An Exploratory Study of the Impact of e-Service Process on Online Customer Satisfaction

Sulin Ba Department of Operations and Information Management

School of Business University of Connecticut

Storrs, CT 06269 Tel: 860-486-6311 Fax: 860-486-4839

Sulin.ba@uconn.edu

Wayne C. Johansson U.S. Department of Homeland Security

Los Angeles International Airport Los Angeles, CA 90045

Tel: 323-436-2772 wcjla@sbcglobal.net

Forthcoming in Production and Operations Management

An Exploratory Study of the Impact of e-Service Process on Online Customer Satisfaction

Abstract

Although extensive academic research has examined the dynamics of interpersonal

interactions between service providers and customers, much less research has investigated

customer service encounters through technological interfaces such as the Web in electronic

commerce transactions. Corporate websites have become an important point of contact with

customers for many companies. Service has been described as one of the most important

attributes for online business to influence traffic and sales. However, more research is needed to

understand how web-based technological capabilities of the services affect customer evaluations

of service value and how to determine the technological capabilities embedded in the e-service

for customer satisfaction. In this paper, we propose viewing the interface between online buyers

and sellers through the lens of service management in order to identify and explain possible

determinants of online customer satisfaction. A company’s website is considered its electronic

Service Delivery System (eSDS). We look at this eSDS from its process point of view and

examine how an eSDS affects customer satisfaction. Our findings indicate that as the eSDS

process improves, a customer’s perception of the website’s ease of use increases, leading to

increased service value and perceived control over the process, which increases customer

satisfaction. The research provides evidence that the technological capabilities embedded in the

website processes are an important factor in determining service quality and ultimately online

customer satisfaction.

Keywords: Web-based Technological Capabilities, Technology Design of e-Service

Process, Online Customer Satisfaction, Electronic Service Delivery System (eSDS)

1

1. Introduction

Technology advancement is revolutionizing the way business is conducted and reshaping

how companies interact with their customers. This phenomenon is particularly evident in the

domain of electronic commerce (EC). Companies have realized electronic commerce not only is

a way of reducing costs through automation and increased efficiency, but, more importantly, also

is a means to expand revenues through enhanced customer service. Corporate websites provide

an important interface through which customers and firms interact with each other. This

interface has several characteristics that are uncommon to the traditional forms of buyer/seller

interaction (e.g., face-to-face or telephone). With little or no human intervention, the capabilities

embedded in the website process technology enable a consumer to locate a product or service,

assess its utility, and purchase it practically whenever and wherever it is convenient. Indeed, the

Internet technology has dramatically impacted the service creation process.

Yet, surveys of online customers consistently indicate that a big percentage is not

satisfied with the interaction (ICSA 2001, Bednarz 2003). As pointed out by Meuter et al.

(2000), although extensive academic research has examined the dynamics of interpersonal

interactions between service providers and customers, much less research has investigated

customer service encounters through technological interfaces. There has been research on

website design to improve customer satisfaction. Most studies, however, focus on website

navigation, information content, download speed, information presentation, etc. (e.g., McKinney

et al. 2002, Palmer 2002). More research is needed to better understand how services delivered

through technological interfaces such as the Web affect customer evaluations of service value

and how to manage the technology-based service process for customer satisfaction and,

ultimately, for creating strategic advantage.

2

Service has been described as one of the most important attributes for online business to

influence traffic and sales (Lohse and Spiller 1998). Provision of service over electronic

networks is referred to as e-service (Rust and Kannan 2003). Scholars have argued that e-service,

compared with offline service, has the ability to serve consumers more efficiently, at a lower

marginal cost, while simultaneously offering real-time product and/or service-specific

information (Shapiro and Varian 1999).

However, although Internet-based service tools and technologies offer many benefits,

there are inevitably costs associated with developing and delivering e-services. A recent IDC

survey found that in 2002, e-service (such as online order taking and order tracking, payment,

and after-sales support) provision absorbed about 50% of the total investment in new information

technologies at a typical company (Tsikriktsis et al. 2004). E-service entails many different

dimensions and attributes, such as responsiveness of answering customer inquiries, website

security, customization, interactivity, service delivery processes, etc. The rapid increase in e-

service activity creates a challenge for firms: what combination of features should be embedded

in the service technology to satisfy consumers while realistically considering operational and

financial constraints?

The ideal action for Internet companies is to improve and maintain all service quality

attributes that satisfy their customers’ needs and wants. However, given that firms, even large

ones, have limited resources, priorities must be set among alternative technological capabilities

embedded in e-service in making investment decisions based on a company’s business strategies.

Not all technological capabilities have the same effect on customer satisfaction. The key is to

find, among various capabilities, which ones are more crucial to enhancing the level of service

quality. In other words, to be successful, e-services need to identify and focus on developing

3

technology-based features that enhance consumer value. In this manner, firms can understand

what service areas should be emphasized to most effectively improve quality while avoiding

investing valuable resources in technology features that may not pay off.

The justification and deployment of new technology – in the case of this study, the e-

service process – to a new market warrants special attention (Betz 2001). We build on theories

from service management and propose viewing the interface between buyers and sellers (i.e., the

website embedding e-service technology capabilities) through the lens of service management in

order to identify and explain possible determinants of online customer satisfaction. In particular,

the service management literature has identified service process as a critical factor for

influencing service quality. We evaluate the significance of the e-service process in terms of

causality of the level of customer satisfaction: How does a customer’s e-service process

perception about a website affect customer satisfaction?

This research strives to make an important contribution to the management of technology

domain by examining the impact of e-service capabilities on customer satisfaction in the Internet

market. Our insights will help firms to understand whether investing in and managing the e-

service is justified and what factors lead to customer acceptance of the e-service process.

2. Background and Theory

The Service-Profit Chain model (Heskett et al. 1994) hypothesizes relationships between

the Service Delivery System (SDS) (internal service quality, employee satisfaction, retention,

and productivity), customer satisfaction, and profitability. In summary, profit and growth result

from customer loyalty, which develops from customer satisfaction. Customer satisfaction, on the

other hand, is influenced by the service value a customer receives from the service delivery

4

system. This model provides an integrative framework for understanding how a firm’s

investments into service operations are related to customer perceptions of the value they receive

from the firm. The framework has been widely used by practitioners. Academic researchers

have also empirically tested the various links suggested by the model and found support for the

positive effects of the SDS’ performance perceptions on service quality perceptions and

customer behaviors (Kamakura et al. 2002).

With recent technological advances and the explosion of Internet usage, many services

are delivered through a company’s website and customers no longer interact face-to-face with

the service provider. A company’s website thus becomes the service delivery system, which is

critical for a company’s value creation strategy. The focus of this study is to examine how this

new, electronic service delivery system (eSDS) affects customers’ perceived service value and

customer satisfaction.

Roth and Jackson (1995) evaluated the service delivery system in the operations

capabilities-service quality-performance (C-SQ-P) model based on the process and people

capabilities of the SDS: what the system can do and what the outcomes of the service interaction

are, because these two factors are what customers tend to use to make their judgment of the

service system. In their survey of the retail banking industry, Roth and Jackson provided

evidence that the processes of a service delivery system had a greater impact upon service

quality than people capabilities – the knowledge and skills possessed by employees interacting

with customers. This finding leads us to think that it is possible that technology and the business

processes embedded in the technology have a greater significance than human interaction on

customers’ perceived service value, especially in the online environment. A study by Meuter et

al. (2000) indicates that process failure and poor process design are among the major factors

5

leading to a customer’s dissatisfactory evaluation of service in a technology-based service

encounter. In the Internet environment, the technology typically is the website which is also the

service delivery system. Therefore we focus on the eSDS from the service process point of view.

Specifically, service process is conceptualized as a configuration of technological capabilities

through which service providers respond to customer needs; and perceived eSDS process refers

to the customer’s view of how service processes are delivered by the website’s technological

capabilities. Combining the C-SQ-P model with the Service-Profit Chain model and adapting

them to the EC service context, we conjecture the following:

H1a: Service value provided to customers is positively correlated to customers’

perceived eSDS process.

H1b: Customer satisfaction is positively correlated to the service value provided to

customers.

H2: Customer satisfaction is positively correlated to perceived eSDS process.

One important factor that has been identified by marketing researchers to have a

considerable impact in a service process is perceived control (Hui and Bateson 1991), which is

described as the amount of control that a customer feels he has over the process or outcome. Hui

and Bateson (1991) concluded that perceptions of control affect customer satisfaction ratings in a

variety of service situations. The sense of control is especially important to customers in a self-

service setting and could increase the evaluation of the experience (Langeard et al. 1981).

In an EC context, when a customer searches through a company’s website for a particular

product or checks inventory availability, the customer is in fact performing a self-service. More

recent studies on Internet retailing suggest that perceived control encourages Internet usage and

loyalty, resulting in more satisfied customers (Lee and Allaway 2002). Indeed, new advances in

6

technology have made it possible for customers to choose the channel through which to acquire

the product, the channel through which the product will be delivered, and the extent to which

they would like to be involved in the development or delivery of the product. Therefore, their

expectations of perceived control have escalated. Rust and Kannan (2003) believe that

appropriately designed and implemented service technologies can provide customers with more

control in their process of conducting transactions which can increase customer satisfaction. In

this research, we define perceived control as the amount of control the customer has in the

technology-based e-service encounter, such as navigating through the website and determining

the e-service outcome. We hypothesize that:

H3a: The customer’s perceived control in the online service process is positively

correlated to the perceived eSDS process.

H3b: Customer satisfaction is positively correlated to the customer’s perceived control

in the online service process.

Although it has been argued that the capabilities embedded in an e-service technology

provide many potential benefits for customers, if customers think the technology is too difficult

to use, customers may not use the e-service technology at all. Therefore, customer acceptance of

this new technology is critical for firms trying to push more service to the customer side to lower

their service cost and improve their service efficiency. The Technology Acceptance Model

(TAM), which has been widely used to study user acceptance of new technology, argues that

perceived ease of use is one of the key predictors of user acceptance of new technology (Davis

1989). Perceived ease of use is directly related to computer-mediated services and refers to the

extent to which a person believes using the technology will be free of effort. In an e-commerce

7

setting, ease of use has been confirmed as a key factor leading to channel satisfaction (Devaraj el

al. 2002).

Ease of use, however, is dictated by what the system can do and what it allows its

customers to do, i.e., the capabilities embedded in the e-service technology. Usability studies on

online stores have looked at website architecture, design, and various navigation processes to

predict how easy it is for users to achieve what they want to do (Lohse and Spiller 1998, Palmer

2002). A recent study by Chen et al. (2004) indicates that poorly designed website processes

have an adverse influence on the website’s perceived ease of use. Therefore, we hypothesize the

following:

H4a: The customer’s perceived ease of use of a website is positively correlated to the

perceived eSDS process.

H4b: Customer satisfaction is positively correlated to the customer’s perceived ease of

use of the website.

Other than the process capabilities of the SDS, the service literature also looks at how a

service is delivered. Bitner et al. (1994) and Mittal and Lassar (1996) both point out that in

service settings, customer satisfaction and evaluation of service are often influenced by the

quality of the interpersonal interaction between the customer and the service provider. In the EC

context, services are provided through the technology medium and therefore are mostly

impersonal. The level of interaction between the service provider and the customer or among the

customers is normally determined by the degree of interaction embedded in the service

provider’s website – a key technological capability. For example, does the website enable two-

way synchronous or asynchronous exchanges between the customer and the service provider?

Does the website provide a telephone number in case the customer needs to contact the service

8

provider? This kind of reciprocal communication-based interaction is termed interactivity in the

literature and considered an important influence in building up online relationships and total

shopping experiences (Ha and James 1998, Merrilees 2002).

In addition to the social aspect of customer interaction with the service provider, many

websites also offer technological tools that allow customers to receive information tailored to

their specific needs. For example, a customer shopping for a sports utility vehicle can choose to

only receive information comparing different SUV models. This type of interactivity – the

technological capability to create a customized product – is mainly process or system oriented

(see, e.g., McKinney et al. 2002, Palmer 2002), instead of social content oriented, and therefore

is captured in our eSDS process construct. Interactivity in our research focuses on the

interpersonal communication aspect of the construct. Based on the above discussion, we

hypothesize the following:

H5a: Interactivity moderates the relationship between the perceived eSDS process and

service value.

H5b: Interactivity moderates the relationship between service value and customer

satisfaction.

H5c: Interactivity moderates the relationship between the perceived eSDS process and

the perceived ease of use.

H5d: Interactivity moderates the relationship between the customer’s perceived ease of

use and customer satisfaction.

H5e: Interactivity moderates the relationship between the perceived eSDS process and

the customer’s perceived control.

9

H5f: Interactivity moderates the relationship between the customer’s perceived control

and customer satisfaction.

Figure 1 summarizes our research model.

Figure 1. Research Model and Hypotheses

Customer Satisfaction

User’s

of control

Interactivity

Customer Satisfaction

Perceived Control

Interactivity

H1b(+)H1a(+

)

H4b(+)H4a(+)

H3b(+)H3a(+)

H5bH5a

H5dH5c

H5fH5e

Perceived eSDSProcess

PerceivedEase of Use

ServiceValue

H2Customer

Satisfaction

User’s

of control

Interactivity

Customer Satisfaction

Perceived Control

Interactivity

H1b(+)H1a(+

)

H4b(+)H4a(+)

H3b(+)H3a(+)

H5bH5a

H5dH5c

H5fH5e

Perceived eSDSProcess

PerceivedEase of Use

ServiceValue

H2

Three demographic variables, namely gender, Internet experience, and online shopping

experience, are important for control purposes. Prior research has indicated that both gender and

relevant prior experience play a role in how users perceive a technology (e.g., Venkatesh and

Morris 2000). To the extent that perceived ease of use and perceived control may be related to

these demographics, it is necessary to control for them in assessing the true relationship between

these constructs and customer satisfaction. Therefore, these three demographic variables were

included as control variables.

10

3. Methodology

3.1. Research Design

The empirical study was conducted in two phases: a pilot study (n=100, with a response

rate of 79.9%) and the final study. Relying upon extant service instruments and concepts, we

first developed an instrument to measure our constructs. Through pilot testing, we substantiated

the reliability of the construct indices and eliminated redundant questions. The final study

instrument is a much shorter version.

Common method variance can be a potential source of bias in survey research. Therefore,

following suggestions by Podsakoff and Organ (1986), a procedural remedy was used to reduce

method bias by guaranteeing response anonymity. Another procedural remedy was the

separation of predictor and criterion variables psychologically – the items measuring different

constructs were mixed throughout the questionnaire. In addition, almost half of the items on the

instrument were reverse-worded.

Data for the study were collected through self-administered questionnaires over a period

of three weeks. Subjects for the study were students enrolled in four operations and information

management courses at a major private university in the U.S. Two of the courses were

undergraduate courses and two MBA elective courses. All four courses had roughly the same

number of enrollment. Each subject was asked to visit a specific website from six retailing

websites, with the purpose of selecting and possibly purchasing one of three products (a pair of

shoes, a briefcase, and a gift item of their choice). The surveys handed out to each course

consisted of an equal number of the six websites. The surveys were handed out randomly;

therefore neither the subject nor the researchers could know in advance who would be asked to

visit which website, thus creating randomization. Those enrolling in more than one of the four

11

courses were identified beforehand and received the questionnaire only once. The incentive for

participating was extra course credit (2% of course grade).

Heim and Sinha (2002) classify an electronic retailer’s service process into four

categories: service kiosk, service mart, mass service customization, and joint alliance service

customization. According to Heim and Sinha, a service kiosk, which is basically an electronic

brochure, has little or no ability to sell online and extremely limited service process, making a

service kiosk unsuitable for our study. Consequently, we selected six websites, each belonging

to one of the other three categories, hence maximizing the variation in the technological

capabilities embedded in the service processes offered by these e-tailers. Three of the six sites

were general broad-category shopping sites (storerunner.com, nbci.com1, netmarket.com) that

can be considered joint alliance service customization – service products were designed and

delivered via interlinking systems between several companies and service processes represented

operations oriented toward multiple-company delivery of service products. Two other websites,

one luggage retailer (luggageonline.com) and one shoe retailer (shopping.zappos.com), were

chosen as mass service customization sites that were single-company implementation of service

process delivery. Finally, a gift item retailer (123-gifts.com2) was chosen as a service mart that

provided basic technological service capabilities such as searching and ordering products but

lacked more complicated service features such as order tracking.

After the subjects completed an encounter with the website, a survey instrument was self-

administered. The subjects were not required to complete a purchasing transaction. Our final

sample consisted of 149 complete and valid responses out of 170 questionnaires that were

handed out, for a response rate of 87.6%. To determine whether nonresponse bias was an issue,

1 This site has evolved into a general portal instead of a shopping site since the study. 2 No longer in operation as of November 2004.

12

we used the procedure outlined by Armstrong and Overton (1977) to compare early (surveys

returned within the first week) with late responders. No significant differences in any of our

measures were noted.

The age of respondents varied from 20 to 39, with an average age of 24. Of these, 45%

were female and 55% male. 91% of the respondents had prior online shopping experience, with

an average of 8 online purchases. 95% of the respondents used the Internet on a daily basis. In

addition, 60% of them typically used high-speed Internet access (DSL, cable modem, or T1

connection).

3.2. Scale Development

In an effort to provide reliability, the study instrument relied upon existing measures

whenever possible, reverse questions, single barrel questions, the test-retest method, and pre-

testing of the questionnaire. The final survey instrument contained one binary question that

asked the user’s success in finding and/or purchasing the target product, 36 questions (7-point

Likert scale) to evaluate the constructs, and 11 demographic questions.

Perceived e-Service Process. Since there was no existing scale explicitly measuring the

eSDS process from the customer’s point of view, new items were created according to Heim and

Sinha’s taxonomy of e-service process (2002). Heim and Sinha suggest that website navigation,

product information and representation, order processing and fulfillment are major e-service

process dimensions. Since our study did not require the subjects to actually purchase a product,

order fulfillment was not measured. Five items were created to measure the subject’s perception

of various technological capabilities embedded in a website, such as website navigation,

information searching, and product ordering. In addition, Roth and Jackson (1995) and Meuter

et al. (2000) both identify process error as a major source of dissatisfaction in a technology-based

13

self-service encounter. One additional item was thus adapted from Roth and Jackson (1995) to

measure process error.

Service Value. Service value in recent years has been considered a key strategic variable

to help explain consumer purchase behavior and relationship commitment (Patterson and Spreng

1997). In service management and marketing, value is typically defined from the consumer's

perspective. Heskett et al. (1994) defines value as the results customers receive in relation to the

total costs (both the prices and other costs to customers incurred in acquiring the service).

Perceived value is often viewed as the customer's overall assessment of the utility of a product

based on perceptions of what is received and what is given (Zeithaml 1988).

Perceptions of value, however, are not limited to the functional aspects but may include

social, emotional and even epistemic value components. Prior research indicates that three

elements contribute to a consumer’s perception of value: product price, product quality, and

shopping experience. Kerin et al. (1992) investigated the effect price, product quality and

shopping experience had on value perceptions of a retail store (rather than a product), concluding

that the shopping experience had a greater effect on store value than did price or product quality.

Existing scales measuring value, however, overwhelmingly focus on price (see, e.g., Patterson

and Spreng 1997 and Sweeney et al. 1999). In our research, since we were mainly interested in

service value relative to shopping experience (we did not ask our subjects to actually purchase a

product), new items were developed to focus on measuring the “give” and “receive” trade-off in

terms of efforts.

Perceived Ease of Use. Ease of use items were adapted from existing scales. The

perceived ease of use measures first developed by Davis (1989) in the technology acceptance

14

model, then modified by Devaraj et al. (2002) and Koufaris (2002) for online transactions

formed the basis for our scale.

Perceived Control. In developing the scale for perceived control, our key focus was that

the items should refer to a technology-based (i.e., the Internet) service encounter. Bowen and

Johnston (1999) suggest that perceived control emphasizes the importance of the individuals’

subjective assessment of whether they can exercise discretion and influence. Negative

consequences, such as alienation and frustration, result when this basic need is not satisfied.

Therefore, incorporating the scales developed by Koufaris (2002) for online retailing, and Hui

and Bateson (1991) for service encounters, we created a six-item scale for perceived control

which tried to capture the positive as well as negative feelings customers might experience in

online service encounters.

Interactivity. As previously mentioned, most existing scales for interactivity in the

information systems literature mainly refer to websites’ technological capability. Our research

focuses on the communication and social aspects of the online interaction, following Ha and

James’ analysis (1998). The only available scale for communication-based interactivity is by

Merrilees (2002) which has a reported reliability measure of 0.85. After a careful examination of

the scale, we felt that some items (e.g., “the overall shopping experience is very pleasant and

enjoyable”) were more about general satisfaction with the website than the interactivity aspect.

Therefore, we adapted two out of the seven items from that scale.

In their conceptualization of interactivity, Ha and James (1998) list connectedness and

information collection as important dimensions of interactivity. Connectedness refers to the

feeling of being able to link to the outside world and to broaden one's experience whereas

information collection refers to a website’s ability to provide and collect necessary information

15

to and from consumers in a transaction. Based on this conceptualization, we developed four new

items for the interactivity scale.

Customer Satisfaction. Five of the items used to operationalize customer satisfaction

came from Oliver and Swan (1989), which had a reported scale reliability of over 0.95.

Although their scale was developed when measuring respondents’ satisfaction in the context of

new car purchases, the wording of the scale is very general and the items have been used by

others in various online research context (e.g., Devaraj et al. 2002, Janda et al. 2002). Therefore

we adapted the scale and made some wording adjustment to reflect the web technology based

service experience. In addition, we added one more item from McKinney et al. (2002) as a

general measure of overall satisfaction towards a website.

4. Data Analysis

All measurements of the constructs are based upon the respondent’s opinions. Unknown

covariates (e.g. traffic volume on the Internet, Internet service providers’ technologies), which

neither the website nor the customer can control, may have an influence. We rely on a large

sample size to mitigate these unknowns. SPSS 13 for Windows was employed for exploratory

factor analysis. Amos 5 was the structural equation modeling (SEM) package utilized for the

confirmatory factor analysis and for determining relationships among the constructs.

Heteroscedasticity was observed in the scatterplots containing the ease of use index and

unfortunately could not be corrected by various transformations. While not desirable, the

existence of heteroscedasticity does not, however, invalidate the index. More than likely, any

hypothesis testing involving this index will be either too conservative or too sensitive (Hair et al.

1998). Across the three products, none of the indices were found to significantly differ.

16

As mentioned before, common method variance can be a potential source of bias in

survey research. Negative affectivity, in particular, can be a problem for this research since

subjects may react negatively to a website, which could affect all of their responses. One of the

procedures commonly used to test for the presence of common method bias in a data set is

Harman’s one-factor test (Podsakoff et al. 2003) – if a single factor is obtained from an

exploratory factor analysis or if one factor accounts for a majority of the covariance in the

independent and dependent variables, then the threat of common method bias is high. Our factor

analysis did not indicate a single-factor structure that explained significant covariance,

suggesting that common method bias is not a cause for concern in our sample.

4.1. Exploratory Factor Analysis

Responses to the questionnaire were subjected to an exploratory factor analysis. Recent

research has demonstrated the benefits of using exploratory factor analysis as a complement to

theory in specifying the appropriate factor loadings in the measurement model (Gerbing and

Hamilton 1996). The Kaiser-Meyer-Olkin Measure of Sampling Adequacy (MSA) for our data

is .882, well above the .80 level deemed as meritorious for factor analysis (Kaiser 1970).

The principal factor method was used to extract the factors, with a promax (oblique)

rotation. The eigenvalue-one criterion suggested eight factors. However, factors 7 and 8 each

only accounted for less than 3% of the common variance. A scree test suggested six meaningful

factors. Therefore, we decided to retain only six factors, which together explained 60% of the

total variance.

In interpreting the rotated factor pattern, according to Hair et al. (1998), our sample size

requires a factor loading of .45 at the minimum to be significant. A factor loading of .50 or

greater would be considered more ideal. Using this criterion, table 1 presents the questionnaire

17

items and their corresponding factor loadings that are considered significant (.45). Each factor is

labeled in the table for easy interpretation.

(Table 1 - EFA factor loading)

Several items turned out problematic (SAT4, VAL1, VAL2, VAL5, EOU4, PC4, PC6,

INT1, INT2, INT3, and eSDS4). Given the exploratory nature of our study, we decided to first

retain all the items with a .45 or greater factor loading that loaded on their intended constructs.

Problem items were dropped from further analysis. The fact that SAT4 did not load significantly

on customer satisfaction was a surprise, given that the item had been used in prior research. In

addition, although PC4 is typically used as an indicator for perceived control in the literature, the

fact that it loaded on customer satisfaction is consistent with the study by McKinney et al. (2002)

who used the item as an indicator of overall customer satisfaction towards a website and reported

a reliability index of .98 for the scale.

Items VAL1, VAL2, and VAL5 were created in an attempt to capture the “give” and

“receive” trade-off in terms of effort, rather than price. However, these items seemed ambiguous

upon closer examination (for example, the trade-off in VAL2 “the service provided through the

website was very efficient” was not obvious), justifying their exclusion from further data

analysis. INT1, INT2, INT3 also seemed to be more about “preparation” than “interactivity.”

Items VAL6, EOU6 and INT6 all had marginal loadings. However, reliability analyses

indicated that including VAL6 in the value scale would improve the alpha from .482 to .629.

EOU6 and INT6, on the other hand, would reduce the alpha level for their corresponding scale.

Therefore, VAL6 was retained whereas EOU6 and INT6 were dropped.

Next, internal consistency reliability analyses were conducted with the remaining items.

Table 2 reports the results and the summary statistics of all scales.

18

(Table 2 – construct reliability)

Although Nunnally (1978) recommends .70 as the threshold for an acceptable alpha,

Bagozzi and Yi (1988) and Hair et al. (1998) have both endorsed reliabilities as low as 0.60 for

exploratory research when structural equation modeling is used, as is the case with our research.

4.2. Testing the Measurement Model

The model derived from EFA was then submitted to confirmatory factor analysis to

determine model fit. The measurement model was estimated using the maximum likelihood

method, and the chi-square value for the model was statistically significant (chi-square(215,

n=149) = 396.048, p<.001). Technically, this chi-square statistic may be used to test the null

hypothesis that the model fits the data. In practice, however, the statistic is very sensitive to

sample size and departures from multivariate normality, and will often result in the rejection of a

well-fitting model. For this reason, it has become common practice to seek a model with a

relatively small chi-square value, rather than necessarily seek a model with a nonsignificant chi-

square. Many use the informal criterion that the model may be acceptable if the chi-square/df

ratio is less than 2 (Hatcher 1994), a criterion our model met (chi-square/df=1.842).

Another result, however, indicated that there was in fact a problem with the model’s fit:

the factor loading for item PC3 failed to load above at least 0.30 (Hatcher 1994). Therefore, this

item was dropped and the resulting model was tested again.

The overall model fit statistics for the revised model reflected reasonable fit. The chi-

square/df ratio is less than 2 (chi-square=372.377, df=194). CFI, GFI, AGFI and RMSEA are all

within reasonable range, although less than ideal (CFI=.879, GFI=.815, AGFI=.760,

RMSEA=.078). Therefore, the revised model was tentatively accepted as the study’s “final”

measurement model, and a number of tests were conducted to assess its reliability and validity.

19

Construct reliability for five of the six constructs remained the same as listed in table 2.

The only difference is the construct perceived control, which now has three indicators instead of

four (PC3 was deleted from the revised model). The reliability for the scale increased from .686

to .770, suggesting that dropping PC3 was the right choice. To summarize, all six scales now

demonstrated acceptable levels of reliability for exploratory research.

Table 3 reports the standardized factor loadings for the revised model. All factor

loadings were significant (p<.001). This finding provides evidence supporting convergent

validity of the indicators (Anderson and Gerbing 1988).

(Table 3 - CFA factor loading)

Discriminant validity was assessed using two different criteria: the variance extracted test

(Fornell and Larcker 1981) and the chi-square difference test (Anderson and Gerbing 1988). The

variance extracted estimate is a measure of the amount of variance captured by a construct,

relative to the variance due to random measurement error. A construct demonstrates

discriminant validity if its variance extracted estimate is .50 or greater (Fornell and Larcker

1981). Two of our six constructs failed this test – the variance extracted estimate was only .374

for service value and .463 for perceived ease of use. However, Hatcher (1994) cautions that this

test is quite conservative. Therefore, the chi-square difference test was used to further assess the

discriminant validity of these two constructs. We ran multiple models, constraining the

correlations between the construct in question and one other construct to 1 in each model, and

compared each of these models to the original model. The chi-square difference was significant

at the .05 level for all pairs of models, providing evidence of discriminant validity.

20

Combined, these findings generally support the reliability and validity of the constructs

and their indicators. The revised model was therefore retained as the study’s final measurement

model.

4.3. The Structural Model

Jöreskog and Sörbom (1996) state the minimum sample size is a function of the number

of variables, k: k(k-1)/2. Our sample size would permit 17 variables, less than the number

required to avoid identification issues. Therefore, the summated score for each index was used

in the SEM analysis. Baumgartner and Homburg (1996) cite numerous articles employing this

approach as evidence of its acceptance in a variety of academic disciplines. Furthermore,

Netemeyer et al. (1990) report that this approach provides the same results as models with

multiple indicators.

Although the data set under consideration cannot be assumed to have a multivariate

normal distribution, Cortina et al. (2001) state that there is considerable evidence that the

Maximum Likelihood Estimator (MLE) “is robust with respect to many types of violations of the

multivariate normality assumption.” Therefore, since there is no indication of extreme departure

from normality, the MLE was used in the SEM analysis without transforming any of the data

distributions. It is generally accepted that the minimum sample size to ensure appropriate use of

MLE is 100 to 150 (Ding et al. 1995), a requirement that our data set met.

The structural model was developed with the error variance of each measurement

variable set equal to the product of its variance and one minus its reliability coefficient. The path

from the latent variable to the composite indicator is fixed at the square root of the reliability

coefficient, α (Jöreskog and Sörbom 1996, Baumgartner and Homburg 1996).

21

Modeling of the interaction effects followed the procedures developed by Mathieu et al.

(1992). Selection of the Mathieu et al. method was based upon Cortina et al.’s (2001) comment

that this method is especially useful when testing complicated theoretical models that include

both mediated and moderated relationships, as is the case with our proposed model.

All interaction effects involving the interactivity construct were found to be insignificant.

Thus support for our hypotheses H5a through H5f is not provided by the data. Another causal

path also proved to be non-significant: from ease of use to customer satisfaction (H4b). In

addition, none of the control variables (i.e., gender, prior Internet experience, and prior online

shopping experience) was significantly related to perceived ease of use and perceived control.

Goodness of fit indices for the model appear in table 4, in the column headed “Mt: Theoretical

Model.” Values on the CFI, NFI, GFI, and AGFI were acceptable. However, a review of the

model’s residuals revealed that one of the standardized residuals was relatively large (in excess

of 2.0). These results showed that the initial theoretical model was problematic.

(Table 4 – Fit Indices for the Structural Model)

However, the model modification indices produced by Amos indicated that additional

relationships may exist, namely that service value and perceived control might both be

influenced by perceived ease of use. Furthermore, interactivity, instead of being a moderating

variable, might have a direct impact on customer satisfaction. Adding a path from perceived

ease of use to service value seems consistent with the definition of service value in terms of the

“give” versus “receive” tradeoff – when a website is difficult to use, customers might have to

give more effort, decreasing the perceived value the customer receives. In addition, when a

customer thinks a website is easy to use, conceptually, it makes sense that the customer might

also think he has more control. Interactivity, on the other hand, according to a recent study by

22

Lii et al. (2004), is a direct driver of repeat visits. Because the addition of these suggested

relationships could be justified on theoretical grounds, corresponding paths were added to the

theoretical model Mt. The resulting model, revised model Mr, was then estimated.

Fit indices for the revised model are presented in table 4. It can be seen that the fit

indices (i.e., CFI, NFI, GFI, and AGFI) were not only above .9 but also higher than those

displayed by the initial theoretical model. In addition, the revised model produced a non-

significant p value, thus justifying the addition of the new paths. The R2 value showed that

80.2% of the variance in customer satisfaction was accounted for by the relationships in the

model.

The revised model is presented in Figure 2 along with the path coefficients.

Figure 2. Revised Model

Customer Satisfaction

Interactivity

PerceivedEase of Use

ServiceValue

Customer Satisfaction

Perceived Control

Interactivity

H1b(+): 0.394H1a(+

): 0.43

7

H4a(+): 0.587

H3b(+): 0.215**H3a(+): 0.438

Perceived eSDSProcess

H2(+): 0.490

0.473

0.553

-0.15

1*

0.300**

Customer Satisfaction

Interactivity

PerceivedEase of Use

ServiceValue

ServiceValue

Customer Satisfaction

Perceived Control

Interactivity

H1b(+): 0.394H1a(+

): 0.43

7

H4a(+): 0.587

H3b(+): 0.215**H3a(+): 0.438

Perceived eSDSProcess

H2(+): 0.490

0.473

0.553

-0.15

1*

0.300**

Note: ** p<0.05. * p<0.1. All other path coefficients are significant at p<0.001.

23

4.4. Discussion

The revised model is the best model that fits the data collected with the survey instrument.

This new model suggests that interactivity, instead of being a moderator, actually acts as a

mediator between the eSDS process and customer satisfaction. However, there is an unusual

element in this relationship. Specifically, as the eSDS process improves, allowing the customer

to make more decisions and choices in the service process, the interactivity allowed by the

website increases. Interactivity, surprisingly, doesn’t necessarily lead to higher customer

satisfaction. On the contrary, if a customer feels an increasing need to interact with the service

provider, his/her satisfaction with the website decreases. This finding is consistent with that by

Zeithaml et al. (2002) in their focus group discussions regarding important service requirements

in the e-commerce context. Only when customers need special assistance, e.g., there is a process

error, do they feel the need to initiate an interaction with a customer service representative.

Many focus group participants were otherwise only interested in having efficient transactions.

The data also reveals that perceived ease of use has a mediated impact, rather than a

direct impact, on customer satisfaction through both service value and perceived control. That is,

as the user’s perception of a website’s ease of use decreases, the service value they feel they

receive from using the website decreases and the user’s perception of his ability to control the

process decreases. But perceived ease of use has no direct impact on customer satisfaction. We

think the probable reason for this is related to the demographics of our study sample. The

sample population reported homogenous and high values for their self-assessment of their

technical competence.

It is worth noting that our result does not necessarily contradict the TAM model which

theorizes that perceived ease of use directly influences a user’s attitude toward a technology,

24

which would be customer satisfaction in this study. The TAM model suggests that perceived

ease of use influences perceived usefulness because, other things being equal, the easier the

technology is to use, the more useful it can be. Perceived usefulness is conceptually closely

related to the service value construct in our model. Our result that perceived ease of use

influences customer satisfaction through service value is in fact consistent with TAM.

Perceived ease of use is a construct related to customer-specific characteristics. For

example, when a customer is technologically sophisticated and has extensive experience

shopping online, she may feel that she has a great degree of control over how she conducts the

transaction. On the other hand, if a customer is not experienced, she may feel at a loss and not

able to engage in the process. Intuitively, this makes sense. The question that must be answered

is: How sound is the theoretical basis of the relationship? Information systems literature has

examined the relationship between computer users’ self-efficacy – the judgment of one’s ability

to perform a specific task using a computer – and their perceived ease of use of computer

systems, and found significant correlations (e.g., Hong et al. 2002). Computer self-efficacy, on

the other hand, is considered the conceptualization of perceived control (Venkatesh 2000). The

question of whether there is a causal relationship between perceived control and ease of use, and

if so, in which direction, remains unanswered. Further research is certainly needed to examine

the precise relationship between the two constructs.

Since the six websites we used belonged to three types of service processes with regard to

their corresponding technological capabilities, namely service mart, mass service customization,

and joint alliance service customization, a post-hoc analysis was done to determine whether there

were any differences in customer satisfaction by type of website. Our analysis did not yield any

significant result. Given our finding that the eSDS process has a positive influence on customer

25

satisfaction, conceptually, one could speculate that the more comprehensive technological

capabilities embedded in a mass service customization site might lead to happier customers than

the basic website capabilities from a service mart. Although this conjecture is not supported by

our data, we believe this issue is an interesting one and should be explored in future research.

5. Implications and Conclusion

In this paper we have argued that service is critical to the success of electronic commerce.

Building on theories from service management, we have examined what e-service technological

capabilities should be embedded in a firm’s website and what technology features should take

priority. We contribute to the management of technology domain by proposing a theoretical

model that helps firms understand the impact of e-service technological capabilities on online

customer satisfaction. In addition, our model also helps firms to justify their investment in e-

service technology.

5.1. Research Implications

It is important to note that online customer satisfaction has been evaluated along other

dimensions. For example, SERVQUAL is often referred to as an important factor leading to

customer satisfaction (Devaraj et al. 2002). Information quality and availability is another factor

examined by the literature (e.g., McKinney et al. 2002). A key contribution of our research is we

demonstrate that the technological capabilities embedded in the e-service process through which

services are delivered and information is presented to the customer are critical to customer

satisfaction. For example, does the website have the technological capability that allows the

customer the flexibility of customizing the information content? If a website offers an abundant

amount of information but doesn’t allow much user freedom in terms of choosing where to go at

26

the website and what information to see, the customer is unlikely to be satisfied. The

technological capabilities of the eSDS processes, therefore, really are the foundation of other

service measurement parameters.

Although the service management literature has established that the SDS has a direct

impact on service value, traditional service delivery systems are made of very different

components from an electronic SDS. Traditionally it is the service employees that determine the

service value the customers perceive they receive. By examining the linkage between the

electronic SDS and service value, we demonstrate that even without face-to-face interactions

with service employees, the eSDS plays a vital role in customer satisfaction. Furthermore, we

extended the current literature that looks at online customer satisfaction – extant research mainly

focuses on the relationship between service quality and customer satisfaction or system quality

and customer satisfaction. We have demonstrated that the technological capabilities embedded

in e-service processes are really the key factor determining service quality and ultimately online

customer satisfaction.

The capabilities embedded in an e-service technology have different dimensions. For

example, Levitt (1976) draws upon manufacturing sources in using the words “standardized” and

“customized” to define the poles of a service process continuum whereas Shostack (1987) uses

“complexity” and “divergence.” In this research, we did not drill down inside the eSDS to these

different capability dimensions to analyze which ones are the most important in determining

online customer satisfaction. However, conceptually it is possible that some dimensions play a

more significant role than others. Indeed, customization has been identified as an important e-

service technology feature preferred by many online customers (Nunes and Kambil 2001).

27

Therefore, future theoretical investigations are warranted to understand what dimensions of

service processes are important in delivering quality online services.

From a practical point of view, our research provides investment guidance to firms in

their creation of and upgrades for e-service technologies. An e-service website can offer

different technological capabilities. Many companies, however, are financially constrained in

practice in terms of what e-service technology features to focus on. It is therefore important to

identify those features that are critical to customer satisfaction. In addition to interface design

factors identified by prior research, such as site aesthetics, graphics presentation, visual effects,

our research results bring to the foreground the importance of procedural and process design

capabilities embedded into an e-service technology site. Companies deploying e-service really

need to understand that their website is not only an interface with their customers, but also an

information system that embeds their business processes. Having smooth and flexible website

processes means seamless system integration. For example, the website needs to be integrated

with the company’s inventory system so customers can check the availability of products; with

the order tracking system so customers can check their order status, etc. Therefore, presenting a

“pretty face” is only a small part of the whole website design effort. How the whole system is

designed, what technological capabilities to offer, and what service processes are enabled

ultimately determines what service value a company delivers to its customers and how satisfied

the customers are.

Although previous research has argued that website interactivity is an important

technology feature for customer satisfaction, our research only provides limited support for this

argument. Offering real-time interactivity between the service provider and the customer can be

expensive, as it involves more human intervention. Based on our results, we believe that at this

28

point there is not enough justification for companies to spend a significant amount of money on

this aspect of e-service. More research on the role of interactivity in e-service delivery and how

interactivity affects customer satisfaction is clearly needed.

5.2. Limitations and Suggestions for Future Research

There are several ways in which future research could strengthen the results of this study.

First, our survey did not require the subjects to actually carry out the purchase. Therefore the

technology-based service capabilities we examined in this paper do not cover any post-purchase

services such as fulfillment and returns. But practitioners as well as researchers (e.g., Zeithaml

et al. 2002) have voiced that post-purchase service is also critical to customer satisfaction and

may explain why some customers never came back to a certain website. This aspect of e-service

capabilities needs to be captured as well in order to evaluate the overall service value an eSDS

can deliver.

Although prior research indicates that the user’s perception of a system’s ease of use is a

significant direct driver of satisfaction, our empirical testing failed to confirm this hypothesis.

As pointed out by Zeithaml et al. (2002), customer-specific characteristics such as demographics

and psychographics could have a strong impact on perceived service value. Perceived ease of

use is directly influenced by the customer’s technology proficiency. The inability to demonstrate

the significance of this construct, we believe, may be an artifact of the sample population.

Therefore, further research is warranted: the size and heterogeneity of the sample should be

increased by the inclusion of individuals outside of the setting of a higher education institution.

The research design of this study has several limitations. First, the measurement

instrument needs to be fine tuned. As discussed in section 4, quite a few items had high cross-

loadings in the exploratory factor analysis. Although the constructs, e.g., service value,

29

perceived ease of use, and perceived control, are theoretically different, they are also correlated.

Developing scale items that clearly distinguish these constructs from one another is an important

task for future research.

As mentioned in section 4, common method variance can be a potential problem for

survey research. Given that our constructs were all measured by the same method from the same

subjects, this study potentially faces the same bias, although various procedural remedies were

employed to reduce the bias. One way of addressing this problem in future studies is to use an

objective measure of the technological capabilities of an e-service website, instead of examining

it from the customer’s perspective. A more objective measure not only helps to reduce methods

variance but might also yield more insight in how customers view a website’s service value and

how that view ultimately translates into satisfaction, providing the critical link between an

organization’s e-service process design decision and customer response.

As electronic commerce continues to grow, e-service is going to play an even bigger role

in customer satisfaction. Managing e-service technology will become more critical for firms

intending to compete online. Firms need to carefully evaluate their technology-based service

offerings and understand how to design web-based technological capabilities to deliver the type

of services customers demand. Moreover, as technology is constantly evolving, so is

technology-based e-service process and its impact on business strategy. The effective

management of the integration of business and technology is becoming an indispensable part of

many organizations’ value creation strategy. This work is only a first step in trying to

understand the e-service technology and the impact of the technology on customer satisfaction.

We believe this is a promising research area for researchers in the management of technology

domain.

30

References:

Anderson, J.C. and D. W. Gerbing. 1988. Structural equation modeling in practice: a review and

recommended two-step approach. Psychological Bulletin, 103, 411-423.

Armstrong, J. S. and T. S. Overton. 1977. Estimating nonresponse bias in mail surveys. Journal

of Marketing Research, 14(3), 396-402.

Bagozzi, R. P., and Y. Yi. 1988. On the evaluation of structural equation models. Journal of the

Academy of Marketing Science, (16), 74-79.

Baumgartner, H. and C. Homburg. 1996. Applications of structural equation modeling in

marketing and consumer research: a review. International Journal of Research in

Marketing, 13(2), 139-161.

Bednarz, A. 2003. Retailers shore up Web sites for holidays. Network World, December

15, 20(50), 10.

Betz, F. 2001. Executive strategy: strategic management and information technology. New

York: John Wiley & Sons, Inc.

Bitner, M.J., B. H. Booms, and L. A. Mohr. 1994. Critical service encounters: the employee’s

viewpoint. Journal of Marketing, 58, 95-106.

Bowen, D. E. and R. Johnston. 1999. Internal service recovery: developing a new construct.

International Journal of Service Industry Management, 10(2), 118.

Brady, M. and J. J. Cronin. 2001. Customer orientation: effects of customer service perceptions

and outcome behaviors. Journal of Service Research, 3(3), 241-251.

Bryne, B.M. 1998. Structural Equation Modeling with LISREL, PRELIS and SIMPLIS: Basic

Concepts, Applications, and Programming, Lawrence Erlbaum Associates, Inc., Mahwah,

NJ.

31

Chen, L., M. Gillenson, and D. Sherrell. 2004. Consumer acceptance of virtual stores: a

theoretical model and critical success factors for virtual stores. Database for Advances in

Information Systems, 35(2), 8-31.

Cortina, J.M., G. Chen, and W.P. Dunlap. 2001. Testing interaction effects in LISREL:

Examination and illustration of available procedures. Organizational Research Methods,

4(4), 324-360.

Davis, F. 1989. Perceived usefulness, perceived use of ease, and user acceptance of information

technology. MIS Quarterly, 13(3), 319-340.

Devaraj, S., M. Fan, and R. Kohli. 2002. Antecedents of B2C channel satisfaction and preference:

validating e-commerce metrics. Information Systems Research, 13(3), 316-333.

Ding, L., W. F. Velicer, and L. L. Harlow. 1995. Effects of estimation methods, number of

indicators per factor and improper solutions on structural equation modeling fit indices.

Structural Equation Modeling, 2, 119-143.

Fornell, C. and D.F. Larcker. 1981. Evaluating structural equation models with unobservable

variables and measurement errors. Journal of Marketing Research, 18, 39-50.

Gerbing, D.W. and J. G. Hamilton. 1996. Viability of exploratory factor analysis as a precursor

to confirmatory factor analysis. Structural Equation Modeling, 3(1), 62-72.

Ha, L, and L. James. 1998. Interactivity reexamined: a baseline analysis of early business web

sites. Journal of Broadcasting & Electronic Media, 42(4).

Hair, Jr., J. F., R. E. Anderson, R. L. Tatham, and W.C. Black. 1998. Multivariate Data Analysis,

5th edition, Prentice Hall, Upper Saddle River, NJ.

Hatcher, L. 1994. A Step-by-Step Approach to Using SAS for Factor Analysis and Structural

Equation Modeling. Cary, NC: SAS Publishing.

32

Heim, G. and K. Sinha. 2002. Service process configurations in electronic retailing: a taxonomic

analysis of electronic food retailers. Production and Operations Management, 11(1), 54-

74.

Heskett, J.L., T.O. Jones, G.W. Loveman, W.E. Sasser, Jr., and L.A. Schlesinger. 1994. Putting

the service-profit chain to work. Harvard Business Journal, March-April, 164-174.

Hong, W., J. Thong, W. Wong, and K. Y. Tam. 2002. Determinants of user acceptance of digital

libraries: an empirical examination of individual differences and system characteristics.

Journal of Management Information Systems, 18(3): 97-124.

Hui, M.K. and J.E.G. Bateson. 1991. Perceived control and the effects of crowding and

consumer choice on the service experience. Journal of Consumer Research, 18, 174-184.

ICSA (2001) 2001 ICSA/E-Satisfy.com Study of Electronic Customer Service.

http://www.icsa.com/resources/pubs/eSatisfy_toc.cfm.

Janda, S., P. J Trocchia, and K. P Gwinner. 2002. Consumer perceptions of Internet retail service

quality. International Journal of Service Industry Management, 13(5), 412-431.

Jöreskog, K. and D. Sörbom. 1996. LISREL8: User’s Reference Guide, Scientific Software

International, Inc., Chicago.

Kaiser, H. F. 1970. A second-generation little jiffy. Psychometrika, 35: 401-415.

Kamakura, W., V. Mittal, F. de Rosa, and J. Mazzon. 2002. Assessing the service-profit chain.

Marketing Science, 21(3), 294-317.

Kerin, R. A., A. Jain, and D. J. Howard. 1992. Store Shopping Experience and Consumer Price-

Quality-Value Perceptions. Journal of Retailing, 68, 376-397.

Koufaris, M. 2002. Applying the technology acceptance model and flow theory to online

consumer behavior. Information Systems Research, 13(2), 205-223.

33

Langeard, E., J. Bateson, C. Lovelock, and P. Eiglier. 1981. Marketing of services: new insights

from consumers and managers. Report No. 81-104. Marketing Science Institute,

Cambridge, MA.

Lee, J. and A. Allaway (2002) Effects of personal control on adoption of self-service technology

innovations. The Journal of Services Marketing, 16(6), 553-572.

Levitt, T. 1976. The industrialization of service. Harvard Business Review, 54(5), 63-71.

Lii, Y., H. Lim, and L. Tseng. 2004. The effects of web operational factors on marketing

performance. Journal of American Academy of Business, 5(1), 486-494.

Lohse, G. and P. Spiller. 1998. Electronic shopping. Communications of the ACM, 41(7), 81-87.

Mathieu, J.E., S.I. Tannenbaum, and E. Salas. 1992. Influences of individual and situational

characteristics on measures of training effectiveness, Academy of Management Journal,

35, 828-847.

McKinney, V., K. Yoon, and F. Zahedi. 2002. The measurement of Web-customer satisfaction:

An expectation and disconfirmation approach. Information Systems

Research, 13(3), 296-315.

Merrilees, B. 2002. Interactivity design as the key to managing customer relations in e-

commerce. Journal of Relationship Marketing, 1(3-4), 111-126.

Meuter, M., A. Ostrom, R. I. Roundtree, and M. J. Bitner. 2000. Self-service technologies:

understanding customer satisfaction with technology-based service encounters. Journal

of Marketing, 64(3), 50-64.

Mittal, B. and W. M. Lassar. 1996. The role of personalization in service encounters. Journal of

Retailing, 72 (1), 95-109.

34

Netemeyer, R. G., M. Johnston, and S. Burton. 1990. Analysis of Role Conflict and Role

Ambiguity in a Structural Equations Framework. Journal of Applied Psychology, 75(2),

148-157.

Nunes, P. and A. Kambil. 2001. Personalization? No thanks. Harvard Business Review, April,

2-3.

Nunnally, J. C. 1978. Psychometric Theory, New York: McGraw-Hill.

Oliver, R. and J. Swan. 1989. Consumer perceptions of interpersonal equity and satisfaction in

transactions: a field survey approach. Journal of Marketing, 53 (April), 21-35.

Palmer, J. 2002. Web site usability, design, and performance metrics. Information Systems

Research, 13(2), 151-167.

Patterson, P. and R. Spreng. 1997. Modeling the relationship between perceived value,

satisfaction, and repurchase intentions in a business-to-business, service context: an

empirical examination. International Journal of Service Industry Management, 8(5).

Podsakoff, P.M. and D.W. Organ. 1986. Self reports in organizational research: problems and

prospects. Journal of Management, 12(4), 531-544.

Podsakoff, P.M., S. MacKenzie, J. Y. Lee, and N. Podsakoff. 2003. Common method biases in

behavioral research: a critical review of the literature and recommended remedies.

Journal of Applied Psychology, 88(5), 879-903.

Roth, A.V., W.E. Jackson III. 1995. Strategic determinants of service quality and performance:

Evidence from the banking industry. Management Science, 41 (11), 1720-1733

Rust, R. and P.K. Kannan. 2003. E-service: a new paradigm for business in the electronic

environment. Communications of the ACM, 46(6).

35

Seyal, A., M. Rahim, and M. Rahman. 2002. A study of computer attitudes of non-computing

students of technical colleges in Brunei Darussalam. Journal of End User

Computing, 14(2). 40-47.

Shapiro, C. and H. R. Varian. 1999. Information Rules: A Strategic Guide to the Network

Economy. Boston: Harvard Business School Press.

Shostack, G.L. 1987. Service positioning through structural change. Journal of Marketing, 51

(1), 34-43.

Sweeney, J., G. Soutar, and L. Johnson. 1999. The role of perceived risk in the quality-value

relationship: a study in a retail environment. Journal of Retailing, 75(1), 77-105.

Tsikriktsis, N., Lanzolla, G., and Frohlich, M. 2004. Adoption of e-processes by service firms:

an empirical study of antecedents. Production & Operations Management, 13(3), 216-229.

Venkatesh, V. 2000. Determinants of perceived ease of use: integrating control, intrinsic

motivation, and emotion into the technology acceptance model. Information Systems

Research, 11(4), 342-365.

Venkatesh, V. and M. G. Morris. 2000. Why don't men ever stop to ask for directions? Gender,

social influence, and their role in technology acceptance and usage behavior. MIS

Quarterly, 24(1), 115-139.

Zeithaml, V.A. 1988. Consumer perceptions of price, quality and value: a means-end model and

synthesis of evidence. Journal of Marketing, 52(3), 2-22.

Zeithaml, V.A., A. Parasuraman, and A. Malhotra. 2002. Service quality delivery through web

sites: a critical review of extant knowledge. Journal of the Academy of Marketing

Science, 30(4), 362-375.

36

Appendix: Measurement Scales

Constructs and Scale Items Source Perceived eSDS Process eSDS1. The website was difficult to navigate through. New item eSDS2. The number of choices at each step of the process doesn’t need to be changed.

New item

eSDS3. The website ordering process wasn’t complicated. New item eSDS4. I did not experience any errors (e.g., web pages that did not load the first time).

Adapted from Roth and Jackson (1995)

eSDS5. I had trouble finding what I was looking for on the website.

New item

eSDS6. The entire process of searching and buying took a reasonable amount of time.

New item

Service Value VAL1. Using the website was a waste of my time. New item VAL2. The service provided through the website was very efficient.

New item

VAL3. The website required a lot of effort to use. New item VAL4. I was treated fairly. New item VAL5. Very little thought was required to use this website. New item VAL6. The website doesn’t provide value. Brady and Cronin (2001) Perceived Ease of Use EOU1. The user of the website has to be skillful to use the website. Davis (1989) EOU2. The user does not have to be knowledgeable in order to use the site.

New item

EOU3. Using this website was easy. Davis (1989) EOU4. The user needs to be a frequent web user. New item EOU5. My interaction with the website was clear and understandable.

Davis (1989)

EOU6. A user does not need specific knowledge about the company in order to use the website.

New item

Perceived Control PC1. The website limited what I could do. Adapted from Seyal et al.

(2002) PC2. I felt in control at each step and could determine the outcome of the online process.

Koufaris (2002)

PC3. To use the website, I had to input unnecessary information, which was confusing.

Koufaris (2002)

PC4. I felt frustrated at the process of searching and buying. Koufaris (2002) PC5. At the website, I could do what I wanted to when I wanted to. Adapted from Seyal et al.

37

(2002) PC6. The website wasn’t complicated to use. New item Interactivity INT1. Sufficient guidelines were provided. New item INT2. Careful instructions were provided. New item INT3. I always knew what information I needed to provide. New item INT4. The website allows good two-way communication. Merrilees (2002) INT5. Interaction with customer service rep through email or phone is necessary so my question can be answered quickly.

Merrilees (2002)

INT6. Interaction with other customers through chat rooms is beneficial.

New item

Customer Satisfaction SAT1. Using the website pleased me. Oliver and Swan (1989) SAT2. I was content with the procedures for using the website. Oliver and Swan (1989) SAT3. I was very unhappy with the online experience. Oliver and Swan (1989) SAT4. The website did an excellent job for me. Oliver and Swan (1989) SAT5. It is a poor choice to use this website. Oliver and Swan (1989) SAT6. I would never use this website again. McKinney et al. (2002)

38

Table 1. Questionnaire Items and Corresponding Factor Loadings from the Rotated Factor Pattern Matrix

Factor 1 2 3 4 5 6

Questionnaire Item

Customer Satisfaction

Perceived eSDS

Process Service Value

Ease of Use

Perceived Control Interactivity

SAT1 .714 SAT2 .563 SAT3 .926 SAT4 SAT5 .548 SAT6 .750 VAL1 .886 VAL2 .549 VAL3 .681 VAL4 .590 VAL5 VAL6 .469 EOU1 .713 EOU2 .829 EOU3 .721 EOU4 EOU5 .586 EOU6 .474 PC1 .546 PC2 .721 PC3 .603 PC4 .650 PC5 .820 PC6 INT1 INT2 INT3 .508 INT4 .711 INT5 .904 INT6 .450

eSDS1 .785 eSDS2 .679 eSDS3 .655 eSDS4 .729 eSDS5 .500 eSDS6 .600

Note: N=149.

39

Table 2. Summary Statistics and Cronbach’s Alpha for All Scales

Construct Mean S.D. Cronbach’s Alpha Customer Satisfaction (SAT1, SAT2, SAT3, SAT5, SAT6) 4.89 1.20 0.869

Perceived Control (PC1, PC2, PC3, PC5) 4.66 1.07 0.686

(0.770 without PC3) Ease of Use (EOU1, EOU2, EOU3, EOU5) 6.47 1.49 0.766

Service Value (VAL3, VAL4, VAL6) 5.39 1.10 0.629

Interactivity (INT4, INT5) 6.89 2.07 0.739

Perceived eSDS Process (eSDS1, eSDS2, eSDS3, eSDS5, eSDS6) 4.54 1.35 0.824

40

Table 3. Standardized Factor Loadings for the Measurement Model

Item Description Factor Loading For Revised Model

F1: Customer Satisfaction SAT1 .747 SAT2 .729 SAT3 .640 SAT5 .854 SAT6 .803 F2: Perceived eSDS Process ESDS1 .578 ESDS2 .521 ESDS3 .663 ESDS5 .838 ESDS6 .862 F3: Service Value VAL3 .462 VAL4 .550 VAL6 .775 F4: Ease of Use EOU1 .735 EOU2 .533 EOU3 .831 EOU5 .580 F5: Perceived Control PC1 .822 PC2 .780 PC5 .589 F6: Interactivity INT4 .899 INT5 .651

Note. All loadings are significant at the .001 level.

Table 4. Fit Indices for the Structural Model

Criteria Guidelines Bryne (1998)

Mt: Theoretical Model

Mr: Revised Model

χ2 (df) p-value

Small Large

20.034 (6) 0.003

7.516 (5) 0.185

CFI > 0.90 0.947 0.993 RMSEA < 0.08 0.126 0.058

NFI >0.90 .929 0.979 GFI >0.90 .954 0.984

AGFI > 0.80 0.840 0.932

41