16

Click here to load reader

Journal of Career Assessment-2005-Hampton-98-113.pdf

Embed Size (px)

Citation preview

Page 1: Journal of Career Assessment-2005-Hampton-98-113.pdf

7/23/2019 Journal of Career Assessment-2005-Hampton-98-113.pdf

http://slidepdf.com/reader/full/journal-of-career-assessment-2005-hampton-98-113pdf 1/16

Testing for the Structure of the Career

Decision Self-Efficacy Scale–Short Form Among Chinese College Students

Nan Zhang HamptonUniversity of Massachusetts–Boston

This study examines the factor structure of the Career Decision Self-Efficacy

Scale–Short Form (CDSES-SF) among Chinese college students. Two samples of college students from China were used. The original 25-item CDSES-SF was notsupported by the data derived from a sample of 256 Chinese college students(Sample 1). However, a modified 13-item, three-factor model of the CDSES-SFfit the data well: The standardized factor loadings and goodness-of-fit indexes wereadequate and the reliability coefficient was .85. This 13-item model was subse-quently cross-validated with a second sample of Chinese college students (N =157). The cross-validation test revealed that the model was invariant across the twosamples. It is suggested that this revised CDSES-SF may be applied to under-standing Chinese college students’ career decision-making behavior.

Keywords: Career Decision Self-Efficacy Scale–Short Form, confirmatoryfactor analysis, construct validity, Chinese

The Career Decision Self-Efficacy Scale (CDSES) (Taylor & Betz, 1983) isone of the most used instruments in the career development/counseling litera-ture in the United States (Luzzo, 1993; O’Brien, 2003; Solberg et al., 1994). TheCDSES is based on Bandura’s self-efficacy theory, which postulates that individ-uals’ beliefs about their ability to successfully perform a given task are the majormediators of behavior and behavior change (Bandura, 1977, 1986, 1997). In aseminal theory-development paper, Hackett and Betz (1981) proposed that self-

efficacy theory could be applied to understanding vocational behavior. In linewith Bandura’s self-efficacy theory and Hackett and Betz’s (1981) suggestion,Taylor and Betz (1983) developed the CDSES, which measured individuals’confidence in their ability to complete tasks related to making career decisions.

The original CDSES consists of 50 items (tasks) that cover five domains of career choice competence proposed by Crites (1961). These five domains (sub-

Correspondence concerning this article should be address to Nan Zhang Hampton, Graduate College of Education, 100 Morrissey Blvd., Boston, MA 02125; e-mail: [email protected].

JOURNAL OF CAREER ASSESSMENT, Vol. 13 No. 1, February 2005 98–113DOI: 10.1177/1069072704270298© 2005 Sage Publications

98

 at Sunan Kalijaga Yogyakarta PARE on December 16, 2015 jca.sagepub.comDownloaded from 

Page 2: Journal of Career Assessment-2005-Hampton-98-113.pdf

7/23/2019 Journal of Career Assessment-2005-Hampton-98-113.pdf

http://slidepdf.com/reader/full/journal-of-career-assessment-2005-hampton-98-113pdf 2/16

Hampton / CAREER DECISION SELF-EFFICACY SCALE 99

scales) include Self-Appraisal (SA), Occupational Information (OI), Goal Setting(GS), Planning (PL), and Problem Solving (PS). Respondents are asked to rate

their confidence about performing each task on a 10-point Likert-type scale rang-ing from no confidence (0) to complete confidence (9). The scores for each sub-scale (10 items) range from 0 to 90, and the total score for the whole scale (50items) is the sum of the scores of the five subscales that range from 0 to 450.Higher scores correspond with higher self-efficacy expectations with regard tocareer decision-making tasks.

Because the original CDSES is a relatively long questionnaire, a short form of the CDSES (CDSES-SF) was developed by Betz, Klein, and Taylor in 1996. TheCDSES-SF contains 25 items that were taken from the original CDSES. Likethe CDSES, the CDSES-SF has five subscales that measure the same five

domains (Betz et al., 1996). Also, the CDSES-SF is scored according to a 10-point Likert-type scale (scores range from 1 to 10 instead of 0 to 9).

Multiple studies indicate that both the CDSES and the CDSES-SF are reli-able instruments when used with American college students. For example,Taylor and Betz (1983) reported that the alpha reliability coefficients of theCDSES were .97 for the whole scale, .88 for the SA subscale, .89 for the OI sub-scale, .87 for the GS subscale, .89 for the PL subscale, and .86 for the PS sub-scale. Luzzo (1993) indicated that the test-retest reliability of the CDSES was.83. Additionally, Nilsson, Schmidt, and Meek (2002) conducted a reliability

generalization study on the CDSES. They reviewed 41 published journal articlesand seven dissertations that reported the use of the CDSES/CDSES-SF. Theyfound that the alpha reliability coefficients of the CDSES/CDSES-SF rangedfrom .83 to .97 (Nilsson et al., 2002).

Similarly, Betz et al. (1996) reported that the alpha reliability coefficients of the CDSES-SF were .94 (total scale), .73 (SA subscale), .78 (OI subscale), .83(GS subscale), .81 (PL subscale), and .75 (PS subscale). Other researchers indi-cated that the internal consistency reliability coefficients of the total scale of theCDSES-SF ranged from .92 to .97 (Chung, 2002; Osipow & Gati, 1998).

Consistent findings regarding the concurrent validity of the CDSES and the

CDSES-SF were also reported in the literature (Bergeron & Romano, 1994; Betz& Hackett, 1986; Robbins, 1985; Taylor & Betz, 1983; Taylor & Popma, 1990).For example, the CDSES was moderately correlated with career indecision (r =–.45), career certainty (r = .58), career maturity (r = .41), and motivation (r = .42)(Betz et al., 1996; Luzzo, 1993; Niles & Sowa, 1992). Likewise, the CDSES-SFwas found to be moderately correlated with career indecision (r  = –.56) andcareer commitment (r = .45) (Betz et al., 1996; Chung, 2002).

Despite the extensive research on the CDSES and CDSES-SF, researcherscalled for further investigations of the reliability and validity of these scales (Betz

& Luzzo, 1996; Luzzo, 1996; Robbins, 1985). They stated that several problemswere associated with the study of career decision-making self-efficacy. First, con-sistent reporting of the five-factor structure of the CDSES as hypothesized by theauthors of the CDSES was lacking in the literature. Some studies indicated that

 at Sunan Kalijaga Yogyakarta PARE on December 16, 2015 jca.sagepub.comDownloaded from 

Page 3: Journal of Career Assessment-2005-Hampton-98-113.pdf

7/23/2019 Journal of Career Assessment-2005-Hampton-98-113.pdf

http://slidepdf.com/reader/full/journal-of-career-assessment-2005-hampton-98-113pdf 3/16

100 JOURNAL OF CAREER ASSESSMENT / February 2005

the CDSES measured a general factor (Robbins, 1985; Taylor & Betz, 1983;Taylor & Popma, 1990). Others reported that there were five factors underlying

a Hebrew short form version of the CDSES and these five factors were consistentwith the hypothesized five factors in the original CDSES, although the samestudy found that the 50-item Hebrew version of the CDSES full scale was incon-sistent with the five-factor structure (Gati, Osipow, & Fassa, 1994). These incon-sistent findings warrant further explorations of the structure of the CDSES.Second, the majority of studies on the CDSES were conducted with college stu-dents in the United States. There is a strong need to extend the validity of theconstruct of career decision self-efficacy to other populations, especially non-Caucasian populations (Betz et al., 1996; Hackett & Betz, 1995; Lent & Hackett,1987; O’Brien, 2003).

To further explore the component structure of the CDSES, Peterson anddelMas (1998) conducted a study with 418 underprepared students enrolled indevelopmental education. These authors administered the CDSES to the par-ticipants and analyzed the data using a principal component factor analysis, ascree test, and a discontinuity analysis. Drawing on the results of these analyses,Peterson and delMas suggested a two-factor model of the CDSES for underpre-pared college students. This two-factor model of the CDSES consists of 16 items.Factor 1 includes 10 items and was named as information gathering. Factor 2contains 6 items and was named as decision making. Eleven of the 16 items are

parallel with the CDSES-SF of Betz et al. (1996).In a study of cross-cultural equivalence of the CDSES-SF, Creed, Patton, and Watson (2002) investigated the construct validity of the CDSES-SF in 979 Australian and South African high school students. Drawing on the results of aprincipal axis factoring analysis, Creed et al. reported two 23-item three-factormodels each for Australian or South African high school students. These threefactors included information gathering, decision making, and problem solving.Two items of the original 25 items of the CDSES-SF did not load on the above-mentioned three factors. Creed et al. also pointed out that the dominant factorfor the two samples was different although the three factors for both samples were

similar. For the Australians, the dominant factor was information gathering,which contained items from the CDSES-SF occupational information and goalselection subscales. For the South Africans, the dominant factor was decisionmaking, which included items from the CDSES-SF Goal Selection andPlanning subscales. The authors suggested that the difference in the dominantfactor indicated cultural differences between the two samples, and they called forfurther investigations on cross-cultural equivalence of the CDSES-SF (Creedet al., 2002).

Responding to the call for further investigations by Creed et al. (2002), the

present study attempted to explore the utility of the CDSES-SF in Chinese col-lege students. Chinese college students were chosen because they were from aculture that had been defined as the opposite of the American culture (Triandiset al., 1986). Compared with the individual-centered American culture, which

 at Sunan Kalijaga Yogyakarta PARE on December 16, 2015 jca.sagepub.comDownloaded from 

Page 4: Journal of Career Assessment-2005-Hampton-98-113.pdf

7/23/2019 Journal of Career Assessment-2005-Hampton-98-113.pdf

http://slidepdf.com/reader/full/journal-of-career-assessment-2005-hampton-98-113pdf 4/16

Hampton / CAREER DECISION SELF-EFFICACY SCALE 101

emphasizes a sense of individual identity, goals, and values, the Chinese cultureis considered as a collective culture that focuses on group identity, decisions, and

actions (Triandis et al., 1986). Another reason for choosing Chinese students wasthat China was experiencing a transitional period marked by rapid economicgrowth and profound social changes. The substantial increase in unemploymentrate, the heated competition for jobs, and the reform of educational and employ-ment systems have affected Chinese students and changed almost all aspects of their lives. For example, the Chinese government has changed the job assign-ment system, which existed for the past 50 years, for high school and college stu-dents. Instead of waiting for the government to assign a job, Chinese studentsnow need to find their own jobs (Zhang, Hu, & Pope, 2002). These changes havecreated pressures that have not been experienced by Chinese students in the past.

The need for career and other kinds of counseling services is increasing (Zhanget al., 2002). However, there is a shortage of adequate career assessment andcounseling tools that can be used with Chinese students (France, Jin, Huang, Si,& Zhang, 1991; Zhang et al., 2002).

 Although the CDSES-SF appears to be a reliable and valid instrument whenused with American students, few published studies have applied it to investigat-ing career decision behavior among Chinese students. Mau (2002) studied cul-tural differences in career decision self-efficacy between American college stu-dents and Chinese college students from Taiwan. The results of this study

showed that the internal consistency reliability of the CDSES-SF was .92 and thetest-retest reliability was .83, suggesting that the CDSES-SF was a reliable instru-ment when used with Chinese college students from Taiwan (Mau, 2000).However, the structure of the CDSES-SF for college students in the People’sRepublic of China has not been confirmed. Accordingly, the purpose of the pres-ent study was to (a) explore and test the most appropriate measurement model of the CDSES-SF for Chinese college students and (b) cross-validate the structureof this model with an independent sample of Chinese college students.

METHOD

Participants

Two independent samples of college students from a medium city in north-eastern China were used. Sample 1 served as the calibration sample forObjective 1 of the study, in which parameter estimates and overall model fitindexes were obtained for all the models of the CDSES-SF tested in the presentstudy (Betz et al., 1996; Creed et al., 2002; Peterson & delMas, 1998). This sam-

ple consisted of 256 college students. Of them, 118 (46%) were women and 138(54%) were men. Their age ranged from 19 to 25 years (M = 21, SD = 1.45). Onehundred thirty (51%) participants were freshmen and 126 (49%) were juniors.

 at Sunan Kalijaga Yogyakarta PARE on December 16, 2015 jca.sagepub.comDownloaded from 

Page 5: Journal of Career Assessment-2005-Hampton-98-113.pdf

7/23/2019 Journal of Career Assessment-2005-Hampton-98-113.pdf

http://slidepdf.com/reader/full/journal-of-career-assessment-2005-hampton-98-113pdf 5/16

102 JOURNAL OF CAREER ASSESSMENT / February 2005

Sample 2 served as the validation sample. It contained 157 college students froma different university. Of them, 97 (61%) were women and 60 (39%) were men.

Their age ranged from 19 to 25 years (M = 21, SD = 1.65). Eighty-two (52%) par-ticipants were freshmen and 75 (48%) were juniors.

Instruments

In this study, a revised CDSES-SF was used. This instrument contained all 25items from the CDSES-SF (Betz et al., 1996) and four items from Peterson anddelMas’s (1998) 16-item short form of the CDSES. One item (“Find and use theplacement office on campus”) in Peterson and delMas’s short form of the

CDSES was excluded because there was no such service available in the partic-ipants’ schools.

The revised CDSES-SF was translated into Chinese by a bilingual Chinese American graduate student in counseling psychology and a bilingual Chinese American psychologist via back translation (English-Chinese-English). Theback-translated English version was compared with the original version for mean-ing accuracy by a native English speaker who is a graduate student in counselingpsychology, the meanings of several words were clarified, and the instrument wasreworded during this process. A Chinese version of a demographic questionnaire

was also used. It included the participant’s age, sex, and grade.

Procedure

Permission was obtained from the two participating universities in China. Tencollege students (5 men and 5 women) were interviewed using open-ended ver-sions of the revised CDSES-SF. A research assistant read each item on theCDSES-SF to the interviewees and asked them to provide opinions on the appro-priateness of the items. After the interview, minor changes (e.g., deleting oneitem) were made to the scale.

College freshmen were recruited from a mathematics class, which wasrequired for all first-year students. College juniors were recruited from a statisticsclass, which was required for all third-year students at the universities.

 A research assistant who was not the class instructor distributed the researchquestionnaires to 256 college students (Sample 1) during regularly scheduledclass meeting time. All participants completed the questionnaires in class. Theresearch assistant collected the questionnaires after participants completed them.These same procedures were repeated later at the second university with 157 stu-

dents (Sample 2).

 at Sunan Kalijaga Yogyakarta PARE on December 16, 2015 jca.sagepub.comDownloaded from 

Page 6: Journal of Career Assessment-2005-Hampton-98-113.pdf

7/23/2019 Journal of Career Assessment-2005-Hampton-98-113.pdf

http://slidepdf.com/reader/full/journal-of-career-assessment-2005-hampton-98-113pdf 6/16

Hampton / CAREER DECISION SELF-EFFICACY SCALE 103

RESULTS

Means, standard deviations, and Cronbach’s reliability coefficients were cal-culated for the CDSES-SF and subscales using SPSS 10.0. For the total 25-itemCDSES-SF scale, scores ranged from 64 to 223, with a mean score of 155.97(SD = 34.9). The mean of the subscales ranged from 27.43 to 32.39 (see the firstproportion of Table 1). The distributions of all the scores were fairly normal (theskewness and kurtosis <1.0). The reliability of the total scale of the 25-itemCDSES-SF was .91, and the reliability of the subscales ranged from .70 to.78.

Exploratory factor analyses were performed to estimate empirically the num-ber of factors of the 25-item CDSES-SF for Chinese college students. Initially, a

principal axis factoring analysis of variance (PAF) with varimax rotation throughSPSS 10.0 was performed. The results of this PAF were factorially complex andnot interpretable in that many items had double or triple loadings greater than.30 on more than one factor. Following the suggestion of Creed et al. (2002), aPAF with oblique rotation was then conducted. This PAF identified three factorswith 40.44% of the variance accounted for in Sample 1. Factor 1 included Items3, 4, 5, 7, 8, 9, 11, 16, 21, 24, and 25, which explained 33% of the variance(eigenvalue 8.31). Factor 2 contained Items 1, 10, 15, 19, 20, and 22, whichexplained 4.09 of the variance (eigenvalue 1.25). Factor 3 consisted of Items 12,13, 17, and 23, which explained 3.09 of the variance (eigenvalue 1.02). Theremaining five items (Items 2, 6, 12, 14, and 18) had dual loadings greater than.30 on more than one factor.

Based on the results of the PAF, a three-factor solution was selected and testedsubsequently using confirmatory factor analysis (CFA). CFA was used because itallowed me to specify a factor structure a priori, test theoretical expectations of the underlying relations among the variables, and evaluate, via various measuresof fit indexes, how well the proposed measurement models fit the empirical data(Byrne, 2001).

The statistical software AMOS 4 was used to conduct the CFA, and the max-

imum likelihood (ML) method was used in the CFA analyses. The ML methodis the default analysis method of the AMOS 4. Compared with other methods of CFA such as the listwise and pairwise methods, the ML has the advantage of dealing with missing data in that the estimates of ML are consistent, efficient,and theoretically based, and they exhibit the least bias (Arbuckle, 1996).

Following the suggestions of McDonald and Ho (2002), four measures of fitwere used to evaluate how well the data fit the aforementioned model: (a) thechi-square/degrees of freedom (df ) ratio, (b) the goodness-of-fit index (GFI), (c)the comparative-fit-index (CFI), and (d) the root mean square error of approxi-mation (RMSEA). The chi-square statistics are reported in Table 2 but were notused in evaluating goodness of fit because it is affected by model and sample size(Bentler & Bonett, 1980).

For the chi-square/ df ratio, a value below 3 is considered acceptable (Kline,1994). The GFI and CFI statistics range from 0 to 1, and values greater than .90

 at Sunan Kalijaga Yogyakarta PARE on December 16, 2015 jca.sagepub.comDownloaded from 

Page 7: Journal of Career Assessment-2005-Hampton-98-113.pdf

7/23/2019 Journal of Career Assessment-2005-Hampton-98-113.pdf

http://slidepdf.com/reader/full/journal-of-career-assessment-2005-hampton-98-113pdf 7/16

104 JOURNAL OF CAREER ASSESSMENT / February 2005

indicate a good model fit (Byrne, 2001). For RMSEA, a value of .05 or less indi-cates a good fit, a value of .08 indicates a reasonable fit, and a value of .10 or high-

er indicates a poor fit (Byrne, 2001; McDonald & Ho, 2002). These aforemen-tioned criteria were used to examine the model fit in the present study.The five items that had dual loadings were excluded from the CFA. Although

all the item parameters were statistically significant, the model still did not ade-quately describe the sample data. Specifically, the chi-square/ df  ratio was 2.60(acceptable), the GFI and the CFI were .86 and .83, respectively (poor fitting),and the RMSEA was .08 (marginal).

 A decision to respecify and reestimate the model was made to determine amodel that better represented the sample data yet maintained the three-factorstructure. Items that had large modification indexes or large standardized residu-

als were identified. Large modification indexes/standardized residuals are indica-tions of misfit resulting from a complex pattern of multiple loading or a highdegree of overlap in item content (Byrne, 2001). Items were deleted one at a timeto study changes in parameter estimates, modification indexes, and standardizedresiduals. Items were deleted until an acceptably fitting model was obtained,where acceptability was defined as a model that met the criteria of goodness of fit discussed previously.

 A total of seven items were eliminated. The respecified three-factor model of the CDSES-SF scale contained 13 items. Factor 1 consisted of six items, includ-

ing three planning items (Items 3, 7, 24), one self-appraisal item (Item 9), onegoal-setting item (Item 16), and one problem-solving item (Item 8). All theseitems appeared to be related to making decisions, and three of them paralleledwith the decision-making items suggested by Creed et al. (2002). This factor wasthen labeled decision making.

Factor 2 contained four items. Of them, three (Items 1, 10, 19) were original-ly written for the Occupational Information subscale and one (Item 22) for theSelf-Appraisal scale. This factor (Factor 2) was labeled information gathering.There were three items in Factor 3. Two of these items (Items 13 and 17) wereoriginally written for the Problem Solving subscale and one (Item 12) for the

Planning subscale. The third factor was labeled problem solving. Figure 1 showsthe structure of this respecified model of the CDSES-SF.

The descriptive statistics for the respecified scale are shown in the second por-tion of Table 1 for Sample 1. The total scores of the 13-item three-factor modelranged from 33 to 115 with a mean score of 83.71 (SD = 17.49). The mean of the three subscales ranged from 18.78 to 38.73. All the scores were normally dis-tributed (the skewness and kurtosis < 1.0). As a check on reliability, Cronbach’salpha for this respecified model was computed and was found to be .85. The reli-ability coefficients of the three subscales were .77 for the Decision Making sub-

scale, .74 for the Information Gathering subscale, and .69 for the ProblemSolving subscale.The factor loadings for all 13 items were statistically significant (p < .05) and

were moderate in magnitude (Table 3). The correlations between the three fac-tors were moderate (decision making–information gathering = .60, information

 at Sunan Kalijaga Yogyakarta PARE on December 16, 2015 jca.sagepub.comDownloaded from 

Page 8: Journal of Career Assessment-2005-Hampton-98-113.pdf

7/23/2019 Journal of Career Assessment-2005-Hampton-98-113.pdf

http://slidepdf.com/reader/full/journal-of-career-assessment-2005-hampton-98-113pdf 8/16

Hampton / CAREER DECISION SELF-EFFICACY SCALE 105

gathering–problem solving = .64, decision making–problem solving = .77). As

indicated by the following goodness-of-fit indexes, this respecified 13-item modelof the CDSES-SF fits the data well, specifically, chi-square/ df = 2.17, GFI = .93;CFI = .92, and RMSEA = .06 (Table 2).

1DM

Item24r10

1

Item16r81

Item9r61

Item8r5

Item7r4 1

Item3r1

1

1

2IG

Item22r171

Item19r15

1

Item10r131

Item1r12  1

3PS

Item12r231

Item17r211

Item13r20 11

1

1

Figure 1. Factor structure of the 13-item Career Decision Self-Efficacy–Short Form.

Note. 1 DM = Factor 1 decision making; 2 IG = Factor 2 information gathering; 3 PS = Factor 3problem solving.

 at Sunan Kalijaga Yogyakarta PARE on December 16, 2015 jca.sagepub.comDownloaded from 

Page 9: Journal of Career Assessment-2005-Hampton-98-113.pdf

7/23/2019 Journal of Career Assessment-2005-Hampton-98-113.pdf

http://slidepdf.com/reader/full/journal-of-career-assessment-2005-hampton-98-113pdf 9/16

106 JOURNAL OF CAREER ASSESSMENT / February 2005

 When the decision was made to respecify and reestimate the model, the sub-sequent analysis was no longer a confirmatory but an exploratory analysis(Byrnes, 2001). Therefore, it was necessary to cross-validate this respecifiedmodel given the many modifications made in the respecification (Byrne, 2001;Byrne, Shavelson, & Muthen, 1989). Sample 2 of the present study served as thevalidation sample. The descriptive statistics and alpha coefficients were comput-ed from Sample 2 for the 13-item scale and are shown in the third portion of Table 1. The distributions of the scores of the total scale and three subscales were

fairly normal (the skewness and kurtosis < 1.0). When comparing the results fromthe two samples, I noted a slight variability in the means of the total scale andthree subscales. However, these differences are trivial and statistically meaning-

Table 1Descriptive Statistics for the Career Decision

Self-Efficacy Scale for Samples 1 and 2

SA OI GS PL PS IG DM Total

(Sample 1, n = 256)

25-item scale, 5 factorsMean 27.43 28.39 32.39 31.17 30.63 — — 155.97

Standarddeviation 6.41 9.06 7.93 8.29 8.77 — — 34.90

 Alpha .78 .74 .78 .70 .70 — — .91

Items 5.00 5.00 5.00 5.00 5.00 — — 25.00

(Sample 1, n = 256)

13-item respecified scale, 3 factorsMean — — — — 18.78 26.19 38.73 83.71

Standarddeviation — — — — 5.66 6.33 9.10 17.49

 Alpha — — — — .69 .74 .77 .85

Items — — — — 3.00 4.00 6.00 13.00

(Sample 2, n = 157)

13-item respecified scale, 3 factorsMean — — — — 16.80 25.80 39.54 82.15

Standarddeviation — — — — 4.99 6.03 8.04 15.28

 Alpha — — — — .55 .71 .74 .82

Items — — — — 3.00 4.00 6.00 13.00

Note. SA = self-appraisal; OI = occupational information; GS = goal setting; PL = planning; PS =problem solving; IG = information gathering; DM = decision making.

 at Sunan Kalijaga Yogyakarta PARE on December 16, 2015 jca.sagepub.comDownloaded from 

Page 10: Journal of Career Assessment-2005-Hampton-98-113.pdf

7/23/2019 Journal of Career Assessment-2005-Hampton-98-113.pdf

http://slidepdf.com/reader/full/journal-of-career-assessment-2005-hampton-98-113pdf 10/16

Hampton / CAREER DECISION SELF-EFFICACY SCALE 107

less. Overall, the reliability estimates from Sample 2 were similar (.71 to .82) tothe estimates obtained from Sample 1 (.73 to .85). However, the reliability coef-ficient for the Problem Solving subscale for Sample 2 was low (.55), compared

with the estimate obtained from Sample 1 (.69).Four models of invariance were tested hierarchically, where the hierarchy

began with the least restrictive model, a pattern of nonconstrained parametersinvariant across two samples (Model A), followed by a model specifying invariantfactor loadings across samples (Model B), a model specifying invariant factorloadings and invariant factor variances across samples (Model C), and a modelspecifying invariant factor loadings, invariant factor variances, and invariant fac-tor covariances across samples (Model D).

Following Byrne’s (2001) recommendation, the fit indexes used to evaluatemultiple-group measurement invariance tests in this study included the chi-square fit statistics, the chi-square/ df ratio, the CFI, and the RMSEA. If the fitindexes are adequate, reasonable evidence of parameter invariance existed andthe hypothesis of invariance can be retained (Byrne, 2001; Marsh, 1995; Marsh& Hocevar, 1985)

Table 2Goodness-of-Fit Statistics for Different Models

of the Career Decision Self-Efficacy–Short Form

Model   χ2 df p   χ

2/ df  GFI CFI RMSEA  

(Sample 1, n = 256)

5 factors 784.95 265 < .05 2.96 .81 .79 .08

3 factors 134.86 62 < .05 2.17 .93 .92 .06(respecified)

2 factors 426.06 89 < .05 4.78 .81 .73 .12

(Peterson &delMas, 1998)

3 factors 766.05 227 < .05 3.37 .80 .75 .09

(Creed et al.,2002, Australia)

3 factors 854.28 227 < .05 3.76 .77 .72 .10

(Creed et al.,2002, South Africa)

(Sample 2, n = 157)

3 factors 120.75 62 < .05 1.94 .90 .90 .07

(respecified)

Note. df = degrees of freedom; χ2/ df = chi-square/degrees of freedom ratio, and a value below 3 isacceptable; GFI = goodness-of-fit index; CFI = comparative-fit-index, and values greater than .90indicate a good model fit; RMSEA = root mean square error of approximation, and values small-er than .08 indicate a reasonable fit.

 at Sunan Kalijaga Yogyakarta PARE on December 16, 2015 jca.sagepub.comDownloaded from 

Page 11: Journal of Career Assessment-2005-Hampton-98-113.pdf

7/23/2019 Journal of Career Assessment-2005-Hampton-98-113.pdf

http://slidepdf.com/reader/full/journal-of-career-assessment-2005-hampton-98-113pdf 11/16

108 JOURNAL OF CAREER ASSESSMENT / February 2005

The initial test of invariance specified only that the same factor pattern appliedto each sample and it did not involve any equality constraints (Model A). AsTable 4 indicates, the chi-square/ df of Model A is small (2.06), the CFI is .90, andthe RMSEA is .05, providing evidence for the assumption that the factor loadingsare invariant across the two samples. This model was used as the baseline modelfor comparison with subsequent models for which additional invariance con-straints were imposed.

Then, constraints were imposed on factor loading regression paths (Model B),factor variances (Model C), and factor correlations (Model D) step by step, andthese three models were tested one by one separately. The chi-square value of each of the constrained models was compared with the chi-square value of the

Table 3Parameter Estimates for the 13-Item Career Decision

Self-Efficacy–Short Form for Samples 1 and 2Loadings

Sample SampleItem 1 2

Factor 1, decision making

Q3 Make a plan of your goals for the next 5 years. .53 .49

Q7 Determine the steps you need to take to successfully completeyour chosen major. .63 .63

Q8 Persistently work at your major or career goal even when you

get frustrated. .62 .56Q9 Determine what your ideal job would be. .50 .61

Q16 Make a career decision and then do not worry whether it wasright or wrong. .63 .54

Q24 Successfully manage the job interview process. .67 .57

Factor 2, information gathering

Q1 Find information in the library about occupations you areinterested in. .75 .67

Q10 Find out the employment trends for an occupation over

the next 10 years. .75 .70Q19 Talk with a person already employed in the field you are

interested in. .60 .64

Q22 Define the type of lifestyle you would like. .49 .49

Factor 3, problem solving

Q12 Prepare a good resume. .59 .69

Q13 Change majors if you did not like your first choice. .70 .37

Q17 Change occupations if you are not satisfied with the one you enter. .69 .52

 at Sunan Kalijaga Yogyakarta PARE on December 16, 2015 jca.sagepub.comDownloaded from 

Page 12: Journal of Career Assessment-2005-Hampton-98-113.pdf

7/23/2019 Journal of Career Assessment-2005-Hampton-98-113.pdf

http://slidepdf.com/reader/full/journal-of-career-assessment-2005-hampton-98-113pdf 12/16

Hampton / CAREER DECISION SELF-EFFICACY SCALE 109

baseline model by subtracting the chi-square and df for the baseline model fromthe chi-square/ df for each of the constrained models. The comparisons yieldedthree chi-square difference values that were statistically nonsignificant at the .05probability level (Table 4). Additionally, the fit indexes were used to assess invari-ance. As Table 4 indicates, the chi-square/ df ratio of each of the constrained modelsis smaller than 2, the CFI is .90, and the RMSEA is .04. These results indicate that

the same factorial structure holds across the two samples, and the hypothesis of thefactor loading, factor variance, and factor correlation–invariance can be retained. Additionally, CFAs were conducted for three models proposed by Peterson

and delMas (1998) or Creed et al. (2002). Although these models had acceptablereliability coefficients among Chinese students (the alpha reliability coefficientwas .85 for the Peterson model and .91 for the two models of Creed et al.), theindexes of model fit showed that these models fit the sample data poorly (Table 2).

DISCUSSION

The purposes of this study were twofold: to explore and test the most appro-priate measurement model of the CDSES-SF for Chinese students, and to cross-

Table 4Invariance Tests Across Samples for the 13-Item

Career Decision Self-Efficacy–Short Form Scale

Model   χ2 df    ∆χ

2∆ df p   χ

2/ df  CFI RMSEA 

Model A 255.68 124 — — — 2.02 .90 .05

(Baseline)

Model B 264.26 134 8.58 10 > .05 1.97 .90 .04

(Factor loadingconstrained)

Model C 269.57 137 13.99 13 > .05 1.96 .90 .04

(Factor loadingand varianceconstrained)

Model D 274.38 140 18.70 16 > .05 1.96 .90 .04

(Factor loading,variance andcovarianceconstrained)

Note. ∆χ2, difference in values between Model A and Models B, C, and D; ∆df , difference innumber of degrees of freedom between Model A and Models B, C, and D; χ

2/ df = chi-square/ degrees of freedom ratio; CFI = comparative-fit-index; RMSEA = root mean square error of approximation.

 at Sunan Kalijaga Yogyakarta PARE on December 16, 2015 jca.sagepub.comDownloaded from 

Page 13: Journal of Career Assessment-2005-Hampton-98-113.pdf

7/23/2019 Journal of Career Assessment-2005-Hampton-98-113.pdf

http://slidepdf.com/reader/full/journal-of-career-assessment-2005-hampton-98-113pdf 13/16

110 JOURNAL OF CAREER ASSESSMENT / February 2005

validate the structure of this model over an independent sample of Chinese stu-dents. On one hand, the results of this study support that the 25-item CDSES-

SF has high internal consistency reliability when used with Chinese college stu-dents. The alpha reliability coefficient for the total scale was .91, and the alphacoefficients for the five subscales of the CDSES-SF were all moderate and with-in the acceptable range. These results are consistent with previous studies thatalso found the reliability coefficients of the CDSES-SF and its subscales to bemoderate to high (Betz et al., 1996; Creed et al., 2002; Mau, 2000).

On the other hand, the results of this study failed to support the hypothesizedfive-factor structure of the CDSES-SF. Rather, an alternative three-factor modelemerged from an exploratory factor analysis. This alternative model of theCDSES-SF, however, was still inadequate based on the results of a CFA. To

develop a more parsimonious CDSES-SF, the alternative three-factor model wasrespecified, and items that had dual loadings or tended to be redundant in con-tent were eliminated based on modification indices. A 13-item scale was derived,and the results of a CFA suggested that this respecified scale fit the data fromSample 1 quite well. Previous researchers suggested that a number of items in theCDSES-SF might be redundant and could be reduced (Creed et al., 2002). Thefindings of the present study seem to support this suggestion. Although a total of 12 items were deleted from the scale, the respecification provided a model thatbetter described the sample data (GFI increased from .86 to .93 and CFI from

.83 to .92, and RMSEA decreased from .08 to .06). Additionally, the alpha relia-bility coefficient estimated from the r -specified scale was acceptable (.85).The 13-item three-factor model of the CDSES-SF was then cross-validated

with a second sample of Chinese college students. The cross-validation testrevealed that the measurement parameters were invariant across the two samples.This multiple group analysis offered a powerful test of the equivalence of thethree-factor solution across samples because it rigorously assessed measurementproperties (Marsh, 1995). The results of the test also suggest that the Chinese-language version of the 13-item CDSES-SF is reliable and replicable.

 Although the three factors that emerged from the present study are similar to

the factors found in the study by Creed et al. (2002), the composition of each of the three factors is different. For example, Items 9 and 24 are part of the decision-making factor in the present study, but they are part of the information-gatheringfactor in the study by Creed et al. This difference suggests that some of the itemsof the CDSES-SF do not measure the same construct across different cultures.On the other hand, three items (Items 1, 10, and 19) of the CDSES-SF consis-tently measure the construct of occupational-related information gatheringacross several culturally different groups (e.g., American college students in thestudy by Betz et al., 1996; Australia and South African high school students the

study by Creed et al.; and Chinese college students in the present study). Thisconsistency indicates that psychometric and cultural equivalences for some itemsof the CDSES-SF may be assumed.

 Although the present study provides a more parsimonious measurementmodel of the CDSES-SF, the model does not reflect all five domains of career

 at Sunan Kalijaga Yogyakarta PARE on December 16, 2015 jca.sagepub.comDownloaded from 

Page 14: Journal of Career Assessment-2005-Hampton-98-113.pdf

7/23/2019 Journal of Career Assessment-2005-Hampton-98-113.pdf

http://slidepdf.com/reader/full/journal-of-career-assessment-2005-hampton-98-113pdf 14/16

Hampton / CAREER DECISION SELF-EFFICACY SCALE 111

choice competence proposed by Crites (1961). This finding further supports thesuggestion by Creed et al. (2002) that self-appraisal and goal-setting dimensions

may not be adequately reflected in the 25-item CDSES-SF when used with cul-turally different populations. Compared with their American counterparts,Chinese college students live in a collective culture that emphasizes collectivedecisions, and the self, to these students, “is made meaningful primarily in refer-ence to those of social relation” (Mau, 2000). The original CDSES-SF may notreflect this kind of self- and collective-oriented goal selection.

Furthermore, a recent study indicated that there were four factors underly-ing the Chinese College Student Career Selection Efficacy Questionnaire(CCSCSEQ), and these four factors included career information, academicachievement, personality, and social support (Zheng & Zhang, 2002). Although

the career information construct measured by the CCSCSEQ was similar to theoccupational information construct in the CDSES, other three constructs meas-ured by the CCSCSEQ were based on issues concerning college students’ careerchoices and decision-making processes in the social context of China (Zheng &Zhang, 2002). It seems that researchers may need to use both the revisedCDSES-SF and the CCSCSEQ when assessing career decision-making self-efficacy of Chinese college students. Meanwhile, instruments that measureChinese students’ self-appraisal and goal selection may be developed.

The findings of the present study have several implications for research. First,

the 13-item CDSES-SF appears to be a reliable and valid instrument that couldbe applied to understanding Chinese college students’ career decision-makingbehavior. However, continued evaluation of the applicability of the 13-itemCDSES-SF to a Chinese culture is necessary before it can be used in applied set-tings. Second, because this study was a first attempt to confirm the factor struc-ture of the CDSES-SF among Chinese college students, replications of this studythat draw on similar and different populations are needed. For instance, becauseof the privatization of the stated-owned enterprise, many workers have been laidoff in China. They desperately need career counseling and retraining (Zhanget al., 2002). Researchers may include this population in future studies.

Third, globalization has strongly influenced many aspects of people’s life,such as education and career choices. More Chinese students from China, HongKong, and Taiwan have moved to Western countries such as Australia, Canada,England, France, Germany, Japan, New Zealand, and the United States to study,and more Western companies have hired Chinese college graduates both inChina and elsewhere in the world. Research is needed to understand careerdevelopment and decision-making process of international students from theChinese culture and Chinese employees of Western companies. The 13-itemCDSES-SF can be used in such studies. Finally, the limitations of the study

include the use of two convenient samples and the absence of an assessment of the test-retest reliability of the scale. Future investigations should use a bettersampling method (e.g., stratified sampling) and evaluate the test-retest reliabilityof the CDSES whenever possible.

 at Sunan Kalijaga Yogyakarta PARE on December 16, 2015 jca.sagepub.comDownloaded from 

Page 15: Journal of Career Assessment-2005-Hampton-98-113.pdf

7/23/2019 Journal of Career Assessment-2005-Hampton-98-113.pdf

http://slidepdf.com/reader/full/journal-of-career-assessment-2005-hampton-98-113pdf 15/16

112 JOURNAL OF CAREER ASSESSMENT / February 2005

REFERENCES

 Arbuckle, J. L. (1996). Full information estimation in the presence of incomplete data. In G. A.Marcoulides & R. E. Schumacker (Eds.),  Advanced structural and equation modeling: Issuesand techniques (pp. 243-277). Mahwah, NJ: Lawrence Erlbaum.

Bandura, A. (1986). Social foundations of thought and action: A social cognitive theory. EnglewoodCliffs, NJ: Prentice Hall.

Bandura, A. (1997). Self-efficacy: The exercise of control. New York: Freeman.Bentler, P. M., & Bonett, D. G. (1980). Significance tests and goodness of fit in the analysis of 

covariance structures. Psychological Bulletin, 88, 588-606.Bergeron, L. M., & Romano, J. L. (1994). The relationships among career decision-making self-

efficacy, educational indecision, vocational indecision, and gender. Journal of College StudentDevelopment, 35, 19-24.

Betz, N. E., & Hackett, G. (1986). Applications of self-efficacy theory to understanding careerchoice behavior. Journal of Social and Clinical Psychology, 4, 279-289.

Betz, N. E., Klein, K. L., & Taylor, K. M. (1996). Evaluation of a short form of the Career Decision-Making Self-Efficacy Scale. Journal of Career Assessment, 4, 47-57.

Betz, N., & Luzzo, D. A. (1996). Career assessment and the Career Decision-Making Self-EfficacyScale. Journal of Career Assessment, 4, 413-428.

Byrne, B. (2001). Structural equation modeling with AMOS: Basic concepts, applications, and pro- gramming. Mahwah, NJ: Lawrence Erlbaum.

Byrne, B., Shavelson, R. J., & Muthen, B. (1989). Testing for the equivalence of factor covarianceand mean structures: The issue of partial measurement invariance. Psychological Bulletin, 105,456-466.

Crites, J. O. (1961). A model for the measurement of vocational maturity. Journal of CounselingPsychology, 8, 255-259.

Chung, Y. B. (2002). Career decision-making self-efficacy and career commitment: Gender andethnic differences among college students. Journal of Career Development, 28, 277-284.

Creed, P. A., Patton, W., & Watson, M. B. (2002). Cross-cultural equivalence of the CareerDecision-Making Self-Efficacy Scale–Short Form: An Australian and South African compari-son. Journal of Career Assessment, 10, 327-342.

France, M. H., Jin, Y., Huang, K., Si, F., & Zhang, W. (1991). Career needs of Chinese middleschool students and implications for school guidance in the People’s Republic of China. Career Development Quarterly, 40, 155-167.

Gati, I., Osipow, S. H., & Fassa, N. (1994). The scale structure of multi-scale measures: Application

of the split-scale method to the task-specific Occupational Self-Efficacy Scale and the CareerDecision-Making Self-Efficacy Scale. Journal of Career Assessment, 2, 384-397.Hackett, G., & Betz, N. E. (1981). A self-efficacy approach to the career development of women.

Journal of Vocational Behavior , 18, 326-339.Hackett, G., & Betz, N. E. (1995). Self-efficacy and career choice and development. In J. E.

Maddux (Ed.), Self-efficacy, adaptation, & adjustment: Theory, research, and application (pp.249-280). New York: Plenum.

Kline, P. (1994). An easy guide to factor analysis. New York: Routledge.Lent, R. W., & Hackett, G. (1987). Career self-efficacy: Empirical status and future directions.

Journal of Vocational Behavior , 30, 347-382.Luzzo, D. A. (1993). Reliability and validity testing of the career decision-making self-efficacy scale.

Measurement and Evaluation in Counseling and Development, 26, 137-142.Luzzo, D. A. (1996). A psychometric evaluation of the Career Decision-Making Self-Efficacy

Scale. Journal of Counseling and Development, 74, 276-279.

 at Sunan Kalijaga Yogyakarta PARE on December 16, 2015 jca.sagepub.comDownloaded from 

Page 16: Journal of Career Assessment-2005-Hampton-98-113.pdf

7/23/2019 Journal of Career Assessment-2005-Hampton-98-113.pdf

http://slidepdf.com/reader/full/journal-of-career-assessment-2005-hampton-98-113pdf 16/16

Hampton / CAREER DECISION SELF-EFFICACY SCALE 113

McDonald, R. P., & Ho, M. R. (2002). Principles and practice in reporting structural equationanalyses. Psychological Methods, 7, 64-82.

Marsh, H. W. (1995). Confirmatory factor analysis models of factorial invariance: A multifaceted

approach. Structural Equation Modeling, 1, 5-34.Marsh, H. W., & Hocevar, D. (1985). The application of confirmatory factor analysis to the study

of self-concept: First and higher order factor structures and their invariance across age groups.Psychological Bulletin, 97, 562-582.

Mau, W. (2000). Cultural differences in career decision-making styles and self-efficacy. Journal of Vocational Behavior , 57, 365-378.

Niles, S., & Sowa, C. (1992). Mapping the nomological network of career self-efficacy. Career Development Quarterly, 41, 13-21.

Nilsson, J. E., Schmidt, C. K., & Meek, W. D. (2002). Reliability generalization: An examinationof the Career Decision-Making Self-Efficacy Scale. Educational and PsychologicalMeasurement, 62, 647-658.

O’Brien, K. M. (2003). Measuring career self-efficacy promoting confidence and happiness atwork. In S. J. Lopez & C. R. Snyder (Eds.), Positive psychological assessment: A handbook of models and measures (pp. 109-126). Washington, DC: American Psychological Association.

Osipow, S. H., & Gati, I. (1998). Construct and concurrent validity of the Career Decision-MakingDifficulties Questionnaire. Journal of Career Assessment, 6, 347-364.

Peterson, S. L., & delMas, R. C. (1998). The component structure of career decision-making self-efficacy for underprepared college students. Journal of Career Development, 24, 209-225.

Robbins, S. (1985). Validity estimates for the Career Decision-Making Self-Efficacy Scale.Measurement and Evaluation in Counseling and Development, 18, 64-71.

Solberg, V. S., Good, G. E., Nord, D., Holm, C., Hohner, R., Zima, et al. (1994). Assessing careersearch expectations: Development and validation of the Career Search Efficacy Scale. Journal

of Career Assessment, 2, 111-123.Taylor, K. M., & Betz, N. (1983). Applications of self-efficacy theory to the understanding and treat-ment of career indecision. Journal of Vocational Behavior , 22, 63-81.

Taylor, K. M., & Popma, J. (1990). An examination of the relationships among career decision-making self-efficacy, career salience, locus of control, and vocational indecision. Journal of Vocational Behavior , 37, 17-21.

Triandis, H. C., Bontempo, R., Betancourt, H., Bond, M., Leung, K., Brenes, A., et al. (1986). Themeasurement of the ethic aspects of individualism and collectivism across cultures.  AustralianJournal of Psychology, 38, 257-267.

Zhang, W., Hu, X., & Pope, M. (2002).The evolution of career guidance and counseling People’sRepublic of China. Career Development Quarterly, 50, 226-236.

Zheng, R., & Zhang, S. (2002). Confirmatory factor analysis of the structure of self-efficacy in career

decision making in a Chinese college sample. Psychological Science (China), 25(1), 91-92.