compromise in career-related decisions: examining the role of compromise severity

68
COMPROMISES IN CAREER-RELATED DECISIONS 1 APA copyright notice: This article may not exactly replicate the final version published in the APA journal. It is not the copy of record. Wee, S. (in press). Compromises in Career-Related Decisions: Examining the Role of Compromise Severity. Journal of Counseling Psychology. Advance online publication. doi: 10.1037/cou0000037 Compromises in Career-Related Decisions: Examining the Role of Compromise Severity Serena Wee Singapore Management University Author Note Serena Wee, School of Social Sciences, Singapore Management University.

Upload: smu-sg

Post on 30-Jan-2023

0 views

Category:

Documents


0 download

TRANSCRIPT

COMPROMISES IN CAREER-RELATED DECISIONS 1

APA copyright notice: This article may not exactly replicatethe final version published in the APA journal. It is not thecopy of record.

Wee, S. (in press). Compromises in Career-Related Decisions: Examining the Role of Compromise Severity. Journal of Counseling Psychology. Advance online publication. doi: 10.1037/cou0000037

Compromises in Career-Related Decisions: Examining the Role of

Compromise Severity

Serena Wee

Singapore Management University

Author Note

Serena Wee, School of Social Sciences, Singapore

Management University.

COMPROMISES IN CAREER-RELATED DECISIONS 2

The article is based on a portion of Serena Wee’s

doctoral dissertation at the University of Illinois at Urbana-

Champaign.

Correspondence concerning this article should be

addressed to Serena Wee, Singapore Management University,

School of Social Sciences, 90 Stamford Road, Level 4,

Singapore 178903. Email: [email protected]

COMPROMISES IN CAREER-RELATED DECISIONS 3

Abstract

This study tested Gottfredson’s (1996) revised compromise

theory by examining if the relative importance of job sextype,

job prestige, and person-job interest congruence for

predicting job choice changed as the level of compromise

required changes. Using a fully within-persons design,

participants engaged in a simulated occupational choice task

where job sextype and job prestige were manipulated to be

experimentally independent. Participants first categorized

jobs as unacceptable, acceptable, or preferred. Then within

each category, they made further pairwise choices among jobs

in that category. In Study 1, participants were 168 college

seniors (124 women, 44 men) from a large Midwestern

university. In Study 2, participants were 262 (146 women, 116

men) individuals residing in the United States and recruited

via Amazon’s Mechanical Turk platform. Across both studies,

job sextype predicted choice when large compromises were

required. Across both studies, job prestige did not predict

choice when moderate compromises were required. In Study 2 but

not Study 1, person-job interest congruence predicted choice

when minimal compromises were required.

COMPROMISES IN CAREER-RELATED DECISIONS 4

Keywords: career choice, compromise, Gottfredson’s theory

of circumscription and compromise, prestige, sextype

COMPROMISES IN CAREER-RELATED DECISIONS 5

Compromises in Career-Related Decisions: Examining the Role of

Compromise Severity

Work provides a central source of meaning and identity

for most adults (Deaux, 2001; Hulin, 2002), with unemployment

(Murphy & Athanasou, 1999), underemployment (Friedland &

Price, 2003), and job (dis)satisfaction (Hulin & Judge, 2003)

predicting a range of personal and organizational

consequences. Choosing an occupation (or job) is thus an

important goal for many people, one with far reaching

consequences. However, theories of career decision-making and

choice suggest that people often have to compromise their

preferences, and that satisficing rather than optimizing is

the norm (Gati, 1993; Ginzberg, Ginsburg, Axelrad, & Herma,

1951; Gottfredson, 1996). For example, in a nationally

representative longitudinal sample in which adolescents

indicated their job aspirations at age 16, only 24%, 6%, and

3% of those aspiring to be health professionals, scientists,

and engineers, respectively, realized their aspirations by age

33 (Schoon, 2001, Table 5). Similarly, in another longitudinal

study with 8 – 10 grade students, Armstrong and Crombie (2000)

found that individuals whose career aspirations were less

COMPROMISES IN CAREER-RELATED DECISIONS 6

normative (e.g., cross-sextyped) tended to shift their

aspirations over time to match more realistic expectations of

what was accessible and achievable. Although aspirational

changes do not equate to having compromised, they indicate the

increasing importance placed on accessibility and

achievability as the need to choose becomes more pressing.

When people relinquish preferred options for less preferred

options in order to align with an external reality, compromise

occurs. Understanding what gets compromised provides the first

step towards helping people to make better choices, choices

where unnecessary concessions can be minimized.

Theoretical Perspectives on Career Compromise

A developmental perspective is implicit, if not explicit,

in almost all theoretical accounts of compromise. In their

most fundamental forms, they postulate a match between person-

focused and occupation-focused variables. However, whereas

psychological theories (Gati, 1993; Holland, 1997) have tended

to emphasize unique characteristics of a person, such as their

interests, sociological (Blau & Duncan, 1967; Sewell, Haller,

& Ohlendorf, 1970) and economic (Becker, 1964; Cunha &

Heckman, 2007) theories have tended to emphasize either group

COMPROMISES IN CAREER-RELATED DECISIONS 7

membership/identities that characterize a person, such as

their gender and social status, or economic conditions (e.g.,

effects of education) that apply regardless of the individual.

Because Gottfredson (1996) provides an account attempting to

integrate these multiple perspectives, and hence to provide a

more comprehensive explanation for the phenomenon of

compromise, this study focused on testing the predictions she

made regarding the conditional priorities enacted when people

are faced with the necessity to compromise.

Gottfredson’s (1981, 1996) theory of circumscription and

compromise. Gottfredson (1981) argued that occupational

aspirations develop in tandem with a child's maturity in

understanding both self and the larger world. Children

progressively eliminate portions of the world of work that do

not match their conceptions of who they are and where they

exist (or want to exist) within the social world. This

circumscription process results in a range of occupations an

individual finds acceptable. Gottfredson (1981) refers to this

as the zone of acceptable alternatives.

Three substantive aspects are examined in Gottfredson’s

(1981) theory: job sextype, job prestige, and person-job

COMPROMISES IN CAREER-RELATED DECISIONS 8

interest congruence. As an example, imagine a typical female

of average intelligence and social standing. As she develops

an understanding of what occupations are, masculine jobs such

as “truck driver” are ruled out, as they are inconsistent with

her developing gender identity. As she becomes aware of status

differences in society, she starts to consider job prestige.

She also starts to understand her own capabilities (e.g.,

feedback from grades or encouragement from teachers), and

begins to determine what is achievable and what is not.

Occupations circumscribed at this stage are not considered at

later stages, resulting in progressively fewer occupations

deemed acceptable (i.e., within the zone of acceptable

alternatives). According to Gottfredson (1981), it is only now

that individuals start to pay attention to their “internal,

unique selves” (e.g., personality, interests, and goals).

Gottfredson (1981) suggests the compromise process is

similarly influenced by the salience and importance of job

sextype, job prestige, and person-job interest congruence.

Specifically, she assumed that people are most concerned with

maintaining an acceptable social identity, and only

secondarily with fulfilling a more private, psychological

COMPROMISES IN CAREER-RELATED DECISIONS 9

self. Thus she proposed that gender identities are more

strongly protected than status identities, which in turn are

more strongly protected than personal identities. Empirical

tests of these specific predictions (1) have provided support

for the opposite pattern of results—interests were more

important than prestige, and prestige was more important than

sextype (Hesketh, Elmslie, & Kaldor (1990), (2) indicated

moderate concerns for both sextype and prestige, although in

some cases participants seemed more interested in

gaining/protecting prestige than sextype (Leung, 1993; Leung &

Plake, 1993), and (3) indicated greater concerns for prestige

than interests (Holt, 1989; for a more detailed review see

Gottfredson, 1996).

To address the inconsistent findings, Gottfredson (1996,

p. 198 – 200) revised her theory to include compromise

severity as an important moderator of the relative importance

of job sextype, job prestige, and person-job interest

congruence. Consistent with her earlier theory, she proposed

that when individuals have to compromise, they would do so on

the job aspect least threatening to their self-concept. But

additionally, she implied that the greater the compromise

COMPROMISES IN CAREER-RELATED DECISIONS 10

required, the more salient central aspects of the self become

in the decision. Specifically, when large compromises are

required, it is assumed such jobs are unlikely to have met

threshold levels of gender appropriateness, and thus

considerations of sextype are likely to be more important than

either prestige or interests in determining choice. When

moderate compromises are required, it is assumed that jobs

have met at least threshold levels of gender appropriateness,

and thus considerations of prestige are likely to be more

important than either sextype or interests in determining

choice. And when minimal compromises are required, it is

assumed that jobs have met at least threshold levels of gender

appropriateness and status, and thus considerations of

personal interests are likely to be more important than either

sextype or prestige in determining choice. These predictions

are based on the verbal statements made by Gottfredson (1996,

p. 200). In the current study, therefore, the following

hypotheses will be tested:

H1: When individuals choose amongst jobs they consider unacceptable (large

compromise), chosen options are expected to be significantly more sextyped (i.e.,

COMPROMISES IN CAREER-RELATED DECISIONS 11

more feminine for women, more masculine for men) than unchosen options. No

differences are expected on job prestige or interest.

H2: When individuals choose amongst jobs they consider acceptable

(moderate compromise), chosen options are expected to be significantly more

prestigious than unchosen options. No differences are expected on job sextype or

interest.

H3: When individuals choose amongst jobs they consider preferable (minimal

compromise), chosen options are expected to be significantly more consistent with a

person’s interests than unchosen options. No differences are expected on job sextype

or prestige.

A PsycINFO search conducted May 2nd 2014 indicated that to

date, only Blanchard and Lichtenberg (2003) have tested the

revised conditional priorities outlined by Gottfredson (1996).

Their study occurred in two phases. In the first phase,

participants rated 89 jobs on how sextyped, prestigious, and

consistent with personal interests they perceived the job to

be. Two weeks later, they returned to categorize the jobs as

“acceptable”, “uncertain”, or “unacceptable”. (Acceptable jobs

corresponded to the low compromise condition, uncertain jobs

to the moderate compromise condition, and unacceptable jobs to

the high compromise condition.) Then, participants were

COMPROMISES IN CAREER-RELATED DECISIONS 12

randomly assigned to one of the conditions, where they ranked

eight randomly selected jobs from within that condition.

Dependent variables were the participant’s subjective ratings

of the top-ranked job’s sextype, prestige, and fit with their

personal interests. In support of Gottfredson’s (1996)

predictions, Blanchard and Lichtenberg (2003) found that jobs

in the low compromise condition were rated more highly on

interests than on prestige, and more highly in terms of

prestige than sextype. However, they found mixed support for

Gottfredson’s other predictions: (a) jobs in the moderate

compromise condition were higher on prestige than interests,

but prestige ratings did not differ significantly from sextype

ratings; and (b) jobs in the large compromise condition were

higher on sextype than interests, but sextype ratings did not

differ significantly from prestige ratings. They concluded

that, consistent with Goffredson’s prediction, interests tend

to be important when people face relatively minimal

compromises. But, when larger compromises are required, both

prestige and sextype appear to be important, and their results

are inconclusive regarding the relative importance of one

against the other.

COMPROMISES IN CAREER-RELATED DECISIONS 13

However, because the top-ranked jobs were not compared to

the seven other jobs, the relative importance of the different

aspects within each condition was not actually tested. In

addition, the study suffered from several notable limitations.

First, compromise was manipulated between- rather than within-

persons, even though compromise severity as outlined in

Gottfredson (1996) appears to vary meaningfully within persons

(cf. Junk & Armstrong, 2010). Second, because only the top-

ranked occupation was used in the data analysis (Blanchard &

Lichtenberg, 2003, p. 258), almost all the potentially useful

data were discarded. Third, participants’ subjective

perceptions were used as ratings of sextype, prestige, and

interest in jobs, even though these ratings were not very

reliable, as indicated by the test-retest reliabilities over a

two-week period (rsextype = .50, rprestige = .58, rinterest = .68).

Study 1

This study examined Gottfredson’s (1996) conditional

priorities hypothesis regarding compromise. Specifically, this

study tested if the relative importance of job sextype, job

prestige, and person-job interest congruence for predicting

job choice changes as the level of compromise required

COMPROMISES IN CAREER-RELATED DECISIONS 14

changes. Using a similar paradigm to the one outlined in

Blanchard and Lichtenberg (2003), this study improved on the

sole empirical test of Gottfredson’s (1996) conditional

priorities hypothesis in three ways. First, compromise was

manipulated within- rather than between-persons. Apart from

increasing the fidelity of the psychological construct, a

within-persons design controls for extraneous variables

associated with individual differences, providing a greater

chance to detect the effect of increasing levels of compromise

on the relative importance of sextype, prestige, and interest

in determining job choices. Second, none of the information

provided by participants was discarded, which results in more

consistent effects being estimated. Third, participants

simulated the compromise process by making forced choices

among jobs within each compromise condition. The current study

is also intended to complement the study conducted by

Blanchard and Lichtenberg (2003) by using objectively rather

than subjectively derived indicators of sextype, prestige, and

work-activity ratings to characterize the job stimuli; the use

of objectively derived measures could provide greater external

validity.

COMPROMISES IN CAREER-RELATED DECISIONS 15

Method

Participants. Participants were 197 college seniors (127

women, 44 men, 26 non-responses) from a large Midwestern

university. Participants who did not indicate their sex (n =

26), or complete the interest inventory (n = 3 women) were

excluded from further analyses, resulting in complete data

obtained for 168 participants. Participants had a mean age of

21.64 (SD = 0.89; range: 20 – 28 years). College seniors were

recruited via email through the Psychology Advising Office and

The Career Center. The email invited participants to enter a

website where they could fill out a research questionnaire in

return for a chance to win a cash lottery; five prizes of

US$100 each were available. Participants were told the

questionnaire would take 20 – 30 minutes to complete. They

were also told they would rate and choose among a series of

jobs, and provide demographic information about themselves

(i.e., age, sex, current major, and intended job).

Participants came from 12 of 14 colleges in the university,

with a majority (58%) coming from the College of Liberal Arts

and Sciences.

COMPROMISES IN CAREER-RELATED DECISIONS 16

Procedure. Participants were provided with a link to an

online questionnaire where they were presented with the

occupational choice task. In the first section, participants

were presented with a list of 50 jobs (described below and

listed in the Appendix) in randomized order. They were

instructed to indicate their preference for each job by

stating if they would personally find the job unacceptable,

acceptable or preferable. No constraints were placed on the

number of jobs that could be placed in each preference

category. The outcome of this categorization was assumed to

correspond to a participant’s circumscribed social space;

one’s zone of acceptable alternatives includes preferred and

acceptable jobs, and excludes unacceptable jobs. After they

indicated their preferences, they moved on to the second

section, where they were presented with a second list

comprising pairs of jobs. The job pairs were formed within

preference category, and consisted of all possible pairwise

combinations of jobs, up to a maximum of 10 job pairs. If a

condition comprised of less than two jobs, no choice was

presented, and that condition was skipped. If more than 10

pairs of jobs were possible, 10 pairs were randomly selected

COMPROMISES IN CAREER-RELATED DECISIONS 17

and presented to the participant. Participants were instructed

to choose one of the two jobs from each pair. Specifically,

they were asked to indicate “which one of these two jobs you

would be more likely to pick if you were presented with only

these two options”. For example, if someone indicated she

found the computer programmer job unacceptable and the

botanist job acceptable, she would not be asked to choose

between being a computer programmer and a botanist. This was

an attempt to simulate compromise within different regions of

a participant’s social space. After they completed both

sections of the occupational choice task, participants also

completed the O*NET interest profiler (described below), and a

short demographic survey.

Occupational stimuli. The 50 jobs were selected from a

set of 268 jobs that formed the Occupational Preference

Inventory (OPI; Deng, Armstrong, & Rounds, 2007). Jobs listed

in the OPI represented approximately 85% of all jobs in the

United States (US), and hence provided a good approximation of

the population of jobs from which to draw a sample for the

stimuli in the current study. In addition to the job titles

listed in the OPI, the first author Deng Chi-ping also

COMPROMISES IN CAREER-RELATED DECISIONS 18

provided ratings for job sextype, job prestige, and RIASEC

work activity ratings for each job; this information is

provided in the Appendix. A brief summary of how each variable

was operationalized is provided below. Further details about

variable operationalization are provided in Deng et al.

(2007).

Job sextype. Deng et al. (2007) asked college students to

rate how much they liked ([1] = “strongly dislike” to [7] =

“strongly like”) each of the OPI jobs. For each job, sextype

was measured using the difference between the mean ratings

provided by women (n = 572) and the mean ratings provided by

men (n = 266). In their sample, this measure of sextype

correlated .78 with the proportion of females employed across

different jobs (based on the 2003 Current Population Survey; Bureau

of Labor Statistics, 2004a), demonstrating good convergent

validity. This measure was used to index sextype in the

current study. Negative values indicate more masculine

occupations (e.g., electrical engineer [sextype = -1.19]),

positive values indicate more feminine occupations (e.g.,

hostess [sextype = 0.62]), and numbers close to zero indicate

sex-neutral occupations (e.g., hotel manager [sextype = 0.06]).

COMPROMISES IN CAREER-RELATED DECISIONS 19

Job prestige. Information from the O*NET database (O*NET

Resource Center, 2003) was used to develop a measure of

prestige for jobs listed in the OPI (Deng et al., 2007).

Specifically, three job variables were used: (1) vocational

preparation, as indexed by the O*NET job zone ratings

indicating how much preparation ([1] = “little or no

preparation” to [5] = “extensive preparation”) was required to

do the job (Oswald, Campbell, McCloy, Rivkin, & Lewis, 1999);

(2) a composite of recognition and social status, which are

need-reinforcer variables (Dawis & Lofquist, 1984)

characterizing the type of work and work environment involved

in each job, and for which ratings were also obtained from the

O*NET (McCloy et al., 1999); and (3) mean annual salary as

reported in the Occupational Employment Statistics Survey (Bureau of

Labor Statistics, 2004b). Deng et al. (2007) conducted a

principal components analysis (PCA) on these three variables

and obtained a first component accounting for 79% of the total

variance. For each job, they formed a job prestige measure

using the regression-based component score resulting from the

PCA, and this is the measure used to index job prestige in the

current study. Larger values indicate more prestigious

COMPROMISES IN CAREER-RELATED DECISIONS 20

occupations (e.g., political science professor [prestige =

1.42]) and smaller values indicate less prestigious

occupations (e.g., cashier [prestige = -1.89]).

RIASEC work activity ratings. Information from the O*NET

database (O*NET Resource Center, 2003) was also used to assign

RIASEC ratings to jobs listed in the OPI (Deng et al., 2007).

Specifically, RIASEC work activity ratings assigned by subject

matter experts indicated how characteristic ([1] = “not at all

characteristic” to [7] = “extremely characteristic”) the job

was for each RIASEC work environment (Rounds et al., 1999).

The O*NET work activity ratings show high agreement (79%) with

first-letter RIASEC codes from the Dictionary of Holland

Occupational Codes (Eggerth, Bowles, Tunick, & Andrew, 2005;

Gottfredson & Holland, 1996). This was the measure used to

index RIASEC work activity ratings in the current study. The

RIASEC types distribution for the 268 OPI jobs was: R (30%), I

(15%), A (9%), S (14%), E (17%), and C (15%); the distribution

for the 50 jobs in this study was similar: R (36%), I (20%), A

(8%), S (8%), E (16%), and C (12%).

Stimuli selection. The 268 jobs were ranked by job prestige

then divided into five groups (i.e., quintiles) such that the

COMPROMISES IN CAREER-RELATED DECISIONS 21

first group included the bottom 20% of jobs in terms of

prestige and the fifth group included the top 20% of jobs in

terms of prestige. Within each prestige quintile, jobs were

further ranked by job sextype then divided into five sextype

groups/quintiles. Then, jobs were sampled from each sextype

quintile to ensure a wide range on sextype within each

prestige quintile. Ten jobs were selected within each prestige

quintile. This selection process was used to ensure that

prestige and sextype were uncorrelated – their mean

correlation within prestige quintiles was < .05 and their

correlation across the 50 jobs was .12. Forced choice

comparisons were always made within prestige groups.

O*NET Interest Profiler. Participants’ interest in

various work activities was measured using the 60-item O*NET

Interest Profiler (Rounds et al., 1999, Rounds, Su, Lewis, &

Rivkin, 2010). There were 10 items per RIASEC type. Each item

described a work-related activity (e.g., “Study weather

conditions”). Participants indicated how much they would like

([1] = “strongly dislike” to [5] = “strongly like”) to perform

each activity. As reported in the scale validation study

(Rounds et al., 2010; N = 1061), scale reliabilities ranged

COMPROMISES IN CAREER-RELATED DECISIONS 22

from .78 to .87 (M = .81), convergent validity with same-named

scales on the Interest-Finder (Wall & Baker, 1997) ranged from

.74 to .82, and discriminant validity with dissimilar scales

on the Interest-Finder ranged from .12 to .48. Internal

consistency reliability within this sample was: R = .89, I

= .92, A = .89, S = .88, E = .87, and C = .90.

Person-job interest congruence. Person-job interest congruence

refers to the degree to which a job’s RIASEC work activity

ratings (described above) are consistent with a person’s O*NET

Interest Profiler scores. In this study, this was

operationalized as the Mahalanobis distance between the two

vectors of scores. The Mahalanobis distance was used because,

for each person x job combination, it provides a single value

that captures the association between the vector scores while

accounting for the covariation among Holland’s RIASEC

dimensions. Values close to zero indicate a job’s work

activities are consistent with an individual's interests.

Large positive values indicate they are far away from an

individual’s vocational preferences and hence inconsistent

with the individual's interests. For example, a retail sales

worker job (work activity codes: R = 3.66, I = 2.33, A = 2.66, S =

COMPROMISES IN CAREER-RELATED DECISIONS 23

4.33, E = 6.33, C = 3.33) is more consistent with Person A’s

Interest Profiler scores (R = 3, I = 4, A = 4, S = 4, E = 5, C

= 3; Mahalanobis distance = 1.81) than Person B’s scores (R = 5, I

= 5, A = 4, S = 3, E = 2, C = 3; Mahalanobis distance = 3.58).

Analytic Overview. In accordance with the conditional

priorities hypothesized, models were estimated separately for

each preference category; within each category, we examined

the extent to which job sextype, job prestige, and person-job

interest congruence predicted the job that was chosen. Thus,

each hypothesis was tested using a multilevel logistic

regression. The level 1 outcome variable was job choice (i.e.,

within the forced choice pair whether the job was chosen or

not chosen). The level 1 predictors were the job sextype, job

prestige, and person-job interest congruence values associated

with each job. Given that this was a repeated measures design,

these variables were nested within person, with the subject

identifier serving as the level 2 grouping variable. Due to

the nature of the design, where everyone chose 50% of the jobs

and rejected the other 50%, the random effect reflecting

heterogeneity across individual intercepts was nil. Gender was

a level 2 predictor that was allowed to interact with the

COMPROMISES IN CAREER-RELATED DECISIONS 24

level 1 predictors. All hypotheses were tested using the lme4

package in R version 3.03 (Bates, Maechler, Bolker, & Walker,

2014; R Core Team, 2014) to estimate the multilevel logistic

regression models.

Table 1 describes the data used to test Hypotheses 1–3,

presented by gender. Table 2 summarizes the results of the

analyses. For each analysis (e.g., Table 2, large compromise

condition), the fixed effect estimate, gamma (), and the

resulting odds ratio and its 95% confidence interval are

presented. Gamma represents the average effect estimate and is

roughly analogous to conducting a separate logistic regression

for each person and averaging the resulting regression weights

across individuals. The odds ratio (OR = exponential())

represents the change in odds resulting from a unit change in

the predictor. If the OR is greater than one, this indicates

that as the predictor increases, the odds of the outcome

(i.e., choosing a particular option) increases. Conversely, if

the OR is less than one, this indicates that as the predictor

increases, the odds of the outcome decreases. Thus, a 95%

confidence interval for OR that bounds 1.00 indicates a non-

significant effect of the predictor.

COMPROMISES IN CAREER-RELATED DECISIONS 25

Results and Discussion

Preliminary analyses. In terms of their vocational

interests, as compared with women, men expressed significantly

greater interests in Realistic (Mmen = 2.68 & Mwomen = 2.06;

Welch’s t = 5.14, p < .01) and Investigative (Mmen = 3.74 &

Mwomen = 3.17; Welch’s t = 4.29, p < .01) activities, and

significantly lesser interest in Social (Mmen = 3.43 & Mwomen =

3.87; Welch’s t = -3.43, p < .01) activities. Men and women did

not differ on their interests in Artistic (Mmen = 3.26 & Mwomen =

3.34; Welch’s t = -0.56, p = .57), Enterprising (Mmen = 3.12 &

Mwomen = 3.20; Welch’s t = -0.66, p = .51) or Conventional (Mmen

= 2.59 & Mwomen = 2.54; Welch’s t = 0.35, p = .73) activities.

Overall, many more jobs were categorized as unacceptable (46%)

than either acceptable (39%) or preferred (15%). Both men and

women categorized 15% of jobs as preferred, though men found

more options acceptable (42% vs. 38%) and fewer options

unacceptable (43% vs. 47%; 2(2) = 10.66, p < .01).

Focal analyses. H1 states that when individuals choose

amongst jobs they consider unacceptable (large compromise),

chosen options are expected to be significantly more sextyped

(i.e., more feminine for women, more masculine for men) than

COMPROMISES IN CAREER-RELATED DECISIONS 26

unchosen options; no differences are expected on job prestige

or interest. As shown in Table 11, when forced to choose among

unacceptable jobs, both men and women chose the less masculine

option (men: Mchosen = -0.26 vs. Munchosen = -0.31; women: Mchosen = -

0.23 vs. Munchosen = -0.50), as well as the less prestigious

option (men: Mchosen = -0.80 vs. Munchosen = -0.76; women: Mchosen = -

0.72 vs. Munchosen = -0.68). In terms of interests, for men,

chosen options (M = 4.85) were not different from unchosen

options (M = 4.84), but for women, chosen options (M = 5.50)

were a poorer fit to their interests than unchosen options (M =

5.38). This pattern of data is consistent with H1 (at least

for women), but does not consider all three predictors

simultaneously. To test H1, we compared the job sextype, job

prestige, and person-job interest congruence of chosen versus

unchosen jobs when people were forced to choose between jobs

they had previously classified as unacceptable. As shown in

Table 2 (Large condition), consistent with H1, the odds of a

less masculine job being chosen was 1.19 (95% CI: 1.07, 1.32)

times higher than that of a more masculine job. There was also

a significant gender x sextype interaction such that women

1 Sidak adjusted paired-sample t-test results for each comparison is reported in Table 1, and not included here.

COMPROMISES IN CAREER-RELATED DECISIONS 27

were 2.52 (95% CI: 2.23, 2.85) times more likely to choose the

less masculine option over the more masculine, as compared

with men. Also consistent with H1, neither job prestige (OR =

0.96) nor interest congruence (OR = 1.00) significantly

predicted choice when large compromises were required.

H2 states that when individuals choose amongst jobs they

consider acceptable (moderate compromise), chosen options are

expected to be significantly more prestigious than unchosen

options; no differences are expected on job sextype or

interest. As shown in Table 1, when forced to choose among

acceptable jobs, for men, none of the variables predicted

choice: sextype (Mchosen = -0.19 vs. Munchosen = -0.24), prestige

(Mchosen = 0.11 vs. Munchosen = 0.12), and interest (Mchosen = 4.42 vs.

Munchosen = 4.39) were not significantly different for the chosen

and unchosen options. By contrast, for women, all of the

variables predicted choice: the less masculine option (M = -

0.07) was chosen over the more masculine option (M = -0.27),

the less prestigious option (M = 0.17) was chosen over the

more prestigious option (M = 0.18), and the more interest

congruent option (M = 4.81) was chosen over the less interest

congruent option (M = 4.88). This pattern of data suggests (a)

COMPROMISES IN CAREER-RELATED DECISIONS 28

that gender may moderate the relationships between choice and

the various predictors, and (b) that H2 is unlikely to be

supported because for both men and women, it was the less

prestigious jobs that tended to be chosen. To test H2, we

compared the job sextype, job prestige, and person-job

interest congruence of chosen versus unchosen jobs when people

were forced to choose between jobs they had previously

classified as acceptable. As shown in Table 2 (Moderate

condition) and contrary to H2, prestige (OR = 0.98) did not

significantly predict job choice. Additionally, job sextype

was a significant predictor of choice such that less masculine

jobs were 1.16 (95% CI: 1.04, 1.29) times more likely to be

chosen than more masculine jobs, and women were 1.72 (95% CI:

1.51, 1.96) times more likely to choose the less masculine

option over the more masculine option, as compared with men.

Although interest congruence (OR = 1.01) did not significantly

predict choice when moderate compromises were required, women

were more likely (OR = 0.94, 95%CI: 0.89, 0.99) than men to

choose jobs congruent with their vocational interests. That

is, women’s interest profiles were a smaller distance away

COMPROMISES IN CAREER-RELATED DECISIONS 29

from the work activity codes of the chosen option than the

unchosen option.

H3 states that when individuals choose amongst jobs they

consider preferable (minimal compromise), chosen options are

expected to be significantly more consistent with a person’s

interests than unchosen options; no differences are expected

on job sextype or prestige. As shown in Table 1, when forced

to choose among preferred jobs, both men and women chose the

more gender consistent option—men chose the more masculine

option over the less masculine option (Mchosen = -0.22 vs. Munchosen

= -0.15) whereas women chose the more less masculine option

over the more masculine option (Mchosen = 0.00 vs. Munchosen = -

0.08), although these results were only significant for women.

For men, prestige did not differ across chosen and unchosen

options (M = 0.52 for both options), whereas for women, the

chosen option was less prestigious than the unchosen option

(Mchosen = 0.75 vs. Munchosen = 0.76). In terms of interests, for

both men and women, the chosen options (Mmen = 4.67 & Mwomen =

4.63) were not significantly different from the unchosen

options (Mmen = 4.71 & Mwomen = 4.60). This pattern of data is

not consistent with H3 as interest was not a significant

COMPROMISES IN CAREER-RELATED DECISIONS 30

predictor of choice. To test H3, we compared the job sextype,

job prestige, and person-job interest congruence of chosen

versus unchosen jobs when people were forced to choose between

jobs they had previously classified as preferred. Contrary to

H3, interest (OR = 0.99) did not predict job choice. Job

sextype (OR = 0.82) and job prestige (OR = 1.01) also did not

predict choice, although this was qualified by a gender x

sextype interaction such that women were 1.51 (95% CI: 1.18,

1.93) times more likely to choose the more feminine option

over the less feminine option, as compared with men.

Study 2

In Study 1, support was found for the hypothesis that job

sextype, but not job prestige nor person-job interest

congruence predicted job choice when large compromises were

required. Contrary to Gottfredson’s predictions, however, both

men and women preferred the less masculine options. A

straightforward interpretation of this finding was however

complicated by the fact that there were far fewer men (n = 44)

than women (n = 124) in the study. Additionally, the sample

was a relatively homogenous one, comprising college students

who were relatively high on ability and social standing. Thus

COMPROMISES IN CAREER-RELATED DECISIONS 31

in Study 2, we used a more representative sample of the

population at large, one with a greater representation of men

and a wider range in levels of education. In this new sample,

an additional, independent test of Gottfredson’s (1996)

hypotheses was conducted, and the consistency of results

across Study 1 and Study 2 was evaluated.

Method

Participants. Participants were 264 individuals from

within the US. Two individuals who provided data with no

variability were removed (i.e., they provided the same

response to all questions in the occupational choice task, and

the same response to all items on the interest profiler),

resulting in a sample of 262 participants whose data were

analyzed; 44% of the sample (n = 116) were men, the rest (n =

146) were women. The sample varied widely in terms of age: 18

– 25 years old (20.3%, n = 53), 26 – 35 years old (44.1%, n =

115), 36 – 45 years old (14.2%, n = 37), and 46 or older

(21.5%, n = 56), one person did not report age; employment

status: employed full-time (56.4%, n = 145), employed part-

time (17.5%, n = 45), and unemployed (26.1%, n = 67), five

people did not provide employment status; and education

COMPROMISES IN CAREER-RELATED DECISIONS 32

levels: high school or less (10.7%, n = 28), some college

(30.0%, n = 77), Associate’s degree (12.3%, n = 32), Bachelors

degree (36.8%, n = 96), and advanced degree (10.7%, n = 28),

one person did not provide education level. Participants were

recruited via Amazon’s Mechanical Turk online data collection

platform, which has been shown to be a reliable and valid way

of collecting data from a diverse population (Buhrmester,

Kwang, & Gosling, 2011). In order to participate, participants

had to reside within the US. Interested participants were

invited to a website where they could complete the

questionnaire in exchange for US$1.15. They were told the

questionnaire would take 20 – 30 minutes to complete. They

were also told they would rate and choose among a series of

jobs, and provide some demographic information about

themselves (i.e., age, sex, employment status, education

status, and current job).

Procedure and measures. The procedure outlined, and the

measures described, in Study 1 were also used in Study 2.

Analytic overview. The same set of analyses conducted in

Study 1 was also conducted in Study 2.

Results and Discussion

COMPROMISES IN CAREER-RELATED DECISIONS 33

Preliminary analyses. In terms of their vocational

interests, as compared with women, men expressed significantly

greater interests in Realistic (Mmen = 3.04 & Mwomen = 2.61;

Welch’s t = 4.41, p < .01) and Investigative (Mmen = 3.60 &

Mwomen = 3.26; Welch’s t = 3.58, p < .01) activities, and

significantly lesser interest in Social (Mmen = 2.97 & Mwomen =

3.38; Welch’s t = -3.96, p < .01) activities. Men and women

expressed similar interests in Artistic (Mmen = 3.33 & Mwomen =

3.49; Welch’s t = -1.69, p = .09), Enterprising (Mmen = 2.91 &

Mwomen = 3.07; Welch’s t = -1.71, p = .09) and Conventional (Mmen

= 3.16 & Mwomen = 3.26; Welch’s t = -1.01, p = .32) activities.

Overall, more jobs were categorized as acceptable (42%) than

either unacceptable (38%) or preferred (20%). Both men and

women found 38% of the jobs unacceptable, though men found

more options preferred (21% vs. 19%) and fewer options

acceptable (40% vs. 43%; 2(2) = 14.23, p < .01), as compared

with women. (Reported results for men do not total to 100% due

to rounding.)

Focal analyses. H1 states that when individuals choose

amongst jobs they consider unacceptable (large compromise),

chosen options are expected to be significantly more sextyped

COMPROMISES IN CAREER-RELATED DECISIONS 34

(i.e., more feminine for women, more masculine for men) than

unchosen options; no differences are expected on job prestige

or interest. As shown in Table 32, when forced to choose among

unacceptable jobs, women (Mchosen = -0.34 vs. Munchosen = -0.48) but

not men (Mchosen = -0.20 vs. Munchosen = -0.22) chose the

significantly less masculine option. For both men and women,

chosen and unchosen options did not differ in prestige (men:

Mchosen = -0.38 vs. Munchosen = -0.39; women: Mchosen & Munchosen = -0.31)

or interest (men: Mchosen = 4.54 vs. Munchosen = 4.58; women: Mchosen =

4.68 vs. Munchosen = 4.70). Similar to Study 1, this pattern of

data is consistent with H1 (at least for women), but does not

consider all three predictors simultaneously. To test H1, we

compared the job sextype, job prestige, and person-job

interest congruence of chosen versus unchosen jobs when people

were forced to choose between jobs they had previously

classified as unacceptable. As shown in Table 4 (Large

condition), although job sextype (OR = 1.05) did not

significantly predict choice, this result was qualified by a

significant gender x sextype interaction such that women were

1.60 (95% CI: 1.40, 1.83) times more likely than men to choose

2 Sidak adjusted paired-sample t-test results for each comparison is reported in Table 3, and not included here.

COMPROMISES IN CAREER-RELATED DECISIONS 35

the less masculine option over the more masculine option. Also

consistent with H1, job prestige (OR = 1.01) and interest

congruence (OR = 0.98) did not significantly predict choice

when large compromises were required. Consistent with Study 1,

these results provide support for H1, but show that the effect

was stronger for women than for men.

H2 states that when individuals choose amongst jobs they

consider acceptable (moderate compromise), chosen options are

expected to be significantly more prestigious than unchosen

options; no differences are expected on job sextype or

interest. As shown in Table 3, when forced to choose among

acceptable jobs, men chose jobs more consistent with their

interests (Mchosen = 4.08 vs. Munchosen = 4.15), and neither sextype

(Mchosen = -0.25 vs. Munchosen = -0.21) nor prestige (Mchosen & Munchosen = -

0.06) predicted choice. By contrast, women chose less

masculine options (M = -0.12) over more masculine options (M =

-0.20), and neither prestige (Mchosen & Munchosen = -0.12) nor

interest (Mchosen = 4.23 vs. Munchosen = 4.24) predicted choice. This

pattern of data suggests that H2 is unlikely to be supported

because for both men and women, choice did not differ based on

prestige. To test H2, we compared the job sextype, job

COMPROMISES IN CAREER-RELATED DECISIONS 36

prestige, and person-job interest congruence of chosen versus

unchosen jobs when people were forced to choose between jobs

they had previously classified as acceptable. As shown in

Table 4 (Moderate condition) and contrary to H2, prestige (OR

= 1.02) did not significantly predict choice. Additionally,

job sextype was a significant predictor of choice such that

less masculine jobs were less likely (OR = 0.89, 95% CI: 0.81,

0.97) to be chosen than more masculine jobs, and this was

qualified by a significant gender x sextype interaction such

that women were 1.51 (95% CI: 1.34, 1.70) times more likely to

choose the less masculine option over the more masculine

option, as compared with men. Interest congruence (OR = 0.97)

did not significantly predict choice when moderate compromises

were required. Similar to Study 1, no support was found for

H2.

H3 states that when individuals choose amongst jobs they

consider preferable (minimal compromise), chosen options are

expected to be significantly more consistent with a person’s

interests than unchosen options; no differences are expected

on job sextype or prestige. As shown in Table 3, when forced

to choose among preferred jobs, both men (Mchosen = 3.98 vs.

COMPROMISES IN CAREER-RELATED DECISIONS 37

Munchosen = 4.09) and women (Mchosen = 4.14 vs. Munchosen = 4.19) chose

jobs more consistent with their interests, although these

results were not statistically significant for women. For both

men and women, neither sextype (men: Mchosen = -0.27 vs. Munchosen =

-0.24; women: Mchosen = -0.09 vs. Munchosen = -0.12) nor prestige

(men: Mchosen = 0.32 vs. Munchosen = 0.30; women: Mchosen = 0.24 vs.

Munchosen = -0.25) predicted choice. This pattern of data is

consistent with H3, but does not consider all three predictors

simultaneously. To test H3, we compared the job sextype, job

prestige, and person-job interest congruence of chosen versus

unchosen jobs when people were forced to choose between jobs

they had previously classified as preferred. As shown in Table

4 (Minimal condition) and consistent with H3, jobs less

congruent with one’s interest (i.e., with a larger distance

between person’s interest and jobs work activity ratings) were

less (OR = 0.94, 95% CI: 0.89, 1.00) likely to be chosen than

jobs more congruent with one’s interest. Also consistent with

this hypothesis, neither sextype (OR = 0.93) nor prestige (OR

= 1.03) significantly predicted choice when minimal

compromises were required. These results differ from those in

Study 1, which did not find support for H3 but instead found

COMPROMISES IN CAREER-RELATED DECISIONS 38

support for a significant gender x sextype interaction

indicating that although both men (Mchosen = -0.22 vs. Munchosen = -

0.15) and women (Mchosen = 0.00 vs. Munchosen = -0.08) chose more

gender-consistent options, the results were statistically

significant only for women. The relation between sextype and

interest congruence is discussed in the next section, and

proposed as a plausible explanation for the discrepancy in

support for H3 between Study 1 and Study 2.

General Discussion

This study tested Gottfredson’s (1996) hypothesis that as

required compromise increases, people’s attention to job

aspects more central to their self-concept also increases.

Specifically, this study tested if the relative importance of

job sextype, job prestige, and person-job interest congruence

for predicting job choice changes as the level of compromise

required changes. Gottfredson (1996) argued that the

centrality of different concepts reflects when they were

incorporated into one’s self-concept. Based on a typical

developmental progression where notions of gender, then

status, then unique personal characteristics (e.g., vocational

interests) develop, she proposed that gender identities are

COMPROMISES IN CAREER-RELATED DECISIONS 39

more central than status identities, which are in turn more

central than personal identities. Correspondingly, when

choosing amongst jobs, she proposed that people would be most

willing to compromise on interest preferences (minimal

compromise), followed by prestige preferences (moderate

compromise), followed by sextype preferences (large

compromise). Although research has demonstrated the importance

of each aspect for influencing choice (Gottfredson, 1996),

less research (Blanchard & Lichtenberg, 2003) has tested if

level of compromise moderates the importance of each aspect

for predicting job choice.

In general, and consistent with Gottfredson’s (1996)

hypotheses, results across two studies indicated that the

relative importance of job aspects changes as the level of

compromise changes. At a more detailed level, specific

predictions of her theory received more mixed support: (1)

across both studies, job sextype predicted choice when large

compromises were required; (2) across both studies, job

prestige did not predict choice when moderate compromises were

required; and (3) in Study 2 but not Study 1, person-job

COMPROMISES IN CAREER-RELATED DECISIONS 40

interest congruence predicted choice when minimal compromises

were required. Each of these findings is discussed in turn.

First, although job sextype predicted choice when large

compromises were required, these results were driven largely

by women’s choices. In these studies, both men and women chose

less masculine options over more masculine options. These

results provide partial support for Gottfredson’s prediction

that people make choices to protect their gender identity when

large compromises are required; results from women are

consistent with the prediction but results from men are not.

However, the unstandardized difference (D = Mchosen - Munchosen, see

Tables 1 and 3) was much larger for women (D = 0.27 in Study 1

& D = 0.14 in Study 2), than for men (D = 0.05 in Study 1 & D

= 0.02 in Study 2). And further, the sextype of unacceptable

jobs was predominantly masculine (i.e., mean values of sextype

in the unacceptable condition were all negative), suggesting

that these jobs were more consistent with men’s gender

identities than women’s. Thus women’s choices, but not men’s,

reflected a need to protect gender identity. Consistent with

the world of work (Deng et al., 2007), about a third of the

jobs in our occupational choice task were characterized by

COMPROMISES IN CAREER-RELATED DECISIONS 41

Realistic work activities. Such work activities tend to vary

widely on prestige, and to be concentrated in masculine jobs.

However, to effectively test Gottfredson’s (1996) hypothesis

regarding the relative importance of sextype when making large

compromises for men, a stimuli set oversampling Social and

Conventional jobs, i.e., more feminine jobs, might be

required.

Second, even though job prestige did not predict choice

when moderate compromises were required, these results should

be interpreted in light of the very clear differentiation of

jobs by prestige when participants initially categorized these

jobs by preference (see Tables 1 and 3). Across both studies,

preferred occupations were higher in prestige than acceptable

occupations, which in turn were higher in prestige than

unacceptable occupations. These results are consistent with

earlier results (Blanchard & Lichtenberg, 2003, Figure 5)

showing that prestige tends to be an important concern across

levels of compromise required. These results are also

consistent with previous tests of Gottfredson’s (1981)

original predictions (i.e., Leung, 1993; Leung & Plake, 1990).

Taken together, these results suggest that individuals may be

COMPROMISES IN CAREER-RELATED DECISIONS 42

deciding on job suitability based primarily on their

perceptions of how prestigious jobs are. Thus, when forced to

make finer distinctions with each category, the choice set is

effectively equated on prestige, and differences on job

sextype or interest congruence may be found instead.

Third, person-job interest congruence predicted choice

when minimal compromises were required, but only in Study 2.

Although sextype and prestige within the stimuli set were

disentangled by the judicious selection of jobs, their

relations with work activity ratings were confounded. It is

plausible the result obtained in Study 1 is due, at least in

part, to sextype and interest congruence being confounded, as

both men and women tended to choose gender-consistent options.

In support of this explanation, participants in Study 2 also

chose more gender-consistent options (see Table 3), although

in Study 2 the effect was not statistically significant.

Further, samples vary in the composition of individuals with

different profiles of vocational interests: Study 2

participants preferred more jobs than Study 1 participants

(20% vs. 15%). Because there were few options with which to

test the preferences of Study 1 participants, the lack of

COMPROMISES IN CAREER-RELATED DECISIONS 43

effect may also be explained by range restriction. Consistent

with this explanation, preferred options were less congruent

with the interests of Study 1 participants than with the

interests of Study 2 participants (see Tables 1 and 3).

Implications

Although it was difficult to independently account for

the effects of sextype, prestige, and interest on job choice,

there was evidence that sextype and prestige influence job

choice. There was also support for the fact that these aspects

are differentially salient across different levels of

compromise. In career-counseling situations, this suggests

greater emphasis needs to be placed on other job aspects

besides vocational interests, especially when some degree of

compromise is required. For example, individuals who want to

be medical doctors but struggle with the necessary pre-

requisites, nursing would be one alternative that could be

proposed based on the perceived similarity of the work-

activities. However, switching to law school might be the

alternative one might propose based on these research

findings: prestige is comparable across occupations, although

the work activities are fairly different. In this situation,

COMPROMISES IN CAREER-RELATED DECISIONS 44

sextype is unlikely to be compromised. Unlike this situation,

nursing likely represents a sextype compromise for men, and a

prestige compromise for men and women.

For personnel selection researchers, these findings

suggest a greater need to recognize that people are often

circumscribing large swathes of the world of work. Although

selection researchers have traditionally focused on selecting

the “best” individuals from the applicant pools, this research

indicates that greater efforts are required to first recruit

individuals from the general population into the relevant

applicant pools. People can and often do make choices based on

their perceptions of occupations. To the degree that these

perceptions are simply stereotypes (e.g., engineering is for

boys), they may preclude the entry of otherwise capable

individuals into these jobs. Recruiting strategies

highlighting less salient aspects of the job may be one step

in the right direction (Eccles, 2007). For example,

highlighting that engineers frequently work in teams stresses

communion more so than agency, and communion is often viewed

as less masculine than agency (Deaux & LaFrance, 1998).

COMPROMISES IN CAREER-RELATED DECISIONS 45

This paper tested specific job aspects (i.e., sextype,

prestige, and interests) as they relate to compromise and job

choice because these aspects have been shown to explain a

considerable amount of variance in people’s occupational

preferences (Deng et al., 2007). Other theories (e.g., Gati,

1993; Gati, Houminer, & Aviram, 1998) however, take a content-

free approach and focus instead on how the choice is

structured. Such theories complement the current approach. For

example, they indicate that the difficulty of a compromise

decision could be ameliorated by shifting a person’s focus

from occupational alternatives (e.g., travel agent vs. police

investigator) to specific comparisons of aspects (e.g.,

working hours vs. working conditions), or to levels of aspects

(e.g., working indoors vs. working outdoors). In a related

fashion, people may revise their occupational image if they

are provided more experience with the occupation. Activities

such as informational interviews, job shadowing, or

internships might provide opportunities to put flesh on the

bare bones of a stereotype.

Limitations and Future Directions

COMPROMISES IN CAREER-RELATED DECISIONS 46

The occupational choice task was designed to capture a

wide range of jobs. Jobs were selected such that they ranged

across the full spectrum of prestige, and from extremely

masculine to extremely feminine jobs. As a consequence, the

RIASEC types distribution for selected jobs also reflects the

distribution of jobs in the US. As highlighted by Gottfredson

(1996), by the time people reach adolescence, and especially

if they are college-bound adolescents, many of these options

would have been progressively eliminated from consideration.

Thus, even though the set of occupations may seem unrealistic

for some groups of people (e.g. college seniors), such a

stimuli set is needed to effectively test conditional

priorities when at least moderate levels of compromise are

required; it is only with such a diverse pool of items that we

can effectively assess the thresholds of what is acceptable

and what is not. However, because relatively few options were

deemed preferable in both studies, these studies provided less

precision in teasing apart the relative importance of

different job aspects when minimal compromise was required.

Future studies using different groups of people (e.g., gifted

populations, rural populations, lesbian, gay, bisexual or

COMPROMISES IN CAREER-RELATED DECISIONS 47

transgendered populations, etc.) and/or different sets of

occupations are required to accumulate sufficient information

to make more generalizable claims.

In this study, one’s self-reported sex (male or female)

served as a proxy for one’s gender identity even though

gendered experience is not synonymous with sex (Deaux, 2001).

Also, the centrality and importance of one’s gender identity

was assumed, but not measured. And further, the overall

structure and relation of different identities within the self

were assumed based on typical developmental trajectories of

when each identity is incorporated into people’s self-concept

(Gottfredson, 1996). Given the results in the present study

and previous studies, which seem to indicate that (in some

cases) job prestige may be at least as important as job

sextype in predicting choice, testing the centrality and

importance of various aspects of people’s identity is probably

warranted. Future research would do well to explicitly measure

different aspects of identity (e.g., Obst & White, 2005), or

even attempt to capture the ordering and relations among

different aspects of self (Rosenberg, 1997). This suggestion

focuses on better measuring the social identities related to

COMPROMISES IN CAREER-RELATED DECISIONS 48

one’s self-concept, and implies that choice is likely

predicted by individual differences in how cognizant of and

committed to their various identities people are likely to be.

Other individual differences, such as people’s willingness to

compromise (Creed & Blume, 2013; Taylor & Pryor, 1985; Wee,

2013), also warrants future research attention.

COMPROMISES IN CAREER-RELATED DECISIONS 49

References

Armstrong, P. I., & Crombie, G. (2000). Compromises in

adolescents’ occupational aspirations and expectations

from grades 8 to 10. Journal of Vocational Behavior, 56, 82-89.

doi:10.1006/jvbe.1999.1709

Bates, D., Maechler, M., Bolker, B., & Walker, S. (2014).

lme4: Linear mixed-effects models using Eigen and S4. R

package version 1.1-5. Available from http://cran.r-

project.org/package=lme4.

Becker, G. S. (1993). Human capital: A theoretical and empirical analysis,

with special reference to education (3rd ed.). Chicago, IL: University

of Chicago Press.

Blanchard, C. A., & Lichtenberg, J. W. (2003). Compromise in

career decision making: A test of Gottfredson's theory.

Journal of Vocational Behavior, 62, 250-271. doi:10.1016/S0001-

8791(02)00026-X

Blau, P. M. & Duncan, O. D. (1967). The American occupational

structure. New York, NY: Wiley.

Buhrmester, M., Kwang, T., & Gosling, S. D. (2011). Amazon’s

Mechanical Turk a new source of inexpensive yet high-

COMPROMISES IN CAREER-RELATED DECISIONS 50

quality data? Perspectives on Psychological Science, 6, 3-5.

doi:10.1177/1745691610393980

Bureau of Labor Statistics (2004a). Household data: Annual averages.

Current Population Survey. Available from

http://www.bls.gov/cps/cpsaat11.pdf.

Bureau of Labor Statistics (2004b). National employment and wage

data. The Occupational Employment Statistics Survey.

Available from http://www.bls.gov/oes/oes_data.htm.

Creed, P. A., & Blume, K. (2013). Well-being, and action

behaviors in young adults in career transitions. Journal of

Career Assessment, 21, 3-19. doi:10.1177/1069072712453830

Cunha, F. & Heckman, J. J. (2007). Identifying and estimating

the distributions of ex post and ex ante returns to

schooling. Labor Economics, 14, 870-893.

doi:10.1016/j.labeco.2007.06.002

Dawis, R. V., & Lofquist, L. H. (1984). A psychosocial theory of work

adjustment. Minneapolis, MN: University of Minnesota Press.

Deaux, K. (2001). Social identity. In J. Worrell (Ed.),

Encyclopedia of women and gender (Vol. 2). San Diego, CA: Academic

Press.

COMPROMISES IN CAREER-RELATED DECISIONS 51

Deaux, K. & La France, M. (1998). Gender. In D.T. Gilbert,

S.T. Fiske, & G. Lindzey (Eds.), The handbook of social

psychology (4th ed., Vol. 2). New York, NY: McGraw-Hill.

Deng, C. P., Armstrong, P. I., & Rounds, J. (2007) The fit of

Holland's RIASEC model to US occupations. Journal of Vocational

Behavior, 71, 1-22. doi:10.1016/j.jvb.2007.04.002

Eccles, J. S. (2007). Where are all the women? Gender

differences in participation in physical science and

engineering. In S. J. Ceci & W. M. Williams (Eds.), Why

aren't more women in science? Top researchers debate the evidence.

Washington, DC: American Psychological Association.

Eggerth, D. E., Bowles, S. M., Tunick, R. H., & Andrew, M. E.

(2005). Convergent validity of O*NET Holland code

classifications. Journal of Career Assessment, 13, 150-168.

doi:10.1177/1069072704273124

Friedland, D. S., & Price, R. H. (2003). Underemployment:

Consequences for the health and well-being of workers.

American Journal of Community Psychology, 32, 33-45.

Gati, I. (1993). Career compromises. Journal of Counseling Psychology,

40, 416-424. doi:10.1037/0022-0167.40.4.416

COMPROMISES IN CAREER-RELATED DECISIONS 52

Gati, I., Houminer, D., & Aviram, T. (1998). Career

compromises: Framings and their implications. Journal of

Counseling Psychology, 45, 505-514. doi:10.1037/0022-

0167.45.4.505

Ginzberg, E., Ginsburg, S. W., Axelrad, S. & Herma, J. L.

(1951). Occupational choice: An approach to a general theory. New

York, NY: Columbia University Press.

Gottfredson, G. D., & Holland, J. L. (1996). Dictionary of Holland

occupational codes (3rd ed.). Odessa, FL: Psychological Assessment

Resources, Inc.

Gottfredson, L. S. (1981). Circumscription and compromise: A

developmental theory of occupational aspirations. Journal of

Counseling Psychology, 28, 545-579. doi:10.1037/0022-

0167.28.6.545

Gottfredson, L. S. (1996). Gottfredson's theory of

circumscription, compromise. In D. Brown (Ed.), Career

Choice and Development (3rd ed.). San Francisco, CA: Jossey

Bass.

Hesketh, B., Elmslie, S., & Kaldor, W. (1990). Career

compromise: An alternative account to Gottfredson's

COMPROMISES IN CAREER-RELATED DECISIONS 53

theory. Journal of Counseling Psychology, 37, 49-56.

doi:10.1037/0022-0167.37.1.49

Holland, J. L. (1997). Making vocational choices: A theory of vocational

personalities and work environments (3rd ed.). Odessa, FL:

Psychological Assessment Resources.

Holt, P. A. (1989). Differential effect of status and interest

in the process of compromise. Journal of Counseling Psychology, 36,

42-47. doi: 10.1037/0022-0167.36.1.42

Hulin, C. L. (2002). Lessons from industrial and

organizational psychology. In J. M. Brett & F. Drasgow

(eds.), The psychology of work: Theoretically based empirical research.

New York, NY: Lawrence Erlbaum.

Hulin, C. L., & Judge, T. A. (2003). Job attitudes. In W. C.

Borman, D. R. Ilgen, and R. J. Klimoski (Eds.), Handbook of

psychology: Industrial and organizational psychology, Vol. 12. New York,

NY: John Wiley & Sons.

Junk, K. E., & Armstrong, P. I. (2010). Stability of career

aspirations: A longitudinal test of Gottfredson’s theory.

Journal of Career Development, 37(3), 579-598.

doi:10.1177/0894845309350921

COMPROMISES IN CAREER-RELATED DECISIONS 54

Leung, S. A. (1993). Circumscription and compromise: A

replication study with Asian Americans. Journal of Counseling

Psychology, 40, 188-193. doi:10.1037/0022-0167.40.2.188

Leung, S. A., & Plake, B. S. (1990). A choice dilemma approach

for examining the relative importance of sex-type and

prestige preferences in the process of career choice

compromise. Journal of Counseling Psychology, 37, 399-406.

doi:10.1037/0022-0167.37.4.399

McCloy, R., Waug, G., Medsker, G., Wall, J., Rivkin, D., &

Lewis, P. (1999). Determining the occupational reinforcer

patterns for the O*NET occupational units. Raleigh, NC:

National Center for O*Net Development.

Murphy, G. C., & Athanasou, J. A. (1999) The effect of

unemployment on mental health. Journal of Occupational and

Organizational Psychology, 72, 83-99.

doi:10.1348/096317999166518

Obst, P. L., & White, K. M. (2005). Three-dimensional strength

of identification across group memberships: A

confirmatory factor analysis. Self and Identity, 4, 69-80.

doi:10.1080/13576500444000182

COMPROMISES IN CAREER-RELATED DECISIONS 55

O*NET Resource Center (2003). The O*NET analyst database.

O*NET Consortium. Available from

http://www.onetcenter.org/database.htm#archive.

Oswald, F., Campbell, J., McCloy, R., Rivkin, D., & Lewis, P.

(1999). Stratifying occupational units by specific vocational preparation.

Raleigh, NC: National Center for O*NET Development.

R Core Team (2014). R: A language and environment for

statistical computing. R Foundation for Statistical

Computing, Vienna, Austria. Available from http://www.R-

project.org/.

Rosenberg, S. (1997). Multiplicity of Selves. In R. D. Ashmore

and L. Jussim (Ed.), Self and identity: Fundamental issues (pp. 23 –

45). New York, NY: Oxford University Press.

Rounds, J., Smith, T., Hubert, L., Lewis, P., & Rivkin, D.

(1999). Development of occupational interest profiles for O*NET.

Raleigh, NC: National Center for O*NET Development.

Rounds, J., Su, R., Lewis, P., & Rivkin, D. (2010). O*NET

interest profiler short form psychometric characteristics: Summary.

Raleigh, NC: National Center for O*NET Development.

Schoon, I. (2001). Teenage job aspirations and career

attainment in adulthood: A 17-year follow-up of teenagers

COMPROMISES IN CAREER-RELATED DECISIONS 56

who aspired to become scientists, health professionals,

or engineers. International Journal of Behavioral Development, 25,

124-132. doi:10.1080/01650250042000186

Sewell, W. H., Haller, A. O., & Ohlendorf, G. W. (1970). The

educational and early occupational attainment process:

Replication and revision. American Sociological Review, 35, 1014-

1027.

Taylor, N. B., & Pryor, R. G. (1985). Exploring the process of

compromise in career decision making. Journal of Vocational

Behavior, 27, 171-190. doi:10.1016/0001-8791(85)90031-4

Wall, J. E., & Baker, H. E. (1997). The Interest-Finder:

Evidence of validity. Journal of Career Assessment, 5, 255-273.

doi:10.1177/106907279700500301

Wee, S. (2013). Development and initial validation of the

willingness to compromise scale. Journal of Career Assessment, 21,

487-501. doi:10.1177/1069072712475281

COMPROMISES IN CAREER-RELATED DECISIONS 57

Table 1

Means and SDs for Study 1Variables by Preference Category, Gender, and Choice

Variables Men (n = 44) Women (n = 124)U C t U C t

UnacceptableSextype -0.31 -0.26 -3.08 -0.50 -0.23 -34.86

(0.49) (0.56) (0.46) (0.51)Prestige -0.76 -0.80 4.76 -0.68 -0.72 7.66

(1.02) (1.02) (1.04) (1.07)Interest 4.84 4.85 -.50 5.38 5.50 -9.03

(1.33) (1.34) (1.60) (1.63)

AcceptableSextype -0.24 -0.19 -2.61 -0.27 -0.07 -18.22

(0.54) (0.57) (0.54) (0.54)Prestige 0.12 0.11 2.07 0.18 0.17 3.44

(0.93) (0.94) (0.88) (0.90)Interest 4.39 4.42 -1.15 4.88 4.81 3.48

(1.33) (1.34) (1.47) (1.45)

PreferredSextype -0.15 -0.22 1.78 -0.08 0.00 -3.27

(0.62) (0.59) (0.64) (0.65)Prestige 0.52 0.52 0.31 0.76 0.75 2.98

(0.77) (0.78) (0.55) (0.55)Interest 4.71 4.67 0.82 4.60 4.63 -.84

(2.39) (2.48) (1.41) (1.44)Note. U = Unchosen option, C = Chosen option, t = statistic for

the paired sample t-test. Interest = Person-job interest

congruence. Standard deviations are reported in parentheses.

Multiple comparisons Sidak adjustment to control Type I error

rate at = .05 for the 18 tests indicates a critical t value

of + 2.83. Significant t-tests are bolded and italicized.

COMPROMISES IN CAREER-RELATED DECISIONS 58

Table 2

Coefficients of the Multilevel Logistic Models Predicting Chosen Options across Compromise Conditions (Study1)

ParameterLarge Moderate Minimal

OR OR 95% CI OR OR 95% CI OR OR 95% CI

Intercept -

0.0

2

0.98

[0.93;

1.04] 0.0

1 1.01

[0.94;

1.09] 0.0

0 1.00

[0.83;

1.21]

Women 0.10 1.10

[1.03; 1.18] -

0.07 0.93

[0.85;

1.01]

0.0

2 1.02

[0.80;

1.31]

Sextype 0.17 1.19

[1.07; 1.32]

0.15 1.16

[1.04; 1.29] -

0.20 0.82

[0.66;

1.01]

Prestige

-

0.0

4

0.96

[0.91;

1.01] -

0.02 0.98

[0.91;

1.04] 0.0

1 1.01

[0.86;

1.20]

COMPROMISES IN CAREER-RELATED DECISIONS 59

Interest

0.

001.00

[0.96;

1.04]

0.0

1 1.01

[0.97;

1.06]

-

0.01 0.99

[0.94;

1.04]

Women x

Sextype 0.93 2.52

[2.23; 2.85]

0.54 1.72

[1.51; 1.96]

0.41 1.51

[1.18; 1.93]

Women x

Prestige

-

0.0

1

0.99

[0.93;

1.05] 0.0

1 1.01

[0.93;

1.09] -

0.08 0.92

[0.74;

1.15]

Women x

Interest

0.

021.02

[0.98;

1.07] -0.07 0.94

[0.89; 0.99] 0.0

1 1.01

[0.94;

1.09]

Note. OR = Odds ratio, Large compromise corresponds to choosing among unacceptable jobs,

Moderate compromise corresponds to choosing among acceptable jobs, and Minimal compromise

corresponds to choosing among preferred jobs. Gender was dummy coded (men = 0, women =

1), and so the parameter was named to indicate the direction of the effect.

Significant results are bolded and italicized.

COMPROMISES IN CAREER-RELATED DECISIONS 60

Table 3

Means and SDs for Study 2 Variables by Preference Category, Gender, and Choice

Variables Men (n = 116) Women (n = 146)U C t U C t

UnacceptableSextype -0.22 -0.20 -0.96 -0.48 -0.34 -11.20

0.56 0.55 0.51 0.53Prestige -0.39 -0.38 -1.29 -0.31 -0.31 -0.73

1.08 1.08 1.08 1.06Interest 4.58 4.54 1.83 4.70 4.68 0.92

1.55 1.53 1.45 1.45

AcceptableSextype -0.21 -0.25 2.72 -0.20 -0.12 -7.04

0.55 0.55 0.54 0.54Prestige -0.06 -0.06 0.18 -0.12 -0.12 0.15

0.99 0.99 1.03 1.03Interest 4.15 4.08 3.43 4.24 4.23 0.84

1.43 1.36 1.40 1.35

PreferredSextype -0.24 -0.27 1.20 -0.12 -0.09 -1.31

0.56 0.57 0.57 0.59Prestige 0.30 0.32 -2.47 0.25 0.24 2.09

0.88 0.87 0.91 0.92Interest 4.09 3.98 3.61 4.19 4.14 1.66

1.38 1.37 1.46 1.40Note. U = Unchosen option, C = Chosen option, t = statistic for

the paired sample t-test. Interest = Person-job interest

congruence. Standard deviations are reported in parentheses.

Multiple comparisons Sidak adjustment to control Type I error

rate at = .05 for the 18 tests indicates a critical t value

of + 2.83. Significant t-tests are bolded and italicized.

COMPROMISES IN CAREER-RELATED DECISIONS 61

Table 4

Coefficients of the Multilevel Logistic Models Predicting Chosen Options across Compromise Conditions (Study 2)

ParameterLarge Moderate Minimal

OR OR 95% CI OR OR 95% CI OR OR 95% CI

Intercept 0.0

01.00

[0.95;

1.06]

-

0.01 0.99

[0.94;

1.04]

-

0.03 0.97

[0.89;

1.05]

Women0.09 1.09

[1.01; 1.18] -

0.02 0.98

[0.92;

1.05] 0.02 1.02

[0.91;

1.15]

Sextype

0.0

51.05

[0.95;

1.16] -0.12 0.89

[0.81; 0.97] -

0.07 0.93

[0.81;

1.06]

Prestige

0.0

11.01

[0.96;

1.06] 0.02 1.02

[0.96;

1.07] 0.03 1.03

[0.95;

1.12]

Interest -

0.0

0.98 [0.95;

1.02]

-

0.03

0.97 [0.93;

1.00]

-0.06 0.94 [0.89; 1.00]

COMPROMISES IN CAREER-RELATED DECISIONS 62

2

Women x

Sextype0.47 1.60

[1.40; 1.83]

0.41 1.51

[1.34; 1.70]

0.17 1.18

[0.98;

1.42]

Women x

Prestige

-

0.0

4

0.96

[0.90;

1.02] -

0.02 0.98

[0.92;

1.05] -

0.04 0.96

[0.85;

1.08]

Women x

Interest

-

0.0

1

0.99

[0.95;

1.04]

0.01 1.01

[0.97;

1.06]

0.03 1.03

[0.96;

1.11]

Note. OR = Odds ratio, Large compromise corresponds to choosing among unacceptable jobs,

Moderate compromise corresponds to choosing among acceptable jobs, and Minimal compromise

corresponds to choosing among preferred jobs. Gender was dummy coded (men = 0, women =

1), and so the parameter was named to indicate the direction of the effect.

Significant results are bolded and italicized.

COMPROMISES IN CAREER-RELATED DECISIONS 63

AppendixOccupational Titles Used as Experimental Stimuli

Table A1

List of Occupational Titles and Their Prestige, Sextype, and RIASEC Work Activity Ratings

Occupational TitleBlock

Presti

ge

Sextyp

eR I A S E C

Fisher man/woman 1 -2.08 -1.39 6.33 2.66 2.00 2.66 5.00 3.33

Janitor 1 -2.27 -0.41 6.66 1.66 1.66 2.33 2.33 3.00

Meter reader 1 -1.45 0.01 5.66 2.33 1.66 2.33 3.66 6.33

Gas pumping station

operator1 -0.97 -0.47 6.66 2.66 1.66 1.66 2.33 3.33

Host/hostess 1 -1.89 0.62 4.00 1.66 2.00 5.00 5.66 3.66

Rail transportation

worker1 -1.57 -0.61 6.33 2.33 1.66 1.66 2.00 4.00

Trapper 1 -1.85 -0.87 6.66 2.33 2.00 1.66 3.00 3.00

Packer/packager 1 -1.89 -0.22 6.66 1.66 1.66 1.66 2.00 3.00

COMPROMISES IN CAREER-RELATED DECISIONS 64

Private household

worker1 -2.27 0.09 6.66 1.66 2.00 3.33 2.33 3.33

Cashier 1 -1.89 0.13 4.00 2.00 2.00 3.00 4.33 6.00

New accounts clerk 2 -0.58 -0.19 2.66 2.00 1.66 4.33 6.00 6.00

Plasterer 2 -0.52 -0.52 6.66 2.00 1.66 1.66 2.66 3.33

Library assistant 2 -0.61 0.19 4.33 3.33 2.66 3.66 3.66 6.00

Cook 2 -0.71 0.12 6.00 2.66 4.33 3.33 5.66 3.66

Sheetmetal worker 2 -0.49 -0.92 6.66 2.33 1.66 1.66 2.33 2.66

Small engine mechanic 2 -0.68 -1.27 6.66 2.66 1.66 1.66 3.00 3.33

Bank teller 2 -0.71 0.04 3.66 2.33 1.66 3.66 4.66 6.66

Retail sales worker 2 -0.45 0.16 3.66 2.33 2.66 4.33 6.33 3.33

Court reporter 2 -0.53 0.32 2.33 3.00 2.00 3.66 4.33 6.33

Correction officer 2 -0.55 -0.38 6.00 2.00 2.00 5.00 3.66 3.66

Probation officer 3 -0.08 -0.02 3.33 2.66 2.66 6.66 3.33 4.33

COMPROMISES IN CAREER-RELATED DECISIONS 65

Radiological

technologist3 0.02 -0.39 6.00 5.00 2.00 4.00 2.66 3.00

Numerical tool and

process control

programmer

3 -0.10 -0.63 6.00 5.00 2.00 2.00 2.66 5.66

Purchaser/buyer 3 -0.04 0.46 2.66 3.00 2.66 4.00 6.33 6.00

Desktop publisher 3 -0.10 -0.05 6.33 2.33 4.00 1.66 2.33 3.33

Computer/office

repairer3 -0.07 -1.01 6.66 3.33 1.66 1.66 2.00 4.00

Tool and die maker 3 -0.07 -0.79 6.66 3.00 1.66 1.66 2.33 3.66

Adjuster,

investigator, or

collector

3 -0.11 -0.59 2.66 4.66 2.00 4.00 6.66 4.00

COMPROMISES IN CAREER-RELATED DECISIONS 66

Hotel manager 3 -0.10 0.06 3.00 1.66 2.00 4.66 6.33 5.00

Cost estimator 3 -0.01 -0.54 3.33 3.00 2.00 3.66 5.33 6.33

Optician 4 0.65 -0.15 4.66 3.33 3.00 4.00 6.00 5.00

Urban/regional

planner4 0.75 -0.29 3.00 6.00 3.66 3.33 5.33 3.66

Designer 4 0.66 1.24 4.33 3.00 6.66 3.66 5.66 2.33

Anthropologist 4 0.69 -0.17 2.00 6.66 4.33 4.66 2.00 3.33

Computer programmer 4 0.69 -0.85 4.66 6.00 2.33 2.00 2.33 4.33

Botanist 4 0.79 -0.26 5.66 7.00 2.33 2.00 3.00 4.00

Market research

analyst4 0.63 0.13 3.00 6.00 2.66 2.66 5.00 4.00

Librarian 4 0.72 0.18 2.66 4.00 6.00 3.66 3.66 5.00

Farm operator/

manager4 0.57 -0.63 5.33 3.33 2.00 3.00 6.33 4.66

Geologist 4 0.73 -0.66 6.00 7.00 3.00 2.00 2.33 4.00

COMPROMISES IN CAREER-RELATED DECISIONS 67

Electrical engineer 5 1.01 -1.19 6.33 6.33 2.33 2.33 3.33 3.66

Economist 5 1.04 -0.83 3.33 6.33 2.66 3.33 4.66 4.33

Education

administrator5 1.01 0.43 2.00 4.33 3.66 6.66 5.66 3.00

Director 5 1.39 -0.14 3.66 2.66 6.66 4.33 5.66 3.33

Biological scientist 5 1.11 -0.58 5.00 7.00 3.33 2.00 2.33 3.66

Newscaster 5 1.01 -0.11 2.66 4.00 6.66 5.00 4.33 3.00

Occupational

therapist5 0.95 0.78 4.66 4.33 4.00 6.66 3.00 3.00

Meteorologist 5 1.01 -0.61 5.66 7.00 3.00 3.00 2.66 3.66

Political science

professor5 1.42 -0.05 2.33 5.33 4.00 6.00 3.00 2.66

Dietitian 5 1.10 1.09 4.33 6.00 3.00 4.66 5.00 3.66

COMPROMISES IN CAREER-RELATED DECISIONS 68

Note. R = Realistic; I = Investigative; A = Artistic; S = Social; E = Enterprising; C= Conventional