliability were tested

Composite reliability should be equal to or greater than 07 (Hair et al 2010) Convergent

validity was assessed from the measurement model by determining the statistical significance

of each indicatorrsquos estimated pattern coefficient on its specified underlying construct

Discriminant validity was assessed by determining whether the confidence interval around

the correlation estimate between two factors included 100 (Anderson and Gerbing 1988)

For model fit assessment goodness-of-fit index (GFI) comparative fit index (CFI)

and the root mean square error of approximation (RMSEA) were used The model achieves

an acceptable fit to the data when GFI and CFI equals or exceeds 090 and RMSEA values

4 Applying standard MLE to ordinal data such as Likert-type items may produce significant estimation

problems such as inflation of chi-square fit statistics and biased underestimation of parameters and standard

errors (Flora and Curran 2004) Coursey and Pandey (2007a 2007b) used a modified MLE using the asymptotic

covariance matrix and D-WLS for adjusting the ordinal nature of the data They reported that although the

modified MLE findings are slightly different on a few model parameters the differences are not substantive to

the conclusions However model fit estimators were clearly better for the D-WLS 5 Given normally distributed categorical variables MLE can be used with confidence when a variable has four

or more categories If there is a large sample a five-point Likert scale and the skewness and kurtosis of variables

is within acceptable limits it suggests distribution symmetry ―In most cases where the hypothesized model is

well specified and the scaling based on more than three categories it seems unlikely that there will be much

difference between the findings (Byrne 2010 160)

- 14 -

fall below 008 (Hair et al 2010) In general the larger the value of GFI and CFI and the

smaller the value of RMSEA the better the fit of the model (Bollen 1989) We also present

the χ2 test statistic Lower values of χ

2 indicate a better fit and should be nonsignificant but

the χ2 value increases as sample size increases and thus for large sample sizes this statistic

may lead to rejection of a model with good fit (Hair et al 2010)

Results

Using SPSS 180 descriptive statistics were computed for individual items as shown in table

1 One item (APP3) in the APP dimension ―I like to discuss topics regarding public programs

and policies with others was mistakenly omitted in the survey conducted in the United

States and so it was deleted in the statistical analysis In the assessment of normality no

items showed a skew or kurtosis value greater than the cutoffs of |3| or |8| recommended by

Kline (2005) As data were distributed normally MLE as well as D-WLS estimation was used

Based on inspection of item-total correlations one item (COM4) was dropped from further

analysis because it was weakly correlated with the overall index (011) Missing data is an

issue in CFA as it can cause estimation errors All variables in the survey have missing data

rates of approximately 016 to 143 Mean substitution was used to calculate replacement

values for the missing values in each countryrsquos dataset (Hair et al 2010) The deletion of the

two items left 31 items for the next stage of analysis

The five-correlated factor model was tested using CFA The CFA model in the initial

analysis hypothesized a priori that (1) responses to the 30-item PSM measurement

instrument could be explained by five factors such as APP CPV1 CPV2 COM SS (2) each

item would have a nonzero loading on the PSM factor that it was designed to measure and

- 15 -

zero loadings on all other factors (3) the five factors would be correlated and (4)

measurement error terms would be uncorrelated The resulting CFA with MLE suggested that

the initial five-correlated factor model was not a good fit to the data χsup2 (df = 424) = 43699 p

lt 001 GFI = 0899 CFI = 835 RMSEA = 0057 [055 059] At this stage APP was highly

correlated with CPV1 (r = 996 in MLE and r = 988 in D-WLS) suggesting the two

dimensions were not unique and lacked discriminant validity

[Table 3 about here]

Given the disconfirmation of the initial model an effort was made to estimate an

alternative model The two dimensions APP and CPV1 were combined to form a single

dimension Attraction to Public Service (APS) The items of CPV1 may be regarded as

action-oriented by the respondents because dedication to public service community and

common good is underlined in these items Consequently it is reasonable to combine APP

with CPV1 to form APS a dimension that focuses more on disposition to serve the public to

work for the common good and to participate in public policy processes After the overall

number of dimensions was reduced to four using standardized factor loadings and

modification indices as criteria the items with the lowest factor loading in each dimension

were deleted and the items whose modification index indicated association with multiple

dimensions were eliminated A CFA model with the remaining items was iteratively tested

until it achieved a good fit to the data The outcome of this refinement process was a 16-item

index of four factors Overall model fit by every measure is quite strong for the D-WLS

estimation technique The estimation results for the four-dimension model are presented in

table 4

- 16 -

[Table 4 about here]

Cronbachrsquos coefficient alpha for the 16-item PSM index was 860 and the coefficient

alphas for the four dimensions ranged from 645 to 784 The composite reliability of the set

of reflective indicators for each dimension of PSM ranged from 713 to 826 and indicated

adequate internal consistency The resulting factor structure showed a four-factor structure

with all items loading significantly onto their a priori dimension (p lt 001) and the

standardized factor loadings ranging from 0490 to 0766 in MLE and from 515 to 807 in

D-WLS The results provide support for convergent validity The correlation estimates

between the two factors ranged from 0462 to 0848 in MLE and from 505 to 859 in D-

WLS The confidence intervals (plusmn2 standard errors) around the correlation estimates between

the two factors did not include 100 The result provides support for discriminant validity To

test whether the four-dimension model is superior to other designs such as the two or three-

dimension models a series of CFA were conducted and the results show that the four-

dimension model is a significantly better fit than any of the other designs

To test whether the factorial structure of the PSM instrument is equivalent across the

samples from the twelve countries four nested models were tested as part of multi-group

analysis with MLE (Byrne 2010) First the unconstrained model was the baseline model

which relaxed all equality constraints This model hypothesized the same four-factor structure

of the instrument across twelve countries The fit indices for this baseline model were χsup2 (df =

1176) = 24977 p lt 001 GFI = 0901 CFI = 900 RMSEA = 0020 [019 021] These

indices indicate that the four-factor model of the PSM instrument exhibits acceptable fit

across the samples from the twelve countries Second the measurement weights model tested

- 17 -

the invariance of factor loadings across countries by placing equality constraints on these

parameters The fit indices for this model were χsup2 (df = 1308) = 28088 p lt 001 GFI = 0890

CFI = 887 RMSEA = 0020 [019 021] Since the measurement weights model was nested

within the unconstrained model the chi-square difference text ∆χsup2 (132) = 3111 p lt 001

indicated that some equality constraints of factor loadings did not hold across the twelve

countries Further test results suggested that the equality constraints of structural covariances

and measurement residuals were not upheld across the twelve countries It needs to be

stressed that despite the differences in factor loadings and factor covariances the results

support the main assumption that the same four-factor structure holds across different

countries We may assume that people in different cultures understand PSM in the same

fashion

To find differences in the factor structure of PSM among countries we conducted a

series of CFA with each countryrsquos dataset and followed the same procedure as when

analyzing the sample as a whole As table 5 shows although there are great similarities

between the appropriate items in the countries there are also some differences Even though

the items vary according to national differences the four-factor structure had a good fit to the

data Thus the results confirm that the four-factor structure of PSM can be generalized

[Table 5 about here]

The APS dimension covers both an individualrsquos attraction to public participation and

commitment to public interest but all its items refer to some notion of public interest such as

improving public service aiding community tackling social problems meaningful public

service and contributing to the common good The APS dimension captures an individualrsquos

- 18 -

disposition to serve the public interest through public participation When developing the

items some assumed that the items of committing to public interests can be separated from

those of pursuing public values in the CPV dimension The empirical test showed that

respondents tended to understand committing to public interests as behavior-oriented and so

the items for public interest are not differentiated from participation-oriented items in their

perceptions but are different from those for public values The respondents seem to

understand that pursuing public values is more related to personal beliefs or creeds Thus the

CPV dimension is solely focused on public values and the items of pursing public interests

are combined with those of the APP dimension to make the APS dimension

Public values and their relative importance are neither impervious to change over

time nor the same among all countries (Charles de Jong and Ryan 2011 Davis and West

2009) Moreover some values may be in conflict as Martinsen and Beck Joslashrgensen (2010)

discussed in regard to the tensions between accountability and efficiency It is therefore not

possible to include all public values in one dimension (Beck Joslashrgensen and Bozeman 2007)

Accordingly the items of the CPV dimension are developed for estimating individual

commitment to seek public values which there are general consensus about and which are

generally formulated

When all four dimensions of PSM are included in a survey cross-national

equivalence of the survey can be assured The items represent reflections or manifestations of

each dimension and hence are relatively homogeneous and interchangeable Adding or

deleting items does not change the essential nature of the dimension Thus any combination

of items in a dimension can be regarded as equivalent When using structural equation

modeling (SEM) each dimension needs to have at least three reflective indicators in order to

achieve identification or it needs to have at least two reflective indicators and be allowed to

- 19 -

covary with another dimension that has at least two reflective measures With one indicator

the dimension is identified if the loading and the variance of the uniqueness are both fixed

(Bollen 1989 Edwards 2010) Thus the number of items in each dimension is not restricted

when measuring PSM However we recommend using multiple items for each dimension

because more research is needed to investigate the appropriateness of this PSM instrument

with varying samples

The test results showed that model fit estimators were clearly better for the D-WLS

It means that in this study MLE was more rigorous than D-WLS in assessing the fit of the

model The empirical evidence indicated that MLE may be an acceptable technique for

ordinal data when there is a large sample a five-point Likert scale and the skewness and

kurtosis of variables is within acceptable limits as suggested by Byrne (2010)

Discussion

The analysis reaffirmed the four-dimensional structure of PSM but the results were not

entirely consistent with the original CFA we tested We were ultimately able to converge on a

good-fit operational model for PSM that was also consistent with underlying theory The

analysis is not able to resolve the debate about measurement models however for reasons

discussed below

Choosing the correct measurement model (ie reflective vs formative) is important

as measurement model misspecification results not only in misrepresentation of the epistemic

relationships between constructs and their measures but also in biased parameter estimates of

structural relationships (Jarvis et al 2003 Law and Wong 1999 MacKenzie 2003

MacKenzie Podsakoff and Jarvis 2005) A formative measurement model in isolation is

- 20 -

under-identified and cannot be estimated in the case of covariance-based SEM (CB-SEM)

(Bollen 1989) A construct with formative indicators must emit paths to at least two unrelated

reflective indicators two unrelated latent constructs or some combination of the two in order

to solve the problem of under-identification (Jarvis MacKenzie and Podsakoff 2003

MacCallum and Browne 1993 MacKenzie Podsakoff and Jarvis 2005) Thus ―when

dependent constructs change the empirical nature of the formatively measured construct

changes (Wilcox Howell and Breivik 2008 1223) and ―when specifying formative

measurement models what might seem like an innocuous choice of outcomes to achieve

identification can have fundamental ramifications for the meaning of the formative construct

and the inferences based on it (Edwards 2010 7)

Earlier research found that the association between a construct and its measures

should remain stable regardless of the larger causal model in which the constructs and

measures are embedded (Edwards and Bagozzi 2000) The association between a construct

and its measures is generally stable for reflective measures but associations of formative

measures with their construct are unstable The susceptibility of formative measurement to

both interpretational confounding and external inconsistency recommends that formative

measurement is not an equally attractive alternative to reflective measurement and that

whenever possible researchers should opt for reflective measurement (Howell Breivik and

Wilcox 2007a 2007b Kim Shin and Grover 2010) In the case of second-order models

researchers are limited to only a second-order reflective model in CB-SEM (Chin 2010)

Regardless of the measurement debate and the tradeoffs between reflective or

formative specification this study makes a theoretical argument that PSM is a second-order

formative construct This second-order formative factor model (reflective first-order

formative second-order) is also suggested as an alternative model that captures the

- 21 -

multidimensionality of complex concepts and avoids the shortcomings of first-order

formative measurement (Edwards 2010) However this model itself cannot achieve

identification One possible strategy is to develop direct reflective measures of PSM based on

the view that PSM is a general concept that is not only influenced by its specific dimensions

but also can be assessed using general measures (Edwards 2010 Javis MacKenzie and

Podsakoff 2003)6 However these general measures have not yet been developed for the

PSM construct

Another way around the problem of underidentification of formative models is to

apply the component-based approach known as partial least squares path modeling (PLS-PM)

which can model formative indicators (Fornell and Bookstein 1982 Goumltz Liehr-Gobbers

and Krafft 2010) The CB-SEM and the PLS-PM methods have distinctive statistical

characteristics and selecting an approach to SEM depends on the particular research situation

CB-SEM is the method of choice for theory testing while PLS-PM is appropriate for

prediction-oriented applications (Ringle Goumltz Wetzels and Wilson 2009) One advantage of

the PLS method is that it demands significantly fewer requirements compared to that of

covariance structure analyses but nevertheless delivers consistent estimation results Another

asset of the PLS approach is its ability to deal with formative as well as reflective indicators

even within one structural equation model This indicates that the PLS approach can be

appropriate for explorative analysis of structural equation models (Chin 2010 Goumltz Liehr-

Gobbers and Krafft 2010) A Monte Carlo simulation study to compare CB-SEM and PLS-

PM showed that PLS-PM provides a viable approximation of model parameters when the

prerequisites for CB-SEM are not met (Ringle Goumltz Wetzels and Wilson 2009) A formative

6 For example job satisfaction can be assessed with general reflective measures such as ―in general I am

satisfied with my job and the formative dimensions can be assessed with measures that describe satisfaction

with each job facet (Edwards 2010 15)

- 22 -

composite in the form of indexes needs to be tested to know the relative weights (γ-

parameters) of the dimensions of PSM In PLS-PM a second-order formative factor is

directly measured by observed variables for all the first order factors While this approach

repeats the number of manifest variables used the model can be identified without adding

outcome constructs that would be considered inappropriate or irrelevant from a conceptual

standpoint (Chin 2010 Edwards 2010) The test result of PLS-PM using Smart PLS 20

(Ringle Wende and Will 2005) produced the function PSM = 31APS + 36CPV + 28COM

+ 31SS However one consideration is whether and how construct error is modeled for the

formative construct (Diamantopoulos 2006) When using PLS-PM the researcher is making

the assumption that there is no error at the construct level as all constructs in PLS-PM are

modeled without error When using PLS-PM the absence of an error term will tend to

increase the weights therefore inflating their importance (Bollen and Lennox 1991) The

primary statistic for assessing a formative indicator is its weight the partialized effect of the

indicator on its intended construct controlling for the effects of all other indicators of that

construct (Cenfetelli and Bassellier 2009) A formative measure is essentially a multiple

regression with the construct representing the dependent variable and the indicators as the

predictors (Diamantopoulos and Winklhofer 2001) PSM may be regarded as a weighted

linear composite of its dimensions The weights used to create the composites are different

across CE-SEM and PLS-PM The causal dimensional weights generated in CB-SEM are

estimated in a model in which a construct-level error term is specified (eg PSM = γ1APM +

γ2CPI + γ3COM + γ4SS + ζ) whereas the weights in PLS-PM are not (eg PSM = γ1APM +

γ2CPI + γ3COM + γ4SS) This leads to two different conceptualizations of the PSM construct

one in which the causal indicators do not capture the construct in its entirety and another in

which they do (Hardin et al 2010) These differences need to be discussed in future research

- 23 -

in this area

One practical suggestion to measure PSM is to apply unit weights to its dimensions if

we assume the effects of the four dimensions are equivalent A literature review indicates that

unit weights have substantial predictive validity when compared with regression weights The

literature shows that the empirical correlations between scores generated from differential and

unit-weighting approaches are quite high In creating composite scores unit weights can be

an appropriate approach for weighting under many circumstances (Bobko Roth and Buster

2007) Although the uses of unit-weight composites will likely result in the explanation of

less variance than empirical estimations of dimensional weights they will remain consistently

defined across studies and hence suitable for theory testing Alternatively more thorough

discussions on the dimensions of PSM may suggest fixed-weight composites (Hardin et al

2010) A previous study (Kim 2010 22) also suggested ―Overall indexing PSM by

measuring all the dimensions of PSM and simply summing the averages of each dimension

can make it easier to measure PSM in practice as PSM = APS + CPV + COM + SS

An advantage of regarding PSM as a second-order formative construct is that the

dimensions can be analyzed both individually and collectively This is important as the

dimensions are expected to have different antecedents and consequences For instance Perry

(1997) found some significant differences in the influence of independent variables on the

four dimensions of PSM Taylor (2007) confirmed that certain dimensions are more important

than others in influencing work outcomes Moynihan and Pandey (2007a) reported that red

tape is negatively related to one dimension but reform orientation and hierarchical authority

is positively related to another dimension A recent study found that the SS dimension has a

positive relationship with occupational intention but the other dimensions do not (Liu et al

2011) Thus it is valuable to analyze the causes and effects of individual dimensions of PSM

- 24 -

as well as of the PSM construct

Conclusion

The purpose of this study is to develop a measurement instrument for PSM for cross-national

research and comparison On the basis of the sharpened construct of PSM (Kim and

Vandenabeele 2010) using data from an international survey conducted in 12 countries we

found that the four-factor structure of PSM is generalizable across country cases This study

provides the measurement index of PSM with the four dimensions of APS CPV COM and

SS The APS dimension measures the instrumental motives of PSM the CPV dimension

determines the value-based motives of PSM while the COM and SS dimensions gauge the

identification motives and self-sacrifice respectively When all four dimensions of PSM are

used in a survey cross-national equivalence of survey items can be assured The items for

measuring PSM in future research are suggested in Appendix 4

This study has several limitations First looking for samples which were comparable

to the largest possible extent we opted for surveys in local government The results found

and discussed in this study are based on PSM among local government employees The

results might be different when surveying central government employees or other specific

groups in the public sector Thus the potential impact of the sample bias needs to be

considered in understanding this study It also means that the survey items for measuring

PSM should be examined with different samples in various areas and at different government

levels

Second the test results showed that there is a high degree of correlation among the

dimension such as APS CPV and COM Future research should not only investigate whether

- 25 -

the items reflect the core content of the dimensions but also continue to develop more

appropriate items for better discriminant validity of each dimension Third the content of

APP dimension was assumed to be distinct from the CPV dimension that focuses on personal

disposition to pursue public interest and public values However the public interest sub-

dimension was separated from the public values sub-dimension in the CPV dimension and

combined with the APP dimension to form the APS dimension Further research should

analyze the CPV dimension and investigate the appropriateness of the APS dimension

In order to facilitate research internationally the PSM construct needs to be

improved conceptually and operationally Kim and Vandenabeele (2010) have dealt with the

conceptual composition and operational dimension of PSM as well as the measurement

model of PSM The next step is to develop an internationally robust measurement instrument

of PSM This study proposes that scholars use all four of the dimensions in measuring the

conceptual components of PSM such as self-sacrifice instrumental value-based and affective

motives because they collectively represent PSM It provides a set of appropriate items for

measuring PSM and confirms the dimensional structure of PSM The survey results suggest

that people in different cultures understand PSM in the same fashion We know that this study

is simply a step along the way to developing a more robust measure but we hope it

contributes to enhanced cross-national research and comparison along with the generation of

cumulative knowledge internationally

- 26 -

REFERENCES

Anderson James C and David D Gerbing 1988 Structural Equation Modeling in Practice

A Review and Recommended Two-Step Approach Psychological Bulletin 103

411-423

Arbuckle James L 2009 Amos 180 Userrsquos Guide Crawfordville FL Amos Development

Co

Beck Joslashrgensen Torben and Barry Bozeman 2007 Public Values An Inventory

Administration amp Society 39(3) 354-381

Beutel Ann M and Margaret Mooney Marini 1995 Gender and values American

Sociological Review 60 436 - 448

Bobko Philip Philip L Roth and Maury A Buster 2007 The Usefulness of Unit Weights in

Creating Composite Scores A Literature Review Application to Content Validity and

Meta-Analysis Organizational Research Methods 10(4) 689-709

Bollen Kenneth A 1989 Structural Equations with Latent Variables New York Wiley

Bollen K A and R Lennox 1991 Conventional Wisdom on Measurement A Structural

Equation Perspective Psychological Bulletin 110(2) 305-314

Byrne Barbara M 2010 Structural Equation Modeling with Amos Basic Concepts

Applications and Programming 2nd

ed New York Routledge

Camilleri Emanuel 2006 Towards Developing an Organizational Commitment ― Public

Service Motivation Model for Maltese Public Service Employees Public Policy

and Administration 21 63-83

Castaing Seacutebastien 2006 The Effects of Psychological Contract Fulfillment and Public

Service Motivation on Organizational Commitment in the French Civil Service

Public Policy and Administration 21 84-98

Cenfetelli Ronald T and Geneviegraveve Bassellier 2009 Interpretation of Formative

Measurement in Information Systems Research MIS Quarterly 33(4) 689-707

Charles Michael B W Martin de Jong and Neal Ryan 2011 Public Values in Western

- 27 -

Europe A Temporal Perspective American Review of Public Administration 41

75-91

Chin Wynne W 2010 How to Write Up and Report PLS Analysis In Handbook of Partical

Least Squares edited by V Esposito Vinzi et al 655-690 Berlin Springer-Verlag

Coursey David H and Sanjay K Pandey 2007a Public Service Motivation Measurement

Testing an Abridged Version of Perryrsquos Proposed Scale Administration amp Society

39 547-568

Coursey David H and Sanjay K Pandey 2007b Content Domain Measurement and

Validity of the Red Tape Concept American Review of Public Administration 37

342-361

Coursey David H James L Perry Jeffrey L Brudney and Laura Littlepage 2008

Psychometric Verification of Perryrsquos Public Service Motivation Instrument Results

for Volunteer Exemplars Review of Public Personnel Administration 28 79-90

Davis Paul and Karen West 2009 What Do Public Values Mean for Public Action Putting

Public Values in Their Plural Place American Review of Public Administration 39

602-618

DeHart-Davis Leisha Justin Marlowe and Sanjay K Pandey 2006 Gender Dimensions of

Public Service Motivation Public Administration Review 66 873-887

Diamantopoulos Adamantios 2006 The error term in formative measurement models

Interpretation and modeling implications Journal of Modeling in Management 1(1)

7-17

Diamantopoulos Adamantios and Judy A Siguaw 2006 Formative versus Reflective

Indicators in Organizational Measure Development A Comparison and Empirical

Illustration British Journal of Management 17 263-282

Diamantopoulos Adamantios and Heidi M Winklhofer 2001 Index Construction with

Formative Indicators An Alternative to Scale Development Journal of Marketing

Research 38 269-277

Edwards Jeffrey R 2010 The Fallacy of Formative Measurement Organizational Research

- 28 -

Methods Online First published on August 19 2010 as

doi1011771094428110378369

Edwards Jeffrey R and Richard P Bagozzi 2000 On the Nature and Direction of

Relationships between Constructs and Measures Psychological Methods 5 155-

174

Flora David B and Patrick J Curran 2004 An Empirical Evaluation of Alternative

Methods of Estimation for Confirmatory Factor Analysis with Ordinal Data

Psychological Methods 9 466-491

Fornell Claes and Fred L Bookstein 1982 Two structural equations models LISREL and

PLS applied to customer exit-voice theory Journal of Marketing Research 19 440-

452

Giauque David Adrian Ritz Freacutedeacuteric Varone Simon Anderfuhren-Biget and Christian

Waldner 2011 Putting Public Service Motivation into Context A Balance between

Universalism and Particularism Forthcoming in International Review of

Administrative Sciences 77(2)

Goumltz Oliver Kerstin Liehr-Gobbers and Manfred Krafft 2010 Evaluation of Structural

Equation Models Using the Partial Least Squares (PLS) Approach In V Esposito

Vinzi et al (eds) Handbook of Partial Least Squares 691-711 DOI 101007978-3-

540-32827-8_30 Springer-Verlag Berlin Heidelberg

Guinot Christiane Julie Latreille and Michel Tenenhaus 2001 PLS Path Modelling and

Multiple Table Analysis Application to the Cosmetic Habits of Women in Ile-de-

France Chemometrics and Intelligent Laboratory Systems 58 247-259

Hair Joseph F Jr William C Black Barry J Babin and Rolph E Anderson 2010

Multivariate Data Analysis 7th

ed Upper Saddle River NJ Pearson Education

Hardin Andrew M Jerry Cha-Jan Chang Mark A Fuller and Gholamreza Torkzadeh 2010

Formative Measurement and Academic Research In Search of Measurement

Theory Educational and Psychological Measurement Online First published on

June 11 2010 as doi101177001316440370208

- 29 -

Howell Roy D Einar Breivik and James B Wilcox 2007a Reconsidering Formative

Measurement Psychological Methods 12 205-218

Howell Roy D Einar Breivik and James B Wilcox 2007b Is Formative Measurement

Really Measurement Reply to Bollen (2007) and Bagozzi (2007) Psychological

Methods 12 238-245

Jarvis Cheryl Burke Scott B MacKenzie and Philip M Podsakoff 2003 A Critical Review

of Construct Indicators and Measurement Model Misspecification in Marketing and

Consumer Research Journal of Consumer Research 30 199-218

Joumlreskog Karl G and Dag Soumlrbom 2005 LISREL 872 (Computer Software) Lincolnwood

IL Scientific Software International

Kim Gimun Bongsik Shin and Varun Grover 2010 Investigating Two Contradictory Views

of Formative Measurement in Information Systems Research MIS Quarterly 34

345-365

Kim Sangmook 2009a Revising Perryrsquos Measurement Scale of Public Service Motivation

American Review of Public Administration 39 149-163

Kim Sangmook 2009b Testing the Structure of Public Service Motivation in Korea A

Research Note Journal of Public Administration Research and Theory 19 839-

851

Kim Sangmook 2010 Testing a Revised Measurement of Public Service Motivation

Reflective versus Formative Specification Journal of Public Administration

Research and Theory Doi101093jopartmuq048

Kim Sangmook and Wouter Vandenabeele 2010 A strategy for building public service

motivation research internationally Public Administration Review 70 701-709

Kline Rex B 2005 Principles and Practice of Structural Equation Modeling 2nd

ed New

York Guilford Press

Law Kenneth and Chi-Sum Wong 1999 Multidimensional Constructs in Structural

Equation Analysis An Illustration using the Job Perception and Job Satisfaction

Constructs Journal of Management 25 143-160

- 30 -

Leisink Peter and Bram Steijn 2009 Public Service Motivation and Job Performance of

Public Sector Employees in the Netherlands International Review of

Administrative Sciences 75 35-52

Liu Bangcheng Chun Hui Jin Hu Wensheng Yang and Xinli Yu 2011 How well can

Public Service Motivation Connect with Occupational Intention International

Review of Administrative Sciences 77 191-211

Liu Bangcheng Ningyu Tang and Xiaomei Zhu 2008 Public Service Motivation and Job

Satisfaction in China An Investigation of Generalisability and Instrumentality

International Journal of Manpower 29 684-699

MacCallum Robert C and Michael W Browne 1993 The Use of Causal Indicators in

Covariance Structure Models Some Practical Issues Psychological Bulletin 114

533-541

MacKenzie Scott B 2003 The dangers of poor construct conceptualization Journal of the

Academy of Marketing Science 31(3) 323-326

MacKenzie Scott B Philip M Podsakoff and Cheryl Burke Jarvis 2005 The Problem of

Measurement Model Misspecification in Behavioral and Organizational Research

and Some Recommended Solutions Journal of Applied Psychology 90 710-730

Martinsen Dorte Sindbjerg and Torben Beck Joslashrgensen 2010 Accountability as a

Differentiated Value in Supranational Governance American Review of Public

Administration 40 742-760

Moynihan Donald P and Sanjay K Pandey 2007a The Role of Organizations in Fostering

Public Service Motivation Public Administration Review 67 40-53

Moynihan Donald P and Sanjay K Pandey 2007b Finding Workable Levers over Work

Motivation Comparing Job Satisfaction Job Involvement and Organizational

Commitment Administration amp Society 39 803-832

Pandey Sanjay K and Edmund C Stazyk 2008 Antecedents and Correlates of Public

Service Motivation In Motivation in Public Management edited by James L Perry

and Annie Hondeghem 101-117 Oxford Oxford University Press

- 31 -

Perry James L 1996 Measuring Public Service Motivation An Assessment of Construct

Reliability and Validity Journal of Public Administration Research and Theory 6

5-22

Perry James L 1997 Antecedents of Public Service Motivation Journal of Public

Administration Research and Theory 7 181-197

Perry James L 2000 Bringing Society in Toward a Theory of Public Service Motivation

Journal of Public Administration Research and Theory 10 471-488

Perry James L and Annie Hondeghem 2008a Motivation in Public Management Oxford

Oxford University Press

Perry James L and Annie Hondeghem 2008b Directions for Future Theory and Research

In Motivation in Public Management edited by James L Perry and Annie

Hondeghem 294-313 Oxford Oxford University Press

Perry James L Annie Hondeghem and Lois Recascino Wise 2010 Revisiting the

Motivational Bases of Public Service Twenty Years of Research and an Agenda for

the Future Public Administration Review 70(5) 681-690

Perry James L and Wouter Vandenabeele 2008 Behavioral dynamics Institutions

Identities and Self-Regulation In Motivation in Public Management edited by

James L Perry and Annie Hondeghem 56-79 Oxford Oxford University Press

Perry James L and Lois Recascino Wise 1990 The Motivational Bases of Public Service

Public Administration Review 50 367-373

Rainey Hal G 1982 Reward Preference among Public and Private Managers In Search of

the Service Ethic American Review of Public Administration 16 288-302

Ringle Christian M Oliver Goumltz Martin Wetzels and Bradley Wilson 2009 On the Use of

Formative Measurement Specifications in Structural Equation Modeling A Monte

Carlo Simulation Study to Compare Covariance-Based and Partial Least Squares

Model Estimation Methodologies Faculty of Economics and Business Administration

Maastricht Univeristy

Ringle C M S Wende and S Will 2005 Smart PLS 20 (M3) Beta

- 32 -

httpwwwsmartplsde

Ritz Adrian 2011 Attraction to Public Policy Making A Qualitative Inquiry into

Improvements in PSM Measurement Forthcoming in Public Administration

Spector Paul E 1992 Summated Rating Scales Construction An Introduction Newbury

Park CA Sage

Steenbergen Marco R 1996 Compassion and American public opinion An analysis of the

NES Humanitarianism Scale Report to the Board of Overseers of the National

Election Studies

Steffen Patrick R and Kevin S Masters 2005 Does compassion mediate the intrinsic

religion-health relationship Annals of Behavioral Medicine 30 217-223

Taylor Jeannette 2007 The Impact of Public Service Motives on Work Outcomes in

Australia A Comparative Multi-Dimensional Analysis Public Administration 85

931-959

Vandenabeele Wouter 2007 Towards a Theory of Public Service Motivation An

Institutional Approach Public Management Review 9 545-556

Vandenabeele Wouter 2008a Government Calling Public Service Motivation as an Element

in Selecting Government as an Employer of Choice Public Administration 86

1089-1105

Vandenabeele Wouter 2008b Development of a Public Service Motivation Measurement

Scale Corroborating and Extending Perryrsquos Measurement Instrument International

Public Management Journal 11 143-167

Vandenabeele Wouter 2009 The Mediating Effect of Job Satisfaction and Organizational

Commitment on Self-Reported Performance More Robust Evidence of the PSM-

Performance Relationship International Review of Administrative Sciences 75 11-

34

Wilcox James B Roy D Howell and Einar Breivik 2008 Questions about Formative

Measurement Journal of Business Research 61 1219-1228

Wise Lois Recascino 2000 The Public Service Culture In Public Administration Concepts

- 33 -

and Cases 7th ed edited by Richard J Stillman II 342-353 Boston Houghton

Mifflin

Wright Bradley E 2008 Methodological Challenges associated with Public Service

Motivation Research In Motivation in Public Management edited by James L

Perry and Annie Hondeghem 80-98 Oxford Oxford University Press

Wright Bradley E and Sanjay K Pandey 2008 Public Service Motivation and the

Assumption of Person-Organization Fit Testing the Mediating Effect of Value

Congruence Administration amp Society 40 502-521

- 34 -

Table 1

Possible Indicators of PSM and Descriptive Statistics (n = 2868)

Dimensions and items Mean Standard

Deviation

Item-Total

Correlation

Attraction to Public Participation (APP)

APP1 I am interested in helping to improve public service

APP2 I am satisfied when I see people benefiting from the public programs I was

involved in

APP3 I like to discuss topics regarding public programs and policies with others

APP4 I believe that public sector activities contribute to our general welfare

APP5 I admire people who initiate or are involved in activities to aid my

community

APP6 Contributing to public programs and policies helps me realize myself

APP7 It is important to contribute to activities that tackle social problems

412

441

355

393

423

351

409

696

662

917

753

672

881

734

524

421

482

412

493

478

552

Commitment to Public Values (CPV)

Sub-dimension for Public Interests (CPV1)

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

CPI3 I would prefer seeing public officials do what is best for the whole

community even if it harmed my interests

CPI4 Serving the public interest is more important than helping a single individual

Sub-dimension for Public Values (CPV2)

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of public

services

CPV3 It is fundamental that public services respond to the needs of the citizens

CPV4 Decisions regarding public services should be democratic despite the time

and effort it takes

CPV5 Everybody is entitled to a good service even if it costs a lot of money

CPV6 It is fundamental that the interests of future generations are taken into

account when developing public policies

CPV7 To act ethically is essential for public servants

CPV8 I believe that public employees must always be aware of the legitimacy of

their activities

CPV9 I personally identify with the aim of protecting individual liberties and rights

418

409

348

337

428

416

433

394

367

426

441

434

414

695

662

899

995

730

704

710

852

932

689

681

692

763

525

596

480

251

484

437

356

325

376

439

462

402

392

Compassion (COM) COM1 It is difficult for me to contain my feelings when I see people in distress

COM2 I feel sympathetic to the plight of the underprivileged

COM3 I empathize with other people who face difficulties

COM4 I have little compassion for people in need who are unwilling to take the

first step to help themselves

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

354

387

388

299

412

391

951

825

707

1083

755

765

399

511

389

011

487

502

Self-Sacrifice (SS)

SS1 Making a difference to society means more to me than personal achievements

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS5 People should give back to society more than they get from it

SS6 Serving other citizens would give me a good feeling even if no one paid me for

it

SS7 I would agree to a good plan to make a better life for the poor even if it costs

me money

357

329

339

302

345

363

350

904

872

949

982

888

1049

895

508

584

454

531

412

368

546

- 35 -

Table 2

Background of Respondents (n = 2868)

Variables Characteristics Respondents

Sex

Male

Female

NA

1209

1622

37

422

566

13

Length of service

(years)

0 ~ 10

10 ~ 20

20 ~ 30

30+

NA

1126

723

484

491

44

393

252

169

171

15

Organizational status

―Do you supervise employees

No

Yes

NA

1472

1076

320

513

375

112

Country

Australia

Belgium

China

Denmark

France

Italy

Korea

Lithuania

The Netherlands

Switzerland

United Kingdom

USA

249

214

230

249

266

162

253

236

249

250

260

250

87

75

80

87

93

56

88

82

87

87

91

87 Note NA = no answer

- 36 -

Table 3

Standardized factor loadings and correlations for the 31-item PSM index (n = 2868)

Dimensions and items SFL (MLE) SFL (D-WLS) Alpha

APP

APP1

APP2

APP4

APP5

APP6

APP7

597

486

455

558

490

600

657

552

514

625

549

665

699

CPV1

CPI1

CPI2

CPI3

CPI4

588

660

488

289

653

732

535

294

542

CPV2

CPV1

CPV2

CPV3

CPV4

CPV5

CPV6

CPV7

CPV8

CPV9

611

546

436

393

418

534

598

496

460

690

631

505

457

486

601

669

573

530

745

COM

COM1

COM2

COM3

COM5

COM6

496

569

498

597

607

516

651

538

670

684

687

SS

SS1

SS2

SS3

SS4

SS5

SS6

SS7

567

736

558

728

467

409

663

660

762

591

695

515

494

709

782

Measures of fit (MLE) χsup2 (df = 424) = 43699 p lt 001 CFI = 835 GFI = 899

RMSEA = 057 [055 059]

Measure of fit (D-WLS) χsup2 (df = 424) = 39247 p lt 001 CFI = 971 GFI = 976

RMSEA = 054 [052 055]

Inter-factor correlations MLE (D-WLS)

1 2 3 4

1 APS

2 CPV1 996 (988)

3 CPV2 902 (890) 798 (798)

4 COM 787 (783) 758 (762) 779 (783)

5 SS 700 (742) 800 (859) 509 (537) 720 (729)

Note SFL = standardized factor loading All factor loadings and correlations are significant at p lt 001

- 37 -

Table 4

Standardized factor loadings and correlations for the 16-item PSM index (n = 2868)

Dimensions and items SFL

(MLE)

SFL

(D-WLS)

APS

APP5 I admire people who initiate or are involved in activities to aid my

community

APP7 It is important to contribute to activities that tackle social problems

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

611

592

637

666

665

669

690

751

CPV

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of

public services

CPV6 It is fundamental that the interests of future generations are taken

into account when developing public policies

CPV7 To act ethically is essential for public servants

612

546

551

590

698

631

618

658

COM

COM1 It is difficult for me to contain my feelings when I see people in

distress

COM2 I feel sympathetic to the plight of the underprivileged

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

490

546

596

604

515

608

669

679

SS

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS7 I would agree to a good plan to make a better life for the poor even if

it costs me money

763

571

766

678

807

629

747

755

Measures of fit (MLE)

χsup2 (df = 98) = 8490 p lt 001

CFI = 939 GFI = 965

RMSEA = 052 [049 055]

Measure of fit (D-WLS)

χsup2 (df = 98) = 6069 p lt 001

CFI = 988 GFI = 992

RMSEA = 043 [039 046]

Alpha CR (D-WLS) Inter-factor correlations MLE (D-WLS)

1 APS 717 788 1 APS 2 CPV 3 COM

2 CPV 663 747 848 (859)

3 COM 645 713 781 (791) 789 (795)

4 SS 784 826 565 (596) 462 (505) 715 (726)

Note SFL = standardized factor loading Alpha = Cronbachrsquos alpha CR = composite reliability

All factor loadings and correlations are significant at p lt 001

- 38 -

Table 5

Results of CFA in the individual countries Standardized factor loadings and measures of fit

Factors

amp Items

Australia Belgium China Denmark France Italy

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

672

756

799

785

804

943

710

703

597

790

799

730

584

675

685

681

731

742

578

748

674

643

765

739

645

704

678

708

812

779

632

604

765

766

672

853

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

552

593

560

699

727

632

624

605

560

713

709

645

702

602

616

724

686

735

567

560

715

654

605

811

608

658

587

696

755

698

673

620

523

782

697

654

COM

COM1

COM2

COM3

COM5

COM6

623

584

662

688

701

739

732

608

475

808

666

581

578

616

605

649

670

673

699

751

737

679

816

776

568

546

801

626

673

850

637

546

741

715

601

822

SS

SS2

SS3

SS4

SS7

654

794

806

892

767

813

727

693

799

793

725

838

857

820

773

896

855

842

806

749

792

829

753

884

847

731

753

889

816

753

815

692

664

795

782

734

Inter-factor

correlations

093

~

752

163

~

810

320

~

773

323

~

825

452

~

827

497

~

855

323

~

850

321

~

847

498

~

899

536

~

902

425

~

829

476

~

865

Sample size 249 214 230 249 266 162

Measures of Fit

χsup2 638 444 881 732 765 554 817 625 735 565 750 564

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 980 100 939 987 969 997 967 994 975 997 949 995

GFI 961 986 941 986 948 991 949 992 955 992 926 986

RMSEA 036 00 063 050 051 026 053 035 045 026 059 033

- 39 -

Table 5 (Continued)

Factors

amp Items

Korea Lithuania The Netherlands Switzerland UK USA

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

619

701

667

706

781

717

666

512

682

751

605

779

557

629

630

620

719

729

535

621

702

643

694

735

655

725

738

710

804

822

535

621

702

643

694

735

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

594

607

722

722

644

722

674

535

533

790

589

711

568

554

538

655

628

545

627

580

615

711

666

704

800

512

555

873

581

646

627

580

615

711

666

704

COM

COM1

COM2

COM3

COM5

COM6

694

677

550

763

736

627

557

639

428

550

696

534

737

623

508

766

616

608

659

592

683

703

661

769

682

583

729

739

644

803

659

592

683

703

661

769

SS

SS2

SS3

SS4

SS7

776

752

679

779

770

811

707

721

635

725

734

706

814

656

743

820

767

764

745

632

689

788

662

744

671

818

832

830

864

802

745

632

689

788

662

744

Inter-factor

correlations

432

~

726

466

~

746

376

~

788

406

~

902

377

~

852

455

~

836

637

~

818

660

~

828

317

~

852

360

~

852

637

~

818

660

~

828

Sample size 253 236 249 250 260 250

Measures of Fit

χsup2 896 573 612 462 773 616 815 715 769 542 815 715

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 951 996 975 100 958 993 957 991 972 998 957 991

GFI 942 988 960 989 951 987 948 990 955 990 948 990

RMSEA 059 028 034 00 050 034 053 044 048 022 053 044

Note All factor loadings and correlations are significant at p lt 001

- 40 -

Appendix 1 Dimensions and possible items of PSM (the first round)

1 Attraction to Public Participation

(1) I am interested in developing public programs helping my country or community (Kim 2009a)

(2) Sharing my views on public policies with others is attractive to me (Kim 2009a)

(3) I am satisfied when seeing people get benefits from the public programs I was involved in (Kim 2009a)

(4) I like to discuss political subjects with others (Giauque et al 2011)

(5) I think the governmental activities contribute to our welfare

(6) I like to initiate actions to help out my community

(7) I do not care much about politicians (Perry 1996)

2 Commitment to Public Values

Possible items for public interest

(1A) Meaningful public service is very important to me (Perry 1996)

(2A) It is important for me to contribute for the common good (Giauque et al 2011)

(3A) I recognize myself with the mission of protection of liberties and rights endorsed by the nation-state

(Giauque et al 2011)

(4A) I would prefer seeing public officials do what is best for the whole community even if it harmed my

interests (Perry 1996)

(5A) To me serving the public interest is more important than helping other people (Vandenabeele 2008b)

Possible items for public values

(1B) I recognize myself with the promotion of the equality of chances endorsed by the nation-state (Giauque et

al 2011)

(2B) It is fundamental that the public service provided to the citizen is regular and continuous (Giauque et al

2011)

(3B) It is fundamental that public service innovates to respond to the new needs of the citizensclients (Giauque

et al 2011)

(4B) The dignity and well-being of all should be the most important concerns in any society (Steenbergen

1996)

(5B) It is important that public servants account for all the costsexpenses they make (Vandenabeele 2008b)

(6B) We have to do everything in our power to pursue the goal of democracy

3 Compassion

(1) It is difficult for me to contain my feelings when I see people in distress (Perry 1996)

(2) I feel sympathetic to the plight of the underprivileged (Kim 2009a)

(3) I am often reminded by daily events how dependent we are on one another (Perry 1996)

(4) I feel compassion for others in the difficulties they are facing (Steffen and Masters 2005)

(5) It is not really my problem if others are in trouble and need help (Beutel and Marini 1995)

(6) I find it hard to be sympathetic toward starving people in foreign lands when there is so much trouble in our

own country (Beutel and Marini 1995)

(7) I get very upset when I see other people treated unfairly (Beutel and Marini 1995)

(8) I would agree to a good plan to make a better life for the poor even if it cost me money (Beutel and Marini

1995)

(9) To me patriotism includes seeing the welfare of others (Perry 1996)

(10) I care very much about other people

4 Self-Sacrifice

(1) Making a difference in society means more to me than personal achievements (Perry 1996)

(2) I am prepared to make enormous sacrifices for the good of society (Perry 1996)

(3) I believe in putting duty before self (Perry 1996)

(4) I am one of those rare people who would risk personal loss to help someone else (Perry 1996)

(5) I think people should give back to society more than they get from it (Perry 1996)

(6) Serving other citizens would give me a good feeling even if no one paid me for it (Perry 1996)

(7) Much of what I do is for a cause bigger than myself (Perry 1996)

- 41 -

Appendix 2 Newly suggested items through the first round

1 Items for the Attraction to Public Participation Dimension

(1) In the public domain every person should be counted as equal

(2) I enjoy to think and argue in a political way

(3) I am satisfied when seeing my ideas being integrated in public policies

(4) I can easily relate to the way of thinking and arguing of politicians

(5) Working in the public sector I can better realize myself

2 Items for the Public Values Sub-Dimension

(1) Public service should give value for money

(2) Cost considerations should play an important role in decisions about public service provision

(3) It is fundamental that the interests of future generations are taken into account when developing public

policies

(4) To act with correct ethics is for public servants as important as their professional competence

(5) Primarily public servants should be accountable to the public and not to their superiors

3 Items for the Self-Sacrifice Dimension

(1) I would agree to a good plan to make a better life for the poor even if it costs me money

- 42 -

Appendix 3 An example of a summary of comments and suggestions (the second round)

On the item ―I am interested in developing public programs helping my country or community

A It is a good sentence but is it possible for everyone to ―develop public programs Even public officials are

limited to do this

B OK

C or the special group

D Double barrel and difficult to translate to my country ―I am interested in helping to improve public services

E Good but I feel that wording requires revision If we draw sample from local government then country is

redundant ―I am interested in developing public policy programs that aim in helping my community

F 4 (General appreciation from1 to 4) Not clear whether it refers to ones work or to ones concern as a citizen

―(In my work) I am interested in contributing to the development of public etchellip

G Instead of development I would use implementing

H Delete public which is redundant because of the characteristic helping my country or community but add

or implementing ―I am interested in developing or implementing programs helping my country or

community

I Referring to dimension Wording Similar to item 6

Note Names have been changed to alphabets in order to ensure the anonymity of participants

- 43 -

Appendix 4 Dimensions and Items of Public Service Motivation for Future Research

Dimensions and items

Attraction to Public Service (APS)

1 I am interested in helping to improve public service

2 I admire people who initiate or are involved in activities to aid my community

3 It is important to contribute to activities that tackle social problems

4 Meaningful public service is very important to me

5 It is important for me to contribute to the common good

Commitment to Public Values (CPV)

1 I think equal opportunities for citizens are very important

2 It is important that citizens can rely on the continuous provision of public services

3 It is fundamental that the interests of future generations are taken into account when

developing public policies

4 To act ethically is essential for public servants

5 I believe that public employees must always be aware of the legitimacy of their activities

6 I personally identify with the aim of protecting individual liberties and rights

Compassion (COM)

1 It is difficult for me to contain my feelings when I see people in distress

2 I feel sympathetic to the plight of the underprivileged

3 I empathize with other people who face difficulties

4 I get very upset when I see other people being treated unfairly

5 Considering the welfare of others is very important

Self-Sacrifice (SS)

1 I am prepared to make sacrifices for the good of society

2 I believe in putting civic duty before self

3 I am willing to risk personal loss to help society

4 I would agree to a good plan to make a better life for the poor even if it costs me money

- 13 -

of estimation within CFA is MLE a technique which assumes that the observed variables are

continuous and normally distributed These assumptions are not met when the observed data

are discrete as occurs when using ordinal scales as in this study A simulation study (Flora

and Curran 2004) found that D-WLS performs well for ordinal measures across a variety of

conditions especially for smaller samples Coursey and Pandey (2007a 2007b) argued the

importance of using D-WLS and not standard MLE in structural equations for ordinal data4

In most cases however the differences between MLE and D-WLS are not substantive to the

conclusions5 No comparison between standard MLE and D-WLS is provided in this paper

We give results from both MLE and D-WLS and this will aid in confirming the stability and

robustness of the results across varying mathematical specifications

For gauging reliability Cronbachrsquos alpha and composite reliability were tested

Composite reliability should be equal to or greater than 07 (Hair et al 2010) Convergent

validity was assessed from the measurement model by determining the statistical significance

of each indicatorrsquos estimated pattern coefficient on its specified underlying construct

Discriminant validity was assessed by determining whether the confidence interval around

the correlation estimate between two factors included 100 (Anderson and Gerbing 1988)

For model fit assessment goodness-of-fit index (GFI) comparative fit index (CFI)

and the root mean square error of approximation (RMSEA) were used The model achieves

an acceptable fit to the data when GFI and CFI equals or exceeds 090 and RMSEA values

4 Applying standard MLE to ordinal data such as Likert-type items may produce significant estimation

problems such as inflation of chi-square fit statistics and biased underestimation of parameters and standard

errors (Flora and Curran 2004) Coursey and Pandey (2007a 2007b) used a modified MLE using the asymptotic

covariance matrix and D-WLS for adjusting the ordinal nature of the data They reported that although the

modified MLE findings are slightly different on a few model parameters the differences are not substantive to

the conclusions However model fit estimators were clearly better for the D-WLS 5 Given normally distributed categorical variables MLE can be used with confidence when a variable has four

or more categories If there is a large sample a five-point Likert scale and the skewness and kurtosis of variables

is within acceptable limits it suggests distribution symmetry ―In most cases where the hypothesized model is

well specified and the scaling based on more than three categories it seems unlikely that there will be much

difference between the findings (Byrne 2010 160)

- 14 -

fall below 008 (Hair et al 2010) In general the larger the value of GFI and CFI and the

smaller the value of RMSEA the better the fit of the model (Bollen 1989) We also present

the χ2 test statistic Lower values of χ

2 indicate a better fit and should be nonsignificant but

the χ2 value increases as sample size increases and thus for large sample sizes this statistic

may lead to rejection of a model with good fit (Hair et al 2010)

Results

Using SPSS 180 descriptive statistics were computed for individual items as shown in table

1 One item (APP3) in the APP dimension ―I like to discuss topics regarding public programs

and policies with others was mistakenly omitted in the survey conducted in the United

States and so it was deleted in the statistical analysis In the assessment of normality no

items showed a skew or kurtosis value greater than the cutoffs of |3| or |8| recommended by

Kline (2005) As data were distributed normally MLE as well as D-WLS estimation was used

Based on inspection of item-total correlations one item (COM4) was dropped from further

analysis because it was weakly correlated with the overall index (011) Missing data is an

issue in CFA as it can cause estimation errors All variables in the survey have missing data

rates of approximately 016 to 143 Mean substitution was used to calculate replacement

values for the missing values in each countryrsquos dataset (Hair et al 2010) The deletion of the

two items left 31 items for the next stage of analysis

The five-correlated factor model was tested using CFA The CFA model in the initial

analysis hypothesized a priori that (1) responses to the 30-item PSM measurement

instrument could be explained by five factors such as APP CPV1 CPV2 COM SS (2) each

item would have a nonzero loading on the PSM factor that it was designed to measure and

- 15 -

zero loadings on all other factors (3) the five factors would be correlated and (4)

measurement error terms would be uncorrelated The resulting CFA with MLE suggested that

the initial five-correlated factor model was not a good fit to the data χsup2 (df = 424) = 43699 p

lt 001 GFI = 0899 CFI = 835 RMSEA = 0057 [055 059] At this stage APP was highly

correlated with CPV1 (r = 996 in MLE and r = 988 in D-WLS) suggesting the two

dimensions were not unique and lacked discriminant validity

[Table 3 about here]

Given the disconfirmation of the initial model an effort was made to estimate an

alternative model The two dimensions APP and CPV1 were combined to form a single

dimension Attraction to Public Service (APS) The items of CPV1 may be regarded as

action-oriented by the respondents because dedication to public service community and

common good is underlined in these items Consequently it is reasonable to combine APP

with CPV1 to form APS a dimension that focuses more on disposition to serve the public to

work for the common good and to participate in public policy processes After the overall

number of dimensions was reduced to four using standardized factor loadings and

modification indices as criteria the items with the lowest factor loading in each dimension

were deleted and the items whose modification index indicated association with multiple

dimensions were eliminated A CFA model with the remaining items was iteratively tested

until it achieved a good fit to the data The outcome of this refinement process was a 16-item

index of four factors Overall model fit by every measure is quite strong for the D-WLS

estimation technique The estimation results for the four-dimension model are presented in

table 4

- 16 -

[Table 4 about here]

Cronbachrsquos coefficient alpha for the 16-item PSM index was 860 and the coefficient

alphas for the four dimensions ranged from 645 to 784 The composite reliability of the set

of reflective indicators for each dimension of PSM ranged from 713 to 826 and indicated

adequate internal consistency The resulting factor structure showed a four-factor structure

with all items loading significantly onto their a priori dimension (p lt 001) and the

standardized factor loadings ranging from 0490 to 0766 in MLE and from 515 to 807 in

D-WLS The results provide support for convergent validity The correlation estimates

between the two factors ranged from 0462 to 0848 in MLE and from 505 to 859 in D-

WLS The confidence intervals (plusmn2 standard errors) around the correlation estimates between

the two factors did not include 100 The result provides support for discriminant validity To

test whether the four-dimension model is superior to other designs such as the two or three-

dimension models a series of CFA were conducted and the results show that the four-

dimension model is a significantly better fit than any of the other designs

To test whether the factorial structure of the PSM instrument is equivalent across the

samples from the twelve countries four nested models were tested as part of multi-group

analysis with MLE (Byrne 2010) First the unconstrained model was the baseline model

which relaxed all equality constraints This model hypothesized the same four-factor structure

of the instrument across twelve countries The fit indices for this baseline model were χsup2 (df =

1176) = 24977 p lt 001 GFI = 0901 CFI = 900 RMSEA = 0020 [019 021] These

indices indicate that the four-factor model of the PSM instrument exhibits acceptable fit

across the samples from the twelve countries Second the measurement weights model tested

- 17 -

the invariance of factor loadings across countries by placing equality constraints on these

parameters The fit indices for this model were χsup2 (df = 1308) = 28088 p lt 001 GFI = 0890

CFI = 887 RMSEA = 0020 [019 021] Since the measurement weights model was nested

within the unconstrained model the chi-square difference text ∆χsup2 (132) = 3111 p lt 001

indicated that some equality constraints of factor loadings did not hold across the twelve

countries Further test results suggested that the equality constraints of structural covariances

and measurement residuals were not upheld across the twelve countries It needs to be

stressed that despite the differences in factor loadings and factor covariances the results

support the main assumption that the same four-factor structure holds across different

countries We may assume that people in different cultures understand PSM in the same

fashion

To find differences in the factor structure of PSM among countries we conducted a

series of CFA with each countryrsquos dataset and followed the same procedure as when

analyzing the sample as a whole As table 5 shows although there are great similarities

between the appropriate items in the countries there are also some differences Even though

the items vary according to national differences the four-factor structure had a good fit to the

data Thus the results confirm that the four-factor structure of PSM can be generalized

[Table 5 about here]

The APS dimension covers both an individualrsquos attraction to public participation and

commitment to public interest but all its items refer to some notion of public interest such as

improving public service aiding community tackling social problems meaningful public

service and contributing to the common good The APS dimension captures an individualrsquos

- 18 -

disposition to serve the public interest through public participation When developing the

items some assumed that the items of committing to public interests can be separated from

those of pursuing public values in the CPV dimension The empirical test showed that

respondents tended to understand committing to public interests as behavior-oriented and so

the items for public interest are not differentiated from participation-oriented items in their

perceptions but are different from those for public values The respondents seem to

understand that pursuing public values is more related to personal beliefs or creeds Thus the

CPV dimension is solely focused on public values and the items of pursing public interests

are combined with those of the APP dimension to make the APS dimension

Public values and their relative importance are neither impervious to change over

time nor the same among all countries (Charles de Jong and Ryan 2011 Davis and West

2009) Moreover some values may be in conflict as Martinsen and Beck Joslashrgensen (2010)

discussed in regard to the tensions between accountability and efficiency It is therefore not

possible to include all public values in one dimension (Beck Joslashrgensen and Bozeman 2007)

Accordingly the items of the CPV dimension are developed for estimating individual

commitment to seek public values which there are general consensus about and which are

generally formulated

When all four dimensions of PSM are included in a survey cross-national

equivalence of the survey can be assured The items represent reflections or manifestations of

each dimension and hence are relatively homogeneous and interchangeable Adding or

deleting items does not change the essential nature of the dimension Thus any combination

of items in a dimension can be regarded as equivalent When using structural equation

modeling (SEM) each dimension needs to have at least three reflective indicators in order to

achieve identification or it needs to have at least two reflective indicators and be allowed to

- 19 -

covary with another dimension that has at least two reflective measures With one indicator

the dimension is identified if the loading and the variance of the uniqueness are both fixed

(Bollen 1989 Edwards 2010) Thus the number of items in each dimension is not restricted

when measuring PSM However we recommend using multiple items for each dimension

because more research is needed to investigate the appropriateness of this PSM instrument

with varying samples

The test results showed that model fit estimators were clearly better for the D-WLS

It means that in this study MLE was more rigorous than D-WLS in assessing the fit of the

model The empirical evidence indicated that MLE may be an acceptable technique for

ordinal data when there is a large sample a five-point Likert scale and the skewness and

kurtosis of variables is within acceptable limits as suggested by Byrne (2010)

Discussion

The analysis reaffirmed the four-dimensional structure of PSM but the results were not

entirely consistent with the original CFA we tested We were ultimately able to converge on a

good-fit operational model for PSM that was also consistent with underlying theory The

analysis is not able to resolve the debate about measurement models however for reasons

discussed below

Choosing the correct measurement model (ie reflective vs formative) is important

as measurement model misspecification results not only in misrepresentation of the epistemic

relationships between constructs and their measures but also in biased parameter estimates of

structural relationships (Jarvis et al 2003 Law and Wong 1999 MacKenzie 2003

MacKenzie Podsakoff and Jarvis 2005) A formative measurement model in isolation is

- 20 -

under-identified and cannot be estimated in the case of covariance-based SEM (CB-SEM)

(Bollen 1989) A construct with formative indicators must emit paths to at least two unrelated

reflective indicators two unrelated latent constructs or some combination of the two in order

to solve the problem of under-identification (Jarvis MacKenzie and Podsakoff 2003

MacCallum and Browne 1993 MacKenzie Podsakoff and Jarvis 2005) Thus ―when

dependent constructs change the empirical nature of the formatively measured construct

changes (Wilcox Howell and Breivik 2008 1223) and ―when specifying formative

measurement models what might seem like an innocuous choice of outcomes to achieve

identification can have fundamental ramifications for the meaning of the formative construct

and the inferences based on it (Edwards 2010 7)

Earlier research found that the association between a construct and its measures

should remain stable regardless of the larger causal model in which the constructs and

measures are embedded (Edwards and Bagozzi 2000) The association between a construct

and its measures is generally stable for reflective measures but associations of formative

measures with their construct are unstable The susceptibility of formative measurement to

both interpretational confounding and external inconsistency recommends that formative

measurement is not an equally attractive alternative to reflective measurement and that

whenever possible researchers should opt for reflective measurement (Howell Breivik and

Wilcox 2007a 2007b Kim Shin and Grover 2010) In the case of second-order models

researchers are limited to only a second-order reflective model in CB-SEM (Chin 2010)

Regardless of the measurement debate and the tradeoffs between reflective or

formative specification this study makes a theoretical argument that PSM is a second-order

formative construct This second-order formative factor model (reflective first-order

formative second-order) is also suggested as an alternative model that captures the

- 21 -

multidimensionality of complex concepts and avoids the shortcomings of first-order

formative measurement (Edwards 2010) However this model itself cannot achieve

identification One possible strategy is to develop direct reflective measures of PSM based on

the view that PSM is a general concept that is not only influenced by its specific dimensions

but also can be assessed using general measures (Edwards 2010 Javis MacKenzie and

Podsakoff 2003)6 However these general measures have not yet been developed for the

PSM construct

Another way around the problem of underidentification of formative models is to

apply the component-based approach known as partial least squares path modeling (PLS-PM)

which can model formative indicators (Fornell and Bookstein 1982 Goumltz Liehr-Gobbers

and Krafft 2010) The CB-SEM and the PLS-PM methods have distinctive statistical

characteristics and selecting an approach to SEM depends on the particular research situation

CB-SEM is the method of choice for theory testing while PLS-PM is appropriate for

prediction-oriented applications (Ringle Goumltz Wetzels and Wilson 2009) One advantage of

the PLS method is that it demands significantly fewer requirements compared to that of

covariance structure analyses but nevertheless delivers consistent estimation results Another

asset of the PLS approach is its ability to deal with formative as well as reflective indicators

even within one structural equation model This indicates that the PLS approach can be

appropriate for explorative analysis of structural equation models (Chin 2010 Goumltz Liehr-

Gobbers and Krafft 2010) A Monte Carlo simulation study to compare CB-SEM and PLS-

PM showed that PLS-PM provides a viable approximation of model parameters when the

prerequisites for CB-SEM are not met (Ringle Goumltz Wetzels and Wilson 2009) A formative

6 For example job satisfaction can be assessed with general reflective measures such as ―in general I am

satisfied with my job and the formative dimensions can be assessed with measures that describe satisfaction

with each job facet (Edwards 2010 15)

- 22 -

composite in the form of indexes needs to be tested to know the relative weights (γ-

parameters) of the dimensions of PSM In PLS-PM a second-order formative factor is

directly measured by observed variables for all the first order factors While this approach

repeats the number of manifest variables used the model can be identified without adding

outcome constructs that would be considered inappropriate or irrelevant from a conceptual

standpoint (Chin 2010 Edwards 2010) The test result of PLS-PM using Smart PLS 20

(Ringle Wende and Will 2005) produced the function PSM = 31APS + 36CPV + 28COM

+ 31SS However one consideration is whether and how construct error is modeled for the

formative construct (Diamantopoulos 2006) When using PLS-PM the researcher is making

the assumption that there is no error at the construct level as all constructs in PLS-PM are

modeled without error When using PLS-PM the absence of an error term will tend to

increase the weights therefore inflating their importance (Bollen and Lennox 1991) The

primary statistic for assessing a formative indicator is its weight the partialized effect of the

indicator on its intended construct controlling for the effects of all other indicators of that

construct (Cenfetelli and Bassellier 2009) A formative measure is essentially a multiple

regression with the construct representing the dependent variable and the indicators as the

predictors (Diamantopoulos and Winklhofer 2001) PSM may be regarded as a weighted

linear composite of its dimensions The weights used to create the composites are different

across CE-SEM and PLS-PM The causal dimensional weights generated in CB-SEM are

estimated in a model in which a construct-level error term is specified (eg PSM = γ1APM +

γ2CPI + γ3COM + γ4SS + ζ) whereas the weights in PLS-PM are not (eg PSM = γ1APM +

γ2CPI + γ3COM + γ4SS) This leads to two different conceptualizations of the PSM construct

one in which the causal indicators do not capture the construct in its entirety and another in

which they do (Hardin et al 2010) These differences need to be discussed in future research

- 23 -

in this area

One practical suggestion to measure PSM is to apply unit weights to its dimensions if

we assume the effects of the four dimensions are equivalent A literature review indicates that

unit weights have substantial predictive validity when compared with regression weights The

literature shows that the empirical correlations between scores generated from differential and

unit-weighting approaches are quite high In creating composite scores unit weights can be

an appropriate approach for weighting under many circumstances (Bobko Roth and Buster

2007) Although the uses of unit-weight composites will likely result in the explanation of

less variance than empirical estimations of dimensional weights they will remain consistently

defined across studies and hence suitable for theory testing Alternatively more thorough

discussions on the dimensions of PSM may suggest fixed-weight composites (Hardin et al

2010) A previous study (Kim 2010 22) also suggested ―Overall indexing PSM by

measuring all the dimensions of PSM and simply summing the averages of each dimension

can make it easier to measure PSM in practice as PSM = APS + CPV + COM + SS

An advantage of regarding PSM as a second-order formative construct is that the

dimensions can be analyzed both individually and collectively This is important as the

dimensions are expected to have different antecedents and consequences For instance Perry

(1997) found some significant differences in the influence of independent variables on the

four dimensions of PSM Taylor (2007) confirmed that certain dimensions are more important

than others in influencing work outcomes Moynihan and Pandey (2007a) reported that red

tape is negatively related to one dimension but reform orientation and hierarchical authority

is positively related to another dimension A recent study found that the SS dimension has a

positive relationship with occupational intention but the other dimensions do not (Liu et al

2011) Thus it is valuable to analyze the causes and effects of individual dimensions of PSM

- 24 -

as well as of the PSM construct

Conclusion

The purpose of this study is to develop a measurement instrument for PSM for cross-national

research and comparison On the basis of the sharpened construct of PSM (Kim and

Vandenabeele 2010) using data from an international survey conducted in 12 countries we

found that the four-factor structure of PSM is generalizable across country cases This study

provides the measurement index of PSM with the four dimensions of APS CPV COM and

SS The APS dimension measures the instrumental motives of PSM the CPV dimension

determines the value-based motives of PSM while the COM and SS dimensions gauge the

identification motives and self-sacrifice respectively When all four dimensions of PSM are

used in a survey cross-national equivalence of survey items can be assured The items for

measuring PSM in future research are suggested in Appendix 4

This study has several limitations First looking for samples which were comparable

to the largest possible extent we opted for surveys in local government The results found

and discussed in this study are based on PSM among local government employees The

results might be different when surveying central government employees or other specific

groups in the public sector Thus the potential impact of the sample bias needs to be

considered in understanding this study It also means that the survey items for measuring

PSM should be examined with different samples in various areas and at different government

levels

Second the test results showed that there is a high degree of correlation among the

dimension such as APS CPV and COM Future research should not only investigate whether

- 25 -

the items reflect the core content of the dimensions but also continue to develop more

appropriate items for better discriminant validity of each dimension Third the content of

APP dimension was assumed to be distinct from the CPV dimension that focuses on personal

disposition to pursue public interest and public values However the public interest sub-

dimension was separated from the public values sub-dimension in the CPV dimension and

combined with the APP dimension to form the APS dimension Further research should

analyze the CPV dimension and investigate the appropriateness of the APS dimension

In order to facilitate research internationally the PSM construct needs to be

improved conceptually and operationally Kim and Vandenabeele (2010) have dealt with the

conceptual composition and operational dimension of PSM as well as the measurement

model of PSM The next step is to develop an internationally robust measurement instrument

of PSM This study proposes that scholars use all four of the dimensions in measuring the

conceptual components of PSM such as self-sacrifice instrumental value-based and affective

motives because they collectively represent PSM It provides a set of appropriate items for

measuring PSM and confirms the dimensional structure of PSM The survey results suggest

that people in different cultures understand PSM in the same fashion We know that this study

is simply a step along the way to developing a more robust measure but we hope it

contributes to enhanced cross-national research and comparison along with the generation of

cumulative knowledge internationally

- 26 -

REFERENCES

Anderson James C and David D Gerbing 1988 Structural Equation Modeling in Practice

A Review and Recommended Two-Step Approach Psychological Bulletin 103

411-423

Arbuckle James L 2009 Amos 180 Userrsquos Guide Crawfordville FL Amos Development

Co

Beck Joslashrgensen Torben and Barry Bozeman 2007 Public Values An Inventory

Administration amp Society 39(3) 354-381

Beutel Ann M and Margaret Mooney Marini 1995 Gender and values American

Sociological Review 60 436 - 448

Bobko Philip Philip L Roth and Maury A Buster 2007 The Usefulness of Unit Weights in

Creating Composite Scores A Literature Review Application to Content Validity and

Meta-Analysis Organizational Research Methods 10(4) 689-709

Bollen Kenneth A 1989 Structural Equations with Latent Variables New York Wiley

Bollen K A and R Lennox 1991 Conventional Wisdom on Measurement A Structural

Equation Perspective Psychological Bulletin 110(2) 305-314

Byrne Barbara M 2010 Structural Equation Modeling with Amos Basic Concepts

Applications and Programming 2nd

ed New York Routledge

Camilleri Emanuel 2006 Towards Developing an Organizational Commitment ― Public

Service Motivation Model for Maltese Public Service Employees Public Policy

and Administration 21 63-83

Castaing Seacutebastien 2006 The Effects of Psychological Contract Fulfillment and Public

Service Motivation on Organizational Commitment in the French Civil Service

Public Policy and Administration 21 84-98

Cenfetelli Ronald T and Geneviegraveve Bassellier 2009 Interpretation of Formative

Measurement in Information Systems Research MIS Quarterly 33(4) 689-707

Charles Michael B W Martin de Jong and Neal Ryan 2011 Public Values in Western

- 27 -

Europe A Temporal Perspective American Review of Public Administration 41

75-91

Chin Wynne W 2010 How to Write Up and Report PLS Analysis In Handbook of Partical

Least Squares edited by V Esposito Vinzi et al 655-690 Berlin Springer-Verlag

Coursey David H and Sanjay K Pandey 2007a Public Service Motivation Measurement

Testing an Abridged Version of Perryrsquos Proposed Scale Administration amp Society

39 547-568

Coursey David H and Sanjay K Pandey 2007b Content Domain Measurement and

Validity of the Red Tape Concept American Review of Public Administration 37

342-361

Coursey David H James L Perry Jeffrey L Brudney and Laura Littlepage 2008

Psychometric Verification of Perryrsquos Public Service Motivation Instrument Results

for Volunteer Exemplars Review of Public Personnel Administration 28 79-90

Davis Paul and Karen West 2009 What Do Public Values Mean for Public Action Putting

Public Values in Their Plural Place American Review of Public Administration 39

602-618

DeHart-Davis Leisha Justin Marlowe and Sanjay K Pandey 2006 Gender Dimensions of

Public Service Motivation Public Administration Review 66 873-887

Diamantopoulos Adamantios 2006 The error term in formative measurement models

Interpretation and modeling implications Journal of Modeling in Management 1(1)

7-17

Diamantopoulos Adamantios and Judy A Siguaw 2006 Formative versus Reflective

Indicators in Organizational Measure Development A Comparison and Empirical

Illustration British Journal of Management 17 263-282

Diamantopoulos Adamantios and Heidi M Winklhofer 2001 Index Construction with

Formative Indicators An Alternative to Scale Development Journal of Marketing

Research 38 269-277

Edwards Jeffrey R 2010 The Fallacy of Formative Measurement Organizational Research

- 28 -

Methods Online First published on August 19 2010 as

doi1011771094428110378369

Edwards Jeffrey R and Richard P Bagozzi 2000 On the Nature and Direction of

Relationships between Constructs and Measures Psychological Methods 5 155-

174

Flora David B and Patrick J Curran 2004 An Empirical Evaluation of Alternative

Methods of Estimation for Confirmatory Factor Analysis with Ordinal Data

Psychological Methods 9 466-491

Fornell Claes and Fred L Bookstein 1982 Two structural equations models LISREL and

PLS applied to customer exit-voice theory Journal of Marketing Research 19 440-

452

Giauque David Adrian Ritz Freacutedeacuteric Varone Simon Anderfuhren-Biget and Christian

Waldner 2011 Putting Public Service Motivation into Context A Balance between

Universalism and Particularism Forthcoming in International Review of

Administrative Sciences 77(2)

Goumltz Oliver Kerstin Liehr-Gobbers and Manfred Krafft 2010 Evaluation of Structural

Equation Models Using the Partial Least Squares (PLS) Approach In V Esposito

Vinzi et al (eds) Handbook of Partial Least Squares 691-711 DOI 101007978-3-

540-32827-8_30 Springer-Verlag Berlin Heidelberg

Guinot Christiane Julie Latreille and Michel Tenenhaus 2001 PLS Path Modelling and

Multiple Table Analysis Application to the Cosmetic Habits of Women in Ile-de-

France Chemometrics and Intelligent Laboratory Systems 58 247-259

Hair Joseph F Jr William C Black Barry J Babin and Rolph E Anderson 2010

Multivariate Data Analysis 7th

ed Upper Saddle River NJ Pearson Education

Hardin Andrew M Jerry Cha-Jan Chang Mark A Fuller and Gholamreza Torkzadeh 2010

Formative Measurement and Academic Research In Search of Measurement

Theory Educational and Psychological Measurement Online First published on

June 11 2010 as doi101177001316440370208

- 29 -

Howell Roy D Einar Breivik and James B Wilcox 2007a Reconsidering Formative

Measurement Psychological Methods 12 205-218

Howell Roy D Einar Breivik and James B Wilcox 2007b Is Formative Measurement

Really Measurement Reply to Bollen (2007) and Bagozzi (2007) Psychological

Methods 12 238-245

Jarvis Cheryl Burke Scott B MacKenzie and Philip M Podsakoff 2003 A Critical Review

of Construct Indicators and Measurement Model Misspecification in Marketing and

Consumer Research Journal of Consumer Research 30 199-218

Joumlreskog Karl G and Dag Soumlrbom 2005 LISREL 872 (Computer Software) Lincolnwood

IL Scientific Software International

Kim Gimun Bongsik Shin and Varun Grover 2010 Investigating Two Contradictory Views

of Formative Measurement in Information Systems Research MIS Quarterly 34

345-365

Kim Sangmook 2009a Revising Perryrsquos Measurement Scale of Public Service Motivation

American Review of Public Administration 39 149-163

Kim Sangmook 2009b Testing the Structure of Public Service Motivation in Korea A

Research Note Journal of Public Administration Research and Theory 19 839-

851

Kim Sangmook 2010 Testing a Revised Measurement of Public Service Motivation

Reflective versus Formative Specification Journal of Public Administration

Research and Theory Doi101093jopartmuq048

Kim Sangmook and Wouter Vandenabeele 2010 A strategy for building public service

motivation research internationally Public Administration Review 70 701-709

Kline Rex B 2005 Principles and Practice of Structural Equation Modeling 2nd

ed New

York Guilford Press

Law Kenneth and Chi-Sum Wong 1999 Multidimensional Constructs in Structural

Equation Analysis An Illustration using the Job Perception and Job Satisfaction

Constructs Journal of Management 25 143-160

- 30 -

Leisink Peter and Bram Steijn 2009 Public Service Motivation and Job Performance of

Public Sector Employees in the Netherlands International Review of

Administrative Sciences 75 35-52

Liu Bangcheng Chun Hui Jin Hu Wensheng Yang and Xinli Yu 2011 How well can

Public Service Motivation Connect with Occupational Intention International

Review of Administrative Sciences 77 191-211

Liu Bangcheng Ningyu Tang and Xiaomei Zhu 2008 Public Service Motivation and Job

Satisfaction in China An Investigation of Generalisability and Instrumentality

International Journal of Manpower 29 684-699

MacCallum Robert C and Michael W Browne 1993 The Use of Causal Indicators in

Covariance Structure Models Some Practical Issues Psychological Bulletin 114

533-541

MacKenzie Scott B 2003 The dangers of poor construct conceptualization Journal of the

Academy of Marketing Science 31(3) 323-326

MacKenzie Scott B Philip M Podsakoff and Cheryl Burke Jarvis 2005 The Problem of

Measurement Model Misspecification in Behavioral and Organizational Research

and Some Recommended Solutions Journal of Applied Psychology 90 710-730

Martinsen Dorte Sindbjerg and Torben Beck Joslashrgensen 2010 Accountability as a

Differentiated Value in Supranational Governance American Review of Public

Administration 40 742-760

Moynihan Donald P and Sanjay K Pandey 2007a The Role of Organizations in Fostering

Public Service Motivation Public Administration Review 67 40-53

Moynihan Donald P and Sanjay K Pandey 2007b Finding Workable Levers over Work

Motivation Comparing Job Satisfaction Job Involvement and Organizational

Commitment Administration amp Society 39 803-832

Pandey Sanjay K and Edmund C Stazyk 2008 Antecedents and Correlates of Public

Service Motivation In Motivation in Public Management edited by James L Perry

and Annie Hondeghem 101-117 Oxford Oxford University Press

- 31 -

Perry James L 1996 Measuring Public Service Motivation An Assessment of Construct

Reliability and Validity Journal of Public Administration Research and Theory 6

5-22

Perry James L 1997 Antecedents of Public Service Motivation Journal of Public

Administration Research and Theory 7 181-197

Perry James L 2000 Bringing Society in Toward a Theory of Public Service Motivation

Journal of Public Administration Research and Theory 10 471-488

Perry James L and Annie Hondeghem 2008a Motivation in Public Management Oxford

Oxford University Press

Perry James L and Annie Hondeghem 2008b Directions for Future Theory and Research

In Motivation in Public Management edited by James L Perry and Annie

Hondeghem 294-313 Oxford Oxford University Press

Perry James L Annie Hondeghem and Lois Recascino Wise 2010 Revisiting the

Motivational Bases of Public Service Twenty Years of Research and an Agenda for

the Future Public Administration Review 70(5) 681-690

Perry James L and Wouter Vandenabeele 2008 Behavioral dynamics Institutions

Identities and Self-Regulation In Motivation in Public Management edited by

James L Perry and Annie Hondeghem 56-79 Oxford Oxford University Press

Perry James L and Lois Recascino Wise 1990 The Motivational Bases of Public Service

Public Administration Review 50 367-373

Rainey Hal G 1982 Reward Preference among Public and Private Managers In Search of

the Service Ethic American Review of Public Administration 16 288-302

Ringle Christian M Oliver Goumltz Martin Wetzels and Bradley Wilson 2009 On the Use of

Formative Measurement Specifications in Structural Equation Modeling A Monte

Carlo Simulation Study to Compare Covariance-Based and Partial Least Squares

Model Estimation Methodologies Faculty of Economics and Business Administration

Maastricht Univeristy

Ringle C M S Wende and S Will 2005 Smart PLS 20 (M3) Beta

- 32 -

httpwwwsmartplsde

Ritz Adrian 2011 Attraction to Public Policy Making A Qualitative Inquiry into

Improvements in PSM Measurement Forthcoming in Public Administration

Spector Paul E 1992 Summated Rating Scales Construction An Introduction Newbury

Park CA Sage

Steenbergen Marco R 1996 Compassion and American public opinion An analysis of the

NES Humanitarianism Scale Report to the Board of Overseers of the National

Election Studies

Steffen Patrick R and Kevin S Masters 2005 Does compassion mediate the intrinsic

religion-health relationship Annals of Behavioral Medicine 30 217-223

Taylor Jeannette 2007 The Impact of Public Service Motives on Work Outcomes in

Australia A Comparative Multi-Dimensional Analysis Public Administration 85

931-959

Vandenabeele Wouter 2007 Towards a Theory of Public Service Motivation An

Institutional Approach Public Management Review 9 545-556

Vandenabeele Wouter 2008a Government Calling Public Service Motivation as an Element

in Selecting Government as an Employer of Choice Public Administration 86

1089-1105

Vandenabeele Wouter 2008b Development of a Public Service Motivation Measurement

Scale Corroborating and Extending Perryrsquos Measurement Instrument International

Public Management Journal 11 143-167

Vandenabeele Wouter 2009 The Mediating Effect of Job Satisfaction and Organizational

Commitment on Self-Reported Performance More Robust Evidence of the PSM-

Performance Relationship International Review of Administrative Sciences 75 11-

34

Wilcox James B Roy D Howell and Einar Breivik 2008 Questions about Formative

Measurement Journal of Business Research 61 1219-1228

Wise Lois Recascino 2000 The Public Service Culture In Public Administration Concepts

- 33 -

and Cases 7th ed edited by Richard J Stillman II 342-353 Boston Houghton

Mifflin

Wright Bradley E 2008 Methodological Challenges associated with Public Service

Motivation Research In Motivation in Public Management edited by James L

Perry and Annie Hondeghem 80-98 Oxford Oxford University Press

Wright Bradley E and Sanjay K Pandey 2008 Public Service Motivation and the

Assumption of Person-Organization Fit Testing the Mediating Effect of Value

Congruence Administration amp Society 40 502-521

- 34 -

Table 1

Possible Indicators of PSM and Descriptive Statistics (n = 2868)

Dimensions and items Mean Standard

Deviation

Item-Total

Correlation

Attraction to Public Participation (APP)

APP1 I am interested in helping to improve public service

APP2 I am satisfied when I see people benefiting from the public programs I was

involved in

APP3 I like to discuss topics regarding public programs and policies with others

APP4 I believe that public sector activities contribute to our general welfare

APP5 I admire people who initiate or are involved in activities to aid my

community

APP6 Contributing to public programs and policies helps me realize myself

APP7 It is important to contribute to activities that tackle social problems

412

441

355

393

423

351

409

696

662

917

753

672

881

734

524

421

482

412

493

478

552

Commitment to Public Values (CPV)

Sub-dimension for Public Interests (CPV1)

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

CPI3 I would prefer seeing public officials do what is best for the whole

community even if it harmed my interests

CPI4 Serving the public interest is more important than helping a single individual

Sub-dimension for Public Values (CPV2)

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of public

services

CPV3 It is fundamental that public services respond to the needs of the citizens

CPV4 Decisions regarding public services should be democratic despite the time

and effort it takes

CPV5 Everybody is entitled to a good service even if it costs a lot of money

CPV6 It is fundamental that the interests of future generations are taken into

account when developing public policies

CPV7 To act ethically is essential for public servants

CPV8 I believe that public employees must always be aware of the legitimacy of

their activities

CPV9 I personally identify with the aim of protecting individual liberties and rights

418

409

348

337

428

416

433

394

367

426

441

434

414

695

662

899

995

730

704

710

852

932

689

681

692

763

525

596

480

251

484

437

356

325

376

439

462

402

392

Compassion (COM) COM1 It is difficult for me to contain my feelings when I see people in distress

COM2 I feel sympathetic to the plight of the underprivileged

COM3 I empathize with other people who face difficulties

COM4 I have little compassion for people in need who are unwilling to take the

first step to help themselves

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

354

387

388

299

412

391

951

825

707

1083

755

765

399

511

389

011

487

502

Self-Sacrifice (SS)

SS1 Making a difference to society means more to me than personal achievements

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS5 People should give back to society more than they get from it

SS6 Serving other citizens would give me a good feeling even if no one paid me for

it

SS7 I would agree to a good plan to make a better life for the poor even if it costs

me money

357

329

339

302

345

363

350

904

872

949

982

888

1049

895

508

584

454

531

412

368

546

- 35 -

Table 2

Background of Respondents (n = 2868)

Variables Characteristics Respondents

Sex

Male

Female

NA

1209

1622

37

422

566

13

Length of service

(years)

0 ~ 10

10 ~ 20

20 ~ 30

30+

NA

1126

723

484

491

44

393

252

169

171

15

Organizational status

―Do you supervise employees

No

Yes

NA

1472

1076

320

513

375

112

Country

Australia

Belgium

China

Denmark

France

Italy

Korea

Lithuania

The Netherlands

Switzerland

United Kingdom

USA

249

214

230

249

266

162

253

236

249

250

260

250

87

75

80

87

93

56

88

82

87

87

91

87 Note NA = no answer

- 36 -

Table 3

Standardized factor loadings and correlations for the 31-item PSM index (n = 2868)

Dimensions and items SFL (MLE) SFL (D-WLS) Alpha

APP

APP1

APP2

APP4

APP5

APP6

APP7

597

486

455

558

490

600

657

552

514

625

549

665

699

CPV1

CPI1

CPI2

CPI3

CPI4

588

660

488

289

653

732

535

294

542

CPV2

CPV1

CPV2

CPV3

CPV4

CPV5

CPV6

CPV7

CPV8

CPV9

611

546

436

393

418

534

598

496

460

690

631

505

457

486

601

669

573

530

745

COM

COM1

COM2

COM3

COM5

COM6

496

569

498

597

607

516

651

538

670

684

687

SS

SS1

SS2

SS3

SS4

SS5

SS6

SS7

567

736

558

728

467

409

663

660

762

591

695

515

494

709

782

Measures of fit (MLE) χsup2 (df = 424) = 43699 p lt 001 CFI = 835 GFI = 899

RMSEA = 057 [055 059]

Measure of fit (D-WLS) χsup2 (df = 424) = 39247 p lt 001 CFI = 971 GFI = 976

RMSEA = 054 [052 055]

Inter-factor correlations MLE (D-WLS)

1 2 3 4

1 APS

2 CPV1 996 (988)

3 CPV2 902 (890) 798 (798)

4 COM 787 (783) 758 (762) 779 (783)

5 SS 700 (742) 800 (859) 509 (537) 720 (729)

Note SFL = standardized factor loading All factor loadings and correlations are significant at p lt 001

- 37 -

Table 4

Standardized factor loadings and correlations for the 16-item PSM index (n = 2868)

Dimensions and items SFL

(MLE)

SFL

(D-WLS)

APS

APP5 I admire people who initiate or are involved in activities to aid my

community

APP7 It is important to contribute to activities that tackle social problems

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

611

592

637

666

665

669

690

751

CPV

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of

public services

CPV6 It is fundamental that the interests of future generations are taken

into account when developing public policies

CPV7 To act ethically is essential for public servants

612

546

551

590

698

631

618

658

COM

COM1 It is difficult for me to contain my feelings when I see people in

distress

COM2 I feel sympathetic to the plight of the underprivileged

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

490

546

596

604

515

608

669

679

SS

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS7 I would agree to a good plan to make a better life for the poor even if

it costs me money

763

571

766

678

807

629

747

755

Measures of fit (MLE)

χsup2 (df = 98) = 8490 p lt 001

CFI = 939 GFI = 965

RMSEA = 052 [049 055]

Measure of fit (D-WLS)

χsup2 (df = 98) = 6069 p lt 001

CFI = 988 GFI = 992

RMSEA = 043 [039 046]

Alpha CR (D-WLS) Inter-factor correlations MLE (D-WLS)

1 APS 717 788 1 APS 2 CPV 3 COM

2 CPV 663 747 848 (859)

3 COM 645 713 781 (791) 789 (795)

4 SS 784 826 565 (596) 462 (505) 715 (726)

Note SFL = standardized factor loading Alpha = Cronbachrsquos alpha CR = composite reliability

All factor loadings and correlations are significant at p lt 001

- 38 -

Table 5

Results of CFA in the individual countries Standardized factor loadings and measures of fit

Factors

amp Items

Australia Belgium China Denmark France Italy

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

672

756

799

785

804

943

710

703

597

790

799

730

584

675

685

681

731

742

578

748

674

643

765

739

645

704

678

708

812

779

632

604

765

766

672

853

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

552

593

560

699

727

632

624

605

560

713

709

645

702

602

616

724

686

735

567

560

715

654

605

811

608

658

587

696

755

698

673

620

523

782

697

654

COM

COM1

COM2

COM3

COM5

COM6

623

584

662

688

701

739

732

608

475

808

666

581

578

616

605

649

670

673

699

751

737

679

816

776

568

546

801

626

673

850

637

546

741

715

601

822

SS

SS2

SS3

SS4

SS7

654

794

806

892

767

813

727

693

799

793

725

838

857

820

773

896

855

842

806

749

792

829

753

884

847

731

753

889

816

753

815

692

664

795

782

734

Inter-factor

correlations

093

~

752

163

~

810

320

~

773

323

~

825

452

~

827

497

~

855

323

~

850

321

~

847

498

~

899

536

~

902

425

~

829

476

~

865

Sample size 249 214 230 249 266 162

Measures of Fit

χsup2 638 444 881 732 765 554 817 625 735 565 750 564

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 980 100 939 987 969 997 967 994 975 997 949 995

GFI 961 986 941 986 948 991 949 992 955 992 926 986

RMSEA 036 00 063 050 051 026 053 035 045 026 059 033

- 39 -

Table 5 (Continued)

Factors

amp Items

Korea Lithuania The Netherlands Switzerland UK USA

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

619

701

667

706

781

717

666

512

682

751

605

779

557

629

630

620

719

729

535

621

702

643

694

735

655

725

738

710

804

822

535

621

702

643

694

735

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

594

607

722

722

644

722

674

535

533

790

589

711

568

554

538

655

628

545

627

580

615

711

666

704

800

512

555

873

581

646

627

580

615

711

666

704

COM

COM1

COM2

COM3

COM5

COM6

694

677

550

763

736

627

557

639

428

550

696

534

737

623

508

766

616

608

659

592

683

703

661

769

682

583

729

739

644

803

659

592

683

703

661

769

SS

SS2

SS3

SS4

SS7

776

752

679

779

770

811

707

721

635

725

734

706

814

656

743

820

767

764

745

632

689

788

662

744

671

818

832

830

864

802

745

632

689

788

662

744

Inter-factor

correlations

432

~

726

466

~

746

376

~

788

406

~

902

377

~

852

455

~

836

637

~

818

660

~

828

317

~

852

360

~

852

637

~

818

660

~

828

Sample size 253 236 249 250 260 250

Measures of Fit

χsup2 896 573 612 462 773 616 815 715 769 542 815 715

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 951 996 975 100 958 993 957 991 972 998 957 991

GFI 942 988 960 989 951 987 948 990 955 990 948 990

RMSEA 059 028 034 00 050 034 053 044 048 022 053 044

Note All factor loadings and correlations are significant at p lt 001

- 40 -

Appendix 1 Dimensions and possible items of PSM (the first round)

1 Attraction to Public Participation

(1) I am interested in developing public programs helping my country or community (Kim 2009a)

(2) Sharing my views on public policies with others is attractive to me (Kim 2009a)

(3) I am satisfied when seeing people get benefits from the public programs I was involved in (Kim 2009a)

(4) I like to discuss political subjects with others (Giauque et al 2011)

(5) I think the governmental activities contribute to our welfare

(6) I like to initiate actions to help out my community

(7) I do not care much about politicians (Perry 1996)

2 Commitment to Public Values

Possible items for public interest

(1A) Meaningful public service is very important to me (Perry 1996)

(2A) It is important for me to contribute for the common good (Giauque et al 2011)

(3A) I recognize myself with the mission of protection of liberties and rights endorsed by the nation-state

(Giauque et al 2011)

(4A) I would prefer seeing public officials do what is best for the whole community even if it harmed my

interests (Perry 1996)

(5A) To me serving the public interest is more important than helping other people (Vandenabeele 2008b)

Possible items for public values

(1B) I recognize myself with the promotion of the equality of chances endorsed by the nation-state (Giauque et

al 2011)

(2B) It is fundamental that the public service provided to the citizen is regular and continuous (Giauque et al

2011)

(3B) It is fundamental that public service innovates to respond to the new needs of the citizensclients (Giauque

et al 2011)

(4B) The dignity and well-being of all should be the most important concerns in any society (Steenbergen

1996)

(5B) It is important that public servants account for all the costsexpenses they make (Vandenabeele 2008b)

(6B) We have to do everything in our power to pursue the goal of democracy

3 Compassion

(1) It is difficult for me to contain my feelings when I see people in distress (Perry 1996)

(2) I feel sympathetic to the plight of the underprivileged (Kim 2009a)

(3) I am often reminded by daily events how dependent we are on one another (Perry 1996)

(4) I feel compassion for others in the difficulties they are facing (Steffen and Masters 2005)

(5) It is not really my problem if others are in trouble and need help (Beutel and Marini 1995)

(6) I find it hard to be sympathetic toward starving people in foreign lands when there is so much trouble in our

own country (Beutel and Marini 1995)

(7) I get very upset when I see other people treated unfairly (Beutel and Marini 1995)

(8) I would agree to a good plan to make a better life for the poor even if it cost me money (Beutel and Marini

1995)

(9) To me patriotism includes seeing the welfare of others (Perry 1996)

(10) I care very much about other people

4 Self-Sacrifice

(1) Making a difference in society means more to me than personal achievements (Perry 1996)

(2) I am prepared to make enormous sacrifices for the good of society (Perry 1996)

(3) I believe in putting duty before self (Perry 1996)

(4) I am one of those rare people who would risk personal loss to help someone else (Perry 1996)

(5) I think people should give back to society more than they get from it (Perry 1996)

(6) Serving other citizens would give me a good feeling even if no one paid me for it (Perry 1996)

(7) Much of what I do is for a cause bigger than myself (Perry 1996)

- 41 -

Appendix 2 Newly suggested items through the first round

1 Items for the Attraction to Public Participation Dimension

(1) In the public domain every person should be counted as equal

(2) I enjoy to think and argue in a political way

(3) I am satisfied when seeing my ideas being integrated in public policies

(4) I can easily relate to the way of thinking and arguing of politicians

(5) Working in the public sector I can better realize myself

2 Items for the Public Values Sub-Dimension

(1) Public service should give value for money

(2) Cost considerations should play an important role in decisions about public service provision

(3) It is fundamental that the interests of future generations are taken into account when developing public

policies

(4) To act with correct ethics is for public servants as important as their professional competence

(5) Primarily public servants should be accountable to the public and not to their superiors

3 Items for the Self-Sacrifice Dimension

(1) I would agree to a good plan to make a better life for the poor even if it costs me money

- 42 -

Appendix 3 An example of a summary of comments and suggestions (the second round)

On the item ―I am interested in developing public programs helping my country or community

A It is a good sentence but is it possible for everyone to ―develop public programs Even public officials are

limited to do this

B OK

C or the special group

D Double barrel and difficult to translate to my country ―I am interested in helping to improve public services

E Good but I feel that wording requires revision If we draw sample from local government then country is

redundant ―I am interested in developing public policy programs that aim in helping my community

F 4 (General appreciation from1 to 4) Not clear whether it refers to ones work or to ones concern as a citizen

―(In my work) I am interested in contributing to the development of public etchellip

G Instead of development I would use implementing

H Delete public which is redundant because of the characteristic helping my country or community but add

or implementing ―I am interested in developing or implementing programs helping my country or

community

I Referring to dimension Wording Similar to item 6

Note Names have been changed to alphabets in order to ensure the anonymity of participants

- 43 -

Appendix 4 Dimensions and Items of Public Service Motivation for Future Research

Dimensions and items

Attraction to Public Service (APS)

1 I am interested in helping to improve public service

2 I admire people who initiate or are involved in activities to aid my community

3 It is important to contribute to activities that tackle social problems

4 Meaningful public service is very important to me

5 It is important for me to contribute to the common good

Commitment to Public Values (CPV)

1 I think equal opportunities for citizens are very important

2 It is important that citizens can rely on the continuous provision of public services

3 It is fundamental that the interests of future generations are taken into account when

developing public policies

4 To act ethically is essential for public servants

5 I believe that public employees must always be aware of the legitimacy of their activities

6 I personally identify with the aim of protecting individual liberties and rights

Compassion (COM)

1 It is difficult for me to contain my feelings when I see people in distress

2 I feel sympathetic to the plight of the underprivileged

3 I empathize with other people who face difficulties

4 I get very upset when I see other people being treated unfairly

5 Considering the welfare of others is very important

Self-Sacrifice (SS)

1 I am prepared to make sacrifices for the good of society

2 I believe in putting civic duty before self

3 I am willing to risk personal loss to help society

4 I would agree to a good plan to make a better life for the poor even if it costs me money

- 14 -

fall below 008 (Hair et al 2010) In general the larger the value of GFI and CFI and the

smaller the value of RMSEA the better the fit of the model (Bollen 1989) We also present

the χ2 test statistic Lower values of χ

2 indicate a better fit and should be nonsignificant but

the χ2 value increases as sample size increases and thus for large sample sizes this statistic

may lead to rejection of a model with good fit (Hair et al 2010)

Results

Using SPSS 180 descriptive statistics were computed for individual items as shown in table

1 One item (APP3) in the APP dimension ―I like to discuss topics regarding public programs

and policies with others was mistakenly omitted in the survey conducted in the United

States and so it was deleted in the statistical analysis In the assessment of normality no

items showed a skew or kurtosis value greater than the cutoffs of |3| or |8| recommended by

Kline (2005) As data were distributed normally MLE as well as D-WLS estimation was used

Based on inspection of item-total correlations one item (COM4) was dropped from further

analysis because it was weakly correlated with the overall index (011) Missing data is an

issue in CFA as it can cause estimation errors All variables in the survey have missing data

rates of approximately 016 to 143 Mean substitution was used to calculate replacement

values for the missing values in each countryrsquos dataset (Hair et al 2010) The deletion of the

two items left 31 items for the next stage of analysis

The five-correlated factor model was tested using CFA The CFA model in the initial

analysis hypothesized a priori that (1) responses to the 30-item PSM measurement

instrument could be explained by five factors such as APP CPV1 CPV2 COM SS (2) each

item would have a nonzero loading on the PSM factor that it was designed to measure and

- 15 -

zero loadings on all other factors (3) the five factors would be correlated and (4)

measurement error terms would be uncorrelated The resulting CFA with MLE suggested that

the initial five-correlated factor model was not a good fit to the data χsup2 (df = 424) = 43699 p

lt 001 GFI = 0899 CFI = 835 RMSEA = 0057 [055 059] At this stage APP was highly

correlated with CPV1 (r = 996 in MLE and r = 988 in D-WLS) suggesting the two

dimensions were not unique and lacked discriminant validity

[Table 3 about here]

Given the disconfirmation of the initial model an effort was made to estimate an

alternative model The two dimensions APP and CPV1 were combined to form a single

dimension Attraction to Public Service (APS) The items of CPV1 may be regarded as

action-oriented by the respondents because dedication to public service community and

common good is underlined in these items Consequently it is reasonable to combine APP

with CPV1 to form APS a dimension that focuses more on disposition to serve the public to

work for the common good and to participate in public policy processes After the overall

number of dimensions was reduced to four using standardized factor loadings and

modification indices as criteria the items with the lowest factor loading in each dimension

were deleted and the items whose modification index indicated association with multiple

dimensions were eliminated A CFA model with the remaining items was iteratively tested

until it achieved a good fit to the data The outcome of this refinement process was a 16-item

index of four factors Overall model fit by every measure is quite strong for the D-WLS

estimation technique The estimation results for the four-dimension model are presented in

table 4

- 16 -

[Table 4 about here]

Cronbachrsquos coefficient alpha for the 16-item PSM index was 860 and the coefficient

alphas for the four dimensions ranged from 645 to 784 The composite reliability of the set

of reflective indicators for each dimension of PSM ranged from 713 to 826 and indicated

adequate internal consistency The resulting factor structure showed a four-factor structure

with all items loading significantly onto their a priori dimension (p lt 001) and the

standardized factor loadings ranging from 0490 to 0766 in MLE and from 515 to 807 in

D-WLS The results provide support for convergent validity The correlation estimates

between the two factors ranged from 0462 to 0848 in MLE and from 505 to 859 in D-

WLS The confidence intervals (plusmn2 standard errors) around the correlation estimates between

the two factors did not include 100 The result provides support for discriminant validity To

test whether the four-dimension model is superior to other designs such as the two or three-

dimension models a series of CFA were conducted and the results show that the four-

dimension model is a significantly better fit than any of the other designs

To test whether the factorial structure of the PSM instrument is equivalent across the

samples from the twelve countries four nested models were tested as part of multi-group

analysis with MLE (Byrne 2010) First the unconstrained model was the baseline model

which relaxed all equality constraints This model hypothesized the same four-factor structure

of the instrument across twelve countries The fit indices for this baseline model were χsup2 (df =

1176) = 24977 p lt 001 GFI = 0901 CFI = 900 RMSEA = 0020 [019 021] These

indices indicate that the four-factor model of the PSM instrument exhibits acceptable fit

across the samples from the twelve countries Second the measurement weights model tested

- 17 -

the invariance of factor loadings across countries by placing equality constraints on these

parameters The fit indices for this model were χsup2 (df = 1308) = 28088 p lt 001 GFI = 0890

CFI = 887 RMSEA = 0020 [019 021] Since the measurement weights model was nested

within the unconstrained model the chi-square difference text ∆χsup2 (132) = 3111 p lt 001

indicated that some equality constraints of factor loadings did not hold across the twelve

countries Further test results suggested that the equality constraints of structural covariances

and measurement residuals were not upheld across the twelve countries It needs to be

stressed that despite the differences in factor loadings and factor covariances the results

support the main assumption that the same four-factor structure holds across different

countries We may assume that people in different cultures understand PSM in the same

fashion

To find differences in the factor structure of PSM among countries we conducted a

series of CFA with each countryrsquos dataset and followed the same procedure as when

analyzing the sample as a whole As table 5 shows although there are great similarities

between the appropriate items in the countries there are also some differences Even though

the items vary according to national differences the four-factor structure had a good fit to the

data Thus the results confirm that the four-factor structure of PSM can be generalized

[Table 5 about here]

The APS dimension covers both an individualrsquos attraction to public participation and

commitment to public interest but all its items refer to some notion of public interest such as

improving public service aiding community tackling social problems meaningful public

service and contributing to the common good The APS dimension captures an individualrsquos

- 18 -

disposition to serve the public interest through public participation When developing the

items some assumed that the items of committing to public interests can be separated from

those of pursuing public values in the CPV dimension The empirical test showed that

respondents tended to understand committing to public interests as behavior-oriented and so

the items for public interest are not differentiated from participation-oriented items in their

perceptions but are different from those for public values The respondents seem to

understand that pursuing public values is more related to personal beliefs or creeds Thus the

CPV dimension is solely focused on public values and the items of pursing public interests

are combined with those of the APP dimension to make the APS dimension

Public values and their relative importance are neither impervious to change over

time nor the same among all countries (Charles de Jong and Ryan 2011 Davis and West

2009) Moreover some values may be in conflict as Martinsen and Beck Joslashrgensen (2010)

discussed in regard to the tensions between accountability and efficiency It is therefore not

possible to include all public values in one dimension (Beck Joslashrgensen and Bozeman 2007)

Accordingly the items of the CPV dimension are developed for estimating individual

commitment to seek public values which there are general consensus about and which are

generally formulated

When all four dimensions of PSM are included in a survey cross-national

equivalence of the survey can be assured The items represent reflections or manifestations of

each dimension and hence are relatively homogeneous and interchangeable Adding or

deleting items does not change the essential nature of the dimension Thus any combination

of items in a dimension can be regarded as equivalent When using structural equation

modeling (SEM) each dimension needs to have at least three reflective indicators in order to

achieve identification or it needs to have at least two reflective indicators and be allowed to

- 19 -

covary with another dimension that has at least two reflective measures With one indicator

the dimension is identified if the loading and the variance of the uniqueness are both fixed

(Bollen 1989 Edwards 2010) Thus the number of items in each dimension is not restricted

when measuring PSM However we recommend using multiple items for each dimension

because more research is needed to investigate the appropriateness of this PSM instrument

with varying samples

The test results showed that model fit estimators were clearly better for the D-WLS

It means that in this study MLE was more rigorous than D-WLS in assessing the fit of the

model The empirical evidence indicated that MLE may be an acceptable technique for

ordinal data when there is a large sample a five-point Likert scale and the skewness and

kurtosis of variables is within acceptable limits as suggested by Byrne (2010)

Discussion

The analysis reaffirmed the four-dimensional structure of PSM but the results were not

entirely consistent with the original CFA we tested We were ultimately able to converge on a

good-fit operational model for PSM that was also consistent with underlying theory The

analysis is not able to resolve the debate about measurement models however for reasons

discussed below

Choosing the correct measurement model (ie reflective vs formative) is important

as measurement model misspecification results not only in misrepresentation of the epistemic

relationships between constructs and their measures but also in biased parameter estimates of

structural relationships (Jarvis et al 2003 Law and Wong 1999 MacKenzie 2003

MacKenzie Podsakoff and Jarvis 2005) A formative measurement model in isolation is

- 20 -

under-identified and cannot be estimated in the case of covariance-based SEM (CB-SEM)

(Bollen 1989) A construct with formative indicators must emit paths to at least two unrelated

reflective indicators two unrelated latent constructs or some combination of the two in order

to solve the problem of under-identification (Jarvis MacKenzie and Podsakoff 2003

MacCallum and Browne 1993 MacKenzie Podsakoff and Jarvis 2005) Thus ―when

dependent constructs change the empirical nature of the formatively measured construct

changes (Wilcox Howell and Breivik 2008 1223) and ―when specifying formative

measurement models what might seem like an innocuous choice of outcomes to achieve

identification can have fundamental ramifications for the meaning of the formative construct

and the inferences based on it (Edwards 2010 7)

Earlier research found that the association between a construct and its measures

should remain stable regardless of the larger causal model in which the constructs and

measures are embedded (Edwards and Bagozzi 2000) The association between a construct

and its measures is generally stable for reflective measures but associations of formative

measures with their construct are unstable The susceptibility of formative measurement to

both interpretational confounding and external inconsistency recommends that formative

measurement is not an equally attractive alternative to reflective measurement and that

whenever possible researchers should opt for reflective measurement (Howell Breivik and

Wilcox 2007a 2007b Kim Shin and Grover 2010) In the case of second-order models

researchers are limited to only a second-order reflective model in CB-SEM (Chin 2010)

Regardless of the measurement debate and the tradeoffs between reflective or

formative specification this study makes a theoretical argument that PSM is a second-order

formative construct This second-order formative factor model (reflective first-order

formative second-order) is also suggested as an alternative model that captures the

- 21 -

multidimensionality of complex concepts and avoids the shortcomings of first-order

formative measurement (Edwards 2010) However this model itself cannot achieve

identification One possible strategy is to develop direct reflective measures of PSM based on

the view that PSM is a general concept that is not only influenced by its specific dimensions

but also can be assessed using general measures (Edwards 2010 Javis MacKenzie and

Podsakoff 2003)6 However these general measures have not yet been developed for the

PSM construct

Another way around the problem of underidentification of formative models is to

apply the component-based approach known as partial least squares path modeling (PLS-PM)

which can model formative indicators (Fornell and Bookstein 1982 Goumltz Liehr-Gobbers

and Krafft 2010) The CB-SEM and the PLS-PM methods have distinctive statistical

characteristics and selecting an approach to SEM depends on the particular research situation

CB-SEM is the method of choice for theory testing while PLS-PM is appropriate for

prediction-oriented applications (Ringle Goumltz Wetzels and Wilson 2009) One advantage of

the PLS method is that it demands significantly fewer requirements compared to that of

covariance structure analyses but nevertheless delivers consistent estimation results Another

asset of the PLS approach is its ability to deal with formative as well as reflective indicators

even within one structural equation model This indicates that the PLS approach can be

appropriate for explorative analysis of structural equation models (Chin 2010 Goumltz Liehr-

Gobbers and Krafft 2010) A Monte Carlo simulation study to compare CB-SEM and PLS-

PM showed that PLS-PM provides a viable approximation of model parameters when the

prerequisites for CB-SEM are not met (Ringle Goumltz Wetzels and Wilson 2009) A formative

6 For example job satisfaction can be assessed with general reflective measures such as ―in general I am

satisfied with my job and the formative dimensions can be assessed with measures that describe satisfaction

with each job facet (Edwards 2010 15)

- 22 -

composite in the form of indexes needs to be tested to know the relative weights (γ-

parameters) of the dimensions of PSM In PLS-PM a second-order formative factor is

directly measured by observed variables for all the first order factors While this approach

repeats the number of manifest variables used the model can be identified without adding

outcome constructs that would be considered inappropriate or irrelevant from a conceptual

standpoint (Chin 2010 Edwards 2010) The test result of PLS-PM using Smart PLS 20

(Ringle Wende and Will 2005) produced the function PSM = 31APS + 36CPV + 28COM

+ 31SS However one consideration is whether and how construct error is modeled for the

formative construct (Diamantopoulos 2006) When using PLS-PM the researcher is making

the assumption that there is no error at the construct level as all constructs in PLS-PM are

modeled without error When using PLS-PM the absence of an error term will tend to

increase the weights therefore inflating their importance (Bollen and Lennox 1991) The

primary statistic for assessing a formative indicator is its weight the partialized effect of the

indicator on its intended construct controlling for the effects of all other indicators of that

construct (Cenfetelli and Bassellier 2009) A formative measure is essentially a multiple

regression with the construct representing the dependent variable and the indicators as the

predictors (Diamantopoulos and Winklhofer 2001) PSM may be regarded as a weighted

linear composite of its dimensions The weights used to create the composites are different

across CE-SEM and PLS-PM The causal dimensional weights generated in CB-SEM are

estimated in a model in which a construct-level error term is specified (eg PSM = γ1APM +

γ2CPI + γ3COM + γ4SS + ζ) whereas the weights in PLS-PM are not (eg PSM = γ1APM +

γ2CPI + γ3COM + γ4SS) This leads to two different conceptualizations of the PSM construct

one in which the causal indicators do not capture the construct in its entirety and another in

which they do (Hardin et al 2010) These differences need to be discussed in future research

- 23 -

in this area

One practical suggestion to measure PSM is to apply unit weights to its dimensions if

we assume the effects of the four dimensions are equivalent A literature review indicates that

unit weights have substantial predictive validity when compared with regression weights The

literature shows that the empirical correlations between scores generated from differential and

unit-weighting approaches are quite high In creating composite scores unit weights can be

an appropriate approach for weighting under many circumstances (Bobko Roth and Buster

2007) Although the uses of unit-weight composites will likely result in the explanation of

less variance than empirical estimations of dimensional weights they will remain consistently

defined across studies and hence suitable for theory testing Alternatively more thorough

discussions on the dimensions of PSM may suggest fixed-weight composites (Hardin et al

2010) A previous study (Kim 2010 22) also suggested ―Overall indexing PSM by

measuring all the dimensions of PSM and simply summing the averages of each dimension

can make it easier to measure PSM in practice as PSM = APS + CPV + COM + SS

An advantage of regarding PSM as a second-order formative construct is that the

dimensions can be analyzed both individually and collectively This is important as the

dimensions are expected to have different antecedents and consequences For instance Perry

(1997) found some significant differences in the influence of independent variables on the

four dimensions of PSM Taylor (2007) confirmed that certain dimensions are more important

than others in influencing work outcomes Moynihan and Pandey (2007a) reported that red

tape is negatively related to one dimension but reform orientation and hierarchical authority

is positively related to another dimension A recent study found that the SS dimension has a

positive relationship with occupational intention but the other dimensions do not (Liu et al

2011) Thus it is valuable to analyze the causes and effects of individual dimensions of PSM

- 24 -

as well as of the PSM construct

Conclusion

The purpose of this study is to develop a measurement instrument for PSM for cross-national

research and comparison On the basis of the sharpened construct of PSM (Kim and

Vandenabeele 2010) using data from an international survey conducted in 12 countries we

found that the four-factor structure of PSM is generalizable across country cases This study

provides the measurement index of PSM with the four dimensions of APS CPV COM and

SS The APS dimension measures the instrumental motives of PSM the CPV dimension

determines the value-based motives of PSM while the COM and SS dimensions gauge the

identification motives and self-sacrifice respectively When all four dimensions of PSM are

used in a survey cross-national equivalence of survey items can be assured The items for

measuring PSM in future research are suggested in Appendix 4

This study has several limitations First looking for samples which were comparable

to the largest possible extent we opted for surveys in local government The results found

and discussed in this study are based on PSM among local government employees The

results might be different when surveying central government employees or other specific

groups in the public sector Thus the potential impact of the sample bias needs to be

considered in understanding this study It also means that the survey items for measuring

PSM should be examined with different samples in various areas and at different government

levels

Second the test results showed that there is a high degree of correlation among the

dimension such as APS CPV and COM Future research should not only investigate whether

- 25 -

the items reflect the core content of the dimensions but also continue to develop more

appropriate items for better discriminant validity of each dimension Third the content of

APP dimension was assumed to be distinct from the CPV dimension that focuses on personal

disposition to pursue public interest and public values However the public interest sub-

dimension was separated from the public values sub-dimension in the CPV dimension and

combined with the APP dimension to form the APS dimension Further research should

analyze the CPV dimension and investigate the appropriateness of the APS dimension

In order to facilitate research internationally the PSM construct needs to be

improved conceptually and operationally Kim and Vandenabeele (2010) have dealt with the

conceptual composition and operational dimension of PSM as well as the measurement

model of PSM The next step is to develop an internationally robust measurement instrument

of PSM This study proposes that scholars use all four of the dimensions in measuring the

conceptual components of PSM such as self-sacrifice instrumental value-based and affective

motives because they collectively represent PSM It provides a set of appropriate items for

measuring PSM and confirms the dimensional structure of PSM The survey results suggest

that people in different cultures understand PSM in the same fashion We know that this study

is simply a step along the way to developing a more robust measure but we hope it

contributes to enhanced cross-national research and comparison along with the generation of

cumulative knowledge internationally

- 26 -

REFERENCES

Anderson James C and David D Gerbing 1988 Structural Equation Modeling in Practice

A Review and Recommended Two-Step Approach Psychological Bulletin 103

411-423

Arbuckle James L 2009 Amos 180 Userrsquos Guide Crawfordville FL Amos Development

Co

Beck Joslashrgensen Torben and Barry Bozeman 2007 Public Values An Inventory

Administration amp Society 39(3) 354-381

Beutel Ann M and Margaret Mooney Marini 1995 Gender and values American

Sociological Review 60 436 - 448

Bobko Philip Philip L Roth and Maury A Buster 2007 The Usefulness of Unit Weights in

Creating Composite Scores A Literature Review Application to Content Validity and

Meta-Analysis Organizational Research Methods 10(4) 689-709

Bollen Kenneth A 1989 Structural Equations with Latent Variables New York Wiley

Bollen K A and R Lennox 1991 Conventional Wisdom on Measurement A Structural

Equation Perspective Psychological Bulletin 110(2) 305-314

Byrne Barbara M 2010 Structural Equation Modeling with Amos Basic Concepts

Applications and Programming 2nd

ed New York Routledge

Camilleri Emanuel 2006 Towards Developing an Organizational Commitment ― Public

Service Motivation Model for Maltese Public Service Employees Public Policy

and Administration 21 63-83

Castaing Seacutebastien 2006 The Effects of Psychological Contract Fulfillment and Public

Service Motivation on Organizational Commitment in the French Civil Service

Public Policy and Administration 21 84-98

Cenfetelli Ronald T and Geneviegraveve Bassellier 2009 Interpretation of Formative

Measurement in Information Systems Research MIS Quarterly 33(4) 689-707

Charles Michael B W Martin de Jong and Neal Ryan 2011 Public Values in Western

- 27 -

Europe A Temporal Perspective American Review of Public Administration 41

75-91

Chin Wynne W 2010 How to Write Up and Report PLS Analysis In Handbook of Partical

Least Squares edited by V Esposito Vinzi et al 655-690 Berlin Springer-Verlag

Coursey David H and Sanjay K Pandey 2007a Public Service Motivation Measurement

Testing an Abridged Version of Perryrsquos Proposed Scale Administration amp Society

39 547-568

Coursey David H and Sanjay K Pandey 2007b Content Domain Measurement and

Validity of the Red Tape Concept American Review of Public Administration 37

342-361

Coursey David H James L Perry Jeffrey L Brudney and Laura Littlepage 2008

Psychometric Verification of Perryrsquos Public Service Motivation Instrument Results

for Volunteer Exemplars Review of Public Personnel Administration 28 79-90

Davis Paul and Karen West 2009 What Do Public Values Mean for Public Action Putting

Public Values in Their Plural Place American Review of Public Administration 39

602-618

DeHart-Davis Leisha Justin Marlowe and Sanjay K Pandey 2006 Gender Dimensions of

Public Service Motivation Public Administration Review 66 873-887

Diamantopoulos Adamantios 2006 The error term in formative measurement models

Interpretation and modeling implications Journal of Modeling in Management 1(1)

7-17

Diamantopoulos Adamantios and Judy A Siguaw 2006 Formative versus Reflective

Indicators in Organizational Measure Development A Comparison and Empirical

Illustration British Journal of Management 17 263-282

Diamantopoulos Adamantios and Heidi M Winklhofer 2001 Index Construction with

Formative Indicators An Alternative to Scale Development Journal of Marketing

Research 38 269-277

Edwards Jeffrey R 2010 The Fallacy of Formative Measurement Organizational Research

- 28 -

Methods Online First published on August 19 2010 as

doi1011771094428110378369

Edwards Jeffrey R and Richard P Bagozzi 2000 On the Nature and Direction of

Relationships between Constructs and Measures Psychological Methods 5 155-

174

Flora David B and Patrick J Curran 2004 An Empirical Evaluation of Alternative

Methods of Estimation for Confirmatory Factor Analysis with Ordinal Data

Psychological Methods 9 466-491

Fornell Claes and Fred L Bookstein 1982 Two structural equations models LISREL and

PLS applied to customer exit-voice theory Journal of Marketing Research 19 440-

452

Giauque David Adrian Ritz Freacutedeacuteric Varone Simon Anderfuhren-Biget and Christian

Waldner 2011 Putting Public Service Motivation into Context A Balance between

Universalism and Particularism Forthcoming in International Review of

Administrative Sciences 77(2)

Goumltz Oliver Kerstin Liehr-Gobbers and Manfred Krafft 2010 Evaluation of Structural

Equation Models Using the Partial Least Squares (PLS) Approach In V Esposito

Vinzi et al (eds) Handbook of Partial Least Squares 691-711 DOI 101007978-3-

540-32827-8_30 Springer-Verlag Berlin Heidelberg

Guinot Christiane Julie Latreille and Michel Tenenhaus 2001 PLS Path Modelling and

Multiple Table Analysis Application to the Cosmetic Habits of Women in Ile-de-

France Chemometrics and Intelligent Laboratory Systems 58 247-259

Hair Joseph F Jr William C Black Barry J Babin and Rolph E Anderson 2010

Multivariate Data Analysis 7th

ed Upper Saddle River NJ Pearson Education

Hardin Andrew M Jerry Cha-Jan Chang Mark A Fuller and Gholamreza Torkzadeh 2010

Formative Measurement and Academic Research In Search of Measurement

Theory Educational and Psychological Measurement Online First published on

June 11 2010 as doi101177001316440370208

- 29 -

Howell Roy D Einar Breivik and James B Wilcox 2007a Reconsidering Formative

Measurement Psychological Methods 12 205-218

Howell Roy D Einar Breivik and James B Wilcox 2007b Is Formative Measurement

Really Measurement Reply to Bollen (2007) and Bagozzi (2007) Psychological

Methods 12 238-245

Jarvis Cheryl Burke Scott B MacKenzie and Philip M Podsakoff 2003 A Critical Review

of Construct Indicators and Measurement Model Misspecification in Marketing and

Consumer Research Journal of Consumer Research 30 199-218

Joumlreskog Karl G and Dag Soumlrbom 2005 LISREL 872 (Computer Software) Lincolnwood

IL Scientific Software International

Kim Gimun Bongsik Shin and Varun Grover 2010 Investigating Two Contradictory Views

of Formative Measurement in Information Systems Research MIS Quarterly 34

345-365

Kim Sangmook 2009a Revising Perryrsquos Measurement Scale of Public Service Motivation

American Review of Public Administration 39 149-163

Kim Sangmook 2009b Testing the Structure of Public Service Motivation in Korea A

Research Note Journal of Public Administration Research and Theory 19 839-

851

Kim Sangmook 2010 Testing a Revised Measurement of Public Service Motivation

Reflective versus Formative Specification Journal of Public Administration

Research and Theory Doi101093jopartmuq048

Kim Sangmook and Wouter Vandenabeele 2010 A strategy for building public service

motivation research internationally Public Administration Review 70 701-709

Kline Rex B 2005 Principles and Practice of Structural Equation Modeling 2nd

ed New

York Guilford Press

Law Kenneth and Chi-Sum Wong 1999 Multidimensional Constructs in Structural

Equation Analysis An Illustration using the Job Perception and Job Satisfaction

Constructs Journal of Management 25 143-160

- 30 -

Leisink Peter and Bram Steijn 2009 Public Service Motivation and Job Performance of

Public Sector Employees in the Netherlands International Review of

Administrative Sciences 75 35-52

Liu Bangcheng Chun Hui Jin Hu Wensheng Yang and Xinli Yu 2011 How well can

Public Service Motivation Connect with Occupational Intention International

Review of Administrative Sciences 77 191-211

Liu Bangcheng Ningyu Tang and Xiaomei Zhu 2008 Public Service Motivation and Job

Satisfaction in China An Investigation of Generalisability and Instrumentality

International Journal of Manpower 29 684-699

MacCallum Robert C and Michael W Browne 1993 The Use of Causal Indicators in

Covariance Structure Models Some Practical Issues Psychological Bulletin 114

533-541

MacKenzie Scott B 2003 The dangers of poor construct conceptualization Journal of the

Academy of Marketing Science 31(3) 323-326

MacKenzie Scott B Philip M Podsakoff and Cheryl Burke Jarvis 2005 The Problem of

Measurement Model Misspecification in Behavioral and Organizational Research

and Some Recommended Solutions Journal of Applied Psychology 90 710-730

Martinsen Dorte Sindbjerg and Torben Beck Joslashrgensen 2010 Accountability as a

Differentiated Value in Supranational Governance American Review of Public

Administration 40 742-760

Moynihan Donald P and Sanjay K Pandey 2007a The Role of Organizations in Fostering

Public Service Motivation Public Administration Review 67 40-53

Moynihan Donald P and Sanjay K Pandey 2007b Finding Workable Levers over Work

Motivation Comparing Job Satisfaction Job Involvement and Organizational

Commitment Administration amp Society 39 803-832

Pandey Sanjay K and Edmund C Stazyk 2008 Antecedents and Correlates of Public

Service Motivation In Motivation in Public Management edited by James L Perry

and Annie Hondeghem 101-117 Oxford Oxford University Press

- 31 -

Perry James L 1996 Measuring Public Service Motivation An Assessment of Construct

Reliability and Validity Journal of Public Administration Research and Theory 6

5-22

Perry James L 1997 Antecedents of Public Service Motivation Journal of Public

Administration Research and Theory 7 181-197

Perry James L 2000 Bringing Society in Toward a Theory of Public Service Motivation

Journal of Public Administration Research and Theory 10 471-488

Perry James L and Annie Hondeghem 2008a Motivation in Public Management Oxford

Oxford University Press

Perry James L and Annie Hondeghem 2008b Directions for Future Theory and Research

In Motivation in Public Management edited by James L Perry and Annie

Hondeghem 294-313 Oxford Oxford University Press

Perry James L Annie Hondeghem and Lois Recascino Wise 2010 Revisiting the

Motivational Bases of Public Service Twenty Years of Research and an Agenda for

the Future Public Administration Review 70(5) 681-690

Perry James L and Wouter Vandenabeele 2008 Behavioral dynamics Institutions

Identities and Self-Regulation In Motivation in Public Management edited by

James L Perry and Annie Hondeghem 56-79 Oxford Oxford University Press

Perry James L and Lois Recascino Wise 1990 The Motivational Bases of Public Service

Public Administration Review 50 367-373

Rainey Hal G 1982 Reward Preference among Public and Private Managers In Search of

the Service Ethic American Review of Public Administration 16 288-302

Ringle Christian M Oliver Goumltz Martin Wetzels and Bradley Wilson 2009 On the Use of

Formative Measurement Specifications in Structural Equation Modeling A Monte

Carlo Simulation Study to Compare Covariance-Based and Partial Least Squares

Model Estimation Methodologies Faculty of Economics and Business Administration

Maastricht Univeristy

Ringle C M S Wende and S Will 2005 Smart PLS 20 (M3) Beta

- 32 -

httpwwwsmartplsde

Ritz Adrian 2011 Attraction to Public Policy Making A Qualitative Inquiry into

Improvements in PSM Measurement Forthcoming in Public Administration

Spector Paul E 1992 Summated Rating Scales Construction An Introduction Newbury

Park CA Sage

Steenbergen Marco R 1996 Compassion and American public opinion An analysis of the

NES Humanitarianism Scale Report to the Board of Overseers of the National

Election Studies

Steffen Patrick R and Kevin S Masters 2005 Does compassion mediate the intrinsic

religion-health relationship Annals of Behavioral Medicine 30 217-223

Taylor Jeannette 2007 The Impact of Public Service Motives on Work Outcomes in

Australia A Comparative Multi-Dimensional Analysis Public Administration 85

931-959

Vandenabeele Wouter 2007 Towards a Theory of Public Service Motivation An

Institutional Approach Public Management Review 9 545-556

Vandenabeele Wouter 2008a Government Calling Public Service Motivation as an Element

in Selecting Government as an Employer of Choice Public Administration 86

1089-1105

Vandenabeele Wouter 2008b Development of a Public Service Motivation Measurement

Scale Corroborating and Extending Perryrsquos Measurement Instrument International

Public Management Journal 11 143-167

Vandenabeele Wouter 2009 The Mediating Effect of Job Satisfaction and Organizational

Commitment on Self-Reported Performance More Robust Evidence of the PSM-

Performance Relationship International Review of Administrative Sciences 75 11-

34

Wilcox James B Roy D Howell and Einar Breivik 2008 Questions about Formative

Measurement Journal of Business Research 61 1219-1228

Wise Lois Recascino 2000 The Public Service Culture In Public Administration Concepts

- 33 -

and Cases 7th ed edited by Richard J Stillman II 342-353 Boston Houghton

Mifflin

Wright Bradley E 2008 Methodological Challenges associated with Public Service

Motivation Research In Motivation in Public Management edited by James L

Perry and Annie Hondeghem 80-98 Oxford Oxford University Press

Wright Bradley E and Sanjay K Pandey 2008 Public Service Motivation and the

Assumption of Person-Organization Fit Testing the Mediating Effect of Value

Congruence Administration amp Society 40 502-521

- 34 -

Table 1

Possible Indicators of PSM and Descriptive Statistics (n = 2868)

Dimensions and items Mean Standard

Deviation

Item-Total

Correlation

Attraction to Public Participation (APP)

APP1 I am interested in helping to improve public service

APP2 I am satisfied when I see people benefiting from the public programs I was

involved in

APP3 I like to discuss topics regarding public programs and policies with others

APP4 I believe that public sector activities contribute to our general welfare

APP5 I admire people who initiate or are involved in activities to aid my

community

APP6 Contributing to public programs and policies helps me realize myself

APP7 It is important to contribute to activities that tackle social problems

412

441

355

393

423

351

409

696

662

917

753

672

881

734

524

421

482

412

493

478

552

Commitment to Public Values (CPV)

Sub-dimension for Public Interests (CPV1)

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

CPI3 I would prefer seeing public officials do what is best for the whole

community even if it harmed my interests

CPI4 Serving the public interest is more important than helping a single individual

Sub-dimension for Public Values (CPV2)

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of public

services

CPV3 It is fundamental that public services respond to the needs of the citizens

CPV4 Decisions regarding public services should be democratic despite the time

and effort it takes

CPV5 Everybody is entitled to a good service even if it costs a lot of money

CPV6 It is fundamental that the interests of future generations are taken into

account when developing public policies

CPV7 To act ethically is essential for public servants

CPV8 I believe that public employees must always be aware of the legitimacy of

their activities

CPV9 I personally identify with the aim of protecting individual liberties and rights

418

409

348

337

428

416

433

394

367

426

441

434

414

695

662

899

995

730

704

710

852

932

689

681

692

763

525

596

480

251

484

437

356

325

376

439

462

402

392

Compassion (COM) COM1 It is difficult for me to contain my feelings when I see people in distress

COM2 I feel sympathetic to the plight of the underprivileged

COM3 I empathize with other people who face difficulties

COM4 I have little compassion for people in need who are unwilling to take the

first step to help themselves

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

354

387

388

299

412

391

951

825

707

1083

755

765

399

511

389

011

487

502

Self-Sacrifice (SS)

SS1 Making a difference to society means more to me than personal achievements

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS5 People should give back to society more than they get from it

SS6 Serving other citizens would give me a good feeling even if no one paid me for

it

SS7 I would agree to a good plan to make a better life for the poor even if it costs

me money

357

329

339

302

345

363

350

904

872

949

982

888

1049

895

508

584

454

531

412

368

546

- 35 -

Table 2

Background of Respondents (n = 2868)

Variables Characteristics Respondents

Sex

Male

Female

NA

1209

1622

37

422

566

13

Length of service

(years)

0 ~ 10

10 ~ 20

20 ~ 30

30+

NA

1126

723

484

491

44

393

252

169

171

15

Organizational status

―Do you supervise employees

No

Yes

NA

1472

1076

320

513

375

112

Country

Australia

Belgium

China

Denmark

France

Italy

Korea

Lithuania

The Netherlands

Switzerland

United Kingdom

USA

249

214

230

249

266

162

253

236

249

250

260

250

87

75

80

87

93

56

88

82

87

87

91

87 Note NA = no answer

- 36 -

Table 3

Standardized factor loadings and correlations for the 31-item PSM index (n = 2868)

Dimensions and items SFL (MLE) SFL (D-WLS) Alpha

APP

APP1

APP2

APP4

APP5

APP6

APP7

597

486

455

558

490

600

657

552

514

625

549

665

699

CPV1

CPI1

CPI2

CPI3

CPI4

588

660

488

289

653

732

535

294

542

CPV2

CPV1

CPV2

CPV3

CPV4

CPV5

CPV6

CPV7

CPV8

CPV9

611

546

436

393

418

534

598

496

460

690

631

505

457

486

601

669

573

530

745

COM

COM1

COM2

COM3

COM5

COM6

496

569

498

597

607

516

651

538

670

684

687

SS

SS1

SS2

SS3

SS4

SS5

SS6

SS7

567

736

558

728

467

409

663

660

762

591

695

515

494

709

782

Measures of fit (MLE) χsup2 (df = 424) = 43699 p lt 001 CFI = 835 GFI = 899

RMSEA = 057 [055 059]

Measure of fit (D-WLS) χsup2 (df = 424) = 39247 p lt 001 CFI = 971 GFI = 976

RMSEA = 054 [052 055]

Inter-factor correlations MLE (D-WLS)

1 2 3 4

1 APS

2 CPV1 996 (988)

3 CPV2 902 (890) 798 (798)

4 COM 787 (783) 758 (762) 779 (783)

5 SS 700 (742) 800 (859) 509 (537) 720 (729)

Note SFL = standardized factor loading All factor loadings and correlations are significant at p lt 001

- 37 -

Table 4

Standardized factor loadings and correlations for the 16-item PSM index (n = 2868)

Dimensions and items SFL

(MLE)

SFL

(D-WLS)

APS

APP5 I admire people who initiate or are involved in activities to aid my

community

APP7 It is important to contribute to activities that tackle social problems

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

611

592

637

666

665

669

690

751

CPV

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of

public services

CPV6 It is fundamental that the interests of future generations are taken

into account when developing public policies

CPV7 To act ethically is essential for public servants

612

546

551

590

698

631

618

658

COM

COM1 It is difficult for me to contain my feelings when I see people in

distress

COM2 I feel sympathetic to the plight of the underprivileged

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

490

546

596

604

515

608

669

679

SS

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS7 I would agree to a good plan to make a better life for the poor even if

it costs me money

763

571

766

678

807

629

747

755

Measures of fit (MLE)

χsup2 (df = 98) = 8490 p lt 001

CFI = 939 GFI = 965

RMSEA = 052 [049 055]

Measure of fit (D-WLS)

χsup2 (df = 98) = 6069 p lt 001

CFI = 988 GFI = 992

RMSEA = 043 [039 046]

Alpha CR (D-WLS) Inter-factor correlations MLE (D-WLS)

1 APS 717 788 1 APS 2 CPV 3 COM

2 CPV 663 747 848 (859)

3 COM 645 713 781 (791) 789 (795)

4 SS 784 826 565 (596) 462 (505) 715 (726)

Note SFL = standardized factor loading Alpha = Cronbachrsquos alpha CR = composite reliability

All factor loadings and correlations are significant at p lt 001

- 38 -

Table 5

Results of CFA in the individual countries Standardized factor loadings and measures of fit

Factors

amp Items

Australia Belgium China Denmark France Italy

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

672

756

799

785

804

943

710

703

597

790

799

730

584

675

685

681

731

742

578

748

674

643

765

739

645

704

678

708

812

779

632

604

765

766

672

853

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

552

593

560

699

727

632

624

605

560

713

709

645

702

602

616

724

686

735

567

560

715

654

605

811

608

658

587

696

755

698

673

620

523

782

697

654

COM

COM1

COM2

COM3

COM5

COM6

623

584

662

688

701

739

732

608

475

808

666

581

578

616

605

649

670

673

699

751

737

679

816

776

568

546

801

626

673

850

637

546

741

715

601

822

SS

SS2

SS3

SS4

SS7

654

794

806

892

767

813

727

693

799

793

725

838

857

820

773

896

855

842

806

749

792

829

753

884

847

731

753

889

816

753

815

692

664

795

782

734

Inter-factor

correlations

093

~

752

163

~

810

320

~

773

323

~

825

452

~

827

497

~

855

323

~

850

321

~

847

498

~

899

536

~

902

425

~

829

476

~

865

Sample size 249 214 230 249 266 162

Measures of Fit

χsup2 638 444 881 732 765 554 817 625 735 565 750 564

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 980 100 939 987 969 997 967 994 975 997 949 995

GFI 961 986 941 986 948 991 949 992 955 992 926 986

RMSEA 036 00 063 050 051 026 053 035 045 026 059 033

- 39 -

Table 5 (Continued)

Factors

amp Items

Korea Lithuania The Netherlands Switzerland UK USA

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

619

701

667

706

781

717

666

512

682

751

605

779

557

629

630

620

719

729

535

621

702

643

694

735

655

725

738

710

804

822

535

621

702

643

694

735

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

594

607

722

722

644

722

674

535

533

790

589

711

568

554

538

655

628

545

627

580

615

711

666

704

800

512

555

873

581

646

627

580

615

711

666

704

COM

COM1

COM2

COM3

COM5

COM6

694

677

550

763

736

627

557

639

428

550

696

534

737

623

508

766

616

608

659

592

683

703

661

769

682

583

729

739

644

803

659

592

683

703

661

769

SS

SS2

SS3

SS4

SS7

776

752

679

779

770

811

707

721

635

725

734

706

814

656

743

820

767

764

745

632

689

788

662

744

671

818

832

830

864

802

745

632

689

788

662

744

Inter-factor

correlations

432

~

726

466

~

746

376

~

788

406

~

902

377

~

852

455

~

836

637

~

818

660

~

828

317

~

852

360

~

852

637

~

818

660

~

828

Sample size 253 236 249 250 260 250

Measures of Fit

χsup2 896 573 612 462 773 616 815 715 769 542 815 715

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 951 996 975 100 958 993 957 991 972 998 957 991

GFI 942 988 960 989 951 987 948 990 955 990 948 990

RMSEA 059 028 034 00 050 034 053 044 048 022 053 044

Note All factor loadings and correlations are significant at p lt 001

- 40 -

Appendix 1 Dimensions and possible items of PSM (the first round)

1 Attraction to Public Participation

(1) I am interested in developing public programs helping my country or community (Kim 2009a)

(2) Sharing my views on public policies with others is attractive to me (Kim 2009a)

(3) I am satisfied when seeing people get benefits from the public programs I was involved in (Kim 2009a)

(4) I like to discuss political subjects with others (Giauque et al 2011)

(5) I think the governmental activities contribute to our welfare

(6) I like to initiate actions to help out my community

(7) I do not care much about politicians (Perry 1996)

2 Commitment to Public Values

Possible items for public interest

(1A) Meaningful public service is very important to me (Perry 1996)

(2A) It is important for me to contribute for the common good (Giauque et al 2011)

(3A) I recognize myself with the mission of protection of liberties and rights endorsed by the nation-state

(Giauque et al 2011)

(4A) I would prefer seeing public officials do what is best for the whole community even if it harmed my

interests (Perry 1996)

(5A) To me serving the public interest is more important than helping other people (Vandenabeele 2008b)

Possible items for public values

(1B) I recognize myself with the promotion of the equality of chances endorsed by the nation-state (Giauque et

al 2011)

(2B) It is fundamental that the public service provided to the citizen is regular and continuous (Giauque et al

2011)

(3B) It is fundamental that public service innovates to respond to the new needs of the citizensclients (Giauque

et al 2011)

(4B) The dignity and well-being of all should be the most important concerns in any society (Steenbergen

1996)

(5B) It is important that public servants account for all the costsexpenses they make (Vandenabeele 2008b)

(6B) We have to do everything in our power to pursue the goal of democracy

3 Compassion

(1) It is difficult for me to contain my feelings when I see people in distress (Perry 1996)

(2) I feel sympathetic to the plight of the underprivileged (Kim 2009a)

(3) I am often reminded by daily events how dependent we are on one another (Perry 1996)

(4) I feel compassion for others in the difficulties they are facing (Steffen and Masters 2005)

(5) It is not really my problem if others are in trouble and need help (Beutel and Marini 1995)

(6) I find it hard to be sympathetic toward starving people in foreign lands when there is so much trouble in our

own country (Beutel and Marini 1995)

(7) I get very upset when I see other people treated unfairly (Beutel and Marini 1995)

(8) I would agree to a good plan to make a better life for the poor even if it cost me money (Beutel and Marini

1995)

(9) To me patriotism includes seeing the welfare of others (Perry 1996)

(10) I care very much about other people

4 Self-Sacrifice

(1) Making a difference in society means more to me than personal achievements (Perry 1996)

(2) I am prepared to make enormous sacrifices for the good of society (Perry 1996)

(3) I believe in putting duty before self (Perry 1996)

(4) I am one of those rare people who would risk personal loss to help someone else (Perry 1996)

(5) I think people should give back to society more than they get from it (Perry 1996)

(6) Serving other citizens would give me a good feeling even if no one paid me for it (Perry 1996)

(7) Much of what I do is for a cause bigger than myself (Perry 1996)

- 41 -

Appendix 2 Newly suggested items through the first round

1 Items for the Attraction to Public Participation Dimension

(1) In the public domain every person should be counted as equal

(2) I enjoy to think and argue in a political way

(3) I am satisfied when seeing my ideas being integrated in public policies

(4) I can easily relate to the way of thinking and arguing of politicians

(5) Working in the public sector I can better realize myself

2 Items for the Public Values Sub-Dimension

(1) Public service should give value for money

(2) Cost considerations should play an important role in decisions about public service provision

(3) It is fundamental that the interests of future generations are taken into account when developing public

policies

(4) To act with correct ethics is for public servants as important as their professional competence

(5) Primarily public servants should be accountable to the public and not to their superiors

3 Items for the Self-Sacrifice Dimension

(1) I would agree to a good plan to make a better life for the poor even if it costs me money

- 42 -

Appendix 3 An example of a summary of comments and suggestions (the second round)

On the item ―I am interested in developing public programs helping my country or community

A It is a good sentence but is it possible for everyone to ―develop public programs Even public officials are

limited to do this

B OK

C or the special group

D Double barrel and difficult to translate to my country ―I am interested in helping to improve public services

E Good but I feel that wording requires revision If we draw sample from local government then country is

redundant ―I am interested in developing public policy programs that aim in helping my community

F 4 (General appreciation from1 to 4) Not clear whether it refers to ones work or to ones concern as a citizen

―(In my work) I am interested in contributing to the development of public etchellip

G Instead of development I would use implementing

H Delete public which is redundant because of the characteristic helping my country or community but add

or implementing ―I am interested in developing or implementing programs helping my country or

community

I Referring to dimension Wording Similar to item 6

Note Names have been changed to alphabets in order to ensure the anonymity of participants

- 43 -

Appendix 4 Dimensions and Items of Public Service Motivation for Future Research

Dimensions and items

Attraction to Public Service (APS)

1 I am interested in helping to improve public service

2 I admire people who initiate or are involved in activities to aid my community

3 It is important to contribute to activities that tackle social problems

4 Meaningful public service is very important to me

5 It is important for me to contribute to the common good

Commitment to Public Values (CPV)

1 I think equal opportunities for citizens are very important

2 It is important that citizens can rely on the continuous provision of public services

3 It is fundamental that the interests of future generations are taken into account when

developing public policies

4 To act ethically is essential for public servants

5 I believe that public employees must always be aware of the legitimacy of their activities

6 I personally identify with the aim of protecting individual liberties and rights

Compassion (COM)

1 It is difficult for me to contain my feelings when I see people in distress

2 I feel sympathetic to the plight of the underprivileged

3 I empathize with other people who face difficulties

4 I get very upset when I see other people being treated unfairly

5 Considering the welfare of others is very important

Self-Sacrifice (SS)

1 I am prepared to make sacrifices for the good of society

2 I believe in putting civic duty before self

3 I am willing to risk personal loss to help society

4 I would agree to a good plan to make a better life for the poor even if it costs me money

- 15 -

zero loadings on all other factors (3) the five factors would be correlated and (4)

measurement error terms would be uncorrelated The resulting CFA with MLE suggested that

the initial five-correlated factor model was not a good fit to the data χsup2 (df = 424) = 43699 p

lt 001 GFI = 0899 CFI = 835 RMSEA = 0057 [055 059] At this stage APP was highly

correlated with CPV1 (r = 996 in MLE and r = 988 in D-WLS) suggesting the two

dimensions were not unique and lacked discriminant validity

[Table 3 about here]

Given the disconfirmation of the initial model an effort was made to estimate an

alternative model The two dimensions APP and CPV1 were combined to form a single

dimension Attraction to Public Service (APS) The items of CPV1 may be regarded as

action-oriented by the respondents because dedication to public service community and

common good is underlined in these items Consequently it is reasonable to combine APP

with CPV1 to form APS a dimension that focuses more on disposition to serve the public to

work for the common good and to participate in public policy processes After the overall

number of dimensions was reduced to four using standardized factor loadings and

modification indices as criteria the items with the lowest factor loading in each dimension

were deleted and the items whose modification index indicated association with multiple

dimensions were eliminated A CFA model with the remaining items was iteratively tested

until it achieved a good fit to the data The outcome of this refinement process was a 16-item

index of four factors Overall model fit by every measure is quite strong for the D-WLS

estimation technique The estimation results for the four-dimension model are presented in

table 4

- 16 -

[Table 4 about here]

Cronbachrsquos coefficient alpha for the 16-item PSM index was 860 and the coefficient

alphas for the four dimensions ranged from 645 to 784 The composite reliability of the set

of reflective indicators for each dimension of PSM ranged from 713 to 826 and indicated

adequate internal consistency The resulting factor structure showed a four-factor structure

with all items loading significantly onto their a priori dimension (p lt 001) and the

standardized factor loadings ranging from 0490 to 0766 in MLE and from 515 to 807 in

D-WLS The results provide support for convergent validity The correlation estimates

between the two factors ranged from 0462 to 0848 in MLE and from 505 to 859 in D-

WLS The confidence intervals (plusmn2 standard errors) around the correlation estimates between

the two factors did not include 100 The result provides support for discriminant validity To

test whether the four-dimension model is superior to other designs such as the two or three-

dimension models a series of CFA were conducted and the results show that the four-

dimension model is a significantly better fit than any of the other designs

To test whether the factorial structure of the PSM instrument is equivalent across the

samples from the twelve countries four nested models were tested as part of multi-group

analysis with MLE (Byrne 2010) First the unconstrained model was the baseline model

which relaxed all equality constraints This model hypothesized the same four-factor structure

of the instrument across twelve countries The fit indices for this baseline model were χsup2 (df =

1176) = 24977 p lt 001 GFI = 0901 CFI = 900 RMSEA = 0020 [019 021] These

indices indicate that the four-factor model of the PSM instrument exhibits acceptable fit

across the samples from the twelve countries Second the measurement weights model tested

- 17 -

the invariance of factor loadings across countries by placing equality constraints on these

parameters The fit indices for this model were χsup2 (df = 1308) = 28088 p lt 001 GFI = 0890

CFI = 887 RMSEA = 0020 [019 021] Since the measurement weights model was nested

within the unconstrained model the chi-square difference text ∆χsup2 (132) = 3111 p lt 001

indicated that some equality constraints of factor loadings did not hold across the twelve

countries Further test results suggested that the equality constraints of structural covariances

and measurement residuals were not upheld across the twelve countries It needs to be

stressed that despite the differences in factor loadings and factor covariances the results

support the main assumption that the same four-factor structure holds across different

countries We may assume that people in different cultures understand PSM in the same

fashion

To find differences in the factor structure of PSM among countries we conducted a

series of CFA with each countryrsquos dataset and followed the same procedure as when

analyzing the sample as a whole As table 5 shows although there are great similarities

between the appropriate items in the countries there are also some differences Even though

the items vary according to national differences the four-factor structure had a good fit to the

data Thus the results confirm that the four-factor structure of PSM can be generalized

[Table 5 about here]

The APS dimension covers both an individualrsquos attraction to public participation and

commitment to public interest but all its items refer to some notion of public interest such as

improving public service aiding community tackling social problems meaningful public

service and contributing to the common good The APS dimension captures an individualrsquos

- 18 -

disposition to serve the public interest through public participation When developing the

items some assumed that the items of committing to public interests can be separated from

those of pursuing public values in the CPV dimension The empirical test showed that

respondents tended to understand committing to public interests as behavior-oriented and so

the items for public interest are not differentiated from participation-oriented items in their

perceptions but are different from those for public values The respondents seem to

understand that pursuing public values is more related to personal beliefs or creeds Thus the

CPV dimension is solely focused on public values and the items of pursing public interests

are combined with those of the APP dimension to make the APS dimension

Public values and their relative importance are neither impervious to change over

time nor the same among all countries (Charles de Jong and Ryan 2011 Davis and West

2009) Moreover some values may be in conflict as Martinsen and Beck Joslashrgensen (2010)

discussed in regard to the tensions between accountability and efficiency It is therefore not

possible to include all public values in one dimension (Beck Joslashrgensen and Bozeman 2007)

Accordingly the items of the CPV dimension are developed for estimating individual

commitment to seek public values which there are general consensus about and which are

generally formulated

When all four dimensions of PSM are included in a survey cross-national

equivalence of the survey can be assured The items represent reflections or manifestations of

each dimension and hence are relatively homogeneous and interchangeable Adding or

deleting items does not change the essential nature of the dimension Thus any combination

of items in a dimension can be regarded as equivalent When using structural equation

modeling (SEM) each dimension needs to have at least three reflective indicators in order to

achieve identification or it needs to have at least two reflective indicators and be allowed to

- 19 -

covary with another dimension that has at least two reflective measures With one indicator

the dimension is identified if the loading and the variance of the uniqueness are both fixed

(Bollen 1989 Edwards 2010) Thus the number of items in each dimension is not restricted

when measuring PSM However we recommend using multiple items for each dimension

because more research is needed to investigate the appropriateness of this PSM instrument

with varying samples

The test results showed that model fit estimators were clearly better for the D-WLS

It means that in this study MLE was more rigorous than D-WLS in assessing the fit of the

model The empirical evidence indicated that MLE may be an acceptable technique for

ordinal data when there is a large sample a five-point Likert scale and the skewness and

kurtosis of variables is within acceptable limits as suggested by Byrne (2010)

Discussion

The analysis reaffirmed the four-dimensional structure of PSM but the results were not

entirely consistent with the original CFA we tested We were ultimately able to converge on a

good-fit operational model for PSM that was also consistent with underlying theory The

analysis is not able to resolve the debate about measurement models however for reasons

discussed below

Choosing the correct measurement model (ie reflective vs formative) is important

as measurement model misspecification results not only in misrepresentation of the epistemic

relationships between constructs and their measures but also in biased parameter estimates of

structural relationships (Jarvis et al 2003 Law and Wong 1999 MacKenzie 2003

MacKenzie Podsakoff and Jarvis 2005) A formative measurement model in isolation is

- 20 -

under-identified and cannot be estimated in the case of covariance-based SEM (CB-SEM)

(Bollen 1989) A construct with formative indicators must emit paths to at least two unrelated

reflective indicators two unrelated latent constructs or some combination of the two in order

to solve the problem of under-identification (Jarvis MacKenzie and Podsakoff 2003

MacCallum and Browne 1993 MacKenzie Podsakoff and Jarvis 2005) Thus ―when

dependent constructs change the empirical nature of the formatively measured construct

changes (Wilcox Howell and Breivik 2008 1223) and ―when specifying formative

measurement models what might seem like an innocuous choice of outcomes to achieve

identification can have fundamental ramifications for the meaning of the formative construct

and the inferences based on it (Edwards 2010 7)

Earlier research found that the association between a construct and its measures

should remain stable regardless of the larger causal model in which the constructs and

measures are embedded (Edwards and Bagozzi 2000) The association between a construct

and its measures is generally stable for reflective measures but associations of formative

measures with their construct are unstable The susceptibility of formative measurement to

both interpretational confounding and external inconsistency recommends that formative

measurement is not an equally attractive alternative to reflective measurement and that

whenever possible researchers should opt for reflective measurement (Howell Breivik and

Wilcox 2007a 2007b Kim Shin and Grover 2010) In the case of second-order models

researchers are limited to only a second-order reflective model in CB-SEM (Chin 2010)

Regardless of the measurement debate and the tradeoffs between reflective or

formative specification this study makes a theoretical argument that PSM is a second-order

formative construct This second-order formative factor model (reflective first-order

formative second-order) is also suggested as an alternative model that captures the

- 21 -

multidimensionality of complex concepts and avoids the shortcomings of first-order

formative measurement (Edwards 2010) However this model itself cannot achieve

identification One possible strategy is to develop direct reflective measures of PSM based on

the view that PSM is a general concept that is not only influenced by its specific dimensions

but also can be assessed using general measures (Edwards 2010 Javis MacKenzie and

Podsakoff 2003)6 However these general measures have not yet been developed for the

PSM construct

Another way around the problem of underidentification of formative models is to

apply the component-based approach known as partial least squares path modeling (PLS-PM)

which can model formative indicators (Fornell and Bookstein 1982 Goumltz Liehr-Gobbers

and Krafft 2010) The CB-SEM and the PLS-PM methods have distinctive statistical

characteristics and selecting an approach to SEM depends on the particular research situation

CB-SEM is the method of choice for theory testing while PLS-PM is appropriate for

prediction-oriented applications (Ringle Goumltz Wetzels and Wilson 2009) One advantage of

the PLS method is that it demands significantly fewer requirements compared to that of

covariance structure analyses but nevertheless delivers consistent estimation results Another

asset of the PLS approach is its ability to deal with formative as well as reflective indicators

even within one structural equation model This indicates that the PLS approach can be

appropriate for explorative analysis of structural equation models (Chin 2010 Goumltz Liehr-

Gobbers and Krafft 2010) A Monte Carlo simulation study to compare CB-SEM and PLS-

PM showed that PLS-PM provides a viable approximation of model parameters when the

prerequisites for CB-SEM are not met (Ringle Goumltz Wetzels and Wilson 2009) A formative

6 For example job satisfaction can be assessed with general reflective measures such as ―in general I am

satisfied with my job and the formative dimensions can be assessed with measures that describe satisfaction

with each job facet (Edwards 2010 15)

- 22 -

composite in the form of indexes needs to be tested to know the relative weights (γ-

parameters) of the dimensions of PSM In PLS-PM a second-order formative factor is

directly measured by observed variables for all the first order factors While this approach

repeats the number of manifest variables used the model can be identified without adding

outcome constructs that would be considered inappropriate or irrelevant from a conceptual

standpoint (Chin 2010 Edwards 2010) The test result of PLS-PM using Smart PLS 20

(Ringle Wende and Will 2005) produced the function PSM = 31APS + 36CPV + 28COM

+ 31SS However one consideration is whether and how construct error is modeled for the

formative construct (Diamantopoulos 2006) When using PLS-PM the researcher is making

the assumption that there is no error at the construct level as all constructs in PLS-PM are

modeled without error When using PLS-PM the absence of an error term will tend to

increase the weights therefore inflating their importance (Bollen and Lennox 1991) The

primary statistic for assessing a formative indicator is its weight the partialized effect of the

indicator on its intended construct controlling for the effects of all other indicators of that

construct (Cenfetelli and Bassellier 2009) A formative measure is essentially a multiple

regression with the construct representing the dependent variable and the indicators as the

predictors (Diamantopoulos and Winklhofer 2001) PSM may be regarded as a weighted

linear composite of its dimensions The weights used to create the composites are different

across CE-SEM and PLS-PM The causal dimensional weights generated in CB-SEM are

estimated in a model in which a construct-level error term is specified (eg PSM = γ1APM +

γ2CPI + γ3COM + γ4SS + ζ) whereas the weights in PLS-PM are not (eg PSM = γ1APM +

γ2CPI + γ3COM + γ4SS) This leads to two different conceptualizations of the PSM construct

one in which the causal indicators do not capture the construct in its entirety and another in

which they do (Hardin et al 2010) These differences need to be discussed in future research

- 23 -

in this area

One practical suggestion to measure PSM is to apply unit weights to its dimensions if

we assume the effects of the four dimensions are equivalent A literature review indicates that

unit weights have substantial predictive validity when compared with regression weights The

literature shows that the empirical correlations between scores generated from differential and

unit-weighting approaches are quite high In creating composite scores unit weights can be

an appropriate approach for weighting under many circumstances (Bobko Roth and Buster

2007) Although the uses of unit-weight composites will likely result in the explanation of

less variance than empirical estimations of dimensional weights they will remain consistently

defined across studies and hence suitable for theory testing Alternatively more thorough

discussions on the dimensions of PSM may suggest fixed-weight composites (Hardin et al

2010) A previous study (Kim 2010 22) also suggested ―Overall indexing PSM by

measuring all the dimensions of PSM and simply summing the averages of each dimension

can make it easier to measure PSM in practice as PSM = APS + CPV + COM + SS

An advantage of regarding PSM as a second-order formative construct is that the

dimensions can be analyzed both individually and collectively This is important as the

dimensions are expected to have different antecedents and consequences For instance Perry

(1997) found some significant differences in the influence of independent variables on the

four dimensions of PSM Taylor (2007) confirmed that certain dimensions are more important

than others in influencing work outcomes Moynihan and Pandey (2007a) reported that red

tape is negatively related to one dimension but reform orientation and hierarchical authority

is positively related to another dimension A recent study found that the SS dimension has a

positive relationship with occupational intention but the other dimensions do not (Liu et al

2011) Thus it is valuable to analyze the causes and effects of individual dimensions of PSM

- 24 -

as well as of the PSM construct

Conclusion

The purpose of this study is to develop a measurement instrument for PSM for cross-national

research and comparison On the basis of the sharpened construct of PSM (Kim and

Vandenabeele 2010) using data from an international survey conducted in 12 countries we

found that the four-factor structure of PSM is generalizable across country cases This study

provides the measurement index of PSM with the four dimensions of APS CPV COM and

SS The APS dimension measures the instrumental motives of PSM the CPV dimension

determines the value-based motives of PSM while the COM and SS dimensions gauge the

identification motives and self-sacrifice respectively When all four dimensions of PSM are

used in a survey cross-national equivalence of survey items can be assured The items for

measuring PSM in future research are suggested in Appendix 4

This study has several limitations First looking for samples which were comparable

to the largest possible extent we opted for surveys in local government The results found

and discussed in this study are based on PSM among local government employees The

results might be different when surveying central government employees or other specific

groups in the public sector Thus the potential impact of the sample bias needs to be

considered in understanding this study It also means that the survey items for measuring

PSM should be examined with different samples in various areas and at different government

levels

Second the test results showed that there is a high degree of correlation among the

dimension such as APS CPV and COM Future research should not only investigate whether

- 25 -

the items reflect the core content of the dimensions but also continue to develop more

appropriate items for better discriminant validity of each dimension Third the content of

APP dimension was assumed to be distinct from the CPV dimension that focuses on personal

disposition to pursue public interest and public values However the public interest sub-

dimension was separated from the public values sub-dimension in the CPV dimension and

combined with the APP dimension to form the APS dimension Further research should

analyze the CPV dimension and investigate the appropriateness of the APS dimension

In order to facilitate research internationally the PSM construct needs to be

improved conceptually and operationally Kim and Vandenabeele (2010) have dealt with the

conceptual composition and operational dimension of PSM as well as the measurement

model of PSM The next step is to develop an internationally robust measurement instrument

of PSM This study proposes that scholars use all four of the dimensions in measuring the

conceptual components of PSM such as self-sacrifice instrumental value-based and affective

motives because they collectively represent PSM It provides a set of appropriate items for

measuring PSM and confirms the dimensional structure of PSM The survey results suggest

that people in different cultures understand PSM in the same fashion We know that this study

is simply a step along the way to developing a more robust measure but we hope it

contributes to enhanced cross-national research and comparison along with the generation of

cumulative knowledge internationally

- 26 -

REFERENCES

Anderson James C and David D Gerbing 1988 Structural Equation Modeling in Practice

A Review and Recommended Two-Step Approach Psychological Bulletin 103

411-423

Arbuckle James L 2009 Amos 180 Userrsquos Guide Crawfordville FL Amos Development

Co

Beck Joslashrgensen Torben and Barry Bozeman 2007 Public Values An Inventory

Administration amp Society 39(3) 354-381

Beutel Ann M and Margaret Mooney Marini 1995 Gender and values American

Sociological Review 60 436 - 448

Bobko Philip Philip L Roth and Maury A Buster 2007 The Usefulness of Unit Weights in

Creating Composite Scores A Literature Review Application to Content Validity and

Meta-Analysis Organizational Research Methods 10(4) 689-709

Bollen Kenneth A 1989 Structural Equations with Latent Variables New York Wiley

Bollen K A and R Lennox 1991 Conventional Wisdom on Measurement A Structural

Equation Perspective Psychological Bulletin 110(2) 305-314

Byrne Barbara M 2010 Structural Equation Modeling with Amos Basic Concepts

Applications and Programming 2nd

ed New York Routledge

Camilleri Emanuel 2006 Towards Developing an Organizational Commitment ― Public

Service Motivation Model for Maltese Public Service Employees Public Policy

and Administration 21 63-83

Castaing Seacutebastien 2006 The Effects of Psychological Contract Fulfillment and Public

Service Motivation on Organizational Commitment in the French Civil Service

Public Policy and Administration 21 84-98

Cenfetelli Ronald T and Geneviegraveve Bassellier 2009 Interpretation of Formative

Measurement in Information Systems Research MIS Quarterly 33(4) 689-707

Charles Michael B W Martin de Jong and Neal Ryan 2011 Public Values in Western

- 27 -

Europe A Temporal Perspective American Review of Public Administration 41

75-91

Chin Wynne W 2010 How to Write Up and Report PLS Analysis In Handbook of Partical

Least Squares edited by V Esposito Vinzi et al 655-690 Berlin Springer-Verlag

Coursey David H and Sanjay K Pandey 2007a Public Service Motivation Measurement

Testing an Abridged Version of Perryrsquos Proposed Scale Administration amp Society

39 547-568

Coursey David H and Sanjay K Pandey 2007b Content Domain Measurement and

Validity of the Red Tape Concept American Review of Public Administration 37

342-361

Coursey David H James L Perry Jeffrey L Brudney and Laura Littlepage 2008

Psychometric Verification of Perryrsquos Public Service Motivation Instrument Results

for Volunteer Exemplars Review of Public Personnel Administration 28 79-90

Davis Paul and Karen West 2009 What Do Public Values Mean for Public Action Putting

Public Values in Their Plural Place American Review of Public Administration 39

602-618

DeHart-Davis Leisha Justin Marlowe and Sanjay K Pandey 2006 Gender Dimensions of

Public Service Motivation Public Administration Review 66 873-887

Diamantopoulos Adamantios 2006 The error term in formative measurement models

Interpretation and modeling implications Journal of Modeling in Management 1(1)

7-17

Diamantopoulos Adamantios and Judy A Siguaw 2006 Formative versus Reflective

Indicators in Organizational Measure Development A Comparison and Empirical

Illustration British Journal of Management 17 263-282

Diamantopoulos Adamantios and Heidi M Winklhofer 2001 Index Construction with

Formative Indicators An Alternative to Scale Development Journal of Marketing

Research 38 269-277

Edwards Jeffrey R 2010 The Fallacy of Formative Measurement Organizational Research

- 28 -

Methods Online First published on August 19 2010 as

doi1011771094428110378369

Edwards Jeffrey R and Richard P Bagozzi 2000 On the Nature and Direction of

Relationships between Constructs and Measures Psychological Methods 5 155-

174

Flora David B and Patrick J Curran 2004 An Empirical Evaluation of Alternative

Methods of Estimation for Confirmatory Factor Analysis with Ordinal Data

Psychological Methods 9 466-491

Fornell Claes and Fred L Bookstein 1982 Two structural equations models LISREL and

PLS applied to customer exit-voice theory Journal of Marketing Research 19 440-

452

Giauque David Adrian Ritz Freacutedeacuteric Varone Simon Anderfuhren-Biget and Christian

Waldner 2011 Putting Public Service Motivation into Context A Balance between

Universalism and Particularism Forthcoming in International Review of

Administrative Sciences 77(2)

Goumltz Oliver Kerstin Liehr-Gobbers and Manfred Krafft 2010 Evaluation of Structural

Equation Models Using the Partial Least Squares (PLS) Approach In V Esposito

Vinzi et al (eds) Handbook of Partial Least Squares 691-711 DOI 101007978-3-

540-32827-8_30 Springer-Verlag Berlin Heidelberg

Guinot Christiane Julie Latreille and Michel Tenenhaus 2001 PLS Path Modelling and

Multiple Table Analysis Application to the Cosmetic Habits of Women in Ile-de-

France Chemometrics and Intelligent Laboratory Systems 58 247-259

Hair Joseph F Jr William C Black Barry J Babin and Rolph E Anderson 2010

Multivariate Data Analysis 7th

ed Upper Saddle River NJ Pearson Education

Hardin Andrew M Jerry Cha-Jan Chang Mark A Fuller and Gholamreza Torkzadeh 2010

Formative Measurement and Academic Research In Search of Measurement

Theory Educational and Psychological Measurement Online First published on

June 11 2010 as doi101177001316440370208

- 29 -

Howell Roy D Einar Breivik and James B Wilcox 2007a Reconsidering Formative

Measurement Psychological Methods 12 205-218

Howell Roy D Einar Breivik and James B Wilcox 2007b Is Formative Measurement

Really Measurement Reply to Bollen (2007) and Bagozzi (2007) Psychological

Methods 12 238-245

Jarvis Cheryl Burke Scott B MacKenzie and Philip M Podsakoff 2003 A Critical Review

of Construct Indicators and Measurement Model Misspecification in Marketing and

Consumer Research Journal of Consumer Research 30 199-218

Joumlreskog Karl G and Dag Soumlrbom 2005 LISREL 872 (Computer Software) Lincolnwood

IL Scientific Software International

Kim Gimun Bongsik Shin and Varun Grover 2010 Investigating Two Contradictory Views

of Formative Measurement in Information Systems Research MIS Quarterly 34

345-365

Kim Sangmook 2009a Revising Perryrsquos Measurement Scale of Public Service Motivation

American Review of Public Administration 39 149-163

Kim Sangmook 2009b Testing the Structure of Public Service Motivation in Korea A

Research Note Journal of Public Administration Research and Theory 19 839-

851

Kim Sangmook 2010 Testing a Revised Measurement of Public Service Motivation

Reflective versus Formative Specification Journal of Public Administration

Research and Theory Doi101093jopartmuq048

Kim Sangmook and Wouter Vandenabeele 2010 A strategy for building public service

motivation research internationally Public Administration Review 70 701-709

Kline Rex B 2005 Principles and Practice of Structural Equation Modeling 2nd

ed New

York Guilford Press

Law Kenneth and Chi-Sum Wong 1999 Multidimensional Constructs in Structural

Equation Analysis An Illustration using the Job Perception and Job Satisfaction

Constructs Journal of Management 25 143-160

- 30 -

Leisink Peter and Bram Steijn 2009 Public Service Motivation and Job Performance of

Public Sector Employees in the Netherlands International Review of

Administrative Sciences 75 35-52

Liu Bangcheng Chun Hui Jin Hu Wensheng Yang and Xinli Yu 2011 How well can

Public Service Motivation Connect with Occupational Intention International

Review of Administrative Sciences 77 191-211

Liu Bangcheng Ningyu Tang and Xiaomei Zhu 2008 Public Service Motivation and Job

Satisfaction in China An Investigation of Generalisability and Instrumentality

International Journal of Manpower 29 684-699

MacCallum Robert C and Michael W Browne 1993 The Use of Causal Indicators in

Covariance Structure Models Some Practical Issues Psychological Bulletin 114

533-541

MacKenzie Scott B 2003 The dangers of poor construct conceptualization Journal of the

Academy of Marketing Science 31(3) 323-326

MacKenzie Scott B Philip M Podsakoff and Cheryl Burke Jarvis 2005 The Problem of

Measurement Model Misspecification in Behavioral and Organizational Research

and Some Recommended Solutions Journal of Applied Psychology 90 710-730

Martinsen Dorte Sindbjerg and Torben Beck Joslashrgensen 2010 Accountability as a

Differentiated Value in Supranational Governance American Review of Public

Administration 40 742-760

Moynihan Donald P and Sanjay K Pandey 2007a The Role of Organizations in Fostering

Public Service Motivation Public Administration Review 67 40-53

Moynihan Donald P and Sanjay K Pandey 2007b Finding Workable Levers over Work

Motivation Comparing Job Satisfaction Job Involvement and Organizational

Commitment Administration amp Society 39 803-832

Pandey Sanjay K and Edmund C Stazyk 2008 Antecedents and Correlates of Public

Service Motivation In Motivation in Public Management edited by James L Perry

and Annie Hondeghem 101-117 Oxford Oxford University Press

- 31 -

Perry James L 1996 Measuring Public Service Motivation An Assessment of Construct

Reliability and Validity Journal of Public Administration Research and Theory 6

5-22

Perry James L 1997 Antecedents of Public Service Motivation Journal of Public

Administration Research and Theory 7 181-197

Perry James L 2000 Bringing Society in Toward a Theory of Public Service Motivation

Journal of Public Administration Research and Theory 10 471-488

Perry James L and Annie Hondeghem 2008a Motivation in Public Management Oxford

Oxford University Press

Perry James L and Annie Hondeghem 2008b Directions for Future Theory and Research

In Motivation in Public Management edited by James L Perry and Annie

Hondeghem 294-313 Oxford Oxford University Press

Perry James L Annie Hondeghem and Lois Recascino Wise 2010 Revisiting the

Motivational Bases of Public Service Twenty Years of Research and an Agenda for

the Future Public Administration Review 70(5) 681-690

Perry James L and Wouter Vandenabeele 2008 Behavioral dynamics Institutions

Identities and Self-Regulation In Motivation in Public Management edited by

James L Perry and Annie Hondeghem 56-79 Oxford Oxford University Press

Perry James L and Lois Recascino Wise 1990 The Motivational Bases of Public Service

Public Administration Review 50 367-373

Rainey Hal G 1982 Reward Preference among Public and Private Managers In Search of

the Service Ethic American Review of Public Administration 16 288-302

Ringle Christian M Oliver Goumltz Martin Wetzels and Bradley Wilson 2009 On the Use of

Formative Measurement Specifications in Structural Equation Modeling A Monte

Carlo Simulation Study to Compare Covariance-Based and Partial Least Squares

Model Estimation Methodologies Faculty of Economics and Business Administration

Maastricht Univeristy

Ringle C M S Wende and S Will 2005 Smart PLS 20 (M3) Beta

- 32 -

httpwwwsmartplsde

Ritz Adrian 2011 Attraction to Public Policy Making A Qualitative Inquiry into

Improvements in PSM Measurement Forthcoming in Public Administration

Spector Paul E 1992 Summated Rating Scales Construction An Introduction Newbury

Park CA Sage

Steenbergen Marco R 1996 Compassion and American public opinion An analysis of the

NES Humanitarianism Scale Report to the Board of Overseers of the National

Election Studies

Steffen Patrick R and Kevin S Masters 2005 Does compassion mediate the intrinsic

religion-health relationship Annals of Behavioral Medicine 30 217-223

Taylor Jeannette 2007 The Impact of Public Service Motives on Work Outcomes in

Australia A Comparative Multi-Dimensional Analysis Public Administration 85

931-959

Vandenabeele Wouter 2007 Towards a Theory of Public Service Motivation An

Institutional Approach Public Management Review 9 545-556

Vandenabeele Wouter 2008a Government Calling Public Service Motivation as an Element

in Selecting Government as an Employer of Choice Public Administration 86

1089-1105

Vandenabeele Wouter 2008b Development of a Public Service Motivation Measurement

Scale Corroborating and Extending Perryrsquos Measurement Instrument International

Public Management Journal 11 143-167

Vandenabeele Wouter 2009 The Mediating Effect of Job Satisfaction and Organizational

Commitment on Self-Reported Performance More Robust Evidence of the PSM-

Performance Relationship International Review of Administrative Sciences 75 11-

34

Wilcox James B Roy D Howell and Einar Breivik 2008 Questions about Formative

Measurement Journal of Business Research 61 1219-1228

Wise Lois Recascino 2000 The Public Service Culture In Public Administration Concepts

- 33 -

and Cases 7th ed edited by Richard J Stillman II 342-353 Boston Houghton

Mifflin

Wright Bradley E 2008 Methodological Challenges associated with Public Service

Motivation Research In Motivation in Public Management edited by James L

Perry and Annie Hondeghem 80-98 Oxford Oxford University Press

Wright Bradley E and Sanjay K Pandey 2008 Public Service Motivation and the

Assumption of Person-Organization Fit Testing the Mediating Effect of Value

Congruence Administration amp Society 40 502-521

- 34 -

Table 1

Possible Indicators of PSM and Descriptive Statistics (n = 2868)

Dimensions and items Mean Standard

Deviation

Item-Total

Correlation

Attraction to Public Participation (APP)

APP1 I am interested in helping to improve public service

APP2 I am satisfied when I see people benefiting from the public programs I was

involved in

APP3 I like to discuss topics regarding public programs and policies with others

APP4 I believe that public sector activities contribute to our general welfare

APP5 I admire people who initiate or are involved in activities to aid my

community

APP6 Contributing to public programs and policies helps me realize myself

APP7 It is important to contribute to activities that tackle social problems

412

441

355

393

423

351

409

696

662

917

753

672

881

734

524

421

482

412

493

478

552

Commitment to Public Values (CPV)

Sub-dimension for Public Interests (CPV1)

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

CPI3 I would prefer seeing public officials do what is best for the whole

community even if it harmed my interests

CPI4 Serving the public interest is more important than helping a single individual

Sub-dimension for Public Values (CPV2)

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of public

services

CPV3 It is fundamental that public services respond to the needs of the citizens

CPV4 Decisions regarding public services should be democratic despite the time

and effort it takes

CPV5 Everybody is entitled to a good service even if it costs a lot of money

CPV6 It is fundamental that the interests of future generations are taken into

account when developing public policies

CPV7 To act ethically is essential for public servants

CPV8 I believe that public employees must always be aware of the legitimacy of

their activities

CPV9 I personally identify with the aim of protecting individual liberties and rights

418

409

348

337

428

416

433

394

367

426

441

434

414

695

662

899

995

730

704

710

852

932

689

681

692

763

525

596

480

251

484

437

356

325

376

439

462

402

392

Compassion (COM) COM1 It is difficult for me to contain my feelings when I see people in distress

COM2 I feel sympathetic to the plight of the underprivileged

COM3 I empathize with other people who face difficulties

COM4 I have little compassion for people in need who are unwilling to take the

first step to help themselves

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

354

387

388

299

412

391

951

825

707

1083

755

765

399

511

389

011

487

502

Self-Sacrifice (SS)

SS1 Making a difference to society means more to me than personal achievements

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS5 People should give back to society more than they get from it

SS6 Serving other citizens would give me a good feeling even if no one paid me for

it

SS7 I would agree to a good plan to make a better life for the poor even if it costs

me money

357

329

339

302

345

363

350

904

872

949

982

888

1049

895

508

584

454

531

412

368

546

- 35 -

Table 2

Background of Respondents (n = 2868)

Variables Characteristics Respondents

Sex

Male

Female

NA

1209

1622

37

422

566

13

Length of service

(years)

0 ~ 10

10 ~ 20

20 ~ 30

30+

NA

1126

723

484

491

44

393

252

169

171

15

Organizational status

―Do you supervise employees

No

Yes

NA

1472

1076

320

513

375

112

Country

Australia

Belgium

China

Denmark

France

Italy

Korea

Lithuania

The Netherlands

Switzerland

United Kingdom

USA

249

214

230

249

266

162

253

236

249

250

260

250

87

75

80

87

93

56

88

82

87

87

91

87 Note NA = no answer

- 36 -

Table 3

Standardized factor loadings and correlations for the 31-item PSM index (n = 2868)

Dimensions and items SFL (MLE) SFL (D-WLS) Alpha

APP

APP1

APP2

APP4

APP5

APP6

APP7

597

486

455

558

490

600

657

552

514

625

549

665

699

CPV1

CPI1

CPI2

CPI3

CPI4

588

660

488

289

653

732

535

294

542

CPV2

CPV1

CPV2

CPV3

CPV4

CPV5

CPV6

CPV7

CPV8

CPV9

611

546

436

393

418

534

598

496

460

690

631

505

457

486

601

669

573

530

745

COM

COM1

COM2

COM3

COM5

COM6

496

569

498

597

607

516

651

538

670

684

687

SS

SS1

SS2

SS3

SS4

SS5

SS6

SS7

567

736

558

728

467

409

663

660

762

591

695

515

494

709

782

Measures of fit (MLE) χsup2 (df = 424) = 43699 p lt 001 CFI = 835 GFI = 899

RMSEA = 057 [055 059]

Measure of fit (D-WLS) χsup2 (df = 424) = 39247 p lt 001 CFI = 971 GFI = 976

RMSEA = 054 [052 055]

Inter-factor correlations MLE (D-WLS)

1 2 3 4

1 APS

2 CPV1 996 (988)

3 CPV2 902 (890) 798 (798)

4 COM 787 (783) 758 (762) 779 (783)

5 SS 700 (742) 800 (859) 509 (537) 720 (729)

Note SFL = standardized factor loading All factor loadings and correlations are significant at p lt 001

- 37 -

Table 4

Standardized factor loadings and correlations for the 16-item PSM index (n = 2868)

Dimensions and items SFL

(MLE)

SFL

(D-WLS)

APS

APP5 I admire people who initiate or are involved in activities to aid my

community

APP7 It is important to contribute to activities that tackle social problems

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

611

592

637

666

665

669

690

751

CPV

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of

public services

CPV6 It is fundamental that the interests of future generations are taken

into account when developing public policies

CPV7 To act ethically is essential for public servants

612

546

551

590

698

631

618

658

COM

COM1 It is difficult for me to contain my feelings when I see people in

distress

COM2 I feel sympathetic to the plight of the underprivileged

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

490

546

596

604

515

608

669

679

SS

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS7 I would agree to a good plan to make a better life for the poor even if

it costs me money

763

571

766

678

807

629

747

755

Measures of fit (MLE)

χsup2 (df = 98) = 8490 p lt 001

CFI = 939 GFI = 965

RMSEA = 052 [049 055]

Measure of fit (D-WLS)

χsup2 (df = 98) = 6069 p lt 001

CFI = 988 GFI = 992

RMSEA = 043 [039 046]

Alpha CR (D-WLS) Inter-factor correlations MLE (D-WLS)

1 APS 717 788 1 APS 2 CPV 3 COM

2 CPV 663 747 848 (859)

3 COM 645 713 781 (791) 789 (795)

4 SS 784 826 565 (596) 462 (505) 715 (726)

Note SFL = standardized factor loading Alpha = Cronbachrsquos alpha CR = composite reliability

All factor loadings and correlations are significant at p lt 001

- 38 -

Table 5

Results of CFA in the individual countries Standardized factor loadings and measures of fit

Factors

amp Items

Australia Belgium China Denmark France Italy

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

672

756

799

785

804

943

710

703

597

790

799

730

584

675

685

681

731

742

578

748

674

643

765

739

645

704

678

708

812

779

632

604

765

766

672

853

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

552

593

560

699

727

632

624

605

560

713

709

645

702

602

616

724

686

735

567

560

715

654

605

811

608

658

587

696

755

698

673

620

523

782

697

654

COM

COM1

COM2

COM3

COM5

COM6

623

584

662

688

701

739

732

608

475

808

666

581

578

616

605

649

670

673

699

751

737

679

816

776

568

546

801

626

673

850

637

546

741

715

601

822

SS

SS2

SS3

SS4

SS7

654

794

806

892

767

813

727

693

799

793

725

838

857

820

773

896

855

842

806

749

792

829

753

884

847

731

753

889

816

753

815

692

664

795

782

734

Inter-factor

correlations

093

~

752

163

~

810

320

~

773

323

~

825

452

~

827

497

~

855

323

~

850

321

~

847

498

~

899

536

~

902

425

~

829

476

~

865

Sample size 249 214 230 249 266 162

Measures of Fit

χsup2 638 444 881 732 765 554 817 625 735 565 750 564

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 980 100 939 987 969 997 967 994 975 997 949 995

GFI 961 986 941 986 948 991 949 992 955 992 926 986

RMSEA 036 00 063 050 051 026 053 035 045 026 059 033

- 39 -

Table 5 (Continued)

Factors

amp Items

Korea Lithuania The Netherlands Switzerland UK USA

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

619

701

667

706

781

717

666

512

682

751

605

779

557

629

630

620

719

729

535

621

702

643

694

735

655

725

738

710

804

822

535

621

702

643

694

735

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

594

607

722

722

644

722

674

535

533

790

589

711

568

554

538

655

628

545

627

580

615

711

666

704

800

512

555

873

581

646

627

580

615

711

666

704

COM

COM1

COM2

COM3

COM5

COM6

694

677

550

763

736

627

557

639

428

550

696

534

737

623

508

766

616

608

659

592

683

703

661

769

682

583

729

739

644

803

659

592

683

703

661

769

SS

SS2

SS3

SS4

SS7

776

752

679

779

770

811

707

721

635

725

734

706

814

656

743

820

767

764

745

632

689

788

662

744

671

818

832

830

864

802

745

632

689

788

662

744

Inter-factor

correlations

432

~

726

466

~

746

376

~

788

406

~

902

377

~

852

455

~

836

637

~

818

660

~

828

317

~

852

360

~

852

637

~

818

660

~

828

Sample size 253 236 249 250 260 250

Measures of Fit

χsup2 896 573 612 462 773 616 815 715 769 542 815 715

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 951 996 975 100 958 993 957 991 972 998 957 991

GFI 942 988 960 989 951 987 948 990 955 990 948 990

RMSEA 059 028 034 00 050 034 053 044 048 022 053 044

Note All factor loadings and correlations are significant at p lt 001

- 40 -

Appendix 1 Dimensions and possible items of PSM (the first round)

1 Attraction to Public Participation

(1) I am interested in developing public programs helping my country or community (Kim 2009a)

(2) Sharing my views on public policies with others is attractive to me (Kim 2009a)

(3) I am satisfied when seeing people get benefits from the public programs I was involved in (Kim 2009a)

(4) I like to discuss political subjects with others (Giauque et al 2011)

(5) I think the governmental activities contribute to our welfare

(6) I like to initiate actions to help out my community

(7) I do not care much about politicians (Perry 1996)

2 Commitment to Public Values

Possible items for public interest

(1A) Meaningful public service is very important to me (Perry 1996)

(2A) It is important for me to contribute for the common good (Giauque et al 2011)

(3A) I recognize myself with the mission of protection of liberties and rights endorsed by the nation-state

(Giauque et al 2011)

(4A) I would prefer seeing public officials do what is best for the whole community even if it harmed my

interests (Perry 1996)

(5A) To me serving the public interest is more important than helping other people (Vandenabeele 2008b)

Possible items for public values

(1B) I recognize myself with the promotion of the equality of chances endorsed by the nation-state (Giauque et

al 2011)

(2B) It is fundamental that the public service provided to the citizen is regular and continuous (Giauque et al

2011)

(3B) It is fundamental that public service innovates to respond to the new needs of the citizensclients (Giauque

et al 2011)

(4B) The dignity and well-being of all should be the most important concerns in any society (Steenbergen

1996)

(5B) It is important that public servants account for all the costsexpenses they make (Vandenabeele 2008b)

(6B) We have to do everything in our power to pursue the goal of democracy

3 Compassion

(1) It is difficult for me to contain my feelings when I see people in distress (Perry 1996)

(2) I feel sympathetic to the plight of the underprivileged (Kim 2009a)

(3) I am often reminded by daily events how dependent we are on one another (Perry 1996)

(4) I feel compassion for others in the difficulties they are facing (Steffen and Masters 2005)

(5) It is not really my problem if others are in trouble and need help (Beutel and Marini 1995)

(6) I find it hard to be sympathetic toward starving people in foreign lands when there is so much trouble in our

own country (Beutel and Marini 1995)

(7) I get very upset when I see other people treated unfairly (Beutel and Marini 1995)

(8) I would agree to a good plan to make a better life for the poor even if it cost me money (Beutel and Marini

1995)

(9) To me patriotism includes seeing the welfare of others (Perry 1996)

(10) I care very much about other people

4 Self-Sacrifice

(1) Making a difference in society means more to me than personal achievements (Perry 1996)

(2) I am prepared to make enormous sacrifices for the good of society (Perry 1996)

(3) I believe in putting duty before self (Perry 1996)

(4) I am one of those rare people who would risk personal loss to help someone else (Perry 1996)

(5) I think people should give back to society more than they get from it (Perry 1996)

(6) Serving other citizens would give me a good feeling even if no one paid me for it (Perry 1996)

(7) Much of what I do is for a cause bigger than myself (Perry 1996)

- 41 -

Appendix 2 Newly suggested items through the first round

1 Items for the Attraction to Public Participation Dimension

(1) In the public domain every person should be counted as equal

(2) I enjoy to think and argue in a political way

(3) I am satisfied when seeing my ideas being integrated in public policies

(4) I can easily relate to the way of thinking and arguing of politicians

(5) Working in the public sector I can better realize myself

2 Items for the Public Values Sub-Dimension

(1) Public service should give value for money

(2) Cost considerations should play an important role in decisions about public service provision

(3) It is fundamental that the interests of future generations are taken into account when developing public

policies

(4) To act with correct ethics is for public servants as important as their professional competence

(5) Primarily public servants should be accountable to the public and not to their superiors

3 Items for the Self-Sacrifice Dimension

(1) I would agree to a good plan to make a better life for the poor even if it costs me money

- 42 -

Appendix 3 An example of a summary of comments and suggestions (the second round)

On the item ―I am interested in developing public programs helping my country or community

A It is a good sentence but is it possible for everyone to ―develop public programs Even public officials are

limited to do this

B OK

C or the special group

D Double barrel and difficult to translate to my country ―I am interested in helping to improve public services

E Good but I feel that wording requires revision If we draw sample from local government then country is

redundant ―I am interested in developing public policy programs that aim in helping my community

F 4 (General appreciation from1 to 4) Not clear whether it refers to ones work or to ones concern as a citizen

―(In my work) I am interested in contributing to the development of public etchellip

G Instead of development I would use implementing

H Delete public which is redundant because of the characteristic helping my country or community but add

or implementing ―I am interested in developing or implementing programs helping my country or

community

I Referring to dimension Wording Similar to item 6

Note Names have been changed to alphabets in order to ensure the anonymity of participants

- 43 -

Appendix 4 Dimensions and Items of Public Service Motivation for Future Research

Dimensions and items

Attraction to Public Service (APS)

1 I am interested in helping to improve public service

2 I admire people who initiate or are involved in activities to aid my community

3 It is important to contribute to activities that tackle social problems

4 Meaningful public service is very important to me

5 It is important for me to contribute to the common good

Commitment to Public Values (CPV)

1 I think equal opportunities for citizens are very important

2 It is important that citizens can rely on the continuous provision of public services

3 It is fundamental that the interests of future generations are taken into account when

developing public policies

4 To act ethically is essential for public servants

5 I believe that public employees must always be aware of the legitimacy of their activities

6 I personally identify with the aim of protecting individual liberties and rights

Compassion (COM)

1 It is difficult for me to contain my feelings when I see people in distress

2 I feel sympathetic to the plight of the underprivileged

3 I empathize with other people who face difficulties

4 I get very upset when I see other people being treated unfairly

5 Considering the welfare of others is very important

Self-Sacrifice (SS)

1 I am prepared to make sacrifices for the good of society

2 I believe in putting civic duty before self

3 I am willing to risk personal loss to help society

4 I would agree to a good plan to make a better life for the poor even if it costs me money

- 16 -

[Table 4 about here]

Cronbachrsquos coefficient alpha for the 16-item PSM index was 860 and the coefficient

alphas for the four dimensions ranged from 645 to 784 The composite reliability of the set

of reflective indicators for each dimension of PSM ranged from 713 to 826 and indicated

adequate internal consistency The resulting factor structure showed a four-factor structure

with all items loading significantly onto their a priori dimension (p lt 001) and the

standardized factor loadings ranging from 0490 to 0766 in MLE and from 515 to 807 in

D-WLS The results provide support for convergent validity The correlation estimates

between the two factors ranged from 0462 to 0848 in MLE and from 505 to 859 in D-

WLS The confidence intervals (plusmn2 standard errors) around the correlation estimates between

the two factors did not include 100 The result provides support for discriminant validity To

test whether the four-dimension model is superior to other designs such as the two or three-

dimension models a series of CFA were conducted and the results show that the four-

dimension model is a significantly better fit than any of the other designs

To test whether the factorial structure of the PSM instrument is equivalent across the

samples from the twelve countries four nested models were tested as part of multi-group

analysis with MLE (Byrne 2010) First the unconstrained model was the baseline model

which relaxed all equality constraints This model hypothesized the same four-factor structure

of the instrument across twelve countries The fit indices for this baseline model were χsup2 (df =

1176) = 24977 p lt 001 GFI = 0901 CFI = 900 RMSEA = 0020 [019 021] These

indices indicate that the four-factor model of the PSM instrument exhibits acceptable fit

across the samples from the twelve countries Second the measurement weights model tested

- 17 -

the invariance of factor loadings across countries by placing equality constraints on these

parameters The fit indices for this model were χsup2 (df = 1308) = 28088 p lt 001 GFI = 0890

CFI = 887 RMSEA = 0020 [019 021] Since the measurement weights model was nested

within the unconstrained model the chi-square difference text ∆χsup2 (132) = 3111 p lt 001

indicated that some equality constraints of factor loadings did not hold across the twelve

countries Further test results suggested that the equality constraints of structural covariances

and measurement residuals were not upheld across the twelve countries It needs to be

stressed that despite the differences in factor loadings and factor covariances the results

support the main assumption that the same four-factor structure holds across different

countries We may assume that people in different cultures understand PSM in the same

fashion

To find differences in the factor structure of PSM among countries we conducted a

series of CFA with each countryrsquos dataset and followed the same procedure as when

analyzing the sample as a whole As table 5 shows although there are great similarities

between the appropriate items in the countries there are also some differences Even though

the items vary according to national differences the four-factor structure had a good fit to the

data Thus the results confirm that the four-factor structure of PSM can be generalized

[Table 5 about here]

The APS dimension covers both an individualrsquos attraction to public participation and

commitment to public interest but all its items refer to some notion of public interest such as

improving public service aiding community tackling social problems meaningful public

service and contributing to the common good The APS dimension captures an individualrsquos

- 18 -

disposition to serve the public interest through public participation When developing the

items some assumed that the items of committing to public interests can be separated from

those of pursuing public values in the CPV dimension The empirical test showed that

respondents tended to understand committing to public interests as behavior-oriented and so

the items for public interest are not differentiated from participation-oriented items in their

perceptions but are different from those for public values The respondents seem to

understand that pursuing public values is more related to personal beliefs or creeds Thus the

CPV dimension is solely focused on public values and the items of pursing public interests

are combined with those of the APP dimension to make the APS dimension

Public values and their relative importance are neither impervious to change over

time nor the same among all countries (Charles de Jong and Ryan 2011 Davis and West

2009) Moreover some values may be in conflict as Martinsen and Beck Joslashrgensen (2010)

discussed in regard to the tensions between accountability and efficiency It is therefore not

possible to include all public values in one dimension (Beck Joslashrgensen and Bozeman 2007)

Accordingly the items of the CPV dimension are developed for estimating individual

commitment to seek public values which there are general consensus about and which are

generally formulated

When all four dimensions of PSM are included in a survey cross-national

equivalence of the survey can be assured The items represent reflections or manifestations of

each dimension and hence are relatively homogeneous and interchangeable Adding or

deleting items does not change the essential nature of the dimension Thus any combination

of items in a dimension can be regarded as equivalent When using structural equation

modeling (SEM) each dimension needs to have at least three reflective indicators in order to

achieve identification or it needs to have at least two reflective indicators and be allowed to

- 19 -

covary with another dimension that has at least two reflective measures With one indicator

the dimension is identified if the loading and the variance of the uniqueness are both fixed

(Bollen 1989 Edwards 2010) Thus the number of items in each dimension is not restricted

when measuring PSM However we recommend using multiple items for each dimension

because more research is needed to investigate the appropriateness of this PSM instrument

with varying samples

The test results showed that model fit estimators were clearly better for the D-WLS

It means that in this study MLE was more rigorous than D-WLS in assessing the fit of the

model The empirical evidence indicated that MLE may be an acceptable technique for

ordinal data when there is a large sample a five-point Likert scale and the skewness and

kurtosis of variables is within acceptable limits as suggested by Byrne (2010)

Discussion

The analysis reaffirmed the four-dimensional structure of PSM but the results were not

entirely consistent with the original CFA we tested We were ultimately able to converge on a

good-fit operational model for PSM that was also consistent with underlying theory The

analysis is not able to resolve the debate about measurement models however for reasons

discussed below

Choosing the correct measurement model (ie reflective vs formative) is important

as measurement model misspecification results not only in misrepresentation of the epistemic

relationships between constructs and their measures but also in biased parameter estimates of

structural relationships (Jarvis et al 2003 Law and Wong 1999 MacKenzie 2003

MacKenzie Podsakoff and Jarvis 2005) A formative measurement model in isolation is

- 20 -

under-identified and cannot be estimated in the case of covariance-based SEM (CB-SEM)

(Bollen 1989) A construct with formative indicators must emit paths to at least two unrelated

reflective indicators two unrelated latent constructs or some combination of the two in order

to solve the problem of under-identification (Jarvis MacKenzie and Podsakoff 2003

MacCallum and Browne 1993 MacKenzie Podsakoff and Jarvis 2005) Thus ―when

dependent constructs change the empirical nature of the formatively measured construct

changes (Wilcox Howell and Breivik 2008 1223) and ―when specifying formative

measurement models what might seem like an innocuous choice of outcomes to achieve

identification can have fundamental ramifications for the meaning of the formative construct

and the inferences based on it (Edwards 2010 7)

Earlier research found that the association between a construct and its measures

should remain stable regardless of the larger causal model in which the constructs and

measures are embedded (Edwards and Bagozzi 2000) The association between a construct

and its measures is generally stable for reflective measures but associations of formative

measures with their construct are unstable The susceptibility of formative measurement to

both interpretational confounding and external inconsistency recommends that formative

measurement is not an equally attractive alternative to reflective measurement and that

whenever possible researchers should opt for reflective measurement (Howell Breivik and

Wilcox 2007a 2007b Kim Shin and Grover 2010) In the case of second-order models

researchers are limited to only a second-order reflective model in CB-SEM (Chin 2010)

Regardless of the measurement debate and the tradeoffs between reflective or

formative specification this study makes a theoretical argument that PSM is a second-order

formative construct This second-order formative factor model (reflective first-order

formative second-order) is also suggested as an alternative model that captures the

- 21 -

multidimensionality of complex concepts and avoids the shortcomings of first-order

formative measurement (Edwards 2010) However this model itself cannot achieve

identification One possible strategy is to develop direct reflective measures of PSM based on

the view that PSM is a general concept that is not only influenced by its specific dimensions

but also can be assessed using general measures (Edwards 2010 Javis MacKenzie and

Podsakoff 2003)6 However these general measures have not yet been developed for the

PSM construct

Another way around the problem of underidentification of formative models is to

apply the component-based approach known as partial least squares path modeling (PLS-PM)

which can model formative indicators (Fornell and Bookstein 1982 Goumltz Liehr-Gobbers

and Krafft 2010) The CB-SEM and the PLS-PM methods have distinctive statistical

characteristics and selecting an approach to SEM depends on the particular research situation

CB-SEM is the method of choice for theory testing while PLS-PM is appropriate for

prediction-oriented applications (Ringle Goumltz Wetzels and Wilson 2009) One advantage of

the PLS method is that it demands significantly fewer requirements compared to that of

covariance structure analyses but nevertheless delivers consistent estimation results Another

asset of the PLS approach is its ability to deal with formative as well as reflective indicators

even within one structural equation model This indicates that the PLS approach can be

appropriate for explorative analysis of structural equation models (Chin 2010 Goumltz Liehr-

Gobbers and Krafft 2010) A Monte Carlo simulation study to compare CB-SEM and PLS-

PM showed that PLS-PM provides a viable approximation of model parameters when the

prerequisites for CB-SEM are not met (Ringle Goumltz Wetzels and Wilson 2009) A formative

6 For example job satisfaction can be assessed with general reflective measures such as ―in general I am

satisfied with my job and the formative dimensions can be assessed with measures that describe satisfaction

with each job facet (Edwards 2010 15)

- 22 -

composite in the form of indexes needs to be tested to know the relative weights (γ-

parameters) of the dimensions of PSM In PLS-PM a second-order formative factor is

directly measured by observed variables for all the first order factors While this approach

repeats the number of manifest variables used the model can be identified without adding

outcome constructs that would be considered inappropriate or irrelevant from a conceptual

standpoint (Chin 2010 Edwards 2010) The test result of PLS-PM using Smart PLS 20

(Ringle Wende and Will 2005) produced the function PSM = 31APS + 36CPV + 28COM

+ 31SS However one consideration is whether and how construct error is modeled for the

formative construct (Diamantopoulos 2006) When using PLS-PM the researcher is making

the assumption that there is no error at the construct level as all constructs in PLS-PM are

modeled without error When using PLS-PM the absence of an error term will tend to

increase the weights therefore inflating their importance (Bollen and Lennox 1991) The

primary statistic for assessing a formative indicator is its weight the partialized effect of the

indicator on its intended construct controlling for the effects of all other indicators of that

construct (Cenfetelli and Bassellier 2009) A formative measure is essentially a multiple

regression with the construct representing the dependent variable and the indicators as the

predictors (Diamantopoulos and Winklhofer 2001) PSM may be regarded as a weighted

linear composite of its dimensions The weights used to create the composites are different

across CE-SEM and PLS-PM The causal dimensional weights generated in CB-SEM are

estimated in a model in which a construct-level error term is specified (eg PSM = γ1APM +

γ2CPI + γ3COM + γ4SS + ζ) whereas the weights in PLS-PM are not (eg PSM = γ1APM +

γ2CPI + γ3COM + γ4SS) This leads to two different conceptualizations of the PSM construct

one in which the causal indicators do not capture the construct in its entirety and another in

which they do (Hardin et al 2010) These differences need to be discussed in future research

- 23 -

in this area

One practical suggestion to measure PSM is to apply unit weights to its dimensions if

we assume the effects of the four dimensions are equivalent A literature review indicates that

unit weights have substantial predictive validity when compared with regression weights The

literature shows that the empirical correlations between scores generated from differential and

unit-weighting approaches are quite high In creating composite scores unit weights can be

an appropriate approach for weighting under many circumstances (Bobko Roth and Buster

2007) Although the uses of unit-weight composites will likely result in the explanation of

less variance than empirical estimations of dimensional weights they will remain consistently

defined across studies and hence suitable for theory testing Alternatively more thorough

discussions on the dimensions of PSM may suggest fixed-weight composites (Hardin et al

2010) A previous study (Kim 2010 22) also suggested ―Overall indexing PSM by

measuring all the dimensions of PSM and simply summing the averages of each dimension

can make it easier to measure PSM in practice as PSM = APS + CPV + COM + SS

An advantage of regarding PSM as a second-order formative construct is that the

dimensions can be analyzed both individually and collectively This is important as the

dimensions are expected to have different antecedents and consequences For instance Perry

(1997) found some significant differences in the influence of independent variables on the

four dimensions of PSM Taylor (2007) confirmed that certain dimensions are more important

than others in influencing work outcomes Moynihan and Pandey (2007a) reported that red

tape is negatively related to one dimension but reform orientation and hierarchical authority

is positively related to another dimension A recent study found that the SS dimension has a

positive relationship with occupational intention but the other dimensions do not (Liu et al

2011) Thus it is valuable to analyze the causes and effects of individual dimensions of PSM

- 24 -

as well as of the PSM construct

Conclusion

The purpose of this study is to develop a measurement instrument for PSM for cross-national

research and comparison On the basis of the sharpened construct of PSM (Kim and

Vandenabeele 2010) using data from an international survey conducted in 12 countries we

found that the four-factor structure of PSM is generalizable across country cases This study

provides the measurement index of PSM with the four dimensions of APS CPV COM and

SS The APS dimension measures the instrumental motives of PSM the CPV dimension

determines the value-based motives of PSM while the COM and SS dimensions gauge the

identification motives and self-sacrifice respectively When all four dimensions of PSM are

used in a survey cross-national equivalence of survey items can be assured The items for

measuring PSM in future research are suggested in Appendix 4

This study has several limitations First looking for samples which were comparable

to the largest possible extent we opted for surveys in local government The results found

and discussed in this study are based on PSM among local government employees The

results might be different when surveying central government employees or other specific

groups in the public sector Thus the potential impact of the sample bias needs to be

considered in understanding this study It also means that the survey items for measuring

PSM should be examined with different samples in various areas and at different government

levels

Second the test results showed that there is a high degree of correlation among the

dimension such as APS CPV and COM Future research should not only investigate whether

- 25 -

the items reflect the core content of the dimensions but also continue to develop more

appropriate items for better discriminant validity of each dimension Third the content of

APP dimension was assumed to be distinct from the CPV dimension that focuses on personal

disposition to pursue public interest and public values However the public interest sub-

dimension was separated from the public values sub-dimension in the CPV dimension and

combined with the APP dimension to form the APS dimension Further research should

analyze the CPV dimension and investigate the appropriateness of the APS dimension

In order to facilitate research internationally the PSM construct needs to be

improved conceptually and operationally Kim and Vandenabeele (2010) have dealt with the

conceptual composition and operational dimension of PSM as well as the measurement

model of PSM The next step is to develop an internationally robust measurement instrument

of PSM This study proposes that scholars use all four of the dimensions in measuring the

conceptual components of PSM such as self-sacrifice instrumental value-based and affective

motives because they collectively represent PSM It provides a set of appropriate items for

measuring PSM and confirms the dimensional structure of PSM The survey results suggest

that people in different cultures understand PSM in the same fashion We know that this study

is simply a step along the way to developing a more robust measure but we hope it

contributes to enhanced cross-national research and comparison along with the generation of

cumulative knowledge internationally

- 26 -

REFERENCES

Anderson James C and David D Gerbing 1988 Structural Equation Modeling in Practice

A Review and Recommended Two-Step Approach Psychological Bulletin 103

411-423

Arbuckle James L 2009 Amos 180 Userrsquos Guide Crawfordville FL Amos Development

Co

Beck Joslashrgensen Torben and Barry Bozeman 2007 Public Values An Inventory

Administration amp Society 39(3) 354-381

Beutel Ann M and Margaret Mooney Marini 1995 Gender and values American

Sociological Review 60 436 - 448

Bobko Philip Philip L Roth and Maury A Buster 2007 The Usefulness of Unit Weights in

Creating Composite Scores A Literature Review Application to Content Validity and

Meta-Analysis Organizational Research Methods 10(4) 689-709

Bollen Kenneth A 1989 Structural Equations with Latent Variables New York Wiley

Bollen K A and R Lennox 1991 Conventional Wisdom on Measurement A Structural

Equation Perspective Psychological Bulletin 110(2) 305-314

Byrne Barbara M 2010 Structural Equation Modeling with Amos Basic Concepts

Applications and Programming 2nd

ed New York Routledge

Camilleri Emanuel 2006 Towards Developing an Organizational Commitment ― Public

Service Motivation Model for Maltese Public Service Employees Public Policy

and Administration 21 63-83

Castaing Seacutebastien 2006 The Effects of Psychological Contract Fulfillment and Public

Service Motivation on Organizational Commitment in the French Civil Service

Public Policy and Administration 21 84-98

Cenfetelli Ronald T and Geneviegraveve Bassellier 2009 Interpretation of Formative

Measurement in Information Systems Research MIS Quarterly 33(4) 689-707

Charles Michael B W Martin de Jong and Neal Ryan 2011 Public Values in Western

- 27 -

Europe A Temporal Perspective American Review of Public Administration 41

75-91

Chin Wynne W 2010 How to Write Up and Report PLS Analysis In Handbook of Partical

Least Squares edited by V Esposito Vinzi et al 655-690 Berlin Springer-Verlag

Coursey David H and Sanjay K Pandey 2007a Public Service Motivation Measurement

Testing an Abridged Version of Perryrsquos Proposed Scale Administration amp Society

39 547-568

Coursey David H and Sanjay K Pandey 2007b Content Domain Measurement and

Validity of the Red Tape Concept American Review of Public Administration 37

342-361

Coursey David H James L Perry Jeffrey L Brudney and Laura Littlepage 2008

Psychometric Verification of Perryrsquos Public Service Motivation Instrument Results

for Volunteer Exemplars Review of Public Personnel Administration 28 79-90

Davis Paul and Karen West 2009 What Do Public Values Mean for Public Action Putting

Public Values in Their Plural Place American Review of Public Administration 39

602-618

DeHart-Davis Leisha Justin Marlowe and Sanjay K Pandey 2006 Gender Dimensions of

Public Service Motivation Public Administration Review 66 873-887

Diamantopoulos Adamantios 2006 The error term in formative measurement models

Interpretation and modeling implications Journal of Modeling in Management 1(1)

7-17

Diamantopoulos Adamantios and Judy A Siguaw 2006 Formative versus Reflective

Indicators in Organizational Measure Development A Comparison and Empirical

Illustration British Journal of Management 17 263-282

Diamantopoulos Adamantios and Heidi M Winklhofer 2001 Index Construction with

Formative Indicators An Alternative to Scale Development Journal of Marketing

Research 38 269-277

Edwards Jeffrey R 2010 The Fallacy of Formative Measurement Organizational Research

- 28 -

Methods Online First published on August 19 2010 as

doi1011771094428110378369

Edwards Jeffrey R and Richard P Bagozzi 2000 On the Nature and Direction of

Relationships between Constructs and Measures Psychological Methods 5 155-

174

Flora David B and Patrick J Curran 2004 An Empirical Evaluation of Alternative

Methods of Estimation for Confirmatory Factor Analysis with Ordinal Data

Psychological Methods 9 466-491

Fornell Claes and Fred L Bookstein 1982 Two structural equations models LISREL and

PLS applied to customer exit-voice theory Journal of Marketing Research 19 440-

452

Giauque David Adrian Ritz Freacutedeacuteric Varone Simon Anderfuhren-Biget and Christian

Waldner 2011 Putting Public Service Motivation into Context A Balance between

Universalism and Particularism Forthcoming in International Review of

Administrative Sciences 77(2)

Goumltz Oliver Kerstin Liehr-Gobbers and Manfred Krafft 2010 Evaluation of Structural

Equation Models Using the Partial Least Squares (PLS) Approach In V Esposito

Vinzi et al (eds) Handbook of Partial Least Squares 691-711 DOI 101007978-3-

540-32827-8_30 Springer-Verlag Berlin Heidelberg

Guinot Christiane Julie Latreille and Michel Tenenhaus 2001 PLS Path Modelling and

Multiple Table Analysis Application to the Cosmetic Habits of Women in Ile-de-

France Chemometrics and Intelligent Laboratory Systems 58 247-259

Hair Joseph F Jr William C Black Barry J Babin and Rolph E Anderson 2010

Multivariate Data Analysis 7th

ed Upper Saddle River NJ Pearson Education

Hardin Andrew M Jerry Cha-Jan Chang Mark A Fuller and Gholamreza Torkzadeh 2010

Formative Measurement and Academic Research In Search of Measurement

Theory Educational and Psychological Measurement Online First published on

June 11 2010 as doi101177001316440370208

- 29 -

Howell Roy D Einar Breivik and James B Wilcox 2007a Reconsidering Formative

Measurement Psychological Methods 12 205-218

Howell Roy D Einar Breivik and James B Wilcox 2007b Is Formative Measurement

Really Measurement Reply to Bollen (2007) and Bagozzi (2007) Psychological

Methods 12 238-245

Jarvis Cheryl Burke Scott B MacKenzie and Philip M Podsakoff 2003 A Critical Review

of Construct Indicators and Measurement Model Misspecification in Marketing and

Consumer Research Journal of Consumer Research 30 199-218

Joumlreskog Karl G and Dag Soumlrbom 2005 LISREL 872 (Computer Software) Lincolnwood

IL Scientific Software International

Kim Gimun Bongsik Shin and Varun Grover 2010 Investigating Two Contradictory Views

of Formative Measurement in Information Systems Research MIS Quarterly 34

345-365

Kim Sangmook 2009a Revising Perryrsquos Measurement Scale of Public Service Motivation

American Review of Public Administration 39 149-163

Kim Sangmook 2009b Testing the Structure of Public Service Motivation in Korea A

Research Note Journal of Public Administration Research and Theory 19 839-

851

Kim Sangmook 2010 Testing a Revised Measurement of Public Service Motivation

Reflective versus Formative Specification Journal of Public Administration

Research and Theory Doi101093jopartmuq048

Kim Sangmook and Wouter Vandenabeele 2010 A strategy for building public service

motivation research internationally Public Administration Review 70 701-709

Kline Rex B 2005 Principles and Practice of Structural Equation Modeling 2nd

ed New

York Guilford Press

Law Kenneth and Chi-Sum Wong 1999 Multidimensional Constructs in Structural

Equation Analysis An Illustration using the Job Perception and Job Satisfaction

Constructs Journal of Management 25 143-160

- 30 -

Leisink Peter and Bram Steijn 2009 Public Service Motivation and Job Performance of

Public Sector Employees in the Netherlands International Review of

Administrative Sciences 75 35-52

Liu Bangcheng Chun Hui Jin Hu Wensheng Yang and Xinli Yu 2011 How well can

Public Service Motivation Connect with Occupational Intention International

Review of Administrative Sciences 77 191-211

Liu Bangcheng Ningyu Tang and Xiaomei Zhu 2008 Public Service Motivation and Job

Satisfaction in China An Investigation of Generalisability and Instrumentality

International Journal of Manpower 29 684-699

MacCallum Robert C and Michael W Browne 1993 The Use of Causal Indicators in

Covariance Structure Models Some Practical Issues Psychological Bulletin 114

533-541

MacKenzie Scott B 2003 The dangers of poor construct conceptualization Journal of the

Academy of Marketing Science 31(3) 323-326

MacKenzie Scott B Philip M Podsakoff and Cheryl Burke Jarvis 2005 The Problem of

Measurement Model Misspecification in Behavioral and Organizational Research

and Some Recommended Solutions Journal of Applied Psychology 90 710-730

Martinsen Dorte Sindbjerg and Torben Beck Joslashrgensen 2010 Accountability as a

Differentiated Value in Supranational Governance American Review of Public

Administration 40 742-760

Moynihan Donald P and Sanjay K Pandey 2007a The Role of Organizations in Fostering

Public Service Motivation Public Administration Review 67 40-53

Moynihan Donald P and Sanjay K Pandey 2007b Finding Workable Levers over Work

Motivation Comparing Job Satisfaction Job Involvement and Organizational

Commitment Administration amp Society 39 803-832

Pandey Sanjay K and Edmund C Stazyk 2008 Antecedents and Correlates of Public

Service Motivation In Motivation in Public Management edited by James L Perry

and Annie Hondeghem 101-117 Oxford Oxford University Press

- 31 -

Perry James L 1996 Measuring Public Service Motivation An Assessment of Construct

Reliability and Validity Journal of Public Administration Research and Theory 6

5-22

Perry James L 1997 Antecedents of Public Service Motivation Journal of Public

Administration Research and Theory 7 181-197

Perry James L 2000 Bringing Society in Toward a Theory of Public Service Motivation

Journal of Public Administration Research and Theory 10 471-488

Perry James L and Annie Hondeghem 2008a Motivation in Public Management Oxford

Oxford University Press

Perry James L and Annie Hondeghem 2008b Directions for Future Theory and Research

In Motivation in Public Management edited by James L Perry and Annie

Hondeghem 294-313 Oxford Oxford University Press

Perry James L Annie Hondeghem and Lois Recascino Wise 2010 Revisiting the

Motivational Bases of Public Service Twenty Years of Research and an Agenda for

the Future Public Administration Review 70(5) 681-690

Perry James L and Wouter Vandenabeele 2008 Behavioral dynamics Institutions

Identities and Self-Regulation In Motivation in Public Management edited by

James L Perry and Annie Hondeghem 56-79 Oxford Oxford University Press

Perry James L and Lois Recascino Wise 1990 The Motivational Bases of Public Service

Public Administration Review 50 367-373

Rainey Hal G 1982 Reward Preference among Public and Private Managers In Search of

the Service Ethic American Review of Public Administration 16 288-302

Ringle Christian M Oliver Goumltz Martin Wetzels and Bradley Wilson 2009 On the Use of

Formative Measurement Specifications in Structural Equation Modeling A Monte

Carlo Simulation Study to Compare Covariance-Based and Partial Least Squares

Model Estimation Methodologies Faculty of Economics and Business Administration

Maastricht Univeristy

Ringle C M S Wende and S Will 2005 Smart PLS 20 (M3) Beta

- 32 -

httpwwwsmartplsde

Ritz Adrian 2011 Attraction to Public Policy Making A Qualitative Inquiry into

Improvements in PSM Measurement Forthcoming in Public Administration

Spector Paul E 1992 Summated Rating Scales Construction An Introduction Newbury

Park CA Sage

Steenbergen Marco R 1996 Compassion and American public opinion An analysis of the

NES Humanitarianism Scale Report to the Board of Overseers of the National

Election Studies

Steffen Patrick R and Kevin S Masters 2005 Does compassion mediate the intrinsic

religion-health relationship Annals of Behavioral Medicine 30 217-223

Taylor Jeannette 2007 The Impact of Public Service Motives on Work Outcomes in

Australia A Comparative Multi-Dimensional Analysis Public Administration 85

931-959

Vandenabeele Wouter 2007 Towards a Theory of Public Service Motivation An

Institutional Approach Public Management Review 9 545-556

Vandenabeele Wouter 2008a Government Calling Public Service Motivation as an Element

in Selecting Government as an Employer of Choice Public Administration 86

1089-1105

Vandenabeele Wouter 2008b Development of a Public Service Motivation Measurement

Scale Corroborating and Extending Perryrsquos Measurement Instrument International

Public Management Journal 11 143-167

Vandenabeele Wouter 2009 The Mediating Effect of Job Satisfaction and Organizational

Commitment on Self-Reported Performance More Robust Evidence of the PSM-

Performance Relationship International Review of Administrative Sciences 75 11-

34

Wilcox James B Roy D Howell and Einar Breivik 2008 Questions about Formative

Measurement Journal of Business Research 61 1219-1228

Wise Lois Recascino 2000 The Public Service Culture In Public Administration Concepts

- 33 -

and Cases 7th ed edited by Richard J Stillman II 342-353 Boston Houghton

Mifflin

Wright Bradley E 2008 Methodological Challenges associated with Public Service

Motivation Research In Motivation in Public Management edited by James L

Perry and Annie Hondeghem 80-98 Oxford Oxford University Press

Wright Bradley E and Sanjay K Pandey 2008 Public Service Motivation and the

Assumption of Person-Organization Fit Testing the Mediating Effect of Value

Congruence Administration amp Society 40 502-521

- 34 -

Table 1

Possible Indicators of PSM and Descriptive Statistics (n = 2868)

Dimensions and items Mean Standard

Deviation

Item-Total

Correlation

Attraction to Public Participation (APP)

APP1 I am interested in helping to improve public service

APP2 I am satisfied when I see people benefiting from the public programs I was

involved in

APP3 I like to discuss topics regarding public programs and policies with others

APP4 I believe that public sector activities contribute to our general welfare

APP5 I admire people who initiate or are involved in activities to aid my

community

APP6 Contributing to public programs and policies helps me realize myself

APP7 It is important to contribute to activities that tackle social problems

412

441

355

393

423

351

409

696

662

917

753

672

881

734

524

421

482

412

493

478

552

Commitment to Public Values (CPV)

Sub-dimension for Public Interests (CPV1)

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

CPI3 I would prefer seeing public officials do what is best for the whole

community even if it harmed my interests

CPI4 Serving the public interest is more important than helping a single individual

Sub-dimension for Public Values (CPV2)

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of public

services

CPV3 It is fundamental that public services respond to the needs of the citizens

CPV4 Decisions regarding public services should be democratic despite the time

and effort it takes

CPV5 Everybody is entitled to a good service even if it costs a lot of money

CPV6 It is fundamental that the interests of future generations are taken into

account when developing public policies

CPV7 To act ethically is essential for public servants

CPV8 I believe that public employees must always be aware of the legitimacy of

their activities

CPV9 I personally identify with the aim of protecting individual liberties and rights

418

409

348

337

428

416

433

394

367

426

441

434

414

695

662

899

995

730

704

710

852

932

689

681

692

763

525

596

480

251

484

437

356

325

376

439

462

402

392

Compassion (COM) COM1 It is difficult for me to contain my feelings when I see people in distress

COM2 I feel sympathetic to the plight of the underprivileged

COM3 I empathize with other people who face difficulties

COM4 I have little compassion for people in need who are unwilling to take the

first step to help themselves

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

354

387

388

299

412

391

951

825

707

1083

755

765

399

511

389

011

487

502

Self-Sacrifice (SS)

SS1 Making a difference to society means more to me than personal achievements

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS5 People should give back to society more than they get from it

SS6 Serving other citizens would give me a good feeling even if no one paid me for

it

SS7 I would agree to a good plan to make a better life for the poor even if it costs

me money

357

329

339

302

345

363

350

904

872

949

982

888

1049

895

508

584

454

531

412

368

546

- 35 -

Table 2

Background of Respondents (n = 2868)

Variables Characteristics Respondents

Sex

Male

Female

NA

1209

1622

37

422

566

13

Length of service

(years)

0 ~ 10

10 ~ 20

20 ~ 30

30+

NA

1126

723

484

491

44

393

252

169

171

15

Organizational status

―Do you supervise employees

No

Yes

NA

1472

1076

320

513

375

112

Country

Australia

Belgium

China

Denmark

France

Italy

Korea

Lithuania

The Netherlands

Switzerland

United Kingdom

USA

249

214

230

249

266

162

253

236

249

250

260

250

87

75

80

87

93

56

88

82

87

87

91

87 Note NA = no answer

- 36 -

Table 3

Standardized factor loadings and correlations for the 31-item PSM index (n = 2868)

Dimensions and items SFL (MLE) SFL (D-WLS) Alpha

APP

APP1

APP2

APP4

APP5

APP6

APP7

597

486

455

558

490

600

657

552

514

625

549

665

699

CPV1

CPI1

CPI2

CPI3

CPI4

588

660

488

289

653

732

535

294

542

CPV2

CPV1

CPV2

CPV3

CPV4

CPV5

CPV6

CPV7

CPV8

CPV9

611

546

436

393

418

534

598

496

460

690

631

505

457

486

601

669

573

530

745

COM

COM1

COM2

COM3

COM5

COM6

496

569

498

597

607

516

651

538

670

684

687

SS

SS1

SS2

SS3

SS4

SS5

SS6

SS7

567

736

558

728

467

409

663

660

762

591

695

515

494

709

782

Measures of fit (MLE) χsup2 (df = 424) = 43699 p lt 001 CFI = 835 GFI = 899

RMSEA = 057 [055 059]

Measure of fit (D-WLS) χsup2 (df = 424) = 39247 p lt 001 CFI = 971 GFI = 976

RMSEA = 054 [052 055]

Inter-factor correlations MLE (D-WLS)

1 2 3 4

1 APS

2 CPV1 996 (988)

3 CPV2 902 (890) 798 (798)

4 COM 787 (783) 758 (762) 779 (783)

5 SS 700 (742) 800 (859) 509 (537) 720 (729)

Note SFL = standardized factor loading All factor loadings and correlations are significant at p lt 001

- 37 -

Table 4

Standardized factor loadings and correlations for the 16-item PSM index (n = 2868)

Dimensions and items SFL

(MLE)

SFL

(D-WLS)

APS

APP5 I admire people who initiate or are involved in activities to aid my

community

APP7 It is important to contribute to activities that tackle social problems

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

611

592

637

666

665

669

690

751

CPV

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of

public services

CPV6 It is fundamental that the interests of future generations are taken

into account when developing public policies

CPV7 To act ethically is essential for public servants

612

546

551

590

698

631

618

658

COM

COM1 It is difficult for me to contain my feelings when I see people in

distress

COM2 I feel sympathetic to the plight of the underprivileged

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

490

546

596

604

515

608

669

679

SS

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS7 I would agree to a good plan to make a better life for the poor even if

it costs me money

763

571

766

678

807

629

747

755

Measures of fit (MLE)

χsup2 (df = 98) = 8490 p lt 001

CFI = 939 GFI = 965

RMSEA = 052 [049 055]

Measure of fit (D-WLS)

χsup2 (df = 98) = 6069 p lt 001

CFI = 988 GFI = 992

RMSEA = 043 [039 046]

Alpha CR (D-WLS) Inter-factor correlations MLE (D-WLS)

1 APS 717 788 1 APS 2 CPV 3 COM

2 CPV 663 747 848 (859)

3 COM 645 713 781 (791) 789 (795)

4 SS 784 826 565 (596) 462 (505) 715 (726)

Note SFL = standardized factor loading Alpha = Cronbachrsquos alpha CR = composite reliability

All factor loadings and correlations are significant at p lt 001

- 38 -

Table 5

Results of CFA in the individual countries Standardized factor loadings and measures of fit

Factors

amp Items

Australia Belgium China Denmark France Italy

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

672

756

799

785

804

943

710

703

597

790

799

730

584

675

685

681

731

742

578

748

674

643

765

739

645

704

678

708

812

779

632

604

765

766

672

853

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

552

593

560

699

727

632

624

605

560

713

709

645

702

602

616

724

686

735

567

560

715

654

605

811

608

658

587

696

755

698

673

620

523

782

697

654

COM

COM1

COM2

COM3

COM5

COM6

623

584

662

688

701

739

732

608

475

808

666

581

578

616

605

649

670

673

699

751

737

679

816

776

568

546

801

626

673

850

637

546

741

715

601

822

SS

SS2

SS3

SS4

SS7

654

794

806

892

767

813

727

693

799

793

725

838

857

820

773

896

855

842

806

749

792

829

753

884

847

731

753

889

816

753

815

692

664

795

782

734

Inter-factor

correlations

093

~

752

163

~

810

320

~

773

323

~

825

452

~

827

497

~

855

323

~

850

321

~

847

498

~

899

536

~

902

425

~

829

476

~

865

Sample size 249 214 230 249 266 162

Measures of Fit

χsup2 638 444 881 732 765 554 817 625 735 565 750 564

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 980 100 939 987 969 997 967 994 975 997 949 995

GFI 961 986 941 986 948 991 949 992 955 992 926 986

RMSEA 036 00 063 050 051 026 053 035 045 026 059 033

- 39 -

Table 5 (Continued)

Factors

amp Items

Korea Lithuania The Netherlands Switzerland UK USA

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

619

701

667

706

781

717

666

512

682

751

605

779

557

629

630

620

719

729

535

621

702

643

694

735

655

725

738

710

804

822

535

621

702

643

694

735

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

594

607

722

722

644

722

674

535

533

790

589

711

568

554

538

655

628

545

627

580

615

711

666

704

800

512

555

873

581

646

627

580

615

711

666

704

COM

COM1

COM2

COM3

COM5

COM6

694

677

550

763

736

627

557

639

428

550

696

534

737

623

508

766

616

608

659

592

683

703

661

769

682

583

729

739

644

803

659

592

683

703

661

769

SS

SS2

SS3

SS4

SS7

776

752

679

779

770

811

707

721

635

725

734

706

814

656

743

820

767

764

745

632

689

788

662

744

671

818

832

830

864

802

745

632

689

788

662

744

Inter-factor

correlations

432

~

726

466

~

746

376

~

788

406

~

902

377

~

852

455

~

836

637

~

818

660

~

828

317

~

852

360

~

852

637

~

818

660

~

828

Sample size 253 236 249 250 260 250

Measures of Fit

χsup2 896 573 612 462 773 616 815 715 769 542 815 715

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 951 996 975 100 958 993 957 991 972 998 957 991

GFI 942 988 960 989 951 987 948 990 955 990 948 990

RMSEA 059 028 034 00 050 034 053 044 048 022 053 044

Note All factor loadings and correlations are significant at p lt 001

- 40 -

Appendix 1 Dimensions and possible items of PSM (the first round)

1 Attraction to Public Participation

(1) I am interested in developing public programs helping my country or community (Kim 2009a)

(2) Sharing my views on public policies with others is attractive to me (Kim 2009a)

(3) I am satisfied when seeing people get benefits from the public programs I was involved in (Kim 2009a)

(4) I like to discuss political subjects with others (Giauque et al 2011)

(5) I think the governmental activities contribute to our welfare

(6) I like to initiate actions to help out my community

(7) I do not care much about politicians (Perry 1996)

2 Commitment to Public Values

Possible items for public interest

(1A) Meaningful public service is very important to me (Perry 1996)

(2A) It is important for me to contribute for the common good (Giauque et al 2011)

(3A) I recognize myself with the mission of protection of liberties and rights endorsed by the nation-state

(Giauque et al 2011)

(4A) I would prefer seeing public officials do what is best for the whole community even if it harmed my

interests (Perry 1996)

(5A) To me serving the public interest is more important than helping other people (Vandenabeele 2008b)

Possible items for public values

(1B) I recognize myself with the promotion of the equality of chances endorsed by the nation-state (Giauque et

al 2011)

(2B) It is fundamental that the public service provided to the citizen is regular and continuous (Giauque et al

2011)

(3B) It is fundamental that public service innovates to respond to the new needs of the citizensclients (Giauque

et al 2011)

(4B) The dignity and well-being of all should be the most important concerns in any society (Steenbergen

1996)

(5B) It is important that public servants account for all the costsexpenses they make (Vandenabeele 2008b)

(6B) We have to do everything in our power to pursue the goal of democracy

3 Compassion

(1) It is difficult for me to contain my feelings when I see people in distress (Perry 1996)

(2) I feel sympathetic to the plight of the underprivileged (Kim 2009a)

(3) I am often reminded by daily events how dependent we are on one another (Perry 1996)

(4) I feel compassion for others in the difficulties they are facing (Steffen and Masters 2005)

(5) It is not really my problem if others are in trouble and need help (Beutel and Marini 1995)

(6) I find it hard to be sympathetic toward starving people in foreign lands when there is so much trouble in our

own country (Beutel and Marini 1995)

(7) I get very upset when I see other people treated unfairly (Beutel and Marini 1995)

(8) I would agree to a good plan to make a better life for the poor even if it cost me money (Beutel and Marini

1995)

(9) To me patriotism includes seeing the welfare of others (Perry 1996)

(10) I care very much about other people

4 Self-Sacrifice

(1) Making a difference in society means more to me than personal achievements (Perry 1996)

(2) I am prepared to make enormous sacrifices for the good of society (Perry 1996)

(3) I believe in putting duty before self (Perry 1996)

(4) I am one of those rare people who would risk personal loss to help someone else (Perry 1996)

(5) I think people should give back to society more than they get from it (Perry 1996)

(6) Serving other citizens would give me a good feeling even if no one paid me for it (Perry 1996)

(7) Much of what I do is for a cause bigger than myself (Perry 1996)

- 41 -

Appendix 2 Newly suggested items through the first round

1 Items for the Attraction to Public Participation Dimension

(1) In the public domain every person should be counted as equal

(2) I enjoy to think and argue in a political way

(3) I am satisfied when seeing my ideas being integrated in public policies

(4) I can easily relate to the way of thinking and arguing of politicians

(5) Working in the public sector I can better realize myself

2 Items for the Public Values Sub-Dimension

(1) Public service should give value for money

(2) Cost considerations should play an important role in decisions about public service provision

(3) It is fundamental that the interests of future generations are taken into account when developing public

policies

(4) To act with correct ethics is for public servants as important as their professional competence

(5) Primarily public servants should be accountable to the public and not to their superiors

3 Items for the Self-Sacrifice Dimension

(1) I would agree to a good plan to make a better life for the poor even if it costs me money

- 42 -

Appendix 3 An example of a summary of comments and suggestions (the second round)

On the item ―I am interested in developing public programs helping my country or community

A It is a good sentence but is it possible for everyone to ―develop public programs Even public officials are

limited to do this

B OK

C or the special group

D Double barrel and difficult to translate to my country ―I am interested in helping to improve public services

E Good but I feel that wording requires revision If we draw sample from local government then country is

redundant ―I am interested in developing public policy programs that aim in helping my community

F 4 (General appreciation from1 to 4) Not clear whether it refers to ones work or to ones concern as a citizen

―(In my work) I am interested in contributing to the development of public etchellip

G Instead of development I would use implementing

H Delete public which is redundant because of the characteristic helping my country or community but add

or implementing ―I am interested in developing or implementing programs helping my country or

community

I Referring to dimension Wording Similar to item 6

Note Names have been changed to alphabets in order to ensure the anonymity of participants

- 43 -

Appendix 4 Dimensions and Items of Public Service Motivation for Future Research

Dimensions and items

Attraction to Public Service (APS)

1 I am interested in helping to improve public service

2 I admire people who initiate or are involved in activities to aid my community

3 It is important to contribute to activities that tackle social problems

4 Meaningful public service is very important to me

5 It is important for me to contribute to the common good

Commitment to Public Values (CPV)

1 I think equal opportunities for citizens are very important

2 It is important that citizens can rely on the continuous provision of public services

3 It is fundamental that the interests of future generations are taken into account when

developing public policies

4 To act ethically is essential for public servants

5 I believe that public employees must always be aware of the legitimacy of their activities

6 I personally identify with the aim of protecting individual liberties and rights

Compassion (COM)

1 It is difficult for me to contain my feelings when I see people in distress

2 I feel sympathetic to the plight of the underprivileged

3 I empathize with other people who face difficulties

4 I get very upset when I see other people being treated unfairly

5 Considering the welfare of others is very important

Self-Sacrifice (SS)

1 I am prepared to make sacrifices for the good of society

2 I believe in putting civic duty before self

3 I am willing to risk personal loss to help society

4 I would agree to a good plan to make a better life for the poor even if it costs me money

- 17 -

the invariance of factor loadings across countries by placing equality constraints on these

parameters The fit indices for this model were χsup2 (df = 1308) = 28088 p lt 001 GFI = 0890

CFI = 887 RMSEA = 0020 [019 021] Since the measurement weights model was nested

within the unconstrained model the chi-square difference text ∆χsup2 (132) = 3111 p lt 001

indicated that some equality constraints of factor loadings did not hold across the twelve

countries Further test results suggested that the equality constraints of structural covariances

and measurement residuals were not upheld across the twelve countries It needs to be

stressed that despite the differences in factor loadings and factor covariances the results

support the main assumption that the same four-factor structure holds across different

countries We may assume that people in different cultures understand PSM in the same

fashion

To find differences in the factor structure of PSM among countries we conducted a

series of CFA with each countryrsquos dataset and followed the same procedure as when

analyzing the sample as a whole As table 5 shows although there are great similarities

between the appropriate items in the countries there are also some differences Even though

the items vary according to national differences the four-factor structure had a good fit to the

data Thus the results confirm that the four-factor structure of PSM can be generalized

[Table 5 about here]

The APS dimension covers both an individualrsquos attraction to public participation and

commitment to public interest but all its items refer to some notion of public interest such as

improving public service aiding community tackling social problems meaningful public

service and contributing to the common good The APS dimension captures an individualrsquos

- 18 -

disposition to serve the public interest through public participation When developing the

items some assumed that the items of committing to public interests can be separated from

those of pursuing public values in the CPV dimension The empirical test showed that

respondents tended to understand committing to public interests as behavior-oriented and so

the items for public interest are not differentiated from participation-oriented items in their

perceptions but are different from those for public values The respondents seem to

understand that pursuing public values is more related to personal beliefs or creeds Thus the

CPV dimension is solely focused on public values and the items of pursing public interests

are combined with those of the APP dimension to make the APS dimension

Public values and their relative importance are neither impervious to change over

time nor the same among all countries (Charles de Jong and Ryan 2011 Davis and West

2009) Moreover some values may be in conflict as Martinsen and Beck Joslashrgensen (2010)

discussed in regard to the tensions between accountability and efficiency It is therefore not

possible to include all public values in one dimension (Beck Joslashrgensen and Bozeman 2007)

Accordingly the items of the CPV dimension are developed for estimating individual

commitment to seek public values which there are general consensus about and which are

generally formulated

When all four dimensions of PSM are included in a survey cross-national

equivalence of the survey can be assured The items represent reflections or manifestations of

each dimension and hence are relatively homogeneous and interchangeable Adding or

deleting items does not change the essential nature of the dimension Thus any combination

of items in a dimension can be regarded as equivalent When using structural equation

modeling (SEM) each dimension needs to have at least three reflective indicators in order to

achieve identification or it needs to have at least two reflective indicators and be allowed to

- 19 -

covary with another dimension that has at least two reflective measures With one indicator

the dimension is identified if the loading and the variance of the uniqueness are both fixed

(Bollen 1989 Edwards 2010) Thus the number of items in each dimension is not restricted

when measuring PSM However we recommend using multiple items for each dimension

because more research is needed to investigate the appropriateness of this PSM instrument

with varying samples

The test results showed that model fit estimators were clearly better for the D-WLS

It means that in this study MLE was more rigorous than D-WLS in assessing the fit of the

model The empirical evidence indicated that MLE may be an acceptable technique for

ordinal data when there is a large sample a five-point Likert scale and the skewness and

kurtosis of variables is within acceptable limits as suggested by Byrne (2010)

Discussion

The analysis reaffirmed the four-dimensional structure of PSM but the results were not

entirely consistent with the original CFA we tested We were ultimately able to converge on a

good-fit operational model for PSM that was also consistent with underlying theory The

analysis is not able to resolve the debate about measurement models however for reasons

discussed below

Choosing the correct measurement model (ie reflective vs formative) is important

as measurement model misspecification results not only in misrepresentation of the epistemic

relationships between constructs and their measures but also in biased parameter estimates of

structural relationships (Jarvis et al 2003 Law and Wong 1999 MacKenzie 2003

MacKenzie Podsakoff and Jarvis 2005) A formative measurement model in isolation is

- 20 -

under-identified and cannot be estimated in the case of covariance-based SEM (CB-SEM)

(Bollen 1989) A construct with formative indicators must emit paths to at least two unrelated

reflective indicators two unrelated latent constructs or some combination of the two in order

to solve the problem of under-identification (Jarvis MacKenzie and Podsakoff 2003

MacCallum and Browne 1993 MacKenzie Podsakoff and Jarvis 2005) Thus ―when

dependent constructs change the empirical nature of the formatively measured construct

changes (Wilcox Howell and Breivik 2008 1223) and ―when specifying formative

measurement models what might seem like an innocuous choice of outcomes to achieve

identification can have fundamental ramifications for the meaning of the formative construct

and the inferences based on it (Edwards 2010 7)

Earlier research found that the association between a construct and its measures

should remain stable regardless of the larger causal model in which the constructs and

measures are embedded (Edwards and Bagozzi 2000) The association between a construct

and its measures is generally stable for reflective measures but associations of formative

measures with their construct are unstable The susceptibility of formative measurement to

both interpretational confounding and external inconsistency recommends that formative

measurement is not an equally attractive alternative to reflective measurement and that

whenever possible researchers should opt for reflective measurement (Howell Breivik and

Wilcox 2007a 2007b Kim Shin and Grover 2010) In the case of second-order models

researchers are limited to only a second-order reflective model in CB-SEM (Chin 2010)

Regardless of the measurement debate and the tradeoffs between reflective or

formative specification this study makes a theoretical argument that PSM is a second-order

formative construct This second-order formative factor model (reflective first-order

formative second-order) is also suggested as an alternative model that captures the

- 21 -

multidimensionality of complex concepts and avoids the shortcomings of first-order

formative measurement (Edwards 2010) However this model itself cannot achieve

identification One possible strategy is to develop direct reflective measures of PSM based on

the view that PSM is a general concept that is not only influenced by its specific dimensions

but also can be assessed using general measures (Edwards 2010 Javis MacKenzie and

Podsakoff 2003)6 However these general measures have not yet been developed for the

PSM construct

Another way around the problem of underidentification of formative models is to

apply the component-based approach known as partial least squares path modeling (PLS-PM)

which can model formative indicators (Fornell and Bookstein 1982 Goumltz Liehr-Gobbers

and Krafft 2010) The CB-SEM and the PLS-PM methods have distinctive statistical

characteristics and selecting an approach to SEM depends on the particular research situation

CB-SEM is the method of choice for theory testing while PLS-PM is appropriate for

prediction-oriented applications (Ringle Goumltz Wetzels and Wilson 2009) One advantage of

the PLS method is that it demands significantly fewer requirements compared to that of

covariance structure analyses but nevertheless delivers consistent estimation results Another

asset of the PLS approach is its ability to deal with formative as well as reflective indicators

even within one structural equation model This indicates that the PLS approach can be

appropriate for explorative analysis of structural equation models (Chin 2010 Goumltz Liehr-

Gobbers and Krafft 2010) A Monte Carlo simulation study to compare CB-SEM and PLS-

PM showed that PLS-PM provides a viable approximation of model parameters when the

prerequisites for CB-SEM are not met (Ringle Goumltz Wetzels and Wilson 2009) A formative

6 For example job satisfaction can be assessed with general reflective measures such as ―in general I am

satisfied with my job and the formative dimensions can be assessed with measures that describe satisfaction

with each job facet (Edwards 2010 15)

- 22 -

composite in the form of indexes needs to be tested to know the relative weights (γ-

parameters) of the dimensions of PSM In PLS-PM a second-order formative factor is

directly measured by observed variables for all the first order factors While this approach

repeats the number of manifest variables used the model can be identified without adding

outcome constructs that would be considered inappropriate or irrelevant from a conceptual

standpoint (Chin 2010 Edwards 2010) The test result of PLS-PM using Smart PLS 20

(Ringle Wende and Will 2005) produced the function PSM = 31APS + 36CPV + 28COM

+ 31SS However one consideration is whether and how construct error is modeled for the

formative construct (Diamantopoulos 2006) When using PLS-PM the researcher is making

the assumption that there is no error at the construct level as all constructs in PLS-PM are

modeled without error When using PLS-PM the absence of an error term will tend to

increase the weights therefore inflating their importance (Bollen and Lennox 1991) The

primary statistic for assessing a formative indicator is its weight the partialized effect of the

indicator on its intended construct controlling for the effects of all other indicators of that

construct (Cenfetelli and Bassellier 2009) A formative measure is essentially a multiple

regression with the construct representing the dependent variable and the indicators as the

predictors (Diamantopoulos and Winklhofer 2001) PSM may be regarded as a weighted

linear composite of its dimensions The weights used to create the composites are different

across CE-SEM and PLS-PM The causal dimensional weights generated in CB-SEM are

estimated in a model in which a construct-level error term is specified (eg PSM = γ1APM +

γ2CPI + γ3COM + γ4SS + ζ) whereas the weights in PLS-PM are not (eg PSM = γ1APM +

γ2CPI + γ3COM + γ4SS) This leads to two different conceptualizations of the PSM construct

one in which the causal indicators do not capture the construct in its entirety and another in

which they do (Hardin et al 2010) These differences need to be discussed in future research

- 23 -

in this area

One practical suggestion to measure PSM is to apply unit weights to its dimensions if

we assume the effects of the four dimensions are equivalent A literature review indicates that

unit weights have substantial predictive validity when compared with regression weights The

literature shows that the empirical correlations between scores generated from differential and

unit-weighting approaches are quite high In creating composite scores unit weights can be

an appropriate approach for weighting under many circumstances (Bobko Roth and Buster

2007) Although the uses of unit-weight composites will likely result in the explanation of

less variance than empirical estimations of dimensional weights they will remain consistently

defined across studies and hence suitable for theory testing Alternatively more thorough

discussions on the dimensions of PSM may suggest fixed-weight composites (Hardin et al

2010) A previous study (Kim 2010 22) also suggested ―Overall indexing PSM by

measuring all the dimensions of PSM and simply summing the averages of each dimension

can make it easier to measure PSM in practice as PSM = APS + CPV + COM + SS

An advantage of regarding PSM as a second-order formative construct is that the

dimensions can be analyzed both individually and collectively This is important as the

dimensions are expected to have different antecedents and consequences For instance Perry

(1997) found some significant differences in the influence of independent variables on the

four dimensions of PSM Taylor (2007) confirmed that certain dimensions are more important

than others in influencing work outcomes Moynihan and Pandey (2007a) reported that red

tape is negatively related to one dimension but reform orientation and hierarchical authority

is positively related to another dimension A recent study found that the SS dimension has a

positive relationship with occupational intention but the other dimensions do not (Liu et al

2011) Thus it is valuable to analyze the causes and effects of individual dimensions of PSM

- 24 -

as well as of the PSM construct

Conclusion

The purpose of this study is to develop a measurement instrument for PSM for cross-national

research and comparison On the basis of the sharpened construct of PSM (Kim and

Vandenabeele 2010) using data from an international survey conducted in 12 countries we

found that the four-factor structure of PSM is generalizable across country cases This study

provides the measurement index of PSM with the four dimensions of APS CPV COM and

SS The APS dimension measures the instrumental motives of PSM the CPV dimension

determines the value-based motives of PSM while the COM and SS dimensions gauge the

identification motives and self-sacrifice respectively When all four dimensions of PSM are

used in a survey cross-national equivalence of survey items can be assured The items for

measuring PSM in future research are suggested in Appendix 4

This study has several limitations First looking for samples which were comparable

to the largest possible extent we opted for surveys in local government The results found

and discussed in this study are based on PSM among local government employees The

results might be different when surveying central government employees or other specific

groups in the public sector Thus the potential impact of the sample bias needs to be

considered in understanding this study It also means that the survey items for measuring

PSM should be examined with different samples in various areas and at different government

levels

Second the test results showed that there is a high degree of correlation among the

dimension such as APS CPV and COM Future research should not only investigate whether

- 25 -

the items reflect the core content of the dimensions but also continue to develop more

appropriate items for better discriminant validity of each dimension Third the content of

APP dimension was assumed to be distinct from the CPV dimension that focuses on personal

disposition to pursue public interest and public values However the public interest sub-

dimension was separated from the public values sub-dimension in the CPV dimension and

combined with the APP dimension to form the APS dimension Further research should

analyze the CPV dimension and investigate the appropriateness of the APS dimension

In order to facilitate research internationally the PSM construct needs to be

improved conceptually and operationally Kim and Vandenabeele (2010) have dealt with the

conceptual composition and operational dimension of PSM as well as the measurement

model of PSM The next step is to develop an internationally robust measurement instrument

of PSM This study proposes that scholars use all four of the dimensions in measuring the

conceptual components of PSM such as self-sacrifice instrumental value-based and affective

motives because they collectively represent PSM It provides a set of appropriate items for

measuring PSM and confirms the dimensional structure of PSM The survey results suggest

that people in different cultures understand PSM in the same fashion We know that this study

is simply a step along the way to developing a more robust measure but we hope it

contributes to enhanced cross-national research and comparison along with the generation of

cumulative knowledge internationally

- 26 -

REFERENCES

Anderson James C and David D Gerbing 1988 Structural Equation Modeling in Practice

A Review and Recommended Two-Step Approach Psychological Bulletin 103

411-423

Arbuckle James L 2009 Amos 180 Userrsquos Guide Crawfordville FL Amos Development

Co

Beck Joslashrgensen Torben and Barry Bozeman 2007 Public Values An Inventory

Administration amp Society 39(3) 354-381

Beutel Ann M and Margaret Mooney Marini 1995 Gender and values American

Sociological Review 60 436 - 448

Bobko Philip Philip L Roth and Maury A Buster 2007 The Usefulness of Unit Weights in

Creating Composite Scores A Literature Review Application to Content Validity and

Meta-Analysis Organizational Research Methods 10(4) 689-709

Bollen Kenneth A 1989 Structural Equations with Latent Variables New York Wiley

Bollen K A and R Lennox 1991 Conventional Wisdom on Measurement A Structural

Equation Perspective Psychological Bulletin 110(2) 305-314

Byrne Barbara M 2010 Structural Equation Modeling with Amos Basic Concepts

Applications and Programming 2nd

ed New York Routledge

Camilleri Emanuel 2006 Towards Developing an Organizational Commitment ― Public

Service Motivation Model for Maltese Public Service Employees Public Policy

and Administration 21 63-83

Castaing Seacutebastien 2006 The Effects of Psychological Contract Fulfillment and Public

Service Motivation on Organizational Commitment in the French Civil Service

Public Policy and Administration 21 84-98

Cenfetelli Ronald T and Geneviegraveve Bassellier 2009 Interpretation of Formative

Measurement in Information Systems Research MIS Quarterly 33(4) 689-707

Charles Michael B W Martin de Jong and Neal Ryan 2011 Public Values in Western

- 27 -

Europe A Temporal Perspective American Review of Public Administration 41

75-91

Chin Wynne W 2010 How to Write Up and Report PLS Analysis In Handbook of Partical

Least Squares edited by V Esposito Vinzi et al 655-690 Berlin Springer-Verlag

Coursey David H and Sanjay K Pandey 2007a Public Service Motivation Measurement

Testing an Abridged Version of Perryrsquos Proposed Scale Administration amp Society

39 547-568

Coursey David H and Sanjay K Pandey 2007b Content Domain Measurement and

Validity of the Red Tape Concept American Review of Public Administration 37

342-361

Coursey David H James L Perry Jeffrey L Brudney and Laura Littlepage 2008

Psychometric Verification of Perryrsquos Public Service Motivation Instrument Results

for Volunteer Exemplars Review of Public Personnel Administration 28 79-90

Davis Paul and Karen West 2009 What Do Public Values Mean for Public Action Putting

Public Values in Their Plural Place American Review of Public Administration 39

602-618

DeHart-Davis Leisha Justin Marlowe and Sanjay K Pandey 2006 Gender Dimensions of

Public Service Motivation Public Administration Review 66 873-887

Diamantopoulos Adamantios 2006 The error term in formative measurement models

Interpretation and modeling implications Journal of Modeling in Management 1(1)

7-17

Diamantopoulos Adamantios and Judy A Siguaw 2006 Formative versus Reflective

Indicators in Organizational Measure Development A Comparison and Empirical

Illustration British Journal of Management 17 263-282

Diamantopoulos Adamantios and Heidi M Winklhofer 2001 Index Construction with

Formative Indicators An Alternative to Scale Development Journal of Marketing

Research 38 269-277

Edwards Jeffrey R 2010 The Fallacy of Formative Measurement Organizational Research

- 28 -

Methods Online First published on August 19 2010 as

doi1011771094428110378369

Edwards Jeffrey R and Richard P Bagozzi 2000 On the Nature and Direction of

Relationships between Constructs and Measures Psychological Methods 5 155-

174

Flora David B and Patrick J Curran 2004 An Empirical Evaluation of Alternative

Methods of Estimation for Confirmatory Factor Analysis with Ordinal Data

Psychological Methods 9 466-491

Fornell Claes and Fred L Bookstein 1982 Two structural equations models LISREL and

PLS applied to customer exit-voice theory Journal of Marketing Research 19 440-

452

Giauque David Adrian Ritz Freacutedeacuteric Varone Simon Anderfuhren-Biget and Christian

Waldner 2011 Putting Public Service Motivation into Context A Balance between

Universalism and Particularism Forthcoming in International Review of

Administrative Sciences 77(2)

Goumltz Oliver Kerstin Liehr-Gobbers and Manfred Krafft 2010 Evaluation of Structural

Equation Models Using the Partial Least Squares (PLS) Approach In V Esposito

Vinzi et al (eds) Handbook of Partial Least Squares 691-711 DOI 101007978-3-

540-32827-8_30 Springer-Verlag Berlin Heidelberg

Guinot Christiane Julie Latreille and Michel Tenenhaus 2001 PLS Path Modelling and

Multiple Table Analysis Application to the Cosmetic Habits of Women in Ile-de-

France Chemometrics and Intelligent Laboratory Systems 58 247-259

Hair Joseph F Jr William C Black Barry J Babin and Rolph E Anderson 2010

Multivariate Data Analysis 7th

ed Upper Saddle River NJ Pearson Education

Hardin Andrew M Jerry Cha-Jan Chang Mark A Fuller and Gholamreza Torkzadeh 2010

Formative Measurement and Academic Research In Search of Measurement

Theory Educational and Psychological Measurement Online First published on

June 11 2010 as doi101177001316440370208

- 29 -

Howell Roy D Einar Breivik and James B Wilcox 2007a Reconsidering Formative

Measurement Psychological Methods 12 205-218

Howell Roy D Einar Breivik and James B Wilcox 2007b Is Formative Measurement

Really Measurement Reply to Bollen (2007) and Bagozzi (2007) Psychological

Methods 12 238-245

Jarvis Cheryl Burke Scott B MacKenzie and Philip M Podsakoff 2003 A Critical Review

of Construct Indicators and Measurement Model Misspecification in Marketing and

Consumer Research Journal of Consumer Research 30 199-218

Joumlreskog Karl G and Dag Soumlrbom 2005 LISREL 872 (Computer Software) Lincolnwood

IL Scientific Software International

Kim Gimun Bongsik Shin and Varun Grover 2010 Investigating Two Contradictory Views

of Formative Measurement in Information Systems Research MIS Quarterly 34

345-365

Kim Sangmook 2009a Revising Perryrsquos Measurement Scale of Public Service Motivation

American Review of Public Administration 39 149-163

Kim Sangmook 2009b Testing the Structure of Public Service Motivation in Korea A

Research Note Journal of Public Administration Research and Theory 19 839-

851

Kim Sangmook 2010 Testing a Revised Measurement of Public Service Motivation

Reflective versus Formative Specification Journal of Public Administration

Research and Theory Doi101093jopartmuq048

Kim Sangmook and Wouter Vandenabeele 2010 A strategy for building public service

motivation research internationally Public Administration Review 70 701-709

Kline Rex B 2005 Principles and Practice of Structural Equation Modeling 2nd

ed New

York Guilford Press

Law Kenneth and Chi-Sum Wong 1999 Multidimensional Constructs in Structural

Equation Analysis An Illustration using the Job Perception and Job Satisfaction

Constructs Journal of Management 25 143-160

- 30 -

Leisink Peter and Bram Steijn 2009 Public Service Motivation and Job Performance of

Public Sector Employees in the Netherlands International Review of

Administrative Sciences 75 35-52

Liu Bangcheng Chun Hui Jin Hu Wensheng Yang and Xinli Yu 2011 How well can

Public Service Motivation Connect with Occupational Intention International

Review of Administrative Sciences 77 191-211

Liu Bangcheng Ningyu Tang and Xiaomei Zhu 2008 Public Service Motivation and Job

Satisfaction in China An Investigation of Generalisability and Instrumentality

International Journal of Manpower 29 684-699

MacCallum Robert C and Michael W Browne 1993 The Use of Causal Indicators in

Covariance Structure Models Some Practical Issues Psychological Bulletin 114

533-541

MacKenzie Scott B 2003 The dangers of poor construct conceptualization Journal of the

Academy of Marketing Science 31(3) 323-326

MacKenzie Scott B Philip M Podsakoff and Cheryl Burke Jarvis 2005 The Problem of

Measurement Model Misspecification in Behavioral and Organizational Research

and Some Recommended Solutions Journal of Applied Psychology 90 710-730

Martinsen Dorte Sindbjerg and Torben Beck Joslashrgensen 2010 Accountability as a

Differentiated Value in Supranational Governance American Review of Public

Administration 40 742-760

Moynihan Donald P and Sanjay K Pandey 2007a The Role of Organizations in Fostering

Public Service Motivation Public Administration Review 67 40-53

Moynihan Donald P and Sanjay K Pandey 2007b Finding Workable Levers over Work

Motivation Comparing Job Satisfaction Job Involvement and Organizational

Commitment Administration amp Society 39 803-832

Pandey Sanjay K and Edmund C Stazyk 2008 Antecedents and Correlates of Public

Service Motivation In Motivation in Public Management edited by James L Perry

and Annie Hondeghem 101-117 Oxford Oxford University Press

- 31 -

Perry James L 1996 Measuring Public Service Motivation An Assessment of Construct

Reliability and Validity Journal of Public Administration Research and Theory 6

5-22

Perry James L 1997 Antecedents of Public Service Motivation Journal of Public

Administration Research and Theory 7 181-197

Perry James L 2000 Bringing Society in Toward a Theory of Public Service Motivation

Journal of Public Administration Research and Theory 10 471-488

Perry James L and Annie Hondeghem 2008a Motivation in Public Management Oxford

Oxford University Press

Perry James L and Annie Hondeghem 2008b Directions for Future Theory and Research

In Motivation in Public Management edited by James L Perry and Annie

Hondeghem 294-313 Oxford Oxford University Press

Perry James L Annie Hondeghem and Lois Recascino Wise 2010 Revisiting the

Motivational Bases of Public Service Twenty Years of Research and an Agenda for

the Future Public Administration Review 70(5) 681-690

Perry James L and Wouter Vandenabeele 2008 Behavioral dynamics Institutions

Identities and Self-Regulation In Motivation in Public Management edited by

James L Perry and Annie Hondeghem 56-79 Oxford Oxford University Press

Perry James L and Lois Recascino Wise 1990 The Motivational Bases of Public Service

Public Administration Review 50 367-373

Rainey Hal G 1982 Reward Preference among Public and Private Managers In Search of

the Service Ethic American Review of Public Administration 16 288-302

Ringle Christian M Oliver Goumltz Martin Wetzels and Bradley Wilson 2009 On the Use of

Formative Measurement Specifications in Structural Equation Modeling A Monte

Carlo Simulation Study to Compare Covariance-Based and Partial Least Squares

Model Estimation Methodologies Faculty of Economics and Business Administration

Maastricht Univeristy

Ringle C M S Wende and S Will 2005 Smart PLS 20 (M3) Beta

- 32 -

httpwwwsmartplsde

Ritz Adrian 2011 Attraction to Public Policy Making A Qualitative Inquiry into

Improvements in PSM Measurement Forthcoming in Public Administration

Spector Paul E 1992 Summated Rating Scales Construction An Introduction Newbury

Park CA Sage

Steenbergen Marco R 1996 Compassion and American public opinion An analysis of the

NES Humanitarianism Scale Report to the Board of Overseers of the National

Election Studies

Steffen Patrick R and Kevin S Masters 2005 Does compassion mediate the intrinsic

religion-health relationship Annals of Behavioral Medicine 30 217-223

Taylor Jeannette 2007 The Impact of Public Service Motives on Work Outcomes in

Australia A Comparative Multi-Dimensional Analysis Public Administration 85

931-959

Vandenabeele Wouter 2007 Towards a Theory of Public Service Motivation An

Institutional Approach Public Management Review 9 545-556

Vandenabeele Wouter 2008a Government Calling Public Service Motivation as an Element

in Selecting Government as an Employer of Choice Public Administration 86

1089-1105

Vandenabeele Wouter 2008b Development of a Public Service Motivation Measurement

Scale Corroborating and Extending Perryrsquos Measurement Instrument International

Public Management Journal 11 143-167

Vandenabeele Wouter 2009 The Mediating Effect of Job Satisfaction and Organizational

Commitment on Self-Reported Performance More Robust Evidence of the PSM-

Performance Relationship International Review of Administrative Sciences 75 11-

34

Wilcox James B Roy D Howell and Einar Breivik 2008 Questions about Formative

Measurement Journal of Business Research 61 1219-1228

Wise Lois Recascino 2000 The Public Service Culture In Public Administration Concepts

- 33 -

and Cases 7th ed edited by Richard J Stillman II 342-353 Boston Houghton

Mifflin

Wright Bradley E 2008 Methodological Challenges associated with Public Service

Motivation Research In Motivation in Public Management edited by James L

Perry and Annie Hondeghem 80-98 Oxford Oxford University Press

Wright Bradley E and Sanjay K Pandey 2008 Public Service Motivation and the

Assumption of Person-Organization Fit Testing the Mediating Effect of Value

Congruence Administration amp Society 40 502-521

- 34 -

Table 1

Possible Indicators of PSM and Descriptive Statistics (n = 2868)

Dimensions and items Mean Standard

Deviation

Item-Total

Correlation

Attraction to Public Participation (APP)

APP1 I am interested in helping to improve public service

APP2 I am satisfied when I see people benefiting from the public programs I was

involved in

APP3 I like to discuss topics regarding public programs and policies with others

APP4 I believe that public sector activities contribute to our general welfare

APP5 I admire people who initiate or are involved in activities to aid my

community

APP6 Contributing to public programs and policies helps me realize myself

APP7 It is important to contribute to activities that tackle social problems

412

441

355

393

423

351

409

696

662

917

753

672

881

734

524

421

482

412

493

478

552

Commitment to Public Values (CPV)

Sub-dimension for Public Interests (CPV1)

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

CPI3 I would prefer seeing public officials do what is best for the whole

community even if it harmed my interests

CPI4 Serving the public interest is more important than helping a single individual

Sub-dimension for Public Values (CPV2)

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of public

services

CPV3 It is fundamental that public services respond to the needs of the citizens

CPV4 Decisions regarding public services should be democratic despite the time

and effort it takes

CPV5 Everybody is entitled to a good service even if it costs a lot of money

CPV6 It is fundamental that the interests of future generations are taken into

account when developing public policies

CPV7 To act ethically is essential for public servants

CPV8 I believe that public employees must always be aware of the legitimacy of

their activities

CPV9 I personally identify with the aim of protecting individual liberties and rights

418

409

348

337

428

416

433

394

367

426

441

434

414

695

662

899

995

730

704

710

852

932

689

681

692

763

525

596

480

251

484

437

356

325

376

439

462

402

392

Compassion (COM) COM1 It is difficult for me to contain my feelings when I see people in distress

COM2 I feel sympathetic to the plight of the underprivileged

COM3 I empathize with other people who face difficulties

COM4 I have little compassion for people in need who are unwilling to take the

first step to help themselves

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

354

387

388

299

412

391

951

825

707

1083

755

765

399

511

389

011

487

502

Self-Sacrifice (SS)

SS1 Making a difference to society means more to me than personal achievements

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS5 People should give back to society more than they get from it

SS6 Serving other citizens would give me a good feeling even if no one paid me for

it

SS7 I would agree to a good plan to make a better life for the poor even if it costs

me money

357

329

339

302

345

363

350

904

872

949

982

888

1049

895

508

584

454

531

412

368

546

- 35 -

Table 2

Background of Respondents (n = 2868)

Variables Characteristics Respondents

Sex

Male

Female

NA

1209

1622

37

422

566

13

Length of service

(years)

0 ~ 10

10 ~ 20

20 ~ 30

30+

NA

1126

723

484

491

44

393

252

169

171

15

Organizational status

―Do you supervise employees

No

Yes

NA

1472

1076

320

513

375

112

Country

Australia

Belgium

China

Denmark

France

Italy

Korea

Lithuania

The Netherlands

Switzerland

United Kingdom

USA

249

214

230

249

266

162

253

236

249

250

260

250

87

75

80

87

93

56

88

82

87

87

91

87 Note NA = no answer

- 36 -

Table 3

Standardized factor loadings and correlations for the 31-item PSM index (n = 2868)

Dimensions and items SFL (MLE) SFL (D-WLS) Alpha

APP

APP1

APP2

APP4

APP5

APP6

APP7

597

486

455

558

490

600

657

552

514

625

549

665

699

CPV1

CPI1

CPI2

CPI3

CPI4

588

660

488

289

653

732

535

294

542

CPV2

CPV1

CPV2

CPV3

CPV4

CPV5

CPV6

CPV7

CPV8

CPV9

611

546

436

393

418

534

598

496

460

690

631

505

457

486

601

669

573

530

745

COM

COM1

COM2

COM3

COM5

COM6

496

569

498

597

607

516

651

538

670

684

687

SS

SS1

SS2

SS3

SS4

SS5

SS6

SS7

567

736

558

728

467

409

663

660

762

591

695

515

494

709

782

Measures of fit (MLE) χsup2 (df = 424) = 43699 p lt 001 CFI = 835 GFI = 899

RMSEA = 057 [055 059]

Measure of fit (D-WLS) χsup2 (df = 424) = 39247 p lt 001 CFI = 971 GFI = 976

RMSEA = 054 [052 055]

Inter-factor correlations MLE (D-WLS)

1 2 3 4

1 APS

2 CPV1 996 (988)

3 CPV2 902 (890) 798 (798)

4 COM 787 (783) 758 (762) 779 (783)

5 SS 700 (742) 800 (859) 509 (537) 720 (729)

Note SFL = standardized factor loading All factor loadings and correlations are significant at p lt 001

- 37 -

Table 4

Standardized factor loadings and correlations for the 16-item PSM index (n = 2868)

Dimensions and items SFL

(MLE)

SFL

(D-WLS)

APS

APP5 I admire people who initiate or are involved in activities to aid my

community

APP7 It is important to contribute to activities that tackle social problems

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

611

592

637

666

665

669

690

751

CPV

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of

public services

CPV6 It is fundamental that the interests of future generations are taken

into account when developing public policies

CPV7 To act ethically is essential for public servants

612

546

551

590

698

631

618

658

COM

COM1 It is difficult for me to contain my feelings when I see people in

distress

COM2 I feel sympathetic to the plight of the underprivileged

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

490

546

596

604

515

608

669

679

SS

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS7 I would agree to a good plan to make a better life for the poor even if

it costs me money

763

571

766

678

807

629

747

755

Measures of fit (MLE)

χsup2 (df = 98) = 8490 p lt 001

CFI = 939 GFI = 965

RMSEA = 052 [049 055]

Measure of fit (D-WLS)

χsup2 (df = 98) = 6069 p lt 001

CFI = 988 GFI = 992

RMSEA = 043 [039 046]

Alpha CR (D-WLS) Inter-factor correlations MLE (D-WLS)

1 APS 717 788 1 APS 2 CPV 3 COM

2 CPV 663 747 848 (859)

3 COM 645 713 781 (791) 789 (795)

4 SS 784 826 565 (596) 462 (505) 715 (726)

Note SFL = standardized factor loading Alpha = Cronbachrsquos alpha CR = composite reliability

All factor loadings and correlations are significant at p lt 001

- 38 -

Table 5

Results of CFA in the individual countries Standardized factor loadings and measures of fit

Factors

amp Items

Australia Belgium China Denmark France Italy

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

672

756

799

785

804

943

710

703

597

790

799

730

584

675

685

681

731

742

578

748

674

643

765

739

645

704

678

708

812

779

632

604

765

766

672

853

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

552

593

560

699

727

632

624

605

560

713

709

645

702

602

616

724

686

735

567

560

715

654

605

811

608

658

587

696

755

698

673

620

523

782

697

654

COM

COM1

COM2

COM3

COM5

COM6

623

584

662

688

701

739

732

608

475

808

666

581

578

616

605

649

670

673

699

751

737

679

816

776

568

546

801

626

673

850

637

546

741

715

601

822

SS

SS2

SS3

SS4

SS7

654

794

806

892

767

813

727

693

799

793

725

838

857

820

773

896

855

842

806

749

792

829

753

884

847

731

753

889

816

753

815

692

664

795

782

734

Inter-factor

correlations

093

~

752

163

~

810

320

~

773

323

~

825

452

~

827

497

~

855

323

~

850

321

~

847

498

~

899

536

~

902

425

~

829

476

~

865

Sample size 249 214 230 249 266 162

Measures of Fit

χsup2 638 444 881 732 765 554 817 625 735 565 750 564

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 980 100 939 987 969 997 967 994 975 997 949 995

GFI 961 986 941 986 948 991 949 992 955 992 926 986

RMSEA 036 00 063 050 051 026 053 035 045 026 059 033

- 39 -

Table 5 (Continued)

Factors

amp Items

Korea Lithuania The Netherlands Switzerland UK USA

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

619

701

667

706

781

717

666

512

682

751

605

779

557

629

630

620

719

729

535

621

702

643

694

735

655

725

738

710

804

822

535

621

702

643

694

735

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

594

607

722

722

644

722

674

535

533

790

589

711

568

554

538

655

628

545

627

580

615

711

666

704

800

512

555

873

581

646

627

580

615

711

666

704

COM

COM1

COM2

COM3

COM5

COM6

694

677

550

763

736

627

557

639

428

550

696

534

737

623

508

766

616

608

659

592

683

703

661

769

682

583

729

739

644

803

659

592

683

703

661

769

SS

SS2

SS3

SS4

SS7

776

752

679

779

770

811

707

721

635

725

734

706

814

656

743

820

767

764

745

632

689

788

662

744

671

818

832

830

864

802

745

632

689

788

662

744

Inter-factor

correlations

432

~

726

466

~

746

376

~

788

406

~

902

377

~

852

455

~

836

637

~

818

660

~

828

317

~

852

360

~

852

637

~

818

660

~

828

Sample size 253 236 249 250 260 250

Measures of Fit

χsup2 896 573 612 462 773 616 815 715 769 542 815 715

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 951 996 975 100 958 993 957 991 972 998 957 991

GFI 942 988 960 989 951 987 948 990 955 990 948 990

RMSEA 059 028 034 00 050 034 053 044 048 022 053 044

Note All factor loadings and correlations are significant at p lt 001

- 40 -

Appendix 1 Dimensions and possible items of PSM (the first round)

1 Attraction to Public Participation

(1) I am interested in developing public programs helping my country or community (Kim 2009a)

(2) Sharing my views on public policies with others is attractive to me (Kim 2009a)

(3) I am satisfied when seeing people get benefits from the public programs I was involved in (Kim 2009a)

(4) I like to discuss political subjects with others (Giauque et al 2011)

(5) I think the governmental activities contribute to our welfare

(6) I like to initiate actions to help out my community

(7) I do not care much about politicians (Perry 1996)

2 Commitment to Public Values

Possible items for public interest

(1A) Meaningful public service is very important to me (Perry 1996)

(2A) It is important for me to contribute for the common good (Giauque et al 2011)

(3A) I recognize myself with the mission of protection of liberties and rights endorsed by the nation-state

(Giauque et al 2011)

(4A) I would prefer seeing public officials do what is best for the whole community even if it harmed my

interests (Perry 1996)

(5A) To me serving the public interest is more important than helping other people (Vandenabeele 2008b)

Possible items for public values

(1B) I recognize myself with the promotion of the equality of chances endorsed by the nation-state (Giauque et

al 2011)

(2B) It is fundamental that the public service provided to the citizen is regular and continuous (Giauque et al

2011)

(3B) It is fundamental that public service innovates to respond to the new needs of the citizensclients (Giauque

et al 2011)

(4B) The dignity and well-being of all should be the most important concerns in any society (Steenbergen

1996)

(5B) It is important that public servants account for all the costsexpenses they make (Vandenabeele 2008b)

(6B) We have to do everything in our power to pursue the goal of democracy

3 Compassion

(1) It is difficult for me to contain my feelings when I see people in distress (Perry 1996)

(2) I feel sympathetic to the plight of the underprivileged (Kim 2009a)

(3) I am often reminded by daily events how dependent we are on one another (Perry 1996)

(4) I feel compassion for others in the difficulties they are facing (Steffen and Masters 2005)

(5) It is not really my problem if others are in trouble and need help (Beutel and Marini 1995)

(6) I find it hard to be sympathetic toward starving people in foreign lands when there is so much trouble in our

own country (Beutel and Marini 1995)

(7) I get very upset when I see other people treated unfairly (Beutel and Marini 1995)

(8) I would agree to a good plan to make a better life for the poor even if it cost me money (Beutel and Marini

1995)

(9) To me patriotism includes seeing the welfare of others (Perry 1996)

(10) I care very much about other people

4 Self-Sacrifice

(1) Making a difference in society means more to me than personal achievements (Perry 1996)

(2) I am prepared to make enormous sacrifices for the good of society (Perry 1996)

(3) I believe in putting duty before self (Perry 1996)

(4) I am one of those rare people who would risk personal loss to help someone else (Perry 1996)

(5) I think people should give back to society more than they get from it (Perry 1996)

(6) Serving other citizens would give me a good feeling even if no one paid me for it (Perry 1996)

(7) Much of what I do is for a cause bigger than myself (Perry 1996)

- 41 -

Appendix 2 Newly suggested items through the first round

1 Items for the Attraction to Public Participation Dimension

(1) In the public domain every person should be counted as equal

(2) I enjoy to think and argue in a political way

(3) I am satisfied when seeing my ideas being integrated in public policies

(4) I can easily relate to the way of thinking and arguing of politicians

(5) Working in the public sector I can better realize myself

2 Items for the Public Values Sub-Dimension

(1) Public service should give value for money

(2) Cost considerations should play an important role in decisions about public service provision

(3) It is fundamental that the interests of future generations are taken into account when developing public

policies

(4) To act with correct ethics is for public servants as important as their professional competence

(5) Primarily public servants should be accountable to the public and not to their superiors

3 Items for the Self-Sacrifice Dimension

(1) I would agree to a good plan to make a better life for the poor even if it costs me money

- 42 -

Appendix 3 An example of a summary of comments and suggestions (the second round)

On the item ―I am interested in developing public programs helping my country or community

A It is a good sentence but is it possible for everyone to ―develop public programs Even public officials are

limited to do this

B OK

C or the special group

D Double barrel and difficult to translate to my country ―I am interested in helping to improve public services

E Good but I feel that wording requires revision If we draw sample from local government then country is

redundant ―I am interested in developing public policy programs that aim in helping my community

F 4 (General appreciation from1 to 4) Not clear whether it refers to ones work or to ones concern as a citizen

―(In my work) I am interested in contributing to the development of public etchellip

G Instead of development I would use implementing

H Delete public which is redundant because of the characteristic helping my country or community but add

or implementing ―I am interested in developing or implementing programs helping my country or

community

I Referring to dimension Wording Similar to item 6

Note Names have been changed to alphabets in order to ensure the anonymity of participants

- 43 -

Appendix 4 Dimensions and Items of Public Service Motivation for Future Research

Dimensions and items

Attraction to Public Service (APS)

1 I am interested in helping to improve public service

2 I admire people who initiate or are involved in activities to aid my community

3 It is important to contribute to activities that tackle social problems

4 Meaningful public service is very important to me

5 It is important for me to contribute to the common good

Commitment to Public Values (CPV)

1 I think equal opportunities for citizens are very important

2 It is important that citizens can rely on the continuous provision of public services

3 It is fundamental that the interests of future generations are taken into account when

developing public policies

4 To act ethically is essential for public servants

5 I believe that public employees must always be aware of the legitimacy of their activities

6 I personally identify with the aim of protecting individual liberties and rights

Compassion (COM)

1 It is difficult for me to contain my feelings when I see people in distress

2 I feel sympathetic to the plight of the underprivileged

3 I empathize with other people who face difficulties

4 I get very upset when I see other people being treated unfairly

5 Considering the welfare of others is very important

Self-Sacrifice (SS)

1 I am prepared to make sacrifices for the good of society

2 I believe in putting civic duty before self

3 I am willing to risk personal loss to help society

4 I would agree to a good plan to make a better life for the poor even if it costs me money

- 18 -

disposition to serve the public interest through public participation When developing the

items some assumed that the items of committing to public interests can be separated from

those of pursuing public values in the CPV dimension The empirical test showed that

respondents tended to understand committing to public interests as behavior-oriented and so

the items for public interest are not differentiated from participation-oriented items in their

perceptions but are different from those for public values The respondents seem to

understand that pursuing public values is more related to personal beliefs or creeds Thus the

CPV dimension is solely focused on public values and the items of pursing public interests

are combined with those of the APP dimension to make the APS dimension

Public values and their relative importance are neither impervious to change over

time nor the same among all countries (Charles de Jong and Ryan 2011 Davis and West

2009) Moreover some values may be in conflict as Martinsen and Beck Joslashrgensen (2010)

discussed in regard to the tensions between accountability and efficiency It is therefore not

possible to include all public values in one dimension (Beck Joslashrgensen and Bozeman 2007)

Accordingly the items of the CPV dimension are developed for estimating individual

commitment to seek public values which there are general consensus about and which are

generally formulated

When all four dimensions of PSM are included in a survey cross-national

equivalence of the survey can be assured The items represent reflections or manifestations of

each dimension and hence are relatively homogeneous and interchangeable Adding or

deleting items does not change the essential nature of the dimension Thus any combination

of items in a dimension can be regarded as equivalent When using structural equation

modeling (SEM) each dimension needs to have at least three reflective indicators in order to

achieve identification or it needs to have at least two reflective indicators and be allowed to

- 19 -

covary with another dimension that has at least two reflective measures With one indicator

the dimension is identified if the loading and the variance of the uniqueness are both fixed

(Bollen 1989 Edwards 2010) Thus the number of items in each dimension is not restricted

when measuring PSM However we recommend using multiple items for each dimension

because more research is needed to investigate the appropriateness of this PSM instrument

with varying samples

The test results showed that model fit estimators were clearly better for the D-WLS

It means that in this study MLE was more rigorous than D-WLS in assessing the fit of the

model The empirical evidence indicated that MLE may be an acceptable technique for

ordinal data when there is a large sample a five-point Likert scale and the skewness and

kurtosis of variables is within acceptable limits as suggested by Byrne (2010)

Discussion

The analysis reaffirmed the four-dimensional structure of PSM but the results were not

entirely consistent with the original CFA we tested We were ultimately able to converge on a

good-fit operational model for PSM that was also consistent with underlying theory The

analysis is not able to resolve the debate about measurement models however for reasons

discussed below

Choosing the correct measurement model (ie reflective vs formative) is important

as measurement model misspecification results not only in misrepresentation of the epistemic

relationships between constructs and their measures but also in biased parameter estimates of

structural relationships (Jarvis et al 2003 Law and Wong 1999 MacKenzie 2003

MacKenzie Podsakoff and Jarvis 2005) A formative measurement model in isolation is

- 20 -

under-identified and cannot be estimated in the case of covariance-based SEM (CB-SEM)

(Bollen 1989) A construct with formative indicators must emit paths to at least two unrelated

reflective indicators two unrelated latent constructs or some combination of the two in order

to solve the problem of under-identification (Jarvis MacKenzie and Podsakoff 2003

MacCallum and Browne 1993 MacKenzie Podsakoff and Jarvis 2005) Thus ―when

dependent constructs change the empirical nature of the formatively measured construct

changes (Wilcox Howell and Breivik 2008 1223) and ―when specifying formative

measurement models what might seem like an innocuous choice of outcomes to achieve

identification can have fundamental ramifications for the meaning of the formative construct

and the inferences based on it (Edwards 2010 7)

Earlier research found that the association between a construct and its measures

should remain stable regardless of the larger causal model in which the constructs and

measures are embedded (Edwards and Bagozzi 2000) The association between a construct

and its measures is generally stable for reflective measures but associations of formative

measures with their construct are unstable The susceptibility of formative measurement to

both interpretational confounding and external inconsistency recommends that formative

measurement is not an equally attractive alternative to reflective measurement and that

whenever possible researchers should opt for reflective measurement (Howell Breivik and

Wilcox 2007a 2007b Kim Shin and Grover 2010) In the case of second-order models

researchers are limited to only a second-order reflective model in CB-SEM (Chin 2010)

Regardless of the measurement debate and the tradeoffs between reflective or

formative specification this study makes a theoretical argument that PSM is a second-order

formative construct This second-order formative factor model (reflective first-order

formative second-order) is also suggested as an alternative model that captures the

- 21 -

multidimensionality of complex concepts and avoids the shortcomings of first-order

formative measurement (Edwards 2010) However this model itself cannot achieve

identification One possible strategy is to develop direct reflective measures of PSM based on

the view that PSM is a general concept that is not only influenced by its specific dimensions

but also can be assessed using general measures (Edwards 2010 Javis MacKenzie and

Podsakoff 2003)6 However these general measures have not yet been developed for the

PSM construct

Another way around the problem of underidentification of formative models is to

apply the component-based approach known as partial least squares path modeling (PLS-PM)

which can model formative indicators (Fornell and Bookstein 1982 Goumltz Liehr-Gobbers

and Krafft 2010) The CB-SEM and the PLS-PM methods have distinctive statistical

characteristics and selecting an approach to SEM depends on the particular research situation

CB-SEM is the method of choice for theory testing while PLS-PM is appropriate for

prediction-oriented applications (Ringle Goumltz Wetzels and Wilson 2009) One advantage of

the PLS method is that it demands significantly fewer requirements compared to that of

covariance structure analyses but nevertheless delivers consistent estimation results Another

asset of the PLS approach is its ability to deal with formative as well as reflective indicators

even within one structural equation model This indicates that the PLS approach can be

appropriate for explorative analysis of structural equation models (Chin 2010 Goumltz Liehr-

Gobbers and Krafft 2010) A Monte Carlo simulation study to compare CB-SEM and PLS-

PM showed that PLS-PM provides a viable approximation of model parameters when the

prerequisites for CB-SEM are not met (Ringle Goumltz Wetzels and Wilson 2009) A formative

6 For example job satisfaction can be assessed with general reflective measures such as ―in general I am

satisfied with my job and the formative dimensions can be assessed with measures that describe satisfaction

with each job facet (Edwards 2010 15)

- 22 -

composite in the form of indexes needs to be tested to know the relative weights (γ-

parameters) of the dimensions of PSM In PLS-PM a second-order formative factor is

directly measured by observed variables for all the first order factors While this approach

repeats the number of manifest variables used the model can be identified without adding

outcome constructs that would be considered inappropriate or irrelevant from a conceptual

standpoint (Chin 2010 Edwards 2010) The test result of PLS-PM using Smart PLS 20

(Ringle Wende and Will 2005) produced the function PSM = 31APS + 36CPV + 28COM

+ 31SS However one consideration is whether and how construct error is modeled for the

formative construct (Diamantopoulos 2006) When using PLS-PM the researcher is making

the assumption that there is no error at the construct level as all constructs in PLS-PM are

modeled without error When using PLS-PM the absence of an error term will tend to

increase the weights therefore inflating their importance (Bollen and Lennox 1991) The

primary statistic for assessing a formative indicator is its weight the partialized effect of the

indicator on its intended construct controlling for the effects of all other indicators of that

construct (Cenfetelli and Bassellier 2009) A formative measure is essentially a multiple

regression with the construct representing the dependent variable and the indicators as the

predictors (Diamantopoulos and Winklhofer 2001) PSM may be regarded as a weighted

linear composite of its dimensions The weights used to create the composites are different

across CE-SEM and PLS-PM The causal dimensional weights generated in CB-SEM are

estimated in a model in which a construct-level error term is specified (eg PSM = γ1APM +

γ2CPI + γ3COM + γ4SS + ζ) whereas the weights in PLS-PM are not (eg PSM = γ1APM +

γ2CPI + γ3COM + γ4SS) This leads to two different conceptualizations of the PSM construct

one in which the causal indicators do not capture the construct in its entirety and another in

which they do (Hardin et al 2010) These differences need to be discussed in future research

- 23 -

in this area

One practical suggestion to measure PSM is to apply unit weights to its dimensions if

we assume the effects of the four dimensions are equivalent A literature review indicates that

unit weights have substantial predictive validity when compared with regression weights The

literature shows that the empirical correlations between scores generated from differential and

unit-weighting approaches are quite high In creating composite scores unit weights can be

an appropriate approach for weighting under many circumstances (Bobko Roth and Buster

2007) Although the uses of unit-weight composites will likely result in the explanation of

less variance than empirical estimations of dimensional weights they will remain consistently

defined across studies and hence suitable for theory testing Alternatively more thorough

discussions on the dimensions of PSM may suggest fixed-weight composites (Hardin et al

2010) A previous study (Kim 2010 22) also suggested ―Overall indexing PSM by

measuring all the dimensions of PSM and simply summing the averages of each dimension

can make it easier to measure PSM in practice as PSM = APS + CPV + COM + SS

An advantage of regarding PSM as a second-order formative construct is that the

dimensions can be analyzed both individually and collectively This is important as the

dimensions are expected to have different antecedents and consequences For instance Perry

(1997) found some significant differences in the influence of independent variables on the

four dimensions of PSM Taylor (2007) confirmed that certain dimensions are more important

than others in influencing work outcomes Moynihan and Pandey (2007a) reported that red

tape is negatively related to one dimension but reform orientation and hierarchical authority

is positively related to another dimension A recent study found that the SS dimension has a

positive relationship with occupational intention but the other dimensions do not (Liu et al

2011) Thus it is valuable to analyze the causes and effects of individual dimensions of PSM

- 24 -

as well as of the PSM construct

Conclusion

The purpose of this study is to develop a measurement instrument for PSM for cross-national

research and comparison On the basis of the sharpened construct of PSM (Kim and

Vandenabeele 2010) using data from an international survey conducted in 12 countries we

found that the four-factor structure of PSM is generalizable across country cases This study

provides the measurement index of PSM with the four dimensions of APS CPV COM and

SS The APS dimension measures the instrumental motives of PSM the CPV dimension

determines the value-based motives of PSM while the COM and SS dimensions gauge the

identification motives and self-sacrifice respectively When all four dimensions of PSM are

used in a survey cross-national equivalence of survey items can be assured The items for

measuring PSM in future research are suggested in Appendix 4

This study has several limitations First looking for samples which were comparable

to the largest possible extent we opted for surveys in local government The results found

and discussed in this study are based on PSM among local government employees The

results might be different when surveying central government employees or other specific

groups in the public sector Thus the potential impact of the sample bias needs to be

considered in understanding this study It also means that the survey items for measuring

PSM should be examined with different samples in various areas and at different government

levels

Second the test results showed that there is a high degree of correlation among the

dimension such as APS CPV and COM Future research should not only investigate whether

- 25 -

the items reflect the core content of the dimensions but also continue to develop more

appropriate items for better discriminant validity of each dimension Third the content of

APP dimension was assumed to be distinct from the CPV dimension that focuses on personal

disposition to pursue public interest and public values However the public interest sub-

dimension was separated from the public values sub-dimension in the CPV dimension and

combined with the APP dimension to form the APS dimension Further research should

analyze the CPV dimension and investigate the appropriateness of the APS dimension

In order to facilitate research internationally the PSM construct needs to be

improved conceptually and operationally Kim and Vandenabeele (2010) have dealt with the

conceptual composition and operational dimension of PSM as well as the measurement

model of PSM The next step is to develop an internationally robust measurement instrument

of PSM This study proposes that scholars use all four of the dimensions in measuring the

conceptual components of PSM such as self-sacrifice instrumental value-based and affective

motives because they collectively represent PSM It provides a set of appropriate items for

measuring PSM and confirms the dimensional structure of PSM The survey results suggest

that people in different cultures understand PSM in the same fashion We know that this study

is simply a step along the way to developing a more robust measure but we hope it

contributes to enhanced cross-national research and comparison along with the generation of

cumulative knowledge internationally

- 26 -

REFERENCES

Anderson James C and David D Gerbing 1988 Structural Equation Modeling in Practice

A Review and Recommended Two-Step Approach Psychological Bulletin 103

411-423

Arbuckle James L 2009 Amos 180 Userrsquos Guide Crawfordville FL Amos Development

Co

Beck Joslashrgensen Torben and Barry Bozeman 2007 Public Values An Inventory

Administration amp Society 39(3) 354-381

Beutel Ann M and Margaret Mooney Marini 1995 Gender and values American

Sociological Review 60 436 - 448

Bobko Philip Philip L Roth and Maury A Buster 2007 The Usefulness of Unit Weights in

Creating Composite Scores A Literature Review Application to Content Validity and

Meta-Analysis Organizational Research Methods 10(4) 689-709

Bollen Kenneth A 1989 Structural Equations with Latent Variables New York Wiley

Bollen K A and R Lennox 1991 Conventional Wisdom on Measurement A Structural

Equation Perspective Psychological Bulletin 110(2) 305-314

Byrne Barbara M 2010 Structural Equation Modeling with Amos Basic Concepts

Applications and Programming 2nd

ed New York Routledge

Camilleri Emanuel 2006 Towards Developing an Organizational Commitment ― Public

Service Motivation Model for Maltese Public Service Employees Public Policy

and Administration 21 63-83

Castaing Seacutebastien 2006 The Effects of Psychological Contract Fulfillment and Public

Service Motivation on Organizational Commitment in the French Civil Service

Public Policy and Administration 21 84-98

Cenfetelli Ronald T and Geneviegraveve Bassellier 2009 Interpretation of Formative

Measurement in Information Systems Research MIS Quarterly 33(4) 689-707

Charles Michael B W Martin de Jong and Neal Ryan 2011 Public Values in Western

- 27 -

Europe A Temporal Perspective American Review of Public Administration 41

75-91

Chin Wynne W 2010 How to Write Up and Report PLS Analysis In Handbook of Partical

Least Squares edited by V Esposito Vinzi et al 655-690 Berlin Springer-Verlag

Coursey David H and Sanjay K Pandey 2007a Public Service Motivation Measurement

Testing an Abridged Version of Perryrsquos Proposed Scale Administration amp Society

39 547-568

Coursey David H and Sanjay K Pandey 2007b Content Domain Measurement and

Validity of the Red Tape Concept American Review of Public Administration 37

342-361

Coursey David H James L Perry Jeffrey L Brudney and Laura Littlepage 2008

Psychometric Verification of Perryrsquos Public Service Motivation Instrument Results

for Volunteer Exemplars Review of Public Personnel Administration 28 79-90

Davis Paul and Karen West 2009 What Do Public Values Mean for Public Action Putting

Public Values in Their Plural Place American Review of Public Administration 39

602-618

DeHart-Davis Leisha Justin Marlowe and Sanjay K Pandey 2006 Gender Dimensions of

Public Service Motivation Public Administration Review 66 873-887

Diamantopoulos Adamantios 2006 The error term in formative measurement models

Interpretation and modeling implications Journal of Modeling in Management 1(1)

7-17

Diamantopoulos Adamantios and Judy A Siguaw 2006 Formative versus Reflective

Indicators in Organizational Measure Development A Comparison and Empirical

Illustration British Journal of Management 17 263-282

Diamantopoulos Adamantios and Heidi M Winklhofer 2001 Index Construction with

Formative Indicators An Alternative to Scale Development Journal of Marketing

Research 38 269-277

Edwards Jeffrey R 2010 The Fallacy of Formative Measurement Organizational Research

- 28 -

Methods Online First published on August 19 2010 as

doi1011771094428110378369

Edwards Jeffrey R and Richard P Bagozzi 2000 On the Nature and Direction of

Relationships between Constructs and Measures Psychological Methods 5 155-

174

Flora David B and Patrick J Curran 2004 An Empirical Evaluation of Alternative

Methods of Estimation for Confirmatory Factor Analysis with Ordinal Data

Psychological Methods 9 466-491

Fornell Claes and Fred L Bookstein 1982 Two structural equations models LISREL and

PLS applied to customer exit-voice theory Journal of Marketing Research 19 440-

452

Giauque David Adrian Ritz Freacutedeacuteric Varone Simon Anderfuhren-Biget and Christian

Waldner 2011 Putting Public Service Motivation into Context A Balance between

Universalism and Particularism Forthcoming in International Review of

Administrative Sciences 77(2)

Goumltz Oliver Kerstin Liehr-Gobbers and Manfred Krafft 2010 Evaluation of Structural

Equation Models Using the Partial Least Squares (PLS) Approach In V Esposito

Vinzi et al (eds) Handbook of Partial Least Squares 691-711 DOI 101007978-3-

540-32827-8_30 Springer-Verlag Berlin Heidelberg

Guinot Christiane Julie Latreille and Michel Tenenhaus 2001 PLS Path Modelling and

Multiple Table Analysis Application to the Cosmetic Habits of Women in Ile-de-

France Chemometrics and Intelligent Laboratory Systems 58 247-259

Hair Joseph F Jr William C Black Barry J Babin and Rolph E Anderson 2010

Multivariate Data Analysis 7th

ed Upper Saddle River NJ Pearson Education

Hardin Andrew M Jerry Cha-Jan Chang Mark A Fuller and Gholamreza Torkzadeh 2010

Formative Measurement and Academic Research In Search of Measurement

Theory Educational and Psychological Measurement Online First published on

June 11 2010 as doi101177001316440370208

- 29 -

Howell Roy D Einar Breivik and James B Wilcox 2007a Reconsidering Formative

Measurement Psychological Methods 12 205-218

Howell Roy D Einar Breivik and James B Wilcox 2007b Is Formative Measurement

Really Measurement Reply to Bollen (2007) and Bagozzi (2007) Psychological

Methods 12 238-245

Jarvis Cheryl Burke Scott B MacKenzie and Philip M Podsakoff 2003 A Critical Review

of Construct Indicators and Measurement Model Misspecification in Marketing and

Consumer Research Journal of Consumer Research 30 199-218

Joumlreskog Karl G and Dag Soumlrbom 2005 LISREL 872 (Computer Software) Lincolnwood

IL Scientific Software International

Kim Gimun Bongsik Shin and Varun Grover 2010 Investigating Two Contradictory Views

of Formative Measurement in Information Systems Research MIS Quarterly 34

345-365

Kim Sangmook 2009a Revising Perryrsquos Measurement Scale of Public Service Motivation

American Review of Public Administration 39 149-163

Kim Sangmook 2009b Testing the Structure of Public Service Motivation in Korea A

Research Note Journal of Public Administration Research and Theory 19 839-

851

Kim Sangmook 2010 Testing a Revised Measurement of Public Service Motivation

Reflective versus Formative Specification Journal of Public Administration

Research and Theory Doi101093jopartmuq048

Kim Sangmook and Wouter Vandenabeele 2010 A strategy for building public service

motivation research internationally Public Administration Review 70 701-709

Kline Rex B 2005 Principles and Practice of Structural Equation Modeling 2nd

ed New

York Guilford Press

Law Kenneth and Chi-Sum Wong 1999 Multidimensional Constructs in Structural

Equation Analysis An Illustration using the Job Perception and Job Satisfaction

Constructs Journal of Management 25 143-160

- 30 -

Leisink Peter and Bram Steijn 2009 Public Service Motivation and Job Performance of

Public Sector Employees in the Netherlands International Review of

Administrative Sciences 75 35-52

Liu Bangcheng Chun Hui Jin Hu Wensheng Yang and Xinli Yu 2011 How well can

Public Service Motivation Connect with Occupational Intention International

Review of Administrative Sciences 77 191-211

Liu Bangcheng Ningyu Tang and Xiaomei Zhu 2008 Public Service Motivation and Job

Satisfaction in China An Investigation of Generalisability and Instrumentality

International Journal of Manpower 29 684-699

MacCallum Robert C and Michael W Browne 1993 The Use of Causal Indicators in

Covariance Structure Models Some Practical Issues Psychological Bulletin 114

533-541

MacKenzie Scott B 2003 The dangers of poor construct conceptualization Journal of the

Academy of Marketing Science 31(3) 323-326

MacKenzie Scott B Philip M Podsakoff and Cheryl Burke Jarvis 2005 The Problem of

Measurement Model Misspecification in Behavioral and Organizational Research

and Some Recommended Solutions Journal of Applied Psychology 90 710-730

Martinsen Dorte Sindbjerg and Torben Beck Joslashrgensen 2010 Accountability as a

Differentiated Value in Supranational Governance American Review of Public

Administration 40 742-760

Moynihan Donald P and Sanjay K Pandey 2007a The Role of Organizations in Fostering

Public Service Motivation Public Administration Review 67 40-53

Moynihan Donald P and Sanjay K Pandey 2007b Finding Workable Levers over Work

Motivation Comparing Job Satisfaction Job Involvement and Organizational

Commitment Administration amp Society 39 803-832

Pandey Sanjay K and Edmund C Stazyk 2008 Antecedents and Correlates of Public

Service Motivation In Motivation in Public Management edited by James L Perry

and Annie Hondeghem 101-117 Oxford Oxford University Press

- 31 -

Perry James L 1996 Measuring Public Service Motivation An Assessment of Construct

Reliability and Validity Journal of Public Administration Research and Theory 6

5-22

Perry James L 1997 Antecedents of Public Service Motivation Journal of Public

Administration Research and Theory 7 181-197

Perry James L 2000 Bringing Society in Toward a Theory of Public Service Motivation

Journal of Public Administration Research and Theory 10 471-488

Perry James L and Annie Hondeghem 2008a Motivation in Public Management Oxford

Oxford University Press

Perry James L and Annie Hondeghem 2008b Directions for Future Theory and Research

In Motivation in Public Management edited by James L Perry and Annie

Hondeghem 294-313 Oxford Oxford University Press

Perry James L Annie Hondeghem and Lois Recascino Wise 2010 Revisiting the

Motivational Bases of Public Service Twenty Years of Research and an Agenda for

the Future Public Administration Review 70(5) 681-690

Perry James L and Wouter Vandenabeele 2008 Behavioral dynamics Institutions

Identities and Self-Regulation In Motivation in Public Management edited by

James L Perry and Annie Hondeghem 56-79 Oxford Oxford University Press

Perry James L and Lois Recascino Wise 1990 The Motivational Bases of Public Service

Public Administration Review 50 367-373

Rainey Hal G 1982 Reward Preference among Public and Private Managers In Search of

the Service Ethic American Review of Public Administration 16 288-302

Ringle Christian M Oliver Goumltz Martin Wetzels and Bradley Wilson 2009 On the Use of

Formative Measurement Specifications in Structural Equation Modeling A Monte

Carlo Simulation Study to Compare Covariance-Based and Partial Least Squares

Model Estimation Methodologies Faculty of Economics and Business Administration

Maastricht Univeristy

Ringle C M S Wende and S Will 2005 Smart PLS 20 (M3) Beta

- 32 -

httpwwwsmartplsde

Ritz Adrian 2011 Attraction to Public Policy Making A Qualitative Inquiry into

Improvements in PSM Measurement Forthcoming in Public Administration

Spector Paul E 1992 Summated Rating Scales Construction An Introduction Newbury

Park CA Sage

Steenbergen Marco R 1996 Compassion and American public opinion An analysis of the

NES Humanitarianism Scale Report to the Board of Overseers of the National

Election Studies

Steffen Patrick R and Kevin S Masters 2005 Does compassion mediate the intrinsic

religion-health relationship Annals of Behavioral Medicine 30 217-223

Taylor Jeannette 2007 The Impact of Public Service Motives on Work Outcomes in

Australia A Comparative Multi-Dimensional Analysis Public Administration 85

931-959

Vandenabeele Wouter 2007 Towards a Theory of Public Service Motivation An

Institutional Approach Public Management Review 9 545-556

Vandenabeele Wouter 2008a Government Calling Public Service Motivation as an Element

in Selecting Government as an Employer of Choice Public Administration 86

1089-1105

Vandenabeele Wouter 2008b Development of a Public Service Motivation Measurement

Scale Corroborating and Extending Perryrsquos Measurement Instrument International

Public Management Journal 11 143-167

Vandenabeele Wouter 2009 The Mediating Effect of Job Satisfaction and Organizational

Commitment on Self-Reported Performance More Robust Evidence of the PSM-

Performance Relationship International Review of Administrative Sciences 75 11-

34

Wilcox James B Roy D Howell and Einar Breivik 2008 Questions about Formative

Measurement Journal of Business Research 61 1219-1228

Wise Lois Recascino 2000 The Public Service Culture In Public Administration Concepts

- 33 -

and Cases 7th ed edited by Richard J Stillman II 342-353 Boston Houghton

Mifflin

Wright Bradley E 2008 Methodological Challenges associated with Public Service

Motivation Research In Motivation in Public Management edited by James L

Perry and Annie Hondeghem 80-98 Oxford Oxford University Press

Wright Bradley E and Sanjay K Pandey 2008 Public Service Motivation and the

Assumption of Person-Organization Fit Testing the Mediating Effect of Value

Congruence Administration amp Society 40 502-521

- 34 -

Table 1

Possible Indicators of PSM and Descriptive Statistics (n = 2868)

Dimensions and items Mean Standard

Deviation

Item-Total

Correlation

Attraction to Public Participation (APP)

APP1 I am interested in helping to improve public service

APP2 I am satisfied when I see people benefiting from the public programs I was

involved in

APP3 I like to discuss topics regarding public programs and policies with others

APP4 I believe that public sector activities contribute to our general welfare

APP5 I admire people who initiate or are involved in activities to aid my

community

APP6 Contributing to public programs and policies helps me realize myself

APP7 It is important to contribute to activities that tackle social problems

412

441

355

393

423

351

409

696

662

917

753

672

881

734

524

421

482

412

493

478

552

Commitment to Public Values (CPV)

Sub-dimension for Public Interests (CPV1)

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

CPI3 I would prefer seeing public officials do what is best for the whole

community even if it harmed my interests

CPI4 Serving the public interest is more important than helping a single individual

Sub-dimension for Public Values (CPV2)

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of public

services

CPV3 It is fundamental that public services respond to the needs of the citizens

CPV4 Decisions regarding public services should be democratic despite the time

and effort it takes

CPV5 Everybody is entitled to a good service even if it costs a lot of money

CPV6 It is fundamental that the interests of future generations are taken into

account when developing public policies

CPV7 To act ethically is essential for public servants

CPV8 I believe that public employees must always be aware of the legitimacy of

their activities

CPV9 I personally identify with the aim of protecting individual liberties and rights

418

409

348

337

428

416

433

394

367

426

441

434

414

695

662

899

995

730

704

710

852

932

689

681

692

763

525

596

480

251

484

437

356

325

376

439

462

402

392

Compassion (COM) COM1 It is difficult for me to contain my feelings when I see people in distress

COM2 I feel sympathetic to the plight of the underprivileged

COM3 I empathize with other people who face difficulties

COM4 I have little compassion for people in need who are unwilling to take the

first step to help themselves

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

354

387

388

299

412

391

951

825

707

1083

755

765

399

511

389

011

487

502

Self-Sacrifice (SS)

SS1 Making a difference to society means more to me than personal achievements

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS5 People should give back to society more than they get from it

SS6 Serving other citizens would give me a good feeling even if no one paid me for

it

SS7 I would agree to a good plan to make a better life for the poor even if it costs

me money

357

329

339

302

345

363

350

904

872

949

982

888

1049

895

508

584

454

531

412

368

546

- 35 -

Table 2

Background of Respondents (n = 2868)

Variables Characteristics Respondents

Sex

Male

Female

NA

1209

1622

37

422

566

13

Length of service

(years)

0 ~ 10

10 ~ 20

20 ~ 30

30+

NA

1126

723

484

491

44

393

252

169

171

15

Organizational status

―Do you supervise employees

No

Yes

NA

1472

1076

320

513

375

112

Country

Australia

Belgium

China

Denmark

France

Italy

Korea

Lithuania

The Netherlands

Switzerland

United Kingdom

USA

249

214

230

249

266

162

253

236

249

250

260

250

87

75

80

87

93

56

88

82

87

87

91

87 Note NA = no answer

- 36 -

Table 3

Standardized factor loadings and correlations for the 31-item PSM index (n = 2868)

Dimensions and items SFL (MLE) SFL (D-WLS) Alpha

APP

APP1

APP2

APP4

APP5

APP6

APP7

597

486

455

558

490

600

657

552

514

625

549

665

699

CPV1

CPI1

CPI2

CPI3

CPI4

588

660

488

289

653

732

535

294

542

CPV2

CPV1

CPV2

CPV3

CPV4

CPV5

CPV6

CPV7

CPV8

CPV9

611

546

436

393

418

534

598

496

460

690

631

505

457

486

601

669

573

530

745

COM

COM1

COM2

COM3

COM5

COM6

496

569

498

597

607

516

651

538

670

684

687

SS

SS1

SS2

SS3

SS4

SS5

SS6

SS7

567

736

558

728

467

409

663

660

762

591

695

515

494

709

782

Measures of fit (MLE) χsup2 (df = 424) = 43699 p lt 001 CFI = 835 GFI = 899

RMSEA = 057 [055 059]

Measure of fit (D-WLS) χsup2 (df = 424) = 39247 p lt 001 CFI = 971 GFI = 976

RMSEA = 054 [052 055]

Inter-factor correlations MLE (D-WLS)

1 2 3 4

1 APS

2 CPV1 996 (988)

3 CPV2 902 (890) 798 (798)

4 COM 787 (783) 758 (762) 779 (783)

5 SS 700 (742) 800 (859) 509 (537) 720 (729)

Note SFL = standardized factor loading All factor loadings and correlations are significant at p lt 001

- 37 -

Table 4

Standardized factor loadings and correlations for the 16-item PSM index (n = 2868)

Dimensions and items SFL

(MLE)

SFL

(D-WLS)

APS

APP5 I admire people who initiate or are involved in activities to aid my

community

APP7 It is important to contribute to activities that tackle social problems

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

611

592

637

666

665

669

690

751

CPV

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of

public services

CPV6 It is fundamental that the interests of future generations are taken

into account when developing public policies

CPV7 To act ethically is essential for public servants

612

546

551

590

698

631

618

658

COM

COM1 It is difficult for me to contain my feelings when I see people in

distress

COM2 I feel sympathetic to the plight of the underprivileged

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

490

546

596

604

515

608

669

679

SS

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS7 I would agree to a good plan to make a better life for the poor even if

it costs me money

763

571

766

678

807

629

747

755

Measures of fit (MLE)

χsup2 (df = 98) = 8490 p lt 001

CFI = 939 GFI = 965

RMSEA = 052 [049 055]

Measure of fit (D-WLS)

χsup2 (df = 98) = 6069 p lt 001

CFI = 988 GFI = 992

RMSEA = 043 [039 046]

Alpha CR (D-WLS) Inter-factor correlations MLE (D-WLS)

1 APS 717 788 1 APS 2 CPV 3 COM

2 CPV 663 747 848 (859)

3 COM 645 713 781 (791) 789 (795)

4 SS 784 826 565 (596) 462 (505) 715 (726)

Note SFL = standardized factor loading Alpha = Cronbachrsquos alpha CR = composite reliability

All factor loadings and correlations are significant at p lt 001

- 38 -

Table 5

Results of CFA in the individual countries Standardized factor loadings and measures of fit

Factors

amp Items

Australia Belgium China Denmark France Italy

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

672

756

799

785

804

943

710

703

597

790

799

730

584

675

685

681

731

742

578

748

674

643

765

739

645

704

678

708

812

779

632

604

765

766

672

853

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

552

593

560

699

727

632

624

605

560

713

709

645

702

602

616

724

686

735

567

560

715

654

605

811

608

658

587

696

755

698

673

620

523

782

697

654

COM

COM1

COM2

COM3

COM5

COM6

623

584

662

688

701

739

732

608

475

808

666

581

578

616

605

649

670

673

699

751

737

679

816

776

568

546

801

626

673

850

637

546

741

715

601

822

SS

SS2

SS3

SS4

SS7

654

794

806

892

767

813

727

693

799

793

725

838

857

820

773

896

855

842

806

749

792

829

753

884

847

731

753

889

816

753

815

692

664

795

782

734

Inter-factor

correlations

093

~

752

163

~

810

320

~

773

323

~

825

452

~

827

497

~

855

323

~

850

321

~

847

498

~

899

536

~

902

425

~

829

476

~

865

Sample size 249 214 230 249 266 162

Measures of Fit

χsup2 638 444 881 732 765 554 817 625 735 565 750 564

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 980 100 939 987 969 997 967 994 975 997 949 995

GFI 961 986 941 986 948 991 949 992 955 992 926 986

RMSEA 036 00 063 050 051 026 053 035 045 026 059 033

- 39 -

Table 5 (Continued)

Factors

amp Items

Korea Lithuania The Netherlands Switzerland UK USA

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

619

701

667

706

781

717

666

512

682

751

605

779

557

629

630

620

719

729

535

621

702

643

694

735

655

725

738

710

804

822

535

621

702

643

694

735

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

594

607

722

722

644

722

674

535

533

790

589

711

568

554

538

655

628

545

627

580

615

711

666

704

800

512

555

873

581

646

627

580

615

711

666

704

COM

COM1

COM2

COM3

COM5

COM6

694

677

550

763

736

627

557

639

428

550

696

534

737

623

508

766

616

608

659

592

683

703

661

769

682

583

729

739

644

803

659

592

683

703

661

769

SS

SS2

SS3

SS4

SS7

776

752

679

779

770

811

707

721

635

725

734

706

814

656

743

820

767

764

745

632

689

788

662

744

671

818

832

830

864

802

745

632

689

788

662

744

Inter-factor

correlations

432

~

726

466

~

746

376

~

788

406

~

902

377

~

852

455

~

836

637

~

818

660

~

828

317

~

852

360

~

852

637

~

818

660

~

828

Sample size 253 236 249 250 260 250

Measures of Fit

χsup2 896 573 612 462 773 616 815 715 769 542 815 715

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 951 996 975 100 958 993 957 991 972 998 957 991

GFI 942 988 960 989 951 987 948 990 955 990 948 990

RMSEA 059 028 034 00 050 034 053 044 048 022 053 044

Note All factor loadings and correlations are significant at p lt 001

- 40 -

Appendix 1 Dimensions and possible items of PSM (the first round)

1 Attraction to Public Participation

(1) I am interested in developing public programs helping my country or community (Kim 2009a)

(2) Sharing my views on public policies with others is attractive to me (Kim 2009a)

(3) I am satisfied when seeing people get benefits from the public programs I was involved in (Kim 2009a)

(4) I like to discuss political subjects with others (Giauque et al 2011)

(5) I think the governmental activities contribute to our welfare

(6) I like to initiate actions to help out my community

(7) I do not care much about politicians (Perry 1996)

2 Commitment to Public Values

Possible items for public interest

(1A) Meaningful public service is very important to me (Perry 1996)

(2A) It is important for me to contribute for the common good (Giauque et al 2011)

(3A) I recognize myself with the mission of protection of liberties and rights endorsed by the nation-state

(Giauque et al 2011)

(4A) I would prefer seeing public officials do what is best for the whole community even if it harmed my

interests (Perry 1996)

(5A) To me serving the public interest is more important than helping other people (Vandenabeele 2008b)

Possible items for public values

(1B) I recognize myself with the promotion of the equality of chances endorsed by the nation-state (Giauque et

al 2011)

(2B) It is fundamental that the public service provided to the citizen is regular and continuous (Giauque et al

2011)

(3B) It is fundamental that public service innovates to respond to the new needs of the citizensclients (Giauque

et al 2011)

(4B) The dignity and well-being of all should be the most important concerns in any society (Steenbergen

1996)

(5B) It is important that public servants account for all the costsexpenses they make (Vandenabeele 2008b)

(6B) We have to do everything in our power to pursue the goal of democracy

3 Compassion

(1) It is difficult for me to contain my feelings when I see people in distress (Perry 1996)

(2) I feel sympathetic to the plight of the underprivileged (Kim 2009a)

(3) I am often reminded by daily events how dependent we are on one another (Perry 1996)

(4) I feel compassion for others in the difficulties they are facing (Steffen and Masters 2005)

(5) It is not really my problem if others are in trouble and need help (Beutel and Marini 1995)

(6) I find it hard to be sympathetic toward starving people in foreign lands when there is so much trouble in our

own country (Beutel and Marini 1995)

(7) I get very upset when I see other people treated unfairly (Beutel and Marini 1995)

(8) I would agree to a good plan to make a better life for the poor even if it cost me money (Beutel and Marini

1995)

(9) To me patriotism includes seeing the welfare of others (Perry 1996)

(10) I care very much about other people

4 Self-Sacrifice

(1) Making a difference in society means more to me than personal achievements (Perry 1996)

(2) I am prepared to make enormous sacrifices for the good of society (Perry 1996)

(3) I believe in putting duty before self (Perry 1996)

(4) I am one of those rare people who would risk personal loss to help someone else (Perry 1996)

(5) I think people should give back to society more than they get from it (Perry 1996)

(6) Serving other citizens would give me a good feeling even if no one paid me for it (Perry 1996)

(7) Much of what I do is for a cause bigger than myself (Perry 1996)

- 41 -

Appendix 2 Newly suggested items through the first round

1 Items for the Attraction to Public Participation Dimension

(1) In the public domain every person should be counted as equal

(2) I enjoy to think and argue in a political way

(3) I am satisfied when seeing my ideas being integrated in public policies

(4) I can easily relate to the way of thinking and arguing of politicians

(5) Working in the public sector I can better realize myself

2 Items for the Public Values Sub-Dimension

(1) Public service should give value for money

(2) Cost considerations should play an important role in decisions about public service provision

(3) It is fundamental that the interests of future generations are taken into account when developing public

policies

(4) To act with correct ethics is for public servants as important as their professional competence

(5) Primarily public servants should be accountable to the public and not to their superiors

3 Items for the Self-Sacrifice Dimension

(1) I would agree to a good plan to make a better life for the poor even if it costs me money

- 42 -

Appendix 3 An example of a summary of comments and suggestions (the second round)

On the item ―I am interested in developing public programs helping my country or community

A It is a good sentence but is it possible for everyone to ―develop public programs Even public officials are

limited to do this

B OK

C or the special group

D Double barrel and difficult to translate to my country ―I am interested in helping to improve public services

E Good but I feel that wording requires revision If we draw sample from local government then country is

redundant ―I am interested in developing public policy programs that aim in helping my community

F 4 (General appreciation from1 to 4) Not clear whether it refers to ones work or to ones concern as a citizen

―(In my work) I am interested in contributing to the development of public etchellip

G Instead of development I would use implementing

H Delete public which is redundant because of the characteristic helping my country or community but add

or implementing ―I am interested in developing or implementing programs helping my country or

community

I Referring to dimension Wording Similar to item 6

Note Names have been changed to alphabets in order to ensure the anonymity of participants

- 43 -

Appendix 4 Dimensions and Items of Public Service Motivation for Future Research

Dimensions and items

Attraction to Public Service (APS)

1 I am interested in helping to improve public service

2 I admire people who initiate or are involved in activities to aid my community

3 It is important to contribute to activities that tackle social problems

4 Meaningful public service is very important to me

5 It is important for me to contribute to the common good

Commitment to Public Values (CPV)

1 I think equal opportunities for citizens are very important

2 It is important that citizens can rely on the continuous provision of public services

3 It is fundamental that the interests of future generations are taken into account when

developing public policies

4 To act ethically is essential for public servants

5 I believe that public employees must always be aware of the legitimacy of their activities

6 I personally identify with the aim of protecting individual liberties and rights

Compassion (COM)

1 It is difficult for me to contain my feelings when I see people in distress

2 I feel sympathetic to the plight of the underprivileged

3 I empathize with other people who face difficulties

4 I get very upset when I see other people being treated unfairly

5 Considering the welfare of others is very important

Self-Sacrifice (SS)

1 I am prepared to make sacrifices for the good of society

2 I believe in putting civic duty before self

3 I am willing to risk personal loss to help society

4 I would agree to a good plan to make a better life for the poor even if it costs me money

- 19 -

covary with another dimension that has at least two reflective measures With one indicator

the dimension is identified if the loading and the variance of the uniqueness are both fixed

(Bollen 1989 Edwards 2010) Thus the number of items in each dimension is not restricted

when measuring PSM However we recommend using multiple items for each dimension

because more research is needed to investigate the appropriateness of this PSM instrument

with varying samples

The test results showed that model fit estimators were clearly better for the D-WLS

It means that in this study MLE was more rigorous than D-WLS in assessing the fit of the

model The empirical evidence indicated that MLE may be an acceptable technique for

ordinal data when there is a large sample a five-point Likert scale and the skewness and

kurtosis of variables is within acceptable limits as suggested by Byrne (2010)

Discussion

The analysis reaffirmed the four-dimensional structure of PSM but the results were not

entirely consistent with the original CFA we tested We were ultimately able to converge on a

good-fit operational model for PSM that was also consistent with underlying theory The

analysis is not able to resolve the debate about measurement models however for reasons

discussed below

Choosing the correct measurement model (ie reflective vs formative) is important

as measurement model misspecification results not only in misrepresentation of the epistemic

relationships between constructs and their measures but also in biased parameter estimates of

structural relationships (Jarvis et al 2003 Law and Wong 1999 MacKenzie 2003

MacKenzie Podsakoff and Jarvis 2005) A formative measurement model in isolation is

- 20 -

under-identified and cannot be estimated in the case of covariance-based SEM (CB-SEM)

(Bollen 1989) A construct with formative indicators must emit paths to at least two unrelated

reflective indicators two unrelated latent constructs or some combination of the two in order

to solve the problem of under-identification (Jarvis MacKenzie and Podsakoff 2003

MacCallum and Browne 1993 MacKenzie Podsakoff and Jarvis 2005) Thus ―when

dependent constructs change the empirical nature of the formatively measured construct

changes (Wilcox Howell and Breivik 2008 1223) and ―when specifying formative

measurement models what might seem like an innocuous choice of outcomes to achieve

identification can have fundamental ramifications for the meaning of the formative construct

and the inferences based on it (Edwards 2010 7)

Earlier research found that the association between a construct and its measures

should remain stable regardless of the larger causal model in which the constructs and

measures are embedded (Edwards and Bagozzi 2000) The association between a construct

and its measures is generally stable for reflective measures but associations of formative

measures with their construct are unstable The susceptibility of formative measurement to

both interpretational confounding and external inconsistency recommends that formative

measurement is not an equally attractive alternative to reflective measurement and that

whenever possible researchers should opt for reflective measurement (Howell Breivik and

Wilcox 2007a 2007b Kim Shin and Grover 2010) In the case of second-order models

researchers are limited to only a second-order reflective model in CB-SEM (Chin 2010)

Regardless of the measurement debate and the tradeoffs between reflective or

formative specification this study makes a theoretical argument that PSM is a second-order

formative construct This second-order formative factor model (reflective first-order

formative second-order) is also suggested as an alternative model that captures the

- 21 -

multidimensionality of complex concepts and avoids the shortcomings of first-order

formative measurement (Edwards 2010) However this model itself cannot achieve

identification One possible strategy is to develop direct reflective measures of PSM based on

the view that PSM is a general concept that is not only influenced by its specific dimensions

but also can be assessed using general measures (Edwards 2010 Javis MacKenzie and

Podsakoff 2003)6 However these general measures have not yet been developed for the

PSM construct

Another way around the problem of underidentification of formative models is to

apply the component-based approach known as partial least squares path modeling (PLS-PM)

which can model formative indicators (Fornell and Bookstein 1982 Goumltz Liehr-Gobbers

and Krafft 2010) The CB-SEM and the PLS-PM methods have distinctive statistical

characteristics and selecting an approach to SEM depends on the particular research situation

CB-SEM is the method of choice for theory testing while PLS-PM is appropriate for

prediction-oriented applications (Ringle Goumltz Wetzels and Wilson 2009) One advantage of

the PLS method is that it demands significantly fewer requirements compared to that of

covariance structure analyses but nevertheless delivers consistent estimation results Another

asset of the PLS approach is its ability to deal with formative as well as reflective indicators

even within one structural equation model This indicates that the PLS approach can be

appropriate for explorative analysis of structural equation models (Chin 2010 Goumltz Liehr-

Gobbers and Krafft 2010) A Monte Carlo simulation study to compare CB-SEM and PLS-

PM showed that PLS-PM provides a viable approximation of model parameters when the

prerequisites for CB-SEM are not met (Ringle Goumltz Wetzels and Wilson 2009) A formative

6 For example job satisfaction can be assessed with general reflective measures such as ―in general I am

satisfied with my job and the formative dimensions can be assessed with measures that describe satisfaction

with each job facet (Edwards 2010 15)

- 22 -

composite in the form of indexes needs to be tested to know the relative weights (γ-

parameters) of the dimensions of PSM In PLS-PM a second-order formative factor is

directly measured by observed variables for all the first order factors While this approach

repeats the number of manifest variables used the model can be identified without adding

outcome constructs that would be considered inappropriate or irrelevant from a conceptual

standpoint (Chin 2010 Edwards 2010) The test result of PLS-PM using Smart PLS 20

(Ringle Wende and Will 2005) produced the function PSM = 31APS + 36CPV + 28COM

+ 31SS However one consideration is whether and how construct error is modeled for the

formative construct (Diamantopoulos 2006) When using PLS-PM the researcher is making

the assumption that there is no error at the construct level as all constructs in PLS-PM are

modeled without error When using PLS-PM the absence of an error term will tend to

increase the weights therefore inflating their importance (Bollen and Lennox 1991) The

primary statistic for assessing a formative indicator is its weight the partialized effect of the

indicator on its intended construct controlling for the effects of all other indicators of that

construct (Cenfetelli and Bassellier 2009) A formative measure is essentially a multiple

regression with the construct representing the dependent variable and the indicators as the

predictors (Diamantopoulos and Winklhofer 2001) PSM may be regarded as a weighted

linear composite of its dimensions The weights used to create the composites are different

across CE-SEM and PLS-PM The causal dimensional weights generated in CB-SEM are

estimated in a model in which a construct-level error term is specified (eg PSM = γ1APM +

γ2CPI + γ3COM + γ4SS + ζ) whereas the weights in PLS-PM are not (eg PSM = γ1APM +

γ2CPI + γ3COM + γ4SS) This leads to two different conceptualizations of the PSM construct

one in which the causal indicators do not capture the construct in its entirety and another in

which they do (Hardin et al 2010) These differences need to be discussed in future research

- 23 -

in this area

One practical suggestion to measure PSM is to apply unit weights to its dimensions if

we assume the effects of the four dimensions are equivalent A literature review indicates that

unit weights have substantial predictive validity when compared with regression weights The

literature shows that the empirical correlations between scores generated from differential and

unit-weighting approaches are quite high In creating composite scores unit weights can be

an appropriate approach for weighting under many circumstances (Bobko Roth and Buster

2007) Although the uses of unit-weight composites will likely result in the explanation of

less variance than empirical estimations of dimensional weights they will remain consistently

defined across studies and hence suitable for theory testing Alternatively more thorough

discussions on the dimensions of PSM may suggest fixed-weight composites (Hardin et al

2010) A previous study (Kim 2010 22) also suggested ―Overall indexing PSM by

measuring all the dimensions of PSM and simply summing the averages of each dimension

can make it easier to measure PSM in practice as PSM = APS + CPV + COM + SS

An advantage of regarding PSM as a second-order formative construct is that the

dimensions can be analyzed both individually and collectively This is important as the

dimensions are expected to have different antecedents and consequences For instance Perry

(1997) found some significant differences in the influence of independent variables on the

four dimensions of PSM Taylor (2007) confirmed that certain dimensions are more important

than others in influencing work outcomes Moynihan and Pandey (2007a) reported that red

tape is negatively related to one dimension but reform orientation and hierarchical authority

is positively related to another dimension A recent study found that the SS dimension has a

positive relationship with occupational intention but the other dimensions do not (Liu et al

2011) Thus it is valuable to analyze the causes and effects of individual dimensions of PSM

- 24 -

as well as of the PSM construct

Conclusion

The purpose of this study is to develop a measurement instrument for PSM for cross-national

research and comparison On the basis of the sharpened construct of PSM (Kim and

Vandenabeele 2010) using data from an international survey conducted in 12 countries we

found that the four-factor structure of PSM is generalizable across country cases This study

provides the measurement index of PSM with the four dimensions of APS CPV COM and

SS The APS dimension measures the instrumental motives of PSM the CPV dimension

determines the value-based motives of PSM while the COM and SS dimensions gauge the

identification motives and self-sacrifice respectively When all four dimensions of PSM are

used in a survey cross-national equivalence of survey items can be assured The items for

measuring PSM in future research are suggested in Appendix 4

This study has several limitations First looking for samples which were comparable

to the largest possible extent we opted for surveys in local government The results found

and discussed in this study are based on PSM among local government employees The

results might be different when surveying central government employees or other specific

groups in the public sector Thus the potential impact of the sample bias needs to be

considered in understanding this study It also means that the survey items for measuring

PSM should be examined with different samples in various areas and at different government

levels

Second the test results showed that there is a high degree of correlation among the

dimension such as APS CPV and COM Future research should not only investigate whether

- 25 -

the items reflect the core content of the dimensions but also continue to develop more

appropriate items for better discriminant validity of each dimension Third the content of

APP dimension was assumed to be distinct from the CPV dimension that focuses on personal

disposition to pursue public interest and public values However the public interest sub-

dimension was separated from the public values sub-dimension in the CPV dimension and

combined with the APP dimension to form the APS dimension Further research should

analyze the CPV dimension and investigate the appropriateness of the APS dimension

In order to facilitate research internationally the PSM construct needs to be

improved conceptually and operationally Kim and Vandenabeele (2010) have dealt with the

conceptual composition and operational dimension of PSM as well as the measurement

model of PSM The next step is to develop an internationally robust measurement instrument

of PSM This study proposes that scholars use all four of the dimensions in measuring the

conceptual components of PSM such as self-sacrifice instrumental value-based and affective

motives because they collectively represent PSM It provides a set of appropriate items for

measuring PSM and confirms the dimensional structure of PSM The survey results suggest

that people in different cultures understand PSM in the same fashion We know that this study

is simply a step along the way to developing a more robust measure but we hope it

contributes to enhanced cross-national research and comparison along with the generation of

cumulative knowledge internationally

- 26 -

REFERENCES

Anderson James C and David D Gerbing 1988 Structural Equation Modeling in Practice

A Review and Recommended Two-Step Approach Psychological Bulletin 103

411-423

Arbuckle James L 2009 Amos 180 Userrsquos Guide Crawfordville FL Amos Development

Co

Beck Joslashrgensen Torben and Barry Bozeman 2007 Public Values An Inventory

Administration amp Society 39(3) 354-381

Beutel Ann M and Margaret Mooney Marini 1995 Gender and values American

Sociological Review 60 436 - 448

Bobko Philip Philip L Roth and Maury A Buster 2007 The Usefulness of Unit Weights in

Creating Composite Scores A Literature Review Application to Content Validity and

Meta-Analysis Organizational Research Methods 10(4) 689-709

Bollen Kenneth A 1989 Structural Equations with Latent Variables New York Wiley

Bollen K A and R Lennox 1991 Conventional Wisdom on Measurement A Structural

Equation Perspective Psychological Bulletin 110(2) 305-314

Byrne Barbara M 2010 Structural Equation Modeling with Amos Basic Concepts

Applications and Programming 2nd

ed New York Routledge

Camilleri Emanuel 2006 Towards Developing an Organizational Commitment ― Public

Service Motivation Model for Maltese Public Service Employees Public Policy

and Administration 21 63-83

Castaing Seacutebastien 2006 The Effects of Psychological Contract Fulfillment and Public

Service Motivation on Organizational Commitment in the French Civil Service

Public Policy and Administration 21 84-98

Cenfetelli Ronald T and Geneviegraveve Bassellier 2009 Interpretation of Formative

Measurement in Information Systems Research MIS Quarterly 33(4) 689-707

Charles Michael B W Martin de Jong and Neal Ryan 2011 Public Values in Western

- 27 -

Europe A Temporal Perspective American Review of Public Administration 41

75-91

Chin Wynne W 2010 How to Write Up and Report PLS Analysis In Handbook of Partical

Least Squares edited by V Esposito Vinzi et al 655-690 Berlin Springer-Verlag

Coursey David H and Sanjay K Pandey 2007a Public Service Motivation Measurement

Testing an Abridged Version of Perryrsquos Proposed Scale Administration amp Society

39 547-568

Coursey David H and Sanjay K Pandey 2007b Content Domain Measurement and

Validity of the Red Tape Concept American Review of Public Administration 37

342-361

Coursey David H James L Perry Jeffrey L Brudney and Laura Littlepage 2008

Psychometric Verification of Perryrsquos Public Service Motivation Instrument Results

for Volunteer Exemplars Review of Public Personnel Administration 28 79-90

Davis Paul and Karen West 2009 What Do Public Values Mean for Public Action Putting

Public Values in Their Plural Place American Review of Public Administration 39

602-618

DeHart-Davis Leisha Justin Marlowe and Sanjay K Pandey 2006 Gender Dimensions of

Public Service Motivation Public Administration Review 66 873-887

Diamantopoulos Adamantios 2006 The error term in formative measurement models

Interpretation and modeling implications Journal of Modeling in Management 1(1)

7-17

Diamantopoulos Adamantios and Judy A Siguaw 2006 Formative versus Reflective

Indicators in Organizational Measure Development A Comparison and Empirical

Illustration British Journal of Management 17 263-282

Diamantopoulos Adamantios and Heidi M Winklhofer 2001 Index Construction with

Formative Indicators An Alternative to Scale Development Journal of Marketing

Research 38 269-277

Edwards Jeffrey R 2010 The Fallacy of Formative Measurement Organizational Research

- 28 -

Methods Online First published on August 19 2010 as

doi1011771094428110378369

Edwards Jeffrey R and Richard P Bagozzi 2000 On the Nature and Direction of

Relationships between Constructs and Measures Psychological Methods 5 155-

174

Flora David B and Patrick J Curran 2004 An Empirical Evaluation of Alternative

Methods of Estimation for Confirmatory Factor Analysis with Ordinal Data

Psychological Methods 9 466-491

Fornell Claes and Fred L Bookstein 1982 Two structural equations models LISREL and

PLS applied to customer exit-voice theory Journal of Marketing Research 19 440-

452

Giauque David Adrian Ritz Freacutedeacuteric Varone Simon Anderfuhren-Biget and Christian

Waldner 2011 Putting Public Service Motivation into Context A Balance between

Universalism and Particularism Forthcoming in International Review of

Administrative Sciences 77(2)

Goumltz Oliver Kerstin Liehr-Gobbers and Manfred Krafft 2010 Evaluation of Structural

Equation Models Using the Partial Least Squares (PLS) Approach In V Esposito

Vinzi et al (eds) Handbook of Partial Least Squares 691-711 DOI 101007978-3-

540-32827-8_30 Springer-Verlag Berlin Heidelberg

Guinot Christiane Julie Latreille and Michel Tenenhaus 2001 PLS Path Modelling and

Multiple Table Analysis Application to the Cosmetic Habits of Women in Ile-de-

France Chemometrics and Intelligent Laboratory Systems 58 247-259

Hair Joseph F Jr William C Black Barry J Babin and Rolph E Anderson 2010

Multivariate Data Analysis 7th

ed Upper Saddle River NJ Pearson Education

Hardin Andrew M Jerry Cha-Jan Chang Mark A Fuller and Gholamreza Torkzadeh 2010

Formative Measurement and Academic Research In Search of Measurement

Theory Educational and Psychological Measurement Online First published on

June 11 2010 as doi101177001316440370208

- 29 -

Howell Roy D Einar Breivik and James B Wilcox 2007a Reconsidering Formative

Measurement Psychological Methods 12 205-218

Howell Roy D Einar Breivik and James B Wilcox 2007b Is Formative Measurement

Really Measurement Reply to Bollen (2007) and Bagozzi (2007) Psychological

Methods 12 238-245

Jarvis Cheryl Burke Scott B MacKenzie and Philip M Podsakoff 2003 A Critical Review

of Construct Indicators and Measurement Model Misspecification in Marketing and

Consumer Research Journal of Consumer Research 30 199-218

Joumlreskog Karl G and Dag Soumlrbom 2005 LISREL 872 (Computer Software) Lincolnwood

IL Scientific Software International

Kim Gimun Bongsik Shin and Varun Grover 2010 Investigating Two Contradictory Views

of Formative Measurement in Information Systems Research MIS Quarterly 34

345-365

Kim Sangmook 2009a Revising Perryrsquos Measurement Scale of Public Service Motivation

American Review of Public Administration 39 149-163

Kim Sangmook 2009b Testing the Structure of Public Service Motivation in Korea A

Research Note Journal of Public Administration Research and Theory 19 839-

851

Kim Sangmook 2010 Testing a Revised Measurement of Public Service Motivation

Reflective versus Formative Specification Journal of Public Administration

Research and Theory Doi101093jopartmuq048

Kim Sangmook and Wouter Vandenabeele 2010 A strategy for building public service

motivation research internationally Public Administration Review 70 701-709

Kline Rex B 2005 Principles and Practice of Structural Equation Modeling 2nd

ed New

York Guilford Press

Law Kenneth and Chi-Sum Wong 1999 Multidimensional Constructs in Structural

Equation Analysis An Illustration using the Job Perception and Job Satisfaction

Constructs Journal of Management 25 143-160

- 30 -

Leisink Peter and Bram Steijn 2009 Public Service Motivation and Job Performance of

Public Sector Employees in the Netherlands International Review of

Administrative Sciences 75 35-52

Liu Bangcheng Chun Hui Jin Hu Wensheng Yang and Xinli Yu 2011 How well can

Public Service Motivation Connect with Occupational Intention International

Review of Administrative Sciences 77 191-211

Liu Bangcheng Ningyu Tang and Xiaomei Zhu 2008 Public Service Motivation and Job

Satisfaction in China An Investigation of Generalisability and Instrumentality

International Journal of Manpower 29 684-699

MacCallum Robert C and Michael W Browne 1993 The Use of Causal Indicators in

Covariance Structure Models Some Practical Issues Psychological Bulletin 114

533-541

MacKenzie Scott B 2003 The dangers of poor construct conceptualization Journal of the

Academy of Marketing Science 31(3) 323-326

MacKenzie Scott B Philip M Podsakoff and Cheryl Burke Jarvis 2005 The Problem of

Measurement Model Misspecification in Behavioral and Organizational Research

and Some Recommended Solutions Journal of Applied Psychology 90 710-730

Martinsen Dorte Sindbjerg and Torben Beck Joslashrgensen 2010 Accountability as a

Differentiated Value in Supranational Governance American Review of Public

Administration 40 742-760

Moynihan Donald P and Sanjay K Pandey 2007a The Role of Organizations in Fostering

Public Service Motivation Public Administration Review 67 40-53

Moynihan Donald P and Sanjay K Pandey 2007b Finding Workable Levers over Work

Motivation Comparing Job Satisfaction Job Involvement and Organizational

Commitment Administration amp Society 39 803-832

Pandey Sanjay K and Edmund C Stazyk 2008 Antecedents and Correlates of Public

Service Motivation In Motivation in Public Management edited by James L Perry

and Annie Hondeghem 101-117 Oxford Oxford University Press

- 31 -

Perry James L 1996 Measuring Public Service Motivation An Assessment of Construct

Reliability and Validity Journal of Public Administration Research and Theory 6

5-22

Perry James L 1997 Antecedents of Public Service Motivation Journal of Public

Administration Research and Theory 7 181-197

Perry James L 2000 Bringing Society in Toward a Theory of Public Service Motivation

Journal of Public Administration Research and Theory 10 471-488

Perry James L and Annie Hondeghem 2008a Motivation in Public Management Oxford

Oxford University Press

Perry James L and Annie Hondeghem 2008b Directions for Future Theory and Research

In Motivation in Public Management edited by James L Perry and Annie

Hondeghem 294-313 Oxford Oxford University Press

Perry James L Annie Hondeghem and Lois Recascino Wise 2010 Revisiting the

Motivational Bases of Public Service Twenty Years of Research and an Agenda for

the Future Public Administration Review 70(5) 681-690

Perry James L and Wouter Vandenabeele 2008 Behavioral dynamics Institutions

Identities and Self-Regulation In Motivation in Public Management edited by

James L Perry and Annie Hondeghem 56-79 Oxford Oxford University Press

Perry James L and Lois Recascino Wise 1990 The Motivational Bases of Public Service

Public Administration Review 50 367-373

Rainey Hal G 1982 Reward Preference among Public and Private Managers In Search of

the Service Ethic American Review of Public Administration 16 288-302

Ringle Christian M Oliver Goumltz Martin Wetzels and Bradley Wilson 2009 On the Use of

Formative Measurement Specifications in Structural Equation Modeling A Monte

Carlo Simulation Study to Compare Covariance-Based and Partial Least Squares

Model Estimation Methodologies Faculty of Economics and Business Administration

Maastricht Univeristy

Ringle C M S Wende and S Will 2005 Smart PLS 20 (M3) Beta

- 32 -

httpwwwsmartplsde

Ritz Adrian 2011 Attraction to Public Policy Making A Qualitative Inquiry into

Improvements in PSM Measurement Forthcoming in Public Administration

Spector Paul E 1992 Summated Rating Scales Construction An Introduction Newbury

Park CA Sage

Steenbergen Marco R 1996 Compassion and American public opinion An analysis of the

NES Humanitarianism Scale Report to the Board of Overseers of the National

Election Studies

Steffen Patrick R and Kevin S Masters 2005 Does compassion mediate the intrinsic

religion-health relationship Annals of Behavioral Medicine 30 217-223

Taylor Jeannette 2007 The Impact of Public Service Motives on Work Outcomes in

Australia A Comparative Multi-Dimensional Analysis Public Administration 85

931-959

Vandenabeele Wouter 2007 Towards a Theory of Public Service Motivation An

Institutional Approach Public Management Review 9 545-556

Vandenabeele Wouter 2008a Government Calling Public Service Motivation as an Element

in Selecting Government as an Employer of Choice Public Administration 86

1089-1105

Vandenabeele Wouter 2008b Development of a Public Service Motivation Measurement

Scale Corroborating and Extending Perryrsquos Measurement Instrument International

Public Management Journal 11 143-167

Vandenabeele Wouter 2009 The Mediating Effect of Job Satisfaction and Organizational

Commitment on Self-Reported Performance More Robust Evidence of the PSM-

Performance Relationship International Review of Administrative Sciences 75 11-

34

Wilcox James B Roy D Howell and Einar Breivik 2008 Questions about Formative

Measurement Journal of Business Research 61 1219-1228

Wise Lois Recascino 2000 The Public Service Culture In Public Administration Concepts

- 33 -

and Cases 7th ed edited by Richard J Stillman II 342-353 Boston Houghton

Mifflin

Wright Bradley E 2008 Methodological Challenges associated with Public Service

Motivation Research In Motivation in Public Management edited by James L

Perry and Annie Hondeghem 80-98 Oxford Oxford University Press

Wright Bradley E and Sanjay K Pandey 2008 Public Service Motivation and the

Assumption of Person-Organization Fit Testing the Mediating Effect of Value

Congruence Administration amp Society 40 502-521

- 34 -

Table 1

Possible Indicators of PSM and Descriptive Statistics (n = 2868)

Dimensions and items Mean Standard

Deviation

Item-Total

Correlation

Attraction to Public Participation (APP)

APP1 I am interested in helping to improve public service

APP2 I am satisfied when I see people benefiting from the public programs I was

involved in

APP3 I like to discuss topics regarding public programs and policies with others

APP4 I believe that public sector activities contribute to our general welfare

APP5 I admire people who initiate or are involved in activities to aid my

community

APP6 Contributing to public programs and policies helps me realize myself

APP7 It is important to contribute to activities that tackle social problems

412

441

355

393

423

351

409

696

662

917

753

672

881

734

524

421

482

412

493

478

552

Commitment to Public Values (CPV)

Sub-dimension for Public Interests (CPV1)

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

CPI3 I would prefer seeing public officials do what is best for the whole

community even if it harmed my interests

CPI4 Serving the public interest is more important than helping a single individual

Sub-dimension for Public Values (CPV2)

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of public

services

CPV3 It is fundamental that public services respond to the needs of the citizens

CPV4 Decisions regarding public services should be democratic despite the time

and effort it takes

CPV5 Everybody is entitled to a good service even if it costs a lot of money

CPV6 It is fundamental that the interests of future generations are taken into

account when developing public policies

CPV7 To act ethically is essential for public servants

CPV8 I believe that public employees must always be aware of the legitimacy of

their activities

CPV9 I personally identify with the aim of protecting individual liberties and rights

418

409

348

337

428

416

433

394

367

426

441

434

414

695

662

899

995

730

704

710

852

932

689

681

692

763

525

596

480

251

484

437

356

325

376

439

462

402

392

Compassion (COM) COM1 It is difficult for me to contain my feelings when I see people in distress

COM2 I feel sympathetic to the plight of the underprivileged

COM3 I empathize with other people who face difficulties

COM4 I have little compassion for people in need who are unwilling to take the

first step to help themselves

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

354

387

388

299

412

391

951

825

707

1083

755

765

399

511

389

011

487

502

Self-Sacrifice (SS)

SS1 Making a difference to society means more to me than personal achievements

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS5 People should give back to society more than they get from it

SS6 Serving other citizens would give me a good feeling even if no one paid me for

it

SS7 I would agree to a good plan to make a better life for the poor even if it costs

me money

357

329

339

302

345

363

350

904

872

949

982

888

1049

895

508

584

454

531

412

368

546

- 35 -

Table 2

Background of Respondents (n = 2868)

Variables Characteristics Respondents

Sex

Male

Female

NA

1209

1622

37

422

566

13

Length of service

(years)

0 ~ 10

10 ~ 20

20 ~ 30

30+

NA

1126

723

484

491

44

393

252

169

171

15

Organizational status

―Do you supervise employees

No

Yes

NA

1472

1076

320

513

375

112

Country

Australia

Belgium

China

Denmark

France

Italy

Korea

Lithuania

The Netherlands

Switzerland

United Kingdom

USA

249

214

230

249

266

162

253

236

249

250

260

250

87

75

80

87

93

56

88

82

87

87

91

87 Note NA = no answer

- 36 -

Table 3

Standardized factor loadings and correlations for the 31-item PSM index (n = 2868)

Dimensions and items SFL (MLE) SFL (D-WLS) Alpha

APP

APP1

APP2

APP4

APP5

APP6

APP7

597

486

455

558

490

600

657

552

514

625

549

665

699

CPV1

CPI1

CPI2

CPI3

CPI4

588

660

488

289

653

732

535

294

542

CPV2

CPV1

CPV2

CPV3

CPV4

CPV5

CPV6

CPV7

CPV8

CPV9

611

546

436

393

418

534

598

496

460

690

631

505

457

486

601

669

573

530

745

COM

COM1

COM2

COM3

COM5

COM6

496

569

498

597

607

516

651

538

670

684

687

SS

SS1

SS2

SS3

SS4

SS5

SS6

SS7

567

736

558

728

467

409

663

660

762

591

695

515

494

709

782

Measures of fit (MLE) χsup2 (df = 424) = 43699 p lt 001 CFI = 835 GFI = 899

RMSEA = 057 [055 059]

Measure of fit (D-WLS) χsup2 (df = 424) = 39247 p lt 001 CFI = 971 GFI = 976

RMSEA = 054 [052 055]

Inter-factor correlations MLE (D-WLS)

1 2 3 4

1 APS

2 CPV1 996 (988)

3 CPV2 902 (890) 798 (798)

4 COM 787 (783) 758 (762) 779 (783)

5 SS 700 (742) 800 (859) 509 (537) 720 (729)

Note SFL = standardized factor loading All factor loadings and correlations are significant at p lt 001

- 37 -

Table 4

Standardized factor loadings and correlations for the 16-item PSM index (n = 2868)

Dimensions and items SFL

(MLE)

SFL

(D-WLS)

APS

APP5 I admire people who initiate or are involved in activities to aid my

community

APP7 It is important to contribute to activities that tackle social problems

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

611

592

637

666

665

669

690

751

CPV

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of

public services

CPV6 It is fundamental that the interests of future generations are taken

into account when developing public policies

CPV7 To act ethically is essential for public servants

612

546

551

590

698

631

618

658

COM

COM1 It is difficult for me to contain my feelings when I see people in

distress

COM2 I feel sympathetic to the plight of the underprivileged

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

490

546

596

604

515

608

669

679

SS

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS7 I would agree to a good plan to make a better life for the poor even if

it costs me money

763

571

766

678

807

629

747

755

Measures of fit (MLE)

χsup2 (df = 98) = 8490 p lt 001

CFI = 939 GFI = 965

RMSEA = 052 [049 055]

Measure of fit (D-WLS)

χsup2 (df = 98) = 6069 p lt 001

CFI = 988 GFI = 992

RMSEA = 043 [039 046]

Alpha CR (D-WLS) Inter-factor correlations MLE (D-WLS)

1 APS 717 788 1 APS 2 CPV 3 COM

2 CPV 663 747 848 (859)

3 COM 645 713 781 (791) 789 (795)

4 SS 784 826 565 (596) 462 (505) 715 (726)

Note SFL = standardized factor loading Alpha = Cronbachrsquos alpha CR = composite reliability

All factor loadings and correlations are significant at p lt 001

- 38 -

Table 5

Results of CFA in the individual countries Standardized factor loadings and measures of fit

Factors

amp Items

Australia Belgium China Denmark France Italy

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

672

756

799

785

804

943

710

703

597

790

799

730

584

675

685

681

731

742

578

748

674

643

765

739

645

704

678

708

812

779

632

604

765

766

672

853

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

552

593

560

699

727

632

624

605

560

713

709

645

702

602

616

724

686

735

567

560

715

654

605

811

608

658

587

696

755

698

673

620

523

782

697

654

COM

COM1

COM2

COM3

COM5

COM6

623

584

662

688

701

739

732

608

475

808

666

581

578

616

605

649

670

673

699

751

737

679

816

776

568

546

801

626

673

850

637

546

741

715

601

822

SS

SS2

SS3

SS4

SS7

654

794

806

892

767

813

727

693

799

793

725

838

857

820

773

896

855

842

806

749

792

829

753

884

847

731

753

889

816

753

815

692

664

795

782

734

Inter-factor

correlations

093

~

752

163

~

810

320

~

773

323

~

825

452

~

827

497

~

855

323

~

850

321

~

847

498

~

899

536

~

902

425

~

829

476

~

865

Sample size 249 214 230 249 266 162

Measures of Fit

χsup2 638 444 881 732 765 554 817 625 735 565 750 564

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 980 100 939 987 969 997 967 994 975 997 949 995

GFI 961 986 941 986 948 991 949 992 955 992 926 986

RMSEA 036 00 063 050 051 026 053 035 045 026 059 033

- 39 -

Table 5 (Continued)

Factors

amp Items

Korea Lithuania The Netherlands Switzerland UK USA

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

619

701

667

706

781

717

666

512

682

751

605

779

557

629

630

620

719

729

535

621

702

643

694

735

655

725

738

710

804

822

535

621

702

643

694

735

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

594

607

722

722

644

722

674

535

533

790

589

711

568

554

538

655

628

545

627

580

615

711

666

704

800

512

555

873

581

646

627

580

615

711

666

704

COM

COM1

COM2

COM3

COM5

COM6

694

677

550

763

736

627

557

639

428

550

696

534

737

623

508

766

616

608

659

592

683

703

661

769

682

583

729

739

644

803

659

592

683

703

661

769

SS

SS2

SS3

SS4

SS7

776

752

679

779

770

811

707

721

635

725

734

706

814

656

743

820

767

764

745

632

689

788

662

744

671

818

832

830

864

802

745

632

689

788

662

744

Inter-factor

correlations

432

~

726

466

~

746

376

~

788

406

~

902

377

~

852

455

~

836

637

~

818

660

~

828

317

~

852

360

~

852

637

~

818

660

~

828

Sample size 253 236 249 250 260 250

Measures of Fit

χsup2 896 573 612 462 773 616 815 715 769 542 815 715

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 951 996 975 100 958 993 957 991 972 998 957 991

GFI 942 988 960 989 951 987 948 990 955 990 948 990

RMSEA 059 028 034 00 050 034 053 044 048 022 053 044

Note All factor loadings and correlations are significant at p lt 001

- 40 -

Appendix 1 Dimensions and possible items of PSM (the first round)

1 Attraction to Public Participation

(1) I am interested in developing public programs helping my country or community (Kim 2009a)

(2) Sharing my views on public policies with others is attractive to me (Kim 2009a)

(3) I am satisfied when seeing people get benefits from the public programs I was involved in (Kim 2009a)

(4) I like to discuss political subjects with others (Giauque et al 2011)

(5) I think the governmental activities contribute to our welfare

(6) I like to initiate actions to help out my community

(7) I do not care much about politicians (Perry 1996)

2 Commitment to Public Values

Possible items for public interest

(1A) Meaningful public service is very important to me (Perry 1996)

(2A) It is important for me to contribute for the common good (Giauque et al 2011)

(3A) I recognize myself with the mission of protection of liberties and rights endorsed by the nation-state

(Giauque et al 2011)

(4A) I would prefer seeing public officials do what is best for the whole community even if it harmed my

interests (Perry 1996)

(5A) To me serving the public interest is more important than helping other people (Vandenabeele 2008b)

Possible items for public values

(1B) I recognize myself with the promotion of the equality of chances endorsed by the nation-state (Giauque et

al 2011)

(2B) It is fundamental that the public service provided to the citizen is regular and continuous (Giauque et al

2011)

(3B) It is fundamental that public service innovates to respond to the new needs of the citizensclients (Giauque

et al 2011)

(4B) The dignity and well-being of all should be the most important concerns in any society (Steenbergen

1996)

(5B) It is important that public servants account for all the costsexpenses they make (Vandenabeele 2008b)

(6B) We have to do everything in our power to pursue the goal of democracy

3 Compassion

(1) It is difficult for me to contain my feelings when I see people in distress (Perry 1996)

(2) I feel sympathetic to the plight of the underprivileged (Kim 2009a)

(3) I am often reminded by daily events how dependent we are on one another (Perry 1996)

(4) I feel compassion for others in the difficulties they are facing (Steffen and Masters 2005)

(5) It is not really my problem if others are in trouble and need help (Beutel and Marini 1995)

(6) I find it hard to be sympathetic toward starving people in foreign lands when there is so much trouble in our

own country (Beutel and Marini 1995)

(7) I get very upset when I see other people treated unfairly (Beutel and Marini 1995)

(8) I would agree to a good plan to make a better life for the poor even if it cost me money (Beutel and Marini

1995)

(9) To me patriotism includes seeing the welfare of others (Perry 1996)

(10) I care very much about other people

4 Self-Sacrifice

(1) Making a difference in society means more to me than personal achievements (Perry 1996)

(2) I am prepared to make enormous sacrifices for the good of society (Perry 1996)

(3) I believe in putting duty before self (Perry 1996)

(4) I am one of those rare people who would risk personal loss to help someone else (Perry 1996)

(5) I think people should give back to society more than they get from it (Perry 1996)

(6) Serving other citizens would give me a good feeling even if no one paid me for it (Perry 1996)

(7) Much of what I do is for a cause bigger than myself (Perry 1996)

- 41 -

Appendix 2 Newly suggested items through the first round

1 Items for the Attraction to Public Participation Dimension

(1) In the public domain every person should be counted as equal

(2) I enjoy to think and argue in a political way

(3) I am satisfied when seeing my ideas being integrated in public policies

(4) I can easily relate to the way of thinking and arguing of politicians

(5) Working in the public sector I can better realize myself

2 Items for the Public Values Sub-Dimension

(1) Public service should give value for money

(2) Cost considerations should play an important role in decisions about public service provision

(3) It is fundamental that the interests of future generations are taken into account when developing public

policies

(4) To act with correct ethics is for public servants as important as their professional competence

(5) Primarily public servants should be accountable to the public and not to their superiors

3 Items for the Self-Sacrifice Dimension

(1) I would agree to a good plan to make a better life for the poor even if it costs me money

- 42 -

Appendix 3 An example of a summary of comments and suggestions (the second round)

On the item ―I am interested in developing public programs helping my country or community

A It is a good sentence but is it possible for everyone to ―develop public programs Even public officials are

limited to do this

B OK

C or the special group

D Double barrel and difficult to translate to my country ―I am interested in helping to improve public services

E Good but I feel that wording requires revision If we draw sample from local government then country is

redundant ―I am interested in developing public policy programs that aim in helping my community

F 4 (General appreciation from1 to 4) Not clear whether it refers to ones work or to ones concern as a citizen

―(In my work) I am interested in contributing to the development of public etchellip

G Instead of development I would use implementing

H Delete public which is redundant because of the characteristic helping my country or community but add

or implementing ―I am interested in developing or implementing programs helping my country or

community

I Referring to dimension Wording Similar to item 6

Note Names have been changed to alphabets in order to ensure the anonymity of participants

- 43 -

Appendix 4 Dimensions and Items of Public Service Motivation for Future Research

Dimensions and items

Attraction to Public Service (APS)

1 I am interested in helping to improve public service

2 I admire people who initiate or are involved in activities to aid my community

3 It is important to contribute to activities that tackle social problems

4 Meaningful public service is very important to me

5 It is important for me to contribute to the common good

Commitment to Public Values (CPV)

1 I think equal opportunities for citizens are very important

2 It is important that citizens can rely on the continuous provision of public services

3 It is fundamental that the interests of future generations are taken into account when

developing public policies

4 To act ethically is essential for public servants

5 I believe that public employees must always be aware of the legitimacy of their activities

6 I personally identify with the aim of protecting individual liberties and rights

Compassion (COM)

1 It is difficult for me to contain my feelings when I see people in distress

2 I feel sympathetic to the plight of the underprivileged

3 I empathize with other people who face difficulties

4 I get very upset when I see other people being treated unfairly

5 Considering the welfare of others is very important

Self-Sacrifice (SS)

1 I am prepared to make sacrifices for the good of society

2 I believe in putting civic duty before self

3 I am willing to risk personal loss to help society

4 I would agree to a good plan to make a better life for the poor even if it costs me money

- 20 -

under-identified and cannot be estimated in the case of covariance-based SEM (CB-SEM)

(Bollen 1989) A construct with formative indicators must emit paths to at least two unrelated

reflective indicators two unrelated latent constructs or some combination of the two in order

to solve the problem of under-identification (Jarvis MacKenzie and Podsakoff 2003

MacCallum and Browne 1993 MacKenzie Podsakoff and Jarvis 2005) Thus ―when

dependent constructs change the empirical nature of the formatively measured construct

changes (Wilcox Howell and Breivik 2008 1223) and ―when specifying formative

measurement models what might seem like an innocuous choice of outcomes to achieve

identification can have fundamental ramifications for the meaning of the formative construct

and the inferences based on it (Edwards 2010 7)

Earlier research found that the association between a construct and its measures

should remain stable regardless of the larger causal model in which the constructs and

measures are embedded (Edwards and Bagozzi 2000) The association between a construct

and its measures is generally stable for reflective measures but associations of formative

measures with their construct are unstable The susceptibility of formative measurement to

both interpretational confounding and external inconsistency recommends that formative

measurement is not an equally attractive alternative to reflective measurement and that

whenever possible researchers should opt for reflective measurement (Howell Breivik and

Wilcox 2007a 2007b Kim Shin and Grover 2010) In the case of second-order models

researchers are limited to only a second-order reflective model in CB-SEM (Chin 2010)

Regardless of the measurement debate and the tradeoffs between reflective or

formative specification this study makes a theoretical argument that PSM is a second-order

formative construct This second-order formative factor model (reflective first-order

formative second-order) is also suggested as an alternative model that captures the

- 21 -

multidimensionality of complex concepts and avoids the shortcomings of first-order

formative measurement (Edwards 2010) However this model itself cannot achieve

identification One possible strategy is to develop direct reflective measures of PSM based on

the view that PSM is a general concept that is not only influenced by its specific dimensions

but also can be assessed using general measures (Edwards 2010 Javis MacKenzie and

Podsakoff 2003)6 However these general measures have not yet been developed for the

PSM construct

Another way around the problem of underidentification of formative models is to

apply the component-based approach known as partial least squares path modeling (PLS-PM)

which can model formative indicators (Fornell and Bookstein 1982 Goumltz Liehr-Gobbers

and Krafft 2010) The CB-SEM and the PLS-PM methods have distinctive statistical

characteristics and selecting an approach to SEM depends on the particular research situation

CB-SEM is the method of choice for theory testing while PLS-PM is appropriate for

prediction-oriented applications (Ringle Goumltz Wetzels and Wilson 2009) One advantage of

the PLS method is that it demands significantly fewer requirements compared to that of

covariance structure analyses but nevertheless delivers consistent estimation results Another

asset of the PLS approach is its ability to deal with formative as well as reflective indicators

even within one structural equation model This indicates that the PLS approach can be

appropriate for explorative analysis of structural equation models (Chin 2010 Goumltz Liehr-

Gobbers and Krafft 2010) A Monte Carlo simulation study to compare CB-SEM and PLS-

PM showed that PLS-PM provides a viable approximation of model parameters when the

prerequisites for CB-SEM are not met (Ringle Goumltz Wetzels and Wilson 2009) A formative

6 For example job satisfaction can be assessed with general reflective measures such as ―in general I am

satisfied with my job and the formative dimensions can be assessed with measures that describe satisfaction

with each job facet (Edwards 2010 15)

- 22 -

composite in the form of indexes needs to be tested to know the relative weights (γ-

parameters) of the dimensions of PSM In PLS-PM a second-order formative factor is

directly measured by observed variables for all the first order factors While this approach

repeats the number of manifest variables used the model can be identified without adding

outcome constructs that would be considered inappropriate or irrelevant from a conceptual

standpoint (Chin 2010 Edwards 2010) The test result of PLS-PM using Smart PLS 20

(Ringle Wende and Will 2005) produced the function PSM = 31APS + 36CPV + 28COM

+ 31SS However one consideration is whether and how construct error is modeled for the

formative construct (Diamantopoulos 2006) When using PLS-PM the researcher is making

the assumption that there is no error at the construct level as all constructs in PLS-PM are

modeled without error When using PLS-PM the absence of an error term will tend to

increase the weights therefore inflating their importance (Bollen and Lennox 1991) The

primary statistic for assessing a formative indicator is its weight the partialized effect of the

indicator on its intended construct controlling for the effects of all other indicators of that

construct (Cenfetelli and Bassellier 2009) A formative measure is essentially a multiple

regression with the construct representing the dependent variable and the indicators as the

predictors (Diamantopoulos and Winklhofer 2001) PSM may be regarded as a weighted

linear composite of its dimensions The weights used to create the composites are different

across CE-SEM and PLS-PM The causal dimensional weights generated in CB-SEM are

estimated in a model in which a construct-level error term is specified (eg PSM = γ1APM +

γ2CPI + γ3COM + γ4SS + ζ) whereas the weights in PLS-PM are not (eg PSM = γ1APM +

γ2CPI + γ3COM + γ4SS) This leads to two different conceptualizations of the PSM construct

one in which the causal indicators do not capture the construct in its entirety and another in

which they do (Hardin et al 2010) These differences need to be discussed in future research

- 23 -

in this area

One practical suggestion to measure PSM is to apply unit weights to its dimensions if

we assume the effects of the four dimensions are equivalent A literature review indicates that

unit weights have substantial predictive validity when compared with regression weights The

literature shows that the empirical correlations between scores generated from differential and

unit-weighting approaches are quite high In creating composite scores unit weights can be

an appropriate approach for weighting under many circumstances (Bobko Roth and Buster

2007) Although the uses of unit-weight composites will likely result in the explanation of

less variance than empirical estimations of dimensional weights they will remain consistently

defined across studies and hence suitable for theory testing Alternatively more thorough

discussions on the dimensions of PSM may suggest fixed-weight composites (Hardin et al

2010) A previous study (Kim 2010 22) also suggested ―Overall indexing PSM by

measuring all the dimensions of PSM and simply summing the averages of each dimension

can make it easier to measure PSM in practice as PSM = APS + CPV + COM + SS

An advantage of regarding PSM as a second-order formative construct is that the

dimensions can be analyzed both individually and collectively This is important as the

dimensions are expected to have different antecedents and consequences For instance Perry

(1997) found some significant differences in the influence of independent variables on the

four dimensions of PSM Taylor (2007) confirmed that certain dimensions are more important

than others in influencing work outcomes Moynihan and Pandey (2007a) reported that red

tape is negatively related to one dimension but reform orientation and hierarchical authority

is positively related to another dimension A recent study found that the SS dimension has a

positive relationship with occupational intention but the other dimensions do not (Liu et al

2011) Thus it is valuable to analyze the causes and effects of individual dimensions of PSM

- 24 -

as well as of the PSM construct

Conclusion

The purpose of this study is to develop a measurement instrument for PSM for cross-national

research and comparison On the basis of the sharpened construct of PSM (Kim and

Vandenabeele 2010) using data from an international survey conducted in 12 countries we

found that the four-factor structure of PSM is generalizable across country cases This study

provides the measurement index of PSM with the four dimensions of APS CPV COM and

SS The APS dimension measures the instrumental motives of PSM the CPV dimension

determines the value-based motives of PSM while the COM and SS dimensions gauge the

identification motives and self-sacrifice respectively When all four dimensions of PSM are

used in a survey cross-national equivalence of survey items can be assured The items for

measuring PSM in future research are suggested in Appendix 4

This study has several limitations First looking for samples which were comparable

to the largest possible extent we opted for surveys in local government The results found

and discussed in this study are based on PSM among local government employees The

results might be different when surveying central government employees or other specific

groups in the public sector Thus the potential impact of the sample bias needs to be

considered in understanding this study It also means that the survey items for measuring

PSM should be examined with different samples in various areas and at different government

levels

Second the test results showed that there is a high degree of correlation among the

dimension such as APS CPV and COM Future research should not only investigate whether

- 25 -

the items reflect the core content of the dimensions but also continue to develop more

appropriate items for better discriminant validity of each dimension Third the content of

APP dimension was assumed to be distinct from the CPV dimension that focuses on personal

disposition to pursue public interest and public values However the public interest sub-

dimension was separated from the public values sub-dimension in the CPV dimension and

combined with the APP dimension to form the APS dimension Further research should

analyze the CPV dimension and investigate the appropriateness of the APS dimension

In order to facilitate research internationally the PSM construct needs to be

improved conceptually and operationally Kim and Vandenabeele (2010) have dealt with the

conceptual composition and operational dimension of PSM as well as the measurement

model of PSM The next step is to develop an internationally robust measurement instrument

of PSM This study proposes that scholars use all four of the dimensions in measuring the

conceptual components of PSM such as self-sacrifice instrumental value-based and affective

motives because they collectively represent PSM It provides a set of appropriate items for

measuring PSM and confirms the dimensional structure of PSM The survey results suggest

that people in different cultures understand PSM in the same fashion We know that this study

is simply a step along the way to developing a more robust measure but we hope it

contributes to enhanced cross-national research and comparison along with the generation of

cumulative knowledge internationally

- 26 -

REFERENCES

Anderson James C and David D Gerbing 1988 Structural Equation Modeling in Practice

A Review and Recommended Two-Step Approach Psychological Bulletin 103

411-423

Arbuckle James L 2009 Amos 180 Userrsquos Guide Crawfordville FL Amos Development

Co

Beck Joslashrgensen Torben and Barry Bozeman 2007 Public Values An Inventory

Administration amp Society 39(3) 354-381

Beutel Ann M and Margaret Mooney Marini 1995 Gender and values American

Sociological Review 60 436 - 448

Bobko Philip Philip L Roth and Maury A Buster 2007 The Usefulness of Unit Weights in

Creating Composite Scores A Literature Review Application to Content Validity and

Meta-Analysis Organizational Research Methods 10(4) 689-709

Bollen Kenneth A 1989 Structural Equations with Latent Variables New York Wiley

Bollen K A and R Lennox 1991 Conventional Wisdom on Measurement A Structural

Equation Perspective Psychological Bulletin 110(2) 305-314

Byrne Barbara M 2010 Structural Equation Modeling with Amos Basic Concepts

Applications and Programming 2nd

ed New York Routledge

Camilleri Emanuel 2006 Towards Developing an Organizational Commitment ― Public

Service Motivation Model for Maltese Public Service Employees Public Policy

and Administration 21 63-83

Castaing Seacutebastien 2006 The Effects of Psychological Contract Fulfillment and Public

Service Motivation on Organizational Commitment in the French Civil Service

Public Policy and Administration 21 84-98

Cenfetelli Ronald T and Geneviegraveve Bassellier 2009 Interpretation of Formative

Measurement in Information Systems Research MIS Quarterly 33(4) 689-707

Charles Michael B W Martin de Jong and Neal Ryan 2011 Public Values in Western

- 27 -

Europe A Temporal Perspective American Review of Public Administration 41

75-91

Chin Wynne W 2010 How to Write Up and Report PLS Analysis In Handbook of Partical

Least Squares edited by V Esposito Vinzi et al 655-690 Berlin Springer-Verlag

Coursey David H and Sanjay K Pandey 2007a Public Service Motivation Measurement

Testing an Abridged Version of Perryrsquos Proposed Scale Administration amp Society

39 547-568

Coursey David H and Sanjay K Pandey 2007b Content Domain Measurement and

Validity of the Red Tape Concept American Review of Public Administration 37

342-361

Coursey David H James L Perry Jeffrey L Brudney and Laura Littlepage 2008

Psychometric Verification of Perryrsquos Public Service Motivation Instrument Results

for Volunteer Exemplars Review of Public Personnel Administration 28 79-90

Davis Paul and Karen West 2009 What Do Public Values Mean for Public Action Putting

Public Values in Their Plural Place American Review of Public Administration 39

602-618

DeHart-Davis Leisha Justin Marlowe and Sanjay K Pandey 2006 Gender Dimensions of

Public Service Motivation Public Administration Review 66 873-887

Diamantopoulos Adamantios 2006 The error term in formative measurement models

Interpretation and modeling implications Journal of Modeling in Management 1(1)

7-17

Diamantopoulos Adamantios and Judy A Siguaw 2006 Formative versus Reflective

Indicators in Organizational Measure Development A Comparison and Empirical

Illustration British Journal of Management 17 263-282

Diamantopoulos Adamantios and Heidi M Winklhofer 2001 Index Construction with

Formative Indicators An Alternative to Scale Development Journal of Marketing

Research 38 269-277

Edwards Jeffrey R 2010 The Fallacy of Formative Measurement Organizational Research

- 28 -

Methods Online First published on August 19 2010 as

doi1011771094428110378369

Edwards Jeffrey R and Richard P Bagozzi 2000 On the Nature and Direction of

Relationships between Constructs and Measures Psychological Methods 5 155-

174

Flora David B and Patrick J Curran 2004 An Empirical Evaluation of Alternative

Methods of Estimation for Confirmatory Factor Analysis with Ordinal Data

Psychological Methods 9 466-491

Fornell Claes and Fred L Bookstein 1982 Two structural equations models LISREL and

PLS applied to customer exit-voice theory Journal of Marketing Research 19 440-

452

Giauque David Adrian Ritz Freacutedeacuteric Varone Simon Anderfuhren-Biget and Christian

Waldner 2011 Putting Public Service Motivation into Context A Balance between

Universalism and Particularism Forthcoming in International Review of

Administrative Sciences 77(2)

Goumltz Oliver Kerstin Liehr-Gobbers and Manfred Krafft 2010 Evaluation of Structural

Equation Models Using the Partial Least Squares (PLS) Approach In V Esposito

Vinzi et al (eds) Handbook of Partial Least Squares 691-711 DOI 101007978-3-

540-32827-8_30 Springer-Verlag Berlin Heidelberg

Guinot Christiane Julie Latreille and Michel Tenenhaus 2001 PLS Path Modelling and

Multiple Table Analysis Application to the Cosmetic Habits of Women in Ile-de-

France Chemometrics and Intelligent Laboratory Systems 58 247-259

Hair Joseph F Jr William C Black Barry J Babin and Rolph E Anderson 2010

Multivariate Data Analysis 7th

ed Upper Saddle River NJ Pearson Education

Hardin Andrew M Jerry Cha-Jan Chang Mark A Fuller and Gholamreza Torkzadeh 2010

Formative Measurement and Academic Research In Search of Measurement

Theory Educational and Psychological Measurement Online First published on

June 11 2010 as doi101177001316440370208

- 29 -

Howell Roy D Einar Breivik and James B Wilcox 2007a Reconsidering Formative

Measurement Psychological Methods 12 205-218

Howell Roy D Einar Breivik and James B Wilcox 2007b Is Formative Measurement

Really Measurement Reply to Bollen (2007) and Bagozzi (2007) Psychological

Methods 12 238-245

Jarvis Cheryl Burke Scott B MacKenzie and Philip M Podsakoff 2003 A Critical Review

of Construct Indicators and Measurement Model Misspecification in Marketing and

Consumer Research Journal of Consumer Research 30 199-218

Joumlreskog Karl G and Dag Soumlrbom 2005 LISREL 872 (Computer Software) Lincolnwood

IL Scientific Software International

Kim Gimun Bongsik Shin and Varun Grover 2010 Investigating Two Contradictory Views

of Formative Measurement in Information Systems Research MIS Quarterly 34

345-365

Kim Sangmook 2009a Revising Perryrsquos Measurement Scale of Public Service Motivation

American Review of Public Administration 39 149-163

Kim Sangmook 2009b Testing the Structure of Public Service Motivation in Korea A

Research Note Journal of Public Administration Research and Theory 19 839-

851

Kim Sangmook 2010 Testing a Revised Measurement of Public Service Motivation

Reflective versus Formative Specification Journal of Public Administration

Research and Theory Doi101093jopartmuq048

Kim Sangmook and Wouter Vandenabeele 2010 A strategy for building public service

motivation research internationally Public Administration Review 70 701-709

Kline Rex B 2005 Principles and Practice of Structural Equation Modeling 2nd

ed New

York Guilford Press

Law Kenneth and Chi-Sum Wong 1999 Multidimensional Constructs in Structural

Equation Analysis An Illustration using the Job Perception and Job Satisfaction

Constructs Journal of Management 25 143-160

- 30 -

Leisink Peter and Bram Steijn 2009 Public Service Motivation and Job Performance of

Public Sector Employees in the Netherlands International Review of

Administrative Sciences 75 35-52

Liu Bangcheng Chun Hui Jin Hu Wensheng Yang and Xinli Yu 2011 How well can

Public Service Motivation Connect with Occupational Intention International

Review of Administrative Sciences 77 191-211

Liu Bangcheng Ningyu Tang and Xiaomei Zhu 2008 Public Service Motivation and Job

Satisfaction in China An Investigation of Generalisability and Instrumentality

International Journal of Manpower 29 684-699

MacCallum Robert C and Michael W Browne 1993 The Use of Causal Indicators in

Covariance Structure Models Some Practical Issues Psychological Bulletin 114

533-541

MacKenzie Scott B 2003 The dangers of poor construct conceptualization Journal of the

Academy of Marketing Science 31(3) 323-326

MacKenzie Scott B Philip M Podsakoff and Cheryl Burke Jarvis 2005 The Problem of

Measurement Model Misspecification in Behavioral and Organizational Research

and Some Recommended Solutions Journal of Applied Psychology 90 710-730

Martinsen Dorte Sindbjerg and Torben Beck Joslashrgensen 2010 Accountability as a

Differentiated Value in Supranational Governance American Review of Public

Administration 40 742-760

Moynihan Donald P and Sanjay K Pandey 2007a The Role of Organizations in Fostering

Public Service Motivation Public Administration Review 67 40-53

Moynihan Donald P and Sanjay K Pandey 2007b Finding Workable Levers over Work

Motivation Comparing Job Satisfaction Job Involvement and Organizational

Commitment Administration amp Society 39 803-832

Pandey Sanjay K and Edmund C Stazyk 2008 Antecedents and Correlates of Public

Service Motivation In Motivation in Public Management edited by James L Perry

and Annie Hondeghem 101-117 Oxford Oxford University Press

- 31 -

Perry James L 1996 Measuring Public Service Motivation An Assessment of Construct

Reliability and Validity Journal of Public Administration Research and Theory 6

5-22

Perry James L 1997 Antecedents of Public Service Motivation Journal of Public

Administration Research and Theory 7 181-197

Perry James L 2000 Bringing Society in Toward a Theory of Public Service Motivation

Journal of Public Administration Research and Theory 10 471-488

Perry James L and Annie Hondeghem 2008a Motivation in Public Management Oxford

Oxford University Press

Perry James L and Annie Hondeghem 2008b Directions for Future Theory and Research

In Motivation in Public Management edited by James L Perry and Annie

Hondeghem 294-313 Oxford Oxford University Press

Perry James L Annie Hondeghem and Lois Recascino Wise 2010 Revisiting the

Motivational Bases of Public Service Twenty Years of Research and an Agenda for

the Future Public Administration Review 70(5) 681-690

Perry James L and Wouter Vandenabeele 2008 Behavioral dynamics Institutions

Identities and Self-Regulation In Motivation in Public Management edited by

James L Perry and Annie Hondeghem 56-79 Oxford Oxford University Press

Perry James L and Lois Recascino Wise 1990 The Motivational Bases of Public Service

Public Administration Review 50 367-373

Rainey Hal G 1982 Reward Preference among Public and Private Managers In Search of

the Service Ethic American Review of Public Administration 16 288-302

Ringle Christian M Oliver Goumltz Martin Wetzels and Bradley Wilson 2009 On the Use of

Formative Measurement Specifications in Structural Equation Modeling A Monte

Carlo Simulation Study to Compare Covariance-Based and Partial Least Squares

Model Estimation Methodologies Faculty of Economics and Business Administration

Maastricht Univeristy

Ringle C M S Wende and S Will 2005 Smart PLS 20 (M3) Beta

- 32 -

httpwwwsmartplsde

Ritz Adrian 2011 Attraction to Public Policy Making A Qualitative Inquiry into

Improvements in PSM Measurement Forthcoming in Public Administration

Spector Paul E 1992 Summated Rating Scales Construction An Introduction Newbury

Park CA Sage

Steenbergen Marco R 1996 Compassion and American public opinion An analysis of the

NES Humanitarianism Scale Report to the Board of Overseers of the National

Election Studies

Steffen Patrick R and Kevin S Masters 2005 Does compassion mediate the intrinsic

religion-health relationship Annals of Behavioral Medicine 30 217-223

Taylor Jeannette 2007 The Impact of Public Service Motives on Work Outcomes in

Australia A Comparative Multi-Dimensional Analysis Public Administration 85

931-959

Vandenabeele Wouter 2007 Towards a Theory of Public Service Motivation An

Institutional Approach Public Management Review 9 545-556

Vandenabeele Wouter 2008a Government Calling Public Service Motivation as an Element

in Selecting Government as an Employer of Choice Public Administration 86

1089-1105

Vandenabeele Wouter 2008b Development of a Public Service Motivation Measurement

Scale Corroborating and Extending Perryrsquos Measurement Instrument International

Public Management Journal 11 143-167

Vandenabeele Wouter 2009 The Mediating Effect of Job Satisfaction and Organizational

Commitment on Self-Reported Performance More Robust Evidence of the PSM-

Performance Relationship International Review of Administrative Sciences 75 11-

34

Wilcox James B Roy D Howell and Einar Breivik 2008 Questions about Formative

Measurement Journal of Business Research 61 1219-1228

Wise Lois Recascino 2000 The Public Service Culture In Public Administration Concepts

- 33 -

and Cases 7th ed edited by Richard J Stillman II 342-353 Boston Houghton

Mifflin

Wright Bradley E 2008 Methodological Challenges associated with Public Service

Motivation Research In Motivation in Public Management edited by James L

Perry and Annie Hondeghem 80-98 Oxford Oxford University Press

Wright Bradley E and Sanjay K Pandey 2008 Public Service Motivation and the

Assumption of Person-Organization Fit Testing the Mediating Effect of Value

Congruence Administration amp Society 40 502-521

- 34 -

Table 1

Possible Indicators of PSM and Descriptive Statistics (n = 2868)

Dimensions and items Mean Standard

Deviation

Item-Total

Correlation

Attraction to Public Participation (APP)

APP1 I am interested in helping to improve public service

APP2 I am satisfied when I see people benefiting from the public programs I was

involved in

APP3 I like to discuss topics regarding public programs and policies with others

APP4 I believe that public sector activities contribute to our general welfare

APP5 I admire people who initiate or are involved in activities to aid my

community

APP6 Contributing to public programs and policies helps me realize myself

APP7 It is important to contribute to activities that tackle social problems

412

441

355

393

423

351

409

696

662

917

753

672

881

734

524

421

482

412

493

478

552

Commitment to Public Values (CPV)

Sub-dimension for Public Interests (CPV1)

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

CPI3 I would prefer seeing public officials do what is best for the whole

community even if it harmed my interests

CPI4 Serving the public interest is more important than helping a single individual

Sub-dimension for Public Values (CPV2)

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of public

services

CPV3 It is fundamental that public services respond to the needs of the citizens

CPV4 Decisions regarding public services should be democratic despite the time

and effort it takes

CPV5 Everybody is entitled to a good service even if it costs a lot of money

CPV6 It is fundamental that the interests of future generations are taken into

account when developing public policies

CPV7 To act ethically is essential for public servants

CPV8 I believe that public employees must always be aware of the legitimacy of

their activities

CPV9 I personally identify with the aim of protecting individual liberties and rights

418

409

348

337

428

416

433

394

367

426

441

434

414

695

662

899

995

730

704

710

852

932

689

681

692

763

525

596

480

251

484

437

356

325

376

439

462

402

392

Compassion (COM) COM1 It is difficult for me to contain my feelings when I see people in distress

COM2 I feel sympathetic to the plight of the underprivileged

COM3 I empathize with other people who face difficulties

COM4 I have little compassion for people in need who are unwilling to take the

first step to help themselves

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

354

387

388

299

412

391

951

825

707

1083

755

765

399

511

389

011

487

502

Self-Sacrifice (SS)

SS1 Making a difference to society means more to me than personal achievements

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS5 People should give back to society more than they get from it

SS6 Serving other citizens would give me a good feeling even if no one paid me for

it

SS7 I would agree to a good plan to make a better life for the poor even if it costs

me money

357

329

339

302

345

363

350

904

872

949

982

888

1049

895

508

584

454

531

412

368

546

- 35 -

Table 2

Background of Respondents (n = 2868)

Variables Characteristics Respondents

Sex

Male

Female

NA

1209

1622

37

422

566

13

Length of service

(years)

0 ~ 10

10 ~ 20

20 ~ 30

30+

NA

1126

723

484

491

44

393

252

169

171

15

Organizational status

―Do you supervise employees

No

Yes

NA

1472

1076

320

513

375

112

Country

Australia

Belgium

China

Denmark

France

Italy

Korea

Lithuania

The Netherlands

Switzerland

United Kingdom

USA

249

214

230

249

266

162

253

236

249

250

260

250

87

75

80

87

93

56

88

82

87

87

91

87 Note NA = no answer

- 36 -

Table 3

Standardized factor loadings and correlations for the 31-item PSM index (n = 2868)

Dimensions and items SFL (MLE) SFL (D-WLS) Alpha

APP

APP1

APP2

APP4

APP5

APP6

APP7

597

486

455

558

490

600

657

552

514

625

549

665

699

CPV1

CPI1

CPI2

CPI3

CPI4

588

660

488

289

653

732

535

294

542

CPV2

CPV1

CPV2

CPV3

CPV4

CPV5

CPV6

CPV7

CPV8

CPV9

611

546

436

393

418

534

598

496

460

690

631

505

457

486

601

669

573

530

745

COM

COM1

COM2

COM3

COM5

COM6

496

569

498

597

607

516

651

538

670

684

687

SS

SS1

SS2

SS3

SS4

SS5

SS6

SS7

567

736

558

728

467

409

663

660

762

591

695

515

494

709

782

Measures of fit (MLE) χsup2 (df = 424) = 43699 p lt 001 CFI = 835 GFI = 899

RMSEA = 057 [055 059]

Measure of fit (D-WLS) χsup2 (df = 424) = 39247 p lt 001 CFI = 971 GFI = 976

RMSEA = 054 [052 055]

Inter-factor correlations MLE (D-WLS)

1 2 3 4

1 APS

2 CPV1 996 (988)

3 CPV2 902 (890) 798 (798)

4 COM 787 (783) 758 (762) 779 (783)

5 SS 700 (742) 800 (859) 509 (537) 720 (729)

Note SFL = standardized factor loading All factor loadings and correlations are significant at p lt 001

- 37 -

Table 4

Standardized factor loadings and correlations for the 16-item PSM index (n = 2868)

Dimensions and items SFL

(MLE)

SFL

(D-WLS)

APS

APP5 I admire people who initiate or are involved in activities to aid my

community

APP7 It is important to contribute to activities that tackle social problems

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

611

592

637

666

665

669

690

751

CPV

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of

public services

CPV6 It is fundamental that the interests of future generations are taken

into account when developing public policies

CPV7 To act ethically is essential for public servants

612

546

551

590

698

631

618

658

COM

COM1 It is difficult for me to contain my feelings when I see people in

distress

COM2 I feel sympathetic to the plight of the underprivileged

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

490

546

596

604

515

608

669

679

SS

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS7 I would agree to a good plan to make a better life for the poor even if

it costs me money

763

571

766

678

807

629

747

755

Measures of fit (MLE)

χsup2 (df = 98) = 8490 p lt 001

CFI = 939 GFI = 965

RMSEA = 052 [049 055]

Measure of fit (D-WLS)

χsup2 (df = 98) = 6069 p lt 001

CFI = 988 GFI = 992

RMSEA = 043 [039 046]

Alpha CR (D-WLS) Inter-factor correlations MLE (D-WLS)

1 APS 717 788 1 APS 2 CPV 3 COM

2 CPV 663 747 848 (859)

3 COM 645 713 781 (791) 789 (795)

4 SS 784 826 565 (596) 462 (505) 715 (726)

Note SFL = standardized factor loading Alpha = Cronbachrsquos alpha CR = composite reliability

All factor loadings and correlations are significant at p lt 001

- 38 -

Table 5

Results of CFA in the individual countries Standardized factor loadings and measures of fit

Factors

amp Items

Australia Belgium China Denmark France Italy

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

672

756

799

785

804

943

710

703

597

790

799

730

584

675

685

681

731

742

578

748

674

643

765

739

645

704

678

708

812

779

632

604

765

766

672

853

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

552

593

560

699

727

632

624

605

560

713

709

645

702

602

616

724

686

735

567

560

715

654

605

811

608

658

587

696

755

698

673

620

523

782

697

654

COM

COM1

COM2

COM3

COM5

COM6

623

584

662

688

701

739

732

608

475

808

666

581

578

616

605

649

670

673

699

751

737

679

816

776

568

546

801

626

673

850

637

546

741

715

601

822

SS

SS2

SS3

SS4

SS7

654

794

806

892

767

813

727

693

799

793

725

838

857

820

773

896

855

842

806

749

792

829

753

884

847

731

753

889

816

753

815

692

664

795

782

734

Inter-factor

correlations

093

~

752

163

~

810

320

~

773

323

~

825

452

~

827

497

~

855

323

~

850

321

~

847

498

~

899

536

~

902

425

~

829

476

~

865

Sample size 249 214 230 249 266 162

Measures of Fit

χsup2 638 444 881 732 765 554 817 625 735 565 750 564

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 980 100 939 987 969 997 967 994 975 997 949 995

GFI 961 986 941 986 948 991 949 992 955 992 926 986

RMSEA 036 00 063 050 051 026 053 035 045 026 059 033

- 39 -

Table 5 (Continued)

Factors

amp Items

Korea Lithuania The Netherlands Switzerland UK USA

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

619

701

667

706

781

717

666

512

682

751

605

779

557

629

630

620

719

729

535

621

702

643

694

735

655

725

738

710

804

822

535

621

702

643

694

735

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

594

607

722

722

644

722

674

535

533

790

589

711

568

554

538

655

628

545

627

580

615

711

666

704

800

512

555

873

581

646

627

580

615

711

666

704

COM

COM1

COM2

COM3

COM5

COM6

694

677

550

763

736

627

557

639

428

550

696

534

737

623

508

766

616

608

659

592

683

703

661

769

682

583

729

739

644

803

659

592

683

703

661

769

SS

SS2

SS3

SS4

SS7

776

752

679

779

770

811

707

721

635

725

734

706

814

656

743

820

767

764

745

632

689

788

662

744

671

818

832

830

864

802

745

632

689

788

662

744

Inter-factor

correlations

432

~

726

466

~

746

376

~

788

406

~

902

377

~

852

455

~

836

637

~

818

660

~

828

317

~

852

360

~

852

637

~

818

660

~

828

Sample size 253 236 249 250 260 250

Measures of Fit

χsup2 896 573 612 462 773 616 815 715 769 542 815 715

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 951 996 975 100 958 993 957 991 972 998 957 991

GFI 942 988 960 989 951 987 948 990 955 990 948 990

RMSEA 059 028 034 00 050 034 053 044 048 022 053 044

Note All factor loadings and correlations are significant at p lt 001

- 40 -

Appendix 1 Dimensions and possible items of PSM (the first round)

1 Attraction to Public Participation

(1) I am interested in developing public programs helping my country or community (Kim 2009a)

(2) Sharing my views on public policies with others is attractive to me (Kim 2009a)

(3) I am satisfied when seeing people get benefits from the public programs I was involved in (Kim 2009a)

(4) I like to discuss political subjects with others (Giauque et al 2011)

(5) I think the governmental activities contribute to our welfare

(6) I like to initiate actions to help out my community

(7) I do not care much about politicians (Perry 1996)

2 Commitment to Public Values

Possible items for public interest

(1A) Meaningful public service is very important to me (Perry 1996)

(2A) It is important for me to contribute for the common good (Giauque et al 2011)

(3A) I recognize myself with the mission of protection of liberties and rights endorsed by the nation-state

(Giauque et al 2011)

(4A) I would prefer seeing public officials do what is best for the whole community even if it harmed my

interests (Perry 1996)

(5A) To me serving the public interest is more important than helping other people (Vandenabeele 2008b)

Possible items for public values

(1B) I recognize myself with the promotion of the equality of chances endorsed by the nation-state (Giauque et

al 2011)

(2B) It is fundamental that the public service provided to the citizen is regular and continuous (Giauque et al

2011)

(3B) It is fundamental that public service innovates to respond to the new needs of the citizensclients (Giauque

et al 2011)

(4B) The dignity and well-being of all should be the most important concerns in any society (Steenbergen

1996)

(5B) It is important that public servants account for all the costsexpenses they make (Vandenabeele 2008b)

(6B) We have to do everything in our power to pursue the goal of democracy

3 Compassion

(1) It is difficult for me to contain my feelings when I see people in distress (Perry 1996)

(2) I feel sympathetic to the plight of the underprivileged (Kim 2009a)

(3) I am often reminded by daily events how dependent we are on one another (Perry 1996)

(4) I feel compassion for others in the difficulties they are facing (Steffen and Masters 2005)

(5) It is not really my problem if others are in trouble and need help (Beutel and Marini 1995)

(6) I find it hard to be sympathetic toward starving people in foreign lands when there is so much trouble in our

own country (Beutel and Marini 1995)

(7) I get very upset when I see other people treated unfairly (Beutel and Marini 1995)

(8) I would agree to a good plan to make a better life for the poor even if it cost me money (Beutel and Marini

1995)

(9) To me patriotism includes seeing the welfare of others (Perry 1996)

(10) I care very much about other people

4 Self-Sacrifice

(1) Making a difference in society means more to me than personal achievements (Perry 1996)

(2) I am prepared to make enormous sacrifices for the good of society (Perry 1996)

(3) I believe in putting duty before self (Perry 1996)

(4) I am one of those rare people who would risk personal loss to help someone else (Perry 1996)

(5) I think people should give back to society more than they get from it (Perry 1996)

(6) Serving other citizens would give me a good feeling even if no one paid me for it (Perry 1996)

(7) Much of what I do is for a cause bigger than myself (Perry 1996)

- 41 -

Appendix 2 Newly suggested items through the first round

1 Items for the Attraction to Public Participation Dimension

(1) In the public domain every person should be counted as equal

(2) I enjoy to think and argue in a political way

(3) I am satisfied when seeing my ideas being integrated in public policies

(4) I can easily relate to the way of thinking and arguing of politicians

(5) Working in the public sector I can better realize myself

2 Items for the Public Values Sub-Dimension

(1) Public service should give value for money

(2) Cost considerations should play an important role in decisions about public service provision

(3) It is fundamental that the interests of future generations are taken into account when developing public

policies

(4) To act with correct ethics is for public servants as important as their professional competence

(5) Primarily public servants should be accountable to the public and not to their superiors

3 Items for the Self-Sacrifice Dimension

(1) I would agree to a good plan to make a better life for the poor even if it costs me money

- 42 -

Appendix 3 An example of a summary of comments and suggestions (the second round)

On the item ―I am interested in developing public programs helping my country or community

A It is a good sentence but is it possible for everyone to ―develop public programs Even public officials are

limited to do this

B OK

C or the special group

D Double barrel and difficult to translate to my country ―I am interested in helping to improve public services

E Good but I feel that wording requires revision If we draw sample from local government then country is

redundant ―I am interested in developing public policy programs that aim in helping my community

F 4 (General appreciation from1 to 4) Not clear whether it refers to ones work or to ones concern as a citizen

―(In my work) I am interested in contributing to the development of public etchellip

G Instead of development I would use implementing

H Delete public which is redundant because of the characteristic helping my country or community but add

or implementing ―I am interested in developing or implementing programs helping my country or

community

I Referring to dimension Wording Similar to item 6

Note Names have been changed to alphabets in order to ensure the anonymity of participants

- 43 -

Appendix 4 Dimensions and Items of Public Service Motivation for Future Research

Dimensions and items

Attraction to Public Service (APS)

1 I am interested in helping to improve public service

2 I admire people who initiate or are involved in activities to aid my community

3 It is important to contribute to activities that tackle social problems

4 Meaningful public service is very important to me

5 It is important for me to contribute to the common good

Commitment to Public Values (CPV)

1 I think equal opportunities for citizens are very important

2 It is important that citizens can rely on the continuous provision of public services

3 It is fundamental that the interests of future generations are taken into account when

developing public policies

4 To act ethically is essential for public servants

5 I believe that public employees must always be aware of the legitimacy of their activities

6 I personally identify with the aim of protecting individual liberties and rights

Compassion (COM)

1 It is difficult for me to contain my feelings when I see people in distress

2 I feel sympathetic to the plight of the underprivileged

3 I empathize with other people who face difficulties

4 I get very upset when I see other people being treated unfairly

5 Considering the welfare of others is very important

Self-Sacrifice (SS)

1 I am prepared to make sacrifices for the good of society

2 I believe in putting civic duty before self

3 I am willing to risk personal loss to help society

4 I would agree to a good plan to make a better life for the poor even if it costs me money

- 21 -

multidimensionality of complex concepts and avoids the shortcomings of first-order

formative measurement (Edwards 2010) However this model itself cannot achieve

identification One possible strategy is to develop direct reflective measures of PSM based on

the view that PSM is a general concept that is not only influenced by its specific dimensions

but also can be assessed using general measures (Edwards 2010 Javis MacKenzie and

Podsakoff 2003)6 However these general measures have not yet been developed for the

PSM construct

Another way around the problem of underidentification of formative models is to

apply the component-based approach known as partial least squares path modeling (PLS-PM)

which can model formative indicators (Fornell and Bookstein 1982 Goumltz Liehr-Gobbers

and Krafft 2010) The CB-SEM and the PLS-PM methods have distinctive statistical

characteristics and selecting an approach to SEM depends on the particular research situation

CB-SEM is the method of choice for theory testing while PLS-PM is appropriate for

prediction-oriented applications (Ringle Goumltz Wetzels and Wilson 2009) One advantage of

the PLS method is that it demands significantly fewer requirements compared to that of

covariance structure analyses but nevertheless delivers consistent estimation results Another

asset of the PLS approach is its ability to deal with formative as well as reflective indicators

even within one structural equation model This indicates that the PLS approach can be

appropriate for explorative analysis of structural equation models (Chin 2010 Goumltz Liehr-

Gobbers and Krafft 2010) A Monte Carlo simulation study to compare CB-SEM and PLS-

PM showed that PLS-PM provides a viable approximation of model parameters when the

prerequisites for CB-SEM are not met (Ringle Goumltz Wetzels and Wilson 2009) A formative

6 For example job satisfaction can be assessed with general reflective measures such as ―in general I am

satisfied with my job and the formative dimensions can be assessed with measures that describe satisfaction

with each job facet (Edwards 2010 15)

- 22 -

composite in the form of indexes needs to be tested to know the relative weights (γ-

parameters) of the dimensions of PSM In PLS-PM a second-order formative factor is

directly measured by observed variables for all the first order factors While this approach

repeats the number of manifest variables used the model can be identified without adding

outcome constructs that would be considered inappropriate or irrelevant from a conceptual

standpoint (Chin 2010 Edwards 2010) The test result of PLS-PM using Smart PLS 20

(Ringle Wende and Will 2005) produced the function PSM = 31APS + 36CPV + 28COM

+ 31SS However one consideration is whether and how construct error is modeled for the

formative construct (Diamantopoulos 2006) When using PLS-PM the researcher is making

the assumption that there is no error at the construct level as all constructs in PLS-PM are

modeled without error When using PLS-PM the absence of an error term will tend to

increase the weights therefore inflating their importance (Bollen and Lennox 1991) The

primary statistic for assessing a formative indicator is its weight the partialized effect of the

indicator on its intended construct controlling for the effects of all other indicators of that

construct (Cenfetelli and Bassellier 2009) A formative measure is essentially a multiple

regression with the construct representing the dependent variable and the indicators as the

predictors (Diamantopoulos and Winklhofer 2001) PSM may be regarded as a weighted

linear composite of its dimensions The weights used to create the composites are different

across CE-SEM and PLS-PM The causal dimensional weights generated in CB-SEM are

estimated in a model in which a construct-level error term is specified (eg PSM = γ1APM +

γ2CPI + γ3COM + γ4SS + ζ) whereas the weights in PLS-PM are not (eg PSM = γ1APM +

γ2CPI + γ3COM + γ4SS) This leads to two different conceptualizations of the PSM construct

one in which the causal indicators do not capture the construct in its entirety and another in

which they do (Hardin et al 2010) These differences need to be discussed in future research

- 23 -

in this area

One practical suggestion to measure PSM is to apply unit weights to its dimensions if

we assume the effects of the four dimensions are equivalent A literature review indicates that

unit weights have substantial predictive validity when compared with regression weights The

literature shows that the empirical correlations between scores generated from differential and

unit-weighting approaches are quite high In creating composite scores unit weights can be

an appropriate approach for weighting under many circumstances (Bobko Roth and Buster

2007) Although the uses of unit-weight composites will likely result in the explanation of

less variance than empirical estimations of dimensional weights they will remain consistently

defined across studies and hence suitable for theory testing Alternatively more thorough

discussions on the dimensions of PSM may suggest fixed-weight composites (Hardin et al

2010) A previous study (Kim 2010 22) also suggested ―Overall indexing PSM by

measuring all the dimensions of PSM and simply summing the averages of each dimension

can make it easier to measure PSM in practice as PSM = APS + CPV + COM + SS

An advantage of regarding PSM as a second-order formative construct is that the

dimensions can be analyzed both individually and collectively This is important as the

dimensions are expected to have different antecedents and consequences For instance Perry

(1997) found some significant differences in the influence of independent variables on the

four dimensions of PSM Taylor (2007) confirmed that certain dimensions are more important

than others in influencing work outcomes Moynihan and Pandey (2007a) reported that red

tape is negatively related to one dimension but reform orientation and hierarchical authority

is positively related to another dimension A recent study found that the SS dimension has a

positive relationship with occupational intention but the other dimensions do not (Liu et al

2011) Thus it is valuable to analyze the causes and effects of individual dimensions of PSM

- 24 -

as well as of the PSM construct

Conclusion

The purpose of this study is to develop a measurement instrument for PSM for cross-national

research and comparison On the basis of the sharpened construct of PSM (Kim and

Vandenabeele 2010) using data from an international survey conducted in 12 countries we

found that the four-factor structure of PSM is generalizable across country cases This study

provides the measurement index of PSM with the four dimensions of APS CPV COM and

SS The APS dimension measures the instrumental motives of PSM the CPV dimension

determines the value-based motives of PSM while the COM and SS dimensions gauge the

identification motives and self-sacrifice respectively When all four dimensions of PSM are

used in a survey cross-national equivalence of survey items can be assured The items for

measuring PSM in future research are suggested in Appendix 4

This study has several limitations First looking for samples which were comparable

to the largest possible extent we opted for surveys in local government The results found

and discussed in this study are based on PSM among local government employees The

results might be different when surveying central government employees or other specific

groups in the public sector Thus the potential impact of the sample bias needs to be

considered in understanding this study It also means that the survey items for measuring

PSM should be examined with different samples in various areas and at different government

levels

Second the test results showed that there is a high degree of correlation among the

dimension such as APS CPV and COM Future research should not only investigate whether

- 25 -

the items reflect the core content of the dimensions but also continue to develop more

appropriate items for better discriminant validity of each dimension Third the content of

APP dimension was assumed to be distinct from the CPV dimension that focuses on personal

disposition to pursue public interest and public values However the public interest sub-

dimension was separated from the public values sub-dimension in the CPV dimension and

combined with the APP dimension to form the APS dimension Further research should

analyze the CPV dimension and investigate the appropriateness of the APS dimension

In order to facilitate research internationally the PSM construct needs to be

improved conceptually and operationally Kim and Vandenabeele (2010) have dealt with the

conceptual composition and operational dimension of PSM as well as the measurement

model of PSM The next step is to develop an internationally robust measurement instrument

of PSM This study proposes that scholars use all four of the dimensions in measuring the

conceptual components of PSM such as self-sacrifice instrumental value-based and affective

motives because they collectively represent PSM It provides a set of appropriate items for

measuring PSM and confirms the dimensional structure of PSM The survey results suggest

that people in different cultures understand PSM in the same fashion We know that this study

is simply a step along the way to developing a more robust measure but we hope it

contributes to enhanced cross-national research and comparison along with the generation of

cumulative knowledge internationally

- 26 -

REFERENCES

Anderson James C and David D Gerbing 1988 Structural Equation Modeling in Practice

A Review and Recommended Two-Step Approach Psychological Bulletin 103

411-423

Arbuckle James L 2009 Amos 180 Userrsquos Guide Crawfordville FL Amos Development

Co

Beck Joslashrgensen Torben and Barry Bozeman 2007 Public Values An Inventory

Administration amp Society 39(3) 354-381

Beutel Ann M and Margaret Mooney Marini 1995 Gender and values American

Sociological Review 60 436 - 448

Bobko Philip Philip L Roth and Maury A Buster 2007 The Usefulness of Unit Weights in

Creating Composite Scores A Literature Review Application to Content Validity and

Meta-Analysis Organizational Research Methods 10(4) 689-709

Bollen Kenneth A 1989 Structural Equations with Latent Variables New York Wiley

Bollen K A and R Lennox 1991 Conventional Wisdom on Measurement A Structural

Equation Perspective Psychological Bulletin 110(2) 305-314

Byrne Barbara M 2010 Structural Equation Modeling with Amos Basic Concepts

Applications and Programming 2nd

ed New York Routledge

Camilleri Emanuel 2006 Towards Developing an Organizational Commitment ― Public

Service Motivation Model for Maltese Public Service Employees Public Policy

and Administration 21 63-83

Castaing Seacutebastien 2006 The Effects of Psychological Contract Fulfillment and Public

Service Motivation on Organizational Commitment in the French Civil Service

Public Policy and Administration 21 84-98

Cenfetelli Ronald T and Geneviegraveve Bassellier 2009 Interpretation of Formative

Measurement in Information Systems Research MIS Quarterly 33(4) 689-707

Charles Michael B W Martin de Jong and Neal Ryan 2011 Public Values in Western

- 27 -

Europe A Temporal Perspective American Review of Public Administration 41

75-91

Chin Wynne W 2010 How to Write Up and Report PLS Analysis In Handbook of Partical

Least Squares edited by V Esposito Vinzi et al 655-690 Berlin Springer-Verlag

Coursey David H and Sanjay K Pandey 2007a Public Service Motivation Measurement

Testing an Abridged Version of Perryrsquos Proposed Scale Administration amp Society

39 547-568

Coursey David H and Sanjay K Pandey 2007b Content Domain Measurement and

Validity of the Red Tape Concept American Review of Public Administration 37

342-361

Coursey David H James L Perry Jeffrey L Brudney and Laura Littlepage 2008

Psychometric Verification of Perryrsquos Public Service Motivation Instrument Results

for Volunteer Exemplars Review of Public Personnel Administration 28 79-90

Davis Paul and Karen West 2009 What Do Public Values Mean for Public Action Putting

Public Values in Their Plural Place American Review of Public Administration 39

602-618

DeHart-Davis Leisha Justin Marlowe and Sanjay K Pandey 2006 Gender Dimensions of

Public Service Motivation Public Administration Review 66 873-887

Diamantopoulos Adamantios 2006 The error term in formative measurement models

Interpretation and modeling implications Journal of Modeling in Management 1(1)

7-17

Diamantopoulos Adamantios and Judy A Siguaw 2006 Formative versus Reflective

Indicators in Organizational Measure Development A Comparison and Empirical

Illustration British Journal of Management 17 263-282

Diamantopoulos Adamantios and Heidi M Winklhofer 2001 Index Construction with

Formative Indicators An Alternative to Scale Development Journal of Marketing

Research 38 269-277

Edwards Jeffrey R 2010 The Fallacy of Formative Measurement Organizational Research

- 28 -

Methods Online First published on August 19 2010 as

doi1011771094428110378369

Edwards Jeffrey R and Richard P Bagozzi 2000 On the Nature and Direction of

Relationships between Constructs and Measures Psychological Methods 5 155-

174

Flora David B and Patrick J Curran 2004 An Empirical Evaluation of Alternative

Methods of Estimation for Confirmatory Factor Analysis with Ordinal Data

Psychological Methods 9 466-491

Fornell Claes and Fred L Bookstein 1982 Two structural equations models LISREL and

PLS applied to customer exit-voice theory Journal of Marketing Research 19 440-

452

Giauque David Adrian Ritz Freacutedeacuteric Varone Simon Anderfuhren-Biget and Christian

Waldner 2011 Putting Public Service Motivation into Context A Balance between

Universalism and Particularism Forthcoming in International Review of

Administrative Sciences 77(2)

Goumltz Oliver Kerstin Liehr-Gobbers and Manfred Krafft 2010 Evaluation of Structural

Equation Models Using the Partial Least Squares (PLS) Approach In V Esposito

Vinzi et al (eds) Handbook of Partial Least Squares 691-711 DOI 101007978-3-

540-32827-8_30 Springer-Verlag Berlin Heidelberg

Guinot Christiane Julie Latreille and Michel Tenenhaus 2001 PLS Path Modelling and

Multiple Table Analysis Application to the Cosmetic Habits of Women in Ile-de-

France Chemometrics and Intelligent Laboratory Systems 58 247-259

Hair Joseph F Jr William C Black Barry J Babin and Rolph E Anderson 2010

Multivariate Data Analysis 7th

ed Upper Saddle River NJ Pearson Education

Hardin Andrew M Jerry Cha-Jan Chang Mark A Fuller and Gholamreza Torkzadeh 2010

Formative Measurement and Academic Research In Search of Measurement

Theory Educational and Psychological Measurement Online First published on

June 11 2010 as doi101177001316440370208

- 29 -

Howell Roy D Einar Breivik and James B Wilcox 2007a Reconsidering Formative

Measurement Psychological Methods 12 205-218

Howell Roy D Einar Breivik and James B Wilcox 2007b Is Formative Measurement

Really Measurement Reply to Bollen (2007) and Bagozzi (2007) Psychological

Methods 12 238-245

Jarvis Cheryl Burke Scott B MacKenzie and Philip M Podsakoff 2003 A Critical Review

of Construct Indicators and Measurement Model Misspecification in Marketing and

Consumer Research Journal of Consumer Research 30 199-218

Joumlreskog Karl G and Dag Soumlrbom 2005 LISREL 872 (Computer Software) Lincolnwood

IL Scientific Software International

Kim Gimun Bongsik Shin and Varun Grover 2010 Investigating Two Contradictory Views

of Formative Measurement in Information Systems Research MIS Quarterly 34

345-365

Kim Sangmook 2009a Revising Perryrsquos Measurement Scale of Public Service Motivation

American Review of Public Administration 39 149-163

Kim Sangmook 2009b Testing the Structure of Public Service Motivation in Korea A

Research Note Journal of Public Administration Research and Theory 19 839-

851

Kim Sangmook 2010 Testing a Revised Measurement of Public Service Motivation

Reflective versus Formative Specification Journal of Public Administration

Research and Theory Doi101093jopartmuq048

Kim Sangmook and Wouter Vandenabeele 2010 A strategy for building public service

motivation research internationally Public Administration Review 70 701-709

Kline Rex B 2005 Principles and Practice of Structural Equation Modeling 2nd

ed New

York Guilford Press

Law Kenneth and Chi-Sum Wong 1999 Multidimensional Constructs in Structural

Equation Analysis An Illustration using the Job Perception and Job Satisfaction

Constructs Journal of Management 25 143-160

- 30 -

Leisink Peter and Bram Steijn 2009 Public Service Motivation and Job Performance of

Public Sector Employees in the Netherlands International Review of

Administrative Sciences 75 35-52

Liu Bangcheng Chun Hui Jin Hu Wensheng Yang and Xinli Yu 2011 How well can

Public Service Motivation Connect with Occupational Intention International

Review of Administrative Sciences 77 191-211

Liu Bangcheng Ningyu Tang and Xiaomei Zhu 2008 Public Service Motivation and Job

Satisfaction in China An Investigation of Generalisability and Instrumentality

International Journal of Manpower 29 684-699

MacCallum Robert C and Michael W Browne 1993 The Use of Causal Indicators in

Covariance Structure Models Some Practical Issues Psychological Bulletin 114

533-541

MacKenzie Scott B 2003 The dangers of poor construct conceptualization Journal of the

Academy of Marketing Science 31(3) 323-326

MacKenzie Scott B Philip M Podsakoff and Cheryl Burke Jarvis 2005 The Problem of

Measurement Model Misspecification in Behavioral and Organizational Research

and Some Recommended Solutions Journal of Applied Psychology 90 710-730

Martinsen Dorte Sindbjerg and Torben Beck Joslashrgensen 2010 Accountability as a

Differentiated Value in Supranational Governance American Review of Public

Administration 40 742-760

Moynihan Donald P and Sanjay K Pandey 2007a The Role of Organizations in Fostering

Public Service Motivation Public Administration Review 67 40-53

Moynihan Donald P and Sanjay K Pandey 2007b Finding Workable Levers over Work

Motivation Comparing Job Satisfaction Job Involvement and Organizational

Commitment Administration amp Society 39 803-832

Pandey Sanjay K and Edmund C Stazyk 2008 Antecedents and Correlates of Public

Service Motivation In Motivation in Public Management edited by James L Perry

and Annie Hondeghem 101-117 Oxford Oxford University Press

- 31 -

Perry James L 1996 Measuring Public Service Motivation An Assessment of Construct

Reliability and Validity Journal of Public Administration Research and Theory 6

5-22

Perry James L 1997 Antecedents of Public Service Motivation Journal of Public

Administration Research and Theory 7 181-197

Perry James L 2000 Bringing Society in Toward a Theory of Public Service Motivation

Journal of Public Administration Research and Theory 10 471-488

Perry James L and Annie Hondeghem 2008a Motivation in Public Management Oxford

Oxford University Press

Perry James L and Annie Hondeghem 2008b Directions for Future Theory and Research

In Motivation in Public Management edited by James L Perry and Annie

Hondeghem 294-313 Oxford Oxford University Press

Perry James L Annie Hondeghem and Lois Recascino Wise 2010 Revisiting the

Motivational Bases of Public Service Twenty Years of Research and an Agenda for

the Future Public Administration Review 70(5) 681-690

Perry James L and Wouter Vandenabeele 2008 Behavioral dynamics Institutions

Identities and Self-Regulation In Motivation in Public Management edited by

James L Perry and Annie Hondeghem 56-79 Oxford Oxford University Press

Perry James L and Lois Recascino Wise 1990 The Motivational Bases of Public Service

Public Administration Review 50 367-373

Rainey Hal G 1982 Reward Preference among Public and Private Managers In Search of

the Service Ethic American Review of Public Administration 16 288-302

Ringle Christian M Oliver Goumltz Martin Wetzels and Bradley Wilson 2009 On the Use of

Formative Measurement Specifications in Structural Equation Modeling A Monte

Carlo Simulation Study to Compare Covariance-Based and Partial Least Squares

Model Estimation Methodologies Faculty of Economics and Business Administration

Maastricht Univeristy

Ringle C M S Wende and S Will 2005 Smart PLS 20 (M3) Beta

- 32 -

httpwwwsmartplsde

Ritz Adrian 2011 Attraction to Public Policy Making A Qualitative Inquiry into

Improvements in PSM Measurement Forthcoming in Public Administration

Spector Paul E 1992 Summated Rating Scales Construction An Introduction Newbury

Park CA Sage

Steenbergen Marco R 1996 Compassion and American public opinion An analysis of the

NES Humanitarianism Scale Report to the Board of Overseers of the National

Election Studies

Steffen Patrick R and Kevin S Masters 2005 Does compassion mediate the intrinsic

religion-health relationship Annals of Behavioral Medicine 30 217-223

Taylor Jeannette 2007 The Impact of Public Service Motives on Work Outcomes in

Australia A Comparative Multi-Dimensional Analysis Public Administration 85

931-959

Vandenabeele Wouter 2007 Towards a Theory of Public Service Motivation An

Institutional Approach Public Management Review 9 545-556

Vandenabeele Wouter 2008a Government Calling Public Service Motivation as an Element

in Selecting Government as an Employer of Choice Public Administration 86

1089-1105

Vandenabeele Wouter 2008b Development of a Public Service Motivation Measurement

Scale Corroborating and Extending Perryrsquos Measurement Instrument International

Public Management Journal 11 143-167

Vandenabeele Wouter 2009 The Mediating Effect of Job Satisfaction and Organizational

Commitment on Self-Reported Performance More Robust Evidence of the PSM-

Performance Relationship International Review of Administrative Sciences 75 11-

34

Wilcox James B Roy D Howell and Einar Breivik 2008 Questions about Formative

Measurement Journal of Business Research 61 1219-1228

Wise Lois Recascino 2000 The Public Service Culture In Public Administration Concepts

- 33 -

and Cases 7th ed edited by Richard J Stillman II 342-353 Boston Houghton

Mifflin

Wright Bradley E 2008 Methodological Challenges associated with Public Service

Motivation Research In Motivation in Public Management edited by James L

Perry and Annie Hondeghem 80-98 Oxford Oxford University Press

Wright Bradley E and Sanjay K Pandey 2008 Public Service Motivation and the

Assumption of Person-Organization Fit Testing the Mediating Effect of Value

Congruence Administration amp Society 40 502-521

- 34 -

Table 1

Possible Indicators of PSM and Descriptive Statistics (n = 2868)

Dimensions and items Mean Standard

Deviation

Item-Total

Correlation

Attraction to Public Participation (APP)

APP1 I am interested in helping to improve public service

APP2 I am satisfied when I see people benefiting from the public programs I was

involved in

APP3 I like to discuss topics regarding public programs and policies with others

APP4 I believe that public sector activities contribute to our general welfare

APP5 I admire people who initiate or are involved in activities to aid my

community

APP6 Contributing to public programs and policies helps me realize myself

APP7 It is important to contribute to activities that tackle social problems

412

441

355

393

423

351

409

696

662

917

753

672

881

734

524

421

482

412

493

478

552

Commitment to Public Values (CPV)

Sub-dimension for Public Interests (CPV1)

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

CPI3 I would prefer seeing public officials do what is best for the whole

community even if it harmed my interests

CPI4 Serving the public interest is more important than helping a single individual

Sub-dimension for Public Values (CPV2)

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of public

services

CPV3 It is fundamental that public services respond to the needs of the citizens

CPV4 Decisions regarding public services should be democratic despite the time

and effort it takes

CPV5 Everybody is entitled to a good service even if it costs a lot of money

CPV6 It is fundamental that the interests of future generations are taken into

account when developing public policies

CPV7 To act ethically is essential for public servants

CPV8 I believe that public employees must always be aware of the legitimacy of

their activities

CPV9 I personally identify with the aim of protecting individual liberties and rights

418

409

348

337

428

416

433

394

367

426

441

434

414

695

662

899

995

730

704

710

852

932

689

681

692

763

525

596

480

251

484

437

356

325

376

439

462

402

392

Compassion (COM) COM1 It is difficult for me to contain my feelings when I see people in distress

COM2 I feel sympathetic to the plight of the underprivileged

COM3 I empathize with other people who face difficulties

COM4 I have little compassion for people in need who are unwilling to take the

first step to help themselves

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

354

387

388

299

412

391

951

825

707

1083

755

765

399

511

389

011

487

502

Self-Sacrifice (SS)

SS1 Making a difference to society means more to me than personal achievements

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS5 People should give back to society more than they get from it

SS6 Serving other citizens would give me a good feeling even if no one paid me for

it

SS7 I would agree to a good plan to make a better life for the poor even if it costs

me money

357

329

339

302

345

363

350

904

872

949

982

888

1049

895

508

584

454

531

412

368

546

- 35 -

Table 2

Background of Respondents (n = 2868)

Variables Characteristics Respondents

Sex

Male

Female

NA

1209

1622

37

422

566

13

Length of service

(years)

0 ~ 10

10 ~ 20

20 ~ 30

30+

NA

1126

723

484

491

44

393

252

169

171

15

Organizational status

―Do you supervise employees

No

Yes

NA

1472

1076

320

513

375

112

Country

Australia

Belgium

China

Denmark

France

Italy

Korea

Lithuania

The Netherlands

Switzerland

United Kingdom

USA

249

214

230

249

266

162

253

236

249

250

260

250

87

75

80

87

93

56

88

82

87

87

91

87 Note NA = no answer

- 36 -

Table 3

Standardized factor loadings and correlations for the 31-item PSM index (n = 2868)

Dimensions and items SFL (MLE) SFL (D-WLS) Alpha

APP

APP1

APP2

APP4

APP5

APP6

APP7

597

486

455

558

490

600

657

552

514

625

549

665

699

CPV1

CPI1

CPI2

CPI3

CPI4

588

660

488

289

653

732

535

294

542

CPV2

CPV1

CPV2

CPV3

CPV4

CPV5

CPV6

CPV7

CPV8

CPV9

611

546

436

393

418

534

598

496

460

690

631

505

457

486

601

669

573

530

745

COM

COM1

COM2

COM3

COM5

COM6

496

569

498

597

607

516

651

538

670

684

687

SS

SS1

SS2

SS3

SS4

SS5

SS6

SS7

567

736

558

728

467

409

663

660

762

591

695

515

494

709

782

Measures of fit (MLE) χsup2 (df = 424) = 43699 p lt 001 CFI = 835 GFI = 899

RMSEA = 057 [055 059]

Measure of fit (D-WLS) χsup2 (df = 424) = 39247 p lt 001 CFI = 971 GFI = 976

RMSEA = 054 [052 055]

Inter-factor correlations MLE (D-WLS)

1 2 3 4

1 APS

2 CPV1 996 (988)

3 CPV2 902 (890) 798 (798)

4 COM 787 (783) 758 (762) 779 (783)

5 SS 700 (742) 800 (859) 509 (537) 720 (729)

Note SFL = standardized factor loading All factor loadings and correlations are significant at p lt 001

- 37 -

Table 4

Standardized factor loadings and correlations for the 16-item PSM index (n = 2868)

Dimensions and items SFL

(MLE)

SFL

(D-WLS)

APS

APP5 I admire people who initiate or are involved in activities to aid my

community

APP7 It is important to contribute to activities that tackle social problems

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

611

592

637

666

665

669

690

751

CPV

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of

public services

CPV6 It is fundamental that the interests of future generations are taken

into account when developing public policies

CPV7 To act ethically is essential for public servants

612

546

551

590

698

631

618

658

COM

COM1 It is difficult for me to contain my feelings when I see people in

distress

COM2 I feel sympathetic to the plight of the underprivileged

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

490

546

596

604

515

608

669

679

SS

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS7 I would agree to a good plan to make a better life for the poor even if

it costs me money

763

571

766

678

807

629

747

755

Measures of fit (MLE)

χsup2 (df = 98) = 8490 p lt 001

CFI = 939 GFI = 965

RMSEA = 052 [049 055]

Measure of fit (D-WLS)

χsup2 (df = 98) = 6069 p lt 001

CFI = 988 GFI = 992

RMSEA = 043 [039 046]

Alpha CR (D-WLS) Inter-factor correlations MLE (D-WLS)

1 APS 717 788 1 APS 2 CPV 3 COM

2 CPV 663 747 848 (859)

3 COM 645 713 781 (791) 789 (795)

4 SS 784 826 565 (596) 462 (505) 715 (726)

Note SFL = standardized factor loading Alpha = Cronbachrsquos alpha CR = composite reliability

All factor loadings and correlations are significant at p lt 001

- 38 -

Table 5

Results of CFA in the individual countries Standardized factor loadings and measures of fit

Factors

amp Items

Australia Belgium China Denmark France Italy

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

672

756

799

785

804

943

710

703

597

790

799

730

584

675

685

681

731

742

578

748

674

643

765

739

645

704

678

708

812

779

632

604

765

766

672

853

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

552

593

560

699

727

632

624

605

560

713

709

645

702

602

616

724

686

735

567

560

715

654

605

811

608

658

587

696

755

698

673

620

523

782

697

654

COM

COM1

COM2

COM3

COM5

COM6

623

584

662

688

701

739

732

608

475

808

666

581

578

616

605

649

670

673

699

751

737

679

816

776

568

546

801

626

673

850

637

546

741

715

601

822

SS

SS2

SS3

SS4

SS7

654

794

806

892

767

813

727

693

799

793

725

838

857

820

773

896

855

842

806

749

792

829

753

884

847

731

753

889

816

753

815

692

664

795

782

734

Inter-factor

correlations

093

~

752

163

~

810

320

~

773

323

~

825

452

~

827

497

~

855

323

~

850

321

~

847

498

~

899

536

~

902

425

~

829

476

~

865

Sample size 249 214 230 249 266 162

Measures of Fit

χsup2 638 444 881 732 765 554 817 625 735 565 750 564

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 980 100 939 987 969 997 967 994 975 997 949 995

GFI 961 986 941 986 948 991 949 992 955 992 926 986

RMSEA 036 00 063 050 051 026 053 035 045 026 059 033

- 39 -

Table 5 (Continued)

Factors

amp Items

Korea Lithuania The Netherlands Switzerland UK USA

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

619

701

667

706

781

717

666

512

682

751

605

779

557

629

630

620

719

729

535

621

702

643

694

735

655

725

738

710

804

822

535

621

702

643

694

735

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

594

607

722

722

644

722

674

535

533

790

589

711

568

554

538

655

628

545

627

580

615

711

666

704

800

512

555

873

581

646

627

580

615

711

666

704

COM

COM1

COM2

COM3

COM5

COM6

694

677

550

763

736

627

557

639

428

550

696

534

737

623

508

766

616

608

659

592

683

703

661

769

682

583

729

739

644

803

659

592

683

703

661

769

SS

SS2

SS3

SS4

SS7

776

752

679

779

770

811

707

721

635

725

734

706

814

656

743

820

767

764

745

632

689

788

662

744

671

818

832

830

864

802

745

632

689

788

662

744

Inter-factor

correlations

432

~

726

466

~

746

376

~

788

406

~

902

377

~

852

455

~

836

637

~

818

660

~

828

317

~

852

360

~

852

637

~

818

660

~

828

Sample size 253 236 249 250 260 250

Measures of Fit

χsup2 896 573 612 462 773 616 815 715 769 542 815 715

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 951 996 975 100 958 993 957 991 972 998 957 991

GFI 942 988 960 989 951 987 948 990 955 990 948 990

RMSEA 059 028 034 00 050 034 053 044 048 022 053 044

Note All factor loadings and correlations are significant at p lt 001

- 40 -

Appendix 1 Dimensions and possible items of PSM (the first round)

1 Attraction to Public Participation

(1) I am interested in developing public programs helping my country or community (Kim 2009a)

(2) Sharing my views on public policies with others is attractive to me (Kim 2009a)

(3) I am satisfied when seeing people get benefits from the public programs I was involved in (Kim 2009a)

(4) I like to discuss political subjects with others (Giauque et al 2011)

(5) I think the governmental activities contribute to our welfare

(6) I like to initiate actions to help out my community

(7) I do not care much about politicians (Perry 1996)

2 Commitment to Public Values

Possible items for public interest

(1A) Meaningful public service is very important to me (Perry 1996)

(2A) It is important for me to contribute for the common good (Giauque et al 2011)

(3A) I recognize myself with the mission of protection of liberties and rights endorsed by the nation-state

(Giauque et al 2011)

(4A) I would prefer seeing public officials do what is best for the whole community even if it harmed my

interests (Perry 1996)

(5A) To me serving the public interest is more important than helping other people (Vandenabeele 2008b)

Possible items for public values

(1B) I recognize myself with the promotion of the equality of chances endorsed by the nation-state (Giauque et

al 2011)

(2B) It is fundamental that the public service provided to the citizen is regular and continuous (Giauque et al

2011)

(3B) It is fundamental that public service innovates to respond to the new needs of the citizensclients (Giauque

et al 2011)

(4B) The dignity and well-being of all should be the most important concerns in any society (Steenbergen

1996)

(5B) It is important that public servants account for all the costsexpenses they make (Vandenabeele 2008b)

(6B) We have to do everything in our power to pursue the goal of democracy

3 Compassion

(1) It is difficult for me to contain my feelings when I see people in distress (Perry 1996)

(2) I feel sympathetic to the plight of the underprivileged (Kim 2009a)

(3) I am often reminded by daily events how dependent we are on one another (Perry 1996)

(4) I feel compassion for others in the difficulties they are facing (Steffen and Masters 2005)

(5) It is not really my problem if others are in trouble and need help (Beutel and Marini 1995)

(6) I find it hard to be sympathetic toward starving people in foreign lands when there is so much trouble in our

own country (Beutel and Marini 1995)

(7) I get very upset when I see other people treated unfairly (Beutel and Marini 1995)

(8) I would agree to a good plan to make a better life for the poor even if it cost me money (Beutel and Marini

1995)

(9) To me patriotism includes seeing the welfare of others (Perry 1996)

(10) I care very much about other people

4 Self-Sacrifice

(1) Making a difference in society means more to me than personal achievements (Perry 1996)

(2) I am prepared to make enormous sacrifices for the good of society (Perry 1996)

(3) I believe in putting duty before self (Perry 1996)

(4) I am one of those rare people who would risk personal loss to help someone else (Perry 1996)

(5) I think people should give back to society more than they get from it (Perry 1996)

(6) Serving other citizens would give me a good feeling even if no one paid me for it (Perry 1996)

(7) Much of what I do is for a cause bigger than myself (Perry 1996)

- 41 -

Appendix 2 Newly suggested items through the first round

1 Items for the Attraction to Public Participation Dimension

(1) In the public domain every person should be counted as equal

(2) I enjoy to think and argue in a political way

(3) I am satisfied when seeing my ideas being integrated in public policies

(4) I can easily relate to the way of thinking and arguing of politicians

(5) Working in the public sector I can better realize myself

2 Items for the Public Values Sub-Dimension

(1) Public service should give value for money

(2) Cost considerations should play an important role in decisions about public service provision

(3) It is fundamental that the interests of future generations are taken into account when developing public

policies

(4) To act with correct ethics is for public servants as important as their professional competence

(5) Primarily public servants should be accountable to the public and not to their superiors

3 Items for the Self-Sacrifice Dimension

(1) I would agree to a good plan to make a better life for the poor even if it costs me money

- 42 -

Appendix 3 An example of a summary of comments and suggestions (the second round)

On the item ―I am interested in developing public programs helping my country or community

A It is a good sentence but is it possible for everyone to ―develop public programs Even public officials are

limited to do this

B OK

C or the special group

D Double barrel and difficult to translate to my country ―I am interested in helping to improve public services

E Good but I feel that wording requires revision If we draw sample from local government then country is

redundant ―I am interested in developing public policy programs that aim in helping my community

F 4 (General appreciation from1 to 4) Not clear whether it refers to ones work or to ones concern as a citizen

―(In my work) I am interested in contributing to the development of public etchellip

G Instead of development I would use implementing

H Delete public which is redundant because of the characteristic helping my country or community but add

or implementing ―I am interested in developing or implementing programs helping my country or

community

I Referring to dimension Wording Similar to item 6

Note Names have been changed to alphabets in order to ensure the anonymity of participants

- 43 -

Appendix 4 Dimensions and Items of Public Service Motivation for Future Research

Dimensions and items

Attraction to Public Service (APS)

1 I am interested in helping to improve public service

2 I admire people who initiate or are involved in activities to aid my community

3 It is important to contribute to activities that tackle social problems

4 Meaningful public service is very important to me

5 It is important for me to contribute to the common good

Commitment to Public Values (CPV)

1 I think equal opportunities for citizens are very important

2 It is important that citizens can rely on the continuous provision of public services

3 It is fundamental that the interests of future generations are taken into account when

developing public policies

4 To act ethically is essential for public servants

5 I believe that public employees must always be aware of the legitimacy of their activities

6 I personally identify with the aim of protecting individual liberties and rights

Compassion (COM)

1 It is difficult for me to contain my feelings when I see people in distress

2 I feel sympathetic to the plight of the underprivileged

3 I empathize with other people who face difficulties

4 I get very upset when I see other people being treated unfairly

5 Considering the welfare of others is very important

Self-Sacrifice (SS)

1 I am prepared to make sacrifices for the good of society

2 I believe in putting civic duty before self

3 I am willing to risk personal loss to help society

4 I would agree to a good plan to make a better life for the poor even if it costs me money

- 22 -

composite in the form of indexes needs to be tested to know the relative weights (γ-

parameters) of the dimensions of PSM In PLS-PM a second-order formative factor is

directly measured by observed variables for all the first order factors While this approach

repeats the number of manifest variables used the model can be identified without adding

outcome constructs that would be considered inappropriate or irrelevant from a conceptual

standpoint (Chin 2010 Edwards 2010) The test result of PLS-PM using Smart PLS 20

(Ringle Wende and Will 2005) produced the function PSM = 31APS + 36CPV + 28COM

+ 31SS However one consideration is whether and how construct error is modeled for the

formative construct (Diamantopoulos 2006) When using PLS-PM the researcher is making

the assumption that there is no error at the construct level as all constructs in PLS-PM are

modeled without error When using PLS-PM the absence of an error term will tend to

increase the weights therefore inflating their importance (Bollen and Lennox 1991) The

primary statistic for assessing a formative indicator is its weight the partialized effect of the

indicator on its intended construct controlling for the effects of all other indicators of that

construct (Cenfetelli and Bassellier 2009) A formative measure is essentially a multiple

regression with the construct representing the dependent variable and the indicators as the

predictors (Diamantopoulos and Winklhofer 2001) PSM may be regarded as a weighted

linear composite of its dimensions The weights used to create the composites are different

across CE-SEM and PLS-PM The causal dimensional weights generated in CB-SEM are

estimated in a model in which a construct-level error term is specified (eg PSM = γ1APM +

γ2CPI + γ3COM + γ4SS + ζ) whereas the weights in PLS-PM are not (eg PSM = γ1APM +

γ2CPI + γ3COM + γ4SS) This leads to two different conceptualizations of the PSM construct

one in which the causal indicators do not capture the construct in its entirety and another in

which they do (Hardin et al 2010) These differences need to be discussed in future research

- 23 -

in this area

One practical suggestion to measure PSM is to apply unit weights to its dimensions if

we assume the effects of the four dimensions are equivalent A literature review indicates that

unit weights have substantial predictive validity when compared with regression weights The

literature shows that the empirical correlations between scores generated from differential and

unit-weighting approaches are quite high In creating composite scores unit weights can be

an appropriate approach for weighting under many circumstances (Bobko Roth and Buster

2007) Although the uses of unit-weight composites will likely result in the explanation of

less variance than empirical estimations of dimensional weights they will remain consistently

defined across studies and hence suitable for theory testing Alternatively more thorough

discussions on the dimensions of PSM may suggest fixed-weight composites (Hardin et al

2010) A previous study (Kim 2010 22) also suggested ―Overall indexing PSM by

measuring all the dimensions of PSM and simply summing the averages of each dimension

can make it easier to measure PSM in practice as PSM = APS + CPV + COM + SS

An advantage of regarding PSM as a second-order formative construct is that the

dimensions can be analyzed both individually and collectively This is important as the

dimensions are expected to have different antecedents and consequences For instance Perry

(1997) found some significant differences in the influence of independent variables on the

four dimensions of PSM Taylor (2007) confirmed that certain dimensions are more important

than others in influencing work outcomes Moynihan and Pandey (2007a) reported that red

tape is negatively related to one dimension but reform orientation and hierarchical authority

is positively related to another dimension A recent study found that the SS dimension has a

positive relationship with occupational intention but the other dimensions do not (Liu et al

2011) Thus it is valuable to analyze the causes and effects of individual dimensions of PSM

- 24 -

as well as of the PSM construct

Conclusion

The purpose of this study is to develop a measurement instrument for PSM for cross-national

research and comparison On the basis of the sharpened construct of PSM (Kim and

Vandenabeele 2010) using data from an international survey conducted in 12 countries we

found that the four-factor structure of PSM is generalizable across country cases This study

provides the measurement index of PSM with the four dimensions of APS CPV COM and

SS The APS dimension measures the instrumental motives of PSM the CPV dimension

determines the value-based motives of PSM while the COM and SS dimensions gauge the

identification motives and self-sacrifice respectively When all four dimensions of PSM are

used in a survey cross-national equivalence of survey items can be assured The items for

measuring PSM in future research are suggested in Appendix 4

This study has several limitations First looking for samples which were comparable

to the largest possible extent we opted for surveys in local government The results found

and discussed in this study are based on PSM among local government employees The

results might be different when surveying central government employees or other specific

groups in the public sector Thus the potential impact of the sample bias needs to be

considered in understanding this study It also means that the survey items for measuring

PSM should be examined with different samples in various areas and at different government

levels

Second the test results showed that there is a high degree of correlation among the

dimension such as APS CPV and COM Future research should not only investigate whether

- 25 -

the items reflect the core content of the dimensions but also continue to develop more

appropriate items for better discriminant validity of each dimension Third the content of

APP dimension was assumed to be distinct from the CPV dimension that focuses on personal

disposition to pursue public interest and public values However the public interest sub-

dimension was separated from the public values sub-dimension in the CPV dimension and

combined with the APP dimension to form the APS dimension Further research should

analyze the CPV dimension and investigate the appropriateness of the APS dimension

In order to facilitate research internationally the PSM construct needs to be

improved conceptually and operationally Kim and Vandenabeele (2010) have dealt with the

conceptual composition and operational dimension of PSM as well as the measurement

model of PSM The next step is to develop an internationally robust measurement instrument

of PSM This study proposes that scholars use all four of the dimensions in measuring the

conceptual components of PSM such as self-sacrifice instrumental value-based and affective

motives because they collectively represent PSM It provides a set of appropriate items for

measuring PSM and confirms the dimensional structure of PSM The survey results suggest

that people in different cultures understand PSM in the same fashion We know that this study

is simply a step along the way to developing a more robust measure but we hope it

contributes to enhanced cross-national research and comparison along with the generation of

cumulative knowledge internationally

- 26 -

REFERENCES

Anderson James C and David D Gerbing 1988 Structural Equation Modeling in Practice

A Review and Recommended Two-Step Approach Psychological Bulletin 103

411-423

Arbuckle James L 2009 Amos 180 Userrsquos Guide Crawfordville FL Amos Development

Co

Beck Joslashrgensen Torben and Barry Bozeman 2007 Public Values An Inventory

Administration amp Society 39(3) 354-381

Beutel Ann M and Margaret Mooney Marini 1995 Gender and values American

Sociological Review 60 436 - 448

Bobko Philip Philip L Roth and Maury A Buster 2007 The Usefulness of Unit Weights in

Creating Composite Scores A Literature Review Application to Content Validity and

Meta-Analysis Organizational Research Methods 10(4) 689-709

Bollen Kenneth A 1989 Structural Equations with Latent Variables New York Wiley

Bollen K A and R Lennox 1991 Conventional Wisdom on Measurement A Structural

Equation Perspective Psychological Bulletin 110(2) 305-314

Byrne Barbara M 2010 Structural Equation Modeling with Amos Basic Concepts

Applications and Programming 2nd

ed New York Routledge

Camilleri Emanuel 2006 Towards Developing an Organizational Commitment ― Public

Service Motivation Model for Maltese Public Service Employees Public Policy

and Administration 21 63-83

Castaing Seacutebastien 2006 The Effects of Psychological Contract Fulfillment and Public

Service Motivation on Organizational Commitment in the French Civil Service

Public Policy and Administration 21 84-98

Cenfetelli Ronald T and Geneviegraveve Bassellier 2009 Interpretation of Formative

Measurement in Information Systems Research MIS Quarterly 33(4) 689-707

Charles Michael B W Martin de Jong and Neal Ryan 2011 Public Values in Western

- 27 -

Europe A Temporal Perspective American Review of Public Administration 41

75-91

Chin Wynne W 2010 How to Write Up and Report PLS Analysis In Handbook of Partical

Least Squares edited by V Esposito Vinzi et al 655-690 Berlin Springer-Verlag

Coursey David H and Sanjay K Pandey 2007a Public Service Motivation Measurement

Testing an Abridged Version of Perryrsquos Proposed Scale Administration amp Society

39 547-568

Coursey David H and Sanjay K Pandey 2007b Content Domain Measurement and

Validity of the Red Tape Concept American Review of Public Administration 37

342-361

Coursey David H James L Perry Jeffrey L Brudney and Laura Littlepage 2008

Psychometric Verification of Perryrsquos Public Service Motivation Instrument Results

for Volunteer Exemplars Review of Public Personnel Administration 28 79-90

Davis Paul and Karen West 2009 What Do Public Values Mean for Public Action Putting

Public Values in Their Plural Place American Review of Public Administration 39

602-618

DeHart-Davis Leisha Justin Marlowe and Sanjay K Pandey 2006 Gender Dimensions of

Public Service Motivation Public Administration Review 66 873-887

Diamantopoulos Adamantios 2006 The error term in formative measurement models

Interpretation and modeling implications Journal of Modeling in Management 1(1)

7-17

Diamantopoulos Adamantios and Judy A Siguaw 2006 Formative versus Reflective

Indicators in Organizational Measure Development A Comparison and Empirical

Illustration British Journal of Management 17 263-282

Diamantopoulos Adamantios and Heidi M Winklhofer 2001 Index Construction with

Formative Indicators An Alternative to Scale Development Journal of Marketing

Research 38 269-277

Edwards Jeffrey R 2010 The Fallacy of Formative Measurement Organizational Research

- 28 -

Methods Online First published on August 19 2010 as

doi1011771094428110378369

Edwards Jeffrey R and Richard P Bagozzi 2000 On the Nature and Direction of

Relationships between Constructs and Measures Psychological Methods 5 155-

174

Flora David B and Patrick J Curran 2004 An Empirical Evaluation of Alternative

Methods of Estimation for Confirmatory Factor Analysis with Ordinal Data

Psychological Methods 9 466-491

Fornell Claes and Fred L Bookstein 1982 Two structural equations models LISREL and

PLS applied to customer exit-voice theory Journal of Marketing Research 19 440-

452

Giauque David Adrian Ritz Freacutedeacuteric Varone Simon Anderfuhren-Biget and Christian

Waldner 2011 Putting Public Service Motivation into Context A Balance between

Universalism and Particularism Forthcoming in International Review of

Administrative Sciences 77(2)

Goumltz Oliver Kerstin Liehr-Gobbers and Manfred Krafft 2010 Evaluation of Structural

Equation Models Using the Partial Least Squares (PLS) Approach In V Esposito

Vinzi et al (eds) Handbook of Partial Least Squares 691-711 DOI 101007978-3-

540-32827-8_30 Springer-Verlag Berlin Heidelberg

Guinot Christiane Julie Latreille and Michel Tenenhaus 2001 PLS Path Modelling and

Multiple Table Analysis Application to the Cosmetic Habits of Women in Ile-de-

France Chemometrics and Intelligent Laboratory Systems 58 247-259

Hair Joseph F Jr William C Black Barry J Babin and Rolph E Anderson 2010

Multivariate Data Analysis 7th

ed Upper Saddle River NJ Pearson Education

Hardin Andrew M Jerry Cha-Jan Chang Mark A Fuller and Gholamreza Torkzadeh 2010

Formative Measurement and Academic Research In Search of Measurement

Theory Educational and Psychological Measurement Online First published on

June 11 2010 as doi101177001316440370208

- 29 -

Howell Roy D Einar Breivik and James B Wilcox 2007a Reconsidering Formative

Measurement Psychological Methods 12 205-218

Howell Roy D Einar Breivik and James B Wilcox 2007b Is Formative Measurement

Really Measurement Reply to Bollen (2007) and Bagozzi (2007) Psychological

Methods 12 238-245

Jarvis Cheryl Burke Scott B MacKenzie and Philip M Podsakoff 2003 A Critical Review

of Construct Indicators and Measurement Model Misspecification in Marketing and

Consumer Research Journal of Consumer Research 30 199-218

Joumlreskog Karl G and Dag Soumlrbom 2005 LISREL 872 (Computer Software) Lincolnwood

IL Scientific Software International

Kim Gimun Bongsik Shin and Varun Grover 2010 Investigating Two Contradictory Views

of Formative Measurement in Information Systems Research MIS Quarterly 34

345-365

Kim Sangmook 2009a Revising Perryrsquos Measurement Scale of Public Service Motivation

American Review of Public Administration 39 149-163

Kim Sangmook 2009b Testing the Structure of Public Service Motivation in Korea A

Research Note Journal of Public Administration Research and Theory 19 839-

851

Kim Sangmook 2010 Testing a Revised Measurement of Public Service Motivation

Reflective versus Formative Specification Journal of Public Administration

Research and Theory Doi101093jopartmuq048

Kim Sangmook and Wouter Vandenabeele 2010 A strategy for building public service

motivation research internationally Public Administration Review 70 701-709

Kline Rex B 2005 Principles and Practice of Structural Equation Modeling 2nd

ed New

York Guilford Press

Law Kenneth and Chi-Sum Wong 1999 Multidimensional Constructs in Structural

Equation Analysis An Illustration using the Job Perception and Job Satisfaction

Constructs Journal of Management 25 143-160

- 30 -

Leisink Peter and Bram Steijn 2009 Public Service Motivation and Job Performance of

Public Sector Employees in the Netherlands International Review of

Administrative Sciences 75 35-52

Liu Bangcheng Chun Hui Jin Hu Wensheng Yang and Xinli Yu 2011 How well can

Public Service Motivation Connect with Occupational Intention International

Review of Administrative Sciences 77 191-211

Liu Bangcheng Ningyu Tang and Xiaomei Zhu 2008 Public Service Motivation and Job

Satisfaction in China An Investigation of Generalisability and Instrumentality

International Journal of Manpower 29 684-699

MacCallum Robert C and Michael W Browne 1993 The Use of Causal Indicators in

Covariance Structure Models Some Practical Issues Psychological Bulletin 114

533-541

MacKenzie Scott B 2003 The dangers of poor construct conceptualization Journal of the

Academy of Marketing Science 31(3) 323-326

MacKenzie Scott B Philip M Podsakoff and Cheryl Burke Jarvis 2005 The Problem of

Measurement Model Misspecification in Behavioral and Organizational Research

and Some Recommended Solutions Journal of Applied Psychology 90 710-730

Martinsen Dorte Sindbjerg and Torben Beck Joslashrgensen 2010 Accountability as a

Differentiated Value in Supranational Governance American Review of Public

Administration 40 742-760

Moynihan Donald P and Sanjay K Pandey 2007a The Role of Organizations in Fostering

Public Service Motivation Public Administration Review 67 40-53

Moynihan Donald P and Sanjay K Pandey 2007b Finding Workable Levers over Work

Motivation Comparing Job Satisfaction Job Involvement and Organizational

Commitment Administration amp Society 39 803-832

Pandey Sanjay K and Edmund C Stazyk 2008 Antecedents and Correlates of Public

Service Motivation In Motivation in Public Management edited by James L Perry

and Annie Hondeghem 101-117 Oxford Oxford University Press

- 31 -

Perry James L 1996 Measuring Public Service Motivation An Assessment of Construct

Reliability and Validity Journal of Public Administration Research and Theory 6

5-22

Perry James L 1997 Antecedents of Public Service Motivation Journal of Public

Administration Research and Theory 7 181-197

Perry James L 2000 Bringing Society in Toward a Theory of Public Service Motivation

Journal of Public Administration Research and Theory 10 471-488

Perry James L and Annie Hondeghem 2008a Motivation in Public Management Oxford

Oxford University Press

Perry James L and Annie Hondeghem 2008b Directions for Future Theory and Research

In Motivation in Public Management edited by James L Perry and Annie

Hondeghem 294-313 Oxford Oxford University Press

Perry James L Annie Hondeghem and Lois Recascino Wise 2010 Revisiting the

Motivational Bases of Public Service Twenty Years of Research and an Agenda for

the Future Public Administration Review 70(5) 681-690

Perry James L and Wouter Vandenabeele 2008 Behavioral dynamics Institutions

Identities and Self-Regulation In Motivation in Public Management edited by

James L Perry and Annie Hondeghem 56-79 Oxford Oxford University Press

Perry James L and Lois Recascino Wise 1990 The Motivational Bases of Public Service

Public Administration Review 50 367-373

Rainey Hal G 1982 Reward Preference among Public and Private Managers In Search of

the Service Ethic American Review of Public Administration 16 288-302

Ringle Christian M Oliver Goumltz Martin Wetzels and Bradley Wilson 2009 On the Use of

Formative Measurement Specifications in Structural Equation Modeling A Monte

Carlo Simulation Study to Compare Covariance-Based and Partial Least Squares

Model Estimation Methodologies Faculty of Economics and Business Administration

Maastricht Univeristy

Ringle C M S Wende and S Will 2005 Smart PLS 20 (M3) Beta

- 32 -

httpwwwsmartplsde

Ritz Adrian 2011 Attraction to Public Policy Making A Qualitative Inquiry into

Improvements in PSM Measurement Forthcoming in Public Administration

Spector Paul E 1992 Summated Rating Scales Construction An Introduction Newbury

Park CA Sage

Steenbergen Marco R 1996 Compassion and American public opinion An analysis of the

NES Humanitarianism Scale Report to the Board of Overseers of the National

Election Studies

Steffen Patrick R and Kevin S Masters 2005 Does compassion mediate the intrinsic

religion-health relationship Annals of Behavioral Medicine 30 217-223

Taylor Jeannette 2007 The Impact of Public Service Motives on Work Outcomes in

Australia A Comparative Multi-Dimensional Analysis Public Administration 85

931-959

Vandenabeele Wouter 2007 Towards a Theory of Public Service Motivation An

Institutional Approach Public Management Review 9 545-556

Vandenabeele Wouter 2008a Government Calling Public Service Motivation as an Element

in Selecting Government as an Employer of Choice Public Administration 86

1089-1105

Vandenabeele Wouter 2008b Development of a Public Service Motivation Measurement

Scale Corroborating and Extending Perryrsquos Measurement Instrument International

Public Management Journal 11 143-167

Vandenabeele Wouter 2009 The Mediating Effect of Job Satisfaction and Organizational

Commitment on Self-Reported Performance More Robust Evidence of the PSM-

Performance Relationship International Review of Administrative Sciences 75 11-

34

Wilcox James B Roy D Howell and Einar Breivik 2008 Questions about Formative

Measurement Journal of Business Research 61 1219-1228

Wise Lois Recascino 2000 The Public Service Culture In Public Administration Concepts

- 33 -

and Cases 7th ed edited by Richard J Stillman II 342-353 Boston Houghton

Mifflin

Wright Bradley E 2008 Methodological Challenges associated with Public Service

Motivation Research In Motivation in Public Management edited by James L

Perry and Annie Hondeghem 80-98 Oxford Oxford University Press

Wright Bradley E and Sanjay K Pandey 2008 Public Service Motivation and the

Assumption of Person-Organization Fit Testing the Mediating Effect of Value

Congruence Administration amp Society 40 502-521

- 34 -

Table 1

Possible Indicators of PSM and Descriptive Statistics (n = 2868)

Dimensions and items Mean Standard

Deviation

Item-Total

Correlation

Attraction to Public Participation (APP)

APP1 I am interested in helping to improve public service

APP2 I am satisfied when I see people benefiting from the public programs I was

involved in

APP3 I like to discuss topics regarding public programs and policies with others

APP4 I believe that public sector activities contribute to our general welfare

APP5 I admire people who initiate or are involved in activities to aid my

community

APP6 Contributing to public programs and policies helps me realize myself

APP7 It is important to contribute to activities that tackle social problems

412

441

355

393

423

351

409

696

662

917

753

672

881

734

524

421

482

412

493

478

552

Commitment to Public Values (CPV)

Sub-dimension for Public Interests (CPV1)

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

CPI3 I would prefer seeing public officials do what is best for the whole

community even if it harmed my interests

CPI4 Serving the public interest is more important than helping a single individual

Sub-dimension for Public Values (CPV2)

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of public

services

CPV3 It is fundamental that public services respond to the needs of the citizens

CPV4 Decisions regarding public services should be democratic despite the time

and effort it takes

CPV5 Everybody is entitled to a good service even if it costs a lot of money

CPV6 It is fundamental that the interests of future generations are taken into

account when developing public policies

CPV7 To act ethically is essential for public servants

CPV8 I believe that public employees must always be aware of the legitimacy of

their activities

CPV9 I personally identify with the aim of protecting individual liberties and rights

418

409

348

337

428

416

433

394

367

426

441

434

414

695

662

899

995

730

704

710

852

932

689

681

692

763

525

596

480

251

484

437

356

325

376

439

462

402

392

Compassion (COM) COM1 It is difficult for me to contain my feelings when I see people in distress

COM2 I feel sympathetic to the plight of the underprivileged

COM3 I empathize with other people who face difficulties

COM4 I have little compassion for people in need who are unwilling to take the

first step to help themselves

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

354

387

388

299

412

391

951

825

707

1083

755

765

399

511

389

011

487

502

Self-Sacrifice (SS)

SS1 Making a difference to society means more to me than personal achievements

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS5 People should give back to society more than they get from it

SS6 Serving other citizens would give me a good feeling even if no one paid me for

it

SS7 I would agree to a good plan to make a better life for the poor even if it costs

me money

357

329

339

302

345

363

350

904

872

949

982

888

1049

895

508

584

454

531

412

368

546

- 35 -

Table 2

Background of Respondents (n = 2868)

Variables Characteristics Respondents

Sex

Male

Female

NA

1209

1622

37

422

566

13

Length of service

(years)

0 ~ 10

10 ~ 20

20 ~ 30

30+

NA

1126

723

484

491

44

393

252

169

171

15

Organizational status

―Do you supervise employees

No

Yes

NA

1472

1076

320

513

375

112

Country

Australia

Belgium

China

Denmark

France

Italy

Korea

Lithuania

The Netherlands

Switzerland

United Kingdom

USA

249

214

230

249

266

162

253

236

249

250

260

250

87

75

80

87

93

56

88

82

87

87

91

87 Note NA = no answer

- 36 -

Table 3

Standardized factor loadings and correlations for the 31-item PSM index (n = 2868)

Dimensions and items SFL (MLE) SFL (D-WLS) Alpha

APP

APP1

APP2

APP4

APP5

APP6

APP7

597

486

455

558

490

600

657

552

514

625

549

665

699

CPV1

CPI1

CPI2

CPI3

CPI4

588

660

488

289

653

732

535

294

542

CPV2

CPV1

CPV2

CPV3

CPV4

CPV5

CPV6

CPV7

CPV8

CPV9

611

546

436

393

418

534

598

496

460

690

631

505

457

486

601

669

573

530

745

COM

COM1

COM2

COM3

COM5

COM6

496

569

498

597

607

516

651

538

670

684

687

SS

SS1

SS2

SS3

SS4

SS5

SS6

SS7

567

736

558

728

467

409

663

660

762

591

695

515

494

709

782

Measures of fit (MLE) χsup2 (df = 424) = 43699 p lt 001 CFI = 835 GFI = 899

RMSEA = 057 [055 059]

Measure of fit (D-WLS) χsup2 (df = 424) = 39247 p lt 001 CFI = 971 GFI = 976

RMSEA = 054 [052 055]

Inter-factor correlations MLE (D-WLS)

1 2 3 4

1 APS

2 CPV1 996 (988)

3 CPV2 902 (890) 798 (798)

4 COM 787 (783) 758 (762) 779 (783)

5 SS 700 (742) 800 (859) 509 (537) 720 (729)

Note SFL = standardized factor loading All factor loadings and correlations are significant at p lt 001

- 37 -

Table 4

Standardized factor loadings and correlations for the 16-item PSM index (n = 2868)

Dimensions and items SFL

(MLE)

SFL

(D-WLS)

APS

APP5 I admire people who initiate or are involved in activities to aid my

community

APP7 It is important to contribute to activities that tackle social problems

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

611

592

637

666

665

669

690

751

CPV

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of

public services

CPV6 It is fundamental that the interests of future generations are taken

into account when developing public policies

CPV7 To act ethically is essential for public servants

612

546

551

590

698

631

618

658

COM

COM1 It is difficult for me to contain my feelings when I see people in

distress

COM2 I feel sympathetic to the plight of the underprivileged

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

490

546

596

604

515

608

669

679

SS

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS7 I would agree to a good plan to make a better life for the poor even if

it costs me money

763

571

766

678

807

629

747

755

Measures of fit (MLE)

χsup2 (df = 98) = 8490 p lt 001

CFI = 939 GFI = 965

RMSEA = 052 [049 055]

Measure of fit (D-WLS)

χsup2 (df = 98) = 6069 p lt 001

CFI = 988 GFI = 992

RMSEA = 043 [039 046]

Alpha CR (D-WLS) Inter-factor correlations MLE (D-WLS)

1 APS 717 788 1 APS 2 CPV 3 COM

2 CPV 663 747 848 (859)

3 COM 645 713 781 (791) 789 (795)

4 SS 784 826 565 (596) 462 (505) 715 (726)

Note SFL = standardized factor loading Alpha = Cronbachrsquos alpha CR = composite reliability

All factor loadings and correlations are significant at p lt 001

- 38 -

Table 5

Results of CFA in the individual countries Standardized factor loadings and measures of fit

Factors

amp Items

Australia Belgium China Denmark France Italy

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

672

756

799

785

804

943

710

703

597

790

799

730

584

675

685

681

731

742

578

748

674

643

765

739

645

704

678

708

812

779

632

604

765

766

672

853

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

552

593

560

699

727

632

624

605

560

713

709

645

702

602

616

724

686

735

567

560

715

654

605

811

608

658

587

696

755

698

673

620

523

782

697

654

COM

COM1

COM2

COM3

COM5

COM6

623

584

662

688

701

739

732

608

475

808

666

581

578

616

605

649

670

673

699

751

737

679

816

776

568

546

801

626

673

850

637

546

741

715

601

822

SS

SS2

SS3

SS4

SS7

654

794

806

892

767

813

727

693

799

793

725

838

857

820

773

896

855

842

806

749

792

829

753

884

847

731

753

889

816

753

815

692

664

795

782

734

Inter-factor

correlations

093

~

752

163

~

810

320

~

773

323

~

825

452

~

827

497

~

855

323

~

850

321

~

847

498

~

899

536

~

902

425

~

829

476

~

865

Sample size 249 214 230 249 266 162

Measures of Fit

χsup2 638 444 881 732 765 554 817 625 735 565 750 564

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 980 100 939 987 969 997 967 994 975 997 949 995

GFI 961 986 941 986 948 991 949 992 955 992 926 986

RMSEA 036 00 063 050 051 026 053 035 045 026 059 033

- 39 -

Table 5 (Continued)

Factors

amp Items

Korea Lithuania The Netherlands Switzerland UK USA

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

619

701

667

706

781

717

666

512

682

751

605

779

557

629

630

620

719

729

535

621

702

643

694

735

655

725

738

710

804

822

535

621

702

643

694

735

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

594

607

722

722

644

722

674

535

533

790

589

711

568

554

538

655

628

545

627

580

615

711

666

704

800

512

555

873

581

646

627

580

615

711

666

704

COM

COM1

COM2

COM3

COM5

COM6

694

677

550

763

736

627

557

639

428

550

696

534

737

623

508

766

616

608

659

592

683

703

661

769

682

583

729

739

644

803

659

592

683

703

661

769

SS

SS2

SS3

SS4

SS7

776

752

679

779

770

811

707

721

635

725

734

706

814

656

743

820

767

764

745

632

689

788

662

744

671

818

832

830

864

802

745

632

689

788

662

744

Inter-factor

correlations

432

~

726

466

~

746

376

~

788

406

~

902

377

~

852

455

~

836

637

~

818

660

~

828

317

~

852

360

~

852

637

~

818

660

~

828

Sample size 253 236 249 250 260 250

Measures of Fit

χsup2 896 573 612 462 773 616 815 715 769 542 815 715

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 951 996 975 100 958 993 957 991 972 998 957 991

GFI 942 988 960 989 951 987 948 990 955 990 948 990

RMSEA 059 028 034 00 050 034 053 044 048 022 053 044

Note All factor loadings and correlations are significant at p lt 001

- 40 -

Appendix 1 Dimensions and possible items of PSM (the first round)

1 Attraction to Public Participation

(1) I am interested in developing public programs helping my country or community (Kim 2009a)

(2) Sharing my views on public policies with others is attractive to me (Kim 2009a)

(3) I am satisfied when seeing people get benefits from the public programs I was involved in (Kim 2009a)

(4) I like to discuss political subjects with others (Giauque et al 2011)

(5) I think the governmental activities contribute to our welfare

(6) I like to initiate actions to help out my community

(7) I do not care much about politicians (Perry 1996)

2 Commitment to Public Values

Possible items for public interest

(1A) Meaningful public service is very important to me (Perry 1996)

(2A) It is important for me to contribute for the common good (Giauque et al 2011)

(3A) I recognize myself with the mission of protection of liberties and rights endorsed by the nation-state

(Giauque et al 2011)

(4A) I would prefer seeing public officials do what is best for the whole community even if it harmed my

interests (Perry 1996)

(5A) To me serving the public interest is more important than helping other people (Vandenabeele 2008b)

Possible items for public values

(1B) I recognize myself with the promotion of the equality of chances endorsed by the nation-state (Giauque et

al 2011)

(2B) It is fundamental that the public service provided to the citizen is regular and continuous (Giauque et al

2011)

(3B) It is fundamental that public service innovates to respond to the new needs of the citizensclients (Giauque

et al 2011)

(4B) The dignity and well-being of all should be the most important concerns in any society (Steenbergen

1996)

(5B) It is important that public servants account for all the costsexpenses they make (Vandenabeele 2008b)

(6B) We have to do everything in our power to pursue the goal of democracy

3 Compassion

(1) It is difficult for me to contain my feelings when I see people in distress (Perry 1996)

(2) I feel sympathetic to the plight of the underprivileged (Kim 2009a)

(3) I am often reminded by daily events how dependent we are on one another (Perry 1996)

(4) I feel compassion for others in the difficulties they are facing (Steffen and Masters 2005)

(5) It is not really my problem if others are in trouble and need help (Beutel and Marini 1995)

(6) I find it hard to be sympathetic toward starving people in foreign lands when there is so much trouble in our

own country (Beutel and Marini 1995)

(7) I get very upset when I see other people treated unfairly (Beutel and Marini 1995)

(8) I would agree to a good plan to make a better life for the poor even if it cost me money (Beutel and Marini

1995)

(9) To me patriotism includes seeing the welfare of others (Perry 1996)

(10) I care very much about other people

4 Self-Sacrifice

(1) Making a difference in society means more to me than personal achievements (Perry 1996)

(2) I am prepared to make enormous sacrifices for the good of society (Perry 1996)

(3) I believe in putting duty before self (Perry 1996)

(4) I am one of those rare people who would risk personal loss to help someone else (Perry 1996)

(5) I think people should give back to society more than they get from it (Perry 1996)

(6) Serving other citizens would give me a good feeling even if no one paid me for it (Perry 1996)

(7) Much of what I do is for a cause bigger than myself (Perry 1996)

- 41 -

Appendix 2 Newly suggested items through the first round

1 Items for the Attraction to Public Participation Dimension

(1) In the public domain every person should be counted as equal

(2) I enjoy to think and argue in a political way

(3) I am satisfied when seeing my ideas being integrated in public policies

(4) I can easily relate to the way of thinking and arguing of politicians

(5) Working in the public sector I can better realize myself

2 Items for the Public Values Sub-Dimension

(1) Public service should give value for money

(2) Cost considerations should play an important role in decisions about public service provision

(3) It is fundamental that the interests of future generations are taken into account when developing public

policies

(4) To act with correct ethics is for public servants as important as their professional competence

(5) Primarily public servants should be accountable to the public and not to their superiors

3 Items for the Self-Sacrifice Dimension

(1) I would agree to a good plan to make a better life for the poor even if it costs me money

- 42 -

Appendix 3 An example of a summary of comments and suggestions (the second round)

On the item ―I am interested in developing public programs helping my country or community

A It is a good sentence but is it possible for everyone to ―develop public programs Even public officials are

limited to do this

B OK

C or the special group

D Double barrel and difficult to translate to my country ―I am interested in helping to improve public services

E Good but I feel that wording requires revision If we draw sample from local government then country is

redundant ―I am interested in developing public policy programs that aim in helping my community

F 4 (General appreciation from1 to 4) Not clear whether it refers to ones work or to ones concern as a citizen

―(In my work) I am interested in contributing to the development of public etchellip

G Instead of development I would use implementing

H Delete public which is redundant because of the characteristic helping my country or community but add

or implementing ―I am interested in developing or implementing programs helping my country or

community

I Referring to dimension Wording Similar to item 6

Note Names have been changed to alphabets in order to ensure the anonymity of participants

- 43 -

Appendix 4 Dimensions and Items of Public Service Motivation for Future Research

Dimensions and items

Attraction to Public Service (APS)

1 I am interested in helping to improve public service

2 I admire people who initiate or are involved in activities to aid my community

3 It is important to contribute to activities that tackle social problems

4 Meaningful public service is very important to me

5 It is important for me to contribute to the common good

Commitment to Public Values (CPV)

1 I think equal opportunities for citizens are very important

2 It is important that citizens can rely on the continuous provision of public services

3 It is fundamental that the interests of future generations are taken into account when

developing public policies

4 To act ethically is essential for public servants

5 I believe that public employees must always be aware of the legitimacy of their activities

6 I personally identify with the aim of protecting individual liberties and rights

Compassion (COM)

1 It is difficult for me to contain my feelings when I see people in distress

2 I feel sympathetic to the plight of the underprivileged

3 I empathize with other people who face difficulties

4 I get very upset when I see other people being treated unfairly

5 Considering the welfare of others is very important

Self-Sacrifice (SS)

1 I am prepared to make sacrifices for the good of society

2 I believe in putting civic duty before self

3 I am willing to risk personal loss to help society

4 I would agree to a good plan to make a better life for the poor even if it costs me money

- 23 -

in this area

One practical suggestion to measure PSM is to apply unit weights to its dimensions if

we assume the effects of the four dimensions are equivalent A literature review indicates that

unit weights have substantial predictive validity when compared with regression weights The

literature shows that the empirical correlations between scores generated from differential and

unit-weighting approaches are quite high In creating composite scores unit weights can be

an appropriate approach for weighting under many circumstances (Bobko Roth and Buster

2007) Although the uses of unit-weight composites will likely result in the explanation of

less variance than empirical estimations of dimensional weights they will remain consistently

defined across studies and hence suitable for theory testing Alternatively more thorough

discussions on the dimensions of PSM may suggest fixed-weight composites (Hardin et al

2010) A previous study (Kim 2010 22) also suggested ―Overall indexing PSM by

measuring all the dimensions of PSM and simply summing the averages of each dimension

can make it easier to measure PSM in practice as PSM = APS + CPV + COM + SS

An advantage of regarding PSM as a second-order formative construct is that the

dimensions can be analyzed both individually and collectively This is important as the

dimensions are expected to have different antecedents and consequences For instance Perry

(1997) found some significant differences in the influence of independent variables on the

four dimensions of PSM Taylor (2007) confirmed that certain dimensions are more important

than others in influencing work outcomes Moynihan and Pandey (2007a) reported that red

tape is negatively related to one dimension but reform orientation and hierarchical authority

is positively related to another dimension A recent study found that the SS dimension has a

positive relationship with occupational intention but the other dimensions do not (Liu et al

2011) Thus it is valuable to analyze the causes and effects of individual dimensions of PSM

- 24 -

as well as of the PSM construct

Conclusion

The purpose of this study is to develop a measurement instrument for PSM for cross-national

research and comparison On the basis of the sharpened construct of PSM (Kim and

Vandenabeele 2010) using data from an international survey conducted in 12 countries we

found that the four-factor structure of PSM is generalizable across country cases This study

provides the measurement index of PSM with the four dimensions of APS CPV COM and

SS The APS dimension measures the instrumental motives of PSM the CPV dimension

determines the value-based motives of PSM while the COM and SS dimensions gauge the

identification motives and self-sacrifice respectively When all four dimensions of PSM are

used in a survey cross-national equivalence of survey items can be assured The items for

measuring PSM in future research are suggested in Appendix 4

This study has several limitations First looking for samples which were comparable

to the largest possible extent we opted for surveys in local government The results found

and discussed in this study are based on PSM among local government employees The

results might be different when surveying central government employees or other specific

groups in the public sector Thus the potential impact of the sample bias needs to be

considered in understanding this study It also means that the survey items for measuring

PSM should be examined with different samples in various areas and at different government

levels

Second the test results showed that there is a high degree of correlation among the

dimension such as APS CPV and COM Future research should not only investigate whether

- 25 -

the items reflect the core content of the dimensions but also continue to develop more

appropriate items for better discriminant validity of each dimension Third the content of

APP dimension was assumed to be distinct from the CPV dimension that focuses on personal

disposition to pursue public interest and public values However the public interest sub-

dimension was separated from the public values sub-dimension in the CPV dimension and

combined with the APP dimension to form the APS dimension Further research should

analyze the CPV dimension and investigate the appropriateness of the APS dimension

In order to facilitate research internationally the PSM construct needs to be

improved conceptually and operationally Kim and Vandenabeele (2010) have dealt with the

conceptual composition and operational dimension of PSM as well as the measurement

model of PSM The next step is to develop an internationally robust measurement instrument

of PSM This study proposes that scholars use all four of the dimensions in measuring the

conceptual components of PSM such as self-sacrifice instrumental value-based and affective

motives because they collectively represent PSM It provides a set of appropriate items for

measuring PSM and confirms the dimensional structure of PSM The survey results suggest

that people in different cultures understand PSM in the same fashion We know that this study

is simply a step along the way to developing a more robust measure but we hope it

contributes to enhanced cross-national research and comparison along with the generation of

cumulative knowledge internationally

- 26 -

REFERENCES

Anderson James C and David D Gerbing 1988 Structural Equation Modeling in Practice

A Review and Recommended Two-Step Approach Psychological Bulletin 103

411-423

Arbuckle James L 2009 Amos 180 Userrsquos Guide Crawfordville FL Amos Development

Co

Beck Joslashrgensen Torben and Barry Bozeman 2007 Public Values An Inventory

Administration amp Society 39(3) 354-381

Beutel Ann M and Margaret Mooney Marini 1995 Gender and values American

Sociological Review 60 436 - 448

Bobko Philip Philip L Roth and Maury A Buster 2007 The Usefulness of Unit Weights in

Creating Composite Scores A Literature Review Application to Content Validity and

Meta-Analysis Organizational Research Methods 10(4) 689-709

Bollen Kenneth A 1989 Structural Equations with Latent Variables New York Wiley

Bollen K A and R Lennox 1991 Conventional Wisdom on Measurement A Structural

Equation Perspective Psychological Bulletin 110(2) 305-314

Byrne Barbara M 2010 Structural Equation Modeling with Amos Basic Concepts

Applications and Programming 2nd

ed New York Routledge

Camilleri Emanuel 2006 Towards Developing an Organizational Commitment ― Public

Service Motivation Model for Maltese Public Service Employees Public Policy

and Administration 21 63-83

Castaing Seacutebastien 2006 The Effects of Psychological Contract Fulfillment and Public

Service Motivation on Organizational Commitment in the French Civil Service

Public Policy and Administration 21 84-98

Cenfetelli Ronald T and Geneviegraveve Bassellier 2009 Interpretation of Formative

Measurement in Information Systems Research MIS Quarterly 33(4) 689-707

Charles Michael B W Martin de Jong and Neal Ryan 2011 Public Values in Western

- 27 -

Europe A Temporal Perspective American Review of Public Administration 41

75-91

Chin Wynne W 2010 How to Write Up and Report PLS Analysis In Handbook of Partical

Least Squares edited by V Esposito Vinzi et al 655-690 Berlin Springer-Verlag

Coursey David H and Sanjay K Pandey 2007a Public Service Motivation Measurement

Testing an Abridged Version of Perryrsquos Proposed Scale Administration amp Society

39 547-568

Coursey David H and Sanjay K Pandey 2007b Content Domain Measurement and

Validity of the Red Tape Concept American Review of Public Administration 37

342-361

Coursey David H James L Perry Jeffrey L Brudney and Laura Littlepage 2008

Psychometric Verification of Perryrsquos Public Service Motivation Instrument Results

for Volunteer Exemplars Review of Public Personnel Administration 28 79-90

Davis Paul and Karen West 2009 What Do Public Values Mean for Public Action Putting

Public Values in Their Plural Place American Review of Public Administration 39

602-618

DeHart-Davis Leisha Justin Marlowe and Sanjay K Pandey 2006 Gender Dimensions of

Public Service Motivation Public Administration Review 66 873-887

Diamantopoulos Adamantios 2006 The error term in formative measurement models

Interpretation and modeling implications Journal of Modeling in Management 1(1)

7-17

Diamantopoulos Adamantios and Judy A Siguaw 2006 Formative versus Reflective

Indicators in Organizational Measure Development A Comparison and Empirical

Illustration British Journal of Management 17 263-282

Diamantopoulos Adamantios and Heidi M Winklhofer 2001 Index Construction with

Formative Indicators An Alternative to Scale Development Journal of Marketing

Research 38 269-277

Edwards Jeffrey R 2010 The Fallacy of Formative Measurement Organizational Research

- 28 -

Methods Online First published on August 19 2010 as

doi1011771094428110378369

Edwards Jeffrey R and Richard P Bagozzi 2000 On the Nature and Direction of

Relationships between Constructs and Measures Psychological Methods 5 155-

174

Flora David B and Patrick J Curran 2004 An Empirical Evaluation of Alternative

Methods of Estimation for Confirmatory Factor Analysis with Ordinal Data

Psychological Methods 9 466-491

Fornell Claes and Fred L Bookstein 1982 Two structural equations models LISREL and

PLS applied to customer exit-voice theory Journal of Marketing Research 19 440-

452

Giauque David Adrian Ritz Freacutedeacuteric Varone Simon Anderfuhren-Biget and Christian

Waldner 2011 Putting Public Service Motivation into Context A Balance between

Universalism and Particularism Forthcoming in International Review of

Administrative Sciences 77(2)

Goumltz Oliver Kerstin Liehr-Gobbers and Manfred Krafft 2010 Evaluation of Structural

Equation Models Using the Partial Least Squares (PLS) Approach In V Esposito

Vinzi et al (eds) Handbook of Partial Least Squares 691-711 DOI 101007978-3-

540-32827-8_30 Springer-Verlag Berlin Heidelberg

Guinot Christiane Julie Latreille and Michel Tenenhaus 2001 PLS Path Modelling and

Multiple Table Analysis Application to the Cosmetic Habits of Women in Ile-de-

France Chemometrics and Intelligent Laboratory Systems 58 247-259

Hair Joseph F Jr William C Black Barry J Babin and Rolph E Anderson 2010

Multivariate Data Analysis 7th

ed Upper Saddle River NJ Pearson Education

Hardin Andrew M Jerry Cha-Jan Chang Mark A Fuller and Gholamreza Torkzadeh 2010

Formative Measurement and Academic Research In Search of Measurement

Theory Educational and Psychological Measurement Online First published on

June 11 2010 as doi101177001316440370208

- 29 -

Howell Roy D Einar Breivik and James B Wilcox 2007a Reconsidering Formative

Measurement Psychological Methods 12 205-218

Howell Roy D Einar Breivik and James B Wilcox 2007b Is Formative Measurement

Really Measurement Reply to Bollen (2007) and Bagozzi (2007) Psychological

Methods 12 238-245

Jarvis Cheryl Burke Scott B MacKenzie and Philip M Podsakoff 2003 A Critical Review

of Construct Indicators and Measurement Model Misspecification in Marketing and

Consumer Research Journal of Consumer Research 30 199-218

Joumlreskog Karl G and Dag Soumlrbom 2005 LISREL 872 (Computer Software) Lincolnwood

IL Scientific Software International

Kim Gimun Bongsik Shin and Varun Grover 2010 Investigating Two Contradictory Views

of Formative Measurement in Information Systems Research MIS Quarterly 34

345-365

Kim Sangmook 2009a Revising Perryrsquos Measurement Scale of Public Service Motivation

American Review of Public Administration 39 149-163

Kim Sangmook 2009b Testing the Structure of Public Service Motivation in Korea A

Research Note Journal of Public Administration Research and Theory 19 839-

851

Kim Sangmook 2010 Testing a Revised Measurement of Public Service Motivation

Reflective versus Formative Specification Journal of Public Administration

Research and Theory Doi101093jopartmuq048

Kim Sangmook and Wouter Vandenabeele 2010 A strategy for building public service

motivation research internationally Public Administration Review 70 701-709

Kline Rex B 2005 Principles and Practice of Structural Equation Modeling 2nd

ed New

York Guilford Press

Law Kenneth and Chi-Sum Wong 1999 Multidimensional Constructs in Structural

Equation Analysis An Illustration using the Job Perception and Job Satisfaction

Constructs Journal of Management 25 143-160

- 30 -

Leisink Peter and Bram Steijn 2009 Public Service Motivation and Job Performance of

Public Sector Employees in the Netherlands International Review of

Administrative Sciences 75 35-52

Liu Bangcheng Chun Hui Jin Hu Wensheng Yang and Xinli Yu 2011 How well can

Public Service Motivation Connect with Occupational Intention International

Review of Administrative Sciences 77 191-211

Liu Bangcheng Ningyu Tang and Xiaomei Zhu 2008 Public Service Motivation and Job

Satisfaction in China An Investigation of Generalisability and Instrumentality

International Journal of Manpower 29 684-699

MacCallum Robert C and Michael W Browne 1993 The Use of Causal Indicators in

Covariance Structure Models Some Practical Issues Psychological Bulletin 114

533-541

MacKenzie Scott B 2003 The dangers of poor construct conceptualization Journal of the

Academy of Marketing Science 31(3) 323-326

MacKenzie Scott B Philip M Podsakoff and Cheryl Burke Jarvis 2005 The Problem of

Measurement Model Misspecification in Behavioral and Organizational Research

and Some Recommended Solutions Journal of Applied Psychology 90 710-730

Martinsen Dorte Sindbjerg and Torben Beck Joslashrgensen 2010 Accountability as a

Differentiated Value in Supranational Governance American Review of Public

Administration 40 742-760

Moynihan Donald P and Sanjay K Pandey 2007a The Role of Organizations in Fostering

Public Service Motivation Public Administration Review 67 40-53

Moynihan Donald P and Sanjay K Pandey 2007b Finding Workable Levers over Work

Motivation Comparing Job Satisfaction Job Involvement and Organizational

Commitment Administration amp Society 39 803-832

Pandey Sanjay K and Edmund C Stazyk 2008 Antecedents and Correlates of Public

Service Motivation In Motivation in Public Management edited by James L Perry

and Annie Hondeghem 101-117 Oxford Oxford University Press

- 31 -

Perry James L 1996 Measuring Public Service Motivation An Assessment of Construct

Reliability and Validity Journal of Public Administration Research and Theory 6

5-22

Perry James L 1997 Antecedents of Public Service Motivation Journal of Public

Administration Research and Theory 7 181-197

Perry James L 2000 Bringing Society in Toward a Theory of Public Service Motivation

Journal of Public Administration Research and Theory 10 471-488

Perry James L and Annie Hondeghem 2008a Motivation in Public Management Oxford

Oxford University Press

Perry James L and Annie Hondeghem 2008b Directions for Future Theory and Research

In Motivation in Public Management edited by James L Perry and Annie

Hondeghem 294-313 Oxford Oxford University Press

Perry James L Annie Hondeghem and Lois Recascino Wise 2010 Revisiting the

Motivational Bases of Public Service Twenty Years of Research and an Agenda for

the Future Public Administration Review 70(5) 681-690

Perry James L and Wouter Vandenabeele 2008 Behavioral dynamics Institutions

Identities and Self-Regulation In Motivation in Public Management edited by

James L Perry and Annie Hondeghem 56-79 Oxford Oxford University Press

Perry James L and Lois Recascino Wise 1990 The Motivational Bases of Public Service

Public Administration Review 50 367-373

Rainey Hal G 1982 Reward Preference among Public and Private Managers In Search of

the Service Ethic American Review of Public Administration 16 288-302

Ringle Christian M Oliver Goumltz Martin Wetzels and Bradley Wilson 2009 On the Use of

Formative Measurement Specifications in Structural Equation Modeling A Monte

Carlo Simulation Study to Compare Covariance-Based and Partial Least Squares

Model Estimation Methodologies Faculty of Economics and Business Administration

Maastricht Univeristy

Ringle C M S Wende and S Will 2005 Smart PLS 20 (M3) Beta

- 32 -

httpwwwsmartplsde

Ritz Adrian 2011 Attraction to Public Policy Making A Qualitative Inquiry into

Improvements in PSM Measurement Forthcoming in Public Administration

Spector Paul E 1992 Summated Rating Scales Construction An Introduction Newbury

Park CA Sage

Steenbergen Marco R 1996 Compassion and American public opinion An analysis of the

NES Humanitarianism Scale Report to the Board of Overseers of the National

Election Studies

Steffen Patrick R and Kevin S Masters 2005 Does compassion mediate the intrinsic

religion-health relationship Annals of Behavioral Medicine 30 217-223

Taylor Jeannette 2007 The Impact of Public Service Motives on Work Outcomes in

Australia A Comparative Multi-Dimensional Analysis Public Administration 85

931-959

Vandenabeele Wouter 2007 Towards a Theory of Public Service Motivation An

Institutional Approach Public Management Review 9 545-556

Vandenabeele Wouter 2008a Government Calling Public Service Motivation as an Element

in Selecting Government as an Employer of Choice Public Administration 86

1089-1105

Vandenabeele Wouter 2008b Development of a Public Service Motivation Measurement

Scale Corroborating and Extending Perryrsquos Measurement Instrument International

Public Management Journal 11 143-167

Vandenabeele Wouter 2009 The Mediating Effect of Job Satisfaction and Organizational

Commitment on Self-Reported Performance More Robust Evidence of the PSM-

Performance Relationship International Review of Administrative Sciences 75 11-

34

Wilcox James B Roy D Howell and Einar Breivik 2008 Questions about Formative

Measurement Journal of Business Research 61 1219-1228

Wise Lois Recascino 2000 The Public Service Culture In Public Administration Concepts

- 33 -

and Cases 7th ed edited by Richard J Stillman II 342-353 Boston Houghton

Mifflin

Wright Bradley E 2008 Methodological Challenges associated with Public Service

Motivation Research In Motivation in Public Management edited by James L

Perry and Annie Hondeghem 80-98 Oxford Oxford University Press

Wright Bradley E and Sanjay K Pandey 2008 Public Service Motivation and the

Assumption of Person-Organization Fit Testing the Mediating Effect of Value

Congruence Administration amp Society 40 502-521

- 34 -

Table 1

Possible Indicators of PSM and Descriptive Statistics (n = 2868)

Dimensions and items Mean Standard

Deviation

Item-Total

Correlation

Attraction to Public Participation (APP)

APP1 I am interested in helping to improve public service

APP2 I am satisfied when I see people benefiting from the public programs I was

involved in

APP3 I like to discuss topics regarding public programs and policies with others

APP4 I believe that public sector activities contribute to our general welfare

APP5 I admire people who initiate or are involved in activities to aid my

community

APP6 Contributing to public programs and policies helps me realize myself

APP7 It is important to contribute to activities that tackle social problems

412

441

355

393

423

351

409

696

662

917

753

672

881

734

524

421

482

412

493

478

552

Commitment to Public Values (CPV)

Sub-dimension for Public Interests (CPV1)

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

CPI3 I would prefer seeing public officials do what is best for the whole

community even if it harmed my interests

CPI4 Serving the public interest is more important than helping a single individual

Sub-dimension for Public Values (CPV2)

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of public

services

CPV3 It is fundamental that public services respond to the needs of the citizens

CPV4 Decisions regarding public services should be democratic despite the time

and effort it takes

CPV5 Everybody is entitled to a good service even if it costs a lot of money

CPV6 It is fundamental that the interests of future generations are taken into

account when developing public policies

CPV7 To act ethically is essential for public servants

CPV8 I believe that public employees must always be aware of the legitimacy of

their activities

CPV9 I personally identify with the aim of protecting individual liberties and rights

418

409

348

337

428

416

433

394

367

426

441

434

414

695

662

899

995

730

704

710

852

932

689

681

692

763

525

596

480

251

484

437

356

325

376

439

462

402

392

Compassion (COM) COM1 It is difficult for me to contain my feelings when I see people in distress

COM2 I feel sympathetic to the plight of the underprivileged

COM3 I empathize with other people who face difficulties

COM4 I have little compassion for people in need who are unwilling to take the

first step to help themselves

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

354

387

388

299

412

391

951

825

707

1083

755

765

399

511

389

011

487

502

Self-Sacrifice (SS)

SS1 Making a difference to society means more to me than personal achievements

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS5 People should give back to society more than they get from it

SS6 Serving other citizens would give me a good feeling even if no one paid me for

it

SS7 I would agree to a good plan to make a better life for the poor even if it costs

me money

357

329

339

302

345

363

350

904

872

949

982

888

1049

895

508

584

454

531

412

368

546

- 35 -

Table 2

Background of Respondents (n = 2868)

Variables Characteristics Respondents

Sex

Male

Female

NA

1209

1622

37

422

566

13

Length of service

(years)

0 ~ 10

10 ~ 20

20 ~ 30

30+

NA

1126

723

484

491

44

393

252

169

171

15

Organizational status

―Do you supervise employees

No

Yes

NA

1472

1076

320

513

375

112

Country

Australia

Belgium

China

Denmark

France

Italy

Korea

Lithuania

The Netherlands

Switzerland

United Kingdom

USA

249

214

230

249

266

162

253

236

249

250

260

250

87

75

80

87

93

56

88

82

87

87

91

87 Note NA = no answer

- 36 -

Table 3

Standardized factor loadings and correlations for the 31-item PSM index (n = 2868)

Dimensions and items SFL (MLE) SFL (D-WLS) Alpha

APP

APP1

APP2

APP4

APP5

APP6

APP7

597

486

455

558

490

600

657

552

514

625

549

665

699

CPV1

CPI1

CPI2

CPI3

CPI4

588

660

488

289

653

732

535

294

542

CPV2

CPV1

CPV2

CPV3

CPV4

CPV5

CPV6

CPV7

CPV8

CPV9

611

546

436

393

418

534

598

496

460

690

631

505

457

486

601

669

573

530

745

COM

COM1

COM2

COM3

COM5

COM6

496

569

498

597

607

516

651

538

670

684

687

SS

SS1

SS2

SS3

SS4

SS5

SS6

SS7

567

736

558

728

467

409

663

660

762

591

695

515

494

709

782

Measures of fit (MLE) χsup2 (df = 424) = 43699 p lt 001 CFI = 835 GFI = 899

RMSEA = 057 [055 059]

Measure of fit (D-WLS) χsup2 (df = 424) = 39247 p lt 001 CFI = 971 GFI = 976

RMSEA = 054 [052 055]

Inter-factor correlations MLE (D-WLS)

1 2 3 4

1 APS

2 CPV1 996 (988)

3 CPV2 902 (890) 798 (798)

4 COM 787 (783) 758 (762) 779 (783)

5 SS 700 (742) 800 (859) 509 (537) 720 (729)

Note SFL = standardized factor loading All factor loadings and correlations are significant at p lt 001

- 37 -

Table 4

Standardized factor loadings and correlations for the 16-item PSM index (n = 2868)

Dimensions and items SFL

(MLE)

SFL

(D-WLS)

APS

APP5 I admire people who initiate or are involved in activities to aid my

community

APP7 It is important to contribute to activities that tackle social problems

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

611

592

637

666

665

669

690

751

CPV

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of

public services

CPV6 It is fundamental that the interests of future generations are taken

into account when developing public policies

CPV7 To act ethically is essential for public servants

612

546

551

590

698

631

618

658

COM

COM1 It is difficult for me to contain my feelings when I see people in

distress

COM2 I feel sympathetic to the plight of the underprivileged

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

490

546

596

604

515

608

669

679

SS

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS7 I would agree to a good plan to make a better life for the poor even if

it costs me money

763

571

766

678

807

629

747

755

Measures of fit (MLE)

χsup2 (df = 98) = 8490 p lt 001

CFI = 939 GFI = 965

RMSEA = 052 [049 055]

Measure of fit (D-WLS)

χsup2 (df = 98) = 6069 p lt 001

CFI = 988 GFI = 992

RMSEA = 043 [039 046]

Alpha CR (D-WLS) Inter-factor correlations MLE (D-WLS)

1 APS 717 788 1 APS 2 CPV 3 COM

2 CPV 663 747 848 (859)

3 COM 645 713 781 (791) 789 (795)

4 SS 784 826 565 (596) 462 (505) 715 (726)

Note SFL = standardized factor loading Alpha = Cronbachrsquos alpha CR = composite reliability

All factor loadings and correlations are significant at p lt 001

- 38 -

Table 5

Results of CFA in the individual countries Standardized factor loadings and measures of fit

Factors

amp Items

Australia Belgium China Denmark France Italy

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

672

756

799

785

804

943

710

703

597

790

799

730

584

675

685

681

731

742

578

748

674

643

765

739

645

704

678

708

812

779

632

604

765

766

672

853

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

552

593

560

699

727

632

624

605

560

713

709

645

702

602

616

724

686

735

567

560

715

654

605

811

608

658

587

696

755

698

673

620

523

782

697

654

COM

COM1

COM2

COM3

COM5

COM6

623

584

662

688

701

739

732

608

475

808

666

581

578

616

605

649

670

673

699

751

737

679

816

776

568

546

801

626

673

850

637

546

741

715

601

822

SS

SS2

SS3

SS4

SS7

654

794

806

892

767

813

727

693

799

793

725

838

857

820

773

896

855

842

806

749

792

829

753

884

847

731

753

889

816

753

815

692

664

795

782

734

Inter-factor

correlations

093

~

752

163

~

810

320

~

773

323

~

825

452

~

827

497

~

855

323

~

850

321

~

847

498

~

899

536

~

902

425

~

829

476

~

865

Sample size 249 214 230 249 266 162

Measures of Fit

χsup2 638 444 881 732 765 554 817 625 735 565 750 564

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 980 100 939 987 969 997 967 994 975 997 949 995

GFI 961 986 941 986 948 991 949 992 955 992 926 986

RMSEA 036 00 063 050 051 026 053 035 045 026 059 033

- 39 -

Table 5 (Continued)

Factors

amp Items

Korea Lithuania The Netherlands Switzerland UK USA

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

619

701

667

706

781

717

666

512

682

751

605

779

557

629

630

620

719

729

535

621

702

643

694

735

655

725

738

710

804

822

535

621

702

643

694

735

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

594

607

722

722

644

722

674

535

533

790

589

711

568

554

538

655

628

545

627

580

615

711

666

704

800

512

555

873

581

646

627

580

615

711

666

704

COM

COM1

COM2

COM3

COM5

COM6

694

677

550

763

736

627

557

639

428

550

696

534

737

623

508

766

616

608

659

592

683

703

661

769

682

583

729

739

644

803

659

592

683

703

661

769

SS

SS2

SS3

SS4

SS7

776

752

679

779

770

811

707

721

635

725

734

706

814

656

743

820

767

764

745

632

689

788

662

744

671

818

832

830

864

802

745

632

689

788

662

744

Inter-factor

correlations

432

~

726

466

~

746

376

~

788

406

~

902

377

~

852

455

~

836

637

~

818

660

~

828

317

~

852

360

~

852

637

~

818

660

~

828

Sample size 253 236 249 250 260 250

Measures of Fit

χsup2 896 573 612 462 773 616 815 715 769 542 815 715

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 951 996 975 100 958 993 957 991 972 998 957 991

GFI 942 988 960 989 951 987 948 990 955 990 948 990

RMSEA 059 028 034 00 050 034 053 044 048 022 053 044

Note All factor loadings and correlations are significant at p lt 001

- 40 -

Appendix 1 Dimensions and possible items of PSM (the first round)

1 Attraction to Public Participation

(1) I am interested in developing public programs helping my country or community (Kim 2009a)

(2) Sharing my views on public policies with others is attractive to me (Kim 2009a)

(3) I am satisfied when seeing people get benefits from the public programs I was involved in (Kim 2009a)

(4) I like to discuss political subjects with others (Giauque et al 2011)

(5) I think the governmental activities contribute to our welfare

(6) I like to initiate actions to help out my community

(7) I do not care much about politicians (Perry 1996)

2 Commitment to Public Values

Possible items for public interest

(1A) Meaningful public service is very important to me (Perry 1996)

(2A) It is important for me to contribute for the common good (Giauque et al 2011)

(3A) I recognize myself with the mission of protection of liberties and rights endorsed by the nation-state

(Giauque et al 2011)

(4A) I would prefer seeing public officials do what is best for the whole community even if it harmed my

interests (Perry 1996)

(5A) To me serving the public interest is more important than helping other people (Vandenabeele 2008b)

Possible items for public values

(1B) I recognize myself with the promotion of the equality of chances endorsed by the nation-state (Giauque et

al 2011)

(2B) It is fundamental that the public service provided to the citizen is regular and continuous (Giauque et al

2011)

(3B) It is fundamental that public service innovates to respond to the new needs of the citizensclients (Giauque

et al 2011)

(4B) The dignity and well-being of all should be the most important concerns in any society (Steenbergen

1996)

(5B) It is important that public servants account for all the costsexpenses they make (Vandenabeele 2008b)

(6B) We have to do everything in our power to pursue the goal of democracy

3 Compassion

(1) It is difficult for me to contain my feelings when I see people in distress (Perry 1996)

(2) I feel sympathetic to the plight of the underprivileged (Kim 2009a)

(3) I am often reminded by daily events how dependent we are on one another (Perry 1996)

(4) I feel compassion for others in the difficulties they are facing (Steffen and Masters 2005)

(5) It is not really my problem if others are in trouble and need help (Beutel and Marini 1995)

(6) I find it hard to be sympathetic toward starving people in foreign lands when there is so much trouble in our

own country (Beutel and Marini 1995)

(7) I get very upset when I see other people treated unfairly (Beutel and Marini 1995)

(8) I would agree to a good plan to make a better life for the poor even if it cost me money (Beutel and Marini

1995)

(9) To me patriotism includes seeing the welfare of others (Perry 1996)

(10) I care very much about other people

4 Self-Sacrifice

(1) Making a difference in society means more to me than personal achievements (Perry 1996)

(2) I am prepared to make enormous sacrifices for the good of society (Perry 1996)

(3) I believe in putting duty before self (Perry 1996)

(4) I am one of those rare people who would risk personal loss to help someone else (Perry 1996)

(5) I think people should give back to society more than they get from it (Perry 1996)

(6) Serving other citizens would give me a good feeling even if no one paid me for it (Perry 1996)

(7) Much of what I do is for a cause bigger than myself (Perry 1996)

- 41 -

Appendix 2 Newly suggested items through the first round

1 Items for the Attraction to Public Participation Dimension

(1) In the public domain every person should be counted as equal

(2) I enjoy to think and argue in a political way

(3) I am satisfied when seeing my ideas being integrated in public policies

(4) I can easily relate to the way of thinking and arguing of politicians

(5) Working in the public sector I can better realize myself

2 Items for the Public Values Sub-Dimension

(1) Public service should give value for money

(2) Cost considerations should play an important role in decisions about public service provision

(3) It is fundamental that the interests of future generations are taken into account when developing public

policies

(4) To act with correct ethics is for public servants as important as their professional competence

(5) Primarily public servants should be accountable to the public and not to their superiors

3 Items for the Self-Sacrifice Dimension

(1) I would agree to a good plan to make a better life for the poor even if it costs me money

- 42 -

Appendix 3 An example of a summary of comments and suggestions (the second round)

On the item ―I am interested in developing public programs helping my country or community

A It is a good sentence but is it possible for everyone to ―develop public programs Even public officials are

limited to do this

B OK

C or the special group

D Double barrel and difficult to translate to my country ―I am interested in helping to improve public services

E Good but I feel that wording requires revision If we draw sample from local government then country is

redundant ―I am interested in developing public policy programs that aim in helping my community

F 4 (General appreciation from1 to 4) Not clear whether it refers to ones work or to ones concern as a citizen

―(In my work) I am interested in contributing to the development of public etchellip

G Instead of development I would use implementing

H Delete public which is redundant because of the characteristic helping my country or community but add

or implementing ―I am interested in developing or implementing programs helping my country or

community

I Referring to dimension Wording Similar to item 6

Note Names have been changed to alphabets in order to ensure the anonymity of participants

- 43 -

Appendix 4 Dimensions and Items of Public Service Motivation for Future Research

Dimensions and items

Attraction to Public Service (APS)

1 I am interested in helping to improve public service

2 I admire people who initiate or are involved in activities to aid my community

3 It is important to contribute to activities that tackle social problems

4 Meaningful public service is very important to me

5 It is important for me to contribute to the common good

Commitment to Public Values (CPV)

1 I think equal opportunities for citizens are very important

2 It is important that citizens can rely on the continuous provision of public services

3 It is fundamental that the interests of future generations are taken into account when

developing public policies

4 To act ethically is essential for public servants

5 I believe that public employees must always be aware of the legitimacy of their activities

6 I personally identify with the aim of protecting individual liberties and rights

Compassion (COM)

1 It is difficult for me to contain my feelings when I see people in distress

2 I feel sympathetic to the plight of the underprivileged

3 I empathize with other people who face difficulties

4 I get very upset when I see other people being treated unfairly

5 Considering the welfare of others is very important

Self-Sacrifice (SS)

1 I am prepared to make sacrifices for the good of society

2 I believe in putting civic duty before self

3 I am willing to risk personal loss to help society

4 I would agree to a good plan to make a better life for the poor even if it costs me money

- 24 -

as well as of the PSM construct

Conclusion

The purpose of this study is to develop a measurement instrument for PSM for cross-national

research and comparison On the basis of the sharpened construct of PSM (Kim and

Vandenabeele 2010) using data from an international survey conducted in 12 countries we

found that the four-factor structure of PSM is generalizable across country cases This study

provides the measurement index of PSM with the four dimensions of APS CPV COM and

SS The APS dimension measures the instrumental motives of PSM the CPV dimension

determines the value-based motives of PSM while the COM and SS dimensions gauge the

identification motives and self-sacrifice respectively When all four dimensions of PSM are

used in a survey cross-national equivalence of survey items can be assured The items for

measuring PSM in future research are suggested in Appendix 4

This study has several limitations First looking for samples which were comparable

to the largest possible extent we opted for surveys in local government The results found

and discussed in this study are based on PSM among local government employees The

results might be different when surveying central government employees or other specific

groups in the public sector Thus the potential impact of the sample bias needs to be

considered in understanding this study It also means that the survey items for measuring

PSM should be examined with different samples in various areas and at different government

levels

Second the test results showed that there is a high degree of correlation among the

dimension such as APS CPV and COM Future research should not only investigate whether

- 25 -

the items reflect the core content of the dimensions but also continue to develop more

appropriate items for better discriminant validity of each dimension Third the content of

APP dimension was assumed to be distinct from the CPV dimension that focuses on personal

disposition to pursue public interest and public values However the public interest sub-

dimension was separated from the public values sub-dimension in the CPV dimension and

combined with the APP dimension to form the APS dimension Further research should

analyze the CPV dimension and investigate the appropriateness of the APS dimension

In order to facilitate research internationally the PSM construct needs to be

improved conceptually and operationally Kim and Vandenabeele (2010) have dealt with the

conceptual composition and operational dimension of PSM as well as the measurement

model of PSM The next step is to develop an internationally robust measurement instrument

of PSM This study proposes that scholars use all four of the dimensions in measuring the

conceptual components of PSM such as self-sacrifice instrumental value-based and affective

motives because they collectively represent PSM It provides a set of appropriate items for

measuring PSM and confirms the dimensional structure of PSM The survey results suggest

that people in different cultures understand PSM in the same fashion We know that this study

is simply a step along the way to developing a more robust measure but we hope it

contributes to enhanced cross-national research and comparison along with the generation of

cumulative knowledge internationally

- 26 -

REFERENCES

Anderson James C and David D Gerbing 1988 Structural Equation Modeling in Practice

A Review and Recommended Two-Step Approach Psychological Bulletin 103

411-423

Arbuckle James L 2009 Amos 180 Userrsquos Guide Crawfordville FL Amos Development

Co

Beck Joslashrgensen Torben and Barry Bozeman 2007 Public Values An Inventory

Administration amp Society 39(3) 354-381

Beutel Ann M and Margaret Mooney Marini 1995 Gender and values American

Sociological Review 60 436 - 448

Bobko Philip Philip L Roth and Maury A Buster 2007 The Usefulness of Unit Weights in

Creating Composite Scores A Literature Review Application to Content Validity and

Meta-Analysis Organizational Research Methods 10(4) 689-709

Bollen Kenneth A 1989 Structural Equations with Latent Variables New York Wiley

Bollen K A and R Lennox 1991 Conventional Wisdom on Measurement A Structural

Equation Perspective Psychological Bulletin 110(2) 305-314

Byrne Barbara M 2010 Structural Equation Modeling with Amos Basic Concepts

Applications and Programming 2nd

ed New York Routledge

Camilleri Emanuel 2006 Towards Developing an Organizational Commitment ― Public

Service Motivation Model for Maltese Public Service Employees Public Policy

and Administration 21 63-83

Castaing Seacutebastien 2006 The Effects of Psychological Contract Fulfillment and Public

Service Motivation on Organizational Commitment in the French Civil Service

Public Policy and Administration 21 84-98

Cenfetelli Ronald T and Geneviegraveve Bassellier 2009 Interpretation of Formative

Measurement in Information Systems Research MIS Quarterly 33(4) 689-707

Charles Michael B W Martin de Jong and Neal Ryan 2011 Public Values in Western

- 27 -

Europe A Temporal Perspective American Review of Public Administration 41

75-91

Chin Wynne W 2010 How to Write Up and Report PLS Analysis In Handbook of Partical

Least Squares edited by V Esposito Vinzi et al 655-690 Berlin Springer-Verlag

Coursey David H and Sanjay K Pandey 2007a Public Service Motivation Measurement

Testing an Abridged Version of Perryrsquos Proposed Scale Administration amp Society

39 547-568

Coursey David H and Sanjay K Pandey 2007b Content Domain Measurement and

Validity of the Red Tape Concept American Review of Public Administration 37

342-361

Coursey David H James L Perry Jeffrey L Brudney and Laura Littlepage 2008

Psychometric Verification of Perryrsquos Public Service Motivation Instrument Results

for Volunteer Exemplars Review of Public Personnel Administration 28 79-90

Davis Paul and Karen West 2009 What Do Public Values Mean for Public Action Putting

Public Values in Their Plural Place American Review of Public Administration 39

602-618

DeHart-Davis Leisha Justin Marlowe and Sanjay K Pandey 2006 Gender Dimensions of

Public Service Motivation Public Administration Review 66 873-887

Diamantopoulos Adamantios 2006 The error term in formative measurement models

Interpretation and modeling implications Journal of Modeling in Management 1(1)

7-17

Diamantopoulos Adamantios and Judy A Siguaw 2006 Formative versus Reflective

Indicators in Organizational Measure Development A Comparison and Empirical

Illustration British Journal of Management 17 263-282

Diamantopoulos Adamantios and Heidi M Winklhofer 2001 Index Construction with

Formative Indicators An Alternative to Scale Development Journal of Marketing

Research 38 269-277

Edwards Jeffrey R 2010 The Fallacy of Formative Measurement Organizational Research

- 28 -

Methods Online First published on August 19 2010 as

doi1011771094428110378369

Edwards Jeffrey R and Richard P Bagozzi 2000 On the Nature and Direction of

Relationships between Constructs and Measures Psychological Methods 5 155-

174

Flora David B and Patrick J Curran 2004 An Empirical Evaluation of Alternative

Methods of Estimation for Confirmatory Factor Analysis with Ordinal Data

Psychological Methods 9 466-491

Fornell Claes and Fred L Bookstein 1982 Two structural equations models LISREL and

PLS applied to customer exit-voice theory Journal of Marketing Research 19 440-

452

Giauque David Adrian Ritz Freacutedeacuteric Varone Simon Anderfuhren-Biget and Christian

Waldner 2011 Putting Public Service Motivation into Context A Balance between

Universalism and Particularism Forthcoming in International Review of

Administrative Sciences 77(2)

Goumltz Oliver Kerstin Liehr-Gobbers and Manfred Krafft 2010 Evaluation of Structural

Equation Models Using the Partial Least Squares (PLS) Approach In V Esposito

Vinzi et al (eds) Handbook of Partial Least Squares 691-711 DOI 101007978-3-

540-32827-8_30 Springer-Verlag Berlin Heidelberg

Guinot Christiane Julie Latreille and Michel Tenenhaus 2001 PLS Path Modelling and

Multiple Table Analysis Application to the Cosmetic Habits of Women in Ile-de-

France Chemometrics and Intelligent Laboratory Systems 58 247-259

Hair Joseph F Jr William C Black Barry J Babin and Rolph E Anderson 2010

Multivariate Data Analysis 7th

ed Upper Saddle River NJ Pearson Education

Hardin Andrew M Jerry Cha-Jan Chang Mark A Fuller and Gholamreza Torkzadeh 2010

Formative Measurement and Academic Research In Search of Measurement

Theory Educational and Psychological Measurement Online First published on

June 11 2010 as doi101177001316440370208

- 29 -

Howell Roy D Einar Breivik and James B Wilcox 2007a Reconsidering Formative

Measurement Psychological Methods 12 205-218

Howell Roy D Einar Breivik and James B Wilcox 2007b Is Formative Measurement

Really Measurement Reply to Bollen (2007) and Bagozzi (2007) Psychological

Methods 12 238-245

Jarvis Cheryl Burke Scott B MacKenzie and Philip M Podsakoff 2003 A Critical Review

of Construct Indicators and Measurement Model Misspecification in Marketing and

Consumer Research Journal of Consumer Research 30 199-218

Joumlreskog Karl G and Dag Soumlrbom 2005 LISREL 872 (Computer Software) Lincolnwood

IL Scientific Software International

Kim Gimun Bongsik Shin and Varun Grover 2010 Investigating Two Contradictory Views

of Formative Measurement in Information Systems Research MIS Quarterly 34

345-365

Kim Sangmook 2009a Revising Perryrsquos Measurement Scale of Public Service Motivation

American Review of Public Administration 39 149-163

Kim Sangmook 2009b Testing the Structure of Public Service Motivation in Korea A

Research Note Journal of Public Administration Research and Theory 19 839-

851

Kim Sangmook 2010 Testing a Revised Measurement of Public Service Motivation

Reflective versus Formative Specification Journal of Public Administration

Research and Theory Doi101093jopartmuq048

Kim Sangmook and Wouter Vandenabeele 2010 A strategy for building public service

motivation research internationally Public Administration Review 70 701-709

Kline Rex B 2005 Principles and Practice of Structural Equation Modeling 2nd

ed New

York Guilford Press

Law Kenneth and Chi-Sum Wong 1999 Multidimensional Constructs in Structural

Equation Analysis An Illustration using the Job Perception and Job Satisfaction

Constructs Journal of Management 25 143-160

- 30 -

Leisink Peter and Bram Steijn 2009 Public Service Motivation and Job Performance of

Public Sector Employees in the Netherlands International Review of

Administrative Sciences 75 35-52

Liu Bangcheng Chun Hui Jin Hu Wensheng Yang and Xinli Yu 2011 How well can

Public Service Motivation Connect with Occupational Intention International

Review of Administrative Sciences 77 191-211

Liu Bangcheng Ningyu Tang and Xiaomei Zhu 2008 Public Service Motivation and Job

Satisfaction in China An Investigation of Generalisability and Instrumentality

International Journal of Manpower 29 684-699

MacCallum Robert C and Michael W Browne 1993 The Use of Causal Indicators in

Covariance Structure Models Some Practical Issues Psychological Bulletin 114

533-541

MacKenzie Scott B 2003 The dangers of poor construct conceptualization Journal of the

Academy of Marketing Science 31(3) 323-326

MacKenzie Scott B Philip M Podsakoff and Cheryl Burke Jarvis 2005 The Problem of

Measurement Model Misspecification in Behavioral and Organizational Research

and Some Recommended Solutions Journal of Applied Psychology 90 710-730

Martinsen Dorte Sindbjerg and Torben Beck Joslashrgensen 2010 Accountability as a

Differentiated Value in Supranational Governance American Review of Public

Administration 40 742-760

Moynihan Donald P and Sanjay K Pandey 2007a The Role of Organizations in Fostering

Public Service Motivation Public Administration Review 67 40-53

Moynihan Donald P and Sanjay K Pandey 2007b Finding Workable Levers over Work

Motivation Comparing Job Satisfaction Job Involvement and Organizational

Commitment Administration amp Society 39 803-832

Pandey Sanjay K and Edmund C Stazyk 2008 Antecedents and Correlates of Public

Service Motivation In Motivation in Public Management edited by James L Perry

and Annie Hondeghem 101-117 Oxford Oxford University Press

- 31 -

Perry James L 1996 Measuring Public Service Motivation An Assessment of Construct

Reliability and Validity Journal of Public Administration Research and Theory 6

5-22

Perry James L 1997 Antecedents of Public Service Motivation Journal of Public

Administration Research and Theory 7 181-197

Perry James L 2000 Bringing Society in Toward a Theory of Public Service Motivation

Journal of Public Administration Research and Theory 10 471-488

Perry James L and Annie Hondeghem 2008a Motivation in Public Management Oxford

Oxford University Press

Perry James L and Annie Hondeghem 2008b Directions for Future Theory and Research

In Motivation in Public Management edited by James L Perry and Annie

Hondeghem 294-313 Oxford Oxford University Press

Perry James L Annie Hondeghem and Lois Recascino Wise 2010 Revisiting the

Motivational Bases of Public Service Twenty Years of Research and an Agenda for

the Future Public Administration Review 70(5) 681-690

Perry James L and Wouter Vandenabeele 2008 Behavioral dynamics Institutions

Identities and Self-Regulation In Motivation in Public Management edited by

James L Perry and Annie Hondeghem 56-79 Oxford Oxford University Press

Perry James L and Lois Recascino Wise 1990 The Motivational Bases of Public Service

Public Administration Review 50 367-373

Rainey Hal G 1982 Reward Preference among Public and Private Managers In Search of

the Service Ethic American Review of Public Administration 16 288-302

Ringle Christian M Oliver Goumltz Martin Wetzels and Bradley Wilson 2009 On the Use of

Formative Measurement Specifications in Structural Equation Modeling A Monte

Carlo Simulation Study to Compare Covariance-Based and Partial Least Squares

Model Estimation Methodologies Faculty of Economics and Business Administration

Maastricht Univeristy

Ringle C M S Wende and S Will 2005 Smart PLS 20 (M3) Beta

- 32 -

httpwwwsmartplsde

Ritz Adrian 2011 Attraction to Public Policy Making A Qualitative Inquiry into

Improvements in PSM Measurement Forthcoming in Public Administration

Spector Paul E 1992 Summated Rating Scales Construction An Introduction Newbury

Park CA Sage

Steenbergen Marco R 1996 Compassion and American public opinion An analysis of the

NES Humanitarianism Scale Report to the Board of Overseers of the National

Election Studies

Steffen Patrick R and Kevin S Masters 2005 Does compassion mediate the intrinsic

religion-health relationship Annals of Behavioral Medicine 30 217-223

Taylor Jeannette 2007 The Impact of Public Service Motives on Work Outcomes in

Australia A Comparative Multi-Dimensional Analysis Public Administration 85

931-959

Vandenabeele Wouter 2007 Towards a Theory of Public Service Motivation An

Institutional Approach Public Management Review 9 545-556

Vandenabeele Wouter 2008a Government Calling Public Service Motivation as an Element

in Selecting Government as an Employer of Choice Public Administration 86

1089-1105

Vandenabeele Wouter 2008b Development of a Public Service Motivation Measurement

Scale Corroborating and Extending Perryrsquos Measurement Instrument International

Public Management Journal 11 143-167

Vandenabeele Wouter 2009 The Mediating Effect of Job Satisfaction and Organizational

Commitment on Self-Reported Performance More Robust Evidence of the PSM-

Performance Relationship International Review of Administrative Sciences 75 11-

34

Wilcox James B Roy D Howell and Einar Breivik 2008 Questions about Formative

Measurement Journal of Business Research 61 1219-1228

Wise Lois Recascino 2000 The Public Service Culture In Public Administration Concepts

- 33 -

and Cases 7th ed edited by Richard J Stillman II 342-353 Boston Houghton

Mifflin

Wright Bradley E 2008 Methodological Challenges associated with Public Service

Motivation Research In Motivation in Public Management edited by James L

Perry and Annie Hondeghem 80-98 Oxford Oxford University Press

Wright Bradley E and Sanjay K Pandey 2008 Public Service Motivation and the

Assumption of Person-Organization Fit Testing the Mediating Effect of Value

Congruence Administration amp Society 40 502-521

- 34 -

Table 1

Possible Indicators of PSM and Descriptive Statistics (n = 2868)

Dimensions and items Mean Standard

Deviation

Item-Total

Correlation

Attraction to Public Participation (APP)

APP1 I am interested in helping to improve public service

APP2 I am satisfied when I see people benefiting from the public programs I was

involved in

APP3 I like to discuss topics regarding public programs and policies with others

APP4 I believe that public sector activities contribute to our general welfare

APP5 I admire people who initiate or are involved in activities to aid my

community

APP6 Contributing to public programs and policies helps me realize myself

APP7 It is important to contribute to activities that tackle social problems

412

441

355

393

423

351

409

696

662

917

753

672

881

734

524

421

482

412

493

478

552

Commitment to Public Values (CPV)

Sub-dimension for Public Interests (CPV1)

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

CPI3 I would prefer seeing public officials do what is best for the whole

community even if it harmed my interests

CPI4 Serving the public interest is more important than helping a single individual

Sub-dimension for Public Values (CPV2)

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of public

services

CPV3 It is fundamental that public services respond to the needs of the citizens

CPV4 Decisions regarding public services should be democratic despite the time

and effort it takes

CPV5 Everybody is entitled to a good service even if it costs a lot of money

CPV6 It is fundamental that the interests of future generations are taken into

account when developing public policies

CPV7 To act ethically is essential for public servants

CPV8 I believe that public employees must always be aware of the legitimacy of

their activities

CPV9 I personally identify with the aim of protecting individual liberties and rights

418

409

348

337

428

416

433

394

367

426

441

434

414

695

662

899

995

730

704

710

852

932

689

681

692

763

525

596

480

251

484

437

356

325

376

439

462

402

392

Compassion (COM) COM1 It is difficult for me to contain my feelings when I see people in distress

COM2 I feel sympathetic to the plight of the underprivileged

COM3 I empathize with other people who face difficulties

COM4 I have little compassion for people in need who are unwilling to take the

first step to help themselves

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

354

387

388

299

412

391

951

825

707

1083

755

765

399

511

389

011

487

502

Self-Sacrifice (SS)

SS1 Making a difference to society means more to me than personal achievements

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS5 People should give back to society more than they get from it

SS6 Serving other citizens would give me a good feeling even if no one paid me for

it

SS7 I would agree to a good plan to make a better life for the poor even if it costs

me money

357

329

339

302

345

363

350

904

872

949

982

888

1049

895

508

584

454

531

412

368

546

- 35 -

Table 2

Background of Respondents (n = 2868)

Variables Characteristics Respondents

Sex

Male

Female

NA

1209

1622

37

422

566

13

Length of service

(years)

0 ~ 10

10 ~ 20

20 ~ 30

30+

NA

1126

723

484

491

44

393

252

169

171

15

Organizational status

―Do you supervise employees

No

Yes

NA

1472

1076

320

513

375

112

Country

Australia

Belgium

China

Denmark

France

Italy

Korea

Lithuania

The Netherlands

Switzerland

United Kingdom

USA

249

214

230

249

266

162

253

236

249

250

260

250

87

75

80

87

93

56

88

82

87

87

91

87 Note NA = no answer

- 36 -

Table 3

Standardized factor loadings and correlations for the 31-item PSM index (n = 2868)

Dimensions and items SFL (MLE) SFL (D-WLS) Alpha

APP

APP1

APP2

APP4

APP5

APP6

APP7

597

486

455

558

490

600

657

552

514

625

549

665

699

CPV1

CPI1

CPI2

CPI3

CPI4

588

660

488

289

653

732

535

294

542

CPV2

CPV1

CPV2

CPV3

CPV4

CPV5

CPV6

CPV7

CPV8

CPV9

611

546

436

393

418

534

598

496

460

690

631

505

457

486

601

669

573

530

745

COM

COM1

COM2

COM3

COM5

COM6

496

569

498

597

607

516

651

538

670

684

687

SS

SS1

SS2

SS3

SS4

SS5

SS6

SS7

567

736

558

728

467

409

663

660

762

591

695

515

494

709

782

Measures of fit (MLE) χsup2 (df = 424) = 43699 p lt 001 CFI = 835 GFI = 899

RMSEA = 057 [055 059]

Measure of fit (D-WLS) χsup2 (df = 424) = 39247 p lt 001 CFI = 971 GFI = 976

RMSEA = 054 [052 055]

Inter-factor correlations MLE (D-WLS)

1 2 3 4

1 APS

2 CPV1 996 (988)

3 CPV2 902 (890) 798 (798)

4 COM 787 (783) 758 (762) 779 (783)

5 SS 700 (742) 800 (859) 509 (537) 720 (729)

Note SFL = standardized factor loading All factor loadings and correlations are significant at p lt 001

- 37 -

Table 4

Standardized factor loadings and correlations for the 16-item PSM index (n = 2868)

Dimensions and items SFL

(MLE)

SFL

(D-WLS)

APS

APP5 I admire people who initiate or are involved in activities to aid my

community

APP7 It is important to contribute to activities that tackle social problems

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

611

592

637

666

665

669

690

751

CPV

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of

public services

CPV6 It is fundamental that the interests of future generations are taken

into account when developing public policies

CPV7 To act ethically is essential for public servants

612

546

551

590

698

631

618

658

COM

COM1 It is difficult for me to contain my feelings when I see people in

distress

COM2 I feel sympathetic to the plight of the underprivileged

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

490

546

596

604

515

608

669

679

SS

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS7 I would agree to a good plan to make a better life for the poor even if

it costs me money

763

571

766

678

807

629

747

755

Measures of fit (MLE)

χsup2 (df = 98) = 8490 p lt 001

CFI = 939 GFI = 965

RMSEA = 052 [049 055]

Measure of fit (D-WLS)

χsup2 (df = 98) = 6069 p lt 001

CFI = 988 GFI = 992

RMSEA = 043 [039 046]

Alpha CR (D-WLS) Inter-factor correlations MLE (D-WLS)

1 APS 717 788 1 APS 2 CPV 3 COM

2 CPV 663 747 848 (859)

3 COM 645 713 781 (791) 789 (795)

4 SS 784 826 565 (596) 462 (505) 715 (726)

Note SFL = standardized factor loading Alpha = Cronbachrsquos alpha CR = composite reliability

All factor loadings and correlations are significant at p lt 001

- 38 -

Table 5

Results of CFA in the individual countries Standardized factor loadings and measures of fit

Factors

amp Items

Australia Belgium China Denmark France Italy

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

672

756

799

785

804

943

710

703

597

790

799

730

584

675

685

681

731

742

578

748

674

643

765

739

645

704

678

708

812

779

632

604

765

766

672

853

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

552

593

560

699

727

632

624

605

560

713

709

645

702

602

616

724

686

735

567

560

715

654

605

811

608

658

587

696

755

698

673

620

523

782

697

654

COM

COM1

COM2

COM3

COM5

COM6

623

584

662

688

701

739

732

608

475

808

666

581

578

616

605

649

670

673

699

751

737

679

816

776

568

546

801

626

673

850

637

546

741

715

601

822

SS

SS2

SS3

SS4

SS7

654

794

806

892

767

813

727

693

799

793

725

838

857

820

773

896

855

842

806

749

792

829

753

884

847

731

753

889

816

753

815

692

664

795

782

734

Inter-factor

correlations

093

~

752

163

~

810

320

~

773

323

~

825

452

~

827

497

~

855

323

~

850

321

~

847

498

~

899

536

~

902

425

~

829

476

~

865

Sample size 249 214 230 249 266 162

Measures of Fit

χsup2 638 444 881 732 765 554 817 625 735 565 750 564

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 980 100 939 987 969 997 967 994 975 997 949 995

GFI 961 986 941 986 948 991 949 992 955 992 926 986

RMSEA 036 00 063 050 051 026 053 035 045 026 059 033

- 39 -

Table 5 (Continued)

Factors

amp Items

Korea Lithuania The Netherlands Switzerland UK USA

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

619

701

667

706

781

717

666

512

682

751

605

779

557

629

630

620

719

729

535

621

702

643

694

735

655

725

738

710

804

822

535

621

702

643

694

735

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

594

607

722

722

644

722

674

535

533

790

589

711

568

554

538

655

628

545

627

580

615

711

666

704

800

512

555

873

581

646

627

580

615

711

666

704

COM

COM1

COM2

COM3

COM5

COM6

694

677

550

763

736

627

557

639

428

550

696

534

737

623

508

766

616

608

659

592

683

703

661

769

682

583

729

739

644

803

659

592

683

703

661

769

SS

SS2

SS3

SS4

SS7

776

752

679

779

770

811

707

721

635

725

734

706

814

656

743

820

767

764

745

632

689

788

662

744

671

818

832

830

864

802

745

632

689

788

662

744

Inter-factor

correlations

432

~

726

466

~

746

376

~

788

406

~

902

377

~

852

455

~

836

637

~

818

660

~

828

317

~

852

360

~

852

637

~

818

660

~

828

Sample size 253 236 249 250 260 250

Measures of Fit

χsup2 896 573 612 462 773 616 815 715 769 542 815 715

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 951 996 975 100 958 993 957 991 972 998 957 991

GFI 942 988 960 989 951 987 948 990 955 990 948 990

RMSEA 059 028 034 00 050 034 053 044 048 022 053 044

Note All factor loadings and correlations are significant at p lt 001

- 40 -

Appendix 1 Dimensions and possible items of PSM (the first round)

1 Attraction to Public Participation

(1) I am interested in developing public programs helping my country or community (Kim 2009a)

(2) Sharing my views on public policies with others is attractive to me (Kim 2009a)

(3) I am satisfied when seeing people get benefits from the public programs I was involved in (Kim 2009a)

(4) I like to discuss political subjects with others (Giauque et al 2011)

(5) I think the governmental activities contribute to our welfare

(6) I like to initiate actions to help out my community

(7) I do not care much about politicians (Perry 1996)

2 Commitment to Public Values

Possible items for public interest

(1A) Meaningful public service is very important to me (Perry 1996)

(2A) It is important for me to contribute for the common good (Giauque et al 2011)

(3A) I recognize myself with the mission of protection of liberties and rights endorsed by the nation-state

(Giauque et al 2011)

(4A) I would prefer seeing public officials do what is best for the whole community even if it harmed my

interests (Perry 1996)

(5A) To me serving the public interest is more important than helping other people (Vandenabeele 2008b)

Possible items for public values

(1B) I recognize myself with the promotion of the equality of chances endorsed by the nation-state (Giauque et

al 2011)

(2B) It is fundamental that the public service provided to the citizen is regular and continuous (Giauque et al

2011)

(3B) It is fundamental that public service innovates to respond to the new needs of the citizensclients (Giauque

et al 2011)

(4B) The dignity and well-being of all should be the most important concerns in any society (Steenbergen

1996)

(5B) It is important that public servants account for all the costsexpenses they make (Vandenabeele 2008b)

(6B) We have to do everything in our power to pursue the goal of democracy

3 Compassion

(1) It is difficult for me to contain my feelings when I see people in distress (Perry 1996)

(2) I feel sympathetic to the plight of the underprivileged (Kim 2009a)

(3) I am often reminded by daily events how dependent we are on one another (Perry 1996)

(4) I feel compassion for others in the difficulties they are facing (Steffen and Masters 2005)

(5) It is not really my problem if others are in trouble and need help (Beutel and Marini 1995)

(6) I find it hard to be sympathetic toward starving people in foreign lands when there is so much trouble in our

own country (Beutel and Marini 1995)

(7) I get very upset when I see other people treated unfairly (Beutel and Marini 1995)

(8) I would agree to a good plan to make a better life for the poor even if it cost me money (Beutel and Marini

1995)

(9) To me patriotism includes seeing the welfare of others (Perry 1996)

(10) I care very much about other people

4 Self-Sacrifice

(1) Making a difference in society means more to me than personal achievements (Perry 1996)

(2) I am prepared to make enormous sacrifices for the good of society (Perry 1996)

(3) I believe in putting duty before self (Perry 1996)

(4) I am one of those rare people who would risk personal loss to help someone else (Perry 1996)

(5) I think people should give back to society more than they get from it (Perry 1996)

(6) Serving other citizens would give me a good feeling even if no one paid me for it (Perry 1996)

(7) Much of what I do is for a cause bigger than myself (Perry 1996)

- 41 -

Appendix 2 Newly suggested items through the first round

1 Items for the Attraction to Public Participation Dimension

(1) In the public domain every person should be counted as equal

(2) I enjoy to think and argue in a political way

(3) I am satisfied when seeing my ideas being integrated in public policies

(4) I can easily relate to the way of thinking and arguing of politicians

(5) Working in the public sector I can better realize myself

2 Items for the Public Values Sub-Dimension

(1) Public service should give value for money

(2) Cost considerations should play an important role in decisions about public service provision

(3) It is fundamental that the interests of future generations are taken into account when developing public

policies

(4) To act with correct ethics is for public servants as important as their professional competence

(5) Primarily public servants should be accountable to the public and not to their superiors

3 Items for the Self-Sacrifice Dimension

(1) I would agree to a good plan to make a better life for the poor even if it costs me money

- 42 -

Appendix 3 An example of a summary of comments and suggestions (the second round)

On the item ―I am interested in developing public programs helping my country or community

A It is a good sentence but is it possible for everyone to ―develop public programs Even public officials are

limited to do this

B OK

C or the special group

D Double barrel and difficult to translate to my country ―I am interested in helping to improve public services

E Good but I feel that wording requires revision If we draw sample from local government then country is

redundant ―I am interested in developing public policy programs that aim in helping my community

F 4 (General appreciation from1 to 4) Not clear whether it refers to ones work or to ones concern as a citizen

―(In my work) I am interested in contributing to the development of public etchellip

G Instead of development I would use implementing

H Delete public which is redundant because of the characteristic helping my country or community but add

or implementing ―I am interested in developing or implementing programs helping my country or

community

I Referring to dimension Wording Similar to item 6

Note Names have been changed to alphabets in order to ensure the anonymity of participants

- 43 -

Appendix 4 Dimensions and Items of Public Service Motivation for Future Research

Dimensions and items

Attraction to Public Service (APS)

1 I am interested in helping to improve public service

2 I admire people who initiate or are involved in activities to aid my community

3 It is important to contribute to activities that tackle social problems

4 Meaningful public service is very important to me

5 It is important for me to contribute to the common good

Commitment to Public Values (CPV)

1 I think equal opportunities for citizens are very important

2 It is important that citizens can rely on the continuous provision of public services

3 It is fundamental that the interests of future generations are taken into account when

developing public policies

4 To act ethically is essential for public servants

5 I believe that public employees must always be aware of the legitimacy of their activities

6 I personally identify with the aim of protecting individual liberties and rights

Compassion (COM)

1 It is difficult for me to contain my feelings when I see people in distress

2 I feel sympathetic to the plight of the underprivileged

3 I empathize with other people who face difficulties

4 I get very upset when I see other people being treated unfairly

5 Considering the welfare of others is very important

Self-Sacrifice (SS)

1 I am prepared to make sacrifices for the good of society

2 I believe in putting civic duty before self

3 I am willing to risk personal loss to help society

4 I would agree to a good plan to make a better life for the poor even if it costs me money

- 25 -

the items reflect the core content of the dimensions but also continue to develop more

appropriate items for better discriminant validity of each dimension Third the content of

APP dimension was assumed to be distinct from the CPV dimension that focuses on personal

disposition to pursue public interest and public values However the public interest sub-

dimension was separated from the public values sub-dimension in the CPV dimension and

combined with the APP dimension to form the APS dimension Further research should

analyze the CPV dimension and investigate the appropriateness of the APS dimension

In order to facilitate research internationally the PSM construct needs to be

improved conceptually and operationally Kim and Vandenabeele (2010) have dealt with the

conceptual composition and operational dimension of PSM as well as the measurement

model of PSM The next step is to develop an internationally robust measurement instrument

of PSM This study proposes that scholars use all four of the dimensions in measuring the

conceptual components of PSM such as self-sacrifice instrumental value-based and affective

motives because they collectively represent PSM It provides a set of appropriate items for

measuring PSM and confirms the dimensional structure of PSM The survey results suggest

that people in different cultures understand PSM in the same fashion We know that this study

is simply a step along the way to developing a more robust measure but we hope it

contributes to enhanced cross-national research and comparison along with the generation of

cumulative knowledge internationally

- 26 -

REFERENCES

Anderson James C and David D Gerbing 1988 Structural Equation Modeling in Practice

A Review and Recommended Two-Step Approach Psychological Bulletin 103

411-423

Arbuckle James L 2009 Amos 180 Userrsquos Guide Crawfordville FL Amos Development

Co

Beck Joslashrgensen Torben and Barry Bozeman 2007 Public Values An Inventory

Administration amp Society 39(3) 354-381

Beutel Ann M and Margaret Mooney Marini 1995 Gender and values American

Sociological Review 60 436 - 448

Bobko Philip Philip L Roth and Maury A Buster 2007 The Usefulness of Unit Weights in

Creating Composite Scores A Literature Review Application to Content Validity and

Meta-Analysis Organizational Research Methods 10(4) 689-709

Bollen Kenneth A 1989 Structural Equations with Latent Variables New York Wiley

Bollen K A and R Lennox 1991 Conventional Wisdom on Measurement A Structural

Equation Perspective Psychological Bulletin 110(2) 305-314

Byrne Barbara M 2010 Structural Equation Modeling with Amos Basic Concepts

Applications and Programming 2nd

ed New York Routledge

Camilleri Emanuel 2006 Towards Developing an Organizational Commitment ― Public

Service Motivation Model for Maltese Public Service Employees Public Policy

and Administration 21 63-83

Castaing Seacutebastien 2006 The Effects of Psychological Contract Fulfillment and Public

Service Motivation on Organizational Commitment in the French Civil Service

Public Policy and Administration 21 84-98

Cenfetelli Ronald T and Geneviegraveve Bassellier 2009 Interpretation of Formative

Measurement in Information Systems Research MIS Quarterly 33(4) 689-707

Charles Michael B W Martin de Jong and Neal Ryan 2011 Public Values in Western

- 27 -

Europe A Temporal Perspective American Review of Public Administration 41

75-91

Chin Wynne W 2010 How to Write Up and Report PLS Analysis In Handbook of Partical

Least Squares edited by V Esposito Vinzi et al 655-690 Berlin Springer-Verlag

Coursey David H and Sanjay K Pandey 2007a Public Service Motivation Measurement

Testing an Abridged Version of Perryrsquos Proposed Scale Administration amp Society

39 547-568

Coursey David H and Sanjay K Pandey 2007b Content Domain Measurement and

Validity of the Red Tape Concept American Review of Public Administration 37

342-361

Coursey David H James L Perry Jeffrey L Brudney and Laura Littlepage 2008

Psychometric Verification of Perryrsquos Public Service Motivation Instrument Results

for Volunteer Exemplars Review of Public Personnel Administration 28 79-90

Davis Paul and Karen West 2009 What Do Public Values Mean for Public Action Putting

Public Values in Their Plural Place American Review of Public Administration 39

602-618

DeHart-Davis Leisha Justin Marlowe and Sanjay K Pandey 2006 Gender Dimensions of

Public Service Motivation Public Administration Review 66 873-887

Diamantopoulos Adamantios 2006 The error term in formative measurement models

Interpretation and modeling implications Journal of Modeling in Management 1(1)

7-17

Diamantopoulos Adamantios and Judy A Siguaw 2006 Formative versus Reflective

Indicators in Organizational Measure Development A Comparison and Empirical

Illustration British Journal of Management 17 263-282

Diamantopoulos Adamantios and Heidi M Winklhofer 2001 Index Construction with

Formative Indicators An Alternative to Scale Development Journal of Marketing

Research 38 269-277

Edwards Jeffrey R 2010 The Fallacy of Formative Measurement Organizational Research

- 28 -

Methods Online First published on August 19 2010 as

doi1011771094428110378369

Edwards Jeffrey R and Richard P Bagozzi 2000 On the Nature and Direction of

Relationships between Constructs and Measures Psychological Methods 5 155-

174

Flora David B and Patrick J Curran 2004 An Empirical Evaluation of Alternative

Methods of Estimation for Confirmatory Factor Analysis with Ordinal Data

Psychological Methods 9 466-491

Fornell Claes and Fred L Bookstein 1982 Two structural equations models LISREL and

PLS applied to customer exit-voice theory Journal of Marketing Research 19 440-

452

Giauque David Adrian Ritz Freacutedeacuteric Varone Simon Anderfuhren-Biget and Christian

Waldner 2011 Putting Public Service Motivation into Context A Balance between

Universalism and Particularism Forthcoming in International Review of

Administrative Sciences 77(2)

Goumltz Oliver Kerstin Liehr-Gobbers and Manfred Krafft 2010 Evaluation of Structural

Equation Models Using the Partial Least Squares (PLS) Approach In V Esposito

Vinzi et al (eds) Handbook of Partial Least Squares 691-711 DOI 101007978-3-

540-32827-8_30 Springer-Verlag Berlin Heidelberg

Guinot Christiane Julie Latreille and Michel Tenenhaus 2001 PLS Path Modelling and

Multiple Table Analysis Application to the Cosmetic Habits of Women in Ile-de-

France Chemometrics and Intelligent Laboratory Systems 58 247-259

Hair Joseph F Jr William C Black Barry J Babin and Rolph E Anderson 2010

Multivariate Data Analysis 7th

ed Upper Saddle River NJ Pearson Education

Hardin Andrew M Jerry Cha-Jan Chang Mark A Fuller and Gholamreza Torkzadeh 2010

Formative Measurement and Academic Research In Search of Measurement

Theory Educational and Psychological Measurement Online First published on

June 11 2010 as doi101177001316440370208

- 29 -

Howell Roy D Einar Breivik and James B Wilcox 2007a Reconsidering Formative

Measurement Psychological Methods 12 205-218

Howell Roy D Einar Breivik and James B Wilcox 2007b Is Formative Measurement

Really Measurement Reply to Bollen (2007) and Bagozzi (2007) Psychological

Methods 12 238-245

Jarvis Cheryl Burke Scott B MacKenzie and Philip M Podsakoff 2003 A Critical Review

of Construct Indicators and Measurement Model Misspecification in Marketing and

Consumer Research Journal of Consumer Research 30 199-218

Joumlreskog Karl G and Dag Soumlrbom 2005 LISREL 872 (Computer Software) Lincolnwood

IL Scientific Software International

Kim Gimun Bongsik Shin and Varun Grover 2010 Investigating Two Contradictory Views

of Formative Measurement in Information Systems Research MIS Quarterly 34

345-365

Kim Sangmook 2009a Revising Perryrsquos Measurement Scale of Public Service Motivation

American Review of Public Administration 39 149-163

Kim Sangmook 2009b Testing the Structure of Public Service Motivation in Korea A

Research Note Journal of Public Administration Research and Theory 19 839-

851

Kim Sangmook 2010 Testing a Revised Measurement of Public Service Motivation

Reflective versus Formative Specification Journal of Public Administration

Research and Theory Doi101093jopartmuq048

Kim Sangmook and Wouter Vandenabeele 2010 A strategy for building public service

motivation research internationally Public Administration Review 70 701-709

Kline Rex B 2005 Principles and Practice of Structural Equation Modeling 2nd

ed New

York Guilford Press

Law Kenneth and Chi-Sum Wong 1999 Multidimensional Constructs in Structural

Equation Analysis An Illustration using the Job Perception and Job Satisfaction

Constructs Journal of Management 25 143-160

- 30 -

Leisink Peter and Bram Steijn 2009 Public Service Motivation and Job Performance of

Public Sector Employees in the Netherlands International Review of

Administrative Sciences 75 35-52

Liu Bangcheng Chun Hui Jin Hu Wensheng Yang and Xinli Yu 2011 How well can

Public Service Motivation Connect with Occupational Intention International

Review of Administrative Sciences 77 191-211

Liu Bangcheng Ningyu Tang and Xiaomei Zhu 2008 Public Service Motivation and Job

Satisfaction in China An Investigation of Generalisability and Instrumentality

International Journal of Manpower 29 684-699

MacCallum Robert C and Michael W Browne 1993 The Use of Causal Indicators in

Covariance Structure Models Some Practical Issues Psychological Bulletin 114

533-541

MacKenzie Scott B 2003 The dangers of poor construct conceptualization Journal of the

Academy of Marketing Science 31(3) 323-326

MacKenzie Scott B Philip M Podsakoff and Cheryl Burke Jarvis 2005 The Problem of

Measurement Model Misspecification in Behavioral and Organizational Research

and Some Recommended Solutions Journal of Applied Psychology 90 710-730

Martinsen Dorte Sindbjerg and Torben Beck Joslashrgensen 2010 Accountability as a

Differentiated Value in Supranational Governance American Review of Public

Administration 40 742-760

Moynihan Donald P and Sanjay K Pandey 2007a The Role of Organizations in Fostering

Public Service Motivation Public Administration Review 67 40-53

Moynihan Donald P and Sanjay K Pandey 2007b Finding Workable Levers over Work

Motivation Comparing Job Satisfaction Job Involvement and Organizational

Commitment Administration amp Society 39 803-832

Pandey Sanjay K and Edmund C Stazyk 2008 Antecedents and Correlates of Public

Service Motivation In Motivation in Public Management edited by James L Perry

and Annie Hondeghem 101-117 Oxford Oxford University Press

- 31 -

Perry James L 1996 Measuring Public Service Motivation An Assessment of Construct

Reliability and Validity Journal of Public Administration Research and Theory 6

5-22

Perry James L 1997 Antecedents of Public Service Motivation Journal of Public

Administration Research and Theory 7 181-197

Perry James L 2000 Bringing Society in Toward a Theory of Public Service Motivation

Journal of Public Administration Research and Theory 10 471-488

Perry James L and Annie Hondeghem 2008a Motivation in Public Management Oxford

Oxford University Press

Perry James L and Annie Hondeghem 2008b Directions for Future Theory and Research

In Motivation in Public Management edited by James L Perry and Annie

Hondeghem 294-313 Oxford Oxford University Press

Perry James L Annie Hondeghem and Lois Recascino Wise 2010 Revisiting the

Motivational Bases of Public Service Twenty Years of Research and an Agenda for

the Future Public Administration Review 70(5) 681-690

Perry James L and Wouter Vandenabeele 2008 Behavioral dynamics Institutions

Identities and Self-Regulation In Motivation in Public Management edited by

James L Perry and Annie Hondeghem 56-79 Oxford Oxford University Press

Perry James L and Lois Recascino Wise 1990 The Motivational Bases of Public Service

Public Administration Review 50 367-373

Rainey Hal G 1982 Reward Preference among Public and Private Managers In Search of

the Service Ethic American Review of Public Administration 16 288-302

Ringle Christian M Oliver Goumltz Martin Wetzels and Bradley Wilson 2009 On the Use of

Formative Measurement Specifications in Structural Equation Modeling A Monte

Carlo Simulation Study to Compare Covariance-Based and Partial Least Squares

Model Estimation Methodologies Faculty of Economics and Business Administration

Maastricht Univeristy

Ringle C M S Wende and S Will 2005 Smart PLS 20 (M3) Beta

- 32 -

httpwwwsmartplsde

Ritz Adrian 2011 Attraction to Public Policy Making A Qualitative Inquiry into

Improvements in PSM Measurement Forthcoming in Public Administration

Spector Paul E 1992 Summated Rating Scales Construction An Introduction Newbury

Park CA Sage

Steenbergen Marco R 1996 Compassion and American public opinion An analysis of the

NES Humanitarianism Scale Report to the Board of Overseers of the National

Election Studies

Steffen Patrick R and Kevin S Masters 2005 Does compassion mediate the intrinsic

religion-health relationship Annals of Behavioral Medicine 30 217-223

Taylor Jeannette 2007 The Impact of Public Service Motives on Work Outcomes in

Australia A Comparative Multi-Dimensional Analysis Public Administration 85

931-959

Vandenabeele Wouter 2007 Towards a Theory of Public Service Motivation An

Institutional Approach Public Management Review 9 545-556

Vandenabeele Wouter 2008a Government Calling Public Service Motivation as an Element

in Selecting Government as an Employer of Choice Public Administration 86

1089-1105

Vandenabeele Wouter 2008b Development of a Public Service Motivation Measurement

Scale Corroborating and Extending Perryrsquos Measurement Instrument International

Public Management Journal 11 143-167

Vandenabeele Wouter 2009 The Mediating Effect of Job Satisfaction and Organizational

Commitment on Self-Reported Performance More Robust Evidence of the PSM-

Performance Relationship International Review of Administrative Sciences 75 11-

34

Wilcox James B Roy D Howell and Einar Breivik 2008 Questions about Formative

Measurement Journal of Business Research 61 1219-1228

Wise Lois Recascino 2000 The Public Service Culture In Public Administration Concepts

- 33 -

and Cases 7th ed edited by Richard J Stillman II 342-353 Boston Houghton

Mifflin

Wright Bradley E 2008 Methodological Challenges associated with Public Service

Motivation Research In Motivation in Public Management edited by James L

Perry and Annie Hondeghem 80-98 Oxford Oxford University Press

Wright Bradley E and Sanjay K Pandey 2008 Public Service Motivation and the

Assumption of Person-Organization Fit Testing the Mediating Effect of Value

Congruence Administration amp Society 40 502-521

- 34 -

Table 1

Possible Indicators of PSM and Descriptive Statistics (n = 2868)

Dimensions and items Mean Standard

Deviation

Item-Total

Correlation

Attraction to Public Participation (APP)

APP1 I am interested in helping to improve public service

APP2 I am satisfied when I see people benefiting from the public programs I was

involved in

APP3 I like to discuss topics regarding public programs and policies with others

APP4 I believe that public sector activities contribute to our general welfare

APP5 I admire people who initiate or are involved in activities to aid my

community

APP6 Contributing to public programs and policies helps me realize myself

APP7 It is important to contribute to activities that tackle social problems

412

441

355

393

423

351

409

696

662

917

753

672

881

734

524

421

482

412

493

478

552

Commitment to Public Values (CPV)

Sub-dimension for Public Interests (CPV1)

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

CPI3 I would prefer seeing public officials do what is best for the whole

community even if it harmed my interests

CPI4 Serving the public interest is more important than helping a single individual

Sub-dimension for Public Values (CPV2)

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of public

services

CPV3 It is fundamental that public services respond to the needs of the citizens

CPV4 Decisions regarding public services should be democratic despite the time

and effort it takes

CPV5 Everybody is entitled to a good service even if it costs a lot of money

CPV6 It is fundamental that the interests of future generations are taken into

account when developing public policies

CPV7 To act ethically is essential for public servants

CPV8 I believe that public employees must always be aware of the legitimacy of

their activities

CPV9 I personally identify with the aim of protecting individual liberties and rights

418

409

348

337

428

416

433

394

367

426

441

434

414

695

662

899

995

730

704

710

852

932

689

681

692

763

525

596

480

251

484

437

356

325

376

439

462

402

392

Compassion (COM) COM1 It is difficult for me to contain my feelings when I see people in distress

COM2 I feel sympathetic to the plight of the underprivileged

COM3 I empathize with other people who face difficulties

COM4 I have little compassion for people in need who are unwilling to take the

first step to help themselves

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

354

387

388

299

412

391

951

825

707

1083

755

765

399

511

389

011

487

502

Self-Sacrifice (SS)

SS1 Making a difference to society means more to me than personal achievements

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS5 People should give back to society more than they get from it

SS6 Serving other citizens would give me a good feeling even if no one paid me for

it

SS7 I would agree to a good plan to make a better life for the poor even if it costs

me money

357

329

339

302

345

363

350

904

872

949

982

888

1049

895

508

584

454

531

412

368

546

- 35 -

Table 2

Background of Respondents (n = 2868)

Variables Characteristics Respondents

Sex

Male

Female

NA

1209

1622

37

422

566

13

Length of service

(years)

0 ~ 10

10 ~ 20

20 ~ 30

30+

NA

1126

723

484

491

44

393

252

169

171

15

Organizational status

―Do you supervise employees

No

Yes

NA

1472

1076

320

513

375

112

Country

Australia

Belgium

China

Denmark

France

Italy

Korea

Lithuania

The Netherlands

Switzerland

United Kingdom

USA

249

214

230

249

266

162

253

236

249

250

260

250

87

75

80

87

93

56

88

82

87

87

91

87 Note NA = no answer

- 36 -

Table 3

Standardized factor loadings and correlations for the 31-item PSM index (n = 2868)

Dimensions and items SFL (MLE) SFL (D-WLS) Alpha

APP

APP1

APP2

APP4

APP5

APP6

APP7

597

486

455

558

490

600

657

552

514

625

549

665

699

CPV1

CPI1

CPI2

CPI3

CPI4

588

660

488

289

653

732

535

294

542

CPV2

CPV1

CPV2

CPV3

CPV4

CPV5

CPV6

CPV7

CPV8

CPV9

611

546

436

393

418

534

598

496

460

690

631

505

457

486

601

669

573

530

745

COM

COM1

COM2

COM3

COM5

COM6

496

569

498

597

607

516

651

538

670

684

687

SS

SS1

SS2

SS3

SS4

SS5

SS6

SS7

567

736

558

728

467

409

663

660

762

591

695

515

494

709

782

Measures of fit (MLE) χsup2 (df = 424) = 43699 p lt 001 CFI = 835 GFI = 899

RMSEA = 057 [055 059]

Measure of fit (D-WLS) χsup2 (df = 424) = 39247 p lt 001 CFI = 971 GFI = 976

RMSEA = 054 [052 055]

Inter-factor correlations MLE (D-WLS)

1 2 3 4

1 APS

2 CPV1 996 (988)

3 CPV2 902 (890) 798 (798)

4 COM 787 (783) 758 (762) 779 (783)

5 SS 700 (742) 800 (859) 509 (537) 720 (729)

Note SFL = standardized factor loading All factor loadings and correlations are significant at p lt 001

- 37 -

Table 4

Standardized factor loadings and correlations for the 16-item PSM index (n = 2868)

Dimensions and items SFL

(MLE)

SFL

(D-WLS)

APS

APP5 I admire people who initiate or are involved in activities to aid my

community

APP7 It is important to contribute to activities that tackle social problems

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

611

592

637

666

665

669

690

751

CPV

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of

public services

CPV6 It is fundamental that the interests of future generations are taken

into account when developing public policies

CPV7 To act ethically is essential for public servants

612

546

551

590

698

631

618

658

COM

COM1 It is difficult for me to contain my feelings when I see people in

distress

COM2 I feel sympathetic to the plight of the underprivileged

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

490

546

596

604

515

608

669

679

SS

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS7 I would agree to a good plan to make a better life for the poor even if

it costs me money

763

571

766

678

807

629

747

755

Measures of fit (MLE)

χsup2 (df = 98) = 8490 p lt 001

CFI = 939 GFI = 965

RMSEA = 052 [049 055]

Measure of fit (D-WLS)

χsup2 (df = 98) = 6069 p lt 001

CFI = 988 GFI = 992

RMSEA = 043 [039 046]

Alpha CR (D-WLS) Inter-factor correlations MLE (D-WLS)

1 APS 717 788 1 APS 2 CPV 3 COM

2 CPV 663 747 848 (859)

3 COM 645 713 781 (791) 789 (795)

4 SS 784 826 565 (596) 462 (505) 715 (726)

Note SFL = standardized factor loading Alpha = Cronbachrsquos alpha CR = composite reliability

All factor loadings and correlations are significant at p lt 001

- 38 -

Table 5

Results of CFA in the individual countries Standardized factor loadings and measures of fit

Factors

amp Items

Australia Belgium China Denmark France Italy

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

672

756

799

785

804

943

710

703

597

790

799

730

584

675

685

681

731

742

578

748

674

643

765

739

645

704

678

708

812

779

632

604

765

766

672

853

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

552

593

560

699

727

632

624

605

560

713

709

645

702

602

616

724

686

735

567

560

715

654

605

811

608

658

587

696

755

698

673

620

523

782

697

654

COM

COM1

COM2

COM3

COM5

COM6

623

584

662

688

701

739

732

608

475

808

666

581

578

616

605

649

670

673

699

751

737

679

816

776

568

546

801

626

673

850

637

546

741

715

601

822

SS

SS2

SS3

SS4

SS7

654

794

806

892

767

813

727

693

799

793

725

838

857

820

773

896

855

842

806

749

792

829

753

884

847

731

753

889

816

753

815

692

664

795

782

734

Inter-factor

correlations

093

~

752

163

~

810

320

~

773

323

~

825

452

~

827

497

~

855

323

~

850

321

~

847

498

~

899

536

~

902

425

~

829

476

~

865

Sample size 249 214 230 249 266 162

Measures of Fit

χsup2 638 444 881 732 765 554 817 625 735 565 750 564

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 980 100 939 987 969 997 967 994 975 997 949 995

GFI 961 986 941 986 948 991 949 992 955 992 926 986

RMSEA 036 00 063 050 051 026 053 035 045 026 059 033

- 39 -

Table 5 (Continued)

Factors

amp Items

Korea Lithuania The Netherlands Switzerland UK USA

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

619

701

667

706

781

717

666

512

682

751

605

779

557

629

630

620

719

729

535

621

702

643

694

735

655

725

738

710

804

822

535

621

702

643

694

735

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

594

607

722

722

644

722

674

535

533

790

589

711

568

554

538

655

628

545

627

580

615

711

666

704

800

512

555

873

581

646

627

580

615

711

666

704

COM

COM1

COM2

COM3

COM5

COM6

694

677

550

763

736

627

557

639

428

550

696

534

737

623

508

766

616

608

659

592

683

703

661

769

682

583

729

739

644

803

659

592

683

703

661

769

SS

SS2

SS3

SS4

SS7

776

752

679

779

770

811

707

721

635

725

734

706

814

656

743

820

767

764

745

632

689

788

662

744

671

818

832

830

864

802

745

632

689

788

662

744

Inter-factor

correlations

432

~

726

466

~

746

376

~

788

406

~

902

377

~

852

455

~

836

637

~

818

660

~

828

317

~

852

360

~

852

637

~

818

660

~

828

Sample size 253 236 249 250 260 250

Measures of Fit

χsup2 896 573 612 462 773 616 815 715 769 542 815 715

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 951 996 975 100 958 993 957 991 972 998 957 991

GFI 942 988 960 989 951 987 948 990 955 990 948 990

RMSEA 059 028 034 00 050 034 053 044 048 022 053 044

Note All factor loadings and correlations are significant at p lt 001

- 40 -

Appendix 1 Dimensions and possible items of PSM (the first round)

1 Attraction to Public Participation

(1) I am interested in developing public programs helping my country or community (Kim 2009a)

(2) Sharing my views on public policies with others is attractive to me (Kim 2009a)

(3) I am satisfied when seeing people get benefits from the public programs I was involved in (Kim 2009a)

(4) I like to discuss political subjects with others (Giauque et al 2011)

(5) I think the governmental activities contribute to our welfare

(6) I like to initiate actions to help out my community

(7) I do not care much about politicians (Perry 1996)

2 Commitment to Public Values

Possible items for public interest

(1A) Meaningful public service is very important to me (Perry 1996)

(2A) It is important for me to contribute for the common good (Giauque et al 2011)

(3A) I recognize myself with the mission of protection of liberties and rights endorsed by the nation-state

(Giauque et al 2011)

(4A) I would prefer seeing public officials do what is best for the whole community even if it harmed my

interests (Perry 1996)

(5A) To me serving the public interest is more important than helping other people (Vandenabeele 2008b)

Possible items for public values

(1B) I recognize myself with the promotion of the equality of chances endorsed by the nation-state (Giauque et

al 2011)

(2B) It is fundamental that the public service provided to the citizen is regular and continuous (Giauque et al

2011)

(3B) It is fundamental that public service innovates to respond to the new needs of the citizensclients (Giauque

et al 2011)

(4B) The dignity and well-being of all should be the most important concerns in any society (Steenbergen

1996)

(5B) It is important that public servants account for all the costsexpenses they make (Vandenabeele 2008b)

(6B) We have to do everything in our power to pursue the goal of democracy

3 Compassion

(1) It is difficult for me to contain my feelings when I see people in distress (Perry 1996)

(2) I feel sympathetic to the plight of the underprivileged (Kim 2009a)

(3) I am often reminded by daily events how dependent we are on one another (Perry 1996)

(4) I feel compassion for others in the difficulties they are facing (Steffen and Masters 2005)

(5) It is not really my problem if others are in trouble and need help (Beutel and Marini 1995)

(6) I find it hard to be sympathetic toward starving people in foreign lands when there is so much trouble in our

own country (Beutel and Marini 1995)

(7) I get very upset when I see other people treated unfairly (Beutel and Marini 1995)

(8) I would agree to a good plan to make a better life for the poor even if it cost me money (Beutel and Marini

1995)

(9) To me patriotism includes seeing the welfare of others (Perry 1996)

(10) I care very much about other people

4 Self-Sacrifice

(1) Making a difference in society means more to me than personal achievements (Perry 1996)

(2) I am prepared to make enormous sacrifices for the good of society (Perry 1996)

(3) I believe in putting duty before self (Perry 1996)

(4) I am one of those rare people who would risk personal loss to help someone else (Perry 1996)

(5) I think people should give back to society more than they get from it (Perry 1996)

(6) Serving other citizens would give me a good feeling even if no one paid me for it (Perry 1996)

(7) Much of what I do is for a cause bigger than myself (Perry 1996)

- 41 -

Appendix 2 Newly suggested items through the first round

1 Items for the Attraction to Public Participation Dimension

(1) In the public domain every person should be counted as equal

(2) I enjoy to think and argue in a political way

(3) I am satisfied when seeing my ideas being integrated in public policies

(4) I can easily relate to the way of thinking and arguing of politicians

(5) Working in the public sector I can better realize myself

2 Items for the Public Values Sub-Dimension

(1) Public service should give value for money

(2) Cost considerations should play an important role in decisions about public service provision

(3) It is fundamental that the interests of future generations are taken into account when developing public

policies

(4) To act with correct ethics is for public servants as important as their professional competence

(5) Primarily public servants should be accountable to the public and not to their superiors

3 Items for the Self-Sacrifice Dimension

(1) I would agree to a good plan to make a better life for the poor even if it costs me money

- 42 -

Appendix 3 An example of a summary of comments and suggestions (the second round)

On the item ―I am interested in developing public programs helping my country or community

A It is a good sentence but is it possible for everyone to ―develop public programs Even public officials are

limited to do this

B OK

C or the special group

D Double barrel and difficult to translate to my country ―I am interested in helping to improve public services

E Good but I feel that wording requires revision If we draw sample from local government then country is

redundant ―I am interested in developing public policy programs that aim in helping my community

F 4 (General appreciation from1 to 4) Not clear whether it refers to ones work or to ones concern as a citizen

―(In my work) I am interested in contributing to the development of public etchellip

G Instead of development I would use implementing

H Delete public which is redundant because of the characteristic helping my country or community but add

or implementing ―I am interested in developing or implementing programs helping my country or

community

I Referring to dimension Wording Similar to item 6

Note Names have been changed to alphabets in order to ensure the anonymity of participants

- 43 -

Appendix 4 Dimensions and Items of Public Service Motivation for Future Research

Dimensions and items

Attraction to Public Service (APS)

1 I am interested in helping to improve public service

2 I admire people who initiate or are involved in activities to aid my community

3 It is important to contribute to activities that tackle social problems

4 Meaningful public service is very important to me

5 It is important for me to contribute to the common good

Commitment to Public Values (CPV)

1 I think equal opportunities for citizens are very important

2 It is important that citizens can rely on the continuous provision of public services

3 It is fundamental that the interests of future generations are taken into account when

developing public policies

4 To act ethically is essential for public servants

5 I believe that public employees must always be aware of the legitimacy of their activities

6 I personally identify with the aim of protecting individual liberties and rights

Compassion (COM)

1 It is difficult for me to contain my feelings when I see people in distress

2 I feel sympathetic to the plight of the underprivileged

3 I empathize with other people who face difficulties

4 I get very upset when I see other people being treated unfairly

5 Considering the welfare of others is very important

Self-Sacrifice (SS)

1 I am prepared to make sacrifices for the good of society

2 I believe in putting civic duty before self

3 I am willing to risk personal loss to help society

4 I would agree to a good plan to make a better life for the poor even if it costs me money

- 26 -

REFERENCES

Anderson James C and David D Gerbing 1988 Structural Equation Modeling in Practice

A Review and Recommended Two-Step Approach Psychological Bulletin 103

411-423

Arbuckle James L 2009 Amos 180 Userrsquos Guide Crawfordville FL Amos Development

Co

Beck Joslashrgensen Torben and Barry Bozeman 2007 Public Values An Inventory

Administration amp Society 39(3) 354-381

Beutel Ann M and Margaret Mooney Marini 1995 Gender and values American

Sociological Review 60 436 - 448

Bobko Philip Philip L Roth and Maury A Buster 2007 The Usefulness of Unit Weights in

Creating Composite Scores A Literature Review Application to Content Validity and

Meta-Analysis Organizational Research Methods 10(4) 689-709

Bollen Kenneth A 1989 Structural Equations with Latent Variables New York Wiley

Bollen K A and R Lennox 1991 Conventional Wisdom on Measurement A Structural

Equation Perspective Psychological Bulletin 110(2) 305-314

Byrne Barbara M 2010 Structural Equation Modeling with Amos Basic Concepts

Applications and Programming 2nd

ed New York Routledge

Camilleri Emanuel 2006 Towards Developing an Organizational Commitment ― Public

Service Motivation Model for Maltese Public Service Employees Public Policy

and Administration 21 63-83

Castaing Seacutebastien 2006 The Effects of Psychological Contract Fulfillment and Public

Service Motivation on Organizational Commitment in the French Civil Service

Public Policy and Administration 21 84-98

Cenfetelli Ronald T and Geneviegraveve Bassellier 2009 Interpretation of Formative

Measurement in Information Systems Research MIS Quarterly 33(4) 689-707

Charles Michael B W Martin de Jong and Neal Ryan 2011 Public Values in Western

- 27 -

Europe A Temporal Perspective American Review of Public Administration 41

75-91

Chin Wynne W 2010 How to Write Up and Report PLS Analysis In Handbook of Partical

Least Squares edited by V Esposito Vinzi et al 655-690 Berlin Springer-Verlag

Coursey David H and Sanjay K Pandey 2007a Public Service Motivation Measurement

Testing an Abridged Version of Perryrsquos Proposed Scale Administration amp Society

39 547-568

Coursey David H and Sanjay K Pandey 2007b Content Domain Measurement and

Validity of the Red Tape Concept American Review of Public Administration 37

342-361

Coursey David H James L Perry Jeffrey L Brudney and Laura Littlepage 2008

Psychometric Verification of Perryrsquos Public Service Motivation Instrument Results

for Volunteer Exemplars Review of Public Personnel Administration 28 79-90

Davis Paul and Karen West 2009 What Do Public Values Mean for Public Action Putting

Public Values in Their Plural Place American Review of Public Administration 39

602-618

DeHart-Davis Leisha Justin Marlowe and Sanjay K Pandey 2006 Gender Dimensions of

Public Service Motivation Public Administration Review 66 873-887

Diamantopoulos Adamantios 2006 The error term in formative measurement models

Interpretation and modeling implications Journal of Modeling in Management 1(1)

7-17

Diamantopoulos Adamantios and Judy A Siguaw 2006 Formative versus Reflective

Indicators in Organizational Measure Development A Comparison and Empirical

Illustration British Journal of Management 17 263-282

Diamantopoulos Adamantios and Heidi M Winklhofer 2001 Index Construction with

Formative Indicators An Alternative to Scale Development Journal of Marketing

Research 38 269-277

Edwards Jeffrey R 2010 The Fallacy of Formative Measurement Organizational Research

- 28 -

Methods Online First published on August 19 2010 as

doi1011771094428110378369

Edwards Jeffrey R and Richard P Bagozzi 2000 On the Nature and Direction of

Relationships between Constructs and Measures Psychological Methods 5 155-

174

Flora David B and Patrick J Curran 2004 An Empirical Evaluation of Alternative

Methods of Estimation for Confirmatory Factor Analysis with Ordinal Data

Psychological Methods 9 466-491

Fornell Claes and Fred L Bookstein 1982 Two structural equations models LISREL and

PLS applied to customer exit-voice theory Journal of Marketing Research 19 440-

452

Giauque David Adrian Ritz Freacutedeacuteric Varone Simon Anderfuhren-Biget and Christian

Waldner 2011 Putting Public Service Motivation into Context A Balance between

Universalism and Particularism Forthcoming in International Review of

Administrative Sciences 77(2)

Goumltz Oliver Kerstin Liehr-Gobbers and Manfred Krafft 2010 Evaluation of Structural

Equation Models Using the Partial Least Squares (PLS) Approach In V Esposito

Vinzi et al (eds) Handbook of Partial Least Squares 691-711 DOI 101007978-3-

540-32827-8_30 Springer-Verlag Berlin Heidelberg

Guinot Christiane Julie Latreille and Michel Tenenhaus 2001 PLS Path Modelling and

Multiple Table Analysis Application to the Cosmetic Habits of Women in Ile-de-

France Chemometrics and Intelligent Laboratory Systems 58 247-259

Hair Joseph F Jr William C Black Barry J Babin and Rolph E Anderson 2010

Multivariate Data Analysis 7th

ed Upper Saddle River NJ Pearson Education

Hardin Andrew M Jerry Cha-Jan Chang Mark A Fuller and Gholamreza Torkzadeh 2010

Formative Measurement and Academic Research In Search of Measurement

Theory Educational and Psychological Measurement Online First published on

June 11 2010 as doi101177001316440370208

- 29 -

Howell Roy D Einar Breivik and James B Wilcox 2007a Reconsidering Formative

Measurement Psychological Methods 12 205-218

Howell Roy D Einar Breivik and James B Wilcox 2007b Is Formative Measurement

Really Measurement Reply to Bollen (2007) and Bagozzi (2007) Psychological

Methods 12 238-245

Jarvis Cheryl Burke Scott B MacKenzie and Philip M Podsakoff 2003 A Critical Review

of Construct Indicators and Measurement Model Misspecification in Marketing and

Consumer Research Journal of Consumer Research 30 199-218

Joumlreskog Karl G and Dag Soumlrbom 2005 LISREL 872 (Computer Software) Lincolnwood

IL Scientific Software International

Kim Gimun Bongsik Shin and Varun Grover 2010 Investigating Two Contradictory Views

of Formative Measurement in Information Systems Research MIS Quarterly 34

345-365

Kim Sangmook 2009a Revising Perryrsquos Measurement Scale of Public Service Motivation

American Review of Public Administration 39 149-163

Kim Sangmook 2009b Testing the Structure of Public Service Motivation in Korea A

Research Note Journal of Public Administration Research and Theory 19 839-

851

Kim Sangmook 2010 Testing a Revised Measurement of Public Service Motivation

Reflective versus Formative Specification Journal of Public Administration

Research and Theory Doi101093jopartmuq048

Kim Sangmook and Wouter Vandenabeele 2010 A strategy for building public service

motivation research internationally Public Administration Review 70 701-709

Kline Rex B 2005 Principles and Practice of Structural Equation Modeling 2nd

ed New

York Guilford Press

Law Kenneth and Chi-Sum Wong 1999 Multidimensional Constructs in Structural

Equation Analysis An Illustration using the Job Perception and Job Satisfaction

Constructs Journal of Management 25 143-160

- 30 -

Leisink Peter and Bram Steijn 2009 Public Service Motivation and Job Performance of

Public Sector Employees in the Netherlands International Review of

Administrative Sciences 75 35-52

Liu Bangcheng Chun Hui Jin Hu Wensheng Yang and Xinli Yu 2011 How well can

Public Service Motivation Connect with Occupational Intention International

Review of Administrative Sciences 77 191-211

Liu Bangcheng Ningyu Tang and Xiaomei Zhu 2008 Public Service Motivation and Job

Satisfaction in China An Investigation of Generalisability and Instrumentality

International Journal of Manpower 29 684-699

MacCallum Robert C and Michael W Browne 1993 The Use of Causal Indicators in

Covariance Structure Models Some Practical Issues Psychological Bulletin 114

533-541

MacKenzie Scott B 2003 The dangers of poor construct conceptualization Journal of the

Academy of Marketing Science 31(3) 323-326

MacKenzie Scott B Philip M Podsakoff and Cheryl Burke Jarvis 2005 The Problem of

Measurement Model Misspecification in Behavioral and Organizational Research

and Some Recommended Solutions Journal of Applied Psychology 90 710-730

Martinsen Dorte Sindbjerg and Torben Beck Joslashrgensen 2010 Accountability as a

Differentiated Value in Supranational Governance American Review of Public

Administration 40 742-760

Moynihan Donald P and Sanjay K Pandey 2007a The Role of Organizations in Fostering

Public Service Motivation Public Administration Review 67 40-53

Moynihan Donald P and Sanjay K Pandey 2007b Finding Workable Levers over Work

Motivation Comparing Job Satisfaction Job Involvement and Organizational

Commitment Administration amp Society 39 803-832

Pandey Sanjay K and Edmund C Stazyk 2008 Antecedents and Correlates of Public

Service Motivation In Motivation in Public Management edited by James L Perry

and Annie Hondeghem 101-117 Oxford Oxford University Press

- 31 -

Perry James L 1996 Measuring Public Service Motivation An Assessment of Construct

Reliability and Validity Journal of Public Administration Research and Theory 6

5-22

Perry James L 1997 Antecedents of Public Service Motivation Journal of Public

Administration Research and Theory 7 181-197

Perry James L 2000 Bringing Society in Toward a Theory of Public Service Motivation

Journal of Public Administration Research and Theory 10 471-488

Perry James L and Annie Hondeghem 2008a Motivation in Public Management Oxford

Oxford University Press

Perry James L and Annie Hondeghem 2008b Directions for Future Theory and Research

In Motivation in Public Management edited by James L Perry and Annie

Hondeghem 294-313 Oxford Oxford University Press

Perry James L Annie Hondeghem and Lois Recascino Wise 2010 Revisiting the

Motivational Bases of Public Service Twenty Years of Research and an Agenda for

the Future Public Administration Review 70(5) 681-690

Perry James L and Wouter Vandenabeele 2008 Behavioral dynamics Institutions

Identities and Self-Regulation In Motivation in Public Management edited by

James L Perry and Annie Hondeghem 56-79 Oxford Oxford University Press

Perry James L and Lois Recascino Wise 1990 The Motivational Bases of Public Service

Public Administration Review 50 367-373

Rainey Hal G 1982 Reward Preference among Public and Private Managers In Search of

the Service Ethic American Review of Public Administration 16 288-302

Ringle Christian M Oliver Goumltz Martin Wetzels and Bradley Wilson 2009 On the Use of

Formative Measurement Specifications in Structural Equation Modeling A Monte

Carlo Simulation Study to Compare Covariance-Based and Partial Least Squares

Model Estimation Methodologies Faculty of Economics and Business Administration

Maastricht Univeristy

Ringle C M S Wende and S Will 2005 Smart PLS 20 (M3) Beta

- 32 -

httpwwwsmartplsde

Ritz Adrian 2011 Attraction to Public Policy Making A Qualitative Inquiry into

Improvements in PSM Measurement Forthcoming in Public Administration

Spector Paul E 1992 Summated Rating Scales Construction An Introduction Newbury

Park CA Sage

Steenbergen Marco R 1996 Compassion and American public opinion An analysis of the

NES Humanitarianism Scale Report to the Board of Overseers of the National

Election Studies

Steffen Patrick R and Kevin S Masters 2005 Does compassion mediate the intrinsic

religion-health relationship Annals of Behavioral Medicine 30 217-223

Taylor Jeannette 2007 The Impact of Public Service Motives on Work Outcomes in

Australia A Comparative Multi-Dimensional Analysis Public Administration 85

931-959

Vandenabeele Wouter 2007 Towards a Theory of Public Service Motivation An

Institutional Approach Public Management Review 9 545-556

Vandenabeele Wouter 2008a Government Calling Public Service Motivation as an Element

in Selecting Government as an Employer of Choice Public Administration 86

1089-1105

Vandenabeele Wouter 2008b Development of a Public Service Motivation Measurement

Scale Corroborating and Extending Perryrsquos Measurement Instrument International

Public Management Journal 11 143-167

Vandenabeele Wouter 2009 The Mediating Effect of Job Satisfaction and Organizational

Commitment on Self-Reported Performance More Robust Evidence of the PSM-

Performance Relationship International Review of Administrative Sciences 75 11-

34

Wilcox James B Roy D Howell and Einar Breivik 2008 Questions about Formative

Measurement Journal of Business Research 61 1219-1228

Wise Lois Recascino 2000 The Public Service Culture In Public Administration Concepts

- 33 -

and Cases 7th ed edited by Richard J Stillman II 342-353 Boston Houghton

Mifflin

Wright Bradley E 2008 Methodological Challenges associated with Public Service

Motivation Research In Motivation in Public Management edited by James L

Perry and Annie Hondeghem 80-98 Oxford Oxford University Press

Wright Bradley E and Sanjay K Pandey 2008 Public Service Motivation and the

Assumption of Person-Organization Fit Testing the Mediating Effect of Value

Congruence Administration amp Society 40 502-521

- 34 -

Table 1

Possible Indicators of PSM and Descriptive Statistics (n = 2868)

Dimensions and items Mean Standard

Deviation

Item-Total

Correlation

Attraction to Public Participation (APP)

APP1 I am interested in helping to improve public service

APP2 I am satisfied when I see people benefiting from the public programs I was

involved in

APP3 I like to discuss topics regarding public programs and policies with others

APP4 I believe that public sector activities contribute to our general welfare

APP5 I admire people who initiate or are involved in activities to aid my

community

APP6 Contributing to public programs and policies helps me realize myself

APP7 It is important to contribute to activities that tackle social problems

412

441

355

393

423

351

409

696

662

917

753

672

881

734

524

421

482

412

493

478

552

Commitment to Public Values (CPV)

Sub-dimension for Public Interests (CPV1)

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

CPI3 I would prefer seeing public officials do what is best for the whole

community even if it harmed my interests

CPI4 Serving the public interest is more important than helping a single individual

Sub-dimension for Public Values (CPV2)

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of public

services

CPV3 It is fundamental that public services respond to the needs of the citizens

CPV4 Decisions regarding public services should be democratic despite the time

and effort it takes

CPV5 Everybody is entitled to a good service even if it costs a lot of money

CPV6 It is fundamental that the interests of future generations are taken into

account when developing public policies

CPV7 To act ethically is essential for public servants

CPV8 I believe that public employees must always be aware of the legitimacy of

their activities

CPV9 I personally identify with the aim of protecting individual liberties and rights

418

409

348

337

428

416

433

394

367

426

441

434

414

695

662

899

995

730

704

710

852

932

689

681

692

763

525

596

480

251

484

437

356

325

376

439

462

402

392

Compassion (COM) COM1 It is difficult for me to contain my feelings when I see people in distress

COM2 I feel sympathetic to the plight of the underprivileged

COM3 I empathize with other people who face difficulties

COM4 I have little compassion for people in need who are unwilling to take the

first step to help themselves

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

354

387

388

299

412

391

951

825

707

1083

755

765

399

511

389

011

487

502

Self-Sacrifice (SS)

SS1 Making a difference to society means more to me than personal achievements

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS5 People should give back to society more than they get from it

SS6 Serving other citizens would give me a good feeling even if no one paid me for

it

SS7 I would agree to a good plan to make a better life for the poor even if it costs

me money

357

329

339

302

345

363

350

904

872

949

982

888

1049

895

508

584

454

531

412

368

546

- 35 -

Table 2

Background of Respondents (n = 2868)

Variables Characteristics Respondents

Sex

Male

Female

NA

1209

1622

37

422

566

13

Length of service

(years)

0 ~ 10

10 ~ 20

20 ~ 30

30+

NA

1126

723

484

491

44

393

252

169

171

15

Organizational status

―Do you supervise employees

No

Yes

NA

1472

1076

320

513

375

112

Country

Australia

Belgium

China

Denmark

France

Italy

Korea

Lithuania

The Netherlands

Switzerland

United Kingdom

USA

249

214

230

249

266

162

253

236

249

250

260

250

87

75

80

87

93

56

88

82

87

87

91

87 Note NA = no answer

- 36 -

Table 3

Standardized factor loadings and correlations for the 31-item PSM index (n = 2868)

Dimensions and items SFL (MLE) SFL (D-WLS) Alpha

APP

APP1

APP2

APP4

APP5

APP6

APP7

597

486

455

558

490

600

657

552

514

625

549

665

699

CPV1

CPI1

CPI2

CPI3

CPI4

588

660

488

289

653

732

535

294

542

CPV2

CPV1

CPV2

CPV3

CPV4

CPV5

CPV6

CPV7

CPV8

CPV9

611

546

436

393

418

534

598

496

460

690

631

505

457

486

601

669

573

530

745

COM

COM1

COM2

COM3

COM5

COM6

496

569

498

597

607

516

651

538

670

684

687

SS

SS1

SS2

SS3

SS4

SS5

SS6

SS7

567

736

558

728

467

409

663

660

762

591

695

515

494

709

782

Measures of fit (MLE) χsup2 (df = 424) = 43699 p lt 001 CFI = 835 GFI = 899

RMSEA = 057 [055 059]

Measure of fit (D-WLS) χsup2 (df = 424) = 39247 p lt 001 CFI = 971 GFI = 976

RMSEA = 054 [052 055]

Inter-factor correlations MLE (D-WLS)

1 2 3 4

1 APS

2 CPV1 996 (988)

3 CPV2 902 (890) 798 (798)

4 COM 787 (783) 758 (762) 779 (783)

5 SS 700 (742) 800 (859) 509 (537) 720 (729)

Note SFL = standardized factor loading All factor loadings and correlations are significant at p lt 001

- 37 -

Table 4

Standardized factor loadings and correlations for the 16-item PSM index (n = 2868)

Dimensions and items SFL

(MLE)

SFL

(D-WLS)

APS

APP5 I admire people who initiate or are involved in activities to aid my

community

APP7 It is important to contribute to activities that tackle social problems

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

611

592

637

666

665

669

690

751

CPV

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of

public services

CPV6 It is fundamental that the interests of future generations are taken

into account when developing public policies

CPV7 To act ethically is essential for public servants

612

546

551

590

698

631

618

658

COM

COM1 It is difficult for me to contain my feelings when I see people in

distress

COM2 I feel sympathetic to the plight of the underprivileged

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

490

546

596

604

515

608

669

679

SS

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS7 I would agree to a good plan to make a better life for the poor even if

it costs me money

763

571

766

678

807

629

747

755

Measures of fit (MLE)

χsup2 (df = 98) = 8490 p lt 001

CFI = 939 GFI = 965

RMSEA = 052 [049 055]

Measure of fit (D-WLS)

χsup2 (df = 98) = 6069 p lt 001

CFI = 988 GFI = 992

RMSEA = 043 [039 046]

Alpha CR (D-WLS) Inter-factor correlations MLE (D-WLS)

1 APS 717 788 1 APS 2 CPV 3 COM

2 CPV 663 747 848 (859)

3 COM 645 713 781 (791) 789 (795)

4 SS 784 826 565 (596) 462 (505) 715 (726)

Note SFL = standardized factor loading Alpha = Cronbachrsquos alpha CR = composite reliability

All factor loadings and correlations are significant at p lt 001

- 38 -

Table 5

Results of CFA in the individual countries Standardized factor loadings and measures of fit

Factors

amp Items

Australia Belgium China Denmark France Italy

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

672

756

799

785

804

943

710

703

597

790

799

730

584

675

685

681

731

742

578

748

674

643

765

739

645

704

678

708

812

779

632

604

765

766

672

853

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

552

593

560

699

727

632

624

605

560

713

709

645

702

602

616

724

686

735

567

560

715

654

605

811

608

658

587

696

755

698

673

620

523

782

697

654

COM

COM1

COM2

COM3

COM5

COM6

623

584

662

688

701

739

732

608

475

808

666

581

578

616

605

649

670

673

699

751

737

679

816

776

568

546

801

626

673

850

637

546

741

715

601

822

SS

SS2

SS3

SS4

SS7

654

794

806

892

767

813

727

693

799

793

725

838

857

820

773

896

855

842

806

749

792

829

753

884

847

731

753

889

816

753

815

692

664

795

782

734

Inter-factor

correlations

093

~

752

163

~

810

320

~

773

323

~

825

452

~

827

497

~

855

323

~

850

321

~

847

498

~

899

536

~

902

425

~

829

476

~

865

Sample size 249 214 230 249 266 162

Measures of Fit

χsup2 638 444 881 732 765 554 817 625 735 565 750 564

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 980 100 939 987 969 997 967 994 975 997 949 995

GFI 961 986 941 986 948 991 949 992 955 992 926 986

RMSEA 036 00 063 050 051 026 053 035 045 026 059 033

- 39 -

Table 5 (Continued)

Factors

amp Items

Korea Lithuania The Netherlands Switzerland UK USA

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

619

701

667

706

781

717

666

512

682

751

605

779

557

629

630

620

719

729

535

621

702

643

694

735

655

725

738

710

804

822

535

621

702

643

694

735

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

594

607

722

722

644

722

674

535

533

790

589

711

568

554

538

655

628

545

627

580

615

711

666

704

800

512

555

873

581

646

627

580

615

711

666

704

COM

COM1

COM2

COM3

COM5

COM6

694

677

550

763

736

627

557

639

428

550

696

534

737

623

508

766

616

608

659

592

683

703

661

769

682

583

729

739

644

803

659

592

683

703

661

769

SS

SS2

SS3

SS4

SS7

776

752

679

779

770

811

707

721

635

725

734

706

814

656

743

820

767

764

745

632

689

788

662

744

671

818

832

830

864

802

745

632

689

788

662

744

Inter-factor

correlations

432

~

726

466

~

746

376

~

788

406

~

902

377

~

852

455

~

836

637

~

818

660

~

828

317

~

852

360

~

852

637

~

818

660

~

828

Sample size 253 236 249 250 260 250

Measures of Fit

χsup2 896 573 612 462 773 616 815 715 769 542 815 715

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 951 996 975 100 958 993 957 991 972 998 957 991

GFI 942 988 960 989 951 987 948 990 955 990 948 990

RMSEA 059 028 034 00 050 034 053 044 048 022 053 044

Note All factor loadings and correlations are significant at p lt 001

- 40 -

Appendix 1 Dimensions and possible items of PSM (the first round)

1 Attraction to Public Participation

(1) I am interested in developing public programs helping my country or community (Kim 2009a)

(2) Sharing my views on public policies with others is attractive to me (Kim 2009a)

(3) I am satisfied when seeing people get benefits from the public programs I was involved in (Kim 2009a)

(4) I like to discuss political subjects with others (Giauque et al 2011)

(5) I think the governmental activities contribute to our welfare

(6) I like to initiate actions to help out my community

(7) I do not care much about politicians (Perry 1996)

2 Commitment to Public Values

Possible items for public interest

(1A) Meaningful public service is very important to me (Perry 1996)

(2A) It is important for me to contribute for the common good (Giauque et al 2011)

(3A) I recognize myself with the mission of protection of liberties and rights endorsed by the nation-state

(Giauque et al 2011)

(4A) I would prefer seeing public officials do what is best for the whole community even if it harmed my

interests (Perry 1996)

(5A) To me serving the public interest is more important than helping other people (Vandenabeele 2008b)

Possible items for public values

(1B) I recognize myself with the promotion of the equality of chances endorsed by the nation-state (Giauque et

al 2011)

(2B) It is fundamental that the public service provided to the citizen is regular and continuous (Giauque et al

2011)

(3B) It is fundamental that public service innovates to respond to the new needs of the citizensclients (Giauque

et al 2011)

(4B) The dignity and well-being of all should be the most important concerns in any society (Steenbergen

1996)

(5B) It is important that public servants account for all the costsexpenses they make (Vandenabeele 2008b)

(6B) We have to do everything in our power to pursue the goal of democracy

3 Compassion

(1) It is difficult for me to contain my feelings when I see people in distress (Perry 1996)

(2) I feel sympathetic to the plight of the underprivileged (Kim 2009a)

(3) I am often reminded by daily events how dependent we are on one another (Perry 1996)

(4) I feel compassion for others in the difficulties they are facing (Steffen and Masters 2005)

(5) It is not really my problem if others are in trouble and need help (Beutel and Marini 1995)

(6) I find it hard to be sympathetic toward starving people in foreign lands when there is so much trouble in our

own country (Beutel and Marini 1995)

(7) I get very upset when I see other people treated unfairly (Beutel and Marini 1995)

(8) I would agree to a good plan to make a better life for the poor even if it cost me money (Beutel and Marini

1995)

(9) To me patriotism includes seeing the welfare of others (Perry 1996)

(10) I care very much about other people

4 Self-Sacrifice

(1) Making a difference in society means more to me than personal achievements (Perry 1996)

(2) I am prepared to make enormous sacrifices for the good of society (Perry 1996)

(3) I believe in putting duty before self (Perry 1996)

(4) I am one of those rare people who would risk personal loss to help someone else (Perry 1996)

(5) I think people should give back to society more than they get from it (Perry 1996)

(6) Serving other citizens would give me a good feeling even if no one paid me for it (Perry 1996)

(7) Much of what I do is for a cause bigger than myself (Perry 1996)

- 41 -

Appendix 2 Newly suggested items through the first round

1 Items for the Attraction to Public Participation Dimension

(1) In the public domain every person should be counted as equal

(2) I enjoy to think and argue in a political way

(3) I am satisfied when seeing my ideas being integrated in public policies

(4) I can easily relate to the way of thinking and arguing of politicians

(5) Working in the public sector I can better realize myself

2 Items for the Public Values Sub-Dimension

(1) Public service should give value for money

(2) Cost considerations should play an important role in decisions about public service provision

(3) It is fundamental that the interests of future generations are taken into account when developing public

policies

(4) To act with correct ethics is for public servants as important as their professional competence

(5) Primarily public servants should be accountable to the public and not to their superiors

3 Items for the Self-Sacrifice Dimension

(1) I would agree to a good plan to make a better life for the poor even if it costs me money

- 42 -

Appendix 3 An example of a summary of comments and suggestions (the second round)

On the item ―I am interested in developing public programs helping my country or community

A It is a good sentence but is it possible for everyone to ―develop public programs Even public officials are

limited to do this

B OK

C or the special group

D Double barrel and difficult to translate to my country ―I am interested in helping to improve public services

E Good but I feel that wording requires revision If we draw sample from local government then country is

redundant ―I am interested in developing public policy programs that aim in helping my community

F 4 (General appreciation from1 to 4) Not clear whether it refers to ones work or to ones concern as a citizen

―(In my work) I am interested in contributing to the development of public etchellip

G Instead of development I would use implementing

H Delete public which is redundant because of the characteristic helping my country or community but add

or implementing ―I am interested in developing or implementing programs helping my country or

community

I Referring to dimension Wording Similar to item 6

Note Names have been changed to alphabets in order to ensure the anonymity of participants

- 43 -

Appendix 4 Dimensions and Items of Public Service Motivation for Future Research

Dimensions and items

Attraction to Public Service (APS)

1 I am interested in helping to improve public service

2 I admire people who initiate or are involved in activities to aid my community

3 It is important to contribute to activities that tackle social problems

4 Meaningful public service is very important to me

5 It is important for me to contribute to the common good

Commitment to Public Values (CPV)

1 I think equal opportunities for citizens are very important

2 It is important that citizens can rely on the continuous provision of public services

3 It is fundamental that the interests of future generations are taken into account when

developing public policies

4 To act ethically is essential for public servants

5 I believe that public employees must always be aware of the legitimacy of their activities

6 I personally identify with the aim of protecting individual liberties and rights

Compassion (COM)

1 It is difficult for me to contain my feelings when I see people in distress

2 I feel sympathetic to the plight of the underprivileged

3 I empathize with other people who face difficulties

4 I get very upset when I see other people being treated unfairly

5 Considering the welfare of others is very important

Self-Sacrifice (SS)

1 I am prepared to make sacrifices for the good of society

2 I believe in putting civic duty before self

3 I am willing to risk personal loss to help society

4 I would agree to a good plan to make a better life for the poor even if it costs me money

- 27 -

Europe A Temporal Perspective American Review of Public Administration 41

75-91

Chin Wynne W 2010 How to Write Up and Report PLS Analysis In Handbook of Partical

Least Squares edited by V Esposito Vinzi et al 655-690 Berlin Springer-Verlag

Coursey David H and Sanjay K Pandey 2007a Public Service Motivation Measurement

Testing an Abridged Version of Perryrsquos Proposed Scale Administration amp Society

39 547-568

Coursey David H and Sanjay K Pandey 2007b Content Domain Measurement and

Validity of the Red Tape Concept American Review of Public Administration 37

342-361

Coursey David H James L Perry Jeffrey L Brudney and Laura Littlepage 2008

Psychometric Verification of Perryrsquos Public Service Motivation Instrument Results

for Volunteer Exemplars Review of Public Personnel Administration 28 79-90

Davis Paul and Karen West 2009 What Do Public Values Mean for Public Action Putting

Public Values in Their Plural Place American Review of Public Administration 39

602-618

DeHart-Davis Leisha Justin Marlowe and Sanjay K Pandey 2006 Gender Dimensions of

Public Service Motivation Public Administration Review 66 873-887

Diamantopoulos Adamantios 2006 The error term in formative measurement models

Interpretation and modeling implications Journal of Modeling in Management 1(1)

7-17

Diamantopoulos Adamantios and Judy A Siguaw 2006 Formative versus Reflective

Indicators in Organizational Measure Development A Comparison and Empirical

Illustration British Journal of Management 17 263-282

Diamantopoulos Adamantios and Heidi M Winklhofer 2001 Index Construction with

Formative Indicators An Alternative to Scale Development Journal of Marketing

Research 38 269-277

Edwards Jeffrey R 2010 The Fallacy of Formative Measurement Organizational Research

- 28 -

Methods Online First published on August 19 2010 as

doi1011771094428110378369

Edwards Jeffrey R and Richard P Bagozzi 2000 On the Nature and Direction of

Relationships between Constructs and Measures Psychological Methods 5 155-

174

Flora David B and Patrick J Curran 2004 An Empirical Evaluation of Alternative

Methods of Estimation for Confirmatory Factor Analysis with Ordinal Data

Psychological Methods 9 466-491

Fornell Claes and Fred L Bookstein 1982 Two structural equations models LISREL and

PLS applied to customer exit-voice theory Journal of Marketing Research 19 440-

452

Giauque David Adrian Ritz Freacutedeacuteric Varone Simon Anderfuhren-Biget and Christian

Waldner 2011 Putting Public Service Motivation into Context A Balance between

Universalism and Particularism Forthcoming in International Review of

Administrative Sciences 77(2)

Goumltz Oliver Kerstin Liehr-Gobbers and Manfred Krafft 2010 Evaluation of Structural

Equation Models Using the Partial Least Squares (PLS) Approach In V Esposito

Vinzi et al (eds) Handbook of Partial Least Squares 691-711 DOI 101007978-3-

540-32827-8_30 Springer-Verlag Berlin Heidelberg

Guinot Christiane Julie Latreille and Michel Tenenhaus 2001 PLS Path Modelling and

Multiple Table Analysis Application to the Cosmetic Habits of Women in Ile-de-

France Chemometrics and Intelligent Laboratory Systems 58 247-259

Hair Joseph F Jr William C Black Barry J Babin and Rolph E Anderson 2010

Multivariate Data Analysis 7th

ed Upper Saddle River NJ Pearson Education

Hardin Andrew M Jerry Cha-Jan Chang Mark A Fuller and Gholamreza Torkzadeh 2010

Formative Measurement and Academic Research In Search of Measurement

Theory Educational and Psychological Measurement Online First published on

June 11 2010 as doi101177001316440370208

- 29 -

Howell Roy D Einar Breivik and James B Wilcox 2007a Reconsidering Formative

Measurement Psychological Methods 12 205-218

Howell Roy D Einar Breivik and James B Wilcox 2007b Is Formative Measurement

Really Measurement Reply to Bollen (2007) and Bagozzi (2007) Psychological

Methods 12 238-245

Jarvis Cheryl Burke Scott B MacKenzie and Philip M Podsakoff 2003 A Critical Review

of Construct Indicators and Measurement Model Misspecification in Marketing and

Consumer Research Journal of Consumer Research 30 199-218

Joumlreskog Karl G and Dag Soumlrbom 2005 LISREL 872 (Computer Software) Lincolnwood

IL Scientific Software International

Kim Gimun Bongsik Shin and Varun Grover 2010 Investigating Two Contradictory Views

of Formative Measurement in Information Systems Research MIS Quarterly 34

345-365

Kim Sangmook 2009a Revising Perryrsquos Measurement Scale of Public Service Motivation

American Review of Public Administration 39 149-163

Kim Sangmook 2009b Testing the Structure of Public Service Motivation in Korea A

Research Note Journal of Public Administration Research and Theory 19 839-

851

Kim Sangmook 2010 Testing a Revised Measurement of Public Service Motivation

Reflective versus Formative Specification Journal of Public Administration

Research and Theory Doi101093jopartmuq048

Kim Sangmook and Wouter Vandenabeele 2010 A strategy for building public service

motivation research internationally Public Administration Review 70 701-709

Kline Rex B 2005 Principles and Practice of Structural Equation Modeling 2nd

ed New

York Guilford Press

Law Kenneth and Chi-Sum Wong 1999 Multidimensional Constructs in Structural

Equation Analysis An Illustration using the Job Perception and Job Satisfaction

Constructs Journal of Management 25 143-160

- 30 -

Leisink Peter and Bram Steijn 2009 Public Service Motivation and Job Performance of

Public Sector Employees in the Netherlands International Review of

Administrative Sciences 75 35-52

Liu Bangcheng Chun Hui Jin Hu Wensheng Yang and Xinli Yu 2011 How well can

Public Service Motivation Connect with Occupational Intention International

Review of Administrative Sciences 77 191-211

Liu Bangcheng Ningyu Tang and Xiaomei Zhu 2008 Public Service Motivation and Job

Satisfaction in China An Investigation of Generalisability and Instrumentality

International Journal of Manpower 29 684-699

MacCallum Robert C and Michael W Browne 1993 The Use of Causal Indicators in

Covariance Structure Models Some Practical Issues Psychological Bulletin 114

533-541

MacKenzie Scott B 2003 The dangers of poor construct conceptualization Journal of the

Academy of Marketing Science 31(3) 323-326

MacKenzie Scott B Philip M Podsakoff and Cheryl Burke Jarvis 2005 The Problem of

Measurement Model Misspecification in Behavioral and Organizational Research

and Some Recommended Solutions Journal of Applied Psychology 90 710-730

Martinsen Dorte Sindbjerg and Torben Beck Joslashrgensen 2010 Accountability as a

Differentiated Value in Supranational Governance American Review of Public

Administration 40 742-760

Moynihan Donald P and Sanjay K Pandey 2007a The Role of Organizations in Fostering

Public Service Motivation Public Administration Review 67 40-53

Moynihan Donald P and Sanjay K Pandey 2007b Finding Workable Levers over Work

Motivation Comparing Job Satisfaction Job Involvement and Organizational

Commitment Administration amp Society 39 803-832

Pandey Sanjay K and Edmund C Stazyk 2008 Antecedents and Correlates of Public

Service Motivation In Motivation in Public Management edited by James L Perry

and Annie Hondeghem 101-117 Oxford Oxford University Press

- 31 -

Perry James L 1996 Measuring Public Service Motivation An Assessment of Construct

Reliability and Validity Journal of Public Administration Research and Theory 6

5-22

Perry James L 1997 Antecedents of Public Service Motivation Journal of Public

Administration Research and Theory 7 181-197

Perry James L 2000 Bringing Society in Toward a Theory of Public Service Motivation

Journal of Public Administration Research and Theory 10 471-488

Perry James L and Annie Hondeghem 2008a Motivation in Public Management Oxford

Oxford University Press

Perry James L and Annie Hondeghem 2008b Directions for Future Theory and Research

In Motivation in Public Management edited by James L Perry and Annie

Hondeghem 294-313 Oxford Oxford University Press

Perry James L Annie Hondeghem and Lois Recascino Wise 2010 Revisiting the

Motivational Bases of Public Service Twenty Years of Research and an Agenda for

the Future Public Administration Review 70(5) 681-690

Perry James L and Wouter Vandenabeele 2008 Behavioral dynamics Institutions

Identities and Self-Regulation In Motivation in Public Management edited by

James L Perry and Annie Hondeghem 56-79 Oxford Oxford University Press

Perry James L and Lois Recascino Wise 1990 The Motivational Bases of Public Service

Public Administration Review 50 367-373

Rainey Hal G 1982 Reward Preference among Public and Private Managers In Search of

the Service Ethic American Review of Public Administration 16 288-302

Ringle Christian M Oliver Goumltz Martin Wetzels and Bradley Wilson 2009 On the Use of

Formative Measurement Specifications in Structural Equation Modeling A Monte

Carlo Simulation Study to Compare Covariance-Based and Partial Least Squares

Model Estimation Methodologies Faculty of Economics and Business Administration

Maastricht Univeristy

Ringle C M S Wende and S Will 2005 Smart PLS 20 (M3) Beta

- 32 -

httpwwwsmartplsde

Ritz Adrian 2011 Attraction to Public Policy Making A Qualitative Inquiry into

Improvements in PSM Measurement Forthcoming in Public Administration

Spector Paul E 1992 Summated Rating Scales Construction An Introduction Newbury

Park CA Sage

Steenbergen Marco R 1996 Compassion and American public opinion An analysis of the

NES Humanitarianism Scale Report to the Board of Overseers of the National

Election Studies

Steffen Patrick R and Kevin S Masters 2005 Does compassion mediate the intrinsic

religion-health relationship Annals of Behavioral Medicine 30 217-223

Taylor Jeannette 2007 The Impact of Public Service Motives on Work Outcomes in

Australia A Comparative Multi-Dimensional Analysis Public Administration 85

931-959

Vandenabeele Wouter 2007 Towards a Theory of Public Service Motivation An

Institutional Approach Public Management Review 9 545-556

Vandenabeele Wouter 2008a Government Calling Public Service Motivation as an Element

in Selecting Government as an Employer of Choice Public Administration 86

1089-1105

Vandenabeele Wouter 2008b Development of a Public Service Motivation Measurement

Scale Corroborating and Extending Perryrsquos Measurement Instrument International

Public Management Journal 11 143-167

Vandenabeele Wouter 2009 The Mediating Effect of Job Satisfaction and Organizational

Commitment on Self-Reported Performance More Robust Evidence of the PSM-

Performance Relationship International Review of Administrative Sciences 75 11-

34

Wilcox James B Roy D Howell and Einar Breivik 2008 Questions about Formative

Measurement Journal of Business Research 61 1219-1228

Wise Lois Recascino 2000 The Public Service Culture In Public Administration Concepts

- 33 -

and Cases 7th ed edited by Richard J Stillman II 342-353 Boston Houghton

Mifflin

Wright Bradley E 2008 Methodological Challenges associated with Public Service

Motivation Research In Motivation in Public Management edited by James L

Perry and Annie Hondeghem 80-98 Oxford Oxford University Press

Wright Bradley E and Sanjay K Pandey 2008 Public Service Motivation and the

Assumption of Person-Organization Fit Testing the Mediating Effect of Value

Congruence Administration amp Society 40 502-521

- 34 -

Table 1

Possible Indicators of PSM and Descriptive Statistics (n = 2868)

Dimensions and items Mean Standard

Deviation

Item-Total

Correlation

Attraction to Public Participation (APP)

APP1 I am interested in helping to improve public service

APP2 I am satisfied when I see people benefiting from the public programs I was

involved in

APP3 I like to discuss topics regarding public programs and policies with others

APP4 I believe that public sector activities contribute to our general welfare

APP5 I admire people who initiate or are involved in activities to aid my

community

APP6 Contributing to public programs and policies helps me realize myself

APP7 It is important to contribute to activities that tackle social problems

412

441

355

393

423

351

409

696

662

917

753

672

881

734

524

421

482

412

493

478

552

Commitment to Public Values (CPV)

Sub-dimension for Public Interests (CPV1)

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

CPI3 I would prefer seeing public officials do what is best for the whole

community even if it harmed my interests

CPI4 Serving the public interest is more important than helping a single individual

Sub-dimension for Public Values (CPV2)

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of public

services

CPV3 It is fundamental that public services respond to the needs of the citizens

CPV4 Decisions regarding public services should be democratic despite the time

and effort it takes

CPV5 Everybody is entitled to a good service even if it costs a lot of money

CPV6 It is fundamental that the interests of future generations are taken into

account when developing public policies

CPV7 To act ethically is essential for public servants

CPV8 I believe that public employees must always be aware of the legitimacy of

their activities

CPV9 I personally identify with the aim of protecting individual liberties and rights

418

409

348

337

428

416

433

394

367

426

441

434

414

695

662

899

995

730

704

710

852

932

689

681

692

763

525

596

480

251

484

437

356

325

376

439

462

402

392

Compassion (COM) COM1 It is difficult for me to contain my feelings when I see people in distress

COM2 I feel sympathetic to the plight of the underprivileged

COM3 I empathize with other people who face difficulties

COM4 I have little compassion for people in need who are unwilling to take the

first step to help themselves

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

354

387

388

299

412

391

951

825

707

1083

755

765

399

511

389

011

487

502

Self-Sacrifice (SS)

SS1 Making a difference to society means more to me than personal achievements

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS5 People should give back to society more than they get from it

SS6 Serving other citizens would give me a good feeling even if no one paid me for

it

SS7 I would agree to a good plan to make a better life for the poor even if it costs

me money

357

329

339

302

345

363

350

904

872

949

982

888

1049

895

508

584

454

531

412

368

546

- 35 -

Table 2

Background of Respondents (n = 2868)

Variables Characteristics Respondents

Sex

Male

Female

NA

1209

1622

37

422

566

13

Length of service

(years)

0 ~ 10

10 ~ 20

20 ~ 30

30+

NA

1126

723

484

491

44

393

252

169

171

15

Organizational status

―Do you supervise employees

No

Yes

NA

1472

1076

320

513

375

112

Country

Australia

Belgium

China

Denmark

France

Italy

Korea

Lithuania

The Netherlands

Switzerland

United Kingdom

USA

249

214

230

249

266

162

253

236

249

250

260

250

87

75

80

87

93

56

88

82

87

87

91

87 Note NA = no answer

- 36 -

Table 3

Standardized factor loadings and correlations for the 31-item PSM index (n = 2868)

Dimensions and items SFL (MLE) SFL (D-WLS) Alpha

APP

APP1

APP2

APP4

APP5

APP6

APP7

597

486

455

558

490

600

657

552

514

625

549

665

699

CPV1

CPI1

CPI2

CPI3

CPI4

588

660

488

289

653

732

535

294

542

CPV2

CPV1

CPV2

CPV3

CPV4

CPV5

CPV6

CPV7

CPV8

CPV9

611

546

436

393

418

534

598

496

460

690

631

505

457

486

601

669

573

530

745

COM

COM1

COM2

COM3

COM5

COM6

496

569

498

597

607

516

651

538

670

684

687

SS

SS1

SS2

SS3

SS4

SS5

SS6

SS7

567

736

558

728

467

409

663

660

762

591

695

515

494

709

782

Measures of fit (MLE) χsup2 (df = 424) = 43699 p lt 001 CFI = 835 GFI = 899

RMSEA = 057 [055 059]

Measure of fit (D-WLS) χsup2 (df = 424) = 39247 p lt 001 CFI = 971 GFI = 976

RMSEA = 054 [052 055]

Inter-factor correlations MLE (D-WLS)

1 2 3 4

1 APS

2 CPV1 996 (988)

3 CPV2 902 (890) 798 (798)

4 COM 787 (783) 758 (762) 779 (783)

5 SS 700 (742) 800 (859) 509 (537) 720 (729)

Note SFL = standardized factor loading All factor loadings and correlations are significant at p lt 001

- 37 -

Table 4

Standardized factor loadings and correlations for the 16-item PSM index (n = 2868)

Dimensions and items SFL

(MLE)

SFL

(D-WLS)

APS

APP5 I admire people who initiate or are involved in activities to aid my

community

APP7 It is important to contribute to activities that tackle social problems

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

611

592

637

666

665

669

690

751

CPV

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of

public services

CPV6 It is fundamental that the interests of future generations are taken

into account when developing public policies

CPV7 To act ethically is essential for public servants

612

546

551

590

698

631

618

658

COM

COM1 It is difficult for me to contain my feelings when I see people in

distress

COM2 I feel sympathetic to the plight of the underprivileged

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

490

546

596

604

515

608

669

679

SS

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS7 I would agree to a good plan to make a better life for the poor even if

it costs me money

763

571

766

678

807

629

747

755

Measures of fit (MLE)

χsup2 (df = 98) = 8490 p lt 001

CFI = 939 GFI = 965

RMSEA = 052 [049 055]

Measure of fit (D-WLS)

χsup2 (df = 98) = 6069 p lt 001

CFI = 988 GFI = 992

RMSEA = 043 [039 046]

Alpha CR (D-WLS) Inter-factor correlations MLE (D-WLS)

1 APS 717 788 1 APS 2 CPV 3 COM

2 CPV 663 747 848 (859)

3 COM 645 713 781 (791) 789 (795)

4 SS 784 826 565 (596) 462 (505) 715 (726)

Note SFL = standardized factor loading Alpha = Cronbachrsquos alpha CR = composite reliability

All factor loadings and correlations are significant at p lt 001

- 38 -

Table 5

Results of CFA in the individual countries Standardized factor loadings and measures of fit

Factors

amp Items

Australia Belgium China Denmark France Italy

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

672

756

799

785

804

943

710

703

597

790

799

730

584

675

685

681

731

742

578

748

674

643

765

739

645

704

678

708

812

779

632

604

765

766

672

853

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

552

593

560

699

727

632

624

605

560

713

709

645

702

602

616

724

686

735

567

560

715

654

605

811

608

658

587

696

755

698

673

620

523

782

697

654

COM

COM1

COM2

COM3

COM5

COM6

623

584

662

688

701

739

732

608

475

808

666

581

578

616

605

649

670

673

699

751

737

679

816

776

568

546

801

626

673

850

637

546

741

715

601

822

SS

SS2

SS3

SS4

SS7

654

794

806

892

767

813

727

693

799

793

725

838

857

820

773

896

855

842

806

749

792

829

753

884

847

731

753

889

816

753

815

692

664

795

782

734

Inter-factor

correlations

093

~

752

163

~

810

320

~

773

323

~

825

452

~

827

497

~

855

323

~

850

321

~

847

498

~

899

536

~

902

425

~

829

476

~

865

Sample size 249 214 230 249 266 162

Measures of Fit

χsup2 638 444 881 732 765 554 817 625 735 565 750 564

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 980 100 939 987 969 997 967 994 975 997 949 995

GFI 961 986 941 986 948 991 949 992 955 992 926 986

RMSEA 036 00 063 050 051 026 053 035 045 026 059 033

- 39 -

Table 5 (Continued)

Factors

amp Items

Korea Lithuania The Netherlands Switzerland UK USA

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

619

701

667

706

781

717

666

512

682

751

605

779

557

629

630

620

719

729

535

621

702

643

694

735

655

725

738

710

804

822

535

621

702

643

694

735

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

594

607

722

722

644

722

674

535

533

790

589

711

568

554

538

655

628

545

627

580

615

711

666

704

800

512

555

873

581

646

627

580

615

711

666

704

COM

COM1

COM2

COM3

COM5

COM6

694

677

550

763

736

627

557

639

428

550

696

534

737

623

508

766

616

608

659

592

683

703

661

769

682

583

729

739

644

803

659

592

683

703

661

769

SS

SS2

SS3

SS4

SS7

776

752

679

779

770

811

707

721

635

725

734

706

814

656

743

820

767

764

745

632

689

788

662

744

671

818

832

830

864

802

745

632

689

788

662

744

Inter-factor

correlations

432

~

726

466

~

746

376

~

788

406

~

902

377

~

852

455

~

836

637

~

818

660

~

828

317

~

852

360

~

852

637

~

818

660

~

828

Sample size 253 236 249 250 260 250

Measures of Fit

χsup2 896 573 612 462 773 616 815 715 769 542 815 715

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 951 996 975 100 958 993 957 991 972 998 957 991

GFI 942 988 960 989 951 987 948 990 955 990 948 990

RMSEA 059 028 034 00 050 034 053 044 048 022 053 044

Note All factor loadings and correlations are significant at p lt 001

- 40 -

Appendix 1 Dimensions and possible items of PSM (the first round)

1 Attraction to Public Participation

(1) I am interested in developing public programs helping my country or community (Kim 2009a)

(2) Sharing my views on public policies with others is attractive to me (Kim 2009a)

(3) I am satisfied when seeing people get benefits from the public programs I was involved in (Kim 2009a)

(4) I like to discuss political subjects with others (Giauque et al 2011)

(5) I think the governmental activities contribute to our welfare

(6) I like to initiate actions to help out my community

(7) I do not care much about politicians (Perry 1996)

2 Commitment to Public Values

Possible items for public interest

(1A) Meaningful public service is very important to me (Perry 1996)

(2A) It is important for me to contribute for the common good (Giauque et al 2011)

(3A) I recognize myself with the mission of protection of liberties and rights endorsed by the nation-state

(Giauque et al 2011)

(4A) I would prefer seeing public officials do what is best for the whole community even if it harmed my

interests (Perry 1996)

(5A) To me serving the public interest is more important than helping other people (Vandenabeele 2008b)

Possible items for public values

(1B) I recognize myself with the promotion of the equality of chances endorsed by the nation-state (Giauque et

al 2011)

(2B) It is fundamental that the public service provided to the citizen is regular and continuous (Giauque et al

2011)

(3B) It is fundamental that public service innovates to respond to the new needs of the citizensclients (Giauque

et al 2011)

(4B) The dignity and well-being of all should be the most important concerns in any society (Steenbergen

1996)

(5B) It is important that public servants account for all the costsexpenses they make (Vandenabeele 2008b)

(6B) We have to do everything in our power to pursue the goal of democracy

3 Compassion

(1) It is difficult for me to contain my feelings when I see people in distress (Perry 1996)

(2) I feel sympathetic to the plight of the underprivileged (Kim 2009a)

(3) I am often reminded by daily events how dependent we are on one another (Perry 1996)

(4) I feel compassion for others in the difficulties they are facing (Steffen and Masters 2005)

(5) It is not really my problem if others are in trouble and need help (Beutel and Marini 1995)

(6) I find it hard to be sympathetic toward starving people in foreign lands when there is so much trouble in our

own country (Beutel and Marini 1995)

(7) I get very upset when I see other people treated unfairly (Beutel and Marini 1995)

(8) I would agree to a good plan to make a better life for the poor even if it cost me money (Beutel and Marini

1995)

(9) To me patriotism includes seeing the welfare of others (Perry 1996)

(10) I care very much about other people

4 Self-Sacrifice

(1) Making a difference in society means more to me than personal achievements (Perry 1996)

(2) I am prepared to make enormous sacrifices for the good of society (Perry 1996)

(3) I believe in putting duty before self (Perry 1996)

(4) I am one of those rare people who would risk personal loss to help someone else (Perry 1996)

(5) I think people should give back to society more than they get from it (Perry 1996)

(6) Serving other citizens would give me a good feeling even if no one paid me for it (Perry 1996)

(7) Much of what I do is for a cause bigger than myself (Perry 1996)

- 41 -

Appendix 2 Newly suggested items through the first round

1 Items for the Attraction to Public Participation Dimension

(1) In the public domain every person should be counted as equal

(2) I enjoy to think and argue in a political way

(3) I am satisfied when seeing my ideas being integrated in public policies

(4) I can easily relate to the way of thinking and arguing of politicians

(5) Working in the public sector I can better realize myself

2 Items for the Public Values Sub-Dimension

(1) Public service should give value for money

(2) Cost considerations should play an important role in decisions about public service provision

(3) It is fundamental that the interests of future generations are taken into account when developing public

policies

(4) To act with correct ethics is for public servants as important as their professional competence

(5) Primarily public servants should be accountable to the public and not to their superiors

3 Items for the Self-Sacrifice Dimension

(1) I would agree to a good plan to make a better life for the poor even if it costs me money

- 42 -

Appendix 3 An example of a summary of comments and suggestions (the second round)

On the item ―I am interested in developing public programs helping my country or community

A It is a good sentence but is it possible for everyone to ―develop public programs Even public officials are

limited to do this

B OK

C or the special group

D Double barrel and difficult to translate to my country ―I am interested in helping to improve public services

E Good but I feel that wording requires revision If we draw sample from local government then country is

redundant ―I am interested in developing public policy programs that aim in helping my community

F 4 (General appreciation from1 to 4) Not clear whether it refers to ones work or to ones concern as a citizen

―(In my work) I am interested in contributing to the development of public etchellip

G Instead of development I would use implementing

H Delete public which is redundant because of the characteristic helping my country or community but add

or implementing ―I am interested in developing or implementing programs helping my country or

community

I Referring to dimension Wording Similar to item 6

Note Names have been changed to alphabets in order to ensure the anonymity of participants

- 43 -

Appendix 4 Dimensions and Items of Public Service Motivation for Future Research

Dimensions and items

Attraction to Public Service (APS)

1 I am interested in helping to improve public service

2 I admire people who initiate or are involved in activities to aid my community

3 It is important to contribute to activities that tackle social problems

4 Meaningful public service is very important to me

5 It is important for me to contribute to the common good

Commitment to Public Values (CPV)

1 I think equal opportunities for citizens are very important

2 It is important that citizens can rely on the continuous provision of public services

3 It is fundamental that the interests of future generations are taken into account when

developing public policies

4 To act ethically is essential for public servants

5 I believe that public employees must always be aware of the legitimacy of their activities

6 I personally identify with the aim of protecting individual liberties and rights

Compassion (COM)

1 It is difficult for me to contain my feelings when I see people in distress

2 I feel sympathetic to the plight of the underprivileged

3 I empathize with other people who face difficulties

4 I get very upset when I see other people being treated unfairly

5 Considering the welfare of others is very important

Self-Sacrifice (SS)

1 I am prepared to make sacrifices for the good of society

2 I believe in putting civic duty before self

3 I am willing to risk personal loss to help society

4 I would agree to a good plan to make a better life for the poor even if it costs me money

- 28 -

Methods Online First published on August 19 2010 as

doi1011771094428110378369

Edwards Jeffrey R and Richard P Bagozzi 2000 On the Nature and Direction of

Relationships between Constructs and Measures Psychological Methods 5 155-

174

Flora David B and Patrick J Curran 2004 An Empirical Evaluation of Alternative

Methods of Estimation for Confirmatory Factor Analysis with Ordinal Data

Psychological Methods 9 466-491

Fornell Claes and Fred L Bookstein 1982 Two structural equations models LISREL and

PLS applied to customer exit-voice theory Journal of Marketing Research 19 440-

452

Giauque David Adrian Ritz Freacutedeacuteric Varone Simon Anderfuhren-Biget and Christian

Waldner 2011 Putting Public Service Motivation into Context A Balance between

Universalism and Particularism Forthcoming in International Review of

Administrative Sciences 77(2)

Goumltz Oliver Kerstin Liehr-Gobbers and Manfred Krafft 2010 Evaluation of Structural

Equation Models Using the Partial Least Squares (PLS) Approach In V Esposito

Vinzi et al (eds) Handbook of Partial Least Squares 691-711 DOI 101007978-3-

540-32827-8_30 Springer-Verlag Berlin Heidelberg

Guinot Christiane Julie Latreille and Michel Tenenhaus 2001 PLS Path Modelling and

Multiple Table Analysis Application to the Cosmetic Habits of Women in Ile-de-

France Chemometrics and Intelligent Laboratory Systems 58 247-259

Hair Joseph F Jr William C Black Barry J Babin and Rolph E Anderson 2010

Multivariate Data Analysis 7th

ed Upper Saddle River NJ Pearson Education

Hardin Andrew M Jerry Cha-Jan Chang Mark A Fuller and Gholamreza Torkzadeh 2010

Formative Measurement and Academic Research In Search of Measurement

Theory Educational and Psychological Measurement Online First published on

June 11 2010 as doi101177001316440370208

- 29 -

Howell Roy D Einar Breivik and James B Wilcox 2007a Reconsidering Formative

Measurement Psychological Methods 12 205-218

Howell Roy D Einar Breivik and James B Wilcox 2007b Is Formative Measurement

Really Measurement Reply to Bollen (2007) and Bagozzi (2007) Psychological

Methods 12 238-245

Jarvis Cheryl Burke Scott B MacKenzie and Philip M Podsakoff 2003 A Critical Review

of Construct Indicators and Measurement Model Misspecification in Marketing and

Consumer Research Journal of Consumer Research 30 199-218

Joumlreskog Karl G and Dag Soumlrbom 2005 LISREL 872 (Computer Software) Lincolnwood

IL Scientific Software International

Kim Gimun Bongsik Shin and Varun Grover 2010 Investigating Two Contradictory Views

of Formative Measurement in Information Systems Research MIS Quarterly 34

345-365

Kim Sangmook 2009a Revising Perryrsquos Measurement Scale of Public Service Motivation

American Review of Public Administration 39 149-163

Kim Sangmook 2009b Testing the Structure of Public Service Motivation in Korea A

Research Note Journal of Public Administration Research and Theory 19 839-

851

Kim Sangmook 2010 Testing a Revised Measurement of Public Service Motivation

Reflective versus Formative Specification Journal of Public Administration

Research and Theory Doi101093jopartmuq048

Kim Sangmook and Wouter Vandenabeele 2010 A strategy for building public service

motivation research internationally Public Administration Review 70 701-709

Kline Rex B 2005 Principles and Practice of Structural Equation Modeling 2nd

ed New

York Guilford Press

Law Kenneth and Chi-Sum Wong 1999 Multidimensional Constructs in Structural

Equation Analysis An Illustration using the Job Perception and Job Satisfaction

Constructs Journal of Management 25 143-160

- 30 -

Leisink Peter and Bram Steijn 2009 Public Service Motivation and Job Performance of

Public Sector Employees in the Netherlands International Review of

Administrative Sciences 75 35-52

Liu Bangcheng Chun Hui Jin Hu Wensheng Yang and Xinli Yu 2011 How well can

Public Service Motivation Connect with Occupational Intention International

Review of Administrative Sciences 77 191-211

Liu Bangcheng Ningyu Tang and Xiaomei Zhu 2008 Public Service Motivation and Job

Satisfaction in China An Investigation of Generalisability and Instrumentality

International Journal of Manpower 29 684-699

MacCallum Robert C and Michael W Browne 1993 The Use of Causal Indicators in

Covariance Structure Models Some Practical Issues Psychological Bulletin 114

533-541

MacKenzie Scott B 2003 The dangers of poor construct conceptualization Journal of the

Academy of Marketing Science 31(3) 323-326

MacKenzie Scott B Philip M Podsakoff and Cheryl Burke Jarvis 2005 The Problem of

Measurement Model Misspecification in Behavioral and Organizational Research

and Some Recommended Solutions Journal of Applied Psychology 90 710-730

Martinsen Dorte Sindbjerg and Torben Beck Joslashrgensen 2010 Accountability as a

Differentiated Value in Supranational Governance American Review of Public

Administration 40 742-760

Moynihan Donald P and Sanjay K Pandey 2007a The Role of Organizations in Fostering

Public Service Motivation Public Administration Review 67 40-53

Moynihan Donald P and Sanjay K Pandey 2007b Finding Workable Levers over Work

Motivation Comparing Job Satisfaction Job Involvement and Organizational

Commitment Administration amp Society 39 803-832

Pandey Sanjay K and Edmund C Stazyk 2008 Antecedents and Correlates of Public

Service Motivation In Motivation in Public Management edited by James L Perry

and Annie Hondeghem 101-117 Oxford Oxford University Press

- 31 -

Perry James L 1996 Measuring Public Service Motivation An Assessment of Construct

Reliability and Validity Journal of Public Administration Research and Theory 6

5-22

Perry James L 1997 Antecedents of Public Service Motivation Journal of Public

Administration Research and Theory 7 181-197

Perry James L 2000 Bringing Society in Toward a Theory of Public Service Motivation

Journal of Public Administration Research and Theory 10 471-488

Perry James L and Annie Hondeghem 2008a Motivation in Public Management Oxford

Oxford University Press

Perry James L and Annie Hondeghem 2008b Directions for Future Theory and Research

In Motivation in Public Management edited by James L Perry and Annie

Hondeghem 294-313 Oxford Oxford University Press

Perry James L Annie Hondeghem and Lois Recascino Wise 2010 Revisiting the

Motivational Bases of Public Service Twenty Years of Research and an Agenda for

the Future Public Administration Review 70(5) 681-690

Perry James L and Wouter Vandenabeele 2008 Behavioral dynamics Institutions

Identities and Self-Regulation In Motivation in Public Management edited by

James L Perry and Annie Hondeghem 56-79 Oxford Oxford University Press

Perry James L and Lois Recascino Wise 1990 The Motivational Bases of Public Service

Public Administration Review 50 367-373

Rainey Hal G 1982 Reward Preference among Public and Private Managers In Search of

the Service Ethic American Review of Public Administration 16 288-302

Ringle Christian M Oliver Goumltz Martin Wetzels and Bradley Wilson 2009 On the Use of

Formative Measurement Specifications in Structural Equation Modeling A Monte

Carlo Simulation Study to Compare Covariance-Based and Partial Least Squares

Model Estimation Methodologies Faculty of Economics and Business Administration

Maastricht Univeristy

Ringle C M S Wende and S Will 2005 Smart PLS 20 (M3) Beta

- 32 -

httpwwwsmartplsde

Ritz Adrian 2011 Attraction to Public Policy Making A Qualitative Inquiry into

Improvements in PSM Measurement Forthcoming in Public Administration

Spector Paul E 1992 Summated Rating Scales Construction An Introduction Newbury

Park CA Sage

Steenbergen Marco R 1996 Compassion and American public opinion An analysis of the

NES Humanitarianism Scale Report to the Board of Overseers of the National

Election Studies

Steffen Patrick R and Kevin S Masters 2005 Does compassion mediate the intrinsic

religion-health relationship Annals of Behavioral Medicine 30 217-223

Taylor Jeannette 2007 The Impact of Public Service Motives on Work Outcomes in

Australia A Comparative Multi-Dimensional Analysis Public Administration 85

931-959

Vandenabeele Wouter 2007 Towards a Theory of Public Service Motivation An

Institutional Approach Public Management Review 9 545-556

Vandenabeele Wouter 2008a Government Calling Public Service Motivation as an Element

in Selecting Government as an Employer of Choice Public Administration 86

1089-1105

Vandenabeele Wouter 2008b Development of a Public Service Motivation Measurement

Scale Corroborating and Extending Perryrsquos Measurement Instrument International

Public Management Journal 11 143-167

Vandenabeele Wouter 2009 The Mediating Effect of Job Satisfaction and Organizational

Commitment on Self-Reported Performance More Robust Evidence of the PSM-

Performance Relationship International Review of Administrative Sciences 75 11-

34

Wilcox James B Roy D Howell and Einar Breivik 2008 Questions about Formative

Measurement Journal of Business Research 61 1219-1228

Wise Lois Recascino 2000 The Public Service Culture In Public Administration Concepts

- 33 -

and Cases 7th ed edited by Richard J Stillman II 342-353 Boston Houghton

Mifflin

Wright Bradley E 2008 Methodological Challenges associated with Public Service

Motivation Research In Motivation in Public Management edited by James L

Perry and Annie Hondeghem 80-98 Oxford Oxford University Press

Wright Bradley E and Sanjay K Pandey 2008 Public Service Motivation and the

Assumption of Person-Organization Fit Testing the Mediating Effect of Value

Congruence Administration amp Society 40 502-521

- 34 -

Table 1

Possible Indicators of PSM and Descriptive Statistics (n = 2868)

Dimensions and items Mean Standard

Deviation

Item-Total

Correlation

Attraction to Public Participation (APP)

APP1 I am interested in helping to improve public service

APP2 I am satisfied when I see people benefiting from the public programs I was

involved in

APP3 I like to discuss topics regarding public programs and policies with others

APP4 I believe that public sector activities contribute to our general welfare

APP5 I admire people who initiate or are involved in activities to aid my

community

APP6 Contributing to public programs and policies helps me realize myself

APP7 It is important to contribute to activities that tackle social problems

412

441

355

393

423

351

409

696

662

917

753

672

881

734

524

421

482

412

493

478

552

Commitment to Public Values (CPV)

Sub-dimension for Public Interests (CPV1)

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

CPI3 I would prefer seeing public officials do what is best for the whole

community even if it harmed my interests

CPI4 Serving the public interest is more important than helping a single individual

Sub-dimension for Public Values (CPV2)

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of public

services

CPV3 It is fundamental that public services respond to the needs of the citizens

CPV4 Decisions regarding public services should be democratic despite the time

and effort it takes

CPV5 Everybody is entitled to a good service even if it costs a lot of money

CPV6 It is fundamental that the interests of future generations are taken into

account when developing public policies

CPV7 To act ethically is essential for public servants

CPV8 I believe that public employees must always be aware of the legitimacy of

their activities

CPV9 I personally identify with the aim of protecting individual liberties and rights

418

409

348

337

428

416

433

394

367

426

441

434

414

695

662

899

995

730

704

710

852

932

689

681

692

763

525

596

480

251

484

437

356

325

376

439

462

402

392

Compassion (COM) COM1 It is difficult for me to contain my feelings when I see people in distress

COM2 I feel sympathetic to the plight of the underprivileged

COM3 I empathize with other people who face difficulties

COM4 I have little compassion for people in need who are unwilling to take the

first step to help themselves

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

354

387

388

299

412

391

951

825

707

1083

755

765

399

511

389

011

487

502

Self-Sacrifice (SS)

SS1 Making a difference to society means more to me than personal achievements

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS5 People should give back to society more than they get from it

SS6 Serving other citizens would give me a good feeling even if no one paid me for

it

SS7 I would agree to a good plan to make a better life for the poor even if it costs

me money

357

329

339

302

345

363

350

904

872

949

982

888

1049

895

508

584

454

531

412

368

546

- 35 -

Table 2

Background of Respondents (n = 2868)

Variables Characteristics Respondents

Sex

Male

Female

NA

1209

1622

37

422

566

13

Length of service

(years)

0 ~ 10

10 ~ 20

20 ~ 30

30+

NA

1126

723

484

491

44

393

252

169

171

15

Organizational status

―Do you supervise employees

No

Yes

NA

1472

1076

320

513

375

112

Country

Australia

Belgium

China

Denmark

France

Italy

Korea

Lithuania

The Netherlands

Switzerland

United Kingdom

USA

249

214

230

249

266

162

253

236

249

250

260

250

87

75

80

87

93

56

88

82

87

87

91

87 Note NA = no answer

- 36 -

Table 3

Standardized factor loadings and correlations for the 31-item PSM index (n = 2868)

Dimensions and items SFL (MLE) SFL (D-WLS) Alpha

APP

APP1

APP2

APP4

APP5

APP6

APP7

597

486

455

558

490

600

657

552

514

625

549

665

699

CPV1

CPI1

CPI2

CPI3

CPI4

588

660

488

289

653

732

535

294

542

CPV2

CPV1

CPV2

CPV3

CPV4

CPV5

CPV6

CPV7

CPV8

CPV9

611

546

436

393

418

534

598

496

460

690

631

505

457

486

601

669

573

530

745

COM

COM1

COM2

COM3

COM5

COM6

496

569

498

597

607

516

651

538

670

684

687

SS

SS1

SS2

SS3

SS4

SS5

SS6

SS7

567

736

558

728

467

409

663

660

762

591

695

515

494

709

782

Measures of fit (MLE) χsup2 (df = 424) = 43699 p lt 001 CFI = 835 GFI = 899

RMSEA = 057 [055 059]

Measure of fit (D-WLS) χsup2 (df = 424) = 39247 p lt 001 CFI = 971 GFI = 976

RMSEA = 054 [052 055]

Inter-factor correlations MLE (D-WLS)

1 2 3 4

1 APS

2 CPV1 996 (988)

3 CPV2 902 (890) 798 (798)

4 COM 787 (783) 758 (762) 779 (783)

5 SS 700 (742) 800 (859) 509 (537) 720 (729)

Note SFL = standardized factor loading All factor loadings and correlations are significant at p lt 001

- 37 -

Table 4

Standardized factor loadings and correlations for the 16-item PSM index (n = 2868)

Dimensions and items SFL

(MLE)

SFL

(D-WLS)

APS

APP5 I admire people who initiate or are involved in activities to aid my

community

APP7 It is important to contribute to activities that tackle social problems

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

611

592

637

666

665

669

690

751

CPV

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of

public services

CPV6 It is fundamental that the interests of future generations are taken

into account when developing public policies

CPV7 To act ethically is essential for public servants

612

546

551

590

698

631

618

658

COM

COM1 It is difficult for me to contain my feelings when I see people in

distress

COM2 I feel sympathetic to the plight of the underprivileged

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

490

546

596

604

515

608

669

679

SS

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS7 I would agree to a good plan to make a better life for the poor even if

it costs me money

763

571

766

678

807

629

747

755

Measures of fit (MLE)

χsup2 (df = 98) = 8490 p lt 001

CFI = 939 GFI = 965

RMSEA = 052 [049 055]

Measure of fit (D-WLS)

χsup2 (df = 98) = 6069 p lt 001

CFI = 988 GFI = 992

RMSEA = 043 [039 046]

Alpha CR (D-WLS) Inter-factor correlations MLE (D-WLS)

1 APS 717 788 1 APS 2 CPV 3 COM

2 CPV 663 747 848 (859)

3 COM 645 713 781 (791) 789 (795)

4 SS 784 826 565 (596) 462 (505) 715 (726)

Note SFL = standardized factor loading Alpha = Cronbachrsquos alpha CR = composite reliability

All factor loadings and correlations are significant at p lt 001

- 38 -

Table 5

Results of CFA in the individual countries Standardized factor loadings and measures of fit

Factors

amp Items

Australia Belgium China Denmark France Italy

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

672

756

799

785

804

943

710

703

597

790

799

730

584

675

685

681

731

742

578

748

674

643

765

739

645

704

678

708

812

779

632

604

765

766

672

853

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

552

593

560

699

727

632

624

605

560

713

709

645

702

602

616

724

686

735

567

560

715

654

605

811

608

658

587

696

755

698

673

620

523

782

697

654

COM

COM1

COM2

COM3

COM5

COM6

623

584

662

688

701

739

732

608

475

808

666

581

578

616

605

649

670

673

699

751

737

679

816

776

568

546

801

626

673

850

637

546

741

715

601

822

SS

SS2

SS3

SS4

SS7

654

794

806

892

767

813

727

693

799

793

725

838

857

820

773

896

855

842

806

749

792

829

753

884

847

731

753

889

816

753

815

692

664

795

782

734

Inter-factor

correlations

093

~

752

163

~

810

320

~

773

323

~

825

452

~

827

497

~

855

323

~

850

321

~

847

498

~

899

536

~

902

425

~

829

476

~

865

Sample size 249 214 230 249 266 162

Measures of Fit

χsup2 638 444 881 732 765 554 817 625 735 565 750 564

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 980 100 939 987 969 997 967 994 975 997 949 995

GFI 961 986 941 986 948 991 949 992 955 992 926 986

RMSEA 036 00 063 050 051 026 053 035 045 026 059 033

- 39 -

Table 5 (Continued)

Factors

amp Items

Korea Lithuania The Netherlands Switzerland UK USA

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

619

701

667

706

781

717

666

512

682

751

605

779

557

629

630

620

719

729

535

621

702

643

694

735

655

725

738

710

804

822

535

621

702

643

694

735

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

594

607

722

722

644

722

674

535

533

790

589

711

568

554

538

655

628

545

627

580

615

711

666

704

800

512

555

873

581

646

627

580

615

711

666

704

COM

COM1

COM2

COM3

COM5

COM6

694

677

550

763

736

627

557

639

428

550

696

534

737

623

508

766

616

608

659

592

683

703

661

769

682

583

729

739

644

803

659

592

683

703

661

769

SS

SS2

SS3

SS4

SS7

776

752

679

779

770

811

707

721

635

725

734

706

814

656

743

820

767

764

745

632

689

788

662

744

671

818

832

830

864

802

745

632

689

788

662

744

Inter-factor

correlations

432

~

726

466

~

746

376

~

788

406

~

902

377

~

852

455

~

836

637

~

818

660

~

828

317

~

852

360

~

852

637

~

818

660

~

828

Sample size 253 236 249 250 260 250

Measures of Fit

χsup2 896 573 612 462 773 616 815 715 769 542 815 715

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 951 996 975 100 958 993 957 991 972 998 957 991

GFI 942 988 960 989 951 987 948 990 955 990 948 990

RMSEA 059 028 034 00 050 034 053 044 048 022 053 044

Note All factor loadings and correlations are significant at p lt 001

- 40 -

Appendix 1 Dimensions and possible items of PSM (the first round)

1 Attraction to Public Participation

(1) I am interested in developing public programs helping my country or community (Kim 2009a)

(2) Sharing my views on public policies with others is attractive to me (Kim 2009a)

(3) I am satisfied when seeing people get benefits from the public programs I was involved in (Kim 2009a)

(4) I like to discuss political subjects with others (Giauque et al 2011)

(5) I think the governmental activities contribute to our welfare

(6) I like to initiate actions to help out my community

(7) I do not care much about politicians (Perry 1996)

2 Commitment to Public Values

Possible items for public interest

(1A) Meaningful public service is very important to me (Perry 1996)

(2A) It is important for me to contribute for the common good (Giauque et al 2011)

(3A) I recognize myself with the mission of protection of liberties and rights endorsed by the nation-state

(Giauque et al 2011)

(4A) I would prefer seeing public officials do what is best for the whole community even if it harmed my

interests (Perry 1996)

(5A) To me serving the public interest is more important than helping other people (Vandenabeele 2008b)

Possible items for public values

(1B) I recognize myself with the promotion of the equality of chances endorsed by the nation-state (Giauque et

al 2011)

(2B) It is fundamental that the public service provided to the citizen is regular and continuous (Giauque et al

2011)

(3B) It is fundamental that public service innovates to respond to the new needs of the citizensclients (Giauque

et al 2011)

(4B) The dignity and well-being of all should be the most important concerns in any society (Steenbergen

1996)

(5B) It is important that public servants account for all the costsexpenses they make (Vandenabeele 2008b)

(6B) We have to do everything in our power to pursue the goal of democracy

3 Compassion

(1) It is difficult for me to contain my feelings when I see people in distress (Perry 1996)

(2) I feel sympathetic to the plight of the underprivileged (Kim 2009a)

(3) I am often reminded by daily events how dependent we are on one another (Perry 1996)

(4) I feel compassion for others in the difficulties they are facing (Steffen and Masters 2005)

(5) It is not really my problem if others are in trouble and need help (Beutel and Marini 1995)

(6) I find it hard to be sympathetic toward starving people in foreign lands when there is so much trouble in our

own country (Beutel and Marini 1995)

(7) I get very upset when I see other people treated unfairly (Beutel and Marini 1995)

(8) I would agree to a good plan to make a better life for the poor even if it cost me money (Beutel and Marini

1995)

(9) To me patriotism includes seeing the welfare of others (Perry 1996)

(10) I care very much about other people

4 Self-Sacrifice

(1) Making a difference in society means more to me than personal achievements (Perry 1996)

(2) I am prepared to make enormous sacrifices for the good of society (Perry 1996)

(3) I believe in putting duty before self (Perry 1996)

(4) I am one of those rare people who would risk personal loss to help someone else (Perry 1996)

(5) I think people should give back to society more than they get from it (Perry 1996)

(6) Serving other citizens would give me a good feeling even if no one paid me for it (Perry 1996)

(7) Much of what I do is for a cause bigger than myself (Perry 1996)

- 41 -

Appendix 2 Newly suggested items through the first round

1 Items for the Attraction to Public Participation Dimension

(1) In the public domain every person should be counted as equal

(2) I enjoy to think and argue in a political way

(3) I am satisfied when seeing my ideas being integrated in public policies

(4) I can easily relate to the way of thinking and arguing of politicians

(5) Working in the public sector I can better realize myself

2 Items for the Public Values Sub-Dimension

(1) Public service should give value for money

(2) Cost considerations should play an important role in decisions about public service provision

(3) It is fundamental that the interests of future generations are taken into account when developing public

policies

(4) To act with correct ethics is for public servants as important as their professional competence

(5) Primarily public servants should be accountable to the public and not to their superiors

3 Items for the Self-Sacrifice Dimension

(1) I would agree to a good plan to make a better life for the poor even if it costs me money

- 42 -

Appendix 3 An example of a summary of comments and suggestions (the second round)

On the item ―I am interested in developing public programs helping my country or community

A It is a good sentence but is it possible for everyone to ―develop public programs Even public officials are

limited to do this

B OK

C or the special group

D Double barrel and difficult to translate to my country ―I am interested in helping to improve public services

E Good but I feel that wording requires revision If we draw sample from local government then country is

redundant ―I am interested in developing public policy programs that aim in helping my community

F 4 (General appreciation from1 to 4) Not clear whether it refers to ones work or to ones concern as a citizen

―(In my work) I am interested in contributing to the development of public etchellip

G Instead of development I would use implementing

H Delete public which is redundant because of the characteristic helping my country or community but add

or implementing ―I am interested in developing or implementing programs helping my country or

community

I Referring to dimension Wording Similar to item 6

Note Names have been changed to alphabets in order to ensure the anonymity of participants

- 43 -

Appendix 4 Dimensions and Items of Public Service Motivation for Future Research

Dimensions and items

Attraction to Public Service (APS)

1 I am interested in helping to improve public service

2 I admire people who initiate or are involved in activities to aid my community

3 It is important to contribute to activities that tackle social problems

4 Meaningful public service is very important to me

5 It is important for me to contribute to the common good

Commitment to Public Values (CPV)

1 I think equal opportunities for citizens are very important

2 It is important that citizens can rely on the continuous provision of public services

3 It is fundamental that the interests of future generations are taken into account when

developing public policies

4 To act ethically is essential for public servants

5 I believe that public employees must always be aware of the legitimacy of their activities

6 I personally identify with the aim of protecting individual liberties and rights

Compassion (COM)

1 It is difficult for me to contain my feelings when I see people in distress

2 I feel sympathetic to the plight of the underprivileged

3 I empathize with other people who face difficulties

4 I get very upset when I see other people being treated unfairly

5 Considering the welfare of others is very important

Self-Sacrifice (SS)

1 I am prepared to make sacrifices for the good of society

2 I believe in putting civic duty before self

3 I am willing to risk personal loss to help society

4 I would agree to a good plan to make a better life for the poor even if it costs me money

- 29 -

Howell Roy D Einar Breivik and James B Wilcox 2007a Reconsidering Formative

Measurement Psychological Methods 12 205-218

Howell Roy D Einar Breivik and James B Wilcox 2007b Is Formative Measurement

Really Measurement Reply to Bollen (2007) and Bagozzi (2007) Psychological

Methods 12 238-245

Jarvis Cheryl Burke Scott B MacKenzie and Philip M Podsakoff 2003 A Critical Review

of Construct Indicators and Measurement Model Misspecification in Marketing and

Consumer Research Journal of Consumer Research 30 199-218

Joumlreskog Karl G and Dag Soumlrbom 2005 LISREL 872 (Computer Software) Lincolnwood

IL Scientific Software International

Kim Gimun Bongsik Shin and Varun Grover 2010 Investigating Two Contradictory Views

of Formative Measurement in Information Systems Research MIS Quarterly 34

345-365

Kim Sangmook 2009a Revising Perryrsquos Measurement Scale of Public Service Motivation

American Review of Public Administration 39 149-163

Kim Sangmook 2009b Testing the Structure of Public Service Motivation in Korea A

Research Note Journal of Public Administration Research and Theory 19 839-

851

Kim Sangmook 2010 Testing a Revised Measurement of Public Service Motivation

Reflective versus Formative Specification Journal of Public Administration

Research and Theory Doi101093jopartmuq048

Kim Sangmook and Wouter Vandenabeele 2010 A strategy for building public service

motivation research internationally Public Administration Review 70 701-709

Kline Rex B 2005 Principles and Practice of Structural Equation Modeling 2nd

ed New

York Guilford Press

Law Kenneth and Chi-Sum Wong 1999 Multidimensional Constructs in Structural

Equation Analysis An Illustration using the Job Perception and Job Satisfaction

Constructs Journal of Management 25 143-160

- 30 -

Leisink Peter and Bram Steijn 2009 Public Service Motivation and Job Performance of

Public Sector Employees in the Netherlands International Review of

Administrative Sciences 75 35-52

Liu Bangcheng Chun Hui Jin Hu Wensheng Yang and Xinli Yu 2011 How well can

Public Service Motivation Connect with Occupational Intention International

Review of Administrative Sciences 77 191-211

Liu Bangcheng Ningyu Tang and Xiaomei Zhu 2008 Public Service Motivation and Job

Satisfaction in China An Investigation of Generalisability and Instrumentality

International Journal of Manpower 29 684-699

MacCallum Robert C and Michael W Browne 1993 The Use of Causal Indicators in

Covariance Structure Models Some Practical Issues Psychological Bulletin 114

533-541

MacKenzie Scott B 2003 The dangers of poor construct conceptualization Journal of the

Academy of Marketing Science 31(3) 323-326

MacKenzie Scott B Philip M Podsakoff and Cheryl Burke Jarvis 2005 The Problem of

Measurement Model Misspecification in Behavioral and Organizational Research

and Some Recommended Solutions Journal of Applied Psychology 90 710-730

Martinsen Dorte Sindbjerg and Torben Beck Joslashrgensen 2010 Accountability as a

Differentiated Value in Supranational Governance American Review of Public

Administration 40 742-760

Moynihan Donald P and Sanjay K Pandey 2007a The Role of Organizations in Fostering

Public Service Motivation Public Administration Review 67 40-53

Moynihan Donald P and Sanjay K Pandey 2007b Finding Workable Levers over Work

Motivation Comparing Job Satisfaction Job Involvement and Organizational

Commitment Administration amp Society 39 803-832

Pandey Sanjay K and Edmund C Stazyk 2008 Antecedents and Correlates of Public

Service Motivation In Motivation in Public Management edited by James L Perry

and Annie Hondeghem 101-117 Oxford Oxford University Press

- 31 -

Perry James L 1996 Measuring Public Service Motivation An Assessment of Construct

Reliability and Validity Journal of Public Administration Research and Theory 6

5-22

Perry James L 1997 Antecedents of Public Service Motivation Journal of Public

Administration Research and Theory 7 181-197

Perry James L 2000 Bringing Society in Toward a Theory of Public Service Motivation

Journal of Public Administration Research and Theory 10 471-488

Perry James L and Annie Hondeghem 2008a Motivation in Public Management Oxford

Oxford University Press

Perry James L and Annie Hondeghem 2008b Directions for Future Theory and Research

In Motivation in Public Management edited by James L Perry and Annie

Hondeghem 294-313 Oxford Oxford University Press

Perry James L Annie Hondeghem and Lois Recascino Wise 2010 Revisiting the

Motivational Bases of Public Service Twenty Years of Research and an Agenda for

the Future Public Administration Review 70(5) 681-690

Perry James L and Wouter Vandenabeele 2008 Behavioral dynamics Institutions

Identities and Self-Regulation In Motivation in Public Management edited by

James L Perry and Annie Hondeghem 56-79 Oxford Oxford University Press

Perry James L and Lois Recascino Wise 1990 The Motivational Bases of Public Service

Public Administration Review 50 367-373

Rainey Hal G 1982 Reward Preference among Public and Private Managers In Search of

the Service Ethic American Review of Public Administration 16 288-302

Ringle Christian M Oliver Goumltz Martin Wetzels and Bradley Wilson 2009 On the Use of

Formative Measurement Specifications in Structural Equation Modeling A Monte

Carlo Simulation Study to Compare Covariance-Based and Partial Least Squares

Model Estimation Methodologies Faculty of Economics and Business Administration

Maastricht Univeristy

Ringle C M S Wende and S Will 2005 Smart PLS 20 (M3) Beta

- 32 -

httpwwwsmartplsde

Ritz Adrian 2011 Attraction to Public Policy Making A Qualitative Inquiry into

Improvements in PSM Measurement Forthcoming in Public Administration

Spector Paul E 1992 Summated Rating Scales Construction An Introduction Newbury

Park CA Sage

Steenbergen Marco R 1996 Compassion and American public opinion An analysis of the

NES Humanitarianism Scale Report to the Board of Overseers of the National

Election Studies

Steffen Patrick R and Kevin S Masters 2005 Does compassion mediate the intrinsic

religion-health relationship Annals of Behavioral Medicine 30 217-223

Taylor Jeannette 2007 The Impact of Public Service Motives on Work Outcomes in

Australia A Comparative Multi-Dimensional Analysis Public Administration 85

931-959

Vandenabeele Wouter 2007 Towards a Theory of Public Service Motivation An

Institutional Approach Public Management Review 9 545-556

Vandenabeele Wouter 2008a Government Calling Public Service Motivation as an Element

in Selecting Government as an Employer of Choice Public Administration 86

1089-1105

Vandenabeele Wouter 2008b Development of a Public Service Motivation Measurement

Scale Corroborating and Extending Perryrsquos Measurement Instrument International

Public Management Journal 11 143-167

Vandenabeele Wouter 2009 The Mediating Effect of Job Satisfaction and Organizational

Commitment on Self-Reported Performance More Robust Evidence of the PSM-

Performance Relationship International Review of Administrative Sciences 75 11-

34

Wilcox James B Roy D Howell and Einar Breivik 2008 Questions about Formative

Measurement Journal of Business Research 61 1219-1228

Wise Lois Recascino 2000 The Public Service Culture In Public Administration Concepts

- 33 -

and Cases 7th ed edited by Richard J Stillman II 342-353 Boston Houghton

Mifflin

Wright Bradley E 2008 Methodological Challenges associated with Public Service

Motivation Research In Motivation in Public Management edited by James L

Perry and Annie Hondeghem 80-98 Oxford Oxford University Press

Wright Bradley E and Sanjay K Pandey 2008 Public Service Motivation and the

Assumption of Person-Organization Fit Testing the Mediating Effect of Value

Congruence Administration amp Society 40 502-521

- 34 -

Table 1

Possible Indicators of PSM and Descriptive Statistics (n = 2868)

Dimensions and items Mean Standard

Deviation

Item-Total

Correlation

Attraction to Public Participation (APP)

APP1 I am interested in helping to improve public service

APP2 I am satisfied when I see people benefiting from the public programs I was

involved in

APP3 I like to discuss topics regarding public programs and policies with others

APP4 I believe that public sector activities contribute to our general welfare

APP5 I admire people who initiate or are involved in activities to aid my

community

APP6 Contributing to public programs and policies helps me realize myself

APP7 It is important to contribute to activities that tackle social problems

412

441

355

393

423

351

409

696

662

917

753

672

881

734

524

421

482

412

493

478

552

Commitment to Public Values (CPV)

Sub-dimension for Public Interests (CPV1)

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

CPI3 I would prefer seeing public officials do what is best for the whole

community even if it harmed my interests

CPI4 Serving the public interest is more important than helping a single individual

Sub-dimension for Public Values (CPV2)

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of public

services

CPV3 It is fundamental that public services respond to the needs of the citizens

CPV4 Decisions regarding public services should be democratic despite the time

and effort it takes

CPV5 Everybody is entitled to a good service even if it costs a lot of money

CPV6 It is fundamental that the interests of future generations are taken into

account when developing public policies

CPV7 To act ethically is essential for public servants

CPV8 I believe that public employees must always be aware of the legitimacy of

their activities

CPV9 I personally identify with the aim of protecting individual liberties and rights

418

409

348

337

428

416

433

394

367

426

441

434

414

695

662

899

995

730

704

710

852

932

689

681

692

763

525

596

480

251

484

437

356

325

376

439

462

402

392

Compassion (COM) COM1 It is difficult for me to contain my feelings when I see people in distress

COM2 I feel sympathetic to the plight of the underprivileged

COM3 I empathize with other people who face difficulties

COM4 I have little compassion for people in need who are unwilling to take the

first step to help themselves

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

354

387

388

299

412

391

951

825

707

1083

755

765

399

511

389

011

487

502

Self-Sacrifice (SS)

SS1 Making a difference to society means more to me than personal achievements

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS5 People should give back to society more than they get from it

SS6 Serving other citizens would give me a good feeling even if no one paid me for

it

SS7 I would agree to a good plan to make a better life for the poor even if it costs

me money

357

329

339

302

345

363

350

904

872

949

982

888

1049

895

508

584

454

531

412

368

546

- 35 -

Table 2

Background of Respondents (n = 2868)

Variables Characteristics Respondents

Sex

Male

Female

NA

1209

1622

37

422

566

13

Length of service

(years)

0 ~ 10

10 ~ 20

20 ~ 30

30+

NA

1126

723

484

491

44

393

252

169

171

15

Organizational status

―Do you supervise employees

No

Yes

NA

1472

1076

320

513

375

112

Country

Australia

Belgium

China

Denmark

France

Italy

Korea

Lithuania

The Netherlands

Switzerland

United Kingdom

USA

249

214

230

249

266

162

253

236

249

250

260

250

87

75

80

87

93

56

88

82

87

87

91

87 Note NA = no answer

- 36 -

Table 3

Standardized factor loadings and correlations for the 31-item PSM index (n = 2868)

Dimensions and items SFL (MLE) SFL (D-WLS) Alpha

APP

APP1

APP2

APP4

APP5

APP6

APP7

597

486

455

558

490

600

657

552

514

625

549

665

699

CPV1

CPI1

CPI2

CPI3

CPI4

588

660

488

289

653

732

535

294

542

CPV2

CPV1

CPV2

CPV3

CPV4

CPV5

CPV6

CPV7

CPV8

CPV9

611

546

436

393

418

534

598

496

460

690

631

505

457

486

601

669

573

530

745

COM

COM1

COM2

COM3

COM5

COM6

496

569

498

597

607

516

651

538

670

684

687

SS

SS1

SS2

SS3

SS4

SS5

SS6

SS7

567

736

558

728

467

409

663

660

762

591

695

515

494

709

782

Measures of fit (MLE) χsup2 (df = 424) = 43699 p lt 001 CFI = 835 GFI = 899

RMSEA = 057 [055 059]

Measure of fit (D-WLS) χsup2 (df = 424) = 39247 p lt 001 CFI = 971 GFI = 976

RMSEA = 054 [052 055]

Inter-factor correlations MLE (D-WLS)

1 2 3 4

1 APS

2 CPV1 996 (988)

3 CPV2 902 (890) 798 (798)

4 COM 787 (783) 758 (762) 779 (783)

5 SS 700 (742) 800 (859) 509 (537) 720 (729)

Note SFL = standardized factor loading All factor loadings and correlations are significant at p lt 001

- 37 -

Table 4

Standardized factor loadings and correlations for the 16-item PSM index (n = 2868)

Dimensions and items SFL

(MLE)

SFL

(D-WLS)

APS

APP5 I admire people who initiate or are involved in activities to aid my

community

APP7 It is important to contribute to activities that tackle social problems

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

611

592

637

666

665

669

690

751

CPV

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of

public services

CPV6 It is fundamental that the interests of future generations are taken

into account when developing public policies

CPV7 To act ethically is essential for public servants

612

546

551

590

698

631

618

658

COM

COM1 It is difficult for me to contain my feelings when I see people in

distress

COM2 I feel sympathetic to the plight of the underprivileged

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

490

546

596

604

515

608

669

679

SS

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS7 I would agree to a good plan to make a better life for the poor even if

it costs me money

763

571

766

678

807

629

747

755

Measures of fit (MLE)

χsup2 (df = 98) = 8490 p lt 001

CFI = 939 GFI = 965

RMSEA = 052 [049 055]

Measure of fit (D-WLS)

χsup2 (df = 98) = 6069 p lt 001

CFI = 988 GFI = 992

RMSEA = 043 [039 046]

Alpha CR (D-WLS) Inter-factor correlations MLE (D-WLS)

1 APS 717 788 1 APS 2 CPV 3 COM

2 CPV 663 747 848 (859)

3 COM 645 713 781 (791) 789 (795)

4 SS 784 826 565 (596) 462 (505) 715 (726)

Note SFL = standardized factor loading Alpha = Cronbachrsquos alpha CR = composite reliability

All factor loadings and correlations are significant at p lt 001

- 38 -

Table 5

Results of CFA in the individual countries Standardized factor loadings and measures of fit

Factors

amp Items

Australia Belgium China Denmark France Italy

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

672

756

799

785

804

943

710

703

597

790

799

730

584

675

685

681

731

742

578

748

674

643

765

739

645

704

678

708

812

779

632

604

765

766

672

853

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

552

593

560

699

727

632

624

605

560

713

709

645

702

602

616

724

686

735

567

560

715

654

605

811

608

658

587

696

755

698

673

620

523

782

697

654

COM

COM1

COM2

COM3

COM5

COM6

623

584

662

688

701

739

732

608

475

808

666

581

578

616

605

649

670

673

699

751

737

679

816

776

568

546

801

626

673

850

637

546

741

715

601

822

SS

SS2

SS3

SS4

SS7

654

794

806

892

767

813

727

693

799

793

725

838

857

820

773

896

855

842

806

749

792

829

753

884

847

731

753

889

816

753

815

692

664

795

782

734

Inter-factor

correlations

093

~

752

163

~

810

320

~

773

323

~

825

452

~

827

497

~

855

323

~

850

321

~

847

498

~

899

536

~

902

425

~

829

476

~

865

Sample size 249 214 230 249 266 162

Measures of Fit

χsup2 638 444 881 732 765 554 817 625 735 565 750 564

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 980 100 939 987 969 997 967 994 975 997 949 995

GFI 961 986 941 986 948 991 949 992 955 992 926 986

RMSEA 036 00 063 050 051 026 053 035 045 026 059 033

- 39 -

Table 5 (Continued)

Factors

amp Items

Korea Lithuania The Netherlands Switzerland UK USA

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

619

701

667

706

781

717

666

512

682

751

605

779

557

629

630

620

719

729

535

621

702

643

694

735

655

725

738

710

804

822

535

621

702

643

694

735

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

594

607

722

722

644

722

674

535

533

790

589

711

568

554

538

655

628

545

627

580

615

711

666

704

800

512

555

873

581

646

627

580

615

711

666

704

COM

COM1

COM2

COM3

COM5

COM6

694

677

550

763

736

627

557

639

428

550

696

534

737

623

508

766

616

608

659

592

683

703

661

769

682

583

729

739

644

803

659

592

683

703

661

769

SS

SS2

SS3

SS4

SS7

776

752

679

779

770

811

707

721

635

725

734

706

814

656

743

820

767

764

745

632

689

788

662

744

671

818

832

830

864

802

745

632

689

788

662

744

Inter-factor

correlations

432

~

726

466

~

746

376

~

788

406

~

902

377

~

852

455

~

836

637

~

818

660

~

828

317

~

852

360

~

852

637

~

818

660

~

828

Sample size 253 236 249 250 260 250

Measures of Fit

χsup2 896 573 612 462 773 616 815 715 769 542 815 715

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 951 996 975 100 958 993 957 991 972 998 957 991

GFI 942 988 960 989 951 987 948 990 955 990 948 990

RMSEA 059 028 034 00 050 034 053 044 048 022 053 044

Note All factor loadings and correlations are significant at p lt 001

- 40 -

Appendix 1 Dimensions and possible items of PSM (the first round)

1 Attraction to Public Participation

(1) I am interested in developing public programs helping my country or community (Kim 2009a)

(2) Sharing my views on public policies with others is attractive to me (Kim 2009a)

(3) I am satisfied when seeing people get benefits from the public programs I was involved in (Kim 2009a)

(4) I like to discuss political subjects with others (Giauque et al 2011)

(5) I think the governmental activities contribute to our welfare

(6) I like to initiate actions to help out my community

(7) I do not care much about politicians (Perry 1996)

2 Commitment to Public Values

Possible items for public interest

(1A) Meaningful public service is very important to me (Perry 1996)

(2A) It is important for me to contribute for the common good (Giauque et al 2011)

(3A) I recognize myself with the mission of protection of liberties and rights endorsed by the nation-state

(Giauque et al 2011)

(4A) I would prefer seeing public officials do what is best for the whole community even if it harmed my

interests (Perry 1996)

(5A) To me serving the public interest is more important than helping other people (Vandenabeele 2008b)

Possible items for public values

(1B) I recognize myself with the promotion of the equality of chances endorsed by the nation-state (Giauque et

al 2011)

(2B) It is fundamental that the public service provided to the citizen is regular and continuous (Giauque et al

2011)

(3B) It is fundamental that public service innovates to respond to the new needs of the citizensclients (Giauque

et al 2011)

(4B) The dignity and well-being of all should be the most important concerns in any society (Steenbergen

1996)

(5B) It is important that public servants account for all the costsexpenses they make (Vandenabeele 2008b)

(6B) We have to do everything in our power to pursue the goal of democracy

3 Compassion

(1) It is difficult for me to contain my feelings when I see people in distress (Perry 1996)

(2) I feel sympathetic to the plight of the underprivileged (Kim 2009a)

(3) I am often reminded by daily events how dependent we are on one another (Perry 1996)

(4) I feel compassion for others in the difficulties they are facing (Steffen and Masters 2005)

(5) It is not really my problem if others are in trouble and need help (Beutel and Marini 1995)

(6) I find it hard to be sympathetic toward starving people in foreign lands when there is so much trouble in our

own country (Beutel and Marini 1995)

(7) I get very upset when I see other people treated unfairly (Beutel and Marini 1995)

(8) I would agree to a good plan to make a better life for the poor even if it cost me money (Beutel and Marini

1995)

(9) To me patriotism includes seeing the welfare of others (Perry 1996)

(10) I care very much about other people

4 Self-Sacrifice

(1) Making a difference in society means more to me than personal achievements (Perry 1996)

(2) I am prepared to make enormous sacrifices for the good of society (Perry 1996)

(3) I believe in putting duty before self (Perry 1996)

(4) I am one of those rare people who would risk personal loss to help someone else (Perry 1996)

(5) I think people should give back to society more than they get from it (Perry 1996)

(6) Serving other citizens would give me a good feeling even if no one paid me for it (Perry 1996)

(7) Much of what I do is for a cause bigger than myself (Perry 1996)

- 41 -

Appendix 2 Newly suggested items through the first round

1 Items for the Attraction to Public Participation Dimension

(1) In the public domain every person should be counted as equal

(2) I enjoy to think and argue in a political way

(3) I am satisfied when seeing my ideas being integrated in public policies

(4) I can easily relate to the way of thinking and arguing of politicians

(5) Working in the public sector I can better realize myself

2 Items for the Public Values Sub-Dimension

(1) Public service should give value for money

(2) Cost considerations should play an important role in decisions about public service provision

(3) It is fundamental that the interests of future generations are taken into account when developing public

policies

(4) To act with correct ethics is for public servants as important as their professional competence

(5) Primarily public servants should be accountable to the public and not to their superiors

3 Items for the Self-Sacrifice Dimension

(1) I would agree to a good plan to make a better life for the poor even if it costs me money

- 42 -

Appendix 3 An example of a summary of comments and suggestions (the second round)

On the item ―I am interested in developing public programs helping my country or community

A It is a good sentence but is it possible for everyone to ―develop public programs Even public officials are

limited to do this

B OK

C or the special group

D Double barrel and difficult to translate to my country ―I am interested in helping to improve public services

E Good but I feel that wording requires revision If we draw sample from local government then country is

redundant ―I am interested in developing public policy programs that aim in helping my community

F 4 (General appreciation from1 to 4) Not clear whether it refers to ones work or to ones concern as a citizen

―(In my work) I am interested in contributing to the development of public etchellip

G Instead of development I would use implementing

H Delete public which is redundant because of the characteristic helping my country or community but add

or implementing ―I am interested in developing or implementing programs helping my country or

community

I Referring to dimension Wording Similar to item 6

Note Names have been changed to alphabets in order to ensure the anonymity of participants

- 43 -

Appendix 4 Dimensions and Items of Public Service Motivation for Future Research

Dimensions and items

Attraction to Public Service (APS)

1 I am interested in helping to improve public service

2 I admire people who initiate or are involved in activities to aid my community

3 It is important to contribute to activities that tackle social problems

4 Meaningful public service is very important to me

5 It is important for me to contribute to the common good

Commitment to Public Values (CPV)

1 I think equal opportunities for citizens are very important

2 It is important that citizens can rely on the continuous provision of public services

3 It is fundamental that the interests of future generations are taken into account when

developing public policies

4 To act ethically is essential for public servants

5 I believe that public employees must always be aware of the legitimacy of their activities

6 I personally identify with the aim of protecting individual liberties and rights

Compassion (COM)

1 It is difficult for me to contain my feelings when I see people in distress

2 I feel sympathetic to the plight of the underprivileged

3 I empathize with other people who face difficulties

4 I get very upset when I see other people being treated unfairly

5 Considering the welfare of others is very important

Self-Sacrifice (SS)

1 I am prepared to make sacrifices for the good of society

2 I believe in putting civic duty before self

3 I am willing to risk personal loss to help society

4 I would agree to a good plan to make a better life for the poor even if it costs me money

- 30 -

Leisink Peter and Bram Steijn 2009 Public Service Motivation and Job Performance of

Public Sector Employees in the Netherlands International Review of

Administrative Sciences 75 35-52

Liu Bangcheng Chun Hui Jin Hu Wensheng Yang and Xinli Yu 2011 How well can

Public Service Motivation Connect with Occupational Intention International

Review of Administrative Sciences 77 191-211

Liu Bangcheng Ningyu Tang and Xiaomei Zhu 2008 Public Service Motivation and Job

Satisfaction in China An Investigation of Generalisability and Instrumentality

International Journal of Manpower 29 684-699

MacCallum Robert C and Michael W Browne 1993 The Use of Causal Indicators in

Covariance Structure Models Some Practical Issues Psychological Bulletin 114

533-541

MacKenzie Scott B 2003 The dangers of poor construct conceptualization Journal of the

Academy of Marketing Science 31(3) 323-326

MacKenzie Scott B Philip M Podsakoff and Cheryl Burke Jarvis 2005 The Problem of

Measurement Model Misspecification in Behavioral and Organizational Research

and Some Recommended Solutions Journal of Applied Psychology 90 710-730

Martinsen Dorte Sindbjerg and Torben Beck Joslashrgensen 2010 Accountability as a

Differentiated Value in Supranational Governance American Review of Public

Administration 40 742-760

Moynihan Donald P and Sanjay K Pandey 2007a The Role of Organizations in Fostering

Public Service Motivation Public Administration Review 67 40-53

Moynihan Donald P and Sanjay K Pandey 2007b Finding Workable Levers over Work

Motivation Comparing Job Satisfaction Job Involvement and Organizational

Commitment Administration amp Society 39 803-832

Pandey Sanjay K and Edmund C Stazyk 2008 Antecedents and Correlates of Public

Service Motivation In Motivation in Public Management edited by James L Perry

and Annie Hondeghem 101-117 Oxford Oxford University Press

- 31 -

Perry James L 1996 Measuring Public Service Motivation An Assessment of Construct

Reliability and Validity Journal of Public Administration Research and Theory 6

5-22

Perry James L 1997 Antecedents of Public Service Motivation Journal of Public

Administration Research and Theory 7 181-197

Perry James L 2000 Bringing Society in Toward a Theory of Public Service Motivation

Journal of Public Administration Research and Theory 10 471-488

Perry James L and Annie Hondeghem 2008a Motivation in Public Management Oxford

Oxford University Press

Perry James L and Annie Hondeghem 2008b Directions for Future Theory and Research

In Motivation in Public Management edited by James L Perry and Annie

Hondeghem 294-313 Oxford Oxford University Press

Perry James L Annie Hondeghem and Lois Recascino Wise 2010 Revisiting the

Motivational Bases of Public Service Twenty Years of Research and an Agenda for

the Future Public Administration Review 70(5) 681-690

Perry James L and Wouter Vandenabeele 2008 Behavioral dynamics Institutions

Identities and Self-Regulation In Motivation in Public Management edited by

James L Perry and Annie Hondeghem 56-79 Oxford Oxford University Press

Perry James L and Lois Recascino Wise 1990 The Motivational Bases of Public Service

Public Administration Review 50 367-373

Rainey Hal G 1982 Reward Preference among Public and Private Managers In Search of

the Service Ethic American Review of Public Administration 16 288-302

Ringle Christian M Oliver Goumltz Martin Wetzels and Bradley Wilson 2009 On the Use of

Formative Measurement Specifications in Structural Equation Modeling A Monte

Carlo Simulation Study to Compare Covariance-Based and Partial Least Squares

Model Estimation Methodologies Faculty of Economics and Business Administration

Maastricht Univeristy

Ringle C M S Wende and S Will 2005 Smart PLS 20 (M3) Beta

- 32 -

httpwwwsmartplsde

Ritz Adrian 2011 Attraction to Public Policy Making A Qualitative Inquiry into

Improvements in PSM Measurement Forthcoming in Public Administration

Spector Paul E 1992 Summated Rating Scales Construction An Introduction Newbury

Park CA Sage

Steenbergen Marco R 1996 Compassion and American public opinion An analysis of the

NES Humanitarianism Scale Report to the Board of Overseers of the National

Election Studies

Steffen Patrick R and Kevin S Masters 2005 Does compassion mediate the intrinsic

religion-health relationship Annals of Behavioral Medicine 30 217-223

Taylor Jeannette 2007 The Impact of Public Service Motives on Work Outcomes in

Australia A Comparative Multi-Dimensional Analysis Public Administration 85

931-959

Vandenabeele Wouter 2007 Towards a Theory of Public Service Motivation An

Institutional Approach Public Management Review 9 545-556

Vandenabeele Wouter 2008a Government Calling Public Service Motivation as an Element

in Selecting Government as an Employer of Choice Public Administration 86

1089-1105

Vandenabeele Wouter 2008b Development of a Public Service Motivation Measurement

Scale Corroborating and Extending Perryrsquos Measurement Instrument International

Public Management Journal 11 143-167

Vandenabeele Wouter 2009 The Mediating Effect of Job Satisfaction and Organizational

Commitment on Self-Reported Performance More Robust Evidence of the PSM-

Performance Relationship International Review of Administrative Sciences 75 11-

34

Wilcox James B Roy D Howell and Einar Breivik 2008 Questions about Formative

Measurement Journal of Business Research 61 1219-1228

Wise Lois Recascino 2000 The Public Service Culture In Public Administration Concepts

- 33 -

and Cases 7th ed edited by Richard J Stillman II 342-353 Boston Houghton

Mifflin

Wright Bradley E 2008 Methodological Challenges associated with Public Service

Motivation Research In Motivation in Public Management edited by James L

Perry and Annie Hondeghem 80-98 Oxford Oxford University Press

Wright Bradley E and Sanjay K Pandey 2008 Public Service Motivation and the

Assumption of Person-Organization Fit Testing the Mediating Effect of Value

Congruence Administration amp Society 40 502-521

- 34 -

Table 1

Possible Indicators of PSM and Descriptive Statistics (n = 2868)

Dimensions and items Mean Standard

Deviation

Item-Total

Correlation

Attraction to Public Participation (APP)

APP1 I am interested in helping to improve public service

APP2 I am satisfied when I see people benefiting from the public programs I was

involved in

APP3 I like to discuss topics regarding public programs and policies with others

APP4 I believe that public sector activities contribute to our general welfare

APP5 I admire people who initiate or are involved in activities to aid my

community

APP6 Contributing to public programs and policies helps me realize myself

APP7 It is important to contribute to activities that tackle social problems

412

441

355

393

423

351

409

696

662

917

753

672

881

734

524

421

482

412

493

478

552

Commitment to Public Values (CPV)

Sub-dimension for Public Interests (CPV1)

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

CPI3 I would prefer seeing public officials do what is best for the whole

community even if it harmed my interests

CPI4 Serving the public interest is more important than helping a single individual

Sub-dimension for Public Values (CPV2)

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of public

services

CPV3 It is fundamental that public services respond to the needs of the citizens

CPV4 Decisions regarding public services should be democratic despite the time

and effort it takes

CPV5 Everybody is entitled to a good service even if it costs a lot of money

CPV6 It is fundamental that the interests of future generations are taken into

account when developing public policies

CPV7 To act ethically is essential for public servants

CPV8 I believe that public employees must always be aware of the legitimacy of

their activities

CPV9 I personally identify with the aim of protecting individual liberties and rights

418

409

348

337

428

416

433

394

367

426

441

434

414

695

662

899

995

730

704

710

852

932

689

681

692

763

525

596

480

251

484

437

356

325

376

439

462

402

392

Compassion (COM) COM1 It is difficult for me to contain my feelings when I see people in distress

COM2 I feel sympathetic to the plight of the underprivileged

COM3 I empathize with other people who face difficulties

COM4 I have little compassion for people in need who are unwilling to take the

first step to help themselves

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

354

387

388

299

412

391

951

825

707

1083

755

765

399

511

389

011

487

502

Self-Sacrifice (SS)

SS1 Making a difference to society means more to me than personal achievements

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS5 People should give back to society more than they get from it

SS6 Serving other citizens would give me a good feeling even if no one paid me for

it

SS7 I would agree to a good plan to make a better life for the poor even if it costs

me money

357

329

339

302

345

363

350

904

872

949

982

888

1049

895

508

584

454

531

412

368

546

- 35 -

Table 2

Background of Respondents (n = 2868)

Variables Characteristics Respondents

Sex

Male

Female

NA

1209

1622

37

422

566

13

Length of service

(years)

0 ~ 10

10 ~ 20

20 ~ 30

30+

NA

1126

723

484

491

44

393

252

169

171

15

Organizational status

―Do you supervise employees

No

Yes

NA

1472

1076

320

513

375

112

Country

Australia

Belgium

China

Denmark

France

Italy

Korea

Lithuania

The Netherlands

Switzerland

United Kingdom

USA

249

214

230

249

266

162

253

236

249

250

260

250

87

75

80

87

93

56

88

82

87

87

91

87 Note NA = no answer

- 36 -

Table 3

Standardized factor loadings and correlations for the 31-item PSM index (n = 2868)

Dimensions and items SFL (MLE) SFL (D-WLS) Alpha

APP

APP1

APP2

APP4

APP5

APP6

APP7

597

486

455

558

490

600

657

552

514

625

549

665

699

CPV1

CPI1

CPI2

CPI3

CPI4

588

660

488

289

653

732

535

294

542

CPV2

CPV1

CPV2

CPV3

CPV4

CPV5

CPV6

CPV7

CPV8

CPV9

611

546

436

393

418

534

598

496

460

690

631

505

457

486

601

669

573

530

745

COM

COM1

COM2

COM3

COM5

COM6

496

569

498

597

607

516

651

538

670

684

687

SS

SS1

SS2

SS3

SS4

SS5

SS6

SS7

567

736

558

728

467

409

663

660

762

591

695

515

494

709

782

Measures of fit (MLE) χsup2 (df = 424) = 43699 p lt 001 CFI = 835 GFI = 899

RMSEA = 057 [055 059]

Measure of fit (D-WLS) χsup2 (df = 424) = 39247 p lt 001 CFI = 971 GFI = 976

RMSEA = 054 [052 055]

Inter-factor correlations MLE (D-WLS)

1 2 3 4

1 APS

2 CPV1 996 (988)

3 CPV2 902 (890) 798 (798)

4 COM 787 (783) 758 (762) 779 (783)

5 SS 700 (742) 800 (859) 509 (537) 720 (729)

Note SFL = standardized factor loading All factor loadings and correlations are significant at p lt 001

- 37 -

Table 4

Standardized factor loadings and correlations for the 16-item PSM index (n = 2868)

Dimensions and items SFL

(MLE)

SFL

(D-WLS)

APS

APP5 I admire people who initiate or are involved in activities to aid my

community

APP7 It is important to contribute to activities that tackle social problems

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

611

592

637

666

665

669

690

751

CPV

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of

public services

CPV6 It is fundamental that the interests of future generations are taken

into account when developing public policies

CPV7 To act ethically is essential for public servants

612

546

551

590

698

631

618

658

COM

COM1 It is difficult for me to contain my feelings when I see people in

distress

COM2 I feel sympathetic to the plight of the underprivileged

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

490

546

596

604

515

608

669

679

SS

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS7 I would agree to a good plan to make a better life for the poor even if

it costs me money

763

571

766

678

807

629

747

755

Measures of fit (MLE)

χsup2 (df = 98) = 8490 p lt 001

CFI = 939 GFI = 965

RMSEA = 052 [049 055]

Measure of fit (D-WLS)

χsup2 (df = 98) = 6069 p lt 001

CFI = 988 GFI = 992

RMSEA = 043 [039 046]

Alpha CR (D-WLS) Inter-factor correlations MLE (D-WLS)

1 APS 717 788 1 APS 2 CPV 3 COM

2 CPV 663 747 848 (859)

3 COM 645 713 781 (791) 789 (795)

4 SS 784 826 565 (596) 462 (505) 715 (726)

Note SFL = standardized factor loading Alpha = Cronbachrsquos alpha CR = composite reliability

All factor loadings and correlations are significant at p lt 001

- 38 -

Table 5

Results of CFA in the individual countries Standardized factor loadings and measures of fit

Factors

amp Items

Australia Belgium China Denmark France Italy

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

672

756

799

785

804

943

710

703

597

790

799

730

584

675

685

681

731

742

578

748

674

643

765

739

645

704

678

708

812

779

632

604

765

766

672

853

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

552

593

560

699

727

632

624

605

560

713

709

645

702

602

616

724

686

735

567

560

715

654

605

811

608

658

587

696

755

698

673

620

523

782

697

654

COM

COM1

COM2

COM3

COM5

COM6

623

584

662

688

701

739

732

608

475

808

666

581

578

616

605

649

670

673

699

751

737

679

816

776

568

546

801

626

673

850

637

546

741

715

601

822

SS

SS2

SS3

SS4

SS7

654

794

806

892

767

813

727

693

799

793

725

838

857

820

773

896

855

842

806

749

792

829

753

884

847

731

753

889

816

753

815

692

664

795

782

734

Inter-factor

correlations

093

~

752

163

~

810

320

~

773

323

~

825

452

~

827

497

~

855

323

~

850

321

~

847

498

~

899

536

~

902

425

~

829

476

~

865

Sample size 249 214 230 249 266 162

Measures of Fit

χsup2 638 444 881 732 765 554 817 625 735 565 750 564

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 980 100 939 987 969 997 967 994 975 997 949 995

GFI 961 986 941 986 948 991 949 992 955 992 926 986

RMSEA 036 00 063 050 051 026 053 035 045 026 059 033

- 39 -

Table 5 (Continued)

Factors

amp Items

Korea Lithuania The Netherlands Switzerland UK USA

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

619

701

667

706

781

717

666

512

682

751

605

779

557

629

630

620

719

729

535

621

702

643

694

735

655

725

738

710

804

822

535

621

702

643

694

735

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

594

607

722

722

644

722

674

535

533

790

589

711

568

554

538

655

628

545

627

580

615

711

666

704

800

512

555

873

581

646

627

580

615

711

666

704

COM

COM1

COM2

COM3

COM5

COM6

694

677

550

763

736

627

557

639

428

550

696

534

737

623

508

766

616

608

659

592

683

703

661

769

682

583

729

739

644

803

659

592

683

703

661

769

SS

SS2

SS3

SS4

SS7

776

752

679

779

770

811

707

721

635

725

734

706

814

656

743

820

767

764

745

632

689

788

662

744

671

818

832

830

864

802

745

632

689

788

662

744

Inter-factor

correlations

432

~

726

466

~

746

376

~

788

406

~

902

377

~

852

455

~

836

637

~

818

660

~

828

317

~

852

360

~

852

637

~

818

660

~

828

Sample size 253 236 249 250 260 250

Measures of Fit

χsup2 896 573 612 462 773 616 815 715 769 542 815 715

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 951 996 975 100 958 993 957 991 972 998 957 991

GFI 942 988 960 989 951 987 948 990 955 990 948 990

RMSEA 059 028 034 00 050 034 053 044 048 022 053 044

Note All factor loadings and correlations are significant at p lt 001

- 40 -

Appendix 1 Dimensions and possible items of PSM (the first round)

1 Attraction to Public Participation

(1) I am interested in developing public programs helping my country or community (Kim 2009a)

(2) Sharing my views on public policies with others is attractive to me (Kim 2009a)

(3) I am satisfied when seeing people get benefits from the public programs I was involved in (Kim 2009a)

(4) I like to discuss political subjects with others (Giauque et al 2011)

(5) I think the governmental activities contribute to our welfare

(6) I like to initiate actions to help out my community

(7) I do not care much about politicians (Perry 1996)

2 Commitment to Public Values

Possible items for public interest

(1A) Meaningful public service is very important to me (Perry 1996)

(2A) It is important for me to contribute for the common good (Giauque et al 2011)

(3A) I recognize myself with the mission of protection of liberties and rights endorsed by the nation-state

(Giauque et al 2011)

(4A) I would prefer seeing public officials do what is best for the whole community even if it harmed my

interests (Perry 1996)

(5A) To me serving the public interest is more important than helping other people (Vandenabeele 2008b)

Possible items for public values

(1B) I recognize myself with the promotion of the equality of chances endorsed by the nation-state (Giauque et

al 2011)

(2B) It is fundamental that the public service provided to the citizen is regular and continuous (Giauque et al

2011)

(3B) It is fundamental that public service innovates to respond to the new needs of the citizensclients (Giauque

et al 2011)

(4B) The dignity and well-being of all should be the most important concerns in any society (Steenbergen

1996)

(5B) It is important that public servants account for all the costsexpenses they make (Vandenabeele 2008b)

(6B) We have to do everything in our power to pursue the goal of democracy

3 Compassion

(1) It is difficult for me to contain my feelings when I see people in distress (Perry 1996)

(2) I feel sympathetic to the plight of the underprivileged (Kim 2009a)

(3) I am often reminded by daily events how dependent we are on one another (Perry 1996)

(4) I feel compassion for others in the difficulties they are facing (Steffen and Masters 2005)

(5) It is not really my problem if others are in trouble and need help (Beutel and Marini 1995)

(6) I find it hard to be sympathetic toward starving people in foreign lands when there is so much trouble in our

own country (Beutel and Marini 1995)

(7) I get very upset when I see other people treated unfairly (Beutel and Marini 1995)

(8) I would agree to a good plan to make a better life for the poor even if it cost me money (Beutel and Marini

1995)

(9) To me patriotism includes seeing the welfare of others (Perry 1996)

(10) I care very much about other people

4 Self-Sacrifice

(1) Making a difference in society means more to me than personal achievements (Perry 1996)

(2) I am prepared to make enormous sacrifices for the good of society (Perry 1996)

(3) I believe in putting duty before self (Perry 1996)

(4) I am one of those rare people who would risk personal loss to help someone else (Perry 1996)

(5) I think people should give back to society more than they get from it (Perry 1996)

(6) Serving other citizens would give me a good feeling even if no one paid me for it (Perry 1996)

(7) Much of what I do is for a cause bigger than myself (Perry 1996)

- 41 -

Appendix 2 Newly suggested items through the first round

1 Items for the Attraction to Public Participation Dimension

(1) In the public domain every person should be counted as equal

(2) I enjoy to think and argue in a political way

(3) I am satisfied when seeing my ideas being integrated in public policies

(4) I can easily relate to the way of thinking and arguing of politicians

(5) Working in the public sector I can better realize myself

2 Items for the Public Values Sub-Dimension

(1) Public service should give value for money

(2) Cost considerations should play an important role in decisions about public service provision

(3) It is fundamental that the interests of future generations are taken into account when developing public

policies

(4) To act with correct ethics is for public servants as important as their professional competence

(5) Primarily public servants should be accountable to the public and not to their superiors

3 Items for the Self-Sacrifice Dimension

(1) I would agree to a good plan to make a better life for the poor even if it costs me money

- 42 -

Appendix 3 An example of a summary of comments and suggestions (the second round)

On the item ―I am interested in developing public programs helping my country or community

A It is a good sentence but is it possible for everyone to ―develop public programs Even public officials are

limited to do this

B OK

C or the special group

D Double barrel and difficult to translate to my country ―I am interested in helping to improve public services

E Good but I feel that wording requires revision If we draw sample from local government then country is

redundant ―I am interested in developing public policy programs that aim in helping my community

F 4 (General appreciation from1 to 4) Not clear whether it refers to ones work or to ones concern as a citizen

―(In my work) I am interested in contributing to the development of public etchellip

G Instead of development I would use implementing

H Delete public which is redundant because of the characteristic helping my country or community but add

or implementing ―I am interested in developing or implementing programs helping my country or

community

I Referring to dimension Wording Similar to item 6

Note Names have been changed to alphabets in order to ensure the anonymity of participants

- 43 -

Appendix 4 Dimensions and Items of Public Service Motivation for Future Research

Dimensions and items

Attraction to Public Service (APS)

1 I am interested in helping to improve public service

2 I admire people who initiate or are involved in activities to aid my community

3 It is important to contribute to activities that tackle social problems

4 Meaningful public service is very important to me

5 It is important for me to contribute to the common good

Commitment to Public Values (CPV)

1 I think equal opportunities for citizens are very important

2 It is important that citizens can rely on the continuous provision of public services

3 It is fundamental that the interests of future generations are taken into account when

developing public policies

4 To act ethically is essential for public servants

5 I believe that public employees must always be aware of the legitimacy of their activities

6 I personally identify with the aim of protecting individual liberties and rights

Compassion (COM)

1 It is difficult for me to contain my feelings when I see people in distress

2 I feel sympathetic to the plight of the underprivileged

3 I empathize with other people who face difficulties

4 I get very upset when I see other people being treated unfairly

5 Considering the welfare of others is very important

Self-Sacrifice (SS)

1 I am prepared to make sacrifices for the good of society

2 I believe in putting civic duty before self

3 I am willing to risk personal loss to help society

4 I would agree to a good plan to make a better life for the poor even if it costs me money

- 31 -

Perry James L 1996 Measuring Public Service Motivation An Assessment of Construct

Reliability and Validity Journal of Public Administration Research and Theory 6

5-22

Perry James L 1997 Antecedents of Public Service Motivation Journal of Public

Administration Research and Theory 7 181-197

Perry James L 2000 Bringing Society in Toward a Theory of Public Service Motivation

Journal of Public Administration Research and Theory 10 471-488

Perry James L and Annie Hondeghem 2008a Motivation in Public Management Oxford

Oxford University Press

Perry James L and Annie Hondeghem 2008b Directions for Future Theory and Research

In Motivation in Public Management edited by James L Perry and Annie

Hondeghem 294-313 Oxford Oxford University Press

Perry James L Annie Hondeghem and Lois Recascino Wise 2010 Revisiting the

Motivational Bases of Public Service Twenty Years of Research and an Agenda for

the Future Public Administration Review 70(5) 681-690

Perry James L and Wouter Vandenabeele 2008 Behavioral dynamics Institutions

Identities and Self-Regulation In Motivation in Public Management edited by

James L Perry and Annie Hondeghem 56-79 Oxford Oxford University Press

Perry James L and Lois Recascino Wise 1990 The Motivational Bases of Public Service

Public Administration Review 50 367-373

Rainey Hal G 1982 Reward Preference among Public and Private Managers In Search of

the Service Ethic American Review of Public Administration 16 288-302

Ringle Christian M Oliver Goumltz Martin Wetzels and Bradley Wilson 2009 On the Use of

Formative Measurement Specifications in Structural Equation Modeling A Monte

Carlo Simulation Study to Compare Covariance-Based and Partial Least Squares

Model Estimation Methodologies Faculty of Economics and Business Administration

Maastricht Univeristy

Ringle C M S Wende and S Will 2005 Smart PLS 20 (M3) Beta

- 32 -

httpwwwsmartplsde

Ritz Adrian 2011 Attraction to Public Policy Making A Qualitative Inquiry into

Improvements in PSM Measurement Forthcoming in Public Administration

Spector Paul E 1992 Summated Rating Scales Construction An Introduction Newbury

Park CA Sage

Steenbergen Marco R 1996 Compassion and American public opinion An analysis of the

NES Humanitarianism Scale Report to the Board of Overseers of the National

Election Studies

Steffen Patrick R and Kevin S Masters 2005 Does compassion mediate the intrinsic

religion-health relationship Annals of Behavioral Medicine 30 217-223

Taylor Jeannette 2007 The Impact of Public Service Motives on Work Outcomes in

Australia A Comparative Multi-Dimensional Analysis Public Administration 85

931-959

Vandenabeele Wouter 2007 Towards a Theory of Public Service Motivation An

Institutional Approach Public Management Review 9 545-556

Vandenabeele Wouter 2008a Government Calling Public Service Motivation as an Element

in Selecting Government as an Employer of Choice Public Administration 86

1089-1105

Vandenabeele Wouter 2008b Development of a Public Service Motivation Measurement

Scale Corroborating and Extending Perryrsquos Measurement Instrument International

Public Management Journal 11 143-167

Vandenabeele Wouter 2009 The Mediating Effect of Job Satisfaction and Organizational

Commitment on Self-Reported Performance More Robust Evidence of the PSM-

Performance Relationship International Review of Administrative Sciences 75 11-

34

Wilcox James B Roy D Howell and Einar Breivik 2008 Questions about Formative

Measurement Journal of Business Research 61 1219-1228

Wise Lois Recascino 2000 The Public Service Culture In Public Administration Concepts

- 33 -

and Cases 7th ed edited by Richard J Stillman II 342-353 Boston Houghton

Mifflin

Wright Bradley E 2008 Methodological Challenges associated with Public Service

Motivation Research In Motivation in Public Management edited by James L

Perry and Annie Hondeghem 80-98 Oxford Oxford University Press

Wright Bradley E and Sanjay K Pandey 2008 Public Service Motivation and the

Assumption of Person-Organization Fit Testing the Mediating Effect of Value

Congruence Administration amp Society 40 502-521

- 34 -

Table 1

Possible Indicators of PSM and Descriptive Statistics (n = 2868)

Dimensions and items Mean Standard

Deviation

Item-Total

Correlation

Attraction to Public Participation (APP)

APP1 I am interested in helping to improve public service

APP2 I am satisfied when I see people benefiting from the public programs I was

involved in

APP3 I like to discuss topics regarding public programs and policies with others

APP4 I believe that public sector activities contribute to our general welfare

APP5 I admire people who initiate or are involved in activities to aid my

community

APP6 Contributing to public programs and policies helps me realize myself

APP7 It is important to contribute to activities that tackle social problems

412

441

355

393

423

351

409

696

662

917

753

672

881

734

524

421

482

412

493

478

552

Commitment to Public Values (CPV)

Sub-dimension for Public Interests (CPV1)

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

CPI3 I would prefer seeing public officials do what is best for the whole

community even if it harmed my interests

CPI4 Serving the public interest is more important than helping a single individual

Sub-dimension for Public Values (CPV2)

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of public

services

CPV3 It is fundamental that public services respond to the needs of the citizens

CPV4 Decisions regarding public services should be democratic despite the time

and effort it takes

CPV5 Everybody is entitled to a good service even if it costs a lot of money

CPV6 It is fundamental that the interests of future generations are taken into

account when developing public policies

CPV7 To act ethically is essential for public servants

CPV8 I believe that public employees must always be aware of the legitimacy of

their activities

CPV9 I personally identify with the aim of protecting individual liberties and rights

418

409

348

337

428

416

433

394

367

426

441

434

414

695

662

899

995

730

704

710

852

932

689

681

692

763

525

596

480

251

484

437

356

325

376

439

462

402

392

Compassion (COM) COM1 It is difficult for me to contain my feelings when I see people in distress

COM2 I feel sympathetic to the plight of the underprivileged

COM3 I empathize with other people who face difficulties

COM4 I have little compassion for people in need who are unwilling to take the

first step to help themselves

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

354

387

388

299

412

391

951

825

707

1083

755

765

399

511

389

011

487

502

Self-Sacrifice (SS)

SS1 Making a difference to society means more to me than personal achievements

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS5 People should give back to society more than they get from it

SS6 Serving other citizens would give me a good feeling even if no one paid me for

it

SS7 I would agree to a good plan to make a better life for the poor even if it costs

me money

357

329

339

302

345

363

350

904

872

949

982

888

1049

895

508

584

454

531

412

368

546

- 35 -

Table 2

Background of Respondents (n = 2868)

Variables Characteristics Respondents

Sex

Male

Female

NA

1209

1622

37

422

566

13

Length of service

(years)

0 ~ 10

10 ~ 20

20 ~ 30

30+

NA

1126

723

484

491

44

393

252

169

171

15

Organizational status

―Do you supervise employees

No

Yes

NA

1472

1076

320

513

375

112

Country

Australia

Belgium

China

Denmark

France

Italy

Korea

Lithuania

The Netherlands

Switzerland

United Kingdom

USA

249

214

230

249

266

162

253

236

249

250

260

250

87

75

80

87

93

56

88

82

87

87

91

87 Note NA = no answer

- 36 -

Table 3

Standardized factor loadings and correlations for the 31-item PSM index (n = 2868)

Dimensions and items SFL (MLE) SFL (D-WLS) Alpha

APP

APP1

APP2

APP4

APP5

APP6

APP7

597

486

455

558

490

600

657

552

514

625

549

665

699

CPV1

CPI1

CPI2

CPI3

CPI4

588

660

488

289

653

732

535

294

542

CPV2

CPV1

CPV2

CPV3

CPV4

CPV5

CPV6

CPV7

CPV8

CPV9

611

546

436

393

418

534

598

496

460

690

631

505

457

486

601

669

573

530

745

COM

COM1

COM2

COM3

COM5

COM6

496

569

498

597

607

516

651

538

670

684

687

SS

SS1

SS2

SS3

SS4

SS5

SS6

SS7

567

736

558

728

467

409

663

660

762

591

695

515

494

709

782

Measures of fit (MLE) χsup2 (df = 424) = 43699 p lt 001 CFI = 835 GFI = 899

RMSEA = 057 [055 059]

Measure of fit (D-WLS) χsup2 (df = 424) = 39247 p lt 001 CFI = 971 GFI = 976

RMSEA = 054 [052 055]

Inter-factor correlations MLE (D-WLS)

1 2 3 4

1 APS

2 CPV1 996 (988)

3 CPV2 902 (890) 798 (798)

4 COM 787 (783) 758 (762) 779 (783)

5 SS 700 (742) 800 (859) 509 (537) 720 (729)

Note SFL = standardized factor loading All factor loadings and correlations are significant at p lt 001

- 37 -

Table 4

Standardized factor loadings and correlations for the 16-item PSM index (n = 2868)

Dimensions and items SFL

(MLE)

SFL

(D-WLS)

APS

APP5 I admire people who initiate or are involved in activities to aid my

community

APP7 It is important to contribute to activities that tackle social problems

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

611

592

637

666

665

669

690

751

CPV

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of

public services

CPV6 It is fundamental that the interests of future generations are taken

into account when developing public policies

CPV7 To act ethically is essential for public servants

612

546

551

590

698

631

618

658

COM

COM1 It is difficult for me to contain my feelings when I see people in

distress

COM2 I feel sympathetic to the plight of the underprivileged

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

490

546

596

604

515

608

669

679

SS

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS7 I would agree to a good plan to make a better life for the poor even if

it costs me money

763

571

766

678

807

629

747

755

Measures of fit (MLE)

χsup2 (df = 98) = 8490 p lt 001

CFI = 939 GFI = 965

RMSEA = 052 [049 055]

Measure of fit (D-WLS)

χsup2 (df = 98) = 6069 p lt 001

CFI = 988 GFI = 992

RMSEA = 043 [039 046]

Alpha CR (D-WLS) Inter-factor correlations MLE (D-WLS)

1 APS 717 788 1 APS 2 CPV 3 COM

2 CPV 663 747 848 (859)

3 COM 645 713 781 (791) 789 (795)

4 SS 784 826 565 (596) 462 (505) 715 (726)

Note SFL = standardized factor loading Alpha = Cronbachrsquos alpha CR = composite reliability

All factor loadings and correlations are significant at p lt 001

- 38 -

Table 5

Results of CFA in the individual countries Standardized factor loadings and measures of fit

Factors

amp Items

Australia Belgium China Denmark France Italy

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

672

756

799

785

804

943

710

703

597

790

799

730

584

675

685

681

731

742

578

748

674

643

765

739

645

704

678

708

812

779

632

604

765

766

672

853

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

552

593

560

699

727

632

624

605

560

713

709

645

702

602

616

724

686

735

567

560

715

654

605

811

608

658

587

696

755

698

673

620

523

782

697

654

COM

COM1

COM2

COM3

COM5

COM6

623

584

662

688

701

739

732

608

475

808

666

581

578

616

605

649

670

673

699

751

737

679

816

776

568

546

801

626

673

850

637

546

741

715

601

822

SS

SS2

SS3

SS4

SS7

654

794

806

892

767

813

727

693

799

793

725

838

857

820

773

896

855

842

806

749

792

829

753

884

847

731

753

889

816

753

815

692

664

795

782

734

Inter-factor

correlations

093

~

752

163

~

810

320

~

773

323

~

825

452

~

827

497

~

855

323

~

850

321

~

847

498

~

899

536

~

902

425

~

829

476

~

865

Sample size 249 214 230 249 266 162

Measures of Fit

χsup2 638 444 881 732 765 554 817 625 735 565 750 564

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 980 100 939 987 969 997 967 994 975 997 949 995

GFI 961 986 941 986 948 991 949 992 955 992 926 986

RMSEA 036 00 063 050 051 026 053 035 045 026 059 033

- 39 -

Table 5 (Continued)

Factors

amp Items

Korea Lithuania The Netherlands Switzerland UK USA

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

619

701

667

706

781

717

666

512

682

751

605

779

557

629

630

620

719

729

535

621

702

643

694

735

655

725

738

710

804

822

535

621

702

643

694

735

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

594

607

722

722

644

722

674

535

533

790

589

711

568

554

538

655

628

545

627

580

615

711

666

704

800

512

555

873

581

646

627

580

615

711

666

704

COM

COM1

COM2

COM3

COM5

COM6

694

677

550

763

736

627

557

639

428

550

696

534

737

623

508

766

616

608

659

592

683

703

661

769

682

583

729

739

644

803

659

592

683

703

661

769

SS

SS2

SS3

SS4

SS7

776

752

679

779

770

811

707

721

635

725

734

706

814

656

743

820

767

764

745

632

689

788

662

744

671

818

832

830

864

802

745

632

689

788

662

744

Inter-factor

correlations

432

~

726

466

~

746

376

~

788

406

~

902

377

~

852

455

~

836

637

~

818

660

~

828

317

~

852

360

~

852

637

~

818

660

~

828

Sample size 253 236 249 250 260 250

Measures of Fit

χsup2 896 573 612 462 773 616 815 715 769 542 815 715

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 951 996 975 100 958 993 957 991 972 998 957 991

GFI 942 988 960 989 951 987 948 990 955 990 948 990

RMSEA 059 028 034 00 050 034 053 044 048 022 053 044

Note All factor loadings and correlations are significant at p lt 001

- 40 -

Appendix 1 Dimensions and possible items of PSM (the first round)

1 Attraction to Public Participation

(1) I am interested in developing public programs helping my country or community (Kim 2009a)

(2) Sharing my views on public policies with others is attractive to me (Kim 2009a)

(3) I am satisfied when seeing people get benefits from the public programs I was involved in (Kim 2009a)

(4) I like to discuss political subjects with others (Giauque et al 2011)

(5) I think the governmental activities contribute to our welfare

(6) I like to initiate actions to help out my community

(7) I do not care much about politicians (Perry 1996)

2 Commitment to Public Values

Possible items for public interest

(1A) Meaningful public service is very important to me (Perry 1996)

(2A) It is important for me to contribute for the common good (Giauque et al 2011)

(3A) I recognize myself with the mission of protection of liberties and rights endorsed by the nation-state

(Giauque et al 2011)

(4A) I would prefer seeing public officials do what is best for the whole community even if it harmed my

interests (Perry 1996)

(5A) To me serving the public interest is more important than helping other people (Vandenabeele 2008b)

Possible items for public values

(1B) I recognize myself with the promotion of the equality of chances endorsed by the nation-state (Giauque et

al 2011)

(2B) It is fundamental that the public service provided to the citizen is regular and continuous (Giauque et al

2011)

(3B) It is fundamental that public service innovates to respond to the new needs of the citizensclients (Giauque

et al 2011)

(4B) The dignity and well-being of all should be the most important concerns in any society (Steenbergen

1996)

(5B) It is important that public servants account for all the costsexpenses they make (Vandenabeele 2008b)

(6B) We have to do everything in our power to pursue the goal of democracy

3 Compassion

(1) It is difficult for me to contain my feelings when I see people in distress (Perry 1996)

(2) I feel sympathetic to the plight of the underprivileged (Kim 2009a)

(3) I am often reminded by daily events how dependent we are on one another (Perry 1996)

(4) I feel compassion for others in the difficulties they are facing (Steffen and Masters 2005)

(5) It is not really my problem if others are in trouble and need help (Beutel and Marini 1995)

(6) I find it hard to be sympathetic toward starving people in foreign lands when there is so much trouble in our

own country (Beutel and Marini 1995)

(7) I get very upset when I see other people treated unfairly (Beutel and Marini 1995)

(8) I would agree to a good plan to make a better life for the poor even if it cost me money (Beutel and Marini

1995)

(9) To me patriotism includes seeing the welfare of others (Perry 1996)

(10) I care very much about other people

4 Self-Sacrifice

(1) Making a difference in society means more to me than personal achievements (Perry 1996)

(2) I am prepared to make enormous sacrifices for the good of society (Perry 1996)

(3) I believe in putting duty before self (Perry 1996)

(4) I am one of those rare people who would risk personal loss to help someone else (Perry 1996)

(5) I think people should give back to society more than they get from it (Perry 1996)

(6) Serving other citizens would give me a good feeling even if no one paid me for it (Perry 1996)

(7) Much of what I do is for a cause bigger than myself (Perry 1996)

- 41 -

Appendix 2 Newly suggested items through the first round

1 Items for the Attraction to Public Participation Dimension

(1) In the public domain every person should be counted as equal

(2) I enjoy to think and argue in a political way

(3) I am satisfied when seeing my ideas being integrated in public policies

(4) I can easily relate to the way of thinking and arguing of politicians

(5) Working in the public sector I can better realize myself

2 Items for the Public Values Sub-Dimension

(1) Public service should give value for money

(2) Cost considerations should play an important role in decisions about public service provision

(3) It is fundamental that the interests of future generations are taken into account when developing public

policies

(4) To act with correct ethics is for public servants as important as their professional competence

(5) Primarily public servants should be accountable to the public and not to their superiors

3 Items for the Self-Sacrifice Dimension

(1) I would agree to a good plan to make a better life for the poor even if it costs me money

- 42 -

Appendix 3 An example of a summary of comments and suggestions (the second round)

On the item ―I am interested in developing public programs helping my country or community

A It is a good sentence but is it possible for everyone to ―develop public programs Even public officials are

limited to do this

B OK

C or the special group

D Double barrel and difficult to translate to my country ―I am interested in helping to improve public services

E Good but I feel that wording requires revision If we draw sample from local government then country is

redundant ―I am interested in developing public policy programs that aim in helping my community

F 4 (General appreciation from1 to 4) Not clear whether it refers to ones work or to ones concern as a citizen

―(In my work) I am interested in contributing to the development of public etchellip

G Instead of development I would use implementing

H Delete public which is redundant because of the characteristic helping my country or community but add

or implementing ―I am interested in developing or implementing programs helping my country or

community

I Referring to dimension Wording Similar to item 6

Note Names have been changed to alphabets in order to ensure the anonymity of participants

- 43 -

Appendix 4 Dimensions and Items of Public Service Motivation for Future Research

Dimensions and items

Attraction to Public Service (APS)

1 I am interested in helping to improve public service

2 I admire people who initiate or are involved in activities to aid my community

3 It is important to contribute to activities that tackle social problems

4 Meaningful public service is very important to me

5 It is important for me to contribute to the common good

Commitment to Public Values (CPV)

1 I think equal opportunities for citizens are very important

2 It is important that citizens can rely on the continuous provision of public services

3 It is fundamental that the interests of future generations are taken into account when

developing public policies

4 To act ethically is essential for public servants

5 I believe that public employees must always be aware of the legitimacy of their activities

6 I personally identify with the aim of protecting individual liberties and rights

Compassion (COM)

1 It is difficult for me to contain my feelings when I see people in distress

2 I feel sympathetic to the plight of the underprivileged

3 I empathize with other people who face difficulties

4 I get very upset when I see other people being treated unfairly

5 Considering the welfare of others is very important

Self-Sacrifice (SS)

1 I am prepared to make sacrifices for the good of society

2 I believe in putting civic duty before self

3 I am willing to risk personal loss to help society

4 I would agree to a good plan to make a better life for the poor even if it costs me money

- 32 -

httpwwwsmartplsde

Ritz Adrian 2011 Attraction to Public Policy Making A Qualitative Inquiry into

Improvements in PSM Measurement Forthcoming in Public Administration

Spector Paul E 1992 Summated Rating Scales Construction An Introduction Newbury

Park CA Sage

Steenbergen Marco R 1996 Compassion and American public opinion An analysis of the

NES Humanitarianism Scale Report to the Board of Overseers of the National

Election Studies

Steffen Patrick R and Kevin S Masters 2005 Does compassion mediate the intrinsic

religion-health relationship Annals of Behavioral Medicine 30 217-223

Taylor Jeannette 2007 The Impact of Public Service Motives on Work Outcomes in

Australia A Comparative Multi-Dimensional Analysis Public Administration 85

931-959

Vandenabeele Wouter 2007 Towards a Theory of Public Service Motivation An

Institutional Approach Public Management Review 9 545-556

Vandenabeele Wouter 2008a Government Calling Public Service Motivation as an Element

in Selecting Government as an Employer of Choice Public Administration 86

1089-1105

Vandenabeele Wouter 2008b Development of a Public Service Motivation Measurement

Scale Corroborating and Extending Perryrsquos Measurement Instrument International

Public Management Journal 11 143-167

Vandenabeele Wouter 2009 The Mediating Effect of Job Satisfaction and Organizational

Commitment on Self-Reported Performance More Robust Evidence of the PSM-

Performance Relationship International Review of Administrative Sciences 75 11-

34

Wilcox James B Roy D Howell and Einar Breivik 2008 Questions about Formative

Measurement Journal of Business Research 61 1219-1228

Wise Lois Recascino 2000 The Public Service Culture In Public Administration Concepts

- 33 -

and Cases 7th ed edited by Richard J Stillman II 342-353 Boston Houghton

Mifflin

Wright Bradley E 2008 Methodological Challenges associated with Public Service

Motivation Research In Motivation in Public Management edited by James L

Perry and Annie Hondeghem 80-98 Oxford Oxford University Press

Wright Bradley E and Sanjay K Pandey 2008 Public Service Motivation and the

Assumption of Person-Organization Fit Testing the Mediating Effect of Value

Congruence Administration amp Society 40 502-521

- 34 -

Table 1

Possible Indicators of PSM and Descriptive Statistics (n = 2868)

Dimensions and items Mean Standard

Deviation

Item-Total

Correlation

Attraction to Public Participation (APP)

APP1 I am interested in helping to improve public service

APP2 I am satisfied when I see people benefiting from the public programs I was

involved in

APP3 I like to discuss topics regarding public programs and policies with others

APP4 I believe that public sector activities contribute to our general welfare

APP5 I admire people who initiate or are involved in activities to aid my

community

APP6 Contributing to public programs and policies helps me realize myself

APP7 It is important to contribute to activities that tackle social problems

412

441

355

393

423

351

409

696

662

917

753

672

881

734

524

421

482

412

493

478

552

Commitment to Public Values (CPV)

Sub-dimension for Public Interests (CPV1)

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

CPI3 I would prefer seeing public officials do what is best for the whole

community even if it harmed my interests

CPI4 Serving the public interest is more important than helping a single individual

Sub-dimension for Public Values (CPV2)

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of public

services

CPV3 It is fundamental that public services respond to the needs of the citizens

CPV4 Decisions regarding public services should be democratic despite the time

and effort it takes

CPV5 Everybody is entitled to a good service even if it costs a lot of money

CPV6 It is fundamental that the interests of future generations are taken into

account when developing public policies

CPV7 To act ethically is essential for public servants

CPV8 I believe that public employees must always be aware of the legitimacy of

their activities

CPV9 I personally identify with the aim of protecting individual liberties and rights

418

409

348

337

428

416

433

394

367

426

441

434

414

695

662

899

995

730

704

710

852

932

689

681

692

763

525

596

480

251

484

437

356

325

376

439

462

402

392

Compassion (COM) COM1 It is difficult for me to contain my feelings when I see people in distress

COM2 I feel sympathetic to the plight of the underprivileged

COM3 I empathize with other people who face difficulties

COM4 I have little compassion for people in need who are unwilling to take the

first step to help themselves

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

354

387

388

299

412

391

951

825

707

1083

755

765

399

511

389

011

487

502

Self-Sacrifice (SS)

SS1 Making a difference to society means more to me than personal achievements

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS5 People should give back to society more than they get from it

SS6 Serving other citizens would give me a good feeling even if no one paid me for

it

SS7 I would agree to a good plan to make a better life for the poor even if it costs

me money

357

329

339

302

345

363

350

904

872

949

982

888

1049

895

508

584

454

531

412

368

546

- 35 -

Table 2

Background of Respondents (n = 2868)

Variables Characteristics Respondents

Sex

Male

Female

NA

1209

1622

37

422

566

13

Length of service

(years)

0 ~ 10

10 ~ 20

20 ~ 30

30+

NA

1126

723

484

491

44

393

252

169

171

15

Organizational status

―Do you supervise employees

No

Yes

NA

1472

1076

320

513

375

112

Country

Australia

Belgium

China

Denmark

France

Italy

Korea

Lithuania

The Netherlands

Switzerland

United Kingdom

USA

249

214

230

249

266

162

253

236

249

250

260

250

87

75

80

87

93

56

88

82

87

87

91

87 Note NA = no answer

- 36 -

Table 3

Standardized factor loadings and correlations for the 31-item PSM index (n = 2868)

Dimensions and items SFL (MLE) SFL (D-WLS) Alpha

APP

APP1

APP2

APP4

APP5

APP6

APP7

597

486

455

558

490

600

657

552

514

625

549

665

699

CPV1

CPI1

CPI2

CPI3

CPI4

588

660

488

289

653

732

535

294

542

CPV2

CPV1

CPV2

CPV3

CPV4

CPV5

CPV6

CPV7

CPV8

CPV9

611

546

436

393

418

534

598

496

460

690

631

505

457

486

601

669

573

530

745

COM

COM1

COM2

COM3

COM5

COM6

496

569

498

597

607

516

651

538

670

684

687

SS

SS1

SS2

SS3

SS4

SS5

SS6

SS7

567

736

558

728

467

409

663

660

762

591

695

515

494

709

782

Measures of fit (MLE) χsup2 (df = 424) = 43699 p lt 001 CFI = 835 GFI = 899

RMSEA = 057 [055 059]

Measure of fit (D-WLS) χsup2 (df = 424) = 39247 p lt 001 CFI = 971 GFI = 976

RMSEA = 054 [052 055]

Inter-factor correlations MLE (D-WLS)

1 2 3 4

1 APS

2 CPV1 996 (988)

3 CPV2 902 (890) 798 (798)

4 COM 787 (783) 758 (762) 779 (783)

5 SS 700 (742) 800 (859) 509 (537) 720 (729)

Note SFL = standardized factor loading All factor loadings and correlations are significant at p lt 001

- 37 -

Table 4

Standardized factor loadings and correlations for the 16-item PSM index (n = 2868)

Dimensions and items SFL

(MLE)

SFL

(D-WLS)

APS

APP5 I admire people who initiate or are involved in activities to aid my

community

APP7 It is important to contribute to activities that tackle social problems

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

611

592

637

666

665

669

690

751

CPV

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of

public services

CPV6 It is fundamental that the interests of future generations are taken

into account when developing public policies

CPV7 To act ethically is essential for public servants

612

546

551

590

698

631

618

658

COM

COM1 It is difficult for me to contain my feelings when I see people in

distress

COM2 I feel sympathetic to the plight of the underprivileged

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

490

546

596

604

515

608

669

679

SS

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS7 I would agree to a good plan to make a better life for the poor even if

it costs me money

763

571

766

678

807

629

747

755

Measures of fit (MLE)

χsup2 (df = 98) = 8490 p lt 001

CFI = 939 GFI = 965

RMSEA = 052 [049 055]

Measure of fit (D-WLS)

χsup2 (df = 98) = 6069 p lt 001

CFI = 988 GFI = 992

RMSEA = 043 [039 046]

Alpha CR (D-WLS) Inter-factor correlations MLE (D-WLS)

1 APS 717 788 1 APS 2 CPV 3 COM

2 CPV 663 747 848 (859)

3 COM 645 713 781 (791) 789 (795)

4 SS 784 826 565 (596) 462 (505) 715 (726)

Note SFL = standardized factor loading Alpha = Cronbachrsquos alpha CR = composite reliability

All factor loadings and correlations are significant at p lt 001

- 38 -

Table 5

Results of CFA in the individual countries Standardized factor loadings and measures of fit

Factors

amp Items

Australia Belgium China Denmark France Italy

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

672

756

799

785

804

943

710

703

597

790

799

730

584

675

685

681

731

742

578

748

674

643

765

739

645

704

678

708

812

779

632

604

765

766

672

853

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

552

593

560

699

727

632

624

605

560

713

709

645

702

602

616

724

686

735

567

560

715

654

605

811

608

658

587

696

755

698

673

620

523

782

697

654

COM

COM1

COM2

COM3

COM5

COM6

623

584

662

688

701

739

732

608

475

808

666

581

578

616

605

649

670

673

699

751

737

679

816

776

568

546

801

626

673

850

637

546

741

715

601

822

SS

SS2

SS3

SS4

SS7

654

794

806

892

767

813

727

693

799

793

725

838

857

820

773

896

855

842

806

749

792

829

753

884

847

731

753

889

816

753

815

692

664

795

782

734

Inter-factor

correlations

093

~

752

163

~

810

320

~

773

323

~

825

452

~

827

497

~

855

323

~

850

321

~

847

498

~

899

536

~

902

425

~

829

476

~

865

Sample size 249 214 230 249 266 162

Measures of Fit

χsup2 638 444 881 732 765 554 817 625 735 565 750 564

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 980 100 939 987 969 997 967 994 975 997 949 995

GFI 961 986 941 986 948 991 949 992 955 992 926 986

RMSEA 036 00 063 050 051 026 053 035 045 026 059 033

- 39 -

Table 5 (Continued)

Factors

amp Items

Korea Lithuania The Netherlands Switzerland UK USA

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

619

701

667

706

781

717

666

512

682

751

605

779

557

629

630

620

719

729

535

621

702

643

694

735

655

725

738

710

804

822

535

621

702

643

694

735

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

594

607

722

722

644

722

674

535

533

790

589

711

568

554

538

655

628

545

627

580

615

711

666

704

800

512

555

873

581

646

627

580

615

711

666

704

COM

COM1

COM2

COM3

COM5

COM6

694

677

550

763

736

627

557

639

428

550

696

534

737

623

508

766

616

608

659

592

683

703

661

769

682

583

729

739

644

803

659

592

683

703

661

769

SS

SS2

SS3

SS4

SS7

776

752

679

779

770

811

707

721

635

725

734

706

814

656

743

820

767

764

745

632

689

788

662

744

671

818

832

830

864

802

745

632

689

788

662

744

Inter-factor

correlations

432

~

726

466

~

746

376

~

788

406

~

902

377

~

852

455

~

836

637

~

818

660

~

828

317

~

852

360

~

852

637

~

818

660

~

828

Sample size 253 236 249 250 260 250

Measures of Fit

χsup2 896 573 612 462 773 616 815 715 769 542 815 715

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 951 996 975 100 958 993 957 991 972 998 957 991

GFI 942 988 960 989 951 987 948 990 955 990 948 990

RMSEA 059 028 034 00 050 034 053 044 048 022 053 044

Note All factor loadings and correlations are significant at p lt 001

- 40 -

Appendix 1 Dimensions and possible items of PSM (the first round)

1 Attraction to Public Participation

(1) I am interested in developing public programs helping my country or community (Kim 2009a)

(2) Sharing my views on public policies with others is attractive to me (Kim 2009a)

(3) I am satisfied when seeing people get benefits from the public programs I was involved in (Kim 2009a)

(4) I like to discuss political subjects with others (Giauque et al 2011)

(5) I think the governmental activities contribute to our welfare

(6) I like to initiate actions to help out my community

(7) I do not care much about politicians (Perry 1996)

2 Commitment to Public Values

Possible items for public interest

(1A) Meaningful public service is very important to me (Perry 1996)

(2A) It is important for me to contribute for the common good (Giauque et al 2011)

(3A) I recognize myself with the mission of protection of liberties and rights endorsed by the nation-state

(Giauque et al 2011)

(4A) I would prefer seeing public officials do what is best for the whole community even if it harmed my

interests (Perry 1996)

(5A) To me serving the public interest is more important than helping other people (Vandenabeele 2008b)

Possible items for public values

(1B) I recognize myself with the promotion of the equality of chances endorsed by the nation-state (Giauque et

al 2011)

(2B) It is fundamental that the public service provided to the citizen is regular and continuous (Giauque et al

2011)

(3B) It is fundamental that public service innovates to respond to the new needs of the citizensclients (Giauque

et al 2011)

(4B) The dignity and well-being of all should be the most important concerns in any society (Steenbergen

1996)

(5B) It is important that public servants account for all the costsexpenses they make (Vandenabeele 2008b)

(6B) We have to do everything in our power to pursue the goal of democracy

3 Compassion

(1) It is difficult for me to contain my feelings when I see people in distress (Perry 1996)

(2) I feel sympathetic to the plight of the underprivileged (Kim 2009a)

(3) I am often reminded by daily events how dependent we are on one another (Perry 1996)

(4) I feel compassion for others in the difficulties they are facing (Steffen and Masters 2005)

(5) It is not really my problem if others are in trouble and need help (Beutel and Marini 1995)

(6) I find it hard to be sympathetic toward starving people in foreign lands when there is so much trouble in our

own country (Beutel and Marini 1995)

(7) I get very upset when I see other people treated unfairly (Beutel and Marini 1995)

(8) I would agree to a good plan to make a better life for the poor even if it cost me money (Beutel and Marini

1995)

(9) To me patriotism includes seeing the welfare of others (Perry 1996)

(10) I care very much about other people

4 Self-Sacrifice

(1) Making a difference in society means more to me than personal achievements (Perry 1996)

(2) I am prepared to make enormous sacrifices for the good of society (Perry 1996)

(3) I believe in putting duty before self (Perry 1996)

(4) I am one of those rare people who would risk personal loss to help someone else (Perry 1996)

(5) I think people should give back to society more than they get from it (Perry 1996)

(6) Serving other citizens would give me a good feeling even if no one paid me for it (Perry 1996)

(7) Much of what I do is for a cause bigger than myself (Perry 1996)

- 41 -

Appendix 2 Newly suggested items through the first round

1 Items for the Attraction to Public Participation Dimension

(1) In the public domain every person should be counted as equal

(2) I enjoy to think and argue in a political way

(3) I am satisfied when seeing my ideas being integrated in public policies

(4) I can easily relate to the way of thinking and arguing of politicians

(5) Working in the public sector I can better realize myself

2 Items for the Public Values Sub-Dimension

(1) Public service should give value for money

(2) Cost considerations should play an important role in decisions about public service provision

(3) It is fundamental that the interests of future generations are taken into account when developing public

policies

(4) To act with correct ethics is for public servants as important as their professional competence

(5) Primarily public servants should be accountable to the public and not to their superiors

3 Items for the Self-Sacrifice Dimension

(1) I would agree to a good plan to make a better life for the poor even if it costs me money

- 42 -

Appendix 3 An example of a summary of comments and suggestions (the second round)

On the item ―I am interested in developing public programs helping my country or community

A It is a good sentence but is it possible for everyone to ―develop public programs Even public officials are

limited to do this

B OK

C or the special group

D Double barrel and difficult to translate to my country ―I am interested in helping to improve public services

E Good but I feel that wording requires revision If we draw sample from local government then country is

redundant ―I am interested in developing public policy programs that aim in helping my community

F 4 (General appreciation from1 to 4) Not clear whether it refers to ones work or to ones concern as a citizen

―(In my work) I am interested in contributing to the development of public etchellip

G Instead of development I would use implementing

H Delete public which is redundant because of the characteristic helping my country or community but add

or implementing ―I am interested in developing or implementing programs helping my country or

community

I Referring to dimension Wording Similar to item 6

Note Names have been changed to alphabets in order to ensure the anonymity of participants

- 43 -

Appendix 4 Dimensions and Items of Public Service Motivation for Future Research

Dimensions and items

Attraction to Public Service (APS)

1 I am interested in helping to improve public service

2 I admire people who initiate or are involved in activities to aid my community

3 It is important to contribute to activities that tackle social problems

4 Meaningful public service is very important to me

5 It is important for me to contribute to the common good

Commitment to Public Values (CPV)

1 I think equal opportunities for citizens are very important

2 It is important that citizens can rely on the continuous provision of public services

3 It is fundamental that the interests of future generations are taken into account when

developing public policies

4 To act ethically is essential for public servants

5 I believe that public employees must always be aware of the legitimacy of their activities

6 I personally identify with the aim of protecting individual liberties and rights

Compassion (COM)

1 It is difficult for me to contain my feelings when I see people in distress

2 I feel sympathetic to the plight of the underprivileged

3 I empathize with other people who face difficulties

4 I get very upset when I see other people being treated unfairly

5 Considering the welfare of others is very important

Self-Sacrifice (SS)

1 I am prepared to make sacrifices for the good of society

2 I believe in putting civic duty before self

3 I am willing to risk personal loss to help society

4 I would agree to a good plan to make a better life for the poor even if it costs me money

- 33 -

and Cases 7th ed edited by Richard J Stillman II 342-353 Boston Houghton

Mifflin

Wright Bradley E 2008 Methodological Challenges associated with Public Service

Motivation Research In Motivation in Public Management edited by James L

Perry and Annie Hondeghem 80-98 Oxford Oxford University Press

Wright Bradley E and Sanjay K Pandey 2008 Public Service Motivation and the

Assumption of Person-Organization Fit Testing the Mediating Effect of Value

Congruence Administration amp Society 40 502-521

- 34 -

Table 1

Possible Indicators of PSM and Descriptive Statistics (n = 2868)

Dimensions and items Mean Standard

Deviation

Item-Total

Correlation

Attraction to Public Participation (APP)

APP1 I am interested in helping to improve public service

APP2 I am satisfied when I see people benefiting from the public programs I was

involved in

APP3 I like to discuss topics regarding public programs and policies with others

APP4 I believe that public sector activities contribute to our general welfare

APP5 I admire people who initiate or are involved in activities to aid my

community

APP6 Contributing to public programs and policies helps me realize myself

APP7 It is important to contribute to activities that tackle social problems

412

441

355

393

423

351

409

696

662

917

753

672

881

734

524

421

482

412

493

478

552

Commitment to Public Values (CPV)

Sub-dimension for Public Interests (CPV1)

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

CPI3 I would prefer seeing public officials do what is best for the whole

community even if it harmed my interests

CPI4 Serving the public interest is more important than helping a single individual

Sub-dimension for Public Values (CPV2)

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of public

services

CPV3 It is fundamental that public services respond to the needs of the citizens

CPV4 Decisions regarding public services should be democratic despite the time

and effort it takes

CPV5 Everybody is entitled to a good service even if it costs a lot of money

CPV6 It is fundamental that the interests of future generations are taken into

account when developing public policies

CPV7 To act ethically is essential for public servants

CPV8 I believe that public employees must always be aware of the legitimacy of

their activities

CPV9 I personally identify with the aim of protecting individual liberties and rights

418

409

348

337

428

416

433

394

367

426

441

434

414

695

662

899

995

730

704

710

852

932

689

681

692

763

525

596

480

251

484

437

356

325

376

439

462

402

392

Compassion (COM) COM1 It is difficult for me to contain my feelings when I see people in distress

COM2 I feel sympathetic to the plight of the underprivileged

COM3 I empathize with other people who face difficulties

COM4 I have little compassion for people in need who are unwilling to take the

first step to help themselves

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

354

387

388

299

412

391

951

825

707

1083

755

765

399

511

389

011

487

502

Self-Sacrifice (SS)

SS1 Making a difference to society means more to me than personal achievements

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS5 People should give back to society more than they get from it

SS6 Serving other citizens would give me a good feeling even if no one paid me for

it

SS7 I would agree to a good plan to make a better life for the poor even if it costs

me money

357

329

339

302

345

363

350

904

872

949

982

888

1049

895

508

584

454

531

412

368

546

- 35 -

Table 2

Background of Respondents (n = 2868)

Variables Characteristics Respondents

Sex

Male

Female

NA

1209

1622

37

422

566

13

Length of service

(years)

0 ~ 10

10 ~ 20

20 ~ 30

30+

NA

1126

723

484

491

44

393

252

169

171

15

Organizational status

―Do you supervise employees

No

Yes

NA

1472

1076

320

513

375

112

Country

Australia

Belgium

China

Denmark

France

Italy

Korea

Lithuania

The Netherlands

Switzerland

United Kingdom

USA

249

214

230

249

266

162

253

236

249

250

260

250

87

75

80

87

93

56

88

82

87

87

91

87 Note NA = no answer

- 36 -

Table 3

Standardized factor loadings and correlations for the 31-item PSM index (n = 2868)

Dimensions and items SFL (MLE) SFL (D-WLS) Alpha

APP

APP1

APP2

APP4

APP5

APP6

APP7

597

486

455

558

490

600

657

552

514

625

549

665

699

CPV1

CPI1

CPI2

CPI3

CPI4

588

660

488

289

653

732

535

294

542

CPV2

CPV1

CPV2

CPV3

CPV4

CPV5

CPV6

CPV7

CPV8

CPV9

611

546

436

393

418

534

598

496

460

690

631

505

457

486

601

669

573

530

745

COM

COM1

COM2

COM3

COM5

COM6

496

569

498

597

607

516

651

538

670

684

687

SS

SS1

SS2

SS3

SS4

SS5

SS6

SS7

567

736

558

728

467

409

663

660

762

591

695

515

494

709

782

Measures of fit (MLE) χsup2 (df = 424) = 43699 p lt 001 CFI = 835 GFI = 899

RMSEA = 057 [055 059]

Measure of fit (D-WLS) χsup2 (df = 424) = 39247 p lt 001 CFI = 971 GFI = 976

RMSEA = 054 [052 055]

Inter-factor correlations MLE (D-WLS)

1 2 3 4

1 APS

2 CPV1 996 (988)

3 CPV2 902 (890) 798 (798)

4 COM 787 (783) 758 (762) 779 (783)

5 SS 700 (742) 800 (859) 509 (537) 720 (729)

Note SFL = standardized factor loading All factor loadings and correlations are significant at p lt 001

- 37 -

Table 4

Standardized factor loadings and correlations for the 16-item PSM index (n = 2868)

Dimensions and items SFL

(MLE)

SFL

(D-WLS)

APS

APP5 I admire people who initiate or are involved in activities to aid my

community

APP7 It is important to contribute to activities that tackle social problems

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

611

592

637

666

665

669

690

751

CPV

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of

public services

CPV6 It is fundamental that the interests of future generations are taken

into account when developing public policies

CPV7 To act ethically is essential for public servants

612

546

551

590

698

631

618

658

COM

COM1 It is difficult for me to contain my feelings when I see people in

distress

COM2 I feel sympathetic to the plight of the underprivileged

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

490

546

596

604

515

608

669

679

SS

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS7 I would agree to a good plan to make a better life for the poor even if

it costs me money

763

571

766

678

807

629

747

755

Measures of fit (MLE)

χsup2 (df = 98) = 8490 p lt 001

CFI = 939 GFI = 965

RMSEA = 052 [049 055]

Measure of fit (D-WLS)

χsup2 (df = 98) = 6069 p lt 001

CFI = 988 GFI = 992

RMSEA = 043 [039 046]

Alpha CR (D-WLS) Inter-factor correlations MLE (D-WLS)

1 APS 717 788 1 APS 2 CPV 3 COM

2 CPV 663 747 848 (859)

3 COM 645 713 781 (791) 789 (795)

4 SS 784 826 565 (596) 462 (505) 715 (726)

Note SFL = standardized factor loading Alpha = Cronbachrsquos alpha CR = composite reliability

All factor loadings and correlations are significant at p lt 001

- 38 -

Table 5

Results of CFA in the individual countries Standardized factor loadings and measures of fit

Factors

amp Items

Australia Belgium China Denmark France Italy

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

672

756

799

785

804

943

710

703

597

790

799

730

584

675

685

681

731

742

578

748

674

643

765

739

645

704

678

708

812

779

632

604

765

766

672

853

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

552

593

560

699

727

632

624

605

560

713

709

645

702

602

616

724

686

735

567

560

715

654

605

811

608

658

587

696

755

698

673

620

523

782

697

654

COM

COM1

COM2

COM3

COM5

COM6

623

584

662

688

701

739

732

608

475

808

666

581

578

616

605

649

670

673

699

751

737

679

816

776

568

546

801

626

673

850

637

546

741

715

601

822

SS

SS2

SS3

SS4

SS7

654

794

806

892

767

813

727

693

799

793

725

838

857

820

773

896

855

842

806

749

792

829

753

884

847

731

753

889

816

753

815

692

664

795

782

734

Inter-factor

correlations

093

~

752

163

~

810

320

~

773

323

~

825

452

~

827

497

~

855

323

~

850

321

~

847

498

~

899

536

~

902

425

~

829

476

~

865

Sample size 249 214 230 249 266 162

Measures of Fit

χsup2 638 444 881 732 765 554 817 625 735 565 750 564

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 980 100 939 987 969 997 967 994 975 997 949 995

GFI 961 986 941 986 948 991 949 992 955 992 926 986

RMSEA 036 00 063 050 051 026 053 035 045 026 059 033

- 39 -

Table 5 (Continued)

Factors

amp Items

Korea Lithuania The Netherlands Switzerland UK USA

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

619

701

667

706

781

717

666

512

682

751

605

779

557

629

630

620

719

729

535

621

702

643

694

735

655

725

738

710

804

822

535

621

702

643

694

735

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

594

607

722

722

644

722

674

535

533

790

589

711

568

554

538

655

628

545

627

580

615

711

666

704

800

512

555

873

581

646

627

580

615

711

666

704

COM

COM1

COM2

COM3

COM5

COM6

694

677

550

763

736

627

557

639

428

550

696

534

737

623

508

766

616

608

659

592

683

703

661

769

682

583

729

739

644

803

659

592

683

703

661

769

SS

SS2

SS3

SS4

SS7

776

752

679

779

770

811

707

721

635

725

734

706

814

656

743

820

767

764

745

632

689

788

662

744

671

818

832

830

864

802

745

632

689

788

662

744

Inter-factor

correlations

432

~

726

466

~

746

376

~

788

406

~

902

377

~

852

455

~

836

637

~

818

660

~

828

317

~

852

360

~

852

637

~

818

660

~

828

Sample size 253 236 249 250 260 250

Measures of Fit

χsup2 896 573 612 462 773 616 815 715 769 542 815 715

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 951 996 975 100 958 993 957 991 972 998 957 991

GFI 942 988 960 989 951 987 948 990 955 990 948 990

RMSEA 059 028 034 00 050 034 053 044 048 022 053 044

Note All factor loadings and correlations are significant at p lt 001

- 40 -

Appendix 1 Dimensions and possible items of PSM (the first round)

1 Attraction to Public Participation

(1) I am interested in developing public programs helping my country or community (Kim 2009a)

(2) Sharing my views on public policies with others is attractive to me (Kim 2009a)

(3) I am satisfied when seeing people get benefits from the public programs I was involved in (Kim 2009a)

(4) I like to discuss political subjects with others (Giauque et al 2011)

(5) I think the governmental activities contribute to our welfare

(6) I like to initiate actions to help out my community

(7) I do not care much about politicians (Perry 1996)

2 Commitment to Public Values

Possible items for public interest

(1A) Meaningful public service is very important to me (Perry 1996)

(2A) It is important for me to contribute for the common good (Giauque et al 2011)

(3A) I recognize myself with the mission of protection of liberties and rights endorsed by the nation-state

(Giauque et al 2011)

(4A) I would prefer seeing public officials do what is best for the whole community even if it harmed my

interests (Perry 1996)

(5A) To me serving the public interest is more important than helping other people (Vandenabeele 2008b)

Possible items for public values

(1B) I recognize myself with the promotion of the equality of chances endorsed by the nation-state (Giauque et

al 2011)

(2B) It is fundamental that the public service provided to the citizen is regular and continuous (Giauque et al

2011)

(3B) It is fundamental that public service innovates to respond to the new needs of the citizensclients (Giauque

et al 2011)

(4B) The dignity and well-being of all should be the most important concerns in any society (Steenbergen

1996)

(5B) It is important that public servants account for all the costsexpenses they make (Vandenabeele 2008b)

(6B) We have to do everything in our power to pursue the goal of democracy

3 Compassion

(1) It is difficult for me to contain my feelings when I see people in distress (Perry 1996)

(2) I feel sympathetic to the plight of the underprivileged (Kim 2009a)

(3) I am often reminded by daily events how dependent we are on one another (Perry 1996)

(4) I feel compassion for others in the difficulties they are facing (Steffen and Masters 2005)

(5) It is not really my problem if others are in trouble and need help (Beutel and Marini 1995)

(6) I find it hard to be sympathetic toward starving people in foreign lands when there is so much trouble in our

own country (Beutel and Marini 1995)

(7) I get very upset when I see other people treated unfairly (Beutel and Marini 1995)

(8) I would agree to a good plan to make a better life for the poor even if it cost me money (Beutel and Marini

1995)

(9) To me patriotism includes seeing the welfare of others (Perry 1996)

(10) I care very much about other people

4 Self-Sacrifice

(1) Making a difference in society means more to me than personal achievements (Perry 1996)

(2) I am prepared to make enormous sacrifices for the good of society (Perry 1996)

(3) I believe in putting duty before self (Perry 1996)

(4) I am one of those rare people who would risk personal loss to help someone else (Perry 1996)

(5) I think people should give back to society more than they get from it (Perry 1996)

(6) Serving other citizens would give me a good feeling even if no one paid me for it (Perry 1996)

(7) Much of what I do is for a cause bigger than myself (Perry 1996)

- 41 -

Appendix 2 Newly suggested items through the first round

1 Items for the Attraction to Public Participation Dimension

(1) In the public domain every person should be counted as equal

(2) I enjoy to think and argue in a political way

(3) I am satisfied when seeing my ideas being integrated in public policies

(4) I can easily relate to the way of thinking and arguing of politicians

(5) Working in the public sector I can better realize myself

2 Items for the Public Values Sub-Dimension

(1) Public service should give value for money

(2) Cost considerations should play an important role in decisions about public service provision

(3) It is fundamental that the interests of future generations are taken into account when developing public

policies

(4) To act with correct ethics is for public servants as important as their professional competence

(5) Primarily public servants should be accountable to the public and not to their superiors

3 Items for the Self-Sacrifice Dimension

(1) I would agree to a good plan to make a better life for the poor even if it costs me money

- 42 -

Appendix 3 An example of a summary of comments and suggestions (the second round)

On the item ―I am interested in developing public programs helping my country or community

A It is a good sentence but is it possible for everyone to ―develop public programs Even public officials are

limited to do this

B OK

C or the special group

D Double barrel and difficult to translate to my country ―I am interested in helping to improve public services

E Good but I feel that wording requires revision If we draw sample from local government then country is

redundant ―I am interested in developing public policy programs that aim in helping my community

F 4 (General appreciation from1 to 4) Not clear whether it refers to ones work or to ones concern as a citizen

―(In my work) I am interested in contributing to the development of public etchellip

G Instead of development I would use implementing

H Delete public which is redundant because of the characteristic helping my country or community but add

or implementing ―I am interested in developing or implementing programs helping my country or

community

I Referring to dimension Wording Similar to item 6

Note Names have been changed to alphabets in order to ensure the anonymity of participants

- 43 -

Appendix 4 Dimensions and Items of Public Service Motivation for Future Research

Dimensions and items

Attraction to Public Service (APS)

1 I am interested in helping to improve public service

2 I admire people who initiate or are involved in activities to aid my community

3 It is important to contribute to activities that tackle social problems

4 Meaningful public service is very important to me

5 It is important for me to contribute to the common good

Commitment to Public Values (CPV)

1 I think equal opportunities for citizens are very important

2 It is important that citizens can rely on the continuous provision of public services

3 It is fundamental that the interests of future generations are taken into account when

developing public policies

4 To act ethically is essential for public servants

5 I believe that public employees must always be aware of the legitimacy of their activities

6 I personally identify with the aim of protecting individual liberties and rights

Compassion (COM)

1 It is difficult for me to contain my feelings when I see people in distress

2 I feel sympathetic to the plight of the underprivileged

3 I empathize with other people who face difficulties

4 I get very upset when I see other people being treated unfairly

5 Considering the welfare of others is very important

Self-Sacrifice (SS)

1 I am prepared to make sacrifices for the good of society

2 I believe in putting civic duty before self

3 I am willing to risk personal loss to help society

4 I would agree to a good plan to make a better life for the poor even if it costs me money

- 34 -

Table 1

Possible Indicators of PSM and Descriptive Statistics (n = 2868)

Dimensions and items Mean Standard

Deviation

Item-Total

Correlation

Attraction to Public Participation (APP)

APP1 I am interested in helping to improve public service

APP2 I am satisfied when I see people benefiting from the public programs I was

involved in

APP3 I like to discuss topics regarding public programs and policies with others

APP4 I believe that public sector activities contribute to our general welfare

APP5 I admire people who initiate or are involved in activities to aid my

community

APP6 Contributing to public programs and policies helps me realize myself

APP7 It is important to contribute to activities that tackle social problems

412

441

355

393

423

351

409

696

662

917

753

672

881

734

524

421

482

412

493

478

552

Commitment to Public Values (CPV)

Sub-dimension for Public Interests (CPV1)

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

CPI3 I would prefer seeing public officials do what is best for the whole

community even if it harmed my interests

CPI4 Serving the public interest is more important than helping a single individual

Sub-dimension for Public Values (CPV2)

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of public

services

CPV3 It is fundamental that public services respond to the needs of the citizens

CPV4 Decisions regarding public services should be democratic despite the time

and effort it takes

CPV5 Everybody is entitled to a good service even if it costs a lot of money

CPV6 It is fundamental that the interests of future generations are taken into

account when developing public policies

CPV7 To act ethically is essential for public servants

CPV8 I believe that public employees must always be aware of the legitimacy of

their activities

CPV9 I personally identify with the aim of protecting individual liberties and rights

418

409

348

337

428

416

433

394

367

426

441

434

414

695

662

899

995

730

704

710

852

932

689

681

692

763

525

596

480

251

484

437

356

325

376

439

462

402

392

Compassion (COM) COM1 It is difficult for me to contain my feelings when I see people in distress

COM2 I feel sympathetic to the plight of the underprivileged

COM3 I empathize with other people who face difficulties

COM4 I have little compassion for people in need who are unwilling to take the

first step to help themselves

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

354

387

388

299

412

391

951

825

707

1083

755

765

399

511

389

011

487

502

Self-Sacrifice (SS)

SS1 Making a difference to society means more to me than personal achievements

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS5 People should give back to society more than they get from it

SS6 Serving other citizens would give me a good feeling even if no one paid me for

it

SS7 I would agree to a good plan to make a better life for the poor even if it costs

me money

357

329

339

302

345

363

350

904

872

949

982

888

1049

895

508

584

454

531

412

368

546

- 35 -

Table 2

Background of Respondents (n = 2868)

Variables Characteristics Respondents

Sex

Male

Female

NA

1209

1622

37

422

566

13

Length of service

(years)

0 ~ 10

10 ~ 20

20 ~ 30

30+

NA

1126

723

484

491

44

393

252

169

171

15

Organizational status

―Do you supervise employees

No

Yes

NA

1472

1076

320

513

375

112

Country

Australia

Belgium

China

Denmark

France

Italy

Korea

Lithuania

The Netherlands

Switzerland

United Kingdom

USA

249

214

230

249

266

162

253

236

249

250

260

250

87

75

80

87

93

56

88

82

87

87

91

87 Note NA = no answer

- 36 -

Table 3

Standardized factor loadings and correlations for the 31-item PSM index (n = 2868)

Dimensions and items SFL (MLE) SFL (D-WLS) Alpha

APP

APP1

APP2

APP4

APP5

APP6

APP7

597

486

455

558

490

600

657

552

514

625

549

665

699

CPV1

CPI1

CPI2

CPI3

CPI4

588

660

488

289

653

732

535

294

542

CPV2

CPV1

CPV2

CPV3

CPV4

CPV5

CPV6

CPV7

CPV8

CPV9

611

546

436

393

418

534

598

496

460

690

631

505

457

486

601

669

573

530

745

COM

COM1

COM2

COM3

COM5

COM6

496

569

498

597

607

516

651

538

670

684

687

SS

SS1

SS2

SS3

SS4

SS5

SS6

SS7

567

736

558

728

467

409

663

660

762

591

695

515

494

709

782

Measures of fit (MLE) χsup2 (df = 424) = 43699 p lt 001 CFI = 835 GFI = 899

RMSEA = 057 [055 059]

Measure of fit (D-WLS) χsup2 (df = 424) = 39247 p lt 001 CFI = 971 GFI = 976

RMSEA = 054 [052 055]

Inter-factor correlations MLE (D-WLS)

1 2 3 4

1 APS

2 CPV1 996 (988)

3 CPV2 902 (890) 798 (798)

4 COM 787 (783) 758 (762) 779 (783)

5 SS 700 (742) 800 (859) 509 (537) 720 (729)

Note SFL = standardized factor loading All factor loadings and correlations are significant at p lt 001

- 37 -

Table 4

Standardized factor loadings and correlations for the 16-item PSM index (n = 2868)

Dimensions and items SFL

(MLE)

SFL

(D-WLS)

APS

APP5 I admire people who initiate or are involved in activities to aid my

community

APP7 It is important to contribute to activities that tackle social problems

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

611

592

637

666

665

669

690

751

CPV

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of

public services

CPV6 It is fundamental that the interests of future generations are taken

into account when developing public policies

CPV7 To act ethically is essential for public servants

612

546

551

590

698

631

618

658

COM

COM1 It is difficult for me to contain my feelings when I see people in

distress

COM2 I feel sympathetic to the plight of the underprivileged

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

490

546

596

604

515

608

669

679

SS

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS7 I would agree to a good plan to make a better life for the poor even if

it costs me money

763

571

766

678

807

629

747

755

Measures of fit (MLE)

χsup2 (df = 98) = 8490 p lt 001

CFI = 939 GFI = 965

RMSEA = 052 [049 055]

Measure of fit (D-WLS)

χsup2 (df = 98) = 6069 p lt 001

CFI = 988 GFI = 992

RMSEA = 043 [039 046]

Alpha CR (D-WLS) Inter-factor correlations MLE (D-WLS)

1 APS 717 788 1 APS 2 CPV 3 COM

2 CPV 663 747 848 (859)

3 COM 645 713 781 (791) 789 (795)

4 SS 784 826 565 (596) 462 (505) 715 (726)

Note SFL = standardized factor loading Alpha = Cronbachrsquos alpha CR = composite reliability

All factor loadings and correlations are significant at p lt 001

- 38 -

Table 5

Results of CFA in the individual countries Standardized factor loadings and measures of fit

Factors

amp Items

Australia Belgium China Denmark France Italy

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

672

756

799

785

804

943

710

703

597

790

799

730

584

675

685

681

731

742

578

748

674

643

765

739

645

704

678

708

812

779

632

604

765

766

672

853

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

552

593

560

699

727

632

624

605

560

713

709

645

702

602

616

724

686

735

567

560

715

654

605

811

608

658

587

696

755

698

673

620

523

782

697

654

COM

COM1

COM2

COM3

COM5

COM6

623

584

662

688

701

739

732

608

475

808

666

581

578

616

605

649

670

673

699

751

737

679

816

776

568

546

801

626

673

850

637

546

741

715

601

822

SS

SS2

SS3

SS4

SS7

654

794

806

892

767

813

727

693

799

793

725

838

857

820

773

896

855

842

806

749

792

829

753

884

847

731

753

889

816

753

815

692

664

795

782

734

Inter-factor

correlations

093

~

752

163

~

810

320

~

773

323

~

825

452

~

827

497

~

855

323

~

850

321

~

847

498

~

899

536

~

902

425

~

829

476

~

865

Sample size 249 214 230 249 266 162

Measures of Fit

χsup2 638 444 881 732 765 554 817 625 735 565 750 564

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 980 100 939 987 969 997 967 994 975 997 949 995

GFI 961 986 941 986 948 991 949 992 955 992 926 986

RMSEA 036 00 063 050 051 026 053 035 045 026 059 033

- 39 -

Table 5 (Continued)

Factors

amp Items

Korea Lithuania The Netherlands Switzerland UK USA

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

619

701

667

706

781

717

666

512

682

751

605

779

557

629

630

620

719

729

535

621

702

643

694

735

655

725

738

710

804

822

535

621

702

643

694

735

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

594

607

722

722

644

722

674

535

533

790

589

711

568

554

538

655

628

545

627

580

615

711

666

704

800

512

555

873

581

646

627

580

615

711

666

704

COM

COM1

COM2

COM3

COM5

COM6

694

677

550

763

736

627

557

639

428

550

696

534

737

623

508

766

616

608

659

592

683

703

661

769

682

583

729

739

644

803

659

592

683

703

661

769

SS

SS2

SS3

SS4

SS7

776

752

679

779

770

811

707

721

635

725

734

706

814

656

743

820

767

764

745

632

689

788

662

744

671

818

832

830

864

802

745

632

689

788

662

744

Inter-factor

correlations

432

~

726

466

~

746

376

~

788

406

~

902

377

~

852

455

~

836

637

~

818

660

~

828

317

~

852

360

~

852

637

~

818

660

~

828

Sample size 253 236 249 250 260 250

Measures of Fit

χsup2 896 573 612 462 773 616 815 715 769 542 815 715

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 951 996 975 100 958 993 957 991 972 998 957 991

GFI 942 988 960 989 951 987 948 990 955 990 948 990

RMSEA 059 028 034 00 050 034 053 044 048 022 053 044

Note All factor loadings and correlations are significant at p lt 001

- 40 -

Appendix 1 Dimensions and possible items of PSM (the first round)

1 Attraction to Public Participation

(1) I am interested in developing public programs helping my country or community (Kim 2009a)

(2) Sharing my views on public policies with others is attractive to me (Kim 2009a)

(3) I am satisfied when seeing people get benefits from the public programs I was involved in (Kim 2009a)

(4) I like to discuss political subjects with others (Giauque et al 2011)

(5) I think the governmental activities contribute to our welfare

(6) I like to initiate actions to help out my community

(7) I do not care much about politicians (Perry 1996)

2 Commitment to Public Values

Possible items for public interest

(1A) Meaningful public service is very important to me (Perry 1996)

(2A) It is important for me to contribute for the common good (Giauque et al 2011)

(3A) I recognize myself with the mission of protection of liberties and rights endorsed by the nation-state

(Giauque et al 2011)

(4A) I would prefer seeing public officials do what is best for the whole community even if it harmed my

interests (Perry 1996)

(5A) To me serving the public interest is more important than helping other people (Vandenabeele 2008b)

Possible items for public values

(1B) I recognize myself with the promotion of the equality of chances endorsed by the nation-state (Giauque et

al 2011)

(2B) It is fundamental that the public service provided to the citizen is regular and continuous (Giauque et al

2011)

(3B) It is fundamental that public service innovates to respond to the new needs of the citizensclients (Giauque

et al 2011)

(4B) The dignity and well-being of all should be the most important concerns in any society (Steenbergen

1996)

(5B) It is important that public servants account for all the costsexpenses they make (Vandenabeele 2008b)

(6B) We have to do everything in our power to pursue the goal of democracy

3 Compassion

(1) It is difficult for me to contain my feelings when I see people in distress (Perry 1996)

(2) I feel sympathetic to the plight of the underprivileged (Kim 2009a)

(3) I am often reminded by daily events how dependent we are on one another (Perry 1996)

(4) I feel compassion for others in the difficulties they are facing (Steffen and Masters 2005)

(5) It is not really my problem if others are in trouble and need help (Beutel and Marini 1995)

(6) I find it hard to be sympathetic toward starving people in foreign lands when there is so much trouble in our

own country (Beutel and Marini 1995)

(7) I get very upset when I see other people treated unfairly (Beutel and Marini 1995)

(8) I would agree to a good plan to make a better life for the poor even if it cost me money (Beutel and Marini

1995)

(9) To me patriotism includes seeing the welfare of others (Perry 1996)

(10) I care very much about other people

4 Self-Sacrifice

(1) Making a difference in society means more to me than personal achievements (Perry 1996)

(2) I am prepared to make enormous sacrifices for the good of society (Perry 1996)

(3) I believe in putting duty before self (Perry 1996)

(4) I am one of those rare people who would risk personal loss to help someone else (Perry 1996)

(5) I think people should give back to society more than they get from it (Perry 1996)

(6) Serving other citizens would give me a good feeling even if no one paid me for it (Perry 1996)

(7) Much of what I do is for a cause bigger than myself (Perry 1996)

- 41 -

Appendix 2 Newly suggested items through the first round

1 Items for the Attraction to Public Participation Dimension

(1) In the public domain every person should be counted as equal

(2) I enjoy to think and argue in a political way

(3) I am satisfied when seeing my ideas being integrated in public policies

(4) I can easily relate to the way of thinking and arguing of politicians

(5) Working in the public sector I can better realize myself

2 Items for the Public Values Sub-Dimension

(1) Public service should give value for money

(2) Cost considerations should play an important role in decisions about public service provision

(3) It is fundamental that the interests of future generations are taken into account when developing public

policies

(4) To act with correct ethics is for public servants as important as their professional competence

(5) Primarily public servants should be accountable to the public and not to their superiors

3 Items for the Self-Sacrifice Dimension

(1) I would agree to a good plan to make a better life for the poor even if it costs me money

- 42 -

Appendix 3 An example of a summary of comments and suggestions (the second round)

On the item ―I am interested in developing public programs helping my country or community

A It is a good sentence but is it possible for everyone to ―develop public programs Even public officials are

limited to do this

B OK

C or the special group

D Double barrel and difficult to translate to my country ―I am interested in helping to improve public services

E Good but I feel that wording requires revision If we draw sample from local government then country is

redundant ―I am interested in developing public policy programs that aim in helping my community

F 4 (General appreciation from1 to 4) Not clear whether it refers to ones work or to ones concern as a citizen

―(In my work) I am interested in contributing to the development of public etchellip

G Instead of development I would use implementing

H Delete public which is redundant because of the characteristic helping my country or community but add

or implementing ―I am interested in developing or implementing programs helping my country or

community

I Referring to dimension Wording Similar to item 6

Note Names have been changed to alphabets in order to ensure the anonymity of participants

- 43 -

Appendix 4 Dimensions and Items of Public Service Motivation for Future Research

Dimensions and items

Attraction to Public Service (APS)

1 I am interested in helping to improve public service

2 I admire people who initiate or are involved in activities to aid my community

3 It is important to contribute to activities that tackle social problems

4 Meaningful public service is very important to me

5 It is important for me to contribute to the common good

Commitment to Public Values (CPV)

1 I think equal opportunities for citizens are very important

2 It is important that citizens can rely on the continuous provision of public services

3 It is fundamental that the interests of future generations are taken into account when

developing public policies

4 To act ethically is essential for public servants

5 I believe that public employees must always be aware of the legitimacy of their activities

6 I personally identify with the aim of protecting individual liberties and rights

Compassion (COM)

1 It is difficult for me to contain my feelings when I see people in distress

2 I feel sympathetic to the plight of the underprivileged

3 I empathize with other people who face difficulties

4 I get very upset when I see other people being treated unfairly

5 Considering the welfare of others is very important

Self-Sacrifice (SS)

1 I am prepared to make sacrifices for the good of society

2 I believe in putting civic duty before self

3 I am willing to risk personal loss to help society

4 I would agree to a good plan to make a better life for the poor even if it costs me money

- 35 -

Table 2

Background of Respondents (n = 2868)

Variables Characteristics Respondents

Sex

Male

Female

NA

1209

1622

37

422

566

13

Length of service

(years)

0 ~ 10

10 ~ 20

20 ~ 30

30+

NA

1126

723

484

491

44

393

252

169

171

15

Organizational status

―Do you supervise employees

No

Yes

NA

1472

1076

320

513

375

112

Country

Australia

Belgium

China

Denmark

France

Italy

Korea

Lithuania

The Netherlands

Switzerland

United Kingdom

USA

249

214

230

249

266

162

253

236

249

250

260

250

87

75

80

87

93

56

88

82

87

87

91

87 Note NA = no answer

- 36 -

Table 3

Standardized factor loadings and correlations for the 31-item PSM index (n = 2868)

Dimensions and items SFL (MLE) SFL (D-WLS) Alpha

APP

APP1

APP2

APP4

APP5

APP6

APP7

597

486

455

558

490

600

657

552

514

625

549

665

699

CPV1

CPI1

CPI2

CPI3

CPI4

588

660

488

289

653

732

535

294

542

CPV2

CPV1

CPV2

CPV3

CPV4

CPV5

CPV6

CPV7

CPV8

CPV9

611

546

436

393

418

534

598

496

460

690

631

505

457

486

601

669

573

530

745

COM

COM1

COM2

COM3

COM5

COM6

496

569

498

597

607

516

651

538

670

684

687

SS

SS1

SS2

SS3

SS4

SS5

SS6

SS7

567

736

558

728

467

409

663

660

762

591

695

515

494

709

782

Measures of fit (MLE) χsup2 (df = 424) = 43699 p lt 001 CFI = 835 GFI = 899

RMSEA = 057 [055 059]

Measure of fit (D-WLS) χsup2 (df = 424) = 39247 p lt 001 CFI = 971 GFI = 976

RMSEA = 054 [052 055]

Inter-factor correlations MLE (D-WLS)

1 2 3 4

1 APS

2 CPV1 996 (988)

3 CPV2 902 (890) 798 (798)

4 COM 787 (783) 758 (762) 779 (783)

5 SS 700 (742) 800 (859) 509 (537) 720 (729)

Note SFL = standardized factor loading All factor loadings and correlations are significant at p lt 001

- 37 -

Table 4

Standardized factor loadings and correlations for the 16-item PSM index (n = 2868)

Dimensions and items SFL

(MLE)

SFL

(D-WLS)

APS

APP5 I admire people who initiate or are involved in activities to aid my

community

APP7 It is important to contribute to activities that tackle social problems

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

611

592

637

666

665

669

690

751

CPV

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of

public services

CPV6 It is fundamental that the interests of future generations are taken

into account when developing public policies

CPV7 To act ethically is essential for public servants

612

546

551

590

698

631

618

658

COM

COM1 It is difficult for me to contain my feelings when I see people in

distress

COM2 I feel sympathetic to the plight of the underprivileged

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

490

546

596

604

515

608

669

679

SS

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS7 I would agree to a good plan to make a better life for the poor even if

it costs me money

763

571

766

678

807

629

747

755

Measures of fit (MLE)

χsup2 (df = 98) = 8490 p lt 001

CFI = 939 GFI = 965

RMSEA = 052 [049 055]

Measure of fit (D-WLS)

χsup2 (df = 98) = 6069 p lt 001

CFI = 988 GFI = 992

RMSEA = 043 [039 046]

Alpha CR (D-WLS) Inter-factor correlations MLE (D-WLS)

1 APS 717 788 1 APS 2 CPV 3 COM

2 CPV 663 747 848 (859)

3 COM 645 713 781 (791) 789 (795)

4 SS 784 826 565 (596) 462 (505) 715 (726)

Note SFL = standardized factor loading Alpha = Cronbachrsquos alpha CR = composite reliability

All factor loadings and correlations are significant at p lt 001

- 38 -

Table 5

Results of CFA in the individual countries Standardized factor loadings and measures of fit

Factors

amp Items

Australia Belgium China Denmark France Italy

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

672

756

799

785

804

943

710

703

597

790

799

730

584

675

685

681

731

742

578

748

674

643

765

739

645

704

678

708

812

779

632

604

765

766

672

853

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

552

593

560

699

727

632

624

605

560

713

709

645

702

602

616

724

686

735

567

560

715

654

605

811

608

658

587

696

755

698

673

620

523

782

697

654

COM

COM1

COM2

COM3

COM5

COM6

623

584

662

688

701

739

732

608

475

808

666

581

578

616

605

649

670

673

699

751

737

679

816

776

568

546

801

626

673

850

637

546

741

715

601

822

SS

SS2

SS3

SS4

SS7

654

794

806

892

767

813

727

693

799

793

725

838

857

820

773

896

855

842

806

749

792

829

753

884

847

731

753

889

816

753

815

692

664

795

782

734

Inter-factor

correlations

093

~

752

163

~

810

320

~

773

323

~

825

452

~

827

497

~

855

323

~

850

321

~

847

498

~

899

536

~

902

425

~

829

476

~

865

Sample size 249 214 230 249 266 162

Measures of Fit

χsup2 638 444 881 732 765 554 817 625 735 565 750 564

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 980 100 939 987 969 997 967 994 975 997 949 995

GFI 961 986 941 986 948 991 949 992 955 992 926 986

RMSEA 036 00 063 050 051 026 053 035 045 026 059 033

- 39 -

Table 5 (Continued)

Factors

amp Items

Korea Lithuania The Netherlands Switzerland UK USA

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

619

701

667

706

781

717

666

512

682

751

605

779

557

629

630

620

719

729

535

621

702

643

694

735

655

725

738

710

804

822

535

621

702

643

694

735

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

594

607

722

722

644

722

674

535

533

790

589

711

568

554

538

655

628

545

627

580

615

711

666

704

800

512

555

873

581

646

627

580

615

711

666

704

COM

COM1

COM2

COM3

COM5

COM6

694

677

550

763

736

627

557

639

428

550

696

534

737

623

508

766

616

608

659

592

683

703

661

769

682

583

729

739

644

803

659

592

683

703

661

769

SS

SS2

SS3

SS4

SS7

776

752

679

779

770

811

707

721

635

725

734

706

814

656

743

820

767

764

745

632

689

788

662

744

671

818

832

830

864

802

745

632

689

788

662

744

Inter-factor

correlations

432

~

726

466

~

746

376

~

788

406

~

902

377

~

852

455

~

836

637

~

818

660

~

828

317

~

852

360

~

852

637

~

818

660

~

828

Sample size 253 236 249 250 260 250

Measures of Fit

χsup2 896 573 612 462 773 616 815 715 769 542 815 715

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 951 996 975 100 958 993 957 991 972 998 957 991

GFI 942 988 960 989 951 987 948 990 955 990 948 990

RMSEA 059 028 034 00 050 034 053 044 048 022 053 044

Note All factor loadings and correlations are significant at p lt 001

- 40 -

Appendix 1 Dimensions and possible items of PSM (the first round)

1 Attraction to Public Participation

(1) I am interested in developing public programs helping my country or community (Kim 2009a)

(2) Sharing my views on public policies with others is attractive to me (Kim 2009a)

(3) I am satisfied when seeing people get benefits from the public programs I was involved in (Kim 2009a)

(4) I like to discuss political subjects with others (Giauque et al 2011)

(5) I think the governmental activities contribute to our welfare

(6) I like to initiate actions to help out my community

(7) I do not care much about politicians (Perry 1996)

2 Commitment to Public Values

Possible items for public interest

(1A) Meaningful public service is very important to me (Perry 1996)

(2A) It is important for me to contribute for the common good (Giauque et al 2011)

(3A) I recognize myself with the mission of protection of liberties and rights endorsed by the nation-state

(Giauque et al 2011)

(4A) I would prefer seeing public officials do what is best for the whole community even if it harmed my

interests (Perry 1996)

(5A) To me serving the public interest is more important than helping other people (Vandenabeele 2008b)

Possible items for public values

(1B) I recognize myself with the promotion of the equality of chances endorsed by the nation-state (Giauque et

al 2011)

(2B) It is fundamental that the public service provided to the citizen is regular and continuous (Giauque et al

2011)

(3B) It is fundamental that public service innovates to respond to the new needs of the citizensclients (Giauque

et al 2011)

(4B) The dignity and well-being of all should be the most important concerns in any society (Steenbergen

1996)

(5B) It is important that public servants account for all the costsexpenses they make (Vandenabeele 2008b)

(6B) We have to do everything in our power to pursue the goal of democracy

3 Compassion

(1) It is difficult for me to contain my feelings when I see people in distress (Perry 1996)

(2) I feel sympathetic to the plight of the underprivileged (Kim 2009a)

(3) I am often reminded by daily events how dependent we are on one another (Perry 1996)

(4) I feel compassion for others in the difficulties they are facing (Steffen and Masters 2005)

(5) It is not really my problem if others are in trouble and need help (Beutel and Marini 1995)

(6) I find it hard to be sympathetic toward starving people in foreign lands when there is so much trouble in our

own country (Beutel and Marini 1995)

(7) I get very upset when I see other people treated unfairly (Beutel and Marini 1995)

(8) I would agree to a good plan to make a better life for the poor even if it cost me money (Beutel and Marini

1995)

(9) To me patriotism includes seeing the welfare of others (Perry 1996)

(10) I care very much about other people

4 Self-Sacrifice

(1) Making a difference in society means more to me than personal achievements (Perry 1996)

(2) I am prepared to make enormous sacrifices for the good of society (Perry 1996)

(3) I believe in putting duty before self (Perry 1996)

(4) I am one of those rare people who would risk personal loss to help someone else (Perry 1996)

(5) I think people should give back to society more than they get from it (Perry 1996)

(6) Serving other citizens would give me a good feeling even if no one paid me for it (Perry 1996)

(7) Much of what I do is for a cause bigger than myself (Perry 1996)

- 41 -

Appendix 2 Newly suggested items through the first round

1 Items for the Attraction to Public Participation Dimension

(1) In the public domain every person should be counted as equal

(2) I enjoy to think and argue in a political way

(3) I am satisfied when seeing my ideas being integrated in public policies

(4) I can easily relate to the way of thinking and arguing of politicians

(5) Working in the public sector I can better realize myself

2 Items for the Public Values Sub-Dimension

(1) Public service should give value for money

(2) Cost considerations should play an important role in decisions about public service provision

(3) It is fundamental that the interests of future generations are taken into account when developing public

policies

(4) To act with correct ethics is for public servants as important as their professional competence

(5) Primarily public servants should be accountable to the public and not to their superiors

3 Items for the Self-Sacrifice Dimension

(1) I would agree to a good plan to make a better life for the poor even if it costs me money

- 42 -

Appendix 3 An example of a summary of comments and suggestions (the second round)

On the item ―I am interested in developing public programs helping my country or community

A It is a good sentence but is it possible for everyone to ―develop public programs Even public officials are

limited to do this

B OK

C or the special group

D Double barrel and difficult to translate to my country ―I am interested in helping to improve public services

E Good but I feel that wording requires revision If we draw sample from local government then country is

redundant ―I am interested in developing public policy programs that aim in helping my community

F 4 (General appreciation from1 to 4) Not clear whether it refers to ones work or to ones concern as a citizen

―(In my work) I am interested in contributing to the development of public etchellip

G Instead of development I would use implementing

H Delete public which is redundant because of the characteristic helping my country or community but add

or implementing ―I am interested in developing or implementing programs helping my country or

community

I Referring to dimension Wording Similar to item 6

Note Names have been changed to alphabets in order to ensure the anonymity of participants

- 43 -

Appendix 4 Dimensions and Items of Public Service Motivation for Future Research

Dimensions and items

Attraction to Public Service (APS)

1 I am interested in helping to improve public service

2 I admire people who initiate or are involved in activities to aid my community

3 It is important to contribute to activities that tackle social problems

4 Meaningful public service is very important to me

5 It is important for me to contribute to the common good

Commitment to Public Values (CPV)

1 I think equal opportunities for citizens are very important

2 It is important that citizens can rely on the continuous provision of public services

3 It is fundamental that the interests of future generations are taken into account when

developing public policies

4 To act ethically is essential for public servants

5 I believe that public employees must always be aware of the legitimacy of their activities

6 I personally identify with the aim of protecting individual liberties and rights

Compassion (COM)

1 It is difficult for me to contain my feelings when I see people in distress

2 I feel sympathetic to the plight of the underprivileged

3 I empathize with other people who face difficulties

4 I get very upset when I see other people being treated unfairly

5 Considering the welfare of others is very important

Self-Sacrifice (SS)

1 I am prepared to make sacrifices for the good of society

2 I believe in putting civic duty before self

3 I am willing to risk personal loss to help society

4 I would agree to a good plan to make a better life for the poor even if it costs me money

- 36 -

Table 3

Standardized factor loadings and correlations for the 31-item PSM index (n = 2868)

Dimensions and items SFL (MLE) SFL (D-WLS) Alpha

APP

APP1

APP2

APP4

APP5

APP6

APP7

597

486

455

558

490

600

657

552

514

625

549

665

699

CPV1

CPI1

CPI2

CPI3

CPI4

588

660

488

289

653

732

535

294

542

CPV2

CPV1

CPV2

CPV3

CPV4

CPV5

CPV6

CPV7

CPV8

CPV9

611

546

436

393

418

534

598

496

460

690

631

505

457

486

601

669

573

530

745

COM

COM1

COM2

COM3

COM5

COM6

496

569

498

597

607

516

651

538

670

684

687

SS

SS1

SS2

SS3

SS4

SS5

SS6

SS7

567

736

558

728

467

409

663

660

762

591

695

515

494

709

782

Measures of fit (MLE) χsup2 (df = 424) = 43699 p lt 001 CFI = 835 GFI = 899

RMSEA = 057 [055 059]

Measure of fit (D-WLS) χsup2 (df = 424) = 39247 p lt 001 CFI = 971 GFI = 976

RMSEA = 054 [052 055]

Inter-factor correlations MLE (D-WLS)

1 2 3 4

1 APS

2 CPV1 996 (988)

3 CPV2 902 (890) 798 (798)

4 COM 787 (783) 758 (762) 779 (783)

5 SS 700 (742) 800 (859) 509 (537) 720 (729)

Note SFL = standardized factor loading All factor loadings and correlations are significant at p lt 001

- 37 -

Table 4

Standardized factor loadings and correlations for the 16-item PSM index (n = 2868)

Dimensions and items SFL

(MLE)

SFL

(D-WLS)

APS

APP5 I admire people who initiate or are involved in activities to aid my

community

APP7 It is important to contribute to activities that tackle social problems

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

611

592

637

666

665

669

690

751

CPV

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of

public services

CPV6 It is fundamental that the interests of future generations are taken

into account when developing public policies

CPV7 To act ethically is essential for public servants

612

546

551

590

698

631

618

658

COM

COM1 It is difficult for me to contain my feelings when I see people in

distress

COM2 I feel sympathetic to the plight of the underprivileged

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

490

546

596

604

515

608

669

679

SS

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS7 I would agree to a good plan to make a better life for the poor even if

it costs me money

763

571

766

678

807

629

747

755

Measures of fit (MLE)

χsup2 (df = 98) = 8490 p lt 001

CFI = 939 GFI = 965

RMSEA = 052 [049 055]

Measure of fit (D-WLS)

χsup2 (df = 98) = 6069 p lt 001

CFI = 988 GFI = 992

RMSEA = 043 [039 046]

Alpha CR (D-WLS) Inter-factor correlations MLE (D-WLS)

1 APS 717 788 1 APS 2 CPV 3 COM

2 CPV 663 747 848 (859)

3 COM 645 713 781 (791) 789 (795)

4 SS 784 826 565 (596) 462 (505) 715 (726)

Note SFL = standardized factor loading Alpha = Cronbachrsquos alpha CR = composite reliability

All factor loadings and correlations are significant at p lt 001

- 38 -

Table 5

Results of CFA in the individual countries Standardized factor loadings and measures of fit

Factors

amp Items

Australia Belgium China Denmark France Italy

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

672

756

799

785

804

943

710

703

597

790

799

730

584

675

685

681

731

742

578

748

674

643

765

739

645

704

678

708

812

779

632

604

765

766

672

853

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

552

593

560

699

727

632

624

605

560

713

709

645

702

602

616

724

686

735

567

560

715

654

605

811

608

658

587

696

755

698

673

620

523

782

697

654

COM

COM1

COM2

COM3

COM5

COM6

623

584

662

688

701

739

732

608

475

808

666

581

578

616

605

649

670

673

699

751

737

679

816

776

568

546

801

626

673

850

637

546

741

715

601

822

SS

SS2

SS3

SS4

SS7

654

794

806

892

767

813

727

693

799

793

725

838

857

820

773

896

855

842

806

749

792

829

753

884

847

731

753

889

816

753

815

692

664

795

782

734

Inter-factor

correlations

093

~

752

163

~

810

320

~

773

323

~

825

452

~

827

497

~

855

323

~

850

321

~

847

498

~

899

536

~

902

425

~

829

476

~

865

Sample size 249 214 230 249 266 162

Measures of Fit

χsup2 638 444 881 732 765 554 817 625 735 565 750 564

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 980 100 939 987 969 997 967 994 975 997 949 995

GFI 961 986 941 986 948 991 949 992 955 992 926 986

RMSEA 036 00 063 050 051 026 053 035 045 026 059 033

- 39 -

Table 5 (Continued)

Factors

amp Items

Korea Lithuania The Netherlands Switzerland UK USA

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

619

701

667

706

781

717

666

512

682

751

605

779

557

629

630

620

719

729

535

621

702

643

694

735

655

725

738

710

804

822

535

621

702

643

694

735

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

594

607

722

722

644

722

674

535

533

790

589

711

568

554

538

655

628

545

627

580

615

711

666

704

800

512

555

873

581

646

627

580

615

711

666

704

COM

COM1

COM2

COM3

COM5

COM6

694

677

550

763

736

627

557

639

428

550

696

534

737

623

508

766

616

608

659

592

683

703

661

769

682

583

729

739

644

803

659

592

683

703

661

769

SS

SS2

SS3

SS4

SS7

776

752

679

779

770

811

707

721

635

725

734

706

814

656

743

820

767

764

745

632

689

788

662

744

671

818

832

830

864

802

745

632

689

788

662

744

Inter-factor

correlations

432

~

726

466

~

746

376

~

788

406

~

902

377

~

852

455

~

836

637

~

818

660

~

828

317

~

852

360

~

852

637

~

818

660

~

828

Sample size 253 236 249 250 260 250

Measures of Fit

χsup2 896 573 612 462 773 616 815 715 769 542 815 715

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 951 996 975 100 958 993 957 991 972 998 957 991

GFI 942 988 960 989 951 987 948 990 955 990 948 990

RMSEA 059 028 034 00 050 034 053 044 048 022 053 044

Note All factor loadings and correlations are significant at p lt 001

- 40 -

Appendix 1 Dimensions and possible items of PSM (the first round)

1 Attraction to Public Participation

(1) I am interested in developing public programs helping my country or community (Kim 2009a)

(2) Sharing my views on public policies with others is attractive to me (Kim 2009a)

(3) I am satisfied when seeing people get benefits from the public programs I was involved in (Kim 2009a)

(4) I like to discuss political subjects with others (Giauque et al 2011)

(5) I think the governmental activities contribute to our welfare

(6) I like to initiate actions to help out my community

(7) I do not care much about politicians (Perry 1996)

2 Commitment to Public Values

Possible items for public interest

(1A) Meaningful public service is very important to me (Perry 1996)

(2A) It is important for me to contribute for the common good (Giauque et al 2011)

(3A) I recognize myself with the mission of protection of liberties and rights endorsed by the nation-state

(Giauque et al 2011)

(4A) I would prefer seeing public officials do what is best for the whole community even if it harmed my

interests (Perry 1996)

(5A) To me serving the public interest is more important than helping other people (Vandenabeele 2008b)

Possible items for public values

(1B) I recognize myself with the promotion of the equality of chances endorsed by the nation-state (Giauque et

al 2011)

(2B) It is fundamental that the public service provided to the citizen is regular and continuous (Giauque et al

2011)

(3B) It is fundamental that public service innovates to respond to the new needs of the citizensclients (Giauque

et al 2011)

(4B) The dignity and well-being of all should be the most important concerns in any society (Steenbergen

1996)

(5B) It is important that public servants account for all the costsexpenses they make (Vandenabeele 2008b)

(6B) We have to do everything in our power to pursue the goal of democracy

3 Compassion

(1) It is difficult for me to contain my feelings when I see people in distress (Perry 1996)

(2) I feel sympathetic to the plight of the underprivileged (Kim 2009a)

(3) I am often reminded by daily events how dependent we are on one another (Perry 1996)

(4) I feel compassion for others in the difficulties they are facing (Steffen and Masters 2005)

(5) It is not really my problem if others are in trouble and need help (Beutel and Marini 1995)

(6) I find it hard to be sympathetic toward starving people in foreign lands when there is so much trouble in our

own country (Beutel and Marini 1995)

(7) I get very upset when I see other people treated unfairly (Beutel and Marini 1995)

(8) I would agree to a good plan to make a better life for the poor even if it cost me money (Beutel and Marini

1995)

(9) To me patriotism includes seeing the welfare of others (Perry 1996)

(10) I care very much about other people

4 Self-Sacrifice

(1) Making a difference in society means more to me than personal achievements (Perry 1996)

(2) I am prepared to make enormous sacrifices for the good of society (Perry 1996)

(3) I believe in putting duty before self (Perry 1996)

(4) I am one of those rare people who would risk personal loss to help someone else (Perry 1996)

(5) I think people should give back to society more than they get from it (Perry 1996)

(6) Serving other citizens would give me a good feeling even if no one paid me for it (Perry 1996)

(7) Much of what I do is for a cause bigger than myself (Perry 1996)

- 41 -

Appendix 2 Newly suggested items through the first round

1 Items for the Attraction to Public Participation Dimension

(1) In the public domain every person should be counted as equal

(2) I enjoy to think and argue in a political way

(3) I am satisfied when seeing my ideas being integrated in public policies

(4) I can easily relate to the way of thinking and arguing of politicians

(5) Working in the public sector I can better realize myself

2 Items for the Public Values Sub-Dimension

(1) Public service should give value for money

(2) Cost considerations should play an important role in decisions about public service provision

(3) It is fundamental that the interests of future generations are taken into account when developing public

policies

(4) To act with correct ethics is for public servants as important as their professional competence

(5) Primarily public servants should be accountable to the public and not to their superiors

3 Items for the Self-Sacrifice Dimension

(1) I would agree to a good plan to make a better life for the poor even if it costs me money

- 42 -

Appendix 3 An example of a summary of comments and suggestions (the second round)

On the item ―I am interested in developing public programs helping my country or community

A It is a good sentence but is it possible for everyone to ―develop public programs Even public officials are

limited to do this

B OK

C or the special group

D Double barrel and difficult to translate to my country ―I am interested in helping to improve public services

E Good but I feel that wording requires revision If we draw sample from local government then country is

redundant ―I am interested in developing public policy programs that aim in helping my community

F 4 (General appreciation from1 to 4) Not clear whether it refers to ones work or to ones concern as a citizen

―(In my work) I am interested in contributing to the development of public etchellip

G Instead of development I would use implementing

H Delete public which is redundant because of the characteristic helping my country or community but add

or implementing ―I am interested in developing or implementing programs helping my country or

community

I Referring to dimension Wording Similar to item 6

Note Names have been changed to alphabets in order to ensure the anonymity of participants

- 43 -

Appendix 4 Dimensions and Items of Public Service Motivation for Future Research

Dimensions and items

Attraction to Public Service (APS)

1 I am interested in helping to improve public service

2 I admire people who initiate or are involved in activities to aid my community

3 It is important to contribute to activities that tackle social problems

4 Meaningful public service is very important to me

5 It is important for me to contribute to the common good

Commitment to Public Values (CPV)

1 I think equal opportunities for citizens are very important

2 It is important that citizens can rely on the continuous provision of public services

3 It is fundamental that the interests of future generations are taken into account when

developing public policies

4 To act ethically is essential for public servants

5 I believe that public employees must always be aware of the legitimacy of their activities

6 I personally identify with the aim of protecting individual liberties and rights

Compassion (COM)

1 It is difficult for me to contain my feelings when I see people in distress

2 I feel sympathetic to the plight of the underprivileged

3 I empathize with other people who face difficulties

4 I get very upset when I see other people being treated unfairly

5 Considering the welfare of others is very important

Self-Sacrifice (SS)

1 I am prepared to make sacrifices for the good of society

2 I believe in putting civic duty before self

3 I am willing to risk personal loss to help society

4 I would agree to a good plan to make a better life for the poor even if it costs me money

- 37 -

Table 4

Standardized factor loadings and correlations for the 16-item PSM index (n = 2868)

Dimensions and items SFL

(MLE)

SFL

(D-WLS)

APS

APP5 I admire people who initiate or are involved in activities to aid my

community

APP7 It is important to contribute to activities that tackle social problems

CPI1 Meaningful public service is very important to me

CPI2 It is important for me to contribute to the common good

611

592

637

666

665

669

690

751

CPV

CPV1 I think equal opportunities for citizens are very important

CPV2 It is important that citizens can rely on the continuous provision of

public services

CPV6 It is fundamental that the interests of future generations are taken

into account when developing public policies

CPV7 To act ethically is essential for public servants

612

546

551

590

698

631

618

658

COM

COM1 It is difficult for me to contain my feelings when I see people in

distress

COM2 I feel sympathetic to the plight of the underprivileged

COM5 I get very upset when I see other people being treated unfairly

COM6 Considering the welfare of others is very important

490

546

596

604

515

608

669

679

SS

SS2 I am prepared to make sacrifices for the good of society

SS3 I believe in putting civic duty before self

SS4 I am willing to risk personal loss to help society

SS7 I would agree to a good plan to make a better life for the poor even if

it costs me money

763

571

766

678

807

629

747

755

Measures of fit (MLE)

χsup2 (df = 98) = 8490 p lt 001

CFI = 939 GFI = 965

RMSEA = 052 [049 055]

Measure of fit (D-WLS)

χsup2 (df = 98) = 6069 p lt 001

CFI = 988 GFI = 992

RMSEA = 043 [039 046]

Alpha CR (D-WLS) Inter-factor correlations MLE (D-WLS)

1 APS 717 788 1 APS 2 CPV 3 COM

2 CPV 663 747 848 (859)

3 COM 645 713 781 (791) 789 (795)

4 SS 784 826 565 (596) 462 (505) 715 (726)

Note SFL = standardized factor loading Alpha = Cronbachrsquos alpha CR = composite reliability

All factor loadings and correlations are significant at p lt 001

- 38 -

Table 5

Results of CFA in the individual countries Standardized factor loadings and measures of fit

Factors

amp Items

Australia Belgium China Denmark France Italy

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

672

756

799

785

804

943

710

703

597

790

799

730

584

675

685

681

731

742

578

748

674

643

765

739

645

704

678

708

812

779

632

604

765

766

672

853

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

552

593

560

699

727

632

624

605

560

713

709

645

702

602

616

724

686

735

567

560

715

654

605

811

608

658

587

696

755

698

673

620

523

782

697

654

COM

COM1

COM2

COM3

COM5

COM6

623

584

662

688

701

739

732

608

475

808

666

581

578

616

605

649

670

673

699

751

737

679

816

776

568

546

801

626

673

850

637

546

741

715

601

822

SS

SS2

SS3

SS4

SS7

654

794

806

892

767

813

727

693

799

793

725

838

857

820

773

896

855

842

806

749

792

829

753

884

847

731

753

889

816

753

815

692

664

795

782

734

Inter-factor

correlations

093

~

752

163

~

810

320

~

773

323

~

825

452

~

827

497

~

855

323

~

850

321

~

847

498

~

899

536

~

902

425

~

829

476

~

865

Sample size 249 214 230 249 266 162

Measures of Fit

χsup2 638 444 881 732 765 554 817 625 735 565 750 564

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 980 100 939 987 969 997 967 994 975 997 949 995

GFI 961 986 941 986 948 991 949 992 955 992 926 986

RMSEA 036 00 063 050 051 026 053 035 045 026 059 033

- 39 -

Table 5 (Continued)

Factors

amp Items

Korea Lithuania The Netherlands Switzerland UK USA

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

619

701

667

706

781

717

666

512

682

751

605

779

557

629

630

620

719

729

535

621

702

643

694

735

655

725

738

710

804

822

535

621

702

643

694

735

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

594

607

722

722

644

722

674

535

533

790

589

711

568

554

538

655

628

545

627

580

615

711

666

704

800

512

555

873

581

646

627

580

615

711

666

704

COM

COM1

COM2

COM3

COM5

COM6

694

677

550

763

736

627

557

639

428

550

696

534

737

623

508

766

616

608

659

592

683

703

661

769

682

583

729

739

644

803

659

592

683

703

661

769

SS

SS2

SS3

SS4

SS7

776

752

679

779

770

811

707

721

635

725

734

706

814

656

743

820

767

764

745

632

689

788

662

744

671

818

832

830

864

802

745

632

689

788

662

744

Inter-factor

correlations

432

~

726

466

~

746

376

~

788

406

~

902

377

~

852

455

~

836

637

~

818

660

~

828

317

~

852

360

~

852

637

~

818

660

~

828

Sample size 253 236 249 250 260 250

Measures of Fit

χsup2 896 573 612 462 773 616 815 715 769 542 815 715

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 951 996 975 100 958 993 957 991 972 998 957 991

GFI 942 988 960 989 951 987 948 990 955 990 948 990

RMSEA 059 028 034 00 050 034 053 044 048 022 053 044

Note All factor loadings and correlations are significant at p lt 001

- 40 -

Appendix 1 Dimensions and possible items of PSM (the first round)

1 Attraction to Public Participation

(1) I am interested in developing public programs helping my country or community (Kim 2009a)

(2) Sharing my views on public policies with others is attractive to me (Kim 2009a)

(3) I am satisfied when seeing people get benefits from the public programs I was involved in (Kim 2009a)

(4) I like to discuss political subjects with others (Giauque et al 2011)

(5) I think the governmental activities contribute to our welfare

(6) I like to initiate actions to help out my community

(7) I do not care much about politicians (Perry 1996)

2 Commitment to Public Values

Possible items for public interest

(1A) Meaningful public service is very important to me (Perry 1996)

(2A) It is important for me to contribute for the common good (Giauque et al 2011)

(3A) I recognize myself with the mission of protection of liberties and rights endorsed by the nation-state

(Giauque et al 2011)

(4A) I would prefer seeing public officials do what is best for the whole community even if it harmed my

interests (Perry 1996)

(5A) To me serving the public interest is more important than helping other people (Vandenabeele 2008b)

Possible items for public values

(1B) I recognize myself with the promotion of the equality of chances endorsed by the nation-state (Giauque et

al 2011)

(2B) It is fundamental that the public service provided to the citizen is regular and continuous (Giauque et al

2011)

(3B) It is fundamental that public service innovates to respond to the new needs of the citizensclients (Giauque

et al 2011)

(4B) The dignity and well-being of all should be the most important concerns in any society (Steenbergen

1996)

(5B) It is important that public servants account for all the costsexpenses they make (Vandenabeele 2008b)

(6B) We have to do everything in our power to pursue the goal of democracy

3 Compassion

(1) It is difficult for me to contain my feelings when I see people in distress (Perry 1996)

(2) I feel sympathetic to the plight of the underprivileged (Kim 2009a)

(3) I am often reminded by daily events how dependent we are on one another (Perry 1996)

(4) I feel compassion for others in the difficulties they are facing (Steffen and Masters 2005)

(5) It is not really my problem if others are in trouble and need help (Beutel and Marini 1995)

(6) I find it hard to be sympathetic toward starving people in foreign lands when there is so much trouble in our

own country (Beutel and Marini 1995)

(7) I get very upset when I see other people treated unfairly (Beutel and Marini 1995)

(8) I would agree to a good plan to make a better life for the poor even if it cost me money (Beutel and Marini

1995)

(9) To me patriotism includes seeing the welfare of others (Perry 1996)

(10) I care very much about other people

4 Self-Sacrifice

(1) Making a difference in society means more to me than personal achievements (Perry 1996)

(2) I am prepared to make enormous sacrifices for the good of society (Perry 1996)

(3) I believe in putting duty before self (Perry 1996)

(4) I am one of those rare people who would risk personal loss to help someone else (Perry 1996)

(5) I think people should give back to society more than they get from it (Perry 1996)

(6) Serving other citizens would give me a good feeling even if no one paid me for it (Perry 1996)

(7) Much of what I do is for a cause bigger than myself (Perry 1996)

- 41 -

Appendix 2 Newly suggested items through the first round

1 Items for the Attraction to Public Participation Dimension

(1) In the public domain every person should be counted as equal

(2) I enjoy to think and argue in a political way

(3) I am satisfied when seeing my ideas being integrated in public policies

(4) I can easily relate to the way of thinking and arguing of politicians

(5) Working in the public sector I can better realize myself

2 Items for the Public Values Sub-Dimension

(1) Public service should give value for money

(2) Cost considerations should play an important role in decisions about public service provision

(3) It is fundamental that the interests of future generations are taken into account when developing public

policies

(4) To act with correct ethics is for public servants as important as their professional competence

(5) Primarily public servants should be accountable to the public and not to their superiors

3 Items for the Self-Sacrifice Dimension

(1) I would agree to a good plan to make a better life for the poor even if it costs me money

- 42 -

Appendix 3 An example of a summary of comments and suggestions (the second round)

On the item ―I am interested in developing public programs helping my country or community

A It is a good sentence but is it possible for everyone to ―develop public programs Even public officials are

limited to do this

B OK

C or the special group

D Double barrel and difficult to translate to my country ―I am interested in helping to improve public services

E Good but I feel that wording requires revision If we draw sample from local government then country is

redundant ―I am interested in developing public policy programs that aim in helping my community

F 4 (General appreciation from1 to 4) Not clear whether it refers to ones work or to ones concern as a citizen

―(In my work) I am interested in contributing to the development of public etchellip

G Instead of development I would use implementing

H Delete public which is redundant because of the characteristic helping my country or community but add

or implementing ―I am interested in developing or implementing programs helping my country or

community

I Referring to dimension Wording Similar to item 6

Note Names have been changed to alphabets in order to ensure the anonymity of participants

- 43 -

Appendix 4 Dimensions and Items of Public Service Motivation for Future Research

Dimensions and items

Attraction to Public Service (APS)

1 I am interested in helping to improve public service

2 I admire people who initiate or are involved in activities to aid my community

3 It is important to contribute to activities that tackle social problems

4 Meaningful public service is very important to me

5 It is important for me to contribute to the common good

Commitment to Public Values (CPV)

1 I think equal opportunities for citizens are very important

2 It is important that citizens can rely on the continuous provision of public services

3 It is fundamental that the interests of future generations are taken into account when

developing public policies

4 To act ethically is essential for public servants

5 I believe that public employees must always be aware of the legitimacy of their activities

6 I personally identify with the aim of protecting individual liberties and rights

Compassion (COM)

1 It is difficult for me to contain my feelings when I see people in distress

2 I feel sympathetic to the plight of the underprivileged

3 I empathize with other people who face difficulties

4 I get very upset when I see other people being treated unfairly

5 Considering the welfare of others is very important

Self-Sacrifice (SS)

1 I am prepared to make sacrifices for the good of society

2 I believe in putting civic duty before self

3 I am willing to risk personal loss to help society

4 I would agree to a good plan to make a better life for the poor even if it costs me money

- 38 -

Table 5

Results of CFA in the individual countries Standardized factor loadings and measures of fit

Factors

amp Items

Australia Belgium China Denmark France Italy

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

672

756

799

785

804

943

710

703

597

790

799

730

584

675

685

681

731

742

578

748

674

643

765

739

645

704

678

708

812

779

632

604

765

766

672

853

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

552

593

560

699

727

632

624

605

560

713

709

645

702

602

616

724

686

735

567

560

715

654

605

811

608

658

587

696

755

698

673

620

523

782

697

654

COM

COM1

COM2

COM3

COM5

COM6

623

584

662

688

701

739

732

608

475

808

666

581

578

616

605

649

670

673

699

751

737

679

816

776

568

546

801

626

673

850

637

546

741

715

601

822

SS

SS2

SS3

SS4

SS7

654

794

806

892

767

813

727

693

799

793

725

838

857

820

773

896

855

842

806

749

792

829

753

884

847

731

753

889

816

753

815

692

664

795

782

734

Inter-factor

correlations

093

~

752

163

~

810

320

~

773

323

~

825

452

~

827

497

~

855

323

~

850

321

~

847

498

~

899

536

~

902

425

~

829

476

~

865

Sample size 249 214 230 249 266 162

Measures of Fit

χsup2 638 444 881 732 765 554 817 625 735 565 750 564

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 980 100 939 987 969 997 967 994 975 997 949 995

GFI 961 986 941 986 948 991 949 992 955 992 926 986

RMSEA 036 00 063 050 051 026 053 035 045 026 059 033

- 39 -

Table 5 (Continued)

Factors

amp Items

Korea Lithuania The Netherlands Switzerland UK USA

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

619

701

667

706

781

717

666

512

682

751

605

779

557

629

630

620

719

729

535

621

702

643

694

735

655

725

738

710

804

822

535

621

702

643

694

735

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

594

607

722

722

644

722

674

535

533

790

589

711

568

554

538

655

628

545

627

580

615

711

666

704

800

512

555

873

581

646

627

580

615

711

666

704

COM

COM1

COM2

COM3

COM5

COM6

694

677

550

763

736

627

557

639

428

550

696

534

737

623

508

766

616

608

659

592

683

703

661

769

682

583

729

739

644

803

659

592

683

703

661

769

SS

SS2

SS3

SS4

SS7

776

752

679

779

770

811

707

721

635

725

734

706

814

656

743

820

767

764

745

632

689

788

662

744

671

818

832

830

864

802

745

632

689

788

662

744

Inter-factor

correlations

432

~

726

466

~

746

376

~

788

406

~

902

377

~

852

455

~

836

637

~

818

660

~

828

317

~

852

360

~

852

637

~

818

660

~

828

Sample size 253 236 249 250 260 250

Measures of Fit

χsup2 896 573 612 462 773 616 815 715 769 542 815 715

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 951 996 975 100 958 993 957 991 972 998 957 991

GFI 942 988 960 989 951 987 948 990 955 990 948 990

RMSEA 059 028 034 00 050 034 053 044 048 022 053 044

Note All factor loadings and correlations are significant at p lt 001

- 40 -

Appendix 1 Dimensions and possible items of PSM (the first round)

1 Attraction to Public Participation

(1) I am interested in developing public programs helping my country or community (Kim 2009a)

(2) Sharing my views on public policies with others is attractive to me (Kim 2009a)

(3) I am satisfied when seeing people get benefits from the public programs I was involved in (Kim 2009a)

(4) I like to discuss political subjects with others (Giauque et al 2011)

(5) I think the governmental activities contribute to our welfare

(6) I like to initiate actions to help out my community

(7) I do not care much about politicians (Perry 1996)

2 Commitment to Public Values

Possible items for public interest

(1A) Meaningful public service is very important to me (Perry 1996)

(2A) It is important for me to contribute for the common good (Giauque et al 2011)

(3A) I recognize myself with the mission of protection of liberties and rights endorsed by the nation-state

(Giauque et al 2011)

(4A) I would prefer seeing public officials do what is best for the whole community even if it harmed my

interests (Perry 1996)

(5A) To me serving the public interest is more important than helping other people (Vandenabeele 2008b)

Possible items for public values

(1B) I recognize myself with the promotion of the equality of chances endorsed by the nation-state (Giauque et

al 2011)

(2B) It is fundamental that the public service provided to the citizen is regular and continuous (Giauque et al

2011)

(3B) It is fundamental that public service innovates to respond to the new needs of the citizensclients (Giauque

et al 2011)

(4B) The dignity and well-being of all should be the most important concerns in any society (Steenbergen

1996)

(5B) It is important that public servants account for all the costsexpenses they make (Vandenabeele 2008b)

(6B) We have to do everything in our power to pursue the goal of democracy

3 Compassion

(1) It is difficult for me to contain my feelings when I see people in distress (Perry 1996)

(2) I feel sympathetic to the plight of the underprivileged (Kim 2009a)

(3) I am often reminded by daily events how dependent we are on one another (Perry 1996)

(4) I feel compassion for others in the difficulties they are facing (Steffen and Masters 2005)

(5) It is not really my problem if others are in trouble and need help (Beutel and Marini 1995)

(6) I find it hard to be sympathetic toward starving people in foreign lands when there is so much trouble in our

own country (Beutel and Marini 1995)

(7) I get very upset when I see other people treated unfairly (Beutel and Marini 1995)

(8) I would agree to a good plan to make a better life for the poor even if it cost me money (Beutel and Marini

1995)

(9) To me patriotism includes seeing the welfare of others (Perry 1996)

(10) I care very much about other people

4 Self-Sacrifice

(1) Making a difference in society means more to me than personal achievements (Perry 1996)

(2) I am prepared to make enormous sacrifices for the good of society (Perry 1996)

(3) I believe in putting duty before self (Perry 1996)

(4) I am one of those rare people who would risk personal loss to help someone else (Perry 1996)

(5) I think people should give back to society more than they get from it (Perry 1996)

(6) Serving other citizens would give me a good feeling even if no one paid me for it (Perry 1996)

(7) Much of what I do is for a cause bigger than myself (Perry 1996)

- 41 -

Appendix 2 Newly suggested items through the first round

1 Items for the Attraction to Public Participation Dimension

(1) In the public domain every person should be counted as equal

(2) I enjoy to think and argue in a political way

(3) I am satisfied when seeing my ideas being integrated in public policies

(4) I can easily relate to the way of thinking and arguing of politicians

(5) Working in the public sector I can better realize myself

2 Items for the Public Values Sub-Dimension

(1) Public service should give value for money

(2) Cost considerations should play an important role in decisions about public service provision

(3) It is fundamental that the interests of future generations are taken into account when developing public

policies

(4) To act with correct ethics is for public servants as important as their professional competence

(5) Primarily public servants should be accountable to the public and not to their superiors

3 Items for the Self-Sacrifice Dimension

(1) I would agree to a good plan to make a better life for the poor even if it costs me money

- 42 -

Appendix 3 An example of a summary of comments and suggestions (the second round)

On the item ―I am interested in developing public programs helping my country or community

A It is a good sentence but is it possible for everyone to ―develop public programs Even public officials are

limited to do this

B OK

C or the special group

D Double barrel and difficult to translate to my country ―I am interested in helping to improve public services

E Good but I feel that wording requires revision If we draw sample from local government then country is

redundant ―I am interested in developing public policy programs that aim in helping my community

F 4 (General appreciation from1 to 4) Not clear whether it refers to ones work or to ones concern as a citizen

―(In my work) I am interested in contributing to the development of public etchellip

G Instead of development I would use implementing

H Delete public which is redundant because of the characteristic helping my country or community but add

or implementing ―I am interested in developing or implementing programs helping my country or

community

I Referring to dimension Wording Similar to item 6

Note Names have been changed to alphabets in order to ensure the anonymity of participants

- 43 -

Appendix 4 Dimensions and Items of Public Service Motivation for Future Research

Dimensions and items

Attraction to Public Service (APS)

1 I am interested in helping to improve public service

2 I admire people who initiate or are involved in activities to aid my community

3 It is important to contribute to activities that tackle social problems

4 Meaningful public service is very important to me

5 It is important for me to contribute to the common good

Commitment to Public Values (CPV)

1 I think equal opportunities for citizens are very important

2 It is important that citizens can rely on the continuous provision of public services

3 It is fundamental that the interests of future generations are taken into account when

developing public policies

4 To act ethically is essential for public servants

5 I believe that public employees must always be aware of the legitimacy of their activities

6 I personally identify with the aim of protecting individual liberties and rights

Compassion (COM)

1 It is difficult for me to contain my feelings when I see people in distress

2 I feel sympathetic to the plight of the underprivileged

3 I empathize with other people who face difficulties

4 I get very upset when I see other people being treated unfairly

5 Considering the welfare of others is very important

Self-Sacrifice (SS)

1 I am prepared to make sacrifices for the good of society

2 I believe in putting civic duty before self

3 I am willing to risk personal loss to help society

4 I would agree to a good plan to make a better life for the poor even if it costs me money

- 39 -

Table 5 (Continued)

Factors

amp Items

Korea Lithuania The Netherlands Switzerland UK USA

MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS MLE D-WLS

APS

APP1

APP5

APP7

CPI1

CPI2

619

701

667

706

781

717

666

512

682

751

605

779

557

629

630

620

719

729

535

621

702

643

694

735

655

725

738

710

804

822

535

621

702

643

694

735

CPV

CPV1

CPV2

CPV6

CPV7

CPV8

CPV9

594

607

722

722

644

722

674

535

533

790

589

711

568

554

538

655

628

545

627

580

615

711

666

704

800

512

555

873

581

646

627

580

615

711

666

704

COM

COM1

COM2

COM3

COM5

COM6

694

677

550

763

736

627

557

639

428

550

696

534

737

623

508

766

616

608

659

592

683

703

661

769

682

583

729

739

644

803

659

592

683

703

661

769

SS

SS2

SS3

SS4

SS7

776

752

679

779

770

811

707

721

635

725

734

706

814

656

743

820

767

764

745

632

689

788

662

744

671

818

832

830

864

802

745

632

689

788

662

744

Inter-factor

correlations

432

~

726

466

~

746

376

~

788

406

~

902

377

~

852

455

~

836

637

~

818

660

~

828

317

~

852

360

~

852

637

~

818

660

~

828

Sample size 253 236 249 250 260 250

Measures of Fit

χsup2 896 573 612 462 773 616 815 715 769 542 815 715

df 48 48 48 48 48 48 48 48 48 48 48 48

CFI 951 996 975 100 958 993 957 991 972 998 957 991

GFI 942 988 960 989 951 987 948 990 955 990 948 990

RMSEA 059 028 034 00 050 034 053 044 048 022 053 044

Note All factor loadings and correlations are significant at p lt 001

- 40 -

Appendix 1 Dimensions and possible items of PSM (the first round)

1 Attraction to Public Participation

(1) I am interested in developing public programs helping my country or community (Kim 2009a)

(2) Sharing my views on public policies with others is attractive to me (Kim 2009a)

(3) I am satisfied when seeing people get benefits from the public programs I was involved in (Kim 2009a)

(4) I like to discuss political subjects with others (Giauque et al 2011)

(5) I think the governmental activities contribute to our welfare

(6) I like to initiate actions to help out my community

(7) I do not care much about politicians (Perry 1996)

2 Commitment to Public Values

Possible items for public interest

(1A) Meaningful public service is very important to me (Perry 1996)

(2A) It is important for me to contribute for the common good (Giauque et al 2011)

(3A) I recognize myself with the mission of protection of liberties and rights endorsed by the nation-state

(Giauque et al 2011)

(4A) I would prefer seeing public officials do what is best for the whole community even if it harmed my

interests (Perry 1996)

(5A) To me serving the public interest is more important than helping other people (Vandenabeele 2008b)

Possible items for public values

(1B) I recognize myself with the promotion of the equality of chances endorsed by the nation-state (Giauque et

al 2011)

(2B) It is fundamental that the public service provided to the citizen is regular and continuous (Giauque et al

2011)

(3B) It is fundamental that public service innovates to respond to the new needs of the citizensclients (Giauque

et al 2011)

(4B) The dignity and well-being of all should be the most important concerns in any society (Steenbergen

1996)

(5B) It is important that public servants account for all the costsexpenses they make (Vandenabeele 2008b)

(6B) We have to do everything in our power to pursue the goal of democracy

3 Compassion

(1) It is difficult for me to contain my feelings when I see people in distress (Perry 1996)

(2) I feel sympathetic to the plight of the underprivileged (Kim 2009a)

(3) I am often reminded by daily events how dependent we are on one another (Perry 1996)

(4) I feel compassion for others in the difficulties they are facing (Steffen and Masters 2005)

(5) It is not really my problem if others are in trouble and need help (Beutel and Marini 1995)

(6) I find it hard to be sympathetic toward starving people in foreign lands when there is so much trouble in our

own country (Beutel and Marini 1995)

(7) I get very upset when I see other people treated unfairly (Beutel and Marini 1995)

(8) I would agree to a good plan to make a better life for the poor even if it cost me money (Beutel and Marini

1995)

(9) To me patriotism includes seeing the welfare of others (Perry 1996)

(10) I care very much about other people

4 Self-Sacrifice

(1) Making a difference in society means more to me than personal achievements (Perry 1996)

(2) I am prepared to make enormous sacrifices for the good of society (Perry 1996)

(3) I believe in putting duty before self (Perry 1996)

(4) I am one of those rare people who would risk personal loss to help someone else (Perry 1996)

(5) I think people should give back to society more than they get from it (Perry 1996)

(6) Serving other citizens would give me a good feeling even if no one paid me for it (Perry 1996)

(7) Much of what I do is for a cause bigger than myself (Perry 1996)

- 41 -

Appendix 2 Newly suggested items through the first round

1 Items for the Attraction to Public Participation Dimension

(1) In the public domain every person should be counted as equal

(2) I enjoy to think and argue in a political way

(3) I am satisfied when seeing my ideas being integrated in public policies

(4) I can easily relate to the way of thinking and arguing of politicians

(5) Working in the public sector I can better realize myself

2 Items for the Public Values Sub-Dimension

(1) Public service should give value for money

(2) Cost considerations should play an important role in decisions about public service provision

(3) It is fundamental that the interests of future generations are taken into account when developing public

policies

(4) To act with correct ethics is for public servants as important as their professional competence

(5) Primarily public servants should be accountable to the public and not to their superiors

3 Items for the Self-Sacrifice Dimension

(1) I would agree to a good plan to make a better life for the poor even if it costs me money

- 42 -

Appendix 3 An example of a summary of comments and suggestions (the second round)

On the item ―I am interested in developing public programs helping my country or community

A It is a good sentence but is it possible for everyone to ―develop public programs Even public officials are

limited to do this

B OK

C or the special group

D Double barrel and difficult to translate to my country ―I am interested in helping to improve public services

E Good but I feel that wording requires revision If we draw sample from local government then country is

redundant ―I am interested in developing public policy programs that aim in helping my community

F 4 (General appreciation from1 to 4) Not clear whether it refers to ones work or to ones concern as a citizen

―(In my work) I am interested in contributing to the development of public etchellip

G Instead of development I would use implementing

H Delete public which is redundant because of the characteristic helping my country or community but add

or implementing ―I am interested in developing or implementing programs helping my country or

community

I Referring to dimension Wording Similar to item 6

Note Names have been changed to alphabets in order to ensure the anonymity of participants

- 43 -

Appendix 4 Dimensions and Items of Public Service Motivation for Future Research

Dimensions and items

Attraction to Public Service (APS)

1 I am interested in helping to improve public service

2 I admire people who initiate or are involved in activities to aid my community

3 It is important to contribute to activities that tackle social problems

4 Meaningful public service is very important to me

5 It is important for me to contribute to the common good

Commitment to Public Values (CPV)

1 I think equal opportunities for citizens are very important

2 It is important that citizens can rely on the continuous provision of public services

3 It is fundamental that the interests of future generations are taken into account when

developing public policies

4 To act ethically is essential for public servants

5 I believe that public employees must always be aware of the legitimacy of their activities

6 I personally identify with the aim of protecting individual liberties and rights

Compassion (COM)

1 It is difficult for me to contain my feelings when I see people in distress

2 I feel sympathetic to the plight of the underprivileged

3 I empathize with other people who face difficulties

4 I get very upset when I see other people being treated unfairly

5 Considering the welfare of others is very important

Self-Sacrifice (SS)

1 I am prepared to make sacrifices for the good of society

2 I believe in putting civic duty before self

3 I am willing to risk personal loss to help society

4 I would agree to a good plan to make a better life for the poor even if it costs me money

- 40 -

Appendix 1 Dimensions and possible items of PSM (the first round)

1 Attraction to Public Participation

(1) I am interested in developing public programs helping my country or community (Kim 2009a)

(2) Sharing my views on public policies with others is attractive to me (Kim 2009a)

(3) I am satisfied when seeing people get benefits from the public programs I was involved in (Kim 2009a)

(4) I like to discuss political subjects with others (Giauque et al 2011)

(5) I think the governmental activities contribute to our welfare

(6) I like to initiate actions to help out my community

(7) I do not care much about politicians (Perry 1996)

2 Commitment to Public Values

Possible items for public interest

(1A) Meaningful public service is very important to me (Perry 1996)

(2A) It is important for me to contribute for the common good (Giauque et al 2011)

(3A) I recognize myself with the mission of protection of liberties and rights endorsed by the nation-state

(Giauque et al 2011)

(4A) I would prefer seeing public officials do what is best for the whole community even if it harmed my

interests (Perry 1996)

(5A) To me serving the public interest is more important than helping other people (Vandenabeele 2008b)

Possible items for public values

(1B) I recognize myself with the promotion of the equality of chances endorsed by the nation-state (Giauque et

al 2011)

(2B) It is fundamental that the public service provided to the citizen is regular and continuous (Giauque et al

2011)

(3B) It is fundamental that public service innovates to respond to the new needs of the citizensclients (Giauque

et al 2011)

(4B) The dignity and well-being of all should be the most important concerns in any society (Steenbergen

1996)

(5B) It is important that public servants account for all the costsexpenses they make (Vandenabeele 2008b)

(6B) We have to do everything in our power to pursue the goal of democracy

3 Compassion

(1) It is difficult for me to contain my feelings when I see people in distress (Perry 1996)

(2) I feel sympathetic to the plight of the underprivileged (Kim 2009a)

(3) I am often reminded by daily events how dependent we are on one another (Perry 1996)

(4) I feel compassion for others in the difficulties they are facing (Steffen and Masters 2005)

(5) It is not really my problem if others are in trouble and need help (Beutel and Marini 1995)

(6) I find it hard to be sympathetic toward starving people in foreign lands when there is so much trouble in our

own country (Beutel and Marini 1995)

(7) I get very upset when I see other people treated unfairly (Beutel and Marini 1995)

(8) I would agree to a good plan to make a better life for the poor even if it cost me money (Beutel and Marini

1995)

(9) To me patriotism includes seeing the welfare of others (Perry 1996)

(10) I care very much about other people

4 Self-Sacrifice

(1) Making a difference in society means more to me than personal achievements (Perry 1996)

(2) I am prepared to make enormous sacrifices for the good of society (Perry 1996)

(3) I believe in putting duty before self (Perry 1996)

(4) I am one of those rare people who would risk personal loss to help someone else (Perry 1996)

(5) I think people should give back to society more than they get from it (Perry 1996)

(6) Serving other citizens would give me a good feeling even if no one paid me for it (Perry 1996)

(7) Much of what I do is for a cause bigger than myself (Perry 1996)

- 41 -

Appendix 2 Newly suggested items through the first round

1 Items for the Attraction to Public Participation Dimension

(1) In the public domain every person should be counted as equal

(2) I enjoy to think and argue in a political way

(3) I am satisfied when seeing my ideas being integrated in public policies

(4) I can easily relate to the way of thinking and arguing of politicians

(5) Working in the public sector I can better realize myself

2 Items for the Public Values Sub-Dimension

(1) Public service should give value for money

(2) Cost considerations should play an important role in decisions about public service provision

(3) It is fundamental that the interests of future generations are taken into account when developing public

policies

(4) To act with correct ethics is for public servants as important as their professional competence

(5) Primarily public servants should be accountable to the public and not to their superiors

3 Items for the Self-Sacrifice Dimension

(1) I would agree to a good plan to make a better life for the poor even if it costs me money

- 42 -

Appendix 3 An example of a summary of comments and suggestions (the second round)

On the item ―I am interested in developing public programs helping my country or community

A It is a good sentence but is it possible for everyone to ―develop public programs Even public officials are

limited to do this

B OK

C or the special group

D Double barrel and difficult to translate to my country ―I am interested in helping to improve public services

E Good but I feel that wording requires revision If we draw sample from local government then country is

redundant ―I am interested in developing public policy programs that aim in helping my community

F 4 (General appreciation from1 to 4) Not clear whether it refers to ones work or to ones concern as a citizen

―(In my work) I am interested in contributing to the development of public etchellip

G Instead of development I would use implementing

H Delete public which is redundant because of the characteristic helping my country or community but add

or implementing ―I am interested in developing or implementing programs helping my country or

community

I Referring to dimension Wording Similar to item 6

Note Names have been changed to alphabets in order to ensure the anonymity of participants

- 43 -

Appendix 4 Dimensions and Items of Public Service Motivation for Future Research

Dimensions and items

Attraction to Public Service (APS)

1 I am interested in helping to improve public service

2 I admire people who initiate or are involved in activities to aid my community

3 It is important to contribute to activities that tackle social problems

4 Meaningful public service is very important to me

5 It is important for me to contribute to the common good

Commitment to Public Values (CPV)

1 I think equal opportunities for citizens are very important

2 It is important that citizens can rely on the continuous provision of public services

3 It is fundamental that the interests of future generations are taken into account when

developing public policies

4 To act ethically is essential for public servants

5 I believe that public employees must always be aware of the legitimacy of their activities

6 I personally identify with the aim of protecting individual liberties and rights

Compassion (COM)

1 It is difficult for me to contain my feelings when I see people in distress

2 I feel sympathetic to the plight of the underprivileged

3 I empathize with other people who face difficulties

4 I get very upset when I see other people being treated unfairly

5 Considering the welfare of others is very important

Self-Sacrifice (SS)

1 I am prepared to make sacrifices for the good of society

2 I believe in putting civic duty before self

3 I am willing to risk personal loss to help society

4 I would agree to a good plan to make a better life for the poor even if it costs me money

- 41 -

Appendix 2 Newly suggested items through the first round

1 Items for the Attraction to Public Participation Dimension

(1) In the public domain every person should be counted as equal

(2) I enjoy to think and argue in a political way

(3) I am satisfied when seeing my ideas being integrated in public policies

(4) I can easily relate to the way of thinking and arguing of politicians

(5) Working in the public sector I can better realize myself

2 Items for the Public Values Sub-Dimension

(1) Public service should give value for money

(2) Cost considerations should play an important role in decisions about public service provision

(3) It is fundamental that the interests of future generations are taken into account when developing public

policies

(4) To act with correct ethics is for public servants as important as their professional competence

(5) Primarily public servants should be accountable to the public and not to their superiors

3 Items for the Self-Sacrifice Dimension

(1) I would agree to a good plan to make a better life for the poor even if it costs me money

- 42 -

Appendix 3 An example of a summary of comments and suggestions (the second round)

On the item ―I am interested in developing public programs helping my country or community

A It is a good sentence but is it possible for everyone to ―develop public programs Even public officials are

limited to do this

B OK

C or the special group

D Double barrel and difficult to translate to my country ―I am interested in helping to improve public services

E Good but I feel that wording requires revision If we draw sample from local government then country is

redundant ―I am interested in developing public policy programs that aim in helping my community

F 4 (General appreciation from1 to 4) Not clear whether it refers to ones work or to ones concern as a citizen

―(In my work) I am interested in contributing to the development of public etchellip

G Instead of development I would use implementing

H Delete public which is redundant because of the characteristic helping my country or community but add

or implementing ―I am interested in developing or implementing programs helping my country or

community

I Referring to dimension Wording Similar to item 6

Note Names have been changed to alphabets in order to ensure the anonymity of participants

- 43 -

Appendix 4 Dimensions and Items of Public Service Motivation for Future Research

Dimensions and items

Attraction to Public Service (APS)

1 I am interested in helping to improve public service

2 I admire people who initiate or are involved in activities to aid my community

3 It is important to contribute to activities that tackle social problems

4 Meaningful public service is very important to me

5 It is important for me to contribute to the common good

Commitment to Public Values (CPV)

1 I think equal opportunities for citizens are very important

2 It is important that citizens can rely on the continuous provision of public services

3 It is fundamental that the interests of future generations are taken into account when

developing public policies

4 To act ethically is essential for public servants

5 I believe that public employees must always be aware of the legitimacy of their activities

6 I personally identify with the aim of protecting individual liberties and rights

Compassion (COM)

1 It is difficult for me to contain my feelings when I see people in distress

2 I feel sympathetic to the plight of the underprivileged

3 I empathize with other people who face difficulties

4 I get very upset when I see other people being treated unfairly

5 Considering the welfare of others is very important

Self-Sacrifice (SS)

1 I am prepared to make sacrifices for the good of society

2 I believe in putting civic duty before self

3 I am willing to risk personal loss to help society

4 I would agree to a good plan to make a better life for the poor even if it costs me money

- 42 -

Appendix 3 An example of a summary of comments and suggestions (the second round)

On the item ―I am interested in developing public programs helping my country or community

A It is a good sentence but is it possible for everyone to ―develop public programs Even public officials are

limited to do this

B OK

C or the special group

D Double barrel and difficult to translate to my country ―I am interested in helping to improve public services

E Good but I feel that wording requires revision If we draw sample from local government then country is

redundant ―I am interested in developing public policy programs that aim in helping my community

F 4 (General appreciation from1 to 4) Not clear whether it refers to ones work or to ones concern as a citizen

―(In my work) I am interested in contributing to the development of public etchellip

G Instead of development I would use implementing

H Delete public which is redundant because of the characteristic helping my country or community but add

or implementing ―I am interested in developing or implementing programs helping my country or

community

I Referring to dimension Wording Similar to item 6

Note Names have been changed to alphabets in order to ensure the anonymity of participants

- 43 -

Appendix 4 Dimensions and Items of Public Service Motivation for Future Research

Dimensions and items

Attraction to Public Service (APS)

1 I am interested in helping to improve public service

2 I admire people who initiate or are involved in activities to aid my community

3 It is important to contribute to activities that tackle social problems

4 Meaningful public service is very important to me

5 It is important for me to contribute to the common good

Commitment to Public Values (CPV)

1 I think equal opportunities for citizens are very important

2 It is important that citizens can rely on the continuous provision of public services

3 It is fundamental that the interests of future generations are taken into account when

developing public policies

4 To act ethically is essential for public servants

5 I believe that public employees must always be aware of the legitimacy of their activities

6 I personally identify with the aim of protecting individual liberties and rights

Compassion (COM)

1 It is difficult for me to contain my feelings when I see people in distress

2 I feel sympathetic to the plight of the underprivileged

3 I empathize with other people who face difficulties

4 I get very upset when I see other people being treated unfairly

5 Considering the welfare of others is very important

Self-Sacrifice (SS)

1 I am prepared to make sacrifices for the good of society

2 I believe in putting civic duty before self

3 I am willing to risk personal loss to help society

4 I would agree to a good plan to make a better life for the poor even if it costs me money

- 43 -

Appendix 4 Dimensions and Items of Public Service Motivation for Future Research

Dimensions and items

Attraction to Public Service (APS)

1 I am interested in helping to improve public service

2 I admire people who initiate or are involved in activities to aid my community

3 It is important to contribute to activities that tackle social problems

4 Meaningful public service is very important to me

5 It is important for me to contribute to the common good

Commitment to Public Values (CPV)

1 I think equal opportunities for citizens are very important

2 It is important that citizens can rely on the continuous provision of public services

3 It is fundamental that the interests of future generations are taken into account when

developing public policies

4 To act ethically is essential for public servants

5 I believe that public employees must always be aware of the legitimacy of their activities

6 I personally identify with the aim of protecting individual liberties and rights

Compassion (COM)

1 It is difficult for me to contain my feelings when I see people in distress

2 I feel sympathetic to the plight of the underprivileged

3 I empathize with other people who face difficulties

4 I get very upset when I see other people being treated unfairly

5 Considering the welfare of others is very important

Self-Sacrifice (SS)

1 I am prepared to make sacrifices for the good of society

2 I believe in putting civic duty before self

3 I am willing to risk personal loss to help society

4 I would agree to a good plan to make a better life for the poor even if it costs me money