See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/12532316

Personality and Job Satisfaction: The Mediating Role of Job Characteristics

Article in Journal of Applied Psychology · April 2000

DOI: 10.1037/0021-9010.85.2.237 · Source: PubMed

CITATIONS READS
988 6,266

3 authors, including:

Edwin A. Locke
University of Maryland, College Park
263 PUBLICATIONS 76,376 CITATIONS

SEE PROFILE

All content following this page was uploaded by Edwin A. Locke on 16 September 2014.

The user has requested enhancement of the downloaded file.

Journal of Applied Psychology Copyright 2000 by the American Psychological Association, Inc.
2000, Vol. 85, No. 2, 237-249 0021-9010/00/S5.00 DOI: 10.1037//0021-9010.85.2.237

Personality and Job Satisfaction: The Mediating Role of Job Characteristics

Timothy A. Judge and Joyce E. Bono Edwin A. Locke

University of Iowa University of Maryland

This study tested a model of the relationship between core self-evaluations, intrinsic job characteristics,
and job satisfaction. Core self-evaluations was assumed to be a broad personality concept manifested in 4
specific traits: self-esteem, generalized self-efficacy, locus of control, and low neuroticism. The model
hypothesized that both subjective (perceived) job characteristics and job complexity mediate the rela-
tionship between core self-evaluations and job satisfaction. Two studies were conducted to test the
model. Results from Study 1 supported the hypothesized model but also suggested that alternative models
fit the data well. Results from Study 2 revealed that core self-evaluations measured in childhood and in
early adulthood were linked to job satisfaction measured in middle adulthood. Furthermore, in Study 2
job complexity mediated part of the relationship between both assessments of core self-evaluations and
job satisfaction.

In the decade since Staw, Bell, and Clausen (1986) discovered roticism. According to Judge et al. (1997), these specific traits
a link between childhood personality and job satisfaction later in indicate a single, higher order factor that they argued forms the
life, there has been considerable interest in the relationship be- basis for other, more specific evaluations. In a test of this theory on
tween individual dispositions and job satisfaction. Although this three diverse samples, Judge, Locke, Durham, and Kluger (1998)
literature has had its critics, an accumulating body of research demonstrated that individuals with positive self-evaluations were
suggests that variance in job satisfaction across individuals can be more likely to assess their job satisfaction at higher levels than
traced to measures of affective temperament (House, Shane, & individuals with less positive self-evaluations. Furthermore, Judge
Herald, 1996; Motowidlo, 1996). There is even evidence that the et al. (1998) found that the link between core self-evaluations and
job satisfaction levels of identical twins reared apart are similar, job satisfaction was mediated by perceptions of intrinsic job char-
which suggests a possible genetic basis for job satisfaction (Arvey, acteristics. Drawing from Hackman and Oldham (1980), Judge et
Bouchard, Segal, & Abraham, 1989). More recently, researchers al. (1998) considered intrinsic job characteristics to include five
have begun to explore the psychological processes that might core job dimensions (identity, variety, feedback, autonomy, and
underlie dispositional sources of job satisfaction. For example, significance). Individuals with positive self-evaluations rated their
Weiss and Cropanzano (1996) suggested that affective tempera- work as higher on these core dimensions, and thus were more
ment may influence the experience of emotionally significant satisfied with their jobs.
events at work, which in turn influence job satisfaction. Similarly, Though the Judge et al. (1998) study is a first step toward
both Brief (1998) and Motowidlo (1996) have recently offered elucidating the role of intrinsic job characteristics in the relation-
theoretical models in an attempt to illuminate the relationship ship between personality and job satisfaction, their findings are
between dispositions and job satisfaction. limited in that they used only perceptual measures of work char-
Continuing this theoretical development, Judge, Locke, and acteristics. The job characteristics literature has clearly shown that
Durham (1997) offered a theory linking "core evaluations" of the perceptual measures of intrinsic job characteristics do not perfectly
self to job satisfaction. Judge et al. (1997) defined core self- reflect job complexity (Spector & Jex, 1991). Furthermore, though
evaluations as fundamental assessments that individuals make perceptual measures of job characteristics correlate more highly
about themselves and their self-worth. Incorporated into their with job satisfaction than do objective measures, perceptual mea-
concept of core self-evaluations are four dispositional traits: self- sures have been criticized for their potential contamination by
esteem, generalized self-efficacy, locus of control, and low neu- common method variance (Click, Jenkins, & Gupta, 1986). Thus,
it is not clear from Judge et al.'s findings to what degree positive
self-evaluations are related to increased job complexity as opposed
Timothy A. Judge and Joyce E. Bono, Department of Management and (or in addition) to enhanced perceptions of work characteristics.
Organizations, Henry B. Tippie College of Business, University of Iowa; On the basis of Judge et al.'s research, it is possible that the
Edwin A. Locke, College of Business and Management, University of relationship between core self-evaluations and job characteristics
Maryland.
is purely the result of a perceptual process. That is, individuals
We thank Terry Boles and Mick Mount for comments on an earlier
version of this article.
with positive self-evaluations may see their jobs as more challeng-
Correspondence concerning this article should be addressed to Timothy ing simply because they are predisposed to perceive all aspects of
A. Judge, Department of Management and Organizations, Henry B. Tippie their jobs positively. According to this explanation, there would
College of Business, University of Iowa, Iowa City, Iowa 52242. Elec- not be a link between core self-evaluations and the actual charac-
tronic mail may be sent to [email protected]. teristics of jobs held (i.e., positive individuals do not really have
237
238 JUDGE, BONO, AND LOCKE

jobs that are more challenging, they simply view their jobs as more ence more objectively positive events in their lives, whereas neg-
challenging). Thus, it is critical to understanding the role of core atively disposed individuals actually experience more negative
self-evaluations in job satisfaction to begin to sort out differences events (Magnus, Diener, Fujita, & Pavot, 1993). As Magnusson
in perceptions from differences in actual jobs held. (1990) noted, "An individual's view of himself or herself... with
The purpose of this study was to advance our understanding of respect to self-evaluation (overall approval and acceptance of
the relationship between core self-evaluations, job characteristics, himself or herself), plays a central role in the process of interaction
and job satisfaction in several ways. First, we were interested in with the environment" (p. 201). Thus, it can be argued that
whether core self-evaluations are linked to objective measures of individuals with positive core evaluations would be attracted to
job complexity—do people with positive core evaluations actually challenging jobs because they see the potential for greater intrinsic
hold more complex jobs? Second, we investigated whether core rewards, whereas individuals with a negative self-concept could be
self-evaluations relate to perceptions of job characteristics once the expected to focus on the difficulty and potential for failure of
trait's relationship with objective job characteristics (complexity) challenging work, thus avoiding it.
is controlled. Finally, we sought to replicate Judge et al. (1998) by In addition, Bandura's theory of self-regulation also supports a
using a longitudinal design, which should provide greater confi- link between core self-evaluations and job complexity (although it
dence in the causal nature and temporal stability of the results. In should be noted that Bandura is not a trait theorist). Bandura's
the following section of this article we develop a hypothesized theory predicts that individuals' beliefs about their capabilities to
model of the relationships among core self-evaluations, perceived perform a task will influence their motivation to seek out or avoid
job characteristics, job complexity, and job satisfaction. the task. As Bandura (1997) noted, "People avoid activities and
environments they believe exceed their capabilities, but they
Hypothesized Model readily undertake activities and pick social environments they
judge themselves capable of handling. The higher the perceived
In an attempt to investigate the degree to which job character-
self-efficacy, the more challenging the activities they select"
istics are related to core self-evaluations and to satisfaction, we
(p. 160). Thus, individuals with a positive self-concept should be
hypothesized a structural model including both direct and indirect
more willing to take on enriched jobs because they believe in their
relations of core self-evaluations and job characteristics with job
satisfaction. Figure 1 contains the hypothesized model. With the ability to handle the challenges the job provides. Furthermore, one
exception of the indirect (mediated) relations, each hypothesized might also view the link between core self-evaluations and job
relationship in the model is discussed below. challenge as the process by which individuals with positive self-
concepts gain control over their work environment, as has been
suggested with respect to self-efficacy (Bandura, 1997) and locus
Core Self-Evaluations and Job Complexity of control (Spector, 1982).
We are aware of no previous research that has investigated a There are several other ways in which core self-evaluations may
link between core self-evaluations and job complexity. However, be linked to job complexity. In a test of the impact of self-esteem
there is some theoretical support for such a link. One source of on goal difficulty, Levy and Baumgardner (1991) found that indi-
theoretical support is interactional psychology. As Diener, Larsen, viduals high in self-esteem chose more difficult goals. These
and Emmons (1984) pointed out in developing their interactional findings were consistent with those of Hall and Foster (1977), who
theory, individuals seek out situations on the basis of their perso- found a relationship not only between self-esteem and goals but
nological predispositions. Positively disposed individuals experi- also between self-esteem and task involvement (which was related

Figure 1. Hypothesized model. AH hypothesized linkages are hypothesized to be positive.

 PERSONALITY AND JOB SATISFACTION 239

to subsequent goal setting). Lending further support to the notion environment that would lead to positive conclusions about their
of a relationship between core self-evaluations and job complexity, work, whereas individuals with negative core self-evaluations
Spector (1982) noted that locus of control is related to greater might attend to negative aspects of their jobs. Thus, on the basis of
efforts toward goal achievement and perseverance in the face of previous findings and theoretical support, we expected that core
failure. Because complex jobs implicitly present more challenging self-evaluations would be related to perceived job characteristics.
goals for individuals, these findings are consistent with a hypoth-
esized link between core self-evaluations and job complexity. Job Complexity and Perceived Job Characteristics
Several of the traits that compose self-evaluations have also
been linked to the way individuals cope with complex tasks. For As noted earlier, there is evidence that the relationship between
example, Spector (1982) reported that high levels of anxiety (one perceptions of job characteristics and job complexity is not perfect
of the two primary facets of neuroticism) cause individuals to (Spector & Jex, 1991). However, the existence of a relationship
experience performance deficits on complex tasks but not on between objective and subjective work characteristics is well es-
simple tasks. In addition, both locus of control and self-efficacy tablished. Fried and Ferris (1987), in their meta-analytic investi-
have been shown to affect coping and perseverance in the face of gation, noted, "The data clearly suggest that objective and per-
obstacles (C. R. Anderson, 1977; Bandura, 1997). Furthermore, ceived job characteristics are related" (p. 309). James and Jones
there are other, less direct and more long-range ways that core (1980) found support for their hypothesis that perceptions of job
self-evaluations may be related to more complex jobs. Racket and characteristics are influenced "causally and positively" (p. 125) by
Betz (1981) found self-efficacy to be an important variable in their job complexity. Although James and Jones cautioned the reader
career choice model. Although the focus of their work was ana- not to assume that job perceptions represent veridical descriptions
lyzing the different career choices made by men and women, they of job characteristics, they concluded that objective measures of
found that regardless of gender, self-efficacy within a particular job complexity do influence perceptions of job characteristics.
skill set was correlated with interest and entry into careers that Furthermore, Gerhart (1988) found that perceptions of job char-
required that skill set. Thus, the early socialization experiences and acteristics are a positive function of job complexity. On the basis
educational preparation of individuals high on core self- of these findings, we expected that job complexity would be
evaluations may lead to self-selection into more complex jobs. related to perceived job characteristics.
In summary, there are a number of ways that core self-
evaluations might be expected to be related to job complexity. Perceived Job Characteristics and Job Satisfaction
First, individuals high on core self-evaluations will be more likely
to seek out and attempt complex jobs. Second, they might be The job characteristics model (Hackman & Oldham, 1980)
expected to exert more effort (because of goal-setting activities proposed job satisfaction as one of the essential outcomes resulting
and higher task involvement) and be less likely to withdraw from from intrinsically enriched jobs. According to the job characteris-
complex jobs if they experience failure (because they believe in tics model, intrinsic work characteristics positively affect job sat-
their abilities). Finally, their high coping skills, particularly for isfaction through a perceptual process. Specific job characteristics
complex tasks, may lead to better performance, resulting in in- (e.g., skill variety, task significance) lead to positive psychological
creased attempts to keep and attain more enriched jobs. For these states such as feelings of meaningfulness and responsibility, which
reasons, we hypothesized a positive link between core self- in turn lead to satisfaction with the job. Research clearly supports
evaluations and job complexity. the link between perceived work characteristics and job satisfac-
tion. Two meta-analyses indicated a positive, moderately strong
Core Self-Evaluations and Perceived Job Characteristics correlation between perceptual measures of intrinsic job charac-
teristics and job satisfaction (Fried & Ferris, 1987; Loher, Noe,
There is considerable evidence that perceptions of work char- Moeller, & Fitzgerald, 1985). Thus, consistent with Judge et al.
acteristics are related to dispositions. Research has shown that (1997), we hypothesized that subjective job characteristics would
positively disposed individuals rate characteristics of the task or be positively related to job satisfaction.
the job as more enriched than do less positively disposed individ-
uals (Brief, Butcher, & Roberson, 1995; James & Jones, 1980; Job Complexity and Job Satisfaction
Judge et al., 1998; Kraiger, Billings, & Isen, 1989; Necowitz &
Roznowski, 1994). Conceptually, several basic areas of research Although the process of translating job characteristics into job
support the link between personality and perceptions of work satisfaction is a perceptual one, the job characteristics model
characteristics. Research shows that individuals prone to the ex- assumes that these perceptions originate from job conditions. If
perience of positive emotions respond favorably to situations de- this assumption is valid, there should be a positive association
signed to induce positive affect, whereas individuals predisposed between objective measures (i.e., nonperceptual) of work charac-
to experience negative emotions and negative self-appraisals are teristics and job satisfaction. However, there have been few tests of
less likely to respond positively to such situations (Larsen & the relationship between job complexity and job satisfaction. Al-
Ketelaar, 1991). Similarly, self-verification theory (Swann, Stein- though some studies found only a small relationship between
Seroussi, & Giesler, 1992) indicates that individuals seek to verify complexity and satisfaction (Spector & Jex, 1991), Fried and
their self-concepts by selecting situations that will supply them Ferris's (1987) review suggested a consistently positive relation-
with feedback that reinforces that self-concept. In the job context, ship between objective measures of job characteristics and job
Judge et al. (1997) noted that individuals with positive core eval- satisfaction. Although job complexity explained less of the vari-
uations might seek out and categorize information in their work ance in job satisfaction than did subjective measures of job char-
240 JUDGE, BONO, AND LOCKE

acteristics, the relationship was both positive and significant across We received 424 completed surveys from respondents (39 surveys were
all studies Fried and Ferris reviewed. On the basis of these small returned as undeliverable), representing a 22% response rate. We received
but consistently positive relationships, we hypothesized a direct 389 significant-other surveys, which indicated that for 92% of the respon-
positive relationship between job complexity and job satisfaction. dents a significant-other survey was also returned. Removal of surveys
with missing data resulted in a total of 384 usable responses with 351
We note that it is possible that the relationship between job
matching significant-other reports. Sixty-five percent of significant-other
complexity and job satisfaction is entirely mediated through sub-
surveys were completed by spouses, with the remaining 35% completed by
jective perceptions of job characteristics. Although this explana- siblings (2%), parents (3%), friends (21%), and others (9%). No mean
tion seems reasonable, no previous research has investigated this difference on any study variable was observed between those respondents
relationship. Thus, we hypothesized a direct relationship between who had a significant-other survey returned and those who did not.
job complexity and job satisfaction, although we did recognize and The mean age of respondents was 41 years. On average, respondents
investigate the possibility that this relationship is mediated by worked 46 hr per week and had held 1.9 jobs over the past 5 years. Mean
perceived job characteristics. annual salary for respondents was $40,940. Twenty-four percent of respon-
dents had a high school diploma or less, 55% had some college or a
bachelor's degree, and 21% had some graduate credit or a graduate degree.
Core Self-Evaluations and Job Satisfaction Sixty-four percent of respondents were male and 66% were married. When
we compared our sample information to U.S. Census data for the city
Consistent with the previous hypotheses, an indirect link from
surveyed, our sample seemed reasonably representative of the working
self-evaluations to job satisfaction (mediated by job characteris-
population in terms of age, area of residence (measured by zip code),
tics) would be expected. However, there is also evidence of a education level, marital status, and hours worked per week. Members of
direct relationship between these concepts. Judge et al. (1998) our sample did earn somewhat higher salaries than indicated by the census
hypothesized and found a direct link between core self-evaluations data. It is important to note that although our inclusion criteria limited our
and job satisfaction. In fact, in most cases the direct relationship initial mailing to households with incomes over $20,000, 24% of the
was larger than the mediated relationship. A direct link between individuals who responded earned incomes of $20,000 or less. Nonethe-
core self-evaluations and job satisfaction, within the variables less, it is possible that our sampling procedure produced a somewhat
measured in this model, is consistent with the direct effects model unrepresentative sample, at least with respect to some characteristics.
discussed by Judge et al. (1997). In that model, core self-
evaluations influence job satisfaction through a process of emo- Measures
tional generalization—individuals' positive feelings about them- Self-esteem. The first core evaluations trait discussed by Judge et al.
selves spill over onto their jobs. To be sure, these direct effects (1997) is self-esteem. Self-esteem was measured using Rosenberg's (1965)
may not be purely direct; they may be mediated by cognitive 10-item self-esteem scale, which includes items such as "I feel that I am a
processes or other mediating variables. However, because this person of worth, at least on an equal basis with others" and "At times I
study does not focus on, and thus does not include, cognitive think I am no good at all" (reverse scored). Scores for individual items,
processes or other factors that may mediate the relationship be- which ranged from 1 (strongly disagree) to 5 (strongly agree), were
tween core self-evaluations and job satisfaction, we hypothesized summed to produce a single self-esteem score for each respondent. Sig-
a direct relationship between core self-evaluations and job nificant others completed these same 10 self-esteem items.
Generalized self-efficacy. As the second core evaluations trait dis-
satisfaction.
cussed by Judge et al. (1997), generalized self-efficacy was measured using
In an effort to investigate the robustness of the hypothesized
seven items from a scale developed by Judge et al. (1998). Respondents
model, we conducted two separate studies. Because the data col- were asked to use a 1 (strongly disagree) to 5 (strongly agree) scale to
lection procedures for each study were quite different, we present indicate their level of agreement with statements such as "I am strong
separate method and results sections for each study. enough to overcome life's struggles" and "I often feel that there is nothing
that I can do well" (reverse scored). Each individual's scores on the seven
Study 1 items were summed to form a single generalized self-efficacy score.
Significant others completed these same seven generalized self-efficacy
items.
Method
Locus of control. To measure the locus-of-control component of core
Data and Procedure self-evaluations, four items measuring internal locus of control were taken
from the Internality, Powerful Others, and Chance Scale (Levenson, 1981).
Participants in Study 1 were randomly selected from all zip codes of a Individuals were asked to indicate their agreement with statements regard-
midsized midwestem city. In an attempt to minimize the number of surveys ing the extent to which they have control over events in their lives, such as
sent to households without working adults, we purchased a mailing list that "When I get what I want, it's usually because I am lucky" (reverse scored)
was limited to individuals between the ages of 24 and 58 (excluding most and "My life is determined by my own actions." As with the other core
college students and retirees), and with a household income of $20,000 or self-evaluations measures, the scale ranged from 1 (strongly disagree) to 5
more (so as to exclude those working part time). Surveys were mailed (strongly agree), with higher scores representing an internal locus of
to 1,981 men and women along with a cover letter assuring participants that control. Scores for each item were summed to produce a single locus-of-
individual responses were confidential. Included in the mailing was a control score for each respondent. For the significant-other survey, these
second survey to be completed by a significant other. Significant others same four items were used.
were instructed to complete their surveys away from the focal person and Neuroticism. To measure the final component of core self-evaluations,
to return it directly to the researchers in a separate postage-paid envelope neuroticism, we used the 12-item Eysenck Personality Inventory (Eysenck
that was included with the questionnaire. Questionnaires were numbered so & Eysenck, 1968) Neuroticism scale. Individuals were asked to indicate
that significant-other responses could be matched with those of respon- their agreement with statements concerning the degree to which they
dents. In return for their participation, respondents were offered the op- experience feelings of irritability, nervousness, worry, embarrassment, or
portunity to enter their names into a drawing for $200. guilt, such as "I am a nervous person" and "I am a worrier." The same 1-5
 PERSONALITY AND JOB SATISFACTION 241

response scale used with the other core self-evaluation traits was used for response formats, they were standardized and combined into a three-item
the Neuroticism scale. Individual items were combined to form a single composite measure of job satisfaction.
neuroticism score for each respondent. Significant others completed the
same 12 neuroticism items.
Perceived job characteristics. Perceptions of work characteristics (i.e.,
Covariance Structure Analysis
autonomy, feedback, task variety, identity, and significance) were mea- Covariance structure analysis, which was estimated in the present study
sured using a 14-item version of the Job Diagnostic Survey (JDS; Hackman by using LISREL 8 (Joreskog & Sorbom, 1993), was used to test the
& Oldham, 1980). Respondents were asked to use a scale ranging from 1 structure of the core self-evaluations concept as well as the hypothesized
(very inaccurate) to 7 (very accurate) to assess the accuracy of statements model. To prevent the interpretational problems inherent in simultaneous
such as "The job is quite simple and repetitive" (reverse scored) and "The estimation of measurement and structural models (J. C. Anderson &
job gives me considerable opportunity for independence and freedom in Gerbing, 1988), we used confirmatory factor analysis to test the measure-
how I do the work." The JDS was designed to be a multidimensional scale; ment model separately prior to simultaneous estimation of the measure-
however, job characteristics measures have been shown to collapse to a ment and the structural models. The first structural model tested (the
single dimension (Dunham, Aldag, & Brief, 1977). Dunham (1976, p. 408) self-report model) is based on self-reports of all study variables, with the
found that the "most parsimonious factorial solution was a single-factor exception of job complexity. The second model (the mixed data model) is
solution representing job complexity." Also, Loher et al.'s (1985) meta- the same as the self-report model, except that significant-other reports of
analytic results support the unidimensional nature of job characteristics core evaluations were used. The advantage of the mixed data model is that
measures. Therefore, the 14 individual item responses were summed to it allows inferences about the relations among the concepts to be mostly
form a single perceived job characteristics score for each respondent. free of percept-percept inflation. The relationship between perceived job
Job complexity. Respondents were asked to report the title of their characteristics and job satisfaction is susceptible to this inflation; however,
current job. These job titles were assigned a three-digit occupational code we considered it impractical to measure these two concepts by any other
according to the 1970 U.S. Census occupational coding scheme. That means.
census code was converted to a complexity score on the basis of complex- We specified the structural model by allowing the four core self-
ity scores derived by Roos and Treiman (1980) from the fourth edition of evaluation traits to load on a latent factor and the two job satisfaction
the Dictionary of Occupational Titles (DOT). Roos and Treiman submit- measures to load on another latent factor (to allow this factor to be
ted 46 items from the DOT to exploratory factor analysis and identified a identified, one of the loadings was fixed at 1.0). Because perceived job
factor they labeled Substantive Complexity. characteristics was measured with a multi-item scale, correction for mea-
There are two primary (non-incumbent-based) methods that have been surement error was based on its estimated reliability. The reliability of the
used to assess job complexity. The first is job analysis, a procedure that Z>0r-based measure of job complexity has been extensively analyzed; we
involves the direct observation of workers and the activities they perform. used Cain and Green's (1983) average estimate of interrater reliability (.69)
Although job analysis is the more thorough and complete non-incumbent- in correcting job complexity for measurement error. In the covariance
based method of rating job complexity, it is not feasible in studies that structure model, the measurement error for all directly observed variables
attempt to assess job complexity across multiple jobs in many organiza- (in Study 1, perceived job characteristics and job complexity; in Study 2,
tions. Therefore, consistent with Gerhart (1987), Adelmann (1987), and all three variables in the model) was fixed as the variance of the variable
Spector and Jex (1991), we used the second method, the DOr-based coding times one minus the reliability of the variable.
for job complexity. One clear advantage to this particular measure of job When interpreting the results of covariance structure analysis, it is
characteristics is the fact that it is based not on reports of workers or their important to evaluate the model by using several indices of overall fit.
colleagues but on independent assessments of job descriptions and on-site Accordingly, we report the following fit statistics: chi-square, root mean
job observations. However, it should be noted that DOT codes are based on square error of approximation (RSEA), goodness-of-fit index (GFI),
occupations rather than on jobs. For example, the DOT code for police normed fit index (NFI), non-normed fit index (NNFI), incremental fit index
officer represents the average complexity of the job of a police officer (IFI), and relative fit index (RFI; Medsker, Williams, & Holahan, 1994).
across police departments, rather than the job complexity of the specific Although levels of the chi-square statistic cannot be interpreted indepen-
police officer who responded to our survey. However, reliability estimates dently of the sample size, rules of thumb suggest that the RSEA should be
for DOT codes are calculated to reflect both interrater agreement and error no greater than .10, whereas values of GFI, NFI, NNFI, IFI, and RFI should
introduced by aggregating information about several jobs to the occupa- be greater than .90 (Medsker et al., 1994). Finally, we also report two fit
tional level. Results have suggested that there is less variation in complex- statistics that correct for lack of parsimony, the adjusted goodness-of-fit
ity across different jobs in the same occupation than across occupations in index (AGFI) and the parsimony normed fit index (PNFI). Rules of thumb
the same organization, supporting the validity of the DOT measurement for judging the latter statistics have not been established.
system. Psychometric information on the DOT measure can be found in
Cain and Green (1983) and Gerhart (1985).
Results
Job satisfaction. We obtained two assessments of job satisfaction.
Overall job satisfaction was measured with five items taken from the Descriptive Statistics, Scale Reliabilities,
Brayfield-Rothe (1951) measure of job satisfaction. These five items were
and Intercorrelations
"I feel fairly satisfied with my present job," "Most days I am enthusiastic
about my work," "Each day at work seems like it will never end" (reverse Table 1 presents descriptive statistics, scale reliabilities, and
scored), "I find real enjoyment in my work," and "I consider my job to be
intercorrelations of study variables. To permit examination of
rather unpleasant" (reverse scored). Responses to the Brayfield-Rothe
unconnected relations among the concepts in the model, also in-
items were evaluated on a 1 (strongly disagree) to 7 (strongly agree) scale.
In addition, global job satisfaction was also measured using the three-item
cluded in Table 1 are two composite core self-evaluations indices
scale developed by Judge, Boudreau, and Bretz (1994), which includes the (one for self-reports and one for significant-other reports) that
nongraphic version of the G. M. Faces Scale (Scarpello & Campbell, 1983, were equally weighted combinations of the four core traits and a
replaced faces with affective state descriptions), an adapted version of the similar composite index for the two individual measures of job
Fordyce Percent Time Happy Item, and the Gallup Poll measure of job satisfaction. As was the case with Judge et al. (1998), locus of
satisfaction. Because these three items were measured using different control displayed the lowest correlations with the other core traits
242 JUDGE, BONO, AND LOCKE

Table 1
Descriptive Statistics, Scale Reliabilities, and Intercorrelations of Study 1 Variables

Variable M SD 1 10 11 12 13 14 15

1. Self-esteem 41.52 6.29 87

2. Generalized self-efficacy 29.49 4.21 77 86
3. Locus of control 14.95 2.51 46 47 70
4. Neuroticism 27.14 8.59 -65 -60 -29 90
5. Core self-evaluations composite 0.00 3.27 89 88 69 -79 84
6. Self-esteem, SOR 41.35 6.06 44 41 26 -41 47 86
7. Self-efficacy, SOR 29.41 4.39 41 40 26 -38 86
45 80
8. Locus of control, SOR 15.51 2.43 16 15 29 -12 22 35 41 70
9. Neuroticism, SOR 26.17 8.38 -33 -28 -12 50 -38 -62 -58 -22 90
10. Core self-evaluations composite, SOR 0.00 3.15 43 39 29 -44 48 88 88 63 -77 80
11. Perceived job characteristics 70.77 11.81 36 36 29 -27 39 26 21 13 -13 23 73
12. Job complexity 5.13 2.11 20 16 13 -16 20 24 18 -02 -19 19 23 69"
13. Job satisfaction (Brayfield & Rothe, 25.88 6.89 40 31 27 -27 39 22 17 12 -07 18 59 16 89
1951)
14. Job satisfaction (Judge et al., 1994) 0.00 2.67 43 32 27 -28 40 24 20 11 -11 21 59 16 84 86
15. Job satisfaction composite 0.00 1.92 43 33 28 -29 41 24 19 12 -09 19 62 17 96 96 91

Note. Coefficient alpha reliability estimates are on the diagonal. Decimals are omitted from correlations and reliabilities. N = 348. SOR = significant-
other reports, p £ .05 at r = .11. p < .01 at r = .14.
a
From Cain and Green (1983).

and with the other variables. Correlations between self-reports and trol displayed lower factor loadings than the other core traits, and
significant-other reports of core evaluations were moderate (rang- locus of control contributed less to the core concept. Thus, it might
ing from .29 for locus of control to .50 for neuroticism). Although be asked whether locus of control is necessary to form the core
it is not reported in Table 1, it is worth noting that self-other self-evaluations concept. As suggested by a reviewer, to test this
correlations were only slightly (r = .02) higher when the spouse possibility we constrained the second-order loading involving lo-
was the significant other. Overall, correlations between study cus of control to zero. If locus of control contributes little to the
variables were in the direction expected. core concept, the fit of the model will not be reduced. However,
estimation of this model revealed that constraining the locus-of-
Confirmatory Factor Analysis of Core Traits control second-order loading to zero significantly reduced the fit of
the second-order model, Ax^l, N = 414) = 107.62, p < .01.
To analyze the factor structure of the data, we conducted a Constraining the second-order loadings of the other core traits to
second-order factor analysis. (Because similar results were ob- zero reduced the fit of the model to an even greater degree. Thus,
tained for the significant-other reports of the core traits, for sim- within the confines of this measurement model, it appears that all
plicity we confine our discussion to self-reports of the core traits.) four traits are important elements of the core concept, though
In the first stage of the analysis, we created three parcels for each future research should investigate the adequacy of locus of control
of the core traits. These parcels were formed from the individual in identifying the core self-evaluations concept, which may also
scale items (e.g., two of the three self-esteem parcels were formed require looking at the adequacy of locus-of-control measures.
by adding three randomly selected items from the self-esteem
scale, and the third parcel was formed by adding the remaining
four items). When each of these parcel sets was constrained to load Structural Model Results
on its respective traits, results indicated that this model fit the data
reasonably well. The fit statistics were as follows: ^(48, N = LISREL estimates for the self-report model, which relates self-
414) = 123.08 (p < .01), RSEA = .06, GFI = .95, AGFI = .92, reports of the core evaluations to perceived job characteristics and
NFI = .96, NNFI = .96, IFI = .97, RFI = .94, PNFI = .70. In the job complexity and to job satisfaction, are provided in the upper
second stage of this analysis, following Joreskog and Sorbom row of Figure 3. Results show that core self-evaluations had a
(1989, p. 160), we conducted a second-order factor analysis in moderately strong and significant relationship both with percep-
which the four core traits contributed to an overall core self- tions of job characteristics and with job complexity. Core self-
evaluations factor. This second-order model appeared to fit the evaluations also displayed a direct relation with job satisfaction.
data acceptably, ^(54, N = 414) = 159.00 (p < .01), RSEA = Results also show that job complexity was related to perceived job
.07, GFI = .94, AGFI = .91, NFI = .95, NNFI = .96, IFI = .96, characteristics. Finally, perceptions of job characteristics and job
RFI = .94, PNFI = .77. The factor loadings from the first- and satisfaction were strongly related, whereas job complexity had no
second-order factor analyses from the self-report model are dis- direct relationship with job satisfaction. However, it is important to
played in Figure 2. note that job complexity had a significant indirect relationship to
As the second-order factor analysis results show, self-esteem job satisfaction (r = .13, p < .05), as mediated through perceived
and self-efficacy were nearly perfectly correlated with the core job characteristics. Fit statistics for the self-report model were as
self-evaluations factor. Furthermore, the parcels for locus of con- follows: x*(l6, N = 384) = 26.41 (p = .05), RSEA = .04, GFI =
 PERSONALITY AND JOB SATISFACTION 243

Self-Esteem #1

Self-Esteem #2

Self-Esteem #3

Self-Efficacy #1

Self-Efficacy #2

Self-Efficacy #3

Neuroticism #1

Neuroticism #2

Neuroticism #3

Locus of Control #1

Locus of Control #2

Locus of Control #3

Figure 2. Second-order factor analysis results—self-report data. First-order factor loadings were estimated
from first-order factor analysis. * p < .01.

.98, AGFI = .96, NFI = .98, NNFI = .99, IFI = .99, RFI = .97, data model were comparable to those of the serf-report model,
PNFI = .56. ^(16, N = 351) = 21.95 (ns), RSEA = .03, GFI = .98, AGFI =
LISREL estimates for the mixed data model, which relates .96, NFI = .98, NNFI = .99, IFI = 1.00, RFI = .97, PNFI = .56.
significant-other reports of the core evaluations to self-reports of The second and third columns of Table 2 contain the direct,
job characteristics, job complexity, and job satisfaction, are pro- indirect, and total (direct + indirect) relationships of core self-
vided in the lower row of Figure 3. Results show that core evaluations with job satisfaction for the self-report and mixed data
evaluations (as reported by significant others) had a significant models, respectively. The results show that most of the significant
relationship to job complexity. However, the relationship between relationship between core self-evaluations and job satisfaction was
core evaluations and perceptions of job characteristics was weaker mediated by perceived job characteristics and job complexity.
in this model, leading to the inference that some of the relationship Given the presence of significant indirect relationships and the fact
between core evaluations and perceptions of work characteristics that more than half of the total relationship is mediated, we can
in the self-report model may be based on common method vari- infer from these results that job characteristics mediated the rela-
ance. Another difference from the self-report model was that core tionship between core self-evaluations and job satisfaction.
self-evaluations no longer had a significant direct relationship with
job satisfaction. Results also show that job complexity had a Alternative Model Tests
relationship, in this model, with perceptions of job characteristics
similar to that found in the self-report model. Finally, consistent To investigate the validity of the hypothesized model, we con-
with the self-report model, perceptions of work characteristics and sidered three alternative models. One alternative model stipulated
job satisfaction were strongly related, whereas job complexity had no direct relationship between core self-evaluations and job satis-
no direct relationship with job satisfaction. However, as in the faction. This model is plausible if the relationship between core
self-report model, job complexity had a significant indirect rela- self-evaluations and job satisfaction is entirely mediated by job
tionship with job satisfaction (r = .19, p < .05) as mediated complexity and perceived job characteristics. A second alternative
through perceived job characteristics. Fit statistics for the mixed model stipulated no direct relationship between job complexity
244 JUDGE, BONO, AND LOCKE

Perceived Job
Characteristics
.63*
i N.'6*
.24*7
.20* _^—-^
.24* X^
Core Self- ( Job
Evaluations H Satisfaction
.04 ^—_,^__-^'

.26*\
.28*
Job
L /
Complexity

Figure 3. LISREL results—Study 1. Estimates in top row represent results from self-report model; estimates
in bottom row represent results from mixed data model. * p < .01.

and job satisfaction. This model is also a plausible alternative the model that drops the link between job complexity and job
because it is quite possible that the effects of job complexity are satisfaction, results indicated that dropping the link did not de-
completely mediated through perceptions of those characteristics. crease the fit of the model for either the self-report, Ay2(l, N =
Because these more parsimonious models are nested within the 384) = 0.20, ns, or mixed data, A^l, N = 351) = 0.61, ns,
hypothesized model, the difference in chi-square between these models. Thus, it appears that two of the direct links in the hypoth-
models can be tested for significance. esized model could be eliminated (thus simplifying the model)
The previous two alternative models entailed removing paths to without causing it great damage.
determine whether a more parsimonious model was possible. A
third alternative model would posit a different order of relation- Study 2
ships, such that job satisfaction mediates the relationship between
core self-evaluations and perceived job characteristics (as opposed Study 2 was conducted in an attempt to examine the extent to
to the hypothesized model, which hypothesized that perceived job which the relationships between core self-evaluations, job charac-
characteristics mediated the relationship between core self- teristics, and job satisfaction are robust over time. This study uses
evaluations and job satisfaction). This model is justified in light of longitudinal data collected over a period of 30 years to test the
research that suggests that the relationship between perceived job hypothesized model.
characteristics and job satisfaction may be bidirectional or from
job satisfaction to perceived job characteristics (James & Jones, Method
1980; James & Tetrick, 1986). This third model would reverse the
Data and Procedure
relationship between perceived job characteristics and job satis-
faction, so that job satisfaction influences (rather than is influenced The data for this study were obtained from the Intergenerational Studies
by) perceived job characteristics. Although this model is a reason- (IGS), administered by the Institute of Human Development, University of
able alternative to the hypothesized model, it would fit the data
exactly the same as the hypothesized model because it would Table 2
simply reverse the link between perceived job characteristics and Direct, Indirect, and Total Relationships Between Core Self-
job satisfaction. Thus, it was not tested in this study but is con- Evaluations and Job Satisfaction
sidered in the Discussion section.
Study 1 Study 2
In the model that drops the link between core self-evaluations
and job satisfaction, results indicate that dropping the link signif- Self-report Mixed data Childhood Adulthood
icantly decreased the fit of the model for the self-report model, Relationship model model model model
Ax^l, N = 384) = 15.38, p < .01, but not for the mixed data 22**
Direct .04 .20 .34**
model, A^2(l, N = 351) = 0.48, ns. Even though the A^2 statistic Indirect .28** .22** .16** .11**
was significant for the self-report model, the other standardized fit Total .50** .26** .36** .45**
statistics showed little difference (the mean difference was .015). Proportion of relationship
Furthermore, the fit statistics that take parsimony into account, mediated .56 .85 .44 .24
AGFI and PNFI, were actually higher for the alternative model Note. Proportion of relationship mediated was calculated by dividing the
(which suggests that the decrease in fit because of the added indirect relationship by the total relationship.
constraint was more than offset by the increase in parsimony). For **p < .01.
 PERSONALITY AND JOB SATISFACTION 245

California at Berkeley. The IGS were a combination of three studies scale include "Is self-defeating" (reverse scored), "Is satisfied with self,"
conducted at the Institute, beginning in the early 1920s. Because, for the and "Is self-pitying" (reverse scored). Responses to these 8 items were
most part, the same measures were collected in the three studies, they were summed to form a measure of core self-evaluations for each time period.
combined in the analyses. The participants included in our study were The reliability of this scale was a = .72 for children ages 13 and 16, and
predominantly White and were roughly evenly divided between males and a = .74 for the adult assessment. Childhood core self-evaluations were
females. The average participant had at least some college education, and assessed as the average of the age 13 and age 16 assessments, whereas
approximately 60% were bom into middle-class homes. Finally, 85% of adulthood core self-evaluations were assessed with the single (ages 30-38)
the participants were married and had an average of 2.7 children. (For a early adulthood assessment.
more complete description of the IGS procedures and measures, see Block, Job satisfaction. Overall job satisfaction was measured when partici-
1971, and Clausen, 1993.) pants were 41-50 years old, with an eight-item scale that asked participants
The IGS participants' personality was assessed twice during the partic- to report their satisfaction with various facets of their job (e.g., income,
ipants' childhood—when children were approximately 13 years old and supervision, job security, coworkers). Interviewers coded responses on a 1
again when participants were 16 years old. Participants' personality was (dislike it very much) to 5 (like it very much) scale. Responses to these eight
also assessed in adulthood—to preserve the temporal ordering of the items were summed to form a measure of overall job satisfaction. The
variables, when adulthood measures of personality were used in the anal- reliability of this scale was a = .92.
yses, we used only the early adulthood personality measures, which were Job complexity. At the same time participants' job satisfaction was
collected when the participants were 30-38 years old. During the course of measured, job complexity was measured by matching the DOT rating of
the IGS, participants' job satisfaction was assessed with a multi-item job titles to the participants' jobs. The DOT evaluates job complexity in
measure once, when participants were 41-50 years old. (Because of the terms of complexity in dealing with people (rated on a 0-8 scale anchored
intensive nature of the data collection, the adult interviews took place over by mentoring [0] and taking instructions [8]), data (rated on a 0-6 scale
approximately 8-year increments.) anchored by synthesizing [0] and coordinating [6]), and things (rated on a
There were 192 individuals who had complete personality data (age 13, 0-7 scale anchored by setting up [0] and handling [7]). These three facets
age 16, and ages 30-38). Of the individuals who had complete personality were summed (a = .72) to form an overall measure of job complexity.
data, 107 also had complete job satisfaction data. Of the individuals whose After summing the facets, the scale was reverse scored to make high scores
personality was assessed when they were 30-38 years old, 151 individuals indicate more complex jobs.
also had complete job satisfaction data. Post hoc analyses indicated that the
core self-evaluations for those who had incomplete data were not different Results
from the evaluations for those who had complete data.
Table 3 presents descriptive statistics, scale reliabilities, and
Measures intercorrelations of study variables. Correlations between study
variables were consistent with Study 1 results. Specifically, core
Core self-evaluations. On the basis of the interview transcripts and
self-evaluations, measured in both childhood and early adulthood,
archives for each participant, expert psychologists trained in personality
assessment were asked to sort 104 personality descriptors into nine cate- were significantly correlated with job complexity and job satisfac-
gories, ranging from most descriptive to least descriptive of the partici- tion. Job complexity was also significantly correlated with job
pants' personality. These categories were then numbered, from 9 being the satisfaction, at a level higher than that in Study 1. Results also
most descriptive to 1 being the least descriptive. To eliminate the possi- revealed relatively strong correlations between the childhood as-
bility that assessors' subsequent personality ratings of a participant were sessments of core self-evaluations and the early adulthood assess-
influenced by earlier ratings, no assessor evaluated the same participant ment. From the time individuals were teenagers to when they were
over more than one time period. Because the multiple assessors rated each in their 30s, the correlation between their core self-evaluations was
participant according to the Q set items, it is possible to estimate interrater .46. When corrected for unreliability, this correlation rose to .62.
reliability of the ratings. Across all assessments, the average reliabilities
ranged from .72 to .78. Thus, the assessors were reliable in their assessment
of participants' personality. Participants' scores on each item were com- Structural Model Results
puted as the average score across the assessors.
We tested two models in this study. One model related the
The 104 items measured many aspects of participants' personality. Of
these items, 8 met the requirements of core self-evaluations measures: (a) childhood assessment of core self-evaluations to job complexity
self-focused, (b) evaluative, and (c) carried a positive or negative affective and job satisfaction; the other model related the early adult assess-
connotation. For example, descriptions such as "Is hostile," "Is sociable," ment (ages 30-38) to these concepts. Covariance structure analy-
and "Is productive; gets things done" were excluded because they did not sis procedures in this study were consistent with those in Study 1.
meet one or more of the evaluative criteria. Sample items from the 8-item LISREL estimates for the hypothesized models are provided in

Table 3
Descriptive Statistics, Scale Reliabilities, and Intercorrelations of Study 2 Variables

Variable M SD 1
1. Core self-evaluations: childhood 0.00 1.76 74
2. Core self-evaluations: ages 30-38 0.00 1.00 46 74
3. Job complexity 5.21 2.35 22 19 72
4. Job satisfaction 27.96 5.71 27 43 41 92

Note. Decimals are omitted from correlations and reliability coefficients. Coefficient alpha reliability estimates
are on the diagonal. All correlations are significant at p < .05. Listwise N = 107.
246 JUDGE, BONO, AND LOCKE

Figure 4. Results indicate that core self-evaluations had a signif- Study 1, it appears that a direct link between core self-evaluations
icant relationship with job complexity in both models. However, and job satisfaction is not required. However, unlike Study 1,
core self-evaluations had a significant direct relationship with job Study 2 results indicated that there must be a direct link between
satisfaction only in the early adult model (in which core evalua- job complexity and job satisfaction for the model fit to be ade-
tions were assessed when participants were ages 30-38). Results quate. This was probably due to the fact that perceived job char-
also showed that job complexity was significantly related to job acteristics were not measured in Study 2.
satisfaction in both models. The fit statistics for the childhood
model were as follows: ^(3, N = 107) = 2.70 (ns), RSEA = .01, Discussion
GFI = .99, AGFI = .96, NFI = .97, NNFI = 1.00, IFI = 1.00,
RFI = .93, PNFI = .48. For the adult model, the fit statistics were The primary contribution of these studies was to reveal that job
as follows: )?(2, N = 151) = 0.11, ns, RSEA = .01, GFI = 1.00, complexity—the actual attainment of challenging jobs—was an
AGFI = 1.00, NFI = 1.00, NNFI = 1.00, IFI = 1.00, RFI = 1.00, important explanatory variable in the relationship between core
PNFI = .67. self-evaluations and job satisfaction, and to show that the relation-
The last two columns of Table 2 contain the direct, indirect, and ship between core self-evaluations and job satisfaction persisted
total (direct + indirect) relations of the core self-evaluations over time. Brief (1998) discussed two different models of job
concept to job satisfaction for the childhood and adulthood models. satisfaction: top-down, in which satisfaction is derived from how
As with Study 1, results indicated a significant indirect relationship one interprets one's environment, and bottom-up, in which satis-
between core self-evaluations and job satisfaction. Also similar to faction is derived from the experience of more positive job con-
Study 1, results were somewhat inconsistent regarding the direct ditions. Previous research has supported the top-down model
relationship. There was also some inconsistency in the percentage (Judge et al., 1998). At the same time, results of the present studies
of the relationship that was mediated by job characteristics. In appear to support the bottom-up model. Regardless of which
total, results indicated that at least part of the relationship between alternative model is adopted, core self-evaluations had a signifi-
core self-evaluations and job satisfaction was mediated by job cant total relationship with job satisfaction in all four models
characteristics, though the exact magnitude of the mediation is not tested, and similarly, core self-evaluations had a direct relationship
clear. with job complexity. Because job complexity, core self-
evaluations, and job satisfaction were measured with independent
methods and, in the case of Study 2, core self-evaluations were
Alternative Model Tests
measured before job satisfaction (in the case of the childhood
In Study 2, we investigated two of the same alternative models model, 30 years before job complexity and job satisfaction), con-
as in Study 1 (with no direct relationship between core self- fidence can be placed in the results.
evaluations and job satisfaction and no direct relationship between Results from Study 2 indicated that core self-evaluations were
job complexity and job satisfaction). For the model that dropped related to job satisfaction over time. As is to be expected, the
the link between core self-evaluations and job satisfaction, results relationship was stronger when core evaluations were measured in
indicated that dropping the link significantly decreased the fit of adulthood, but it is impressive that independent childhood assess-
the adulthood model, A^2(l, N = 151) = 13.69, p < .01, but not ments correlated with job satisfaction 30 years later. Only one
the childhood model, A^2(l, N = 107) = 2.82, ns. For the model previous study (Staw et al., 1986) has related IGS data to job
that dropped the link between job complexity and job satisfaction, satisfaction, and similar results were found with respect to Staw et
results indicated that dropping this link decreased the fit of both al.'s measure of affective disposition, which they acknowledged to
the childhood and adulthood models, A^(l, N = 107) = 12.66, have some conceptual ambiguities. However, Staw et al. did not
p < .01, and A^(l, N = 151) = 18.81, p < .01, respectively. The link personality to job complexity.
parsimony fit statistics increased for the alternative models that Results from Study 2 also provided insight into the stability of
dropped the core self-evaluations link but decreased in the alter- the core self-evaluations concept. Research on the Big Five per-
native models that dropped the job complexity link. Thus, as in sonality traits suggests that the average correlation between the

Figure 4. LISREL results—Study 2. Estimates in top row represent results using childhood personality ratings;
estimates in bottom row represent results using early adult (ages 30-38) personality ratings. * p < .01.
 PERSONALITY AND JOB SATISFACTION 247

traits is .58 over roughly the same time period (20 years) as that in other hand, significant-other reports may be less accurate because
Study 2 (Costa & McCrae, 1994). The stability of core self- significant others have no opportunity to directly observe the type
evaluations in Study 2 was somewhat lower: .46 (.62 when cor- of internal trait we were measuring with self-evaluations.
rected for unreliability). The slightly lower stability may be due to Use of peer reports is a common practice in personality research.
the fact that the time interval included an individual's formative As noted by Funder and Colvin (1997), one of the reasons re-
years (whereas Costa & McCrae's, 1994, data only considered searchers study self-other agreement is to investigate the degree to
stability in adult personality), or it may be due to the fact that some which self-enhancement or self-serving biases affect self-reports
traits may be somewhat less stable than others (House et al., 1996). of personality. Although this area continues to be investigated,
Thus, although the stability of core self-evaluations is moderately evidence does suggest that some degree of self-enhancement oc-
high, it is not so high to suggest that it is immutable to change. We curs (Funder & Colvin, 1997). For this reason, we assessed core
are aware of no previous research on the stability of the core self-evaluations with both self-reports and other reports, though
self-evaluations concept, so future research should explore this the latter certainly have limitations of their own. As Costa and
issue further. McCrae (1992) noted, the joint inclusion of self-reports and other
Some degree of confidence can be placed in the interpretation of reports provides valuable information, particularly when the
our results because we used independent sources of data to elim- sources produce mostly equivalent results. In the case of Study 1,
inate response-response bias and a longitudinal design to support there are some differences in the results of the self- and other-
the assumed causal ordering of the variables. However, with report models.
respect to several important aspects of the model, causal inferences It is possible that the best way to assess another person's core
cannot be drawn. Specifically, it is just as likely that the hypoth- self-evaluations would be a clinical interview, and this is what we
esized link from perceived job characteristics to job satisfaction is did in Study 2. It is interesting to note that, across the two studies,
also (or instead) in the opposite direction, from job satisfaction to all three sources of data (self-, significant-other, and clinical rat-
perceived job characteristics. Ideally, we would have tested a ings) produced generally similar results—in all cases, core self-
nonrecursive model that tested a reciprocal relationship between evaluations were significantly related to both job satisfaction and
the two concepts, something that has been done in past research job complexity. Because the sources are not purely equivalent, we
(James & Jones, 1980; James & Tetrick, 1986). Unfortunately, think future research should continue to use multiple sources of
such tests require a number of instrumental variables that uniquely core self-evaluations. However, it would also be useful for future
influence each concept; it was not possible to use such variables in research to model the causes of agreement and disagreement
this study. Thus, little weight can be placed on the hypothesized among the sources of core self-evaluations data. This mimics a
causal ordering of the perceived job characteristics-job satisfac- more general call for more research on the causes and meaning of
tion relationship. For this reason, the results involving perceived self-other agreement recently voiced by Funder and Colvin
job characteristics are merely exploratory. Only experimental stud- (1997).
ies involving core self-evaluations, perceived task characteristics, In interpreting the results of the present studies, it is important
and task satisfaction would confirm the hypothesized causal to note that job complexity had a direct relationship with job
structure. satisfaction only in Study 2. This should not be surprising consid-
More weight can be placed on the relationship between core ering that virtually all of the effects of environmental conditions
self-evaluations and job complexity, which is where the unique are mediated by conscious perceptions (although there could be
contribution of this study lies. For the first time, it has been shown subconscious mechanisms involved in some cases). However, a
that part of the reason individuals with positive core self- cautionary note should be sounded here. Although the measure of
evaluations perceive more challenging jobs and report higher job complexity was correlated with perceptions, it is not a com-
levels of job satisfaction is that they actually have obtained more plete measure. A job title or brief job description does not provide
complex (and thus more challenging and intrinsically enriching) a detailed picture of a job's characteristics because jobs with the
jobs. Judge et al.'s (1997) theory of core evaluations suggests three same title can be very different. As House et al. (1996) noted, the
possible paths from core self-evaluations to job satisfaction: direct, somewhat imprecise measurement of job complexity makes it
indirect through job attribute perceptions, and indirect through more difficult to find associations with other variables. Thus, the
on-the-job actions taken to make the job more rewarding (e.g., imperfections in our measure of job complexity suggest that the
showing initiative). This study provides the first general support results might be conservative. It would be useful for future re-
for the action mediator. search to investigate whether individuals with a positive self-
Correlations between self- and significant-other reports of core concept attain more complex job duties within the same type of
self-evaluations in Study 1 were far from perfect (the mean cor- job.
rected correlation between the core traits was .49), and self-reports The response rate for Study 1 is a limitation of the study.
of core self-evaluations had a higher relationship with job satis- Because the majority of individuals receiving surveys did not
faction than the significant-other reports. One might question the respond, it raises the question of whether the same relationships
use of significant-other reports for self-evaluations. Whereas an would be observed if all individuals had responded. Viswesvaran,
individual's own core self-evaluations can be perceived directly, Barrick, and Ones (1995) developed a methodology to determine
from the perspective of others they must be inferred. However, the response level of nonrespondents that would invalidate the
errors can be made in both cases. On the one hand, self-reports conclusions on the basis of the survey of respondents. Using their
may not be completely accurate because of the possibility of formula, we estimated that the correlation between core self-
self-enhancement. Further, in this case, self-reports also introduce evaluations and job satisfaction among nonrespondents would
the possibility of response-response bias in the results. On the have to be .03 to render the correlations that were observed (.41 for
248 JUDGE, BONO, AND LOCKE

self-reported core self-evaluations, .19 for significant-other re- affectivity on job satisfaction in a field experiment. Organizational
ported core evaluations; see Table 1) nonsignificant. The nonre- Behavior and Human Decision Processes, 62, 55-62.
spondent correlation between core self-evaluations and job com- Cain, P. S., & Green, B. F. (1983). Reliabilities of selected ratings available
plexity that would render the observed correlation nonsignificant is from the Dictionary of Occupational Titles. Journal of Applied Psychol-
ogy, 68, 155-165.
.09. Because both of these values are well below the observed
Clausen, J. A. (1993). American lives: Looking back at the children of the
correlations (see Table 1), the relationships among nonrespondents
Great Depression. New York: Free Press.
would have to be quite different than those that were observed in
Costa, P. T., Jr., & McCrae, R. R. (1994). Set like plaster? Evidence for the
this study. Furthermore, the representativeness of the sample—all stability of adult personality. In T. F. Heatherton & J. L. Weinberger
data were collected from residents of a single Midwestern city—is (Eds.), Can personality change? (pp. 21-40). Washington, DC: Amer-
a limitation. It is critical that future research replicate these results ican Psychological Association.
with more diverse, particularly international, samples. Diener, E., Larsen, R. J., & Emmons, R. A. (1984). Person X Situation
Future studies should attempt to build on the theory of core interactions: Choice of situations and congruence response models.
evaluations offered by Judge et al. (1998). Although our findings Journal of Personality and Social Psychology, 47, 580-592.
demonstrate a relationship between core self-evaluations and job Dunham, R. B. (1976). The measurement and dimensionality of job char-
complexity (and thus extend previous findings that considered acteristics. Journal of Applied Psychology, 61, 404-409.
Dunham, R. B., Aldag, R. J., & Brief, A. P. (1977). Dimensionality of task
only perceptions of job characteristics), they do not shed light on
design as measured by the Job Diagnostic Survey. Academy of Manage-
the processes leading to the relationship. There are a number of
ment Journal, 20, 209-223.
mechanisms that may link core self-evaluations to job complexity. Eysenck, H. J., & Eysenck, S. B. G. (1968). Manual for the Eysenck
Job choice was one of the actions specifically mentioned by Judge Personality Inventory. San Diego, CA: Educational and Industrial Test-
et al. (1997). People who are generally confident in themselves ing Service.
should be more likely to think they can get challenging jobs. Fried, Y., & Ferris, G. R. (1987). The validity of the job characteristics
People with high self-esteem also tend to have better social net- model: A review and meta-analysis. Personnel Psychology, 40, 287-
works and make more favorable impressions on others (Locke, 322.
McClear, & Knight, 1996), enhancing their ability to obtain com- Funder, D. C., & Colvin, C. R. (1997). Congruence of others' and self-
plex jobs. Other possible mechanisms linking core self-evaluations judgments of personality. In R. Hogan, J. Johnson, & S. Briggs (Eds.),
and job complexity include job behaviors such as goal setting, goal Handbook of personality psychology (pp. 617-647). San Diego, CA:
Academic Press.
commitment, effort and tenacity in the face of setbacks, and coping
Gerhart, B. A. (1985). Sources of variance in perceptions of job complex-
with negative events at work. Other actions that might assist
ity. Unpublished doctoral dissertation, University of Wisconsin—Mad-
positively disposed individuals in obtaining complex jobs include ison.
efforts to improve one's skills and to exercise leadership. Finally, Gerhart, B. (1987). How important are dispositional factors as determinants
given the link between core self-evaluations and job complexity of job satisfaction? Implications for job design and other personnel
found in two separate studies here, and given the somewhat inexact programs. Journal of Applied Psychology, 72, 366-373.
measurement of job complexity, more work is needed on the Gerhart, B. (1988). Sources of variance in incumbent perceptions of job
construct validity of job complexity. Thus, although the present complexity. Journal of Applied Psychology, 73, 154-162.
study extends recent work on core self-evaluations, there is a need Click, W. H., Jenkins, G. D., Jr., & Gupta, N. (1986). Method versus
for further extension. substance: How strong are underlying relationships between job char-
acteristics and attitudinal outcomes? Academy of Management Jour-
nal, 29, 441-464.
References Hacket, G., & Betz, N. E. (1981). A self-efficacy approach to the career
development of women. Journal of Vocational Behavior, 18, 326-339.
Adelmann, P. K. (1987). Occupational complexity, control, and personal Hackman, J., & Oldham, G. (1980). Work redesign. Reading, MA:
income: Their relation to psychological well-being in men and women. Addison-Wesley.
Journal of Applied Psychology, 72, 529-537. Hall, D. T., & Foster, L. W. (1977). A psychological success cycle and goal
Anderson, C. R. (1977). Locus of control, coping behaviors, and perfor- setting: Goals, performance, and attitudes. Academy of Management
mance in a stress setting: A longitudinal study. Journal of Applied Journal, 20, 282-290.
Psychology, 62, 446-451. House, R. J., Shane, S. A., & Herold, D. M. (1996). Rumors of the death
Anderson, J. C., & Gerbing, D. W. (1988). Structural equation modeling in of dispositional research are vastly exaggerated. Academy of Manage-
practice: A review and recommended two-step approach. Psychological ment Review, 21, 203-224.
Bulletin, 103, 411-423. James, L. R., & Jones, A. P. (1980). Perceived job characteristics and job
Arvey, R. D., Bouchard, T. J., Segal, N. L., & Abraham, L. M. (1989). Job satisfaction: An examination of reciprocal causation. Personnel Psychol-
satisfaction: Environmental and genetic components. Journal of Applied ogy, 33, 97-135.
Psychology, 74, 187-192. James, L. R., & Tetrick, L. E. (1986). Confirmatory analytic tests of three
Bandura, A. (1997). Self-efficacy: The exercise of control. New York: causal models relating job perceptions to job satisfaction. Journal of
Freeman. Applied Psychology, 71, 77-82.
Block, J. (1971). Lives through time. Berkeley, CA: Bancroft Books. Joreskog, K. G., & Sorbom, D. (1989). USREL 7: A guide to the program
Brayfield, A. H., & Rothe, H. F. (1951). An index of job satisfaction. and applications. Chicago: SPSS.
Journal of Applied Psychology, 35, 307-311. Joreskog, K. G., & Sorbom, D. (1993). USREL 8 user's reference guide.
Brief, A. P. (1998). Attitudes in and around organizations. Thousand Oaks, Chicago: Scientific Software.
CA: Sage. Judge, T. A., Boudreau, J. W., & Bretz, R. D. (1994). Job and life attitudes
Brief, A. P., Butcher, A., & Roberson, L. (1995). Cookies, disposition, and of male executives. Journal of Applied Psychology, 79, 767-782.
job attitudes: The effects of positive mood inducing events and negative Judge, T. A., Locke, E. A., & Durham, C. C. (1997). The dispositional
 PERSONALITY AND JOB SATISFACTION 249

causes of job satisfaction: A core evaluations approach. Research in Necowitz, L. B., & Roznowski, M. (1994). Negative affectivity and job
Organizational Behavior, 19, 151-188. satisfaction: Cognitive processes underlying the relationship and effects
Judge, T. A., Locke, E. A., Durham, C. C., & Kluger, A. N. (1998). on employee behaviors. Journal of Vocational Behavior, 45, 270-294.
Dispositional effects on job and life satisfaction: The role of core Roos, P. A., & Treiman, D. J. (1980). Worker functions and worker traits
evaluations. Journal of Applied Psychology, 83, 17-34. for the 1970 U.S. census classification. In A. R. Miller, D. J. Treiman,
Kraiger, K., Billings, R. S., & Isen, A. M. (1989). The influence of positive P. S. Cain, & P. A. Roos (Eds.), Work, jobs, and occupations: A critical
affective states on task perceptions and satisfaction. Organizational review of the Dictionary of Occupational Titles (pp. 336-389). Wash-
Behavior and Human Decision Processes, 44, 12—25. ington, DC: National Academy Press.
Larsen, R. J., & Ketelaar, T. (1991). Personality and susceptibility to
Rosenberg, M. (1965). Society and the adolescent self-image. Princeton,
positive and negative emotional states. Journal of Personality and Social
NJ: Princeton University Press.
Psychology, 61, 132-140.
Scarpello, V., & Campbell, J. P. (1983). Job satisfaction: Are all the parts
Levenson, H. (1981). Differentiating among internality, powerful others,
there? Personnel Psychology, 36, 577-600.
and chance. In H. M. Lefcourt (Ed.), Research with the locus of control
construct (pp. 15-63). New York: Academic Press. Spector, P. E. (1982). Behavior in organizations as a function of employ-
Levy, P. E., & Baumgardner, A. H. (1991). Effects of self-esteem and ee's locus of control. Psychological Bulletin, 91, 482-497.
gender on goal choice. Journal of Organizational Behavior, 12, 529- Spector, P. E., & Jex, S. M. (1991). Relations of job characteristics from
541. multiple data sources with employee affect, absence, turnover intentions,
Locke, E. A., McClear, K., & Knight, D. (1996). Self-esteem and work. and health. Journal of Applied Psychology, 76, 46-53.
International Review of Industrial & Organizational Psychology, 11, Staw, B. M., Bell, N. E., & Clausen, J. A. (1986). The dispositional
1-32. approach to job attitudes: A lifetime longitudinal test. Administrative
Loher, B. T., Noe, R. A., Moeller, N. L., & Fitzgerald, M. P. (1985). A Science Quarterly, 31, 56-77.
meta-analysis of the relation of job characteristics to job satisfaction. Swann, W. B., Jr., Stein-Seroussi, A., & Giesler, R. B. (1992). Why people
Journal of Applied Psychology, 70, 280-289. self-verify. Journal of Personality and Social Psychology, 62, 392-401.
Magnus, K., Diener, E., Fujita, F., & Pavot, W. (1993). Extraversion and Viswesvaran, C., Barrick, M. R., & Ones, D. S. (1995). How definitive are
neuroticism as predictors of objective life events: A longitudinal anal- conclusions based on survey data: Estimating robustness to nonresponse.
ysis. Journal of Personality and Social Psychology, 65, 1046-1053. Personnel Psychology, 46, 551-567.
Magnusson, D. (1990). Personality development from an interactional Weiss, H. M., & Cropanzano, R. (1996). Affective events theory: A
perspective. In L. A. Pervin (Ed.), Handbook of personality (pp. 193- theoretical discussion of the structure, causes, and consequences of
222). New York: Guilford Press. affective experiences at work. Research in Organizational Behavior, 18,
Medsker, G. J., Williams, L. J., & Holahan, P. J. (1994). A review of
1-74.
current practices for evaluating causal models in organizational behavior
and human resources management research. Journal of Management, 20,
439_464.
Motowidlo, S. J. (1996). Orientation toward the job and organization. In Received June 24, 1998
K. R. Murphy (Ed.), Individual differences and behavior in organiza- Revision received May 8, 1999
tions (pp. 175-208). San Francisco: Jossey Bass. Accepted May 8, 1999

View publication stats