Professional Documents
Culture Documents
İncele Moderated Mediation
İncele Moderated Mediation
İncele Moderated Mediation
Revised 12/13/12
Accepted 01/01/13
DOI: 10.1002/j.2161-0045.2014.00070.x
Boram Kim, Sun Hee Jang, Sun Hwa Jung, Bo Hyun Lee, and Sang Min Lee,
Department of Education, Korea University, Seoul, South Korea; Ana Puig, Of-
fice of Educational Research, University of Florida. This work was supported by a
grant (NRF-2011-330-B00167) from the National Research Foundation of Ko-
rea. Correspondence concerning this article should be addressed to Sang Min Lee,
Department of Education, Korea University, Anam-Dong, Seongbuk-Gu, Seoul,
South Korea (e-mail: leesang@korea.ac.kr).
© 2014 by the National Career Development Association. All rights reserved.
Method
Participants
Questionnaires were distributed to 229 Korean college students in gen-
eral education courses at a large university in Seoul, the capital city of
South Korea. Participants were recruited from general psychology classes
Career Decision
Planned Happenstance Self-Efficacy Planned Happenstance
Skills Skills
FIGURE 1
Theoretical Model
Measures
Occupational engagement. The Korean version of the Occupational En-
gagement Scale for Students (OES-S; Cox, 2008) was used to measure
the participants’ current status of career engagement. This scale has yet
to be validated, but it was empirically reviewed by Jung (2011) to show
that the Korean version reflected the intentions of the original scale.
The OES-S is a 14-item single-factor instrument. Sample items include
“I talk about my career choices with family or friends” and “I attend
presentations or talks related to a career I might find interesting.” There
is no item coded inversely, and each item is rated on a 5-point Likert
scale, with responses ranging from 1 (I do not agree at all) to 5 (I highly
agree). Jung reported Cronbach’s alpha, the internal consistency of this
scale, as .85; it was .89 in our study.
Career decision self-efficacy. To measure career decision self-efficacy, we
used the short form of the Career Decision Self-Efficacy Scale (CDSES-SF;
Betz, Klein, & Taylor, 1996). In our study, we used the Korean version
of the CDSES-SF, which has been validated by Lee and Lee (2000). The
CDSES-SF has 25 items and uses a 5-point Likert scale ranging from
1 (strongly disagree) to 5 (strongly agree). The CDSES-SF was used to
measure how much self-efficacy the respondents had in making career
decisions. The instrument has five subscales (Gathering Occupational
Information, Goal Selection, Plans for Implementation, Problem-Solving,
and Self-Appraisal), and there are five items for each. Sample items are
“I can choose one job among the jobs I am considering” and “I can
tell what the ideal job is for me.” A higher score means higher career
decision self-efficacy. Cronbach’s alpha was reported as .92 by Lee and
Lee; it was .93 in our study.
Career decision certainty. The Career Decision Scale (CDS; Osipow,
Carney, Winer, Yanico, & Koschir, 1976) was used to measure career
decision certainty. Ko (1992) validated the Korean version of the CDS
with a sample of Korean college students. The scale consists of 18 items
and uses a 4-point Likert scale ranging from 1 (strongly disagree) to 4
(strongly agree). The scale is divided into two levels: career certainty
(e.g., “I decided my career and feel comfortable about it”) and career
indecision (e.g., “I cannot make my career decision since I do not know
about my aptitudes and abilities well”). Career indecision level scores
Data Analyses
Two interlinked steps were tested in our study: a simple mediation model
and a moderated mediation model. The moderated mediation model
was tested by integrating the proposed moderator variable (planned
happenstance skills in our study) into the model. As Aiken and West
(1991) suggested, all continuous variables were mean centered prior
to the analyses to reduce the multicollinearity between main effects
and interaction. For the first step of analysis, simple mediation models
were examined by using an application provided by Preacher and Hayes
(2004). Preacher and Hayes developed an SPSS macro that facilitates
estimation of the indirect effect ab, both with a normal theory approach
(i.e., the Sobel test) and with a bootstrap approach to obtain confidence
intervals (CIs). The SPSS macro also incorporates the stepwise proce-
dure described by Baron and Kenny (1986). In the next step, we again
utilized an SPSS macro designed by Preacher and colleagues (2007). By
using this macro, we predicted that the implementation of the recom-
mended bootstrapping methods would be facilitated and that a method
for probing the significance of the conditional indirect effects would be
provided at different values of the moderator variable.
Results
Means, standard deviations, and intercorrelations among variables for
all participants are presented in Table 1. An examination of the correla-
tions showed that career engagement was positively related to career
decision self-efficacy (r = .58, p < .001), planned happenstance skills
(r = .46, p < .001), and career decision certainty (r = .18, p < .01).
Results also indicated that career decision self-efficacy positively cor-
related with planned happenstance skills (r = .45, p < .001) and career
decision certainty (r = .49, p < .001). It is interesting that no statistically
significant correlation was found between planned happenstance skills
and career decision certainty (r = .01, ns).
Next, as shown in Table 2, the results of the simple mediation model
indicated that career engagement was significantly related to career
decision self-efficacy (B = .50, t = 10.32, p < .001). The relationship
between career decision self-efficacy and career decision certainty was
also significant (B = .45, t = 8.12, p < .001) while controlling for career
engagement. Furthermore, career engagement was negatively associ-
ated with career decision certainty when controlling for career decision
self-efficacy (B = –.10, t = –2.16, p < .05). Although the direct effect
between career engagement and career decision certainty was originally
positive when controlling for career decision self-efficacy, the direction
Table 2
Regression Results for Simple Mediation
Variable B SE t z p 95% CI
Direct and total effects
Career decision certainty
regressed on career
engagement .12 .04 2.69 .008
Career decision self-
efficacy regressed on
career engagement .50 .05 10.32 .000
Career decision certainty
regressed on career
decision self-efficacy,
controlling for career
engagement .45 .06 8.12 .000
Career decision certainty
regressed on career
engagement, controlling
for career decision self-
efficacy –.10 .05 –2.16 .032
Indirect effect (IE)a .04 6.36 .000 [.15, .29]
Bootstrap results for IEb .04 0.31 [.15, .31]
ACACDQ_v62_n1_0314TEXT.indd 63
Variable B SE t p IE SE z p Percentile BC BCa
Career Decision Self-Efficacy (CDSE)
Constant –0.03 .03 –0.98 .331
Career engagement (CE) 0.38 .05 7.34 .000
Planned happenstance skills (PH) 0.25 .07 3.63 .000
CE × PH 0.20 .07 2.75 .006
Career Decision Certainty
Constant 2.47 .03 89.93 .000
CE –0.06 .05 –1.30 .194
PH –0.23 .06 –3.99 .000
CE × PH –0.05 .08 –0.66 .509
CDSE 0.51 .06 8.95 .000
CDSE × PH 0.11 .09 1.24 .215
Conditional Indirect Effect at Specified Values of PH
Value of PH
–1 SD (–.52) .13 .04 3.47 .001 [.08, .33] [.09, .34] [.09, .34]
M (.00) .20 .04 5.65 .000 [.19, .42] [.20, .42] [.20, .43]
+1 SD (.52) .28 .05 5.31 .000 [.25, .60] [.26, .62] [.26, .61]
Conditional Indirect Effect at Range of Values of PH
Value of PH
–1.45 .03 .05 0.67 .503
–1.29 .05 .05 0.99 .322
–1.12 .06 .04 1.37 .172
–0.96 .08 .04 1.81 .071
63
(Continued on next page)
2/10/2014 12:02:52 PM
64
Table 3 (Continued)
ACACDQ_v62_n1_0314TEXT.indd 64
Regression Results for Conditional Indirect Effect
Bootstrap 95% CI
Variable B SE t p IE SE z p Percentile BC BCa
Conditional Indirect Effect at Range of Values of PH
Value of PH (Continued)
–0.80 .09 .04 2.34 .020
–0.63 .11 .04 2.96 .003
–0.47 .13 .04 3.69 .000
–0.31 .15 .03 4.47 .000
–0.14 .18 .03 5.20 .000
0.02 .20 .04 5.69 .000
0.18 .23 .04 5.83 .000
0.35 .25 .04 5.67 .000
0.51 .28 .05 5.33 .000
0.67 .31 .06 4.95 .000
0.84 .34 .07 4.58 .000
1.00 .37 .09 4.24 .000
1.16 .40 .10 3.96 .000
1.33 .43 .12 3.71 .000
1.49 .47 .13 3.50 .001
Note. Unstandardized regression coefficients are reported. Bootstrap sample size = 10,000. CI = confidence interval; IE = indirect effect; BC = bias corrected;
BCa = bias corrected and accelerated.
2/10/2014 12:02:52 PM
= Low planned happenstance skills
= High planned happenstance skills
0.5 —
0.4 —
0.3 —
Career Decision Self-Efficacy
0.2 —
0.1 —
Low High
0.0 —
–0.1 —
–0.2 —
–0.3 —
–0.4 —
—
—
M – 1 SD M + 1 SD
Career Engagement
Figure 2
Interaction Effect of Planned Happenstance Skills Between
Career Engagement and Career Decision Self-Efficacy
Note. Unstandardized regression coefficients are used.
Discussion
The purpose of our research was to identify how planned happenstance
skills play a role in the relationships among career engagement, career
decision self-efficacy, and career decision certainty. Using the moderated
mediation model, we analyzed the relationships among career engage-
ment, career decision self-efficacy, career decision certainty, and planned
happenstance skills. First, the results of our study indicated that the
planned happenstance skills variable was significantly correlated with
career engagement and career decision self-efficacy. On the other hand,
no significant relationship was found between planned happenstance
skills and career decision certainty. That is, individuals who have planned
happenstance skills are actively involved in career exploration activities
and feel confident about the career decision-making process but do
not necessarily make future career decisions. This finding needs to be
interpreted in the context of Krumboltz’s (2009) happenstance learning
theory, which suggests that the goal of career counseling is not for clients
to make a career decision by declaring their future lifetime occupation
but for clients to reframe being undecided into open-mindedness. The
individuals who have planned happenstance skills keep career options
always open, which allows them to create, recognize, and seize new
References
Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and interpreting interac-
tions. Thousand Oaks, CA: Sage.