18
7/21/2019 Journal of Career Assessment-2002-Betz-283-300.pdf http://slidepdf.com/reader/full/journal-of-career-assessment-2002-betz-283-300pdf 1/18 The Utility of Measures of Self-Efficacy for the Holland Themes in African American and European American College Students Nancy E. Betz Laurie R. Gwilliam The Ohio State University Three inventories of self-efficacy or confidence with respect to the six Holland themes were compared in samples of European American and African American college students. Results indicated that the reliabilities of all subscales were suffi- ciently high for use in research and that all subscales were as reliable for African  Americans as for European Americans. There were significant gender differences in confidence levels but few race differences. A multitrait-multimethod analysis indicated differential degrees of convergent and discriminant validity and the pres- ence of method variance across the scales examined herein. Keywords: Career self-efficacy, Holland’s vocational theory, African Americans, Holland confidence themes, career development of college students  As manifested by the appearance of several recent special issues of the Journal of Career Assessment focusing on career assessment and self-efficacy, applications of Bandura’s (1977, 1986) self-efficacy theory to career psychology represent one of the most visible and flourishing areas of vocational research today. Self-effica- cy expectations, our beliefs concerning our competence in specific behavioral domains, are postulated to influence our choices of, performance in, and per- sistence in areas of endeavor requiring or using those behavioral competencies. Betz (2000) has reviewed the career domains to which self-efficacy theory has been applied, and Lent, Brown, and Hackett (1994, 2000) have developed a comprehensive model of career choice intentions and goals using self-efficacy as a central construct. Much current attention is being paid to the assessment of self-efficacy relative to the activity domains postulated in Holland’s (1985, 1997) vocational theory. This is in part due to the increasing use of parallel measures of vocational inter- est and self-efficacy in career counseling (e.g., Betz & Borgen, 2000). Evidence Correspondence regarding this article should be directed to Nancy E. Betz, Professor, Department of Psychology, 110 Townshend Hall, The Ohio State University, 1885 Neil Avenue, Columbus, OH 43210-1222; e-mail: [email protected]. JOURNAL OF CAREER ASSESSMENT, Vol. 10 No. 3, August 2002 283–300 © 2002 Sage Publications 283  at Sunan Kalijaga Yogyakarta PARE on December 16, 2015  jca.sagepub.com Downloaded from 

Journal of Career Assessment-2002-Betz-283-300.pdf

Embed Size (px)

Citation preview

Page 1: Journal of Career Assessment-2002-Betz-283-300.pdf

7/21/2019 Journal of Career Assessment-2002-Betz-283-300.pdf

http://slidepdf.com/reader/full/journal-of-career-assessment-2002-betz-283-300pdf 1/18

The Utility of Measures of Self-Efficacy forthe Holland Themes in African Americanand European American College Students

Nancy E. BetzLaurie R. GwilliamThe Ohio State University

Three inventories of self-efficacy or confidence with respect to the six Holland

themes were compared in samples of European American and African Americancollege students. Results indicated that the reliabilities of all subscales were suffi-ciently high for use in research and that all subscales were as reliable for African

 Americans as for European Americans. There were significant gender differencesin confidence levels but few race differences. A multitrait-multimethod analysisindicated differential degrees of convergent and discriminant validity and the pres-ence of method variance across the scales examined herein.

Keywords: Career self-efficacy, Holland’s vocational theory, African Americans,Holland confidence themes, career development of college students

 As manifested by the appearance of several recent special issues of the Journalof Career Assessment focusing on career assessment and self-efficacy, applications

of Bandura’s (1977, 1986) self-efficacy theory to career psychology represent oneof the most visible and flourishing areas of vocational research today. Self-effica-cy expectations, our beliefs concerning our competence in specific behavioraldomains, are postulated to influence our choices of, performance in, and per-sistence in areas of endeavor requiring or using those behavioral competencies.Betz (2000) has reviewed the career domains to which self-efficacy theory hasbeen applied, and Lent, Brown, and Hackett (1994, 2000) have developed acomprehensive model of career choice intentions and goals using self-efficacy asa central construct.

Much current attention is being paid to the assessment of self-efficacy relativeto the activity domains postulated in Holland’s (1985, 1997) vocational theory.This is in part due to the increasing use of parallel measures of vocational inter-est and self-efficacy in career counseling (e.g., Betz & Borgen, 2000). Evidence

Correspondence regarding this article should be directed to Nancy E. Betz, Professor, Department of Psychology, 110 Townshend Hall, The Ohio State University, 1885 Neil Avenue, Columbus, OH 43210-1222;e-mail: [email protected].

JOURNAL OF CAREER ASSESSMENT, Vol. 10 No. 3, August 2002 283–300© 2002 Sage Publications

283

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

Page 2: Journal of Career Assessment-2002-Betz-283-300.pdf

7/21/2019 Journal of Career Assessment-2002-Betz-283-300.pdf

http://slidepdf.com/reader/full/journal-of-career-assessment-2002-betz-283-300pdf 2/18

284 JOURNAL OF CAREER ASSESSMENT / August 2002

supporting the usefulness of self-efficacy as well as interests in career counseling

has been accumulating since Betz and Hackett’s (1981) first study applying self-efficacy theory to the study of career behavior. In that first study, Betz andHackett found that both self-efficacy and interests predicted the nature and rangeof career options considered by college students.

Since the first study, a number of other studies have provided additional evi-dence for the incremental predictive ability of perceived self-efficacy (e.g.,Lapan, Boggs, & Morrill, 1989; Lapan, Shaughnessy, & Boggs, 1996). Probablythe largest scale study was that of Donnay and Borgen (1999), who used scoreson the six general occupational (Holland) themes (GOTs) of the Strong InterestInventory (SII) and the six general confidence themes (GCTs) of the SkillsConfidence Inventory (SCI) (Betz, Borgen, & Harmon, 1996; Betz, Harmon, &Borgen, 1996) to predict occupational group membership among 1,105 adultsemployed in 21 occupations. Using discriminant analysis in both a validation anda cross-validation sample, Donnay and Borgen found significant incrementalvalidity for the self-efficacy scales (the GCTs) beyond that accounted for by theGOTs. As a separate multivariate set, the GCTs were somewhat more powerfulthan the GOTs in separating the 21 occupational groups (six times greater thanchance versus five times greater than chance). In percentage of variance terms,GOTs accounted for 70% of occupational differences, GCTs accounted for 82%of occupational differences, and the combined GOTs and GCTs accounted for91% of occupational differences.

Isaacs, Borgen, Donnay, and Hansen (1997) investigated the extent to whichself-efficacy added to the prediction of college major, beyond the predictive effi-

cacy of interests. Using the Holland interest themes from the SII and the GCTsfrom the SCI, they found that self-efficacy, and the combined synergistic effectof interests and self-efficacy, added to the prediction of the Holland theme of thecollege major.

Given findings of the predictive utility of both interests and self-efficacy, it isappropriate that means for joint use of such measures in career counseling arenow increasingly available. There are now several measures of self-efficacy withrespect to one or more of the six Holland (1985) types. Lapan et al. (1989) devel-oped a measure of self-efficacy with respect to the RIASEC areas to examine thedegree to which gender differences in self-efficacy mediated gender differencesin interests. Lapan et al. introduced and described each theme as an occupa-tional category, below which examples of occupations in that category were given

(for example, pharmacist, computer programmer, mathematician, and physicistfor the Investigative theme). The focus of Lapan et al’s analysis was Realistic andInvestigative self-efficacy, which were significantly related to Realistic and Investi-gative interests, respectively. Path analyses indicated that lower perceived self-efficacy was a major factor in women’s lower Realistic and Investigative interests.

Lenox and Subich (1994) also developed measures of self-efficacy with respectto Holland’s themes. Six 5-item scales were developed using activities represent-ing each theme. Only the Realistic, Investigative, and Enterprising scales were

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

Page 3: Journal of Career Assessment-2002-Betz-283-300.pdf

7/21/2019 Journal of Career Assessment-2002-Betz-283-300.pdf

http://slidepdf.com/reader/full/journal-of-career-assessment-2002-betz-283-300pdf 3/18

Betz, Gwilliam / HOLLAND THEME SELF-EFFICACY 285

investigated further, however, because the Artistic, Social, and Conventional

scales were characterized by restriction in range. Lenox and Subich found gen-eral support for linear relationships between self-efficacy and interests, althoughthey found slight curvilinearity for the Realistic and Investigative scales.

The SCI (Betz, Borgen, et al., 1996) is a 60-item measure of self-efficacy, with10 items assessing self-efficacy for each of the six RIASEC areas. As described inBetz, Borgen, et al. (1996) and Betz, Harmon, et al. (1996), the SCI was devel-oped using samples of about 1,100 employed adults and 700 college students.Each confidence scale consists of 10 activities, tasks, or school subjects associat-ed with the relevant Holland theme. The SCI has been shown to significantly dis-criminate occupational groups (e.g., Betz, Borgen, Kaplan, & Harmon, 1998;Harmon et al., 1996) and to make a significant incremental contribution to theprediction of occupational group membership (Donnay & Borgen, 1999; Tracey& Hopkins, 2001).

Because at least three inventories of confidence with respect to the six Hollandthemes are available, their psychometric characteristics, interrelationships, andrelative degree to which they yield gender and/or race differences would be help-ful for further research uses or uses in career assessment and counseling. Thepresent study was designed to compare the three available measures—the Self-Efficacy Questionnaire (SEQ), the SCI, and the Lapan et al. measure, the Self-Efficacy Rating Scale (SERS)—in a sample of college students.

More specifically, the first major purpose of the present study was to examinethe construct validity of the three instruments using the methodology providedby the multitrait-multimethod matrix (Campbell & Fiske, 1959). Using this

matrix with multiple measures of the same constructs allows the simultaneousinvestigation of convergent validity (that is, the extent to which different meas-ures of the same construct are related) and discriminant validity (the extent towhich measures of different constructs are unrelated). The advantage of a multi-trait-multimethod matrix is that it also allows the examination of method vari-ance (that is, the degree to which scales are correlated because they use the samemethod of measurement rather than because they share valid trait variance).

 As defined by Campbell and Fiske (1959), the monotrait/heteromethod val-ues, indicative of convergent validity, should be greater than the monomethodheterotrait values, indicative of method variance. The convergent validity coeffi-cients should also be greater than heterotrait-heteromethod values, which shareneither trait variance nor method variance. Finally, the patterns of correlations

among a set of traits within a method should be similar from method to method.The second major purpose of the study was to examine the reliability and

validity of the scale in a sample of African American college students. There iswidespread agreement that the psychometric quality and utility of commonlyused measures must be demonstrated for groups other than those samples inwhich the scales were originally normed (see Fouad, 1993; Samuda, 1998;Subich, 1996). Because these scales were originally developed in samples domi-nated by European American college students, their utility in other groups is an

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

Page 4: Journal of Career Assessment-2002-Betz-283-300.pdf

7/21/2019 Journal of Career Assessment-2002-Betz-283-300.pdf

http://slidepdf.com/reader/full/journal-of-career-assessment-2002-betz-283-300pdf 4/18

286 JOURNAL OF CAREER ASSESSMENT / August 2002

important question. This study was designed to contribute to knowledge of the

psychometric characteristics of these scales within a group of African Americansand to compare those to the psychometric characteristics within a group of European American students.

METHOD

Measures

SCI. The SCI (Betz, Borgen, et al., 1996) is a 60-item instrument designed tomeasure self-efficacy with respect to the six Holland themes. It consists of six 10-

item subscales, each of which addresses a single theme (Realistic, Investigative, Artistic, Social, Enterprising, or Conventional). Within each subscale, several items address activities and tasks (e.g., “solve

abstract puzzles”), and others address college-level courses (e.g., “Industrial Arts”). For each item, the respondent is asked to rate his or her current confi-dence in his or her ability to successfully complete the named activity or to suc-cessfully complete a college course on the subject in question. Responses aregiven using a 5-point Likert-type rating scale ranging from 1 (no confidence at all)to 5 (complete confidence). A total subscale score is obtained by summing scoresacross all 10 items and then dividing the result by 10. Possible subscale scoresrange from 1 to 5, with higher scores indicating greater self-efficacy for theHolland theme in question. Because each subscale measures a theoretically dis-

tinct construct, no full scale score is generated.Previously reported values of coefficient alpha for the 10-item subscales of the

SCI range from .79 to .87 in student samples and from .84 to .88 in an adult sam-ple (Betz, Harmon, et al., 1996). Three-week test-retest reliability coefficients forthe six subscales have been reported to range from .83 to .87 (Parsons & Betz,1998). There is extensive evidence for the validity of the SCI, including signifi-cant and moderately sized relationships with same-named Holland interestscores on the SII (Betz, Harmon, et al., 1996) and with probability of concurrentemployment in the predicted Holland area (Harmon et al., 1996).

SEQ. Similar to the SCI, the SEQ (Lenox & Subich, 1994) was designed tomeasure self-efficacy expectations with respect to each of the six Holland themes.

The SEQ consists of six 5-item subscales, one for each of the Holland themes,resulting in a total of 30 items. Items within each subscale ask the respondent torate his or her confidence in his or her ability to complete activities that requirea particular Holland-theme-related skill, such as “Use algebra to solve mathe-matical problems” or “Design clothing, furniture, or posters.” Responses are givenon a 10-point Likert-type response scale, ranging from 1 (completely unsure) to 10(completely sure). Total subscale scores are obtained by summing scores across all10 items. Possible subscale scores thus range from 5 to 50, with higher scores

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

Page 5: Journal of Career Assessment-2002-Betz-283-300.pdf

7/21/2019 Journal of Career Assessment-2002-Betz-283-300.pdf

http://slidepdf.com/reader/full/journal-of-career-assessment-2002-betz-283-300pdf 5/18

Betz, Gwilliam / HOLLAND THEME SELF-EFFICACY 287

indicating greater self-efficacy. Once again, because subscales represent theoret-

ically distinct types of self-efficacy expectations, no full scale score is generated.Lenox and Subich (1994) reported data obtained from a pilot study using the

Realistic, Investigative, and Enterprising subscales of the SEQ. Values of coeffi-cient alpha drawn from this pilot study range from .79 to .88. In terms of validi-ty, the authors reported that scores on the Realistic, Investigative, andEnterprising subscales of the SEQ show a moderately strong correlation withscores on the corresponding GOTs of the SII (Lenox & Subich, 1994). Lenoxand Subich did not use the Social, Artistic, and Conventional scores because of reported restriction in range.

Self-efficacy ratings for GOTs. Like the SCI and the SEQ, this self-efficacyrating instrument (SER) (Lapan et al., 1989) is also designed to measure self-

efficacy expectations for each of the six Holland themes. The SER consists of 95items overall, divided into 12 subscales, 2 subscales for each Holland theme. Thenumber of items contained in each pair of subscales varies across Hollandthemes, ranging from 9 items for the Conventional theme to 20 items for theInvestigative theme.

The SER consists of two separate subscales, one addressing self-efficacy foreducational requirements and the other addressing self-efficacy for job duties.Furthermore, each subscale of the SER is composed of two parts. The first con-sists of a list of occupational titles; the respondent is asked to indicate whether heor she could complete the educational requirements or the job duties for eachoccupation on the list. Responses are given on a yes or no basis. This response

method corresponds to Bandura’s (1977) original definition of level of self-efficacyexpectations and is based directly on the original (Betz & Hackett, 1981) study of occupational self-efficacy expectations. The second part of each subscale consistsof a single item asking respondents to rate their confidence, on a 7-point scale, intheir ability to complete either the educational requirements or the job duties of the list of occupations as a whole. This response continuum more closely resem-bles Bandura’s confidence rating, a measure of strength of self-efficacy expecta-tions, except that Bandura associated strength ratings with individual levelresponses rather than level responses for a whole group of items. Lapan et al.(1989) obtained scores by multiplying the sum of “yeses” to the educationalrequirements for a group of occupations by its overall confidence rating andadding that sum to the product of the number of yeses to the job duties for that

same set of occupations and the confidence rating for that occupational group.Given that this unconventional multiplicative scoring method confounds the

level versus strength definitions and specific versus general definitions of self-efficacy, we decided to retain only the “level” measurement of specific behaviors,in this case occupational titles. Accordingly, the SERS was scored by adding thenumber of “yes” responses for educational requirements to the number of “yes”responses for job duties for each Holland theme. This method corresponds moreclosely to Bandura’s (1977) original definition of level of self-efficacy, as indicat-

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

Page 6: Journal of Career Assessment-2002-Betz-283-300.pdf

7/21/2019 Journal of Career Assessment-2002-Betz-283-300.pdf

http://slidepdf.com/reader/full/journal-of-career-assessment-2002-betz-283-300pdf 6/18

Page 7: Journal of Career Assessment-2002-Betz-283-300.pdf

7/21/2019 Journal of Career Assessment-2002-Betz-283-300.pdf

http://slidepdf.com/reader/full/journal-of-career-assessment-2002-betz-283-300pdf 7/18

Betz, Gwilliam / HOLLAND THEME SELF-EFFICACY 289

Hispanics did not constitute sufficiently sized subgroups. Multivariate analysis of 

variance, with gender and race/ethnicity as fixed factors, was done separately foreach inventory of Holland self-efficacy scores (SCI, SEQ, SERS), followed bypost hoc univariate tests in the case of significant multivariate F statistics. Thecorrelations among all of the measures were arranged in a multitrait-multimethodmatrix. Correlations were also calculated within African American and European American subgroups.

RESULTS

Table 1 shows sample items for each Holland theme area from each of the

three inventories examined herein. Table 2 provides values of coefficient alphafor all scales and also provides overall scale means and standard deviations. Asshown in the table, all of the scales met Nunnally and Bernstein’s (1994) criteri-on of a minimum alpha of .70 for use in research, although not all met standardcriteria for applied (counseling) use (alpha = .80, at a minimum). The SCI scaleshad values of alpha ranging from .83 (Investigative) to .86 (Realistic and Social).Reliabilities of the SEQ scales ranged from .70 (Conventional) to .91 (Realistic),with Artistic also somewhat lower (.79). Finally, values of SERS scale reliabilitiesranged from .78 (Realistic) to .91(Investigative). However, alpha values arestrongly associated with numbers of items in the scale, so differences across thesix SERS subscales may be primarily artifactual in nature.

The Artistic and Conventional scales were among those (along with Social)

criticized by the scale authors Lenox and Subich (1994) as characterized byrestriction in score range, which would also reduce reliability. In the present setof data, the standard deviations of Social and Conventional (1.3 and 1.4) weresmaller than those associated with other scales, and the mean for Social (8.23) wasclosest to the theoretical maximum of 10.0 (complete confidence). Thus, restric-tion in range is shown most clearly herein for the Social SEQ score. From anoth-er vantage point, skewness (negative, showing a ceiling effect) is extreme in theSEQ Social and to a lesser extent Enterprising and Conventional confidence scores.Our data did not indicate restriction of range for Artistic self-efficacy scores.

Table 3 presents the multitrait-multimethod matrix showing the correlationsamong the three inventories of Holland theme self-efficacy. Diagonal values, rep-resenting reliability coefficients, were presented earlier in this section and willnot be repeated here. The first criterion of Campbell and Fiske (1959) is that theconvergent validity values be statistically significant. Given the large N in thisstudy, statistical significance is an inappropriate criterion because values as small asr =.10 would be significant at p .05. Rather, to be indicative of convergent valid-ity, a value of .40 was specified. This is somewhat arbitrary but, in terms of sharedvariance, would mean that 16% of variance between the two measures of thesame construct was shared. Validity coefficients should be high enough to lendsome confidence to the assertion that the two measures share common variance.

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

Page 8: Journal of Career Assessment-2002-Betz-283-300.pdf

7/21/2019 Journal of Career Assessment-2002-Betz-283-300.pdf

http://slidepdf.com/reader/full/journal-of-career-assessment-2002-betz-283-300pdf 8/18

290 JOURNAL OF CAREER ASSESSMENT / August 2002

The convergent validity coefficients are underlined in Table 3. As shown inTable 3, convergent validity values range from .59 (Conventional) to .81(Realistic) between the SCI and SEQ, from .37 (Social) to .55 (Investigative)between the SCI and SERS, and from .38 (Conventional) to .58 (Artistic)between the SEQ and SERS. Using the arbitrary value of .40, the convergentvalidities with the SERS are slightly below that for both Social and Conventional.Thus, the SCI and SEQ are providing the most similar measurement, in com-parison to the SERS. Averaging for themes across measures, validities are highestfor Artistic and Investigative (.63) and lowest for Conventional (.45).

The second requirement of the multitrait-multimethod matrix is that the con-vergent validities are higher than either the correlations between different traitsmeasured the same way (monomethod/heterotrait, or method variance) or betweendifferent traits measured differently (heteromethod/heterotrait). These particu-lar criteria are complicated in this study by the hexagonal arrangement of Holland’sthemes, with higher associations expected between adjacent themes. It is expect-ed that these correlations would be intermediate in magnitude between the con-vergent validity correlations and those between nonadjacent Holland themes.

Thus, “heterotrait” was defined as nonadjacent Holland themes, including thefollowing nonadjacent trait pairs: Realistic/Artistic, Realistic/Social, Realistic/ Enterprising, Investigative/Social, Investigative/Enterprising, Investigative/ 

Table 1

Sample Items From Measures Examined

Measure and Subscale Item(s)

Skills Confidence Inventory

 Activities Meet new people (Social)

Design sets for a play (Artistic)

School subjects Calculus (Investigative)

Finance (Conventional)

Self-Efficacy Questionnaire

 Activities Make simple electrical repairs (Realistic)

 Write stories or poetry (Artistic)

Manage a sales campaign (Enterprising)Self-Efficacy Rating Scale

Occupational titles(Educational requirements and job duties) Veterinarian (Realistic)

Social worker (Social)

Buyer (Enterprising)

Musician (Artistic)

 Accountant (Conventional)

Chemist (Investigative)

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

Page 9: Journal of Career Assessment-2002-Betz-283-300.pdf

7/21/2019 Journal of Career Assessment-2002-Betz-283-300.pdf

http://slidepdf.com/reader/full/journal-of-career-assessment-2002-betz-283-300pdf 9/18

Betz, Gwilliam / HOLLAND THEME SELF-EFFICACY 291

Conventional, Artistic/Enterprising, Artistic/Conventional, and Social/Conventional.These pairs of correlations were used in constructing Table 4, which shows theaverage monomethod/heterotrait and heteromethod/heterotrait correlations. Asshown in the table, the mean monomethod/heterotrait (nonadjacent trait) corre-lations were substantially and significantly lower than the convergent validity cor-relations for the SCI (M = .28) and SEQ (M =.38) but not for the SER scales(M =.49). The heterotrait/heteromethod correlation for the SCI/SEQ (r = .27)was equal to the mean monomethod/nonadjacent trait correlation within theSCI, suggesting little method variance for the SCI. In contrast, the heteromethodnonadjacent trait correlations involving the SERS were smaller (.15 with theSCI and .16 with the SEQ) in comparison to their monomethod nonadjacenttrait correlations.

The second major purpose of the present study was to examine gender andrace differences in scores, and the relative reliability and validity of the invento-

Table 2

 Values of Coefficient Alpha and MeasuredStandard Deviations for Total Group and by Gender and Race

Total Sample

Coefficient (N = 397)   M SD

Skills Confidence Inventory

Realistic .86 3.18 0.80

Investigative .83 3.31 0.68

 Artistic .85 3.18 0.77

Social .86 3.82 0.63

Enterprising .85 3.42 0.67

Conventional .85 3.43 0.67Self-Efficacy Questionnaire

Realistic .91 6.36 2.2

Investigative .82 6.84 1.7

 Artistic .79 5.78 1.9

Social .81 8.23 1.3

Enterprising .86 7.34 1.6

Conventional .70 7.13 1.4

Self-Efficacy Rating Scale

Realistic (11) .78 1.26 0.48

Investigative (20) .91 1.44 0.43

 Artistic (14) .82 1.27 0.48

Social (12) .82 1.59 0.41Enterprising (17) .89 1.47 0.49

Conventional (9) .82 1.46 0.53

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

Page 10: Journal of Career Assessment-2002-Betz-283-300.pdf

7/21/2019 Journal of Career Assessment-2002-Betz-283-300.pdf

http://slidepdf.com/reader/full/journal-of-career-assessment-2002-betz-283-300pdf 10/18

Table 3Multitrait-Multimethod Matrix

SCI SEQ(Activities + College Course) (Activities) (Occu

Method R I A S E C R I A S E C R I

SCI

Realistic

Investigative .53

 Artistic .41 .26

Social .21 .17 .34

Enterprising .31 .27 .38 .48

Conventional .41 .50 .09 .16 .44

SEQ

Realistic .81 .45 .30 .15 .27 .39

Investigative .42 .72 .14 .15 .25 .54 .51

 Artistic .40 .25 .77 .32 .39 .10 .43 .24Social .20 .15 .31 .74 .50 .15 .24 .23 .39

Enterprising .23 .23 .28 .51 .79 .34 .31 .34 .43 .68

Conventional .32 .42 .20 .36 .48 .59 .44 .55 .37 .53 .63

SER 

Realistic .43 .29 .26 .09 .17 .15 .45 .31 .20 .03 .18 .19

Investigative .38 .55 .18 .06 .16 .21 .38 .54 .13 .03 .20 .30 .63

 Artistic .11 .06 .54 .13 .15 .03 .23 .15 .58 .23 .28 .15 .42 .40

Social .12 .07 .25 .37 .19 .11 .16 .11 .28 .39 .28 .16 .45 .34 .

Enterprising .15 .08 .40 .30 .42 .25 .20 .14 .36 .36 .46 .30 .51 .47 .

Conventional .24 .30 .10 .00 .27 .38 .24 .27 .26 .25 .37 .38 .48 .56 .

Note. SCI = Skills Confidence Inventory; SEQ = Self-Efficacy Questionnaire; SER = Self-Efficacy Rating Scale; R = Realistic; I = S = Social; E = Enterprising; C = Conventional. For N = 384, values of .10, .13, and .17 are statistically significant at the .05, .01, an

 2    9    2   

Page 11: Journal of Career Assessment-2002-Betz-283-300.pdf

7/21/2019 Journal of Career Assessment-2002-Betz-283-300.pdf

http://slidepdf.com/reader/full/journal-of-career-assessment-2002-betz-283-300pdf 11/18

Betz, Gwilliam / HOLLAND THEME SELF-EFFICACY 293

ries for African Americans versus European Americans. Two-way MANOVAs(Gender Race) were conducted on the six (Holland) scores from the SCI,SEQ, and SERS. Results of the MANOVA for the SCI indicated significant mul-tivariate effects for gender, F(6, 346) = 10.48, p   .001, and for race, F(6, 346) =4.47, p   .001, based on Pillai’s trace V statistic (see Haase & Ellis, 1987). Therewere no interaction effects. Results for the SEQ also indicated significant gender,F(6, 354) = 16.8, p   .001, and race effects, F(6, 354) = 3.4, p   .01. Again therewere no interaction effects. The results for the SERS, however, indicated signifi-cant effects only for gender, F(6, 343) = 11.2, p   .001. There were no effects forrace or any interaction effects.

Examining Tables 5 and 6 shows the results of univariate F tests that followedfindings of significant multivariate Fs. As shown in Table 5, men reported higher

Realistic and Investigative confidence on all four sets of scores examined herein—the SCI, SEQ, and SERS—whereas women reported higher Social confidenceon two of the three inventories (all but the SCI, where gender differences haveoccurred in previous samples). Men were more confident on the SCI Enterprisingand Conventional themes than were women. No gender differences occurred on Artistic confidence.

Comparisons of African Americans and European Americans, shown in Table6, indicate that the former group reported significantly higher confidence rela-

Table 4

Comparisons of Convergent Validity, Monomethod/Heterotrait,and Heteromethod/Heterotrait Correlations

Mean r  Minimum Maximuma

Convergent validity

SCI – SEQ .74 .59 .81

SCI – SERS .45 .37 .55

SEQ – SERS .47 .38 .58

Monomethod/nonadjacent trait

Skills Confidence Inventory .30 .09 .50 (IC)

Self-Efficacy Questionnaire .38 .23 .55 (IC)

Self-Efficacy Rating Scale .49 .34 .66 (AE)Heteromethod/nonadjacent trait

SCI/SEQ .27 .15 .42 (IC)

SCI/SERS .15 .00 .40 (AE)

SEQ/SERS .16 .11 .36 (AE)

Note. SCI = Skills Confidence Inventory; SEQ = Self-Efficacy Questionnaire; SERS = Self-Efficacy Rating Scale; I = Investigative; C = Conventional; A = Artistic; E = Enterprising.

a. Shown in parentheses are the Holland code pairs where the highest nonadjacent trait correla-tions were found.

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

Page 12: Journal of Career Assessment-2002-Betz-283-300.pdf

7/21/2019 Journal of Career Assessment-2002-Betz-283-300.pdf

http://slidepdf.com/reader/full/journal-of-career-assessment-2002-betz-283-300pdf 12/18

294 JOURNAL OF CAREER ASSESSMENT / August 2002

tive to the Artistic, Social, and Enterprising themes of the SCI and SEQ,although there were no significant race differences on the SERS.

Table 7 provides values of coefficient alpha within race and gender subgroups.Overall, there were few differences across racial and gender groups. The onlyevidence of a slight difference on the SCI was Enterprising confidence, where

the alpha was higher (.87) for males than for the other three subgroups (femalesand African American or European Americans, separately). For the SEQ, relia-bility of the Social and Conventional theses was lower in the male and, to a less-er extent, the European American samples herein than in the female or African American samples. Contrary to the SEQ, where Social and Conventional wereslightly less reliable among the male and African American samples, Social andConventional on the SERS were slightly less reliable among the female and African American samples. It should be kept in mind, though, that these differ-

Table 5

Score Means and Standard Deviations and Resultsof Univariate Analyses of Significant Multivariate Gender Effects

Males Females(N = 125) (N = 274)

Measure   M SD M SD F

Skills Confidence Inventory

Realistic 3.56 .78 2.99 .74 41.2***

Investigative 3.54 .66 3.2 .65 17.0***

 Artistic 3.2 .71 3.2 .79 0.06

Social 3.7 .64 3.8 .62 0.78

Enterprising 3.6 .70 3.3 .64 6.7**

Conventional 3.7 .61 3.3 .65 22.7***

Self-Efficacy Questionnaire

Realistic 7.6 1.9 5.8 2.1 40.3***

Investigative 7.6 1.6 6.5 1.7 24.7***

 Artistic 5.7 2.0 5.8 1.8 0.03

Social 7.9 1.3 8.4 1.3 7.6**

Enterprising 7.5 1.6 7.3 1.6 0.53

Conventional 7.2 1.3 7.1 1.5 0.004

Self-Efficacy Rating Scale

Realistic 1.4 .46 1.2 .46 23.2***

Investigative 1.5 .39 1.4 .44 7.6**

 Artistic 1.3 .49 1.3 .47 0.11Social 1.5 .49 1.6 .36 9.9**

Enterprising 1.5 .49 1.5 .49 0.02

Conventional 1.5 .55 1.4 .52 0.03

**p   .01. ***p   .001.

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

Page 13: Journal of Career Assessment-2002-Betz-283-300.pdf

7/21/2019 Journal of Career Assessment-2002-Betz-283-300.pdf

http://slidepdf.com/reader/full/journal-of-career-assessment-2002-betz-283-300pdf 13/18

Betz, Gwilliam / HOLLAND THEME SELF-EFFICACY 295

ences were small and were not statistically significant. Overall, they suggest sim-ilarly reliable scales across gender and race subgroups. For example, mean alphasacross the six subscales in each inventory were as follows: SCI (.85 for African Americans vs. .84 for European Americans), SEQ (.83 vs. .80, respectively), andSERS (.84 for both groups).

Finally, Table 8 provides summaries of the multitrait-multimethod matriceswithin African American and European American groups.1 It may be noted thatalthough the convergent validity correlations are slightly higher among African

Table 6

Means and Standard Deviationsof Scores for European and African Americans

 African European American American(N = 111) (N = 252)

Measure   M SD M SD F

Skills Confidence Inventory

Realistic 3.2 .80 3.2 .80 0.65

Investigative 3.4 .63 3.3 .69 0.15

 Artistic 3.4 .75 3.1 .74 7.7**

Social 4.0 .62 3.7 .62 9.1**

Enterprising 3.6 .67 3.3 .62 12.0***

Conventional 3.6 .70 3.3 .63 3.7

Self-Efficacy Questionnaire

Realistic 6.6 2.2 6.2 2.2 0.80

Investigative 7.1 1.7 6.6 1.7 3.05

 Artistic 6.4 1.9 5.5 1.8 16.8***

Social 8.5 1.4 8.1 1.2 4.9*

Enterprising 7.7 1.7 7.2 1.6 6.3**

Conventional 7.4 1.5 7.0 1.3 0.10

Self-Efficacy Rating Scale

Realistic 1.3 .49 1.2 .46 3.10

Investigative 1.5 .44 1.4 .42 2.5 Artistic 1.3 .46 1.3 .47 1.2

Social 1.6 .37 1.6 .40 1.6

Enterprising 1.5 .47 1.5 .49 0.10

Conventional 1.5 .50 1.5 .53 0.19

Note. The multivariate Fs for race for the Skills Confidence Inventory and Self-Efficacy Question-naire were statistically significant, permitting interpretation of univariate F statistics. Those forrace for the Self-Efficacy Rating Scale were not statistically significant, so univariate Fs should notbe interpreted.

*p   .05. **p   .01. ***p   .001.

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

Page 14: Journal of Career Assessment-2002-Betz-283-300.pdf

7/21/2019 Journal of Career Assessment-2002-Betz-283-300.pdf

http://slidepdf.com/reader/full/journal-of-career-assessment-2002-betz-283-300pdf 14/18

296 JOURNAL OF CAREER ASSESSMENT / August 2002

 Americans, so are the monomethod heterotrait values, indicative of method vari-ance. Using the  z test for the difference between independent correlations, thedifference in the mean values within the SCI (.45 for African Americans and .20for European Americans) is statistically significant (p   .01), as is that betweenthe heteromethod-heterotrait correlation for the SCI/SERS for African Americans (r = .28) versus European Americans (r = .07). Although not statisti-

cally significant, all other comparisons also indicate higher correlations among African Americans.

DISCUSSION

The present analyses of inventories of self-efficacy or confidence with respectto the six Holland themes indicated, first, that all subscales were adequately reli-

Table 7

 Values of Coefficient Alpha Within Gender and Race Subgroups

 African EuropeanMales Females American American

Scale (N = 124) (N = 273) (N = 110) (N = 274)

Skills Confidence Inventory

Realistic .87 .85 .88 .87

Investigative .83 .82 .80 .84

 Artistic .81 .85 .85 .84

Social .87 .86 .87 .86

Enterprising .87 .80 .82 .80

Conventional .83 .84 .86 .83

Self-Efficacy Questionnaire

Realistic .91 .89 .89 .91

Investigative .78 .82 .81 .83

 Artistic .82 .78 .79 .78

Social .74 .84 .86 .79

Enterprising .86 .86 .87 .85

Conventional .62 .70 .74 .65

Self-Efficacy Rating Scale

Realistic .78 .78 .80 .77

Investigative .89 .91 .92 .90

 Artistic .82 .82 .82 .82

Social .87 .78 .80 .83Enterprising .90 .89 .88 .89

Conventional .85 .80 .79 .83

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

Page 15: Journal of Career Assessment-2002-Betz-283-300.pdf

7/21/2019 Journal of Career Assessment-2002-Betz-283-300.pdf

http://slidepdf.com/reader/full/journal-of-career-assessment-2002-betz-283-300pdf 15/18

Betz, Gwilliam / HOLLAND THEME SELF-EFFICACY 297

able for research purposes (alphas above .70), but several scales were not suffi-ciently reliable for counseling uses, as suggested by a minimum alpha of .80. Theunsatisfactory scales from a counseling standpoint were the Conventional and Artistic scales of the SEQ (both of which had been judged unsatisfactory by their

authors, Lenox & Subich, 1994) and the Realistic scale of the SERS. The relia-bilities of the six SCI subscales were more consistent (.83 to .86) than the morevariable values found for the other inventories (.70 to .91 for the SEQ and .78 to.91 for the SERS). The reliabilities of the SEQ, however, are based on short (5-item) scales, whereas those of the SERS are based on from 9 (Conventional) to20 (Investigative) occupations each, with a mean scale length of 14 occupations.

Multitrait-multimethod analyses were used to examine the extent of conver-gent validity, the degree to which convergent validity coefficients were largerthan correlations of different subscales within the same inventory (monomethod-heterotrait correlations postulated to indicate the presence of method variance),and correlations of different scales assessed using different measures. These lattertwo sets of values, if lower than the convergent validity coefficients, provided evi-dence for discriminant validity. Using these criteria, there was stronger evidencefor the validity of the SCI and SEQ relative to the SERS. The average correla-tion of same-named confidence scales across the SCI and SEQ was .74 versusmean correlations of .45 and .47 when each was paired with the SERS. For theSCI and SEQ, their convergent validity correlation was significantly greater thanthe monomethod heterotrait correlations indicating method variance (.30 and.38 for the SCI and SEQ, respectively). In contrast, the mean convergent validi-ty correlations of the SERS (.45 with the SCI and .47 with the SEQ) were not

Table 8

Comparisons of Correlations for African and European Americans

 African Americans European Americans

Convergent Validity

SCI – SEQ .76 .72

SCI – SERS .51 .45

SEQ – SERS .51 .42

Monomethod Nonadjacent Trait

SCI .45 .20

SEQ .47 .32

SERS .53 .48

Heteromethod Nonadjacent TraitSCI – SEQ .36 .19

SCI – SERS .28 .07

SEQ – SERS .29 .13

Note. SCI = Skills Confidence Inventory; SEQ = Self-Efficacy Questionnaire; SERS = Self-Efficacy Rating Scale.

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

Page 16: Journal of Career Assessment-2002-Betz-283-300.pdf

7/21/2019 Journal of Career Assessment-2002-Betz-283-300.pdf

http://slidepdf.com/reader/full/journal-of-career-assessment-2002-betz-283-300pdf 16/18

298 JOURNAL OF CAREER ASSESSMENT / August 2002

significantly different from the method variance correlations (M = .49). The con-

vergent validity correlations for the SERS were, however, greater than the het-erotrait-heteromethod correlations, providing some (though less convincing) evi-dence for construct validity.

In addition to examination of the characteristics of these inventories in thetotal group, examination of their psychometric characteristics within the samplesof African American and European American samples was of interest. Alpha reli-abilities were as high or slightly higher for African Americans as they were forEuropean Americans; comparisons of .85 versus .84, .83 versus .80, and .84 forboth groups characterized mean alphas for African Americans versus European Americans on the SCI, SEQ, and SERS. The results also suggested a relative lackof race differences on measures of confidence with respect to the Holland themesand, in this sample, at least suggest that African Americans may feel somewhatgreater confidence than European Americans relative to the Artistic, Social, andEnterprising dimensions of Holland’s vocational theory. However, these differ-ences occurred only on the inventories using activities and school subjects ratherthan on that using occupational titles.

Evidence for convergent validity indicated comparability between the tworacial groups studied herein, but the use of the multitrait-multimethod matrix sug-gested lower levels of discriminant validity and greater method variance for the African American versus European American sample. Monomethod-heterotraitcorrelations for African Americans ranged from .45 to .53 across the three inven-tories, indicating probably greater relevance of method variance to the resultingscores. And in fact, although the convergent validity correlations for the SCI and

SEQ were significantly greater than their method variance values (monomethod/ heterotrait), the convergent validities of pairs including the SERS (SCI-SERSand SEQ-SERS) were not different from the method variance values within the African American sample.

The present study provides several findings that suggest the utility of the mul-titrait-multimethod matrix. First, the analyses suggested that item type (whetheractivities and school subjects, as in the SEQ and SCI) versus occupational titlesdoes influence self-perceived confidence for the six Holland theme areas. Notonly was this finding supported by the convergent validity correlations, but alsothe presence of race differences (in favor of African Americans) when activitiesitems versus occupations items were used is of interest. Furthermore, the multitrait-multimethod analysis suggested some problems in the discriminant validity of 

these inventories when used with African Americans. Whether this suggests thatself-efficacy is a more generalized characteristic among African Americans thanEuropean Americans warrants further study.

 A finding consistent with previous research (see Gasser et al., 2001) is that thehighest nonadjacent trait correlations occur between Investigative andConventional, especially as the content of the latter theme more consistentlyincludes computer activities. For example, the correlations between Investigativeand Conventional herein were .50 (SCI), .55 (SEQ), and .56 (SER), which were

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

Page 17: Journal of Career Assessment-2002-Betz-283-300.pdf

7/21/2019 Journal of Career Assessment-2002-Betz-283-300.pdf

http://slidepdf.com/reader/full/journal-of-career-assessment-2002-betz-283-300pdf 17/18

Betz, Gwilliam / HOLLAND THEME SELF-EFFICACY 299

actually higher than the majority of convergent validity correlations (i.e., the

same trait) using the SERS. These were the highest nonadjacent trait correla-tions found of nine such nonadjacent trait pairs.

 Although not the focus of the study, gender comparisons of scores were alsodone, using MANOVA followed by post hoc univariate tests. These analyses indi-cated gender differences consistent with previous research on confidence andself-efficacy with respect to the Holland activity areas, with males reporting con-sistently higher confidence on Realistic and Investigative on all four inventoriesand greater Enterprising and Conventional confidences on the SCI. Femalesreported higher confidence on Social. Because self-efficacy measures are ideallyused in practice to target individuals for interventions that can increase self-efficacy and, thus, vocational options (see Betz, 1999), score patterns herein con-tinue to suggest the inadequacies in confidence among college-age women.

In summary, the present study has used multitrait-multimethod matrices andother psychometric criteria to provide useful information about the psychomet-ric quality and utility of three inventories of confidence with respect to theHolland vocational themes. Further research might be directed at further explor-ing the confidence differences in favor of African Americans and, at the sametime, the questionable degree of discriminant validity across Holland dimensionsfor the African American sample. Findings that convergent validity is stronglyinfluenced by the type of item used should also be carefully noted in otherresearch of this type.

NOTE

1. Because of space limitations, we have omitted the full multitrait-multimethod matrices with-in African American and European American subgroups. They may, however, be obtained fromthe first author.

REFERENCES

Bandura, A. (1977). Self-efficacy: Toward a unifying theory of behavioral change. PsychologicalReview, 84, 191-215.

Bandura, A. (1986). Social foundations of thought and action: A social cognitive theory. Englewood

Cliffs, NJ: Prentice Hall.Betz, N. E. (1999). Getting clients to act on their interests: Self-efficacy expectations as a mediator

of the implementation of vocational interests. In M. L. Savickas & A. R. Spokane (Eds.),Occupational interests: Their meaning, measurement, and use in counseling (pp. 327-344). Palo

 Alto, CA: Davies-Black.Betz, N. E. (2000). Self-efficacy theory as a basis for career assessment. Journal of Career 

 Assessment, 8, 205-222.Betz, N. E., & Borgen, F. H. (2000). Integrating vocational interests with self-efficacy and person-

al styles. Journal of Career Assessment, 8, 329-338.

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

Page 18: Journal of Career Assessment-2002-Betz-283-300.pdf

7/21/2019 Journal of Career Assessment-2002-Betz-283-300.pdf

http://slidepdf.com/reader/full/journal-of-career-assessment-2002-betz-283-300pdf 18/18

300 JOURNAL OF CAREER ASSESSMENT / August 2002

Betz, N. E., Borgen, F. H., & Harmon, L. W. (1996). Skills Confidence Inventory: Applications and

technical guide. Palo Alto, CA: Consulting Psychologists Press.Betz, N. E., Borgen, F. H., Kaplan, A., & Harmon, L. W. (1998). Gender as a moderator of the

validity and interpretive utility of the Skills Confidence Inventory. Journal of VocationalBehavior , 53, 1-19.

Betz, N. E., & Hackett, G. (1981). The relationship of career-related self-efficacy expectations toperceived career options in college women and men. Journal of Counseling Psychology, 28, 399-410.

Betz, N. E., Harmon, L., & Borgen, F. (1996). The relationships of self-efficacy for the Hollandthemes to gender, occupational group membership, and vocational interests. Journal of Counseling Psychology, 43, 90-98.

Campbell, D. T., & Fiske, D. W. (1959). Convergent and discriminant validation by the multitrait-multimethod matrix. Psychological Bulletin, 56, 81-105.

Donnay, D.A.C., & Borgen, F. H. (1999). The incremental validity of vocational self-efficacy: Anexamination of interest, self-efficacy, and occupation. Journal of Counseling Psychology, 46,432-447.

Fouad, N. A. (1993). Cross cultural vocational assessment. Career Development Quarterly, 42, 4-13.Gasser, C. E., Paulsen, A., Thomas, T., Hofer, K., Betz, N., & Rottinghaus, P. (2001, August).

Expanding the Skills Confidence Inventory with basic confidence scales. Poster presented at theannual convention of the American Psychological Association, San Francisco.

Haase, R. F., & Ellis, M. V. (1987). Multivariate analyses of variance. Journal of CounselingPsychology, 34, 404-413.

Harmon, L., Borgen, F., Bereth, J., King, J., Schauer, D., & Ward, C. (1996). The SkillsConfidence Inventory: A measure of self-efficacy. Journal of Career Assessment, 4, 457-477.

Holland, J. L. (1985). Making vocational choices: A theory of vocational personalities and work envi-ronments (2nd ed.). Englewood Cliffs, NJ: Prentice Hall.

Holland, J. L. (1997). Making vocational choices: A theory of vocational personalities and work envi-ronments (3rd ed.). Odessa, FL: Psychological Assessment Resources.

Isaacs, J., Borgen, F., Donnay, D., & Hansen, T. A. (1997, August). Self-efficacy and interests:Relationships of Holland themes to college major . Poster presented at annual meeting of the

 American Psychological Association, Chicago.Lapan, R. T., Boggs, K. R., & Morrill, W. H. (1989). Self-efficacy as a mediator of investigative and

realistic general occupational themes on the Strong-Campbell Interest Inventory. Journal of Counseling Psychology, 36(2), 176-182.

Lapan, R. T., Shaughnessy, P., & Boggs, K. (1996). Efficacy expectations and vocational interestsas mediators between sex and choice of math/science college majors. Journal of VocationalBehavior , 49, 277-291.

Lenox, R. A., & Subich, L. M. (1994). The relationship between self-efficacy beliefs and invento-ried vocational interests. Career Development Quarterly, 42, 302-313.

Lent, R. W., Brown, S. D., & Hackett, G. (1994). Toward a unifying social cognitive theory of careerand academic interest, choice, and performance. Journal of Vocational Behavior , 45, 79-122.

Lent, R. W., Brown, S. D., & Hackett, G. (2000). Contextual supports and barriers to career choice: A social cognitive analysis. Journal of Counseling Psychology, 47, 36-49.

Nunnally, J., & Bernstein, I. H. (1994). Psychometric theory (3rd ed.). New York: McGraw-Hill.

Parsons, E., & Betz, N. E. (1998). Test-retest reliability and validity studies of the Skills ConfidenceInventory. Journal of Career Assessment, 6(1), 1-12.

Samuda, R. (1998). Psychological testing of American minorities. (2nd ed.). Thousand Oaks, CA: Sage.Subich, L. M. (1996). Addressing diversity in the process of career assessment. In M. L. Savickas

& W. B. Walsh (Eds.), Handbook of career counseling theory and practice (pp. 277-290). Palo Alto, CA: Davies-Black.

Tracey, T.J.G., & Hopkins, N. (2001). Correspondence of interests and abilities with occupationalchoice. Journal of Counseling Psychology, 48, 178-190.