A Revisit of General Self-Efficacy Scale:Uni- or Multi-dimensional?

Mingming Zhou

# Springer Science+Business Media New York 2015

Abstract While a large body of research provides strong evidence that general self-efficacy (GSE) is a unidimensional construct, no study has been conducted to test orcompare its single-factor structure with alternative factor structures (e.g., two-factorstructure). The main purpose of this study was to re-examine its factor structure in aChinese context. The Chinese version of GSE scale was administered to 195 Chineseuniversity students. Personal identity and optimism instruments were also used toexamine the predictive validity of GSE. Exploratory factor analysis determined that atwo-factor solution was best fitting, explaining 54.49 % of the variance, which wasreproduced reliably in the two subsamples. Through a series of factor model testing byconfirmatory factor analysis, findings indicated that the two-factor structure wasreliable, internally valid as demonstrated. Subsequent regression analyses also showedthat only one of the two factors significantly predicted personal identity and optimism,whereas the other one was irrelevant. The findings emphasized the value of revisitingGSE as a uni- or multi-dimensional construct.

Keywords General self-efficacy . Exploratory factor analysis . Confirmatory factoranalysis

Introduction

Self-efficacy, defined as Bbeliefs in one’s capabilities to mobilize the motivation,cognitive resources, and courses of action needed to meet given situational demands^(Wood and Bandura 1989, p. 408), has been studied extensively in different fields, suchas physical activity (Sperber et al. 2014), health (Jerome and McAuley 2013), behav-ioral therapy (Gallagher et al. 2013), training proficiency (Martocchio and Judge 1997),job performance (Stajkovic and Luthans 1998), and academic achievement (Parkeret al. 2014). Such beliefs are Bthe key factor of human agency^ (Bandura 1997, p. 3)and determine whether coping behavior will be initiated, how much effort will be

Curr PsycholDOI 10.1007/s12144-015-9311-4

M. Zhou (*)Faculty of Education, University of Macau, Avenida da Universidade, Taipa, Macau, Chinae-mail: mmzhou@umac.mo

expended, and how long effort will be sustained in the face of obstacles and aversiveexperiences (Bandura 1997).

According to Bandura (1997), self-efficacy beliefs tap on three dimensions: (a) levelor magnitude (the level of task difficulty), (b) strength (the certainty of successfullyperforming a particular level of task difficulty), and (c) generality (the extent to whichmagnitude and strength beliefs generalize across tasks and situations). AlthoughBandura (1997, 2006) argued that self-efficacy measurement should be domain specif-ic; that is, the content of the scale items need to be directly related to the construct thatis being measured, others maintained that measuring self-efficacy at a general level isrelatively stable over time and over domains of functioning (Scheier and Carver 1992),as a psychological trait as opposed to a more conditional and fluctuating state (Eden1988; Gardner and Pierce 1998). Because various experiences of failure and success indifferent domains of functioning may generate more generalized beliefs of self-efficacy(Bosscher and Smit 1998), general self-efficacy (GSE) has great utility for explainingbehavior in less specific contexts (Luszczynska et al. 2005; Scholz et al. 2002) andpredicting general outcomes (Barlow et al. 1996; van der Slot et al. 2010). Since then,GSE has caught much attention in recent years in psychological and personalityresearch (e.g., Cramm et al. 2013; Ebstrup et al. 2011), because it highlights thepossibility to transfer personal efficacy judgments across tasks or activities (Bandura1997; Zimmerman and Cleary 2006). Several definitions were available for GSE.Generally, it refers to global confidence in one’s coping ability across a wide varietyof situations (Schwarzer et al. 1997a). This conceptualization captures differencesamong individuals in their tendency to view themselves as capable of handling varioussituations and meeting task demands in a variety of contexts (Chen et al. 2001).

GSE Measurement

Many measures have been invented for GSE in order to capture individuals’ generalbeliefs about their capabilities to handle different situations by and large, with re-sponses created largely through an aggregation of all successes and experiences, notjust those in a given domain (Eden 1988). Based on the accumulation of priorexperiences, respondents will need to gather information of their past performancerelated life events as described in survey items to reach a general assessment of capacityto handle such situations. Schwarzer and colleagues’ General Self-Efficacy (GSE) scale(Schwarzer and Jerusalem 1995) gained most popularity due to its ease for adminis-tration and accordance to the original conceptualization of the construct. This scaleassesses Bthe strength of an individual’s belief in his/her own ability to respond to novelor difficult situations and to deal with any associated obstacles or setbacks^ (Schwarzerand Jerusalem 1995 p. 35). It has been adapted into 30 languages, with normal(Posadzki et al. 2010; Strobel et al. 2011) and clinical populations (Rottmann et al.2010; Wells and Anderson 2011; Bonsaksen et al. 2012). Reliability analyses producedhighly consistent results, with a high range of internal consistency between an alpha of0.75 and 0.91 (Scholz et al. 2002). The GSE scale demonstrated some stability over a4-month period with a test-retest reliability coefficient of 0.63 (Barlow et al. 1996).Nonetheless, the test-retest reliability dropped to 0.47 among migrants from EastGermany over a 2-year period (Schwarzer 1993). Thus, although the GSE scale has

Curr Psychol

been found to be internally consistent in numerous studies, its test-retest reliability isnot too encouraging.

Validation studies of the GSE scale have also been conducted in 25 nations. Whilenumerous studies with western samples defend the original one-factor structure of thisconstruct (e.g., Leganger et al. 2000; Löve et al. 2012; Scholz et al. 2002) viaexploratory and confirmatory factor analyses, the one-dimensional nature of the scalewith Asian samples was not supported unequivocally. On one hand, cross-culturalresearchers argued that efficacy beliefs operate differently in non-western cultures thanthey do in western cultures (Klassen 2004). The role played by collective efficacy (i.e.,a group’s shared belief) is possibly more salient among those with collectivist orienta-tions found in many Asian countries. On the other hand, the dominant use of confir-matory factor analyses in past studies might have missed alternative factor structures ofGSE. For example, with exploratory factor analyses, Chiu and Tsang (2004) reported atwo-factor structure of the scale with a Hong Kong schizophrenia sample. The twofactors accounted for 69.0 % of the total variance. The authors interpreted the newfactor structure with Hall’s (2000) theory of self-efficacy, which stated that one’s self-efficacy in work-related situations involved both personal and environmental aspects.The personal factor related to self-efficacy included nine items that tapped on self-directing or self-determining behaviors, and the environmental factor included only oneitem that was concerned with one’s self-efficacy in the social context such as expec-tations of others and support from others. Hence, within a limited number of validationstudies with Chinese speaking populations, although a large portion of studies indicatedthat it can be seen as homogeneous and unidimensional (Cheung and Sun 1999; Leungand Leung 2011; Schwarzer et al. 1997a, b; Teo and Kam 2015; Zhang and Schwarzer1995), there is still a clear need for further examination of GSE using multiple methodsof analyses, with non-western samples.

GSE: One Factor or Multiple Factors?

The above brief review somehow invites further thinking about the conceptualization ofthis construct due to inconsistent findings about the factorial structure of GES. In effect,perusal of the GSE items showed that the semantics of the items did not converge intoone factor nicely at the conceptual level; rather, it echoed Schwarzer and Renner’s(2000) two-phase model of self-efficacy to a great extent. In this model, a distinctionwas made between preintentional motivation processes and postintentional volitionprocesses. In the preintentional phase, individuals first develop an intention (or goals)to act. The intention could stem from the belief that: (a) one is at risk forfailure (risk perception), (b) behavioral change would reduce a failure threat(outcome expectancies), and (c) one is sufficiently capable of exercising controlover a difficult behavior (perceived self-efficacy). After individuals set goals,they enter the postintentional phase wherein they plan details and invest effort.Thus, the first phase leads to a behavioral intention and the second leads toactual behavior (Schwarzer and Fuchs 1996).

Schwarzer and Renner (2000) termed self-efficacy that functions in the preintentioalphase as action self-efficacy and the one in the postintentioal phase as coping self-efficacy The distinction between these two types of efficacy is necessary because it is

Curr Psychol

possible that some individuals could be very confident in their ability to set goals andtake initiative (high in action self-efficacy) but not so confident in their ability tomaintain the effort to achieve the goals (low in coping self-efficacy). By contrast,others could have high confidence in their ability to recover from failures to continuegoal pursuit (high in coping self-efficacy) but little confidence in getting started (low inaction self-efficacy). Hence, these two types self-efficacy co-define GSE as a multi-dimensional construct.

GSE may explain a broader range of human behaviours and coping outcomes whenthe context is less specific (Luszczynska et al. 2005). It has been shown to be associatedwith a variety of variables, including future orientation (Strathman et al. 1994), anxietyand depression (Schwarzer 1992), social relations (Bandura 1997) and academicachievement (Bandura 1999). Speaking to this point, a closer look at the proposedfactor structure of GSE showed that action self-efficacy seems to be in overlap to someextent with optimism in terms that both involve the general belief that desired outcomes(i.e., goals) are likely to be achieved. This belief is thought to influence people’s goal-directed behaviors and, ultimately, goal achievement (Rand 2009). Empirically, self-efficacy has been repeatedly found to be positively correlated to optimism (e.g.,Ausbrooks et al. 1995; Chemers et al. 2001; Scholz, et al. 2002). Hence, in the currentstudy, optimism was included for the predictive validity examination. On the otherhand, literature on identity has shed light on its relationship with one’s coping strategy/style. According to Berzonsky’s (1992) process model of identity, the manner in whichindividuals deal with events or stressors that may invalidate or force revisions in theirself-views will vary with their identity. Nario-Redmond et al. (2013) also noted that theidentification of self would be associated with certain coping strategies. For this reason,personal identity was also included as a second variable for examining predictivevalidity.

Aims of the Present Study

There is a serious discrepancy between the conceptualization of GSE as an undiffer-entiated belief in one’s generalized ability as a unitary construct on one hand and themultifactorial structure of the scale on the other. Although GSE has been conceived asunidimensional, the findings summarized above suggest that this assumption needs tobe revisited and examined closely. To borrow from Cronbach’s (1956) terminology,past research has mainly followed the Blumpers^ approach to social science, whichemphasizes increasing bandwidth and prediction, as opposed to the Bsplitters^ ap-proach, which focuses more on increasing fidelity an understanding of the underlyingmechanism that link traits to criteria (also see Chen et al. 2004). Given the fact thatrepeated factor analyses of the hope model have recovered the two components (Snyderet al. 1991), it is reasonable to expect that similar factor structures would emerge fromGSE. While the large body of research provides strong evidence that GSE is aunidimensional construct, there has been no study that has tested or compared itssingle-factor structure with alternative factor structures. Accordingly, the main purposeof the present research was to re-examine the factor structure of GSE, particularly withrespect to the underlying mechanism that constitutes the construct, such that we couldgain a better understanding of the nature and function of GSE.

Curr Psychol

To summarize, although the GSE scale has fairly high internal consistency reliabilityand predictive validity, its unidimensionality and seemingly low content validity makefindings difficult to interpret. Of particular concern is whether the GSE scale capturesself-perceptions of agency and means to outcomes, or a combination of these con-structs. It is not known whether predictions made with the GSE scale in previousstudies are attributable to GSE as a whole or to any one of the constructs. For thesereasons, the GSE scale has not demonstrated sufficient validity to warrant its continueduse without further evaluation. Because the large amount of GSE research has relied onthe GSE scale, the unique contribution of GSE, as distinct from its correlates, to theory,research, and practice remains unclear. Given the methodological and conceptualdeficiencies of GSE, the current study aims to reexamine the factor structure of thisconstruct, and its reliability and validity.

Method

Participants and Procedure

The sample consisted of 195 university students from different departments inthree randomly selected universities in South China, of which 53.7 % weremales. Mean age was 20.53 years (SD=1.31). After the briefing of the objec-tive of the study, students were informed about the importance and voluntarynature of their participation. Consents were collected before the participantsstarted. Participants completed the instrument individually under the researchassistant’s supervision after class hours.

Measures

The 10-item Chinese version of GSE Scale (Zhang and Schwarzer 1995, see Appendix)was administered to assess one’s generalized beliefs of self-efficacy across situations,together with two other instruments for comparing the magnitudes of the relationshipsof GSE subscales with these variables. The GSE scale was administered with a 5-pointLikert scale ranging from 1 (completely disagree) to 5 (completely agree).

Participants were given Nario-Redmond and colleagues’ (2004) 8-item personalidentities scale which measured the level of self which differentiates the individual asunique from others. The Chinese version was used after a back-translation by twobilingual experts following a standardized back-translation procedure (Brislin 1970).The scale was administered with a 9-point Likert scale ranging from 1 (completelydisagree) to 9 (completely agree). Sample items were: BMy sense of being different fromothers.^ and BMy complete individuality.^ A composite score of all the eight items wascalculated to indicate participants’ level of self. The original instrument has beenproven reliable and valid in studies with undergraduates of North American andinternational origin (e.g., Nario-Redmond et al. 2004). A one-factor confirmatory factoranalysis (CFA) provided an acceptable model fit for the data after removingone item due to its poor loading: χ2=23.19, χ2/df=2.11, TLI=0.96, CFI=0.98,IFI=0.98, RMSEA=0.078, SRMR=0.029. The internal consistency reliabilitywas satisfactory (α=0.89).

Curr Psychol

The Chinese version of revised Life Orientation Test by Lai and others (1998) wasalso administered to assess participants’ level of optimism in general. The scale wasadministered with a 5-point Likert scale ranging from 1 (completely disagree) to 5(completely agree). Sample items were: BI’m always optimistic about my future.^ andBOverall, I expect more good things to happen to me than bad.^ A total score wascomputed by adding ratings on the positively worded items to reversed ratings on thenegatively worded items. Previous studies indicated that this measure is reliable andvalid in the Chinese-speaking population (e.g., Lai et al. 1998; Lai and Yue 2000; Laiand Wong 1998). A one-factor CFA provided a great model fit for the data afterremoving one item due to its poor loading: χ2=0.80, χ2/df=0.27, TLI=1.00, CFI=1.00, IFI=1.00, RMSEA=0.000, SRMR=0.011. The internal consistency reliabilitywas acceptable (α=0.67).

Results

The normality of all the items was first examined. Four cases were removed due to highkurtosis (larger than 7, West et al. 1995) for GSE. Together with another six cases withincomplete answers, the final sample size was 185. Izquierdo, Olea, and José Abad(2014) argued that in studies seeking evidence about the internal structure of a scale, itis recommended to carry out a preliminary analysis of the metric quality of the itemswith exploratory factor analyses (EFA). A principal components factor analysis usingvarimax rotation was then performed. The factor loadings presented in Table 1 revealedthat the items loaded clearly on two separate factors, which accounted for 54.49 % ofthe overall variability in scores. Two items (1 and 7) had loadings on another factor thatwere somewhat high; however, when Cronbach’s alpha coefficients were examined forthese items within their corresponding subscale, the coefficients decreased by 0.03 and0.07 for the subscale with the items removed. Thus, I decided to retain the items forCFA.

To further understand the generalizability of the factors extracted, the total samplewas split randomly into two subsamples. Identical EFA procedures were carried out toaccess the stability of the factor structure of the scale. Results showed that the factorsextracted from the two subsamples resembled those extracted from the total sample,with acceptable internal consistency coefficients in each subsample (see Table 1).

Next, a series of CFAwere conducted to test the validity of different factor structuresof GSE. First, a unidimensional CFA mode was investigated wherein all 10 items wereexplained by a single first-order factor. This model reflected the intended dimensionalstructure of the construct (Model 1). Second, a two-factor CFA model was estimated inwhich all cross-loadings were fixed to the value zero, with two correlated factors(Model 2). Third, a model with two correlated factors (i.e., coping self-efficacy andaction self-efficacy) was tested that included a priori target loadings of items on onefactor, while allowing for cross-loadings on the other factor (Model 3). Fourth, asWiesner and Schanding (2013) noted, bi-factor models might be better suited forinvestigating the factor structure of scales that are composed of indicators with cross-loadings. In a bi-factor model, a general factor is posited to account for the common-ality of all manifest variables, with several orthogonal specific factors representing thehypothesized unique influence of the specific factors on subsets of the manifest

Curr Psychol

Tab

le1

GSE

Scaleitemsandcorrespondingfactor

loadings

afterEFA

Item

(originalitem

number)

EFA

(totalsample)

EFA

(sub-sam

ple1)

EFA

(sub-sam

ple2)

Actionself-

efficacy

Copingself-

efficacy

Actionself-

efficacy

Copingself-

efficacy

Actionself-

efficacy

Copingself-

efficacy

Itiseasy

formeto

stickto

myaimsandaccomplishmygoals.(3)

0.79

0.06

0.62

0.23

0.85

0.03

Thanksto

myresourcefulness,Iknow

how

tohandleunforeseen

situations.(5)

0.70

0.32

0.72

0.15

0.66

0.48

Nomatterwhatcomes

myway,I’m

usually

ableto

handleit.

(10)

0.69

0.24

0.63

0.26

0.73

0.28

Iam

confidentthatIcoulddealefficiently

with

unexpected

events.(4)

0.68

0.34

0.80

0.13

0.58

0.52

Ifsomeone

opposesme,Icanfind

means

andwaysto

getwhatIwant.(2)

0.63

0.26

0.54

0.34

0.68

0.24

IcansolvemostproblemsifIinvestthenecessaryeffort.(6)

0.14

0.78

0.25

0.68

0.07

0.81

IfIam

inabind,I

canusually

thinkof

something

todo.(9)

0.10

0.67

-0.03

0.77

0.23

0.57

Icanremaincalm

whenfacing

difficultiesbecauseIcanrely

onmycoping

abilities.(7)

0.39

0.66

0.43

0.65

0.27

0.75

WhenIam

confronted

with

aproblem,I

canusually

find

severalsolutions.(8)

0.30

0.65

0.40

0.59

0.20

0.69

Icanalwaysmanageto

solvedifficultproblemsifItryhard

enough.(1)

0.36

0.58

0.36

0.61

0.35

0.55

Varianceexplained

43.87%

10.62%

42.09%

10.33%

46.00%

11.31%

Cronbach’salphacoefficient

0.80

0.77

0.76

0.77

0.83

0.77

Curr Psychol

variables over and above the general factor. Bi-factor models have been used inresearch on intelligence (e.g., Watkins and Beaujean 2014), psychopathy constructs(e.g., Kimonis et al. 2008), and psychological well-being (e.g., Chen et al. 2013), butthey have yet been applied in investigating self-efficacy. Therefore, a bi-factor modelwas estimated in this study wherein there was a general factor of global self-efficacy(on which all 10 items loaded) and two specific factors of coping self-efficacy andaction self-efficacy (on which only their respective subsets of items loaded), with allfactors being specified as uncorrelated with each other (Model 4). The factor structuresfor these models are illustrated in Fig. 1.

As shown in Table 2, the one-factor model (Model 1) exhibited a poor overall fit(χ2=99.90, χ2/df=2.85). The two-factor model without cross-loadings (Model 2), onthe other hand, showed a great improvement in goodness-of-fit (χ2=67.57, χ2/df=1.99). The two-factor model with cross-loadings (Model 3) revealed a similar level ofmodel fit (χ2=65.60, χ2/df=2.05). Further, The bi-factor model (Model 4) showed amuch better model fit compared to the previous models (χ2=31.96, χ2/df=1.28).However, it is important to note that three of the five factor loadings for the actionself-efficacy factor were not significant, and, contrary to prediction, two of the fivefactor loadings were negative. In addition, the variance of the action self-efficacy factorwas nonsignificant (p=0.21), indicating that the bi-factor model could be misspecified.At this step, Model 2 seemed to be the best model among all the models.

However, a scrutiny of the correlation between the two separate factors (r=0.81)resulted in the concern of multicollinearity in Model 2, which also pointed to thepossibility that these two factors could stem from a global factor. As such, a second-order factor model was examined to determine whether GSE as a global self-efficacyconcept could be drawn from the two first-order factors. Model 5 is the standardsecond-order model (Jöreskog 1970; Rindskopf and Rose 1988), which assumes that

Fig. 1 Factor structures for Models 1 to 6

Curr Psychol

the two first-order factors (i.e., coping self-efficacy and action self-efficacy) account forthe variance in the observed variables and that a second-order factor (i.e., global self-efficacy) accounts for the covariance in the first-order factors. Based on the literature(Marsh and Hocevar 1985; Arnau and Thompson 2000), a second-order model struc-ture can never produce a model fit that is Bbetter^ (i.e., better fitting in terms of themodel fit indices) than a first-order model with correlated factors (such as Model 2).However, a second-order model that rivals the performance of a first-order correlatedmodel can be an attractive alternative (Koufteros, Babbar, and Kaighobadi 2009). Forexample, Model 5 also fit well and was not divergent in performance from Model 2.Given the deficiency (e.g., issues of multicollinearity) of Model 2, Model 5 appeared tobe the most prudent choice. Modification indices showed adding one more correlatederror would further improve the model fit: (χ2=48.45, χ2/df=1.47). This model wasthus retained for further consideration as Model 6, and was selected to be the finalmodel. Factor loadings in Model 6 are presented in Fig. 2.

Finally, in order to examine whether the two first-order factors showed differentpatterns of the relationship with other variables, multiple regression analyses were runwith GSE scores as predictors and participants’ scores on their personal identity andoptimism as outcome variables. Results showed that coping self-efficacy factor pre-dicted 16.1 % of the variance in personal identity (β=0.48, p<.01), whereasaction self-efficacy factor was not related to one’s identity perception (β=−0.06,p>.10). In contrast, action self-efficacy factor predicted 5.4 % of the variancein optimism (β=0.18, p<.05), whereas coping self-efficacy factor was notrelevant (β=0.07, p>.10). This provided further evidence of the differentiationbetween these two factors.

Discussion

The purpose of this study was to examine the factor structure of the GSE scale(Schwarzer and Jerusalem 1995).

Table 2 Comparison of GSE models

Model χ2 χ2 /df RMSEA SRMR CFI IFI TLI AIC BCC

Model 1 One-factor model 99.90 2.85 0.100 0.062 0.89 0.89 0.86 159.90 163.72

Model 2 Two-factor modelwithout cross-loading

67.57 1.99 0.073 0.051 0.94 0.94 0.92 129.57 133.52

Model 3 Two-factor model withcross-loading (Item1 and 7)

65.60 2.05 0.076 0.049 0.94 0.94 0.92 131.60 135.80

Model 4 Bi-factor model 31.96 1.28 0.039 0.034 0.99 0.99 0.98 111.96 117.05

Model 5 Second-order factormodel

67.57 1.99 0.073 0.051 0.94 0.94 0.92 129.57 133.52

Model 6 Second-order factormodel with onecorrelated error

48.45 1.47 0.050 0.044 0.97 0.97 0.96 112.45 116.52

Curr Psychol

Several latent measurement modeling approaches were adopted to gain a betterunderstanding of their strengths and potential boundaries in capturing the underlyingfactor structure of GSE. The findings from the current study provided strong evidencethat a unidimensional measurement model did not adequately capture the underlyingfactor structure of GSE; rather, a multidimensional measurement model provided amore reasonable approximation to the data for the GSE scale. This finding is importantbecause it lends support to Chiu and Tsang’s (2004) findings about the dimensionalityof this scale.

The results revealed that the Chinese version of GSE scale appeared to have a robustsecond–order factor structure that made decent conceptual senses. The second-orderfactor model with one correlated error was chosen to be the final model for boththeoretical and statistical considerations. From the theoretical point of view, the second-order model suggests that there is a logical higher order structure underlying the first-order factors. An individual’s self-efficacy (second-order level of abstraction) can beconceptualized as a composite of two dimensions such as action and coping self-efficacy (first-order latent variables). Within this second-order model, the contributionof each dimension to the higher-level construct can be assessed and delineated ascompared to aggregating all items together in a single composite score. This second–order factor model was further in accordance with Schwarzer and Renner’s (2000) two-phase model of self-efficacy – one factor paralleled with action self-efficacy, whereasthe other factor paralleled coping self-efficacy. Factor analysis results also offeredempirical support for the distinction of these two factors, which represented thetheoretical conception. From the statistical perspective, the second–order factor modelyielded one of the best model fits compared to other alternative models, with decentfactor loadings on each factor. Moreover, the high correlation between the two factorsas shown in Model 2 was worrisome because the composite score of two first-orderfactors would be quite difficult to meaningfully be described and used in subsequenthypothesis testing. Second-order models recognize the contribution and retain theidiosyncratic nature of each first-order construct, and treat such constructs as different

Fig. 2 Factor loadings of the second-order factor model with one correlated error (Model 6)

Curr Psychol

facets of the higher-order construct. Altogether, the results suggested again that GSE bea multidimensional construct.

The results shed some light on previous inconsistent findings. In a recent study byTeo and Kam (2015), who assessed the measurement invariance of the GSE scaleacross two cultures, only partial scalar invariance was achieved because the interceptsof four out of ten items were non-invariant. The authors discussed about the possiblepresence of response bias in cross-cultural mean comparison with this scale, that is,Asians have shown distinct response styles from westerners, such as frequent use ofmoderate response anchors and higher acquiescence response styles (Chen 2008).These response styles could have created unpredictable effects that hinder scalarinvariance. Another possible reason, however, could be that the factorial structure ofthe scale did not hold the same across cultures. In particular, the action self-efficacyfactor as discovered in the current study covered three items out of four that wereidentified as Bproblematic^ in the scalar invariance examination in Teo and Kam’sstudy. This factor emphasized the perception of outcome (e.g., failure) or one’sperceived capacity of controlling over difficulties, which could approximate someaspects of the cultural dimension in a Chinese culture. Heine and Lehman (1995) foundsignificant differences in students’ perceptions of future between Eastern and Westernsamples. Compared to their Western counterparts, the Eastern young people tended toview the future as about average or even worse than average. This cultural variationcould have affected scores on this factor, which could partially explain the partition ofthis factor in the current sample.

There are multiple aspects that need further evaluation. First, according toMundfrom and others’ (2005) recommendation for a minimum sample size of 140,with a moderate level of communalities (ranging from 0.26 to 0.54) and a variable-to-factor ratio of 5, although a sample size of 195 was expected to be adequate for factoranalyses, the subsample size for the EFA was rather limited. Future testing will benecessary to involve a larger and more demographically diverse sample to allow for thedevelopment of a more parsimonious instrument that could be used with multiplepopulations. Second, the current findings greatly differed from Chiu and Tsang’s(2004) two-factor model in GSE. In their model, the personal-environmental distinctionwas not replicated in the current normal sample. It is possible that other measurementmodels can represent the data equally well or even surpass the performance of theirtwo-factor configurations. In the absence of testing for alternative measurementmodels, it is hazardous to suggest that any specification is the most suitable. Last, thepresent analysis was restricted to samples of university students to assure somehomogeneity of the populations under study. This presented problems when the resultswere to be generalized to other age groups. Emerging adults’ life experiences could bequite different from older adults. Therefore, current findings need to be re-examinedwith an older population, who might show a different factor structure of the construct.

As our theories and understanding evolve, it is incumbent upon current and futureresearchers to continue to assess the validity of the measures in use. The current findingevidenced the two-factor structure of GSE in a Chinese sample both on empirical aswell as conceptual grounds. Although the two factors of GSE were found to be distinctfrom each other, it is important that further research should examine the factorialinvariance of GSE across different ethnic groups as well as different age groups, anddirectly test the relationship between the two factors and how they function differently

Curr Psychol

in explaining other constructs that were examined by taking GSE as a single-factorvariable, such as resilience (Schwarzer and Warner 2013), perceived stress and person-ality (Ebstrup et al. 2011).

This research investigation lays a foundation for extending inquiry into the nature ofself-efficacy and provides a different way of examining GSE in the second-order factormodel. Refinements in self-efficacy research have been debating the nature of appro-priate measurement, such as the appropriate degree of specificity in conceptualizationand operationalization of this construct (Maurer 2001). If a researcher is interested inhow self-efficacy influences English learning, feelings of self-efficacy in generallanguage learning, or even at a broader level, efficacy in most life scenarios, mightbe necessary as a foundation to understanding the origin of specific self-efficacy. But afocus on more task-specific efficacy beliefs, such as learning specific software orsolving a given math problem, would be more relevant when we attempt to use it toexplain or predict other task-related variables.

Appendix

References

Arnau, R. C., & Thompson, B. (2000). Second-order confirmatory factor analysis of the WAIS-III.Assessment, 7(3), 237–246.

Ausbrooks, E. P., Thomas, S. P., & Williams, R. L. (1995). Relationships among self-efficacy, optimism, traitanger, and anger expression. Health Values: The Journal of Health Behavior, Education & Promotion,19(4), 46–54.

Table 3 Items for general self-efficacy scale

1 I can always manage to solve difficult problems if I try hard enough.如果我尽力去做的话,我总是能够解决问题的。

2 If someone opposes me, I can find the means and ways to get what I want.即使别人反对我,我仍有办法取得我所想要的。

3 It is easy for me to stick to my aims and accomplish my goals.对我来说,坚持理想和大成目标是轻而易举的。

4 I am confident that I could deal efficiently with unexpected events.我相信能有效地应付任何突如其来的事情。

5 Thanks to my resourcefulness, I know how to handle unforeseen situations.以我的才智,我定能应付意料之外的情况。

6 I can solve most problems if I invest the necessary effort.如果我付出必要的努力,我一定能解决大多数的难题。

7 I can remain calm when facing difficulties because I can rely on my coping abilities.我能冷静地面对困难,因为我可信赖自己处理问题的能力。

8 When I am confronted with a problem, I can usually find several solutions.面对一个难题时,我通常能找到几个解决的方法。

9 If I am in trouble, I can usually think of a solution.有麻烦的时候,我通常能想到一些应付的方法。

10 I can usually handle whatever comes my way.无论什么事在我身上发生,我都能够应付自如。

Curr Psychol

Bandura, A. (1997). Self-efficacy: The exercise of control. New York: Freeman.Bandura, A. (1999). A social cognitive theory of personality. In L. Pervin & O. John (Eds.), Handbook of

personality (2nd ed., pp. 154–196). New York: Guilford Press.Bandura, A. (2006). Guide for constructing self-efficacy scales. In F. Pajares & T. Urdan (Eds.), Self-efficacy

beliefs of adolescents (pp. 307–336). CT: Information Age.Barlow, J. H., Williams, B., & Wright, C. (1996). The generalized self‐efficacy scale in people with arthritis.

Arthritis and Rheumatism, 9(3), 189–196.Berzonsky, M. D. (1992). Identity style and coping strategies. Journal of Personality, 60(4), 771–788.Bonsaksen, T., Lerdal, A., & Fagermoen, M. S. (2012). Factors associated with self-efficacy in persons with

chronic illness. Scandinavian Journal of Psychology, 53(4), 333–339.Bosscher, R. J., & Smit, J. H. (1998). Confirmatory factor analysis of the general self-efficacy scale. Behaviour

Research and Therapy, 36(3), 339–343.Brislin, R. W. (1970). Back-translation for cross-cultural research. Journal of Cross-Cultural Psychology, 1(3),

185–216.Chemers, M. M., Hu, L. T., & Garcia, B. F. (2001). Academic self-efficacy and first year college student

performance and adjustment. Journal of Educational Psychology, 93(1), 55.Chen, F. F. (2008). What happens if we compare chopsticks with forks? The impact of making inappropriate

comparisons in cross-cultural research. Journal of Personality and Social Psychology, 95, 1005–1018.Chen, G., Gully, S. M., & Eden, D. (2001). Validation of a new general self-efficacy scale. Organizational

Research Methods, 4(1), 62–83.Chen, G., Gully, S. M., & Eden, D. (2004). General self‐efficacy and self‐esteem: Toward theoretical and

empirical distinction between correlated self‐evaluations. Journal of organizational Behavior, 25, 375–395.

Chen, F. F., Jing, Y., Hayes, A., & Lee, J. M. (2013). Two concepts or two approaches? A bifactor analysis ofpsychological and subjective well-being. Journal of Happiness Studies, 14(3), 1033–1068.

Cheung, S. K., & Sun, S. Y. (1999). Assessment of optimistic self-beliefs: further validation of the Chineseversion of the General Self-Efficacy Scale. Psychological Reports, 85, 1221–1224.

Chiu, F. P., & Tsang, H. W. (2004). Validation of the Chinese general self-efficacy scale amongindividuals with schizophrenia in Hong Kong. International Journal of Rehabilitation Research,27(2), 159–161.

Cramm, J. M., Strating, M. M., Roebroeck, M. E., & Nieboer, A. P. (2013). The importance of general self-efficacy for the quality of life of adolescents with chronic conditions. Social Indicators Research, 113(1),551–561.

Cronbach, L. J. (1956). Assessment of individual differences. Annual Review of Psychology, 7, 173–196.Ebstrup, J. F., Eplov, L. F., Pisinger, C., & Jørgensen, T. (2011). Association between the five factor

personality traits and perceived stress: is the effect mediated by general self-efficacy? Anxiety, Stressand Coping, 24(4), 407–419.

Eden, D. (1988). Pygmalion, goal setting, and expectancy: compatible ways to raise productivity. Academy ofManagement Review, 13, 639–652.

Gallagher, M. W., Payne, L. A., White, K. S., Shear, K. M., Woods, S. W., Gorman, J. M., & Barlow, D. H.(2013). Mechanisms of change in cognitive behavioral therapy for panic disorder: the unique effects ofself-efficacy and anxiety sensitivity. Behaviour Research and Therapy, 51(11), 767–777.

Gardner, D. G., & Pierce, J. L. (1998). Self-esteem and self-efficacy within the organizational context. Groupand Organizational Management, 23, 48–70.

Hall, G. S. M. (2000). Self-efficacy at work: a quality analysis of personal and environmental factorscontributing to personal efficiency during job-task change. Dissertation Abstracts International A(Humanities and Social Sciences), 61, 916.

Heine, S. J., & Lehman, D. R. (1995). Cultural variation in unrealistic optimism: does the west feel morevulnerable than the east? Journal of Personality and Social Psychology, 68(4), 595–607.

Izquierdo, I., Olea, J., & José Abad, F. (2014). Exploratory factor analysis in validation studies: uses andrecommendations. Psicothema, 26(3), 395–400.

Jerome, G. J., & McAuley, E. (2013). Enrollment and participation in a pilot walking programme: The role ofself-efficacy. Journal of Health Psychology, 18(2), 236–244.

Jöreskog, K. G. (1970). A general method for analysis of covariance structures. Biometrika, 57(2), 239–251.Kimonis, E. R., Frick, P. J., Skeem, J. L., Marsee, M. A., Cruise, K., Munoz, L. C., Aucoin, K. J., &Morris, A.

S. (2008). Assessing callous–unemotional traits in adolescent offenders: Validation of the inventory ofcallous–unemotional traits. International Journal of Law and Psychiatry, 31, 241–252.

Klassen, R. M. (2004). Optimism and realism: a review of self‐efficacy from a cross‐cultural perspective.International Journal of Psychology, 39(3), 205–230.

Curr Psychol

Koufteros, X., Babbar, S., & Kaighobadi, M. (2009). A paradigm for examining second-order factor modelsemploying structural equation modeling. International Journal of Production Economics,120(2), 633–652.

Lai, J. C. L., & Wong, W. S. (1998). Optimism and coping with unemployment among Hong Kong Chinesewomen. Journal of Research in Personality, 32, 454–479.

Lai, C. L., & Yue, X. D. (2000). Measuring optimism in Hong Kong and mainland Chinese with the revisedlife orientation test. Personality and Individual Differences, 28, 781–796.

Lai, J. C. L., Cheung, H., Lee, W. M., & Yu, H. (1998). The utility of the revised life orientation test tomeasure optimism among Hong Kong Chinese. International Journal of Psychology, 33, 45–56.

Leganger, A., Kraft, P., & Roysamb, E. (2000). Perceived self-efficacy in health behaviour research:Conceptualisation, measurement and correlates. Psychology and Health, 15(1), 51–69.

Leung, D. P., & Leung, A. M. (2011). Factor structure and gender invariance of the Chinese general self-efficacy scale among soon-to-be-aged adults. Journal of Advanced Nursing, 67(6), 1383–1392.

Löve, J., Moore, C. D., & Hensing, G. (2012). Validation of the Swedish translation of the general self-efficacy scale. Quality of Life Research, 21(7), 1249–1253.

Luszczynska, A., Gutierrez-Dona, B., & Schwarzer, R. (2005). General self-efficacy in variousdomains of human functioning: Evidence from five countries. International Journal ofPsychology, 40(2), 80–89.

Marsh, H. W., & Hocevar, D. (1985). Application of confirmatory factor analysis of the study of self-concept:first and higher order factor models and their invariance across groups. Psychological Bulletin, 97(3),562–582.

Martocchio, J. J., & Judge, T. A. (1997). Relationships between conscientiousness and learning in employeetraining: mediating influences of self-deception and self-efficacy. Journal of Applied Psychology, 82,764–773.

Maurer, T. J. (2001). Career-relevant learning and development, worker age, and beliefs about self-efficacy fordevelopment. Journal of Management, 27, 123–140.

Mundfrom, D. J., Shaw, D. G., & Tian, L. K. (2005). Minimum sample size recommendations for conductingfactor analyses. International Journal of Testing, 5, 159–168.

Nario-Redmond, M. R., Biernat, M., Eidelman, S., & Palenske, D. J. (2004). The social and personal identitiesscale: a measure of the differential importance ascribed to social and personal self-categorizations. Selfand Identity, 3(2), 143–175.

Nario-Redmond, M. R., Noel, J. G., & Fern, E. (2013). Redefining disability, re-imagining the self:disability identification predicts self-esteem and strategic responses to stigma. Self and Identity,12(5), 468–488.

Parker, P. D., Marsh, H. W., Ciarrochi, J., Marshall, S., & Abduljabbar, A. S. (2014). Juxtaposing math self-efficacy and self-concept as predictors of long-term achievement outcomes. Educational Psychology,34(1), 29–48.

Posadzki, P., Stockl, A., Musonda, P., & Tsouroufli, M. (2010). A mixed-method approach to sense ofcoherence, health behaviors, self-efficacy and optimism: towards the operationalization of positive healthattitudes. Scandinavian Journal of Psychology, 51(3), 246–252.

Rand, K. L. (2009). Hope and optimism: latent structures and influences on grade expectancy and academicperformance. Journal of Personality, 77(1), 231–260.

Rindskopf, D., & Rose, T. (1988). Some theory and applications of confirmatory second-order factor analysis.Multivariate Behavioral Research, 23, 51–67.

Rottmann, N., Dalton, S. O., Christensen, J., Frederiksen, K., & Johansen, C. (2010). Self-efficacy, adjustmentstyle and well-being in breast cancer patients: A longitudinal study. Quality of Life Research, 19(6), 827–836.

Scheier, M. F., & Carver, C. S. (1992). Effects of optimism on psychological and physical well-being:theoretical overview and empirical update. Cognitive Therapy and Research, 16, 201–228.

Scholz, U., Doña, B. G., Sud, S., & Schwarzer, R. (2002). Is general self-efficacy a universal construct?Psychometric findings from 25 countries. European Journal of Psychological Assessment, 18(3), 242–251.

Schwarzer, R. (Ed.). (1992). Self-efficacy: Thought control of action. Washington: Hemisphere.Schwarzer, R. (1993). Measurement of perceived self-efficacy: Psychometric scales for cross-cultural re-

search. Berlin: Freien Universität Berlin.Schwarzer, R., & Fuchs, R. (1996). Self-efficacy and health behaviors. In M. Conner and P. Norman (Eds.),

Predicting health behavior: Research and practice with social cognition models (pp. 163–196). Buckingham, England: Open University Press.

Curr Psychol

Schwarzer, R., & Jerusalem, M. (1995). Generalized self-efficacy scale. Measures in health psychology: auser’s portfolio. Causal and Control Beliefs, 1, 35–37.

Schwarzer, R., & Renner, B. (2000). Social-cognitive predictors of health behavior: action self-efficacy andcoping self-efficacy. Health Psychology, 19(5), 487.

Schwarzer, R., & Warner, L. M. (2013). Perceived self-efficacy and its relationship to resilience. In S. Prince-Embury & D. H. Saklofske (Eds.), Resilience in children, adolescents, and adults: Translating researchinto practice (pp. 139–150). New York: Springer.

Schwarzer, R., Bäßler, J., Kwiatek, P., Schröder, K., & Zhang, J. X. (1997a). The assessment of optimistic self‐beliefs: comparison of the German, Spanish, and Chinese versions of the general self‐efficacy scale.Applied Psychology, 46(1), 69–88.

Schwarzer, R., Born, A., Iwawaki, S., & Lee, Y. M. (1997b). The assessment of optimistic self-beliefs:comparison of the Chinese, Indonesian, Japanese, and Korean versions of the general self-efficacy scale.Psychologia: An International Journal of Psychology in the Orient, 40, 1–13.

Snyder, C. R., Harris, C., Anderson, J. R., Holleran, S. A., Irving, L. M., Sigmon, S. T., ... & Harney, P. (1991).The will and the ways: Development and validation of an individual-differences measure of hope. Journalof Personality and Social Psychology, 60, 570–585.

Sperber, N., Hall, K. S., Allen, K., Devellis, B. M., Lewis, M., & Callahan, L. F. (2014). The role of symptomsand self-efficacy in predicting physical activity change among older adults with arthritis. Journal ofPhysical Activity & Health, 11, 528–535.

Stajkovic, A. D., & Luthans, F. (1998). Self-efficacy and work-related performance: a meta-analysis.Psychological Bulletin, 124, 240–261.

Strathman, A., Gleicher, F., Boninger, D. S., & Edwards, C. S. (1994). The consideration of future conse-quences: weighing immediate and distant outcomes of behavior. Journal of Personality and SocialPsychology, 66, 747–752.

Strobel, M., Tumasjan, A., & Spörrle, M. (2011). Be yourself, believe in yourself, and be happy: self-efficacyas a mediator between personality factors and subjective well-being. Scandinavian Journal of Psychology,52(1), 43–48.

Teo, T., & Kam, C. (2015). A measurement invariance analysis of the general self-efficacy scale on twodifferent cultures. Journal of Psychoeducational Assessment.

van der Slot, W. M., Nieuwenhuijsen, C., van den Berg-Emons, R. J., Wensink-Boonstra, A. E., Stam, H. J., &Roebroeck, M. E. (2010). Participation and health-related quality of life in adults with spastic bilateralcerebral palsy and the role of self-efficacy. Journal of Rehabilitation Medicine, 42, 528–535.

Watkins, M. W., & Beaujean, A. A. (2014). Bifactor structure of the Wechsler preschool and primary scale ofintelligence—fourth edition. School Psychology Quarterly, 29(1), 52–63.

Wells, J. R., & Anderson, S. T. (2011). Self-efficacy and social support in African Americans diagnosed withend stage renal disease. The ABNF Journal: Official Journal of the Association of Black Nursing Facultyin Higher Education, 22(1), 9–12.

West, S. G., Finch, J. F., & Curran, P. J. (1995). Structural equation models with nonnormal variables. In R. H.Hoyle (Ed.), Structural equation modeling: Concepts, issues, and applications (pp. 56–75). ThousandOaks: Sage Publications.

Wiesner, M., & Schanding, G. T. (2013). Exploratory structural equation modeling, bifactor models, andstandard confirmatory factor analysis models: application to the BASC-2 behavioral and emotionalscreening system teacher form. Journal of School Psychology, 51(6), 751–763.

Wood, R., & Bandura, A. (1989). Impact of conceptions of ability on self-regulatory mechanisms and complexdecision making. Journal of Personality and Social Psychology, 56, 407–415.

Zhang, J. X., & Schwarzer, R. (1995). Measuring optimistic self-beliefs: a Chinese adaptation of the generalself-efficacy scale. Psychologia: An International Journal of Psychology in the Orient, 38, 174–181.

Zimmerman, B. J., & Cleary, T. J. (2006). Adolescents’ development of personal agency: The role of self-efficacy beliefs and self-regulatory skills. In F. Pajares & T. Urdan (Eds.), Self-efficacy beliefs ofadolescents (pp. 45–69). Greenwich: Information Age Publishing.

Curr Psychol