Social cognitive predictors of adjustment to engineering

majors across gender and race/ethnicity

Robert W. Lent
a,
, Matthew J. Miller
a
, Paige E. Smith
a
, Bevlee A. Watford
b
, Robert H. Lim
a
,
Kayi Hui
a
, M. Ashley Morrison
a
, Gregory Wilkins
c
, Kevin Williams
d
a
University of Maryland, USA
b
Virginia Tech University, USA
c
Morgan State University, USA
d
Tennessee State University, USA
a r t i c l e i n f o a b s t r a c t
Article history:
Received 8 January 2013
Available online 20 February 2013
We tested a social cognitive model of academic adjustment ina sample of 1377 students enrolled in
engineering schools at two predominantly White and two historically Black state universities. The
model brought together central elements of social cognitive career theory's (SCCT) segmental
models of educational/vocational satisfaction, interest, choice, and performance/persistence (Lent
&Brown, 2006; Lent, Brown, &Hackett, 1994). The model was specifically designed to examine the
interplay between interest and satisfaction in predicting students' intentions to remain in
engineering majors. The integrative model offered good fit to the data bothin the larger sample and
in sub-samples of women, men, and racial/ethnic majority and minority students. The relation of
interests to intended persistence was fully mediated by satisfaction. The findings both extend
research on SCCT and suggest linkages to theories of person-environment fit. Implications for
further research and practice on academic adjustment and persistence in STEM fields are
considered.
2013 Elsevier Inc. All rights reserved.
Keywords:
Self-efficacy
Outcome expectations
Support
Interests
Satisfaction
Persistence
1. Introduction
Vocational psychology researchers have long been concerned with factors that promote or impede STEM career paths and, in
recent years, social cognitive career theory (SCCT; Lent, Brown & Hackett, 1994, 2000) has offered one theoretical platform for
such inquiry. SCCT originally consisted of three segmental models focusing on (a) how basic academic and career interests
develop, (b) how educational and career choices are made, and (c) what factors affect academic and career performance and
persistence. Based on Bandura's (1997) general social cognitive theory, these models involve overlapping variables and processes.
A fourth model was recently added to help explain the factors responsible for educational and occupational satisfaction and other
aspects of positive adjustment to school and work contexts (Lent & Brown, 2006, 2008). The latter model specifically focuses on
the interplay between cognitive, behavioral, contextual, and personality factors within the process of educational and vocational
adjustment.
A number of studies have generally found good support for SCCT's basic interest, choice, and performance models in middle
school (Fouad & Smith, 1996), high school (Lopez, Lent, Brown, & Gore, 1997), and college students (Ferry, Fouad, & Smith, 2000).
Focusing specifically on college students in STEM fields, findings indicate that SCCT helps to account for the interests and major
choice goals of racially diverse female and male students in engineering (Lent, Brown et al., 2003, 2005), biological/life sciences
Journal of Vocational Behavior 83 (2013) 2230
This article is based upon the work supported by the National Science Foundation under grant no. 0827470.
Corresponding author at: Department of Counseling, Higher Education, and Special Education, University of Maryland, College Park, MD 20742, USA.
E-mail address: boblent@umd.edu (R.W. Lent).
0001-8791/$ see front matter 2013 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.jvb.2013.02.006
Contents lists available at SciVerse ScienceDirect
Journal of Vocational Behavior
j our nal homepage: www. el sevi er . com/ l ocat e/ j vb
(Byars-Winston, Estrada, Howard, Davis, & Zalapa, 2010), and computing disciplines (Lent, Lopez, Lopez, & Sheu, 2008; Lent,
Lopez, Sheu, & Lopez, 2011).
Although SCCT's segmental models eachfocus on a particular class of academic and career outcomes, the models were designed to
overlap with one another, for instance, by drawing on a common core of social cognitive elements, such as self-efficacy, outcome
expectations, and contextual supports. The interest and choice models have frequently been studied in tandem(Sheu et al., 2010), in
part because interests are hypothesized to help predict goals. The satisfaction model has, however, been studied apart fromthe other
SCCT models, and only one study has thus far tested it in a STEM sample (Lent, Singley, Sheu, Schmidt, & Schmidt, 2007). Although
academic/work satisfaction may be considered an important outcome in its own right, theorists have also been interested in its
consequences, such as its linkage to decisions to remain in, or leave, particular educational or vocational paths (Dawis, 2005).
The current study was designed to combine elements of SCCT's segmental models into an integrative or holistic model, one that
specifically examines the interplay among interest, satisfaction, and choice variables. It was also intended to build on prior research
testing SCCT with STEMmajors by examining the adequacy of the integrative model as a function of gender and race/ethnicity. While
it is useful to test SCCT in samples of STEM students in general, group-specific tests are necessary to determine the range of the
theory's cross-gender and cross-cultural validity and, potentially, to identify theory-derived methods for engaging and retaining more
diverse persons in STEM fields.
1.1. An integrative model of interest, satisfaction, and choice stability
Interests and satisfaction are both regarded as important constructs in vocational psychology, and the linkage between them has
long intrigued scholars (Hansen, 1984). For example, Holland's (1997) theory posits that individuals will be satisfied with their
occupations to the extent that their interests are congruent with those of others in their work environment. The theory of work
adjustment also acknowledges that interests are one source of work satisfaction (Dawis, 2005), though the theory views interests as
less central to satisfaction than is the correspondence between one's values and the reinforcers provided by his or her work
environment. SCCT focuses on interests and satisfaction in distinctive segmental models, though both variables are assumed to be
informed by an overlapping set of predictors (e.g., outcome expectations, self-efficacy). It, thus, seems reasonable to investigate the
nature of the relationship between interests and satisfaction from a social cognitive perspective, just as prior research has examined
the interface between interests and choice in SCCT.
Following prior vocational inquiry, in SCCT, interests are defined as the extent to which people like particular activities that are
available within given school or work contexts (e.g., I like algebra, I like fixing mechanical things). Satisfaction is defined as
one's overall enjoyment of the school or work context (e.g., I am happy with my job). That is, interests usually refer to specific
behavioral activities, whereas satisfaction refers to affinity for the larger settings in which these activities are situated. When
Environmental
Supports &
Resources
R
2
= .31
Outcome
Expectations
R
2
= .44
Satisfaction
R
2
= .58
Persist
R
2
= .44
Self-Efficacy
Expectations
R
2
= .50
Personality Traits &
Affective Dispositions
Interest
R
2
= .47
(
i
)

.
2
3
*
(a) .33*
(b) .03
(f) .0
9
*
(
g
)
.0
9
*
(h) .22*
(
d
)

.
4
4
*
(
e
)

.
3
2
*

(
l
)

.
3
3
*
(
m
)

.
2
3
*
(n)
.50*
(k) .56*
(j) .47*
(o) .32*
(
p
)
.1
4
*
(c) .30*
(q) -.05
*p < .05, 1-tailed.
Fig. 1. Structural path coefficients for the full sample.
23 R.W. Lent et al. / Journal of Vocational Behavior 83 (2013) 2230
people appraise their overall satisfaction with their work (or school) lives, they are likely to consider an array of factors, which
may include interest in their work tasks as well as other considerations, such as the quality of their interpersonal relations at work
or the adequacy of the compensation they receive. Thus, the opportunity to engage in interesting activities offers one of several
possible routes to school and work satisfaction. That it is not the only such route may help to explain the generally modest
relation of interest congruence to satisfaction in research on Holland's (1997) theory (Nauta, 2013).
In the current study, we examined the interplay between interests and satisfaction relative to intentions to persist in engineering
majors. Persistence is relevant both to SCCT's choice and performance models in that persistence reflects choice stability that is,
whether and how long people choose to remain with an initial choice (or are permitted to do so by environmental gatekeepers).
Hence, persistence may be seen as a reflection of both choice and performance (e.g., good performers are more likely to want to
remain, and to be allowed to remain, in a given academic or work environment; Dawis, 2005).
For the purposes of the current integrative model of interest and satisfaction (see Fig. 1), we maintain that pursuit of one's
interests is likely to serve as a precursor to satisfaction with one's academic or job context (path a). For instance, exposure to an
academic environment that invites expression of one's educational/vocational interests should lead to academic satisfaction. Both
interests (path b) and satisfaction (path c) should, in turn, predict academic or vocational persistence because people tend to
prefer to keep pursuing activities that they like and to remain in environments that are satisfying. Findings support the linkage of
interests to persistence (e.g., Lent, Brown et al., 2003, 2005) and of satisfaction to persistence (Dawis, 2005). What is less clear is
whether the relationship of interests to persistence is fully or partially mediated by satisfaction (i.e., whether satisfaction explains
the linkage of interests to persistence).
The remainder of the variables in the integrative model, and the paths between them, are drawn largely from SCCT's interest,
choice, and satisfaction models. In particular, interest is seen as a consequence of self-efficacy and outcome expectations (paths d and
e, respectively); that is, people tend to become interested in school and work activities when they feel efficacious at performing them
and anticipate that they will lead to valued outcomes. Satisfaction with one's academic or work life is, in addition to pursuit of one's
interests, likely to result from favorable appraisals of self-efficacy, outcome expectations, and contextual supports (paths f though h,
respectively). Trait positive affect is seen as contributing to academic/work satisfaction (path i) as well as to perceptions of
self-efficacy and environmental support (paths j and k, respectively).
Environmental support is posited to be a precursor of both self-efficacy and outcome expectations, and self-efficacy is also
assumed to predict outcome expectations (paths l, m, and n, respectively) (Sheu et al., 2010). Intended persistence is expected to
be predicted by self-efficacy, outcome expectations, and environmental support (paths o, p, and q, respectively), in addition to
interest and satisfaction (paths b and c). Positive affect is assumed to relate to intended persistence indirectly, through its linkages
to other variables in the model.
In the current study, we examined the fit to the data of the interestsatisfactionchoice stability model in a sample of students
planning to major in engineering. The study represents a cross-sectional part of a larger longitudinal (3-year) project aimed at
examining the unfolding academic adjustment process in engineering majors. The current report focuses on students' experiences
during the early stages of their exposure to the STEM academic environment, a time when many are weighing decisions about
whether to remain in or leave this environment. The overall goal of the project is to explore the interplay among theoretical
elements that may promote positive adjustment (defined mainly in terms of satisfaction and choice persistence) over time
elements that may potentially inform interventions designed to attract women and racial/ethnic minority students to, and retain
them within, STEM educational paths.
2. Method
2.1. Participants
Participants were 1377 students enrolled in introductory engineering courses at four state universities. Three of the universities
were located in the mid-Atlantic region and one was located in the Southeastern region of the U.S. Two were predominantly White
institutions (PWIs) and two were historically Black colleges/universities (HBCUs). Thirty-three percent (n=456) of the participants
were women and 67% (n=918) were men (three students did not identify their gender). Given the disparities in university sizes,
most of the participants (87%) were students at the PWIs. Ninety seven percent of the participants were first-year students; the mean
age was 18.42, SD=1.96. In terms of race/ethnicity, 58% were European American, 15% were African American, 20% were Asian
American, and 4% were Hispanic; the remaining 3% listed other racial/ethnic designations or did not report their race/ethnicity. The
students expressed their intent to major in a variety of engineering disciplines, the most frequently mentioned of which were
mechanical (18%), aerospace (11%), civil (11%), chemical (8%), and electrical (7%) engineering. Mean self-reported mathematics SAT
scores were 674.98 (SD=75.04), which are comparable to the scores reported in prior engineering student samples in this line of
research (e.g., Lent et al., 2003, 2007).
2.2. Procedure and instruments
Students were recruited for participation in the study via email messages sent by faculty members/research partners at each of
the universities. In return for their participation, students received a $15 gift certificate, redeemable at an online retail site.
Students were asked to provide their email addresses so that they could be sent the gift certificate as well as invitations to
participate in the subsequent three (longitudinal) waves of data collection. Participant recruitment and measure completion
24 R.W. Lent et al. / Journal of Vocational Behavior 83 (2013) 2230
occurred in the last three weeks of a fall semester so that first-year students could base their survey responses on a reasonable
sample of exposure to STEM coursework. The data were gathered from two cohorts of students in successive academic years.
The online survey presented demographic and academic status questions, along with measures of trait positive affect and
academic (engineering) domain-specific self-efficacy, coping efficacy, outcome expectations, social support, interest, satisfaction,
and intended persistence. (The survey also contained additional measures and open-ended questions regarding students'
academic experiences as part of the larger project's mixed-methods design.) Although the social cognitive measures have been
used in prior research, the faculty research partners at each participating university were invited to reviewand suggest changes to
these measures to ensure that the item content adequately reflected their engineering curricula. Given the adoption of several
new self-efficacy, outcome expectation, and interest items, we subjected all of the items from the social cognitive measures to
exploratory factor analyses to examine their dimensionality prior to testing the structural model shown in Fig. 1. (A summary of
these analyses can be obtained from the first author.) The revised measures produced adequate internal consistency reliability
estimates, with most values above .80 (see Table 1).
2.2.1. Self-efcacy
Self-efficacy was assessed with two measures, an academic milestones scale and a barrier-coping scale. The milestones
self-efficacy scale included four items adapted from Lent et al. (1986). The coping efficacy measure included the 7-item measure
developed by Lent et al. (2003), plus an additional item recommended by our engineering faculty research team. The milestones
scale asked students to indicate their confidence in their ability to perform well academically in engineering (e.g., how much
confidence do you have in your ability to excel in your engineering major over the next semester). On the coping efficacy
measure, participants indicated their confidence in their ability to cope with a variety of barriers, or problems, that engineering
students could potentially experience (e.g., cope with a lack of support from professors or your advisor). Self-efficacy ratings
were obtained on a 10-point scale, ranging from no confidence (0) to complete confidence (9). Scores on each scale were calculated
by dividing the summed item responses by the number of scale items, producing a possible score range of 09. The milestones
scale and the original 7-itemversion of the coping efficacy scale have produced adequate internal consistency reliability estimates
(e.g., coefficient alpha values of .90 and above) in several prior studies and have also shown theory-consistent relations with
measures of outcome expectations, interest, satisfaction, contextual support, and persistence intentions (e.g., Lent et al., 2003,
2007).
2.2.2. Outcome expectations
The outcome expectations measure contained 12 positive outcomes that could result fromearning a BS degree in engineering. The
scale was based on one used in prior research with engineering students (e.g., Lent et al., 2003, 2007). Participants indicated how
strongly they agreed that an engineering degree would lead to each outcome, using a 10-point scale, from strongly disagree (0) to
strongly agree (9). Item responses were summed and divided by 12. The original version of the scale has yielded adequate internal
consistency reliability estimates (.88 and above). Although it has typically correlated as expected with measures of self-efficacy,
interests, and major choice goals in prior research, it has not always accounted for unique predictive variance in multivariate analyses
(e.g., Lent, Brown et al., 2003, 2005). Given the possibility that the scale's items represent a limited range of outcomes associated with
engineering, we decided to supplement it by adapting items from the Work Importance Profiler (U.S. Department of Labor,
Employment and Training Administration, 2002), an online version of the Minnesota Importance Questionnaire (MIQ). The new
items reflect value statements not contained in the original outcome expectations scale. The MIQassesses work needs/values that can
be formatted as outcome expectations. Our exploratory factor analysis indicated that the items of the revised measure loaded on two
factors, reflecting extrinsic outcomes (5 items, e.g., earn an attractive salary) and intrinsic outcomes (7 items, e.g., do work that
gives me a feeling of accomplishment) associated with becoming an engineer.
2.2.3. Environmental supports
Environmental support/resources for students' academic progress in engineering were assessed with a 7-item scale. Lent et al.
(2007) had adapted a generic measure of academic support developed by Lent, Singley et al. (2005) to the context of engineering. The
measure presented participants with a list of conditions that may support your progress in engineering. A sample itemwas, [I] get
Table 1
Means, standard deviations, and correlations among predictor and criterion variables.
Variable 1 2 3 4 5 6 7 8 M SD
1. Milestone self-efficacy 7.97 1.42 .88
2. Coping efficacy .64 7.81 1.33 .89
3. Outcome expec-intrinsic .43 .51 8.54 1.24 .93
4. Outcome expec-extrinsic .32 .38 .61 8.93 .97 .85
5. Social support .39 .44 .46 .33 3.84 .74 .82
6. Interests .40 .42 .52 .27 .36 3.92 .67 .80
7. Satisfaction .47 .46 .54 .27 .52 .54 3.92 .83 .91
8. Intended persistence .45 .45 .49 .30 .35 .42 .53 4.61 .78 .94
9. Positive affect .44 .47 .51 .33 .48 .51 .58 .42 3.82 .64 .89
Note. N=1377. All correlations are significant, pb.001.
25 R.W. Lent et al. / Journal of Vocational Behavior 83 (2013) 2230
encouragement frommy friends for pursuing anengineering major. Participants were asked to indicate their level of agreement with
each statement on a scale from1 (strongly disagree) to 5 (strongly agree). Itemresponses were summed and divided by 7. The generic
(Lent, Singley et al.) and engineering-specific (Lent et al.) versions have produced alpha coefficients of .81 to .84 and correlated as
expected with corresponding measures of self-efficacy, outcome expectations, and academic satisfaction. In the current study, we
employed a 7-item version of the scale because our exploratory factor analysis revealed that two of the original 9 items composed a
separate (decisional support) factor.
2.2.4. Interests
This measure asked participants to indicate their degree of interest in performing six engineering-related activities. Responses
were obtained on a 5-point scale, from very low interest (1) to very high interest (5). Item responses were summed over items and
divided by 6. Lent, Brown et al. (2003, 2005) reported that a 7-itemversion of this scale produced coefficient alphas of .80 to .83 as
well as theory-consistent relations with measures of self-efficacy, outcome expectations, and major choice goals. Based on
suggestions from our engineering research team members, we added two items to this scale in the current study to capture
engineering activities not reflected in the original version (e.g., working on an engineering project team). Our factor analytic
findings led us to the decision to retain six items, all of which loaded highly on a dimension reflecting engineering-related
activities (solving complicated technical problems).
2.2.5. Satisfaction
Academic satisfaction was assessed with a 6-item scale asking participants to indicate the degree to which they felt satisfied
both overall (e.g., I am generally satisfied with my academic life in engineering) and with specific aspects of their academic
experience (e.g., intellectual stimulation) in engineering. The scale had been adapted to the context of engineering by Lent et al.
(2007) froma generic (non-specified major) version developed by Lent, Singley et al. (2005); for example, the termengineering
replaced general references to one's intended major. Responses were obtained along a 5-point (strongly disagree to strongly agree)
scale. Summed itemresponses were divided by 6. The original and engineering-specific versions of the scale have yielded internal
consistency estimates of .86 and above and have correlated in theory consistent ways with intended academic persistence, life
satisfaction, and a variety of social cognitive variables (Lent et al., 2007). We omitted one of the scale's original seven items due to
its cross-loading with another factor.
2.2.6. Positive affect
We used the Positive Affect (PA) scale of the Positive and Negative Affect Schedule (Watson, Clark, & Tellegen, 1988) to assess
the tendency to experience positive emotions across situations. Participants were asked to indicate the extent to which they
generally feel 10 positive emotions (e.g., enthusiastic) along a 5-point continuum (1=very slightly or not at all; 5=extremely). To
calculate scale scores, itemratings were summed and divided by 10. PA has been found to yield adequate internal consistency and
stability coefficients and to correlate as expected with measures of domain and life satisfaction (e.g., Lent, Singley et al., 2005).
2.2.7. Intended persistence
Students' intention to persist in engineering was assessed with a 4-item scale developed by Lent et al. (2003). Participants were
asked to indicate their level of agreement with each statement (e.g., I plan to remain enrolled in an engineering major over the next
semester) along a 5-point scale, ranging from strongly disagree (1) to strongly agree (5). Summed item responses were divided by 4.
The scale has yielded internal consistency reliability estimates of .93 and above (Lent, Brown et al., 2003, 2005) and been found to be
strongly predictive of actual future persistence in engineering (Lent et al., 2003). Because the scale scores were observed to be highly
(negatively) skewed and kurtotic, we used a rank transformation to create more normally distributed scores; the ranks were
subsequently converted to z scores for use in the model testing analyses.
3. Results
Table 1 presents the means, standard deviations, intercorrelations, and internal consistency reliability estimates of the measured
variables in the full sample (N=1377). To explore possible mean differences as a function of cohort, type of university, and gender,
we ran a 2 (first vs. second cohort)2 (HBCU vs. PWI)2 (women vs. men) multivariate analysis of variance. This analysis yielded
significant (pb.01) main effects for cohort (F [9, 1358]=2.49, Wilks' lambda=.98;
2
=.02), university type (F [9, 1358]=6.16,
Wilks' lambda=.96;
2
=.04), and gender (F (9, 1358)=3.42, Wilks' lambda=.98;
2
=.02). The two and three-way interactions
were, however, non-significant. Follow-up univariate analyses of variance on the main effects revealed that the effect sizes of the
statistically significant mean differences were uniformly small (none of the
2
values exceeded.02). We, therefore, combined the data
over the grouping variables in the full sample analyses.
We next conducted a series of structural equation modeling analyses to test the integrative model. These analyses employed
multiple observed indicators of each latent construct to control for measurement error. Self-efficacy (academic milestone and
coping self-efficacy scale scores) and outcome expectations (intrinsic and extrinsic outcome scale scores) were each represented
by two indicators. Item parcels were used to create multiple indicators for each of the other constructs (Bentler & Wu, 2005).
Interests, satisfaction, supports, and positive affect were each represented by three item parcel scores, with 23 items per parcel.
Persistence intentions were indexed by two parcels of 2 items apiece. Variance/covariance matrices (which can be obtained from
the first author) were used as the input data for each model test.
26 R.W. Lent et al. / Journal of Vocational Behavior 83 (2013) 2230
We used three primary indices to assess model-data fit: the comparative fit index (CFI), the root mean squared error of
approximation (RMSEA), and the standardized root mean squared residual (SRMR). CFI values.90 (Hoyle & Panter, 1995), RMSEA
values.08 (Browne & Cudeck, 1992), and SRMR values.08 (Hu & Bentler, 1999) may be considered as indicating acceptable levels
of fit, though higher levels of CFI (.95) and lower levels of RMSEA (.06) are generally preferred. Because of multivariate
non-normality (Mardia's normalized estimate>5), robust maximum likelihood estimation was employed in calculating the CFI and
RMSEA indices.
Tests of the structural paths shown in Fig. 1 were preceded by measurement model tests to confirm that the constructs in the
integrative model represent separate but correlated latent dimensions. In testing the measurement model, one factor loading for
each construct was fixed to 1; all other loadings and paths among the latent constructs were freely estimated. Assuming adequate
fit of the measurement model, structural models tests were run to assess the extent to which the hypothesized paths among the
constructs were consistent with the data.
We used two indices to compare the fit of alternative models: the scaled difference in SatorraBentler (SB)
2
statistic (Satorra &
Bentler, 2001) and the difference in CFI values between models. The SB
2
tests for statistically significant differences between
nested models. However, it has been found to be affected by sample size and model complexity (Cheung & Rensvold, 2002). We,
therefore, also examined the CFI statistic as an indicator of the practical significance of differences in model fit, using CFI
differences>.01 as the criterion for concluding that two models differ to a meaningful degree (Cheung &Rensvold, 2002). We judged
alternative models as invariant in cases where model fit differences did not reach the criterion for practical significance.
3.1. Full sample analysis
A seven-factor measurement model (self-efficacy, outcome expectations, interests, satisfaction, support, intended persistence,
positive affect) produced acceptable fit to the data in the full sample: CFI =.959, RMSEA=.049, SRMR=.033, SB scaled
2
(114,
N=1377)=494.56, pb.001. The structural model also yielded adequate fit, CFI =.951, RMSEA=.053, SRMR=.045, SB scaled

2
(118, N=1377)=576.14, pb.001. Although the structural model did not fit the data as well as the measurement model using
the SB
2
criterion (SB
2
[4] =61.94, pb.001), the difference in fit between the two models was small in a practical sense
(CFI =.008), suggesting that the structural model offered an adequate explanation of the relations among the factors.
As shown in Fig. 1, support was found for nearly all of the hypothesized paths. The only non-significant coefficients were for the
direct paths from interests and supports to intended persistence. These non-significant paths suggest that interests were linked to
intended persistence indirectly through satisfaction, and that the relation of support to intended persistence was mediated by
self-efficacy, outcome expectations, and satisfaction. On balance, interest, satisfaction, and intended persistence were each
well-predicted in the model (R
2
=.47, .58, and .44, respectively). A trimmed model, omitting the two non-significant paths, did not
differ significantly fromthe target structural model, CFI =.951, RMSEA=.053, SRMR=.045, SB scaled
2
(120, N=1377)=578.79,
pb.001, SB
2
[2]=2.53, p>.05, CFI =.00. Students were more likely to express satisfaction with the engineering domain to the
extent that they felt efficacious at and interested in engineering activities, held favorable outcome expectations about becoming an
engineer, perceived support for pursuing this major, and generally experienced positive affect. They were more likely to want to
persist in engineering if they were satisfied with the environment and reported higher self-efficacy and favorable outcome
expectations. The paths predicting interest, self-efficacy, outcome expectations, and environmental supports were consistent with
hypotheses.
3.2. Multiple-group analyses
To examine model invariance across gender and race/ethnicity, we conducted multiple-groups analyses. We began by constraining
indicator-factor loadings to be equal across each grouping variable (e.g., gender) and comparing the fit of this constrained model with a
model in which the factor loadings were allowed to vary by group. Where the two measurement models (constrained and
unconstrained) didnot yieldsubstantially different fit, we assumedthat the constructs were being measuredcomparably across groups.
We then proceeded to assess the invariance of the structural paths across groups. This was done by comparing the fit of a model in
which the values of both the factor loadings and the structural paths were constrained to equality across groups against a model in
which the factor loadings were constrained to equivalence but the structural paths were allowed to vary by group. Results of all of the
multiple-group analyses are presented in Table 2.
3.2.1. Model t by gender
The structural model, testing the hypothesized paths among constructs, achieved adequate fit in each group; for women, CFI =
.947, RMSEA=.054, SRMR=.048, SB scaled
2
(118, N=456)=277.40, pb.001; for men, CFI =.954, RMSEA=.052, SRMR=.046,
SB scaled
2
(118, N=918)=408.12, pb.001. The analyses accounted for large amounts of the predictive variance in satisfaction
and intended persistence in both genders but somewhat more so in the female than male sub-samples (for satisfaction, R
2
=.67 in
women and .54 in men; for persistence, R
2
=.56 in women and .39 in men). The multisample analysis by gender indicated that the
constrained measurement model differed in fit from the unconstrained model on the statistical (SB
2
) but not on the practical
criterion (CFI) of fit comparison, suggesting measurement equivalence across gender. The structural model with constraints on all
paths did not differ in fit from the unconstrained structural model on either criterion; the structural model may thus be viewed as
invariant across gender.
27 R.W. Lent et al. / Journal of Vocational Behavior 83 (2013) 2230
3.2.2. Model t by racial/ethnic group
Finally, we examined the structural model by race/ethnicity, dividing the larger sample into two groups: racial/ethnic majority
(European American; n=802) and minority (African, Hispanic, Asian Americans, and other racial designations; n=568) students. (We
aggregatedthe data inthis way inorder to achieve reasonably large sample sizes for model testing purposes. However, as we note inthe
Discussion section, caution is needed in interpreting these findings to avoid uniformity assumptions regarding race/ethnicity.) The
structural model achieved adequate fit to the data in each sub-sample; for majority students, CFI =.943, RMSEA=.056, SRMR=.047,
SB scaled
2
[118, n=802]=414.29, pb.001; for minority students, CFI =.955, RMSEA=.052, SRMR=.049, SB scaled
2
[118, n=
568]=302.40, pb.001. The SCCT model accounted for relatively large amounts of predictive variance in satisfaction and persistence in
both the majority and minority sub-samples (for satisfaction, R
2
=.60, .56, respectively; for persistence, R
2
=.47, .43).
The multisample analysis revealed that the measurement and structural models were each invariant across the two racial/ethnic
groups. In particular, the constrained measurement model did not differ from the unconstrained measurement model on either fit
difference criterion. Likewise, the constrained structural model did not differ from the unconstrained structural model on either
criterion. Thus, the constructs appear to retain an equivalent latent structure, and the relations among the constructs are similar,
across the two larger racial/ethnic groupings.
4. Discussion
The present study examined the tenability of a model integrating elements of SCCT's interest, choice, performance, andsatisfaction
models. The findings generally provide initial support for this model. In particular, consistent with prior research on SCCT's interest
and choice hypotheses (Sheu et al., 2010), interests were strongly predicted by the combination of self-efficacy and outcome
expectations, and social supports were linked to interests indirectly through their relation to self-efficacy and outcome expectations.
Extending prior researchon SCCT, we found that interests were, in turn, predictive of students' academic domain satisfaction. In other
words, possessing engineering interests (which are presumed to be compatible with those of peers and faculty) is a likely source of
satisfactionwith the academic environment of engineering majors. Moreover, students who were more satisfied withthe engineering
environment were more likely to want to remain in this major. These findings are consistent with the satisfactiontenure hypothesis
in the theory of work adjustment (Dawis, 2005), findings on the relation of job satisfaction to turnover/persistence intentions in adult
workers (Brief, 1998), and previous results linking satisfaction to intended persistence in engineering students (Lent et al., 2007).
Intended persistence has, moreover, been found to be a good predictor of actual persistence in engineering majors in past research
(Lent et al., 2003).
The integrative model of interestssatisfactionchoice may help to clarify the differential roles that interests and satisfaction
play within the academic/work adjustment process. In particular, the two variables may be linked together such that pursuit of
one's interests is likely to lead to satisfaction with one's environment, and satisfaction is, in turn, likely to lead to intentions to
persist within this environment. The pathway from interests to persistence intentions was indirect, or mediated, via satisfaction.
That is, interests may lead to choice stability to the extent that one is satisfied with the larger context in which one's interests are
situated. Stated otherwise, interests may help to draw people toward particular educational/vocational environments;
satisfaction with the environment may then be an important part of what keeps them coming back.
The interestsatisfactionchoice stability model may augment certain aspects of the theory of work adjustment (TWA; Dawis,
2005) and Holland's (1997) theory. While acknowledging the importance of interests, TWA is not entirely clear about the unique
role that interests play in relation to satisfaction, persistence, or other key variables, such as values. The current integrative model
attempts to account for how each of these variables functions together (with values being subsumed by outcome expectations).
Holland's theory maintains that interest congruence leads to satisfaction, which is similar to our hypothesis that pursuit of one's
interests is likely to lead to domain satisfaction. Our model assumes, however, that interest (or interest congruence) is only one of
several routes to satisfaction; other variables in particular, feeling competent to perform requisite tasks, expecting to obtain
favorable outcomes, receiving support from others in the environment, and possessing a general tendency to experience positive
affect can also engender satisfaction.
Table 2
Fit indices for the multiple-group analyses.
Model type by group
2
SB
2
df SB
2
df RMSEA
a
CFI
a
CFI
a
Grouping variable: gender
Measurement model without constraints 740.30 601.28 228 .049 .99
Measurement model with constraints on all loadings 772.97 636.41 239 32.48

11 .049 .99 .00

Structural models without constraints 896.03 742.83 247 .054 .98
Structural model with constraints on all structural paths 918.73 757.00 264 16.09 17 .052 .98 .00
Grouping variable: race/ethnicity
Measurement model without constraints 781.23 675.13 229 .053 .99
Measurement model with constraints on all loadings 797.74 693.76 240 16.85 11 .053 .99 .00
Structural model without constraints 913.74 821.33 248 .058 .98
Structural model with constraints on all structural paths 938.83 841.94 265 21.15 17 .056 .98 .00
a
Derived from robust maximum likelihood estimation.
pb.05.
28 R.W. Lent et al. / Journal of Vocational Behavior 83 (2013) 2230
The current findings suggest that outcome expectations may promote satisfaction both directly and indirectly, by cultivating
interests that, in turn, foster satisfaction. By contrast, a few prior studies have found that outcome expectations did not produce a
unique path to academic satisfaction either in a general sample of college students (Lent, Singley et al., 2005) or in engineering
students (Lent et al., 2007). Considering the possibility that the measure of outcome expectations used in these prior studies
captured only a limited range of desired outcomes, we decided to modify this measure by adding other types of outcomes that
may motivate academic choice and persistence. The revised measure produced a two-factor structure reflecting intrinsic and
extrinsic outcome expectations. The previous version of this measure may have underrepresented the intrinsic rewards of an
engineering career. The newer version accounted for a modest, unique amount of variance in both satisfaction and intended
persistence.
To this point, we have been discussing the results in the sample as a whole. However, a key goal of the study was to examine
the adequacy of the SCCT model as a function of gender and racial/ethnicity. Such model comparisons are important in
determining the range of the model's explanatory utility. Our findings indicated that the model fit the data well and comparably
across the grouping variables, accounting for somewhat more of the variance in women's than men's academic satisfaction and
persistence intentions. These findings thereby add to the small number of studies that have tested the invariance of SCCT models
across gender, race/ethnicity, and other grouping variables (e.g., Lent, Brown et al., 2005; Lent et al., 2008, 2011).
The study's findings need to be interpreted in light of its limitations. First, we need to highlight that the models were tested
with cross-sectional data. Therefore, the study's design cannot support causal inferences regarding the determinants of interest,
satisfaction, or intended persistence. Neither can the cross-sectional design provide a wholly satisfactory test of the model's
meditational paths. Using the test of joint significance (MacKinnon, Lockwood, Hoffman, West, & Sheets, 2002), a significant
indirect effect may be inferred when the paths from the predictor to the mediator and from the mediator to the criterion variable
are significant. While these conditions were met in our analyses (e.g., the indirect pathway from interest to intended persistence
via satisfaction was consistent with this pattern), longitudinal methods can provide a more convincing test of the hypothesized
temporal precedence and mediating pathways among the constructs.
A second limitation is that personality was represented only by positive affect and contextual affordances only by social
support. While these measurement decisions were made partly for pragmatic reasons (i.e., to contain model complexity), it would
be useful to include additional trait (e.g., negative affect, conscientiousness) and environmental attributes (e.g., presence of social
barriers) in future research on the academic adjustment process. Third, sample size considerations prevented us from examining
model-data fit in particular racial/ethnic groups. Our decision to include students of color in a single group was based on the
assumption that they may share some similar experiences as minority members of the engineering profession. However, there
may also be group-specific differences that were obscured by such a decision. For example, Asian American students are
overrepresented in some STEM disciplines, whereas other racial/ethnic minority groups are underrepresented, relative to their
proportions in the general U.S. population (NSF, 2009). Among other things, this may have implications for acquisition of peer and
faculty support or modeling within the engineering environment.
These limitations suggest several directions for future research on academic adjustment from a social cognitive perspective. First,
it would be useful to replicate and extend our findings on the interplay of interests and satisfaction with other samples of students
planning to major in STEMfields. Inparticular, it wouldbe useful to test our hypotheses regarding their differential roles, withinterest
serving as a primary attractor to STEM majors and satisfaction serving as the more proximal determinant of intended persistence.
Second, more research is needed specifically testing the explanatory utility of SCCT across gender and samples of particular racial/
ethnic groups. Such work is important in order to determine the range or limits of the theory's applicability. Third, it would be useful
to test SCCT using longitudinal designs.
In sum, the present findings extend social cognitive inquiry on person and environment factors that may promote or deter
adjustment to engineering majors. We specifically found support for an integrative model that weaves together elements of SCCT's
interest, choice, satisfaction, and performance models. The model fit the data well both in the full sample and in sub-samples of
womenand menand in students of racial/ethnic majority and minority groups. These findings suggest that the SCCT framework offers
potential for explaining the processes through which students become attracted to, and subsequently decide to remain in (or leave),
STEMfields. Further research is needed to test the model's explanatory value as well as the utility of efforts to apply it to educational/
career assessment and intervention.
References
Bandura, A. (1997). Self-efficacy: The exercise of control. New York: W.H. Freeman.
Bentler, P. M., & Wu, E. J. C. (2005). EQS 6.1 for Windows user's guide. Encino, CA: Multivariate Software.
Brief, A. P. (1998). Attitudes in and around organizations. Thousand Oaks, CA: Sage.
Browne, M. W., & Cudeck, R. (1992). Alternative ways of assessing model fit. Sociological Methods and Research, 21, 230258. http://dx.doi.org/10.1177/
0049124192021002005.
Byars-Winston, A., Estrada, Y., Howard, C., Davis, D., & Zalapa, J. (2010). Influence of social cognitive and ethnic variables on academic goals of underrepresented
students in science and engineering: A multiple-groups analysis. Journal of Counseling Psychology, 57, 205218. http://dx.doi.org/10.1037/a0018608.
Cheung, G. W., & Rensvold, R. B. (2002). Evaluating goodness-of-fit indexes for testing measurement invariance. Structural Equation Modeling: A Multidisciplinary
Journal, 9, 233255. http://dx.doi.org/10.1207/S15328007SEM0902_5.
Dawis, R. V. (2005). The Minnesota theory of work adjustment. In S. D. Brown, & R. W. Lent (Eds.), Career development and counseling: Putting theory and research
to work (pp. 323). Hoboken, NJ: Wiley.
Ferry, T. R., Fouad, N. A., & Smith, P. L. (2000). The role of family context in a social cognitive model for career-related choice behavior: A math and science
perspective. Journal of Vocational Behavior, 57, 348364. http://dx.doi.org/10.1006/jvbe.1999.1743.
Fouad, N. A., & Smith, P. L. (1996). A test of a social cognitive model for middle school students: Math and science. Journal of Counseling Psychology, 43, 338346.
29 R.W. Lent et al. / Journal of Vocational Behavior 83 (2013) 2230
Hansen, J. C. (1984). The measurement of vocational interests: Issues and future directions. In S. D. Brown, & R. W. Lent (Eds.), Handbook of counseling psychology
(pp. 99136). New York: Wiley.
Holland, J. L. (1997). Making vocational choices: A theory of vocational personalities and work environments (3rd ed.). Odessa, FL: Psychological Assessment
Resources.
Hoyle, R. H., & Panter, A. T. (1995). Writing about structural equation models. In R. H. Hoyle (Ed.), Structural equation modeling: Concepts, issues, and applications
(pp. 158176). Thousand Oaks, CA: Sage.
Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation
Modeling, 6, 155. http://dx.doi.org/10.1080/10705519909540118.
Lent, R. W., & Brown, S. (2006). Integrating person and situation perspectives on work satisfaction: A social-cognitive view. Journal of Vocational Behavior, 69,
236247. http://dx.doi.org/10.1016/j.jvb.2006.02.006.
Lent, R. W., & Brown, S. (2008). Social cognitive career theory and subjective well- being in the context of work. Journal of Career Assessment, 16, 621. http://dx.doi.org/
10.1177/1069072707305769.
Lent, R. W., Brown, S. D., & Hackett, G. (1994). Toward a unifying social cognitive theory of career and academic interest, choice, and performance [Monograph].
Journal of Vocational Behavior, 45, 79122. http://dx.doi.org/10.1006/jvbe.1994.1027.
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,
3649. http://dx.doi.org/10.1037//0022-0167.47.1.36.
Lent, R. W., Brown, S. D., & Larkin, K. C. (1986). Self-efficacy in the prediction of academic performance and perceived career options. Journal of Counseling
Psychology, 33, 265269.
Lent, R. W., Brown, S. D., Schmidt, J., Brenner, B., Lyons, H., & Treistman, D. (2003). Relation of contextual supports and barriers to choice behavior in engineering
majors: Test of alternative social cognitive models. Journal of Counseling Psychology, 50, 458465. http://dx.doi.org/10.1037/0022-0167.50.4.458.
Lent, R. W., Brown, S. D., Sheu, H., Schmidt, J., Brenner, B. R., Gloster, C. S., et al. (2005). Social cognitive predictors of academic interests and goals in engineering:
Utility for women and students at historically Black universities. Journal of Counseling Psychology, 52, 8492. http://dx.doi.org/10.1037/0022-0167.52.1.84.
Lent, R. W., Lopez, A. M., Lopez, F. G., & Sheu, H. (2008). Social cognitive career theory and the prediction of interests and choice goals in the computing disciplines.
Journal of Vocational Behavior, 73, 5262. http://dx.doi.org/10.1016/j.jvb.2008.01.002.
Lent, R. W., Lopez, F. G., Sheu, H., & Lopez, A. M. (2011). Social cognitive predictors of the interests and choices of computing majors: Applicability to
underrepresented students. Journal of Vocational Behavior, 78, 184192. http://dx.doi.org/10.1016/j.jvb.2010.10.006.
Lent, R. W., Singley, D., Sheu, H., Gainor, K., Brenner, B. R., & Treistman, D. (2005). Social cognitive predictors of domain and life satisfaction: Exploring the
theoretical precursors of subjective well-being. Journal of Counseling Psychology, 52, 429442.
Lent, R. W., Singley, D., Sheu, H., Schmidt, J. A., & Schmidt, L. C. (2007). Relation of social-cognitive factors to academic satisfaction in engineering students. Journal
of Career Assessment, 15, 8797. http://dx.doi.org/10.1177/1069072706294518.
Lopez, F. G., Lent, R. W., Brown, S. D., & Gore, P. A. (1997). Role of social-cognitive expectations in high school students' mathematics-related interest and
performance. Journal of Counseling Psychology, 44, 4452. http://dx.doi.org/10.1037//0022-0167.44.1.44.
MacKinnon, D. P., Lockwood, C. M., Hoffman, J. M., West, S. G., & Sheets, V. (2002). A comparison of methods to test mediation and other intervening variable
effects. Psychological Methods, 7(1), 83104. http://dx.doi.org/10.1037/1082-989X.7.1.83.
National Science Foundation, Division of Science Resources Statistics (2009). Women, minorities, and persons with disabilities in science and engineering: 2009
(NSF 09-305). Arlington, VA: Author (Retrieved July 29, 2010, from http://www.nsf.gov/statistics/wmpd/)
Nauta, M. M. (2013). Holland's theory of vocational choice and adjustment. In S. D. Brown, & R. W. Lent (Eds.), Career development and counseling: Putting theory
and research to work (2nd ed.). Hoboken, NJ: Wiley.
Satorra, A., & Bentler, P. M. (2001). A scaled difference chi-square test statistic for moment structure analysis. Psychometrika, 66, 507514. http://dx.doi.org/
10.1007/BF02296192.
Sheu, H., Lent, R. W., Brown, S. D., Miller, M. J., Hennessy, K. D., & Duffy, R. D. (2010). Testing the choice model of social cognitive career theory across Holland
themes: A meta-analytic path analysis. Journal of Vocational Behavior, 76, 252264. http://dx.doi.org/10.1016/j.jvb.2009.10.015.
U.S. Department of Labor, Employment and Training Administration (2002). Work importance profiler: User's guide. Washington, D.C.: Author.
Watson, D., Clark, L. A., & Tellegen, A. (1988). Development and validation of brief measures of positive and negative affect: The PANAS scales. Journal of
Personality and Social Psychology, 54, 10631070. http://dx.doi.org/10.1037//0022-3514.54.6.1063.
30 R.W. Lent et al. / Journal of Vocational Behavior 83 (2013) 2230