Content uploaded by Stephen E. Humphrey
Author content
All content in this area was uploaded by Stephen E. Humphrey on Mar 23, 2016
Content may be subject to copyright.
Integrating Motivational, Social, and Contextual Work Design Features:
A Meta-Analytic Summary and Theoretical Extension
of the Work Design Literature
Stephen E. Humphrey
Florida State University
Jennifer D. Nahrgang and Frederick P. Morgeson
Michigan State University
The authors developed and meta-analytically examined hypotheses designed to test and extend work
design theory by integrating motivational, social, and work context characteristics. Results from a
summary of 259 studies and 219,625 participants showed that 14 work characteristics explained, on
average, 43% of the variance in the 19 worker attitudes and behaviors examined. For example,
motivational characteristics explained 25% of the variance in subjective performance, 2% in turnover
perceptions, 34% in job satisfaction, 24% in organizational commitment, and 26% in role perception
outcomes. Beyond motivational characteristics, social characteristics explained incremental variances of
9% of the variance in subjective performance, 24% in turnover intentions, 17% in job satisfaction, 40%
in organizational commitment, and 18% in role perception outcomes. Finally, beyond both motivational
and social characteristics, work context characteristics explained incremental variances of 4% in job
satisfaction and 16% in stress. The results of this study suggest numerous opportunities for the continued
development of work design theory and practice.
Keywords: work design, job design, satisfaction, performance, social support
Interest in work design has a long history. Early writings fo-
cused on how the division of labor could increase worker effi-
ciency and productivity (Babbage, 1835; Smith, 1776). The first
systematic treatment of the topic was conducted in the early part of
the 20th century by Gilbreth (1911) and Taylor (1911), who
focused on specialization and simplification in an attempt to max-
imize worker efficiency. Yet, one of the problems of designing
work to maximize efficiency was that it tended to result in de-
creased employee satisfaction, increased turnover and absentee-
ism, and difficulties in managing employees in simplified jobs
(Hackman & Lawler, 1971).
Reacting to this, researchers developed theories focusing on the
motivating features of work (Hackman & Lawler, 1971; Hackman
& Oldham, 1975; Herzberg, Mausner & Snyderman, 1959; Turner
& Lawrence, 1965). The motivational approach forwarded by
these scholars has been influential over the past 30 years (Morge-
son & Campion, 2003). For example, the key articles summarizing
Hackman and Oldham’s job characteristics model and measures
have been cited nearly 2,000 times by researchers (ISI Web of
Knowledge, 2006). Although the model is more than 30 years old
and there are several criticisms of its key propositions and mea-
sures (Johns, Xie, & Fang, 1992; Morgeson & Humphrey, 2006;
Roberts & Glick, 1981; Taber & Taylor, 1990), it retains a central
place in work design theory today. Yet the success of the motiva-
tional approach has had two curious effects on work design re-
search over the past 30 years.
First, it has focused research attention on a limited set of
motivational work features (e.g., skill variety, autonomy). Al-
though these are important work characteristics, other important
aspects of work (such as the social environment and work context)
have been neglected. As a consequence, social and contextual
aspects of work have received less attention. This is unfortunate, as
research in other areas has documented the importance of both the
social environment and work context for a wide range of out-
comes.
Second, the success of the motivational approach has likely
contributed to a general decline of research and theorizing on work
design in the fields of industrial/organizational (I/O) psychology
and management. Because the motivational approach is widely
accepted, it appears that many in the fields of I/O psychology and
management concluded that it was “case closed” with respect to
work design. As Figure 1 demonstrates, work design research
published in top I/O psychology and management journals began
to decline in the late 1980s and has remained at a low level ever
since. However, work design research appears to be alive and well
outside of the top journals in the fields of I/O psychology and
management. Such a decline is not entirely unexpected, as scholars
in philosophy of science have noted that programs of research
traditionally have highly fertile periods, followed by the accumu-
lation of unsolved problems that require changes to the traditional
perspective or the introduction of a radically new perspective in
order to advance and spur new research (Bechtel, 1988; Kuhn,
1970; Laudan, 1977).
One way to invigorate an area of research is to use meta-analytic
techniques to clarify and synthesize existing empirical findings,
Stephen E. Humphrey, Department of Management, Florida State Uni-
versity; Jennifer D. Nahrgang and Frederick P. Morgeson, Eli Broad
Graduate School of Management, Michigan State University.
Thanks to Anne Downey for her proofreading and editing assistance.
Correspondence concerning this article should be addressed to Stephen
E. Humphrey, Department of Management, Florida State University, Tal-
lahassee, FL 32306-1110. E-mail: stephen.humphrey@fsu.edu
Journal of Applied Psychology Copyright 2007 by the American Psychological Association
2007, Vol. 92, No. 5, 1332–1356 0021-9010/07/$12.00 DOI: 10.1037/0021-9010.92.5.1332
1332
test hypotheses at a qualitatively different level, and point to the
best direction for new theoretical developments (Hunter &
Schmidt, 2004). Given the large amount of research on work
design that has occurred both within and outside the fields of I/O
psychology and management, this topic can benefit from meta-
analytic techniques.
The goal of our meta-analytic review is to summarize and
extend the literature on work design for individual-level jobs. We
define work design characteristics as the attributes of the task, job,
and social and organizational environment. Our focus is on work
design, rather than “job design,” because it recognizes that work
consists of the attributes of a job and the link between a job and the
broader work environment (Morgeson & Humphrey, 2006).
First, we replicate and extend Fried and Ferris’s (1987) meta-
analytic summary of the motivational characteristics. Compared to
Fried and Ferris, our analyses include almost 20 years of additional
research studies, 14 additional work design outcomes, and the first
formal meta-analytic test of Hackman and Oldham’s (1976) me-
diation model. Second, we develop a theoretical model examining
four social characteristics of work, an area long neglected in work
design research (Grant, 2007; Morgeson & Humphrey, 2006), and
three work context characteristics that have traditionally been
examined in the human physiology and engineering literatures
(Campion & Thayer, 1985; Parker & Wall, 2001). This expanded
model is summarized in Figure 2. Third, we assess the amount of
unique variance in work outcomes explained by social character-
istics and work context characteristics beyond what is explained by
motivational characteristics.
Testing and Extending Work Design Theory
Testing the Job Characteristics Model
Hackman and Oldham (1976) suggested that five work charac-
teristics make jobs more satisfying for workers: autonomy (i.e., the
freedom an individual has in carrying out work), skill variety (i.e.,
the extent to which an individual must use different skills to
perform his or her job), task identity (i.e., the extent to which an
individual can complete a whole piece of work), task significance
(i.e., the extent to which a job impacts others’ lives), and feedback
from the job (i.e., the extent to which a job imparts information
about an individual’s performance). These work characteristics
were expected to increase positive behavioral (e.g., job perfor-
mance) and attitudinal (e.g., job satisfaction) outcomes and de-
crease negative behavioral outcomes (e.g., absenteeism). In their
meta-analytic examination, Fried and Ferris (1987) found that
these five characteristics were strongly related to job satisfaction,
growth satisfaction, and internal work motivation, with weaker
relationships to job performance and absenteeism. In our expanded
set of research studies we expected to find similar relationships
between these five work characteristics and outcomes.
Hypothesis 1: Autonomy, skill variety, task identity, task
significance, and feedback from the job will be (a) positively
related to job satisfaction, (b) positively related to growth
satisfaction, (c) positively related to internal work motivation,
(d) positively related to job performance, and (e) negatively
related to absenteeism.
0
5
10
15
20
25
30
1969
1971
1973
1975
1977
1979
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003
"Top 7"
Management
Journal
Publications
All Other
Journal
Publications
Figure 1. Work design publications (1969 –2004). Podsakoff, Mackenzie, Bachrach, and Podsakoff (2005)
divided management-related journals (which include the traditional top industrial/organizational psychology
journals) into quartiles based upon the journal’s impact (assessed via citations per article). The top quartile
(seven journals) consisted of Academy of Management Journal, Academy of Management Review, Administra-
tive Science Quarterly, Journal of Applied Psychology, Organizational Behavior and Human Decision Pro-
cesses, Personnel Psychology, and Strategic Management Journal. These Top 7 journals accounted for almost
61% of all citations between 1981 and 1999. Moreover, the Top 7 journals “averaged almost six times more
citations per paper (23.93 vs. 4.54) from 1981 to 1999 than the seven bottom journals” (p. 481).
1333
WORK DESIGN META-ANALYTIC SUMMARY AND EXTENSION
Although Fried and Ferris’s (1987) review was limited to the
five outcomes of work directly specified by Hackman and Oldham
(1976), there is reason to suspect that this set of work character-
istics applies to a broader set of outcomes. In fact, Hackman and
Oldham’s (1976) theoretical model did not specifically argue that
the motivational characteristics would only relate to the five out-
comes. Instead, they argued more generally that these character-
istics would impact positive personal and work outcomes. Thus,
we formally tested whether these characteristics generalize to a
broader set of behavioral (e.g., turnover intentions) and attitudinal
(e.g., organizational commitment, job involvement, supervisor sat-
isfaction) outcomes.
Hypothesis 2: Autonomy, skill variety, task identity, task
significance, and feedback from the job will be (a) positively
related to positive behavioral outcomes, (b) positively related
to positive attitudinal outcomes, and (c) negatively related to
negative behavioral outcomes.
Hackman and Oldham (1976) suggested that motivational work
characteristics impact behavioral and attitudinal outcomes through
their influence on three critical psychological states: experienced
meaningfulness (i.e., the degree to which an employee feels the job
has value and importance), experienced responsibility (i.e., the
degree to which an employee feels liable and accountable for job
results), and knowledge of results (i.e., the degree to which the
employee is aware of his or her level of performance). Specifi-
cally, skill variety, task identity, and task significance are thought
to impact experienced meaningfulness, autonomy is thought to
impact experienced responsibility, and feedback from the job is
thought to impact knowledge of results.
We also expected the critical psychological states to mediate the
relationships between these work characteristics and work out-
comes. Despite the considerable amount of work design research,
however, there has been no comprehensive meta-analytic evalua-
tion of the proposed mediation. Fried and Ferris (1987, p. 305)
examined only corrected correlations between the motivational
work characteristics, the critical psychological states, and out-
comes and explicitly noted that “bivariate correlational analysis
cannot provide a specific test of the mediating hypothesis and is
less appropriate than other more sophisticated statistical tests.”
Using more sophisticated techniques, we performed the first meta-
analytic test of these mediational predictions.
Whereas Hackman and Oldham (1975, 1980) suggested that the
critical psychological states would each independently act as me-
diators, later work suggested that the true mediation model is
different (Johns et al., 1992; Oldham, 1996). In particular, Johns et
al. (p. 667) suggested that experienced meaning was a “particularly
encompassing psychological state,” as it served as a mediator for
all five motivational characteristics. Johns et al. did not argue that
the other two critical psychological states should be removed from
semoctuO kroWscitsiretcarahCngiseD kroW Mediators
Behavioral Outcomes
Performance Objective
Performance - Subjective
Absenteeism
Turnover Intentions
Motivational Characteristics
Autonomy
Work Scheduling Autonomy
Work Methods Autonomy
Decision-Making Autonomy
Skill Variety
Task Variety
Significance
Task Identity
Feedback from the Job
Information Processing
Job Complexity
Specialization
Problem Solving
Critical Psychological States
Experienced Meaningfulness
Responsibility
Attitudinal Outcomes
Satisfaction - Job
Satisfaction - Supervisor
Satisfaction - Coworker
Satisfaction - Compensation
Satisfaction - Growth
Satisfaction - Promotion
Organizational Commitment
Job Involvement
Internal Work Motivation
Social Characteristics
Interdependence
Feedback from Others
Social Support
Interaction Outside the Organization
Role Perception Outcomes
Role Ambiguity
Role Conflict
Work Context Characteristics
Physical Demands
Work Conditions
Er
g
onomics
Well-Being Outcomes
Anxiety
Stress
Burnout/exhaustion
Overload
Knowledge of Results
Experienced
-
Figure 2. Expanded work design model.
1334
HUMPHREY, NAHRGANG, AND MORGESON
mediational models but instead suggested that each contributes in
varying degrees in the mediational process. Following Johns et al.,
we tested an alternative mediational model in which we compared
the simultaneous mediation ability of all three critical psycholog-
ical states.
Hypothesis 3: Experienced meaningfulness, experienced re-
sponsibility, and knowledge of results will mediate the rela-
tionships between (a) autonomy, (b) skill variety, (c) task
identity, (d) task significance, and (e) feedback from the job
and the behavioral and attitudinal outcomes.
Extending the Job Characteristics Model: Additional
Motivational Characteristics
In addition to the five work characteristics identified by Hack-
man and Oldham (1975), five other motivational work character-
istics have been discussed in the literature: task variety, informa-
tion processing, job complexity, specialization, and problem
solving. Task variety (i.e., the extent to which an individual per-
forms different tasks at his or her job) is different from skill
variety, such that skill variety focuses on the skills necessary to
perform a job, whereas task variety focuses on the specific tasks
performed. Thus, task variety is conceptually more similar to the
other four Hackman and Oldham (1975) characteristics (i.e., au-
tonomy, task significance, task identity, and feedback from the
job) than skill variety, in that these four characteristics are con-
cerned with how work is performed and the specific tasks com-
posing a job (Morgeson & Humphrey, 2006). In contrast, skill
variety reflects the knowledge and skills necessary to perform a
job.
It is interesting to note that task variety was not measured in the
most popular work design measure (i.e., the Job Diagnostic Sur-
vey; Hackman & Oldham, 1975) and instead is most frequently
measured via the Job Characteristics Inventory (Sims, Szilagyi, &
Keller, 1976). Because previous researchers often did not concep-
tually separate skill and task variety, it is not surprising that Fried
and Ferris (1987) combined measures of skill and task variety into
the same construct when reviewing the work design literature. Due
to differences in their construct definitions and operationalizations,
however, we separated the two constructs. Yet because task variety
is conceptually similar to autonomy, task significance, task iden-
tity, and feedback from the job, such that they all focus on how
work occurs and the specific tasks being performed, it was rea-
sonable to speculate that task variety would have similar relation-
ships with specified outcomes.
Hypothesis 4: Task variety will be (a) positively related to
positive behavioral outcomes, (b) positively related to posi-
tive attitudinal outcomes, and (c) negatively related to nega-
tive behavioral outcomes.
In contrast to task-focused motivational characteristics, several
motivational characteristics are primarily concerned with the
knowledge demands of work. These include information process-
ing, job complexity, specialization, and problem solving. First,
information processing is the extent to which a job necessitates an
incumbent to focus on and manage information. Past research has
suggested that jobs differ in their level of monitoring and process-
ing of information (Martin & Wall, 1989; Wall & Jackson, 1995;
Wall, Jackson, & Mullarkey, 1995). Higher levels of information
processing were expected to change the requirements for jobs, as
employees require high levels of knowledge in high information
processing jobs in order complete their work (Morgeson & Hum-
phrey, 2006).
Second, job complexity is the extent to which a job is multifac-
eted and difficult to perform. Others have discussed the opposite of
job complexity, namely job simplicity (Campion, 1988; Edwards,
Scully, & Brtek, 2000). Because complex jobs involve the use of
high-level skills, they tend to be mentally demanding. Previous
research has suggested that high job complexity promotes satis-
faction but is also likely to hurt efficiency and promote perceptions
of work overload, as incumbents find that high-complexity work
both engages and overwhelms them.
Third, specialization is the extent to which a job involves the
performance of tasks requiring specific knowledge and skill.
Whereas skill and task variety reflect the breadth of behaviors and
skills involved in a job, specialization represents the depth of
knowledge and skills necessary. Although there is only limited
research surrounding specialization (e.g., Campion, 1988; Ed-
wards et al., 2000), recent research has suggested that increasing
specialization may resolve the tradeoff in work design between
satisfaction and efficiency (Morgeson & Campion, 2002). That is,
although specialization may make work more efficient by simpli-
fying it, it also may make work more motivating by requiring a
depth of knowledge in a specific area.
Fourth, problem solving is the extent to which a job requires the
production of unique solutions or ideas. It is conceptually similar
to creativity in that it involves innovating, solving nonroutine
problems, and dealing with (or preventing) errors (Jackson, Wall,
Martin, & Davids, 1993; Wall, Corbett, Clegg, Jackson, & Martin,
1990).
These four work characteristics were expected to impact a
variety of work outcomes. Past research has suggested that increas-
ing knowledge requirements makes work more intrinsically moti-
vating and promotes positive attitudinal outcomes. For example,
jobs with high problem-solving requirements provide a chance for
employees to perform in challenging, novel situations in which
they can demonstrate and reinforce their sense of competence on
the job (Deci & Ryan, 2000). In these jobs, the employees were
expected to be more motivated and more satisfied.
Whereas these work characteristics were expected to promote
positive attitudinal outcomes, the increased cognitive requirements
associated with these four work characteristics were expected to
harm employee well-being. For example, high-complexity jobs
were expected to tax employees’ cognitive resources, increasing
their stress and perceptions of overload. Finally, we expected that
these work characteristics would increase performance. For exam-
ple, specialization has been shown to increase job efficiency
(Morgeson & Campion, 2002) and therefore performance.
Hypothesis 5: Information processing, job complexity, spe-
cialization, and problem solving will be (a) positively related
to positive behavioral outcomes, (b) positively related to
positive attitudinal outcomes, and (c) negatively related to
negative behavioral outcomes.
1335
WORK DESIGN META-ANALYTIC SUMMARY AND EXTENSION
Finally, relying on Hackman and Oldham’s (1975) model, we
have thus far discussed autonomy as a one-dimensional construct.
However, numerous scholars have suggested that autonomy has
multiple facets with unique predictive qualities (Breaugh, 1985;
Morgeson & Humphrey, 2006). For example, Jackson and col-
leagues (1993) argued that autonomy can be conceptualized as
work scheduling autonomy (i.e., the freedom to control the sched-
uling and timing of work) and work methods autonomy (i.e., the
freedom to control which methods and procedures are utilized).
Additional research has suggested that decision-making autonomy
(i.e., the freedom to make decisions at work) is also an important
component of autonomy (e.g., Karasek et al., 1998).
Although each of these three facets of autonomy was expected
to relate to work outcomes, there was reason to suspect differences
in their impact. For example, as compared to work scheduling
autonomy, job incumbents with high levels of work methods
autonomy should perceive that they have greater influence on how
a task is accomplished. That is, work methods autonomy allows
job incumbents the opportunity to influence the specific behaviors
on the job, whereas work scheduling autonomy just suggests that
an incumbent can influence how the behaviors are ordered. We
therefore hypothesized that the three dimensions of autonomy
would be related to work outcomes. In addition, we examined
whether the relationships between these three dimensions and
work outcomes demonstrate differences in magnitudes on an ex-
ploratory basis.
Hypothesis 6: Work scheduling autonomy, work methods
autonomy, and decision-making autonomy will be (a) posi-
tively related to positive behavioral outcomes, (b) positively
related to positive attitudinal outcomes, and (c) negatively
related to negative behavioral outcomes.
Extending the Job Characteristics Model: Social
Characteristics
Early work design research recognized the importance of the
social environment (Trist & Bamforth, 1951; Turner & Lawrence,
1965). In fact, two social characteristics (dealing with others and
friendship opportunities) were identified and examined (Hackman
& Lawler, 1971). Although it was found that these social charac-
teristics related to satisfaction, their lack of relationships with
behavioral outcomes or motivation tempered enthusiasm. From
that point forward, with the exception of their inclusion in the Job
Characteristics Inventory (Sims et al., 1976), social characteristics
received much less attention in the work design literature (Morge-
son & Campion, 2003). Although social information processing
theory deals with social influences on the perceptions of motiva-
tional work characteristics (Pfeffer, 1981; Salancik & Pfeffer,
1978), it does not discuss social characteristics as substantive work
characteristics.
Recently, researchers have noted that social characteristics are
important components of work (Parker & Wall, 2001) that are
nonredundant with motivational characteristics (Grant, 2007;
Morgeson & Humphrey, 2006). For example, researchers have
noted that relationships between workers are among the most
important determinants of well-being (Myers, 1999) and percep-
tions of meaningful work (Gersick, Bartunek, & Dutton, 2000;
Wrzesniewski, Dutton, & Debebe, 2003). These characteristics
were expected to reduce job stress by buffering workers against
negative job events (Karasek, 1979; Karasek, Triantis, &
Chaudhry, 1982). They may also increase work motivation (Adler
& Kwon, 2002) and prosocial work behaviors (Grant, 2007), as
they promote resilience, security, and positive moods on the job
(Ryan & Deci, 2001).
In addition, part of the growth in interest in social characteristics
may be attributed to the increased use of teams in organizations
(Ilgen, 1999). A team, by definition, has some degree of interde-
pendence amongst its members (Guzzo & Dickson, 1996), pro-
ducing ample opportunities for social interaction. Moreover, a
number of researchers have begun to conduct team task analyses
(e.g., Arthur, Villado, & Bennett, in press) that specifically high-
light the importance of interdependence (Arthur, Edwards, Bell,
Villado, & Bennett, 2005).
We focused on four social characteristics. Interdependence is
the extent to which a job is contingent on others’ work and other
jobs are dependent on the work of the focal job. This dimension
has alternatively been labeled dealing with others (Hackman &
Lawler, 1971). Our focus was solely on task interdependence,
rather than the broader set of interdependencies people may share
(Saavedra, Earley, & Van Dyne, 1993). Feedback from others is
the extent to which other organizational members provide perfor-
mance information. It is different from feedback from the job, as
it focuses more broadly on the interpersonal component of feed-
back rather than the performance information derived directly
from the work itself. Social support is the extent to which a job
provides opportunities for getting assistance and advice from ei-
ther supervisors or coworkers (Karasek, 1979; Karasek et al.,
1998) and includes friendship opportunities on the job (Sims et al.,
1976). Finally, interaction outside the organization is the extent to
which a job requires an incumbent to communicate with people
(e.g., suppliers or customers) external to the organization. Alter-
natively labeled serves the public (Stone & Gueutal, 1985), this
dimension reflects a social component of work linking job incum-
bents to people who are not members of an employee’s organiza-
tion.
Social characteristics are likely to impact a variety of work
outcomes. First, social characteristics were expected to impact
well-being, as social interactions are associated with positive mood
(Watson, 2000). Wanting to form, participate in, and maintain
interpersonal relationships is a fundamental motivation (Baumeis-
ter & Leary, 1995), and having interpersonal relationships helps to
increase well-being (Cohen & Wills, 1985). Research has demon-
strated that social activity, regardless of its nature, extent, duration
or valence, has a positive quality and conveys feelings of energy,
enthusiasm, and general feelings of positive affect (Watson, 1988,
2000). Thus, increases in social aspects of work were expected to
increase incumbent well-being.
Social characteristics should enhance role perceptions, as role
theory suggests that roles become more clearly defined through
greater contact with others (Biddle, 1979). First, highly interde-
pendent jobs provide increased contact and more opportunities to
communicate what each worker requires (Salas, Rozell, Mullen, &
Driskell, 1999), what is expected in return (Seers, Petty, & Cash-
man, 1995), and what each worker is doing (Humphrey, Hollen-
beck, Meyer, & Ilgen, 2007). That is, this contact helps bound
individual roles (Alderfer & Smith, 1982) by clarifying the roles
that each individual fills (Tuckman, 1965). Second, feedback from
1336
HUMPHREY, NAHRGANG, AND MORGESON
others enhances role perceptions by providing opportunities to
negotiate and define roles with the appropriate people who hold
the expectations and provide the performance feedback (Graen,
1976). Third, high levels of social support provide opportunities in
which people can garner advice and assistance from others. This
form of interaction is likely to help people clarify their roles and
address concerns when they experience incompatible expectations.
Fourth, interactions outside an organization allow incumbents to
gain additional (external) insight into the specific tasks they per-
form and provide opportunities for additional, nonredundant feed-
back on their performance.
We also expected social characteristics to impact attitudinal
outcomes. The well-being literature has demonstrated that inter-
actions with others make work more satisfying for an employee
(Ryan & Deci, 2001). In addition, having greater interaction with
others (through greater interdependence or interaction outside the
organization) creates a more complex and challenging job, which
can increase motivation (Kiggundu, 1983). Finally, social support
from coworkers and supervisors has been found to be important in
buffering workers from negative outcomes (Johnson & Hall, 1988;
Karasek et al., 1998), thereby increasing satisfaction.
Finally, we expected that social characteristics would impact
behavioral outcomes. Social characteristics provide a chance for
job incumbents to learn from others. That is, having greater inter-
action with other employees in performance-oriented (i.e., inter-
dependence) and non-performance-oriented contexts (i.e., social
support), having greater interaction outside the organization, and
having opportunities for direct performance-related feedback (i.e.,
feedback from others) provides job incumbents the opportunity to
learn how to perform their job more effectively through the trans-
fer of implicit and explicit knowledge (Berman, Down, & Hill,
2002). In addition, we expected that social characteristics would
decrease absenteeism and turnover, as social interaction can reduce
the negative outcomes associated with work (e.g., stress and over-
load) and promote positive affect in the job incumbents, making
them less likely to want to skip work and more likely to want to
continue to work at the organization (Steers & Mowday, 1981).
Hypothesis 7: Interdependence, feedback from others, social
support, and interaction outside the organization will be (a)
positively related to positive behavioral outcomes, (b) posi-
tively related to positive attitudinal outcomes, and (c) nega-
tively related to negative behavioral outcomes.
Extending the Job Characteristics Model: Work Context
Characteristics
Work context characteristics, which reflect the broad perfor-
mance context, have been virtually ignored in the fields of I/O
psychology and management (Morgeson & Campion, 2003). How-
ever, researchers in related fields such as work physiology, ergo-
nomics, human factors engineering, and biomechanics have stud-
ied the physical and environmental context (Campion & Thayer,
1985). Three work context characteristics are pertinent for our
discussion: physical demands, work conditions, and ergonomics.
Physical demands reflect the amount of physical activity or
effort necessary for a job. Alternatively labeled physical ease
(Edwards et al., 2000), this characteristic measures the physical
strength, endurance, and activity components of a job. Work
conditions reflect aspects of the work environment such as
health hazards, temperature, and noise (Campion & McClel-
land, 1991; Edwards, Scully, & Brtek, 1999). Finally, ergonom-
ics reflects the extent to which work permits appropriate pos-
ture and movement.
There are reasons to expect that work context characteristics
will impact various work outcomes. First, physical demands, work
conditions, and ergonomics reflect the extent to which the job is
designed in terms of biological concerns (Campion & Thayer,
1985). If physical demands increase and work conditions or ergo-
nomics decrease, job incumbents will become increasingly phys-
ically uncomfortable (Campion, 1988), which in turn will hurt
attitudinal outcomes such as job satisfaction. Moreover, if jobs are
uncomfortable and dissatisfying, it is likely that job incumbents
will want to avoid going to work and will instead look for new
jobs.
Hypothesis 8: Physical demands will be (a) negatively related
to positive behavioral outcomes, (b) negatively related to
positive attitudinal outcomes, and (c) positively related to
negative behavioral outcomes.
Hypothesis 9: Work conditions and ergonomics will be (a)
positively related to positive behavioral outcomes, (b) posi-
tively related to positive attitudinal outcomes, and (c) nega-
tively related to negative behavioral outcomes.
Integrating Motivational, Social, and Work Context
Characteristics
Thus far, we have suggested that motivational characteristics,
social characteristics, and work context characteristics impact be-
havioral and attitudinal outcomes. We also expected unique con-
tributions from these work characteristics. Because motivational
characteristics focus on individual job components, social charac-
teristics focus on the interactional components, and work context
characteristics focus on contextual components, they were ex-
pected to have nonredundant effects on behavior and reactions of
workers. Thus, the inclusion of all three sets of characteristics
would explain unique variance in work outcomes.
Although researchers have not traditionally examined the incre-
mental contribution of either social or work context characteristics
above the motivational characteristics, there is some empirical
evidence suggesting that this may occur. For example, Hackman
and Lawler (1971) and Brief and Aldag (1978) found only a
modest correlation between their motivational characteristics and
several social characteristics. Yet they found that satisfaction was
related to both sets of work characteristics. More recently, Morge-
son and Humphrey (2006) found that social support incrementally
predicted job satisfaction, training requirements, and compensa-
tion requirements beyond the motivational characteristics in their
study.
Hypothesis 10: (a) Social characteristics and (b) work context
characteristics will explain unique variance in the behavioral
and attitudinal outcomes, above and beyond motivational
characteristics.
1337
WORK DESIGN META-ANALYTIC SUMMARY AND EXTENSION
Method
Literature Search
A literature search was conducted to identify published articles,
conference papers, and doctoral dissertations that were related to
the design of work. The articles were identified through computer-
based searches of the PsycINFO (1887–2004) and Web of Science
ISI (1970 –2004) databases. Searches included the terms work or
job with keywords such as design, content, redesign, complexity,
characteristics, conditions, dimensions, scope, demands, social
support, enrichment, and interdependence. In addition, keywords
from Hackman and Oldman’s (1975, 1976) job characteristic
model, such as job feedback, skill variety, task identity, task
significance, autonomy, and psychological states, were also used
in the searches. The electronic search was supplemented with a
manual search of reference lists of key empirical and theoretical
articles as well as reference sections from key chapters on work
design and prior meta-analyses. The searches identified more than
8,000 published articles, dissertations, and conference presenta-
tions.
Inclusion Criteria
The abstracts obtained from this initial search were reviewed for
appropriate content and considered for inclusion in the meta-
analysis. After reading the abstracts, we eliminated studies without
data (theoretical work or literature reviews) and studies outside of
the work context. This resulted in an initial population that was
split among the three authors for review. Overall, we examined
677 studies to determine whether each would be included in the
meta-analysis. A number of decision rules were used to determine
which studies would be included. First, a study must have inves-
tigated at least one relationship from the constructs of interest.
Second, the study had to report sufficient results to calculate an
effect size. Third, the study had to be a unique sample that had not
been previously included in the current meta-analysis. If a data set
was used more than once, we coded all of the data from the first
published manuscript utilizing the data set. We then examined
each subsequent study to determine whether it presented unique
information about the data set. If a subsequent study presented
unique information, we coded that information; if a subsequent
study did not present any unique information beyond what was
already coded, we did not include the subsequent study in our
analyses. These inclusion criteria reduced our final study popula-
tion to 259 articles, of which 232 were published articles, 23 were
dissertations, and 4 were conference presentations.
All three authors participated in the coding of the studies. We
each coded approximately one third of the total set of manuscripts.
We independently coded each manuscript and met weekly as a
group to discuss the manuscripts coded that week. During the
weekly meetings, we clarified any ambiguous coding situations
(e.g., whether a variable represented Construct A or Construct B),
discussed whether an article’s data set was unique, and worked to
achieve consensus on any disagreements.
In conducting this meta-analysis, we tried to be as comprehen-
sive as possible in capturing all work design constructs. Our initial
set of work design constructs was derived from the set of charac-
teristics identified by Morgeson and Humphrey (2006). As we
comprehensively reviewed the literature, we started with the 21
dimensions of work design these authors had identified. We then
read the recent reviews of the work design literature in order to
expand our construct domain. Finally, we examined our population
of 677 articles for any additional work design characteristics. At
our weekly meetings, we discussed whether variables found in that
week’s articles should be coded. At the end of this process, we had
a list of 33 potential work design characteristics to be coded. To
develop the list of mediators and outcomes, we started with the
outcomes investigated in the Fried and Ferris (1987) meta-
analysis. We then added all relevant constructs studied in previous
work design studies, following the same process as described
above. This resulted in our coding 36 potential outcomes. In sum,
we were coding a total of 2,346 potential relationships. Given the
theoretical focus of this article and the low k associated with
numerous relationships, we only included a subset of the total
possible number of relationships in our meta-analytic review.
Specifically, we included a work characteristic in our meta-
analytic review if there were at least five studies examining the
relationship between the characteristic and job satisfaction. This
criterion was selected in order to include a large number of work
characteristics yet simultaneously restrict the empirical examina-
tion to only those characteristics for which there was a significant
accumulation of research. Because of the limited number of studies
examining specialization, problem solving, and ergonomics, we
followed our inclusion criterion and thus did not meta-analytically
summarize these characteristics. Nonetheless, in an effort to be as
comprehensive as possible, we coded all of these relationships.
We developed several rules regarding how we would code
variables into constructs. The purpose of these rules was to have a
common rubric for coding frequently encountered variables that
were ambiguously labeled. We developed these rules by examin-
ing the wording of the measures. During our weekly meetings, we
discussed each ambiguous variable and reached consensus on
which constructs were being captured by a specific measure. For
example, we concluded that the Job Diagnostic Survey (Hackman
& Oldham, 1980) variety measure assessed skill variety, whereas
the Job Characteristics Inventory (Sims et al., 1976) variety mea-
sure assessed task variety. The complete set of rules is available
from Stephen E. Humphrey upon request. These procedures re-
sulted in the coding of 6,333 unique correlations across the 259
articles.
In order to check the reliability of the coding, two of the authors
coded a random sample of approximately 10% of the articles (25
out of 259). We found there were only 66 differences in coding
(out of 958 pieces of information coded) on the construct labels,
resulting in an interrater agreement of 93%. For the coding of the
correlations in the studies, we found there were only 70 differences
in coding (out of 958 pieces of information coded), again resulting
in an interrater agreement of 93%. The interrater reliability esti-
mate for the constructs was .98 and for the correlations was .97.
All differences were checked against the original documents to
ensure that only correctly coded information was included in
analyses.
Meta-Analytic Procedures
We used the Schmidt–Hunter psychometric meta-analysis
method (Hunter & Schmidt, 2004) to conduct the meta-analytic
review. For studies with multiple measures of the same construct,
1338
HUMPHREY, NAHRGANG, AND MORGESON
we followed Hunter and Schmidt’s recommendations by creating
composite correlation values. This prevented a study being
“double-counted” in the meta-analysis. In contrast, studies that
included multiple independent samples were separately coded. We
also corrected for measure unreliability. Correlations from indi-
vidual samples were corrected for measurement error in both the
predictor and the criterion scores using Cronbach’s alpha (these
values were provided by the majority of studies). For the studies
missing this reliability coefficient, we used the average value from
the other studies.
Several pieces of information are presented concerning the
population correlation estimates. First, we include both the uncor-
rected (r) and corrected (!) estimates. Second, we include the 95%
confidence interval (CI) for each corrected population correlation.
Finally, we present the number of studies included in determining
the correlation (k) and the total number of participants in the
studies (n).
The following meta-analytic regression procedures were fol-
lowed. First, as sample sizes differed across studies, we followed
previous recommendations (Viswesvaran & Ones, 1995) and uti-
lized the harmonic mean when calculating sample sizes for the
meta-analytic regression. Second, we used ordinary least squares
techniques for meta-analytic regression, as they have less restric-
tive assumptions than maximum likelihood and are more optimal
when the data are in the form of correlations, rather than covari-
ances. Third, as there were many relationships between work
design characteristics and outcomes for which we were not able to
find any studies that examined the relationships, we reduced our
correlation matrix for specific hypothesis tests to include only
constructs for which there was a full matrix. In doing so, we
calculated a new sample size derived from the harmonic mean for
each regression analysis.
Results
Correlation Results
We first examined the relationships between the work design
characteristics. As seen in Table 1, the corrected intercorrelations
were positive in sign (with the exception of the relationships with
the two work context characteristics) and generally moderate in
magnitude (mean !".25). This suggested that although the work
design characteristics were interrelated, they were not so highly
correlated as to be multiple indicators of the same construct. There
were several interesting correlations to note. First, the eight moti-
vational characteristics were more highly correlated with one an-
other (mean !".46) than with either the four social characteristics
(mean !".27) or the two work context characteristics (mean !"
–.05), providing evidence that motivational, social, and work con-
text characteristics were unique sets of characteristics. Second, the
results showed that although feedback from the job and feedback
from others (which have been combined into one scale in some
studies) were correlated (!".57; 95% CI " .51 # .57 # .63),
they shared only approximately one third of the same variance.
Thus, there seemed to be ample evidence for studying the two
constructs independently.
Motivational Characteristics
We sought to replicate Fried and Ferris’s (1987) findings by
hypothesizing that autonomy, skill variety, task identity, task sig-
nificance, and feedback from the job would be (a) positively
related to job satisfaction, (b) positively related to growth satis-
faction, (c) positively related to internal work motivation, (d)
positively related to job performance, and (e) negatively related to
absenteeism. Table 2 presents the results of these analyses. First,
all five motivational characteristics were positively related to job
satisfaction (mean !".41), growth satisfaction (mean !".55),
and internal work motivation (mean !".39). Thus, the results
strongly supported Hypotheses 1a, 1b, and 1c. Second, we note
that autonomy was the only motivational characteristic for which
the 95% CI did not include zero (!".17; 95% CI " .04 # .17 #
.30) for the relationship with objective performance. In contrast,
autonomy, task identity, task significance, and feedback from the
job had non-zero correlations with subjective performance (mean
!".18). Thus, Hypothesis 1d was supported for subjective
performance with four of the five motivational characteristics
discussed in Fried and Ferris. Finally, in testing Hypothesis 1e, we
found that autonomy (!"–.15; 95% CI " –.21 # –.15 # –.09),
task identity (!"–.09; 95% CI " –.17 # –.09 # –.01), and
feedback from the job (!"–.10; 95% CI " –.16 # –.10 # –.03)
were all related to absenteeism, whereas the 95% CIs for skill
variety (!"–.07; 95% CI " –.15 # –.07 # .01) and task
significance (!".06; 95% CI " –.04 # .06 # .17) included zero.
In comparison to Fried and Ferris’s (1987) meta-analysis, the
results of our meta-analysis demonstrated generally stronger rela-
tionships between the motivational characteristics and outcomes.
One notable exception was absenteeism, which demonstrated
markedly smaller magnitude correlations in our review. This dif-
ference may be partially attributable to the fact that our review had
between 8 and 12 studies of relationships with absenteeism, and
the total sample sizes ranged from 1,706 to 2,902 (depending on
the specific work characteristic), whereas Fried and Ferris only had
3 studies and a total sample size of 658 per characteristic. Because
of the larger number of studies and sample sizes in our review, we
were able to produce more accurate coefficient estimates.
Hypothesis 2 predicted that autonomy, skill variety, task iden-
tity, task significance, and feedback from the job would be (a)
positively related to positive behavioral outcomes, (b) positively
related to positive attitudinal outcomes, and (c) negatively related
to negative behavioral outcomes. First, as shown in Table 2, none
of these characteristics were related to turnover intentions (mean
!"–.03). Second, all of the motivational characteristics were
related to supervisor satisfaction (mean !".30), coworker satis-
faction (mean !".39), compensation satisfaction (mean !".19),
and promotion satisfaction (mean !".21), with the strongest
relationships consistently held by autonomy (except for promotion
satisfaction, for which feedback from the job demonstrated the
largest relationship). Third, these five motivational characteristics
were related to organizational commitment (mean !".34) and job
involvement (mean !".29). Fourth, only autonomy (!"–.23;
95% CI " –.35 # –.23 # –.12) and feedback from the job (!"
–.43; 95% CI " –.59 # –.43 # –.28) were related to role
ambiguity, whereas autonomy (!"–.17; 95% CI " –.26 #
–.17 # –.09), feedback from the job (!"–.32; 95% CI " –.42 #
–.32 # –.23), and task identity (!"–.17; 95% CI " –.22 #
–.17 # –.12) were related to role conflict. Fifth, turning to well-
being outcomes, autonomy (!"–.10; 95% CI " –.14 # –.10 #
–.06) and feedback from the job (!"–.32; 95% CI " –.37 #
–.32 # –.27) were related to anxiety; autonomy (!"–.23; 95%
1339
WORK DESIGN META-ANALYTIC SUMMARY AND EXTENSION
Table 1
Interrelationships of Work Design Characteristics
Characteristic
Autonomy
Skill
variety
Task
variety
Task
significance
Task
identity
Feedback
from
the job
Information
processing
Job
complexity Interdependence
Feedback
from
others
Social
support
Interaction
outside the
organization
Physical
demands
Work
conditions
r, !
(95% CI)
r, !
(95% CI)
r, !
(95% CI)
r, !
(95% CI)
r, !
(95% CI)
r, !
(95% CI)
r, !
(95% CI)
r, !
(95% CI)
r, !
(95% CI)
r, !
(95% CI)
r, !
(95% CI)
r, !
(95% CI)
r, !
(95% CI)
r, !
(95% CI)
Skill variety .47, .64
(.60, .69)
—
k, N 100, 58,350
Task variety .34, .46
(.32, .60)
.47, .52
(.47, .56)
—
k, N 21, 8,877 2, 974
Significance .36, .50
(.46, .54)
.44, .62
(.58, .66)
.40, .52
(.28, .75)
—
k, N 100, 41,837 78, 37,758 8, 2,885
Task identity .39, .55
(.51, .59)
.26, .37
(.32, .41)
.27, .39
(.26, .51)
.27, .39
(.34, .43)
—
k, N 111, 43,427 80, 36,334 16, 5,881 83, 37,435
Feedback from
the job
.38, .53
(.49, .56)
.36, .50
(.46, .55)
.30, .40
(.30, .50)
.38, .56
(.51, .61)
.35, .49
(.45, .54)
—
k, N 110, 44,390 79, 36,256 15, 5,765 80, 37,082 92, 41,108
Information
processing
.37, .45
(.41, .49)
.50, .57
(.53, .62)
.43, .48
(.43, .53)
.39, .44
(.39, .50)
$.02, $.02
($.09, .04)
.21, .25
(.18, .32)
—
k, N 6, 1,838 2, 974 2, 974 2, 974 2, 974 2, 974
Job complexity .35, .43
(.28, .58)
.40, .51
(.38, .64)
.54, .62
(.40, .84)
.25, .31
(.19, .42)
.17, .22
(.02, .41)
.17, .21
(.14, .29)
.51, .60
(.51, .68)
—
k, N 14, 4,926 9, 3,194 4, 2,404 9, 3,194 10, 3,703 11, 4,644 2, 974
Interdependence .19, .29
(.13, .44)
.39, .61
(.50, .72)
.14, .18
(.08, .28)
.31, .50
(.43, .57)
.13, .19
(.08, .30)
.26, .41
(.31, .51)
.21, .25
(.19, .31)
.25, .37
(.21, .52)
—
k, N 33, 19,733 23, 16,448 11, 4,695 25, 51,700 28, 17,889 28, 17,889 2, 974 7, 2,610
Feedback from
others
.35, .48
(.40, .56)
.27, .37
(.31, .44)
.07, .10
(.06, .15)
.25, .36
(.30, .42)
.21, .31
(.25, .38)
.40, .57
(.51, .63)
.12, .14
(.08, .20)
.01, .01
($.13, .15)
.23, .33
(.25, .42)
—
k, N 38, 19,915 34, 18,987 5, 1,788 35, 19,101 37, 17,410 43, 17,953 2, 974 2, 974 21, 14,850
Social support .26, .38
(.34, .42)
.24, .36
(.32, .40)
.17, .21
(.11, .31)
.27, .39
(.37, .41)
.18, .24
(.14, .34)
.22, .27
(.16, .37)
.22, .26
(.20, .32)
.08, .12
($.02, .26)
.34, .46
(.41, .51)
.31, .38
(.32, .44)
—
k, N 43, 42,668 10, 19,680 9, 3,339 9, 37,908 12, 4,444 13, 13,153 2, 974 6, 3,115 12, 40,181 5, 1,240
Interaction
outside the
organization
.10, .11
(.00, .22)
.15, .17
(.11, .23)
$.05, $.06
($.28, .17)
.22, .25
(.19, .31)
$.09, $.10
($.17, $.04)
.14, .15
(.07, .24)
.24, .27
(.22, .31)
.17, .19
(.13, .25)
$.01, $.01
($.12, .11)
.07, .08
($.04, .19)
.03, .04
($.01, .09)
—
k, N 4, 2,608 2, 974 3, 2,467 2, 974 2, 974 2, 974 3, 1,626 2, 974 3, 2,467 4, 1,296 3, 31,812
Physical demands $.17, $.22
($.28, $.17)
$.13, $.16
($.23, $.09)
.07, .07
(.01, .13)
$.01, $.01
($.05, .04)
.01, .01
($.07, .10)
$.04, $.04
($.12, .04)
$.19, $.21
($.26, $.16)
$.23, $.27
($.38, $.17)
.11, .13
($.01, .28)
.04, .05
($.05, .14)
$.06, $.08
($.11, $.05)
$.04, $.04
($.08, $.01)
—
k, N 8, 20,154 6, 18,449 3, 2,467 4, 1,848 4, 1,848 5, 2,222 3, 1,626 5, 2,042 5, 3,341 2, 974 7, 18,823 4, 3,119
Work conditions .08, .10
(.00, .20)
$.09, $.12
($.29, .06)
$.07, $.07
($.15, .01)
$.03, $.04
($.10, .03)
$.03, $.03
($.08, .02)
$.01, $.01
($.11, .09)
.20, .23
(.17, .29)
$.06, $.09
($.26, .09)
.08, .09
($.08, .26)
.06, .07
(.00, .13)
.13, .18
(.00, .36)
$.08, $.08
($.27, .10)
$.04, $.04
($.36, .27)
—
k, N 9, 4,659 5, 2,747 3, 2,467 4, 1,848 4, 1,848 6, 2391 2, 974 4, 1,848 5, 3,341 2, 974 8, 2,752 3, 2,467 6, 3,715
Note. r " uncorrected meta-analytic correlation; !"correlation corrected for unreliability; 95% CI " 95% confidence interval around !; k " number of studies for specific correlation; N " total
number of people in the studies for specific correlation.
1340
HUMPHREY, NAHRGANG, AND MORGESON
CI " –.27 # –.23 # –.18), feedback from the job (!"–.21; 95%
CI " –.33 # –.21 # –.08), and task identity (!"–.17; 95% CI "
–.23 # –.17 # –.11) were related to stress; and only task signif-
icance (!".38; 95% CI " .29 # .38 # .47) was related to
overload. In contrast, four characteristics (autonomy, skill variety,
task significance, and task identity) were negatively related to
burnout/exhaustion (mean !"–.26). Taken together, the results
demonstrated that these five motivational characteristics general-
ized to outcomes beyond the five specifically proposed, supporting
Hypotheses 2b and 2c.
Hypothesis 3 stated that experienced meaningfulness, experi-
enced responsibility, and knowledge of results would mediate the
relationships between (a) autonomy, (b) skill variety, (c) task
identity, (d) task significance, and (e) feedback from the job and
the behavioral and attitudinal outcomes. As noted by Baron and
Kenny (1986), mediation is a multistep process. First, the depen-
dent variables are regressed on the independent variables. Our tests
of Hypothesis 1 demonstrated those relationships that successfully
passed this step. In the second step, the mediators are regressed on
the independent variables. The results of this step can be found in
Table 2. For the five characteristics, there were moderate to large
relationships (ranging from !".22 to !".68) between these
characteristics and experienced meaningfulness, experienced re-
sponsibility, and knowledge of results.
The final step of the mediation test was to show that the direct
effect of the motivational characteristics on the outcomes was
reduced with the inclusion of the mediators. To perform this step,
we ran a series of regressions in which we regressed the outcome
of interest simultaneously on both the motivational characteristic
of interest and the predicted mediator. However, because of miss-
ing data between the critical psychological states and several
outcomes, we were only able to test mediation for three outcomes:
subjective performance, job satisfaction, and internal work moti-
vation. Table 3 presents the results of this step.
The first regression demonstrated that although the relationship
between autonomy and subjective performance only decreased
slightly with the inclusion of experienced responsibility (% de-
creased from .23 to .19), the R
2
decreased from .05 to .02, meaning
that autonomy explained less than half of the variance in subjective
performance when experienced responsibility was included. Expe-
rienced meaningfulness served to partially mediate task signifi-
cance (&%: .23 to .21; &R
2
: .05 to .02), task identity (&%: .17 to
.12; &R
2
: .03 to .01), and skill variety (&%: .07 to –.06; &R
2
: .01
to .00). Finally, knowledge of results did not mediate feedback
from the job. Table 3 shows that for both job satisfaction and
internal work motivation, autonomy was mediated by experienced
responsibility; and skill variety, task significance, and task identity
were mediated by experienced meaningfulness. In contrast, feed-
back from the job was only partially mediated by knowledge of
results. Taken together, there was strong support for the mediating
effect of experienced meaningfulness for skill variety, task signif-
icance, and task identity; partial support for the mediating effect of
experienced responsibility for autonomy; and no support for the
mediating effect of knowledge of results for feedback from the job.
Thus, Hypotheses 3a, 3b, 3c, and 3d were supported, whereas
Hypothesis 3e was not supported.
To test the alternative model suggested by Johns et al. (1992),
we compared the theorized model with one that allowed all three
critical states (i.e., experienced meaningfulness, experienced re-
sponsibility, and knowledge of results) to act as mediators. The
results of these analyses are presented in the right side of Table 3.
For subjective performance, the only meaningful difference was
the decrease in the beta of autonomy from .19 (mediated by only
experienced responsibility) to .07 (mediated by all three critical
psychological states). For job satisfaction and internal work mo-
tivation, the major difference was that feedback from the job was
fully mediated with both outcomes. These mediations can be
primarily attributed to experienced meaning, as its inclusion drove
the beta and R
2
values to zero. Thus, the results suggested that
experienced meaning was the “most critical” critical psychological
state, consistent with Johns et al.
Additional Motivational Characteristics
Hypothesis 4 stated that task variety would be (a) positively
related to the positive behavioral outcomes, (b) positively related
to the attitudinal outcomes, and (c) negatively related to the neg-
ative behavioral outcomes. As shown in Table 2, little research has
examined the relationships between task variety and a number of
outcomes (making it impossible to test Hypothesis 4c). Nonethe-
less, the limited research demonstrated several relationships. First,
although task variety was not related to objective performance
(!"–.02; 95% CI " –.14 # –.02 # .10), it was related to
subjective performance (!".23; 95% CI " .16 # .23 # .29),
providing support for Hypothesis 4a. Task variety did not relate to
role ambiguity (!"–.08; 95% CI " –.17 # –.08 # .01) or role
conflict (!".05; 95% CI " –.07 # .05 # .18) but did relate to
overload (!".38; 95% CI " .30 # .38 # .46). In addition, it
related to four satisfaction outcomes, including job satisfaction
(!".46; 95% CI " .35 # .46 # .56), supervisor satisfaction (!"
.31; 95% CI " .21 # .31 # .40), compensation satisfaction (!"
.19; 95% CI " .15 # .19 # .23), and promotion satisfaction (!"
.32; 95% CI " .27 # .32 # .37). Thus, the results provided some
support for Hypothesis 4b.
Hypothesis 5 stated that information processing, job complexity,
specialization, and problem solving would be (a) positively related
to positive behavioral outcomes, (b) positively related to positive
attitudinal outcomes and (c) negatively related to negative behav-
ioral outcomes. As noted in the Method section, there were not
enough studies examining specialization or problem solving to
include these characteristics in our meta-analytic review. More-
over, as shown in Table 2, we were only able to examine the
relationships for information processing and job complexity with a
limited number of outcomes. First, Table 2 shows that information
processing was related to job satisfaction (!".38; 95% CI "
.35 # .38 # .42), as was job complexity (!".37; 95% CI " .22 #
.37 # .52). In addition, job complexity was related to job involve-
ment (!".24; 95% CI " .04 # .24 # .45) and overload (!".59;
95% CI " .52 # .59 # .65), whereas it was not related to anxiety
(!".01; 95% CI " –.09 # .01 # .12). Thus, there was some
limited, preliminary support for Hypothesis 5b. However, because
of the low ns and ks for both work characteristics, it is important
to temper the interpretation of these results.
Hypothesis 6 stated that work scheduling autonomy, work meth-
ods autonomy, and decision-making autonomy would be (a) pos-
itively related to positive behavioral outcomes, (b) positively re-
lated to positive attitudinal outcomes, and (c) negatively related to
negative behavioral outcomes. As shown in Table 4, there have
1341
WORK DESIGN META-ANALYTIC SUMMARY AND EXTENSION
Table 2
Correlations Between Work Characteristics and Outcomes
Outcome
Autonomy
Skill
variety
Task
variety
Task
significance
Task
identity
Feedback
from
the job
Information
processing
Job
complexity Interdependence
Feedback
from
others
Social
support
Interaction
outside the
organization
Physical
demands
Work
conditions
r, !
(95% CI)
r, !
(95% CI)
r, !
(95% CI)
r, !
(95% CI)
r, !
(95% CI)
r, !
(95% CI)
r, !
(95% CI)
r, !
(95% CI)
r, !
(95% CI)
r, !
(95% CI)
r, !
(95% CI)
r, !
(95% CI)
r, !
(95% CI)
r, !
(95% CI)
Behavioral outcomes
Performance—
objective
.14, .17
(.04, .30)
$.03, $.03
($.14, .07)
$.02, $.02
($.14, .10)
.05, .06
($.02, .13)
.09, .09
($.04, .23)
$.06, $.07
($.57, .43)
k, N 9, 1,185 3, 344 5, 613 7, 760 8, 874 3, 557
Performance—
subjective
.18, .23
(.19, .28)
.06, .07
($.01, .15)
.21, .23
(.16, .29)
.16, .23
(.16, .29)
.13, .17
(.14, .19)
.14, .20
(.15, .24)
.32, .37
(.26, .48)
.14, .18
(.05, .31)
.22, .28
(.24, .33)
.09, .12
($.01, .25)
k, N 42, 7,886 26, 5,374 2, 918 20, 3,503 25, 8,055 26, 5,241 2, 268 8, 2,200 9, 1,584 5, 1,369
Absenteeism $.13, $.15
($.21, $.09)
$.06, $.07
($.15, .01)
.05, .06
($.04, .17)
$.08, $.09
($.17, $.01)
$.08, $.10
($.16, $.03)
$.03, $.03
($.17, .10)
$.06, $.09
($.22, .05)
$.06, $.09
($.14, $.03)
k, N 12, 2,972 11, 2,288 8, 1,706 10, 2,154 11, 2,211 2, 464 4, 853 4, 1,252
Turnover
intentions
.00, $.01
($.14, .13)
$.07, $.09
($.21, .04)
$.02, $.03
($.16, .11)
.00, .00
($.09, .08)
$.01, $.02
($.13, .10)
.04, .04
($.05, .13)
$.11, $.17
($.31, $.03)
$.22, $.34
($.46, $.23)
$.26, $.34
($.47, $.22)
k, N 21, 7,721 20, 7,549 17, 6,355 17, 6,355 20, 6,720 2, 464 5, 1,178 8, 1,453 9, 1,886
Role perceptions outcomes
Role ambiguity $.19, $.23
($.35, $.12)
$.06, $.08
($.25, .08)
$.07, $.08
($.17, .01)
$.03, $.03
($.30, .24)
$.09, $.09
($.33, .15)
$.36, $.43
($.59, $.28)
.01, .02
($.16, .19)
.03, .03
($.13, .19)
$.28, $.54
($1.00, .00)
$.25, $.32
($.42, $.21)
.00, .00
($.20, .20)
k, N 21, 8,186 7, 2,538 10, 3,167 7, 1,369 11, 2,873 14, 12,351 4, 1,863 4, 2,216 5, 1,350 15, 14,385 2, 1,333
Role conflict $.14, $.17
($.26, $.09)
.01, .02
($.13, .17)
.00, .05
($.07, .18)
.04, .06
($.11, .24)
$.12, $.17
($.22, $.12)
$.27, $.32
($.42, $.23)
$.03, $.03
($.13, .06)
$.24, $.31
($.36, $.26)
k, N 14, 5,400 7, 2,515 5, 978 7, 1,346 8, 2,036 8, 10,369 2, 415 11, 11,996
Well-being outcomes
Anxiety $.08, $.10
($.14, $.06)
.00, $.01
($.07, .06)
$.02, $.03
($.10, .05)
$.07, $.09
($.20, .01)
$.26, $.32
($.37, $.27)
.01, .01
($.09, .12)
$.19, $.23
($.27, $.19)
k, N 14, 4,414 5, 1,021 5, 1,021 5, 1,021 6, 9,470 3, 882 10, 12,676
Stress $.18, $.23
($.27, $.18)
$.10, $.14
($.32, .05)
.04, .05
($.01, .12)
$.13, $.17
($.23, $.11)
$.15, $.21
($.33, $.08)
$.06, $.09
($.17, $.02)
$.25, $.32
($.39, $.25)
$.22, $.26
($.36, $.16)
$.36, $.42
($.48, $.37)
k, N 13, 12,240 6, 1,673 4, 1,008 4, 1,008 6, 1,212 3, 1,098 7, 1,170 11, 7,946 2, 6,726
Burnout/
exhaustion
$.25, $.30
($.38, $.23)
.06, .07
($.30, .45)
$.24, $.29
($.43, $.16)
$.23, $.28
($.37, $.18)
$.09, $.10
($.48, .28)
$.14, $.17
($.29, $.06)
$.27, $.34
($.40, $.29)
.08, .10
($.17, .37)
k, N 14, 14,825 4, 1,789 2, 756 2, 756 3, 1,130 2, 322 18, 10,647 2, 1,273
Overload .02, .02
($.11, .14)
.33, .38
(.30, .46)
.32, .38
(.29, .47)
.50, .59
(.52, .65)
.08, .10
($.15, .35)
$.10, $.15
($.27, $.03)
k, N 7, 2,961 4, 992 3, 587 2, 1,076 4, 2,520 10, 3,377
1342
HUMPHREY, NAHRGANG, AND MORGESON
Table 2 (continued)
Outcome
Autonomy
Skill
variety
Task
variety
Task
significance
Task
identity
Feedback
from
the job
Information
processing
Job
complexity Interdependence
Feedback
from
others
Social
support
Interaction
outside the
organization
Physical
demands
Work
conditions
r, !
(95% CI)
r, !
(95% CI)
r, !
(95% CI)
r, !
(95% CI)
r, !
(95% CI)
r, !
(95% CI)
r, !
(95% CI)
r, !
(95% CI)
r, !
(95% CI)
r, !
(95% CI)
r, !
(95% CI)
r, !
(95% CI)
r, !
(95% CI)
r, !
(95% CI)
Attitudinal outcomes
Satisfaction-job .37, .48
(.45, .50)
.32, .42
(.38, .46)
.35, .46
(.35, .56)
.31, .41
(.39, .43)
.23, .31
(.29, .32)
.33, .43
(.41, .45)
.31, .38
(.35, .42)
.32, .37
(.22, .52)
.23, .33
(.30, .36)
.32, .42
(.38, .45)
.41, .56
(.52, .59)
.05, .06
(.03, .08)
$.15, $.17
($.26, $.08)
.20, .23
(.07, .39)
k, N 175, 75,364 111, 48,795 27, 8,480 108, 84,141 121, 49,973 126, 60,272 7, 2,490 13, 3,758 41, 53,993 39, 18,551 52, 91,109 5, 32,625 5, 5,201 8, 9,392
Satisfaction-growth .51, .69
(.64, .74)
.46, .61
(.57, .66)
.34, .49
(.43, .55)
.25, .35
(.31, .40)
.41, .55
(.50, .60)
.24, .33
(.29, .38)
.08, .11
($.01, .24)
.52, .78
(.66, .90)
k, N 32, 17,602 31, 15,941 29, 15,395 30, 15,603 31, 15,941 9, 9,370 7, 8,824 3, 1,987
Satisfaction-
supervisor
.30, .40
(.37, .43)
.16, .22
(.18, .25)
.25, .31
(.21, .40)
.18, .25
(.21, .29)
.16, .22
(.18, .26)
.31, .41
(.37, .44)
.14, .19
(.14, .24)
.37, .49
(.42, .56)
.51, .59
(.47, .71)
k, N 31, 20,157 21, 12,482 5, 2,631 19, 12,531 25, 14,246 25, 14,246 13, 11,694 14, 11,109 6, 6,327
Satisfaction-
coworker
.31, .47
(.42, .52)
.26, .39
(.36, .41)
.27, .43
(.40, .46)
.16, .26
(.22, .30)
.27, .41
(.37, .45)
.25, .41
(.35, .46)
.44, .65
(.52, .77)
.41, .64
(.50, .77)
k, N 17, 13,479 15, 11,052 13, 10,740 16, 11,818 16, 11,818 10, 11,071 7, 9,759 4, 2,753
Satisfaction-
compensation
.20, .27
(.23, .31)
.12, .16
(.10, .21)
.14, .19
(.15, .23)
.09, .13
(.10, .15)
.08, .12
(.06, .17)
.19, .26
(.23, .29)
.10, .16
(.13, .18)
.26, .33
(.29, .37)
.15, .24
(.21, .27)
k, N 18, 14,765 15, 11,191 3, 2,417 14, 46,876 16, 12,237 17, 12,406 11, 46,541 7, 9,512 6, 69,294
Satisfaction-
promotion
.13, .19
(.12, .26)
.09, .15
(.11, .19)
.23, .32
(.27, .37)
.08, .14
(.12, .15)
.12, .20
(.16, .24)
.22, .37
(.26, .48)
.08, .15
(.12, .17)
.25, .43
(.20, .66)
.12, .20
(.19, .21)
k, N 8, 3,559 6, 1,926 2, 1,633 5, 37,331 7, 2,692 7, 2,692 5, 37,894 3, 1,510 4, 36,795
Organizational
commitment
.30, .37
(.31, .43)
.23, .28
(.22, .35)
.34, .44
(.39, .48)
.18, .21
(.16, .27)
.29, .33
(.28, .39)
.34, .39
(.37, .40)
.56, .77
(.71, .82)
k, N 15, 6,420 9, 4,799 6, 39,463 9, 4,781 8, 4,665 3, 36,128 12, 69,313
Job involvement .23, .30
(.25, .35)
.24, .30
(.25, .36)
.26, .36
(.29, .43)
.14, .19
(.15, .24)
.20, .26
(.20, .32)
.20, .24
(.04, .45)
.16, .20
(.11, .30)
.13, .17
(.11, .24)
.16, .21
($.01, .42)
k, N 20, 6,502 18, 6,060 15, 5,197 19, 6,282 19, 6,282 2, 1,076 9, 2,585 4, 861 5, 1,493
Internal work
motivation
.27, .38
(.35, .42)
.30, .42
(.39, .46)
.30, .45
(.41, .50)
.17, .26
(.23, .28)
.29, .42
(.39, .46)
.21, .33
(.29, .37)
.22, .31
(.26, .35)
.11, .13
(.04, .23)
k, N 48, 20,835 47, 19,098 41, 18,362 44, 19,013 44, 19,013 13, 10,298 15, 10,186 12, 2,944
Critical psychological states
Experienced
meaning
.41, .60
(.56, .63)
.44, .62
(.60, .64)
.45, .68
(.64, .73)
.24, .37
(.33, .40)
.37, .53
(.47, .59)
.21, .32
(.28, .36)
.28, .38
(.33, .43)
k, N 22, 11,225 23, 11,274 24, 11,444 24, 11,444 22, 11,225 7, 8,824 7, 8,824
Knowledge
of results
.27, .40
(.37, .43)
.15, .22
(.20, .24)
.21, .33
(.31, .36)
.20, .31
(.28, .34)
.46, .67
(.61, .73)
.05, .08
(.05, .10)
.36, .50
(.42, .58)
k, N 22, 11,225 22, 11,225 22, 11,225 22, 11,225 23, 11,366 7, 8,824 7, 8,824
Responsibility .38, .58
(.53, .62)
.33, .49
(.45, .53)
.32, .51
(.47, .56)
.27, .43
(.39, .47)
.33, .49
(.43, .55)
.15, .24
(.21, .28)
.22, .32
(.28, .36)
k, N 23, 11,366 22, 11,225 22, 11,225 22, 11,225 22, 11,225 7, 8,824 7, 8,824
Note. r " uncorrected meta-analytic correlation; !"correlation corrected for unreliability; 95% CI " 95% confidence interval around !; k " number of studies for specific correlation; N " total
number of people in the studies for specific correlation.
1343
WORK DESIGN META-ANALYTIC SUMMARY AND EXTENSION
only been enough studies to summarize the relationships between
these three facets of autonomy and job satisfaction. Even so, this
table provides some interesting information. First, these three
facets of autonomy were fairly strongly correlated (mean !".67),
which was not surprising because they are considered subsets of
the broader autonomy construct. Second, these three facets of
autonomy demonstrated widely different relationships with job
satisfaction. That is, whereas the CI for work scheduling autonomy
included zero (!".11; 95% CI " .00 # .11 # .21), work methods
autonomy had a stronger relationship with job satisfaction (!"
.34; 95% CI " .30 # .34 # .38) and decision-making autonomy
demonstrated an even larger relationship with job satisfaction (!"
.58; 95% CI " .52 # .58 # .65). Moreover, the CIs for each facet
did not overlap, which suggested that the differences in magnitude
of these relationships were meaningfully different. Thus, these
results provided limited support for Hypothesis 6b.
Social Characteristics
Hypothesis 7 stated that interdependence, feedback from others,
social support, and interaction outside the organization would be
(a) positively related to positive behavioral outcomes, (b) posi-
tively related to positive attitudinal outcomes, and (c) negatively
related to negative behavioral outcomes. The results of our meta-
analyses investigating this hypothesis are presented in Table 2. As
shown in Table 2, no studies have investigated the relationship
between these social characteristics and objective performance.
Both interdependence (!".18; 95% CI " .05 # .18 # .31) and
Table 3
Mediation Tests for the Motivational Characteristics
Characteristic
Hypothesized mediator All three mediators
Performance—
subjective
Satisfaction—
job
Internal work
motivation
Performance—
subjective
Satisfaction—
job
Internal work
motivation
% R
2
% R
2
% R
2
% R
2
% R
2
% R
2
Autonomy
Unmediated .23 .05 .48 .23 .38 .14 .23 .05 .48 .23 .38 .14
Mediated .19 .02 .16 .02 $.14 .01 .07 .02 $.04 .00 $.20 .02
Skill variety
Unmediated .07 .01 .42 .18 .42 .18 .07 .01 .42 .18 .42 .18
Mediated $.06 .00 $.17 .02 $.09 .01 $.05 .00 $.16 .02 $.10 .01
Task significance
Unmediated .23 .05 .41 .17 .45 .20 .23 .05 .41 .17 .45 .20
Mediated .21 .02 $.31 .05 $.14 .01 .23 .03 $.33 .06 $.10 .01
Task identity
Unmediated .17 .03 .31 .10 .26 .07 .17 .03 .31 .10 .26 .07
Mediated .12 .01 $.01 .00 $.03 .00 .11 .01 $.07 .00 $.07 .00
Feedback from the job
Unmediated .20 .04 .43 .19 .42 .18 .20 .04 .43 .19 .42 .18
Mediated .22 .03 .39 .15 .39 .15 .18 .02 .02 .00 $.07 .00
Table 4
Correlations Between Autonomy Dimensions and Job Satisfaction
Variable
Work scheduling
autonomy
Work methods
autonomy
Decision-making
autonomy
r, !
(95% CI)
r, !
(95% CI)
r, !
(95% CI)
Work scheduling autonomy —
k, N
Work methods autonomy .56, .68
(.58, .78)
—
k, N 9, 1,987
Decision-making autonomy .53, .71
(.58, .84)
.53, .63
(.59, .66)
—
k, N 7, 1,412 8, 9,731
Satisfaction—job .09, .11
(.00, .21)
.29, .34
(.30, .38)
.50, .58
(.52, .65)
k, N 7, 1,294 12, 11,738 7, 9,008
Note. r " uncorrected meta-analytic correlation; !"correlation corrected for unreliability; 95% CI " 95%
confidence interval around !; k " number of studies for specific correlation; N " total number of people in the
studies for specific correlation.
1344
HUMPHREY, NAHRGANG, AND MORGESON
feedback from others (!".28; 95% CI " .24 # .28 # .33) were
related to subjective performance, whereas social support (!".12;
95% CI " –.01 # .12 # .25) only weakly was, supporting
Hypothesis 7a. Whereas social support (!"–.09; 95% CI "
–.14 # –.09 # –.03) was related to absenteeism, the 95% CI for
interdependence included zero (!"–.09; 95% CI " –.22 #
–.09 # .05). In contrast, interdependence (!"–.17; 95% CI "
–.31 # –.17 # –.03), feedback from others (!"–.34; 95% CI "
–.46 # –.34 # –.23), and social support (!"–.34; 95% CI "
–.47 # –.34 # –.22) were all related to turnover intentions,
providing support for Hypothesis 7c.
All four social characteristics were related to job satisfaction
(mean !".36). Interdependence, feedback from others, and social
support were all related to supervisor satisfaction (mean !".44),
coworker satisfaction (mean !".58), compensation satisfaction
(mean !".24), and promotion satisfaction (mean !".26).
However, for growth satisfaction, only the CIs for interdependence
(!".33; 95% CI " .29 # .33 # .38) and social support (!".78;
95% CI " .66 # .78 # .90) did not include zero, whereas the CI
for feedback from others (!".11; 95% CI " –.01 # .11 # .24)
included zero. Interdependence (!".39; 95% CI " .37 # .39 #
.40) and social support (!".77; 95% CI " .71 # .77 # .82) both
were related to organizational commitment; interdependence (!"
.20; 95% CI " .11 # .20 # .30) and feedback from others (!"
.17; 95% CI " .11 # .17 # .24) were related to job involvement;
and interdependence, feedback from others, and social support
(mean !".26) were all related to internal work motivation.
Finally, we examined the relationships between these social
characteristics and both role perception outcomes and well-being
outcomes. As shown in Table 2, only social support (!"–.32;
95% CI " –.42 # –.32 # –.21) was related to role ambiguity, role
conflict (!"–.31; 95% CI " –.36 # –.31 # –.26), anxiety (!"
–.23; 95% CI " –.27 # –.23 # –.19), and overload (!"–.15;
95% CI " –.27 # –.15 # –.03). All three characteristics were
negatively related to stress (mean !"–.22), whereas only feed-
back from others (!"–.17; 95% CI " –.29 # –.17 # –.06) and
social support (!"–.34; 95% CI " –.40 # –.34 # –.29) were
related to burnout/exhaustion. Taken together, there was partial
support for Hypothesis 7b.
Work Context Characteristics
Hypothesis 8 stated that physical demands would be (a) nega-
tively related to positive behavioral outcomes, (b) negatively re-
lated to positive attitudinal outcomes, and (c) positively related to
negative behavioral outcomes. As shown in Table 2, we only had
enough studies to meta-analyze the relationship between physical
demands and job satisfaction. Physical demands was negatively
related to job satisfaction (!"–.17; 95% CI " –.26 # –.17 #
–.08), providing limited support for Hypothesis 8b.
Hypothesis 9 stated that work conditions and ergonomics would
be (a) positively related to positive behavioral outcomes, (b) pos-
itively related to positive attitudinal outcomes, and (c) negatively
relate to negative behavioral outcomes. As noted in the Method
section, there were not enough studies examining ergonomics to
include this characteristic in our meta-analytic review. As shown
in Table 2, work conditions was positively related to job satisfac-
tion (!".23; 95% CI " .07 # .23 # .39) and negatively related
to stress (!"–.42; 95% CI " –.48 # –.42 # –.37). However, the
95% CIs for both role ambiguity (!".00; 95% CI " –.20 # .00 #
.20) and burnout/exhaustion (!".10; 95% CI " –.17 # .10 # .37)
included zero. Thus, there was limited support for Hypothesis 9b.
Incremental Contribution of Social and Work Context
Characteristics
Hypothesis 10 stated that (a) social characteristics and (b) work
context characteristics would explain unique variance in the be-
havioral and attitudinal outcomes above and beyond motivational
characteristics. To test this hypothesis, we conducted several re-
gressions in which the set of motivational characteristics was
entered in the first step, the set of social characteristics was entered
in the second step, and the set of work context characteristics was
entered in the third step. Results of these regressions are presented
in Table 5.
First, as shown in Table 5, three social characteristics explained
an additional 9% of the variance in subjective performance above
and beyond the 25% of the variance explained by the seven
motivational characteristics. Although the set of social character-
istics only explained an additional 2% of the variance in absen-
teeism, it explained an additional 24% of the variance in turnover
intentions. Thus, the set of social characteristics explained a large
amount of variance beyond the motivational characteristics for two
behavioral outcomes.
Second, the social characteristics explained an additional 40%
of the variance in organizational commitment and 44% of the
variance in job involvement. In contrast, they only explained an
additional 2% of the variance in internal work motivation. In
addition, the social characteristics explained, on average, an addi-
tional 24% of the variance in the six satisfaction outcomes,
whereas the two work context characteristics accounted for an
additional 4% of the variance in job satisfaction beyond the 51%
of the variance explained by the 12 motivational and social char-
acteristics.
Third, the social characteristics explained an additional 24% of
the variance in role ambiguity and 9% of the variance in role
conflict, whereas work conditions did not explain any additional
variance role ambiguity. Turning to the well-being outcomes, the
social characteristics explained an additional 6% of the variance in
anxiety, 9% of the variance in stress, 4% of the variance in
burnout/exhaustion, and 10% of the variance in overload. In ad-
dition, work conditions explained an additional 16% of the vari-
ance in stress and 2% of the variance in burnout/exhaustion. In
sum, the social characteristics explained a considerable amount of
variance beyond the motivational characteristics, supporting Hy-
pothesis 10a. In addition, although there were only limited empir-
ical data for the work context characteristics, the results of our
hierarchical regressions suggested that work context characteris-
tics could explain some additional variance in work outcomes
beyond either motivational or social characteristics, providing
limited support for Hypothesis 10b.
Discussion
Goals of the Meta-Analytic Review
Meta-analytic techniques are uniquely suited for summarizing
and clarifying past research, testing new hypotheses at a qualita-
1345
WORK DESIGN META-ANALYTIC SUMMARY AND EXTENSION
tively different level, and advancing theory (Hunter & Schmidt,
2004). Although work design research has slowed in the I/O
psychology and management fields during the past 20 years, it is
important that researchers continue to investigate this topic, as the
design of work has a profound effect on employees’ behavior,
attitudes, and well-being (Campion, Mumford, Morgeson, & Nahr-
gang, 2005). More than 34% of the variance in performance and
more than 55% of the variance in job satisfaction was a function of
the 14 work characteristics investigated herein. Thus, we had three
goals for our meta-analytic review.
First, we were interested in replicating and extending Fried and
Ferris’s (1987) meta-analytic summary of the work design litera-
ture. To accomplish this goal, we meta-analytically summarized
259 studies, compared to Fried and Ferris’s 76 studies. This
allowed us to test 19 outcomes (vs. their 5) in our meta-analytic
review. The large sample sizes for several relationships (e.g.,
75,364 respondents and 175 studies for the autonomy–job satis-
faction relationship) provided highly stable estimates of the true
population correlation. Thus, we have provided the best estimates
to date for the relationships studied herein.
In addition, we were able to provide the first meta-analytic test
of the job characteristics– critical psychological states– outcomes
mediation model. Whereas Fried and Ferris (1987) were able to
examine only bivariate correlations between the motivational char-
acteristics, mediation processes, and work outcomes, we utilized
Baron and Kenny’s (1986) multistep mediation process. The re-
sults suggest a modified mediation model for the motivational
characteristics in which the primary mediator of the motivational
characteristics–work outcome relationships is experienced mean-
ing. Its inclusion in the mediation model led to the greatest level of
mediation.
Perhaps it should not be surprising that experienced meaning-
fulness is the best mediator of the relationships between motiva-
tional characteristics and work outcomes. Three motivational char-
acteristics (skill variety, task identity, and task significance) have
been hypothesized to impact work outcomes through experienced
meaningfulness (Hackman & Oldham, 1976).
Other psychological research helps us understand why the other
two motivational characteristics (autonomy and feedback from the
job) impact work outcomes through experienced meaningfulness.
The research suggests that the ultimate goal of human beings is to
pursue meaning in our work and nonwork lives (Ryan & Deci,
2001), as experiencing meaning has been found to promote well-
being and happiness (King & Napa, 1998; Zika & Chamberlin,
1992). Authors have suggested that experienced meaning is the
critical mediator between life events and positive outcomes
(Fredrickson, 2003) and that promoting intrinsic motivation is
central to helping people achieve this meaning (Deci & Ryan,
2000). Therefore, all of the motivational characteristics, which are
theoretically and empirically linked to internal work motivation,
should be expected to promote meaning.
Table 5
Incremental Regression Results for Outcomes
Outcome
Motivational
characteristics
Social
characteristics
Work context
characteristics
Total R
2
Step 1 Step 2 Step 3
R
2
&R
2
&R
2
Behavioral outcomes
Performance-objective .08 0.08
Performance-subjective .25 .09 0.34
Absenteeism .06 .02 0.07
Turnover intentions .02 .24 0.26
Attitudinal outcomes
Satisfaction—job .34 .17 .04 0.55
Satisfaction—supervisor .25 .29 0.55
Satisfaction—coworker .29 .33 0.62
Satisfaction—compensation .11 .07 0.18
Satisfaction—growth .56 .44 1.00
Satisfaction—promotion .21 .15 0.35
Organizational commitment .24 .40 0.64
Job involvement .43 .44 0.87
Internal work motivation .27 .02 0.29
Role perception outcomes
Role ambiguity .29 .25 .00 0.54
Role conflict .22 .11 0.33
Well-being outcomes
Anxiety .15 .06 0.20
Stress .14 .09 .16 0.38
Burnout/exhaustion .17 .04 .02 0.23
Overload .54 .10 0.64
Note. Motivational characteristics are autonomy, skill variety, task variety, task significance, task identity,
feedback from the job, information processing, and job complexity. Social characteristics are interdependence,
feedback from others, social support, and interaction outside the organization. Work context characteristics are
physical demands and work conditions. Only those work characteristics that appear with population correlations
in Table 2 are included in specific regressions.
1346
HUMPHREY, NAHRGANG, AND MORGESON
People also find work meaningful if they are able to pursue
cherished goals (Klinger, 1977; Ryff & Singer, 1998). Thus, if one
views the development of meaning as a process of setting and
pursuing important goals, having both autonomy and feedback
from the job is essential. Successful goal completion requires that
employees have flexibility in how goals are pursued (Locke &
Latham, 1990). If employees have autonomy in the decision-
making process leading to goal completion, they will have higher
levels of experienced meaning (Maddi, 1970). More directly, nu-
merous studies have shown that autonomy is critical for creating
self-determination and meaning (Deci & Ryan, 2000). In addition,
it is crucial that employees receive feedback on the progress
toward goal accomplishment (Locke & Latham, 1990). Feedback
from the job provides an opportunity for employees to learn about
their performance level and proximity to their goal. If employees
are successfully moving toward goal accomplishment, experienced
meaning will be enhanced. If employees learn that they are not
moving toward goal accomplishment, having the ability to change
their behavior (i.e., autonomy) will allow them the ability to find
different paths toward goal accomplishment. Thus, having auton-
omy and feedback from the job should promote experienced mean-
ing and positive work outcomes.
Second, we were interested in extending the job characteristics
model by including additional motivational characteristics. In our
theoretical model, we added five additional motivational charac-
teristics to the five motivational characteristics proposed in the job
characteristics model. The results demonstrate that task variety,
job complexity, and information processing impact a variety of
work outcomes. In particular, all three characteristics demon-
strated large relationships with job satisfaction, and both job com-
plexity and task variety were strongly related to overload. How-
ever, our meta-analytic review also highlights the fact that a large
number of relationships (e.g., specialization, problem solving, and
information processing) have not been sufficiently studied. There
also has been limited research examining either task variety or job
complexity. Future research should be conducted to investigate the
impact of these characteristics on work outcomes.
Third, we were also interested in extending the job characteris-
tics model by integrating social and work context characteristics
into a broader theoretical model and meta-analytically estimating
their effects. Our results suggest that the four social and two work
context characteristics have comparable relationships with many
of the same work outcomes as the motivational characteristics and
predict some outcomes that are not predicted by the motivational
characteristics. Moreover, our hierarchical regression analyses
provide evidence for the incremental impact of social and work
context characteristics above and beyond the eight motivational
characteristics. This is made even more remarkable because our
method was a conservative test of this hypothesis. That is, by
entering the set of motivational characteristics in the first step of
the regression, all shared variance between the social, work con-
text, and motivational characteristics was attributed to the eight
motivational characteristics.
Several findings are particularly noteworthy. For example, the
set of social characteristics was strongly related to turnover inten-
tions (&R
2
" .24), whereas the set of motivational characteristics
demonstrated almost no relationship (R
2
" .02) with turnover
intentions. Similarly, whereas the motivational characteristics
were strongly related to organizational commitment (R
2
" .24),
three social characteristics demonstrated even stronger incremental
relationships with it (&R
2
" .40). In contrast, the set of motiva
-
tional characteristics was strongly related to internal work moti-
vation (R
2
" .27), whereas the set of social characteristics dem
-
onstrated almost no unique relationship with it (&R
2
" .02). In
addition, the set of motivational characteristics was strongly re-
lated to overload (R
2
" .54), whereas the set of social character
-
istics explained a comparatively smaller amount of unique vari-
ance in it (&R
2
" .10). For other outcomes, the set of social
characteristics explained an equivalent amount of variance as the
set of motivational characteristics (e.g., supervisor satisfaction,
coworker satisfaction, job involvement, role ambiguity, and
stress). These findings highlight the differing impact of motiva-
tional and social characteristics. In particular, this suggests that
social characteristics provide a unique perspective on work design
beyond motivational characteristics.
In addition, although there were fewer studies examining the
work context characteristics, it is important to note two findings.
First, work conditions explained an incremental 16% of the vari-
ance in stress. The variance explained by this single work charac-
teristic was larger than the variance explained by either the five
motivational characteristics or three social characteristics, high-
lighting the power of work conditions to influence work outcomes.
Second, it is important to note that the two work context charac-
teristics explained unique variance (&R
2
" .04) in job satisfaction
above the variance explained by the eight motivational (R
2
" .34)
and four social (&R
2
" .17) characteristics. Although this is not a
large amount of variance compared to the variance explained by
the 12 other characteristics, it does suggest that attitudinal out-
comes can be influenced by work context conditions.
Future Directions
Our review suggests several future directions for research. First,
although the theories of work design reside at the job level, the
studies of work design have been conducted at the individual level.
In fact, of the 677 studies examined for inclusion in the meta-
analytic review, only 8 provided job-level data. In addition, many
studies were conducted with a limited number of jobs. For exam-
ple, 70 of the 259 studies reported sampling only one job, and an
additional 29 reported sampling only two to four jobs, and almost
50% of the total studies did not report the number of jobs sampled.
This means that the restriction in range within studies (due to the
limited number of jobs) may have reduced the observed correla-
tions between work characteristics and outcomes, producing low-
ered estimates of population correlations. Clearly, future research
should investigate these relationships at the job level.
Second, most research on work design has been conducted such
that employees evaluated both the work characteristics and per-
ceptual outcomes. With the exception of relationships with several
behavioral outcomes, this means that the data likely suffer from
common-source biases that inflate the relationships between con-
structs (Crampton & Wagner, 1994; Podsakoff, MacKenzie, Lee,
& Podsakoff, 2003; Roberts & Glick, 1981). Meta-analyzing data
does not remove these flaws. Moreover, even though we coded for
common-source issues, we were not able to adjust the population
correlation estimates, because very few studies had independent
sources of data. One way to address this problem is to conduct
1347
WORK DESIGN META-ANALYTIC SUMMARY AND EXTENSION
job-level analyses in which respondents evaluate either the work
characteristics or work outcomes, but not both.
Third, although we worked to be as comprehensive as possible
in our summary of the work design literature, we were limited by
the available research. This was evident in the almost complete
lack of research examining specialization, problem solving, or
ergonomics. In several cases, the relationship between a charac-
teristic and an outcome was estimated with a low number of both
studies (k) and job incumbents (n). For example, there were only
three studies (n " 587) that examined the relationship between
task significance and overload. For any relationship in our meta-
analytic review with such a small k and n, it is important to
recognize that the results are only the best approximations of
population coefficients to date. Future research is necessary to
increase the quality of estimates of these population coefficients.
Fourth, as the focus of our research was on how individual job
incumbents reacted to the characteristics of work, rather than on
how teams reacted to their design, our study excluded team-level
research. Nonetheless, it is important to note that several research-
ers have begun to utilize work design characteristics in the inves-
tigation of team-level phenomena (e.g., Bailey, 1998; Janz,
Colquitt, & Noe, 1997). However, research to date has examined
only a subset of the work characteristics discussed herein, primar-
ily focusing on autonomy and interdependence. Future research is
needed to determine the impact of the work characteristics pre-
sented in this meta-analytic review on team-level work design.
Moreover, as moving to the team level creates unique constructs
that may not exist at the individual level (Morgeson & Hofmann,
1999), future research should investigate how characteristics of the
team such as trust and cooperative norms relate to the work
characteristics discussed herein.
Fifth, the nature of work has been marked by dramatic techno-
logical changes, increased competition, and workforce composi-
tion changes (Howard, 1995a; Morgeson & Campion, 2003;
Parker, Wall, & Cordery, 2001). Although some have suggested
that models of work design should expand to address these
changes, there has been no direct investigation of the impact of
these changes. Yet it is clear that work has become more cogni-
tively demanding and complex due to increased technology, in-
creased skill variety, and a shift to knowledge-based work
(Howard, 1995b; Parker & Wall, 2001). In addition, uncertainty
has increased as a result of global competition and changes in
employment contracts, whereas increasingly flexible technologies
have added to operational uncertainty because of the growing
variability and complexity of many work processes (Parker &
Wall, 2001). Work has become increasingly interdependent
through new production technologies and the use of team-based
designs such that workers now have new roles and relationships
(Howard, 1995b; Ilgen, 1999). Finally, certain jobs have different
consequences of failure. For example, an error in the health care
industry may result in an injury or death to the patient. In contrast,
an error in a customer service context may result in a dissatisfied
customer, an unpleasant but not life-threatening outcome. These
changes, taken together, suggest that future research should ac-
tively investigate how changes in work and differences across
industries impact the relationships between work characteristics
and outcomes.
Sixth, our implicit assumption is that the relationship between
work characteristics and well-being outcomes is linear. However,
some research has suggested that the true relationship may be
nonlinear. For example, Xie and Johns (1995) found that both high
and low levels of job complexity were related to high levels of
exhaustion, whereas moderate levels of complexity were related to
low levels of exhaustion. To date there has only been limited
research examining curvilinear relationships between work char-
acteristics and outcomes. Yet these results suggest the need to
examine more complex relationships.
Finally, our results provide clear evidence that there are multiple
options for redesigning work to achieve certain work outcomes.
For example, the two best predictors of job satisfaction were
autonomy and social support. If an organization were interested in
improving job satisfaction, it could improve either job autonomy
or social support. However, as shown by Morgeson and Humphrey
(2006), increasing autonomy can increase compensation and train-
ing requirements, whereas increasing social support does not have
these negative tradeoffs. Thus, organizations may benefit by uti-
lizing the results of our meta-analysis in their work redesign
process to pinpoint those work characteristics that maximize the
outcomes they are interested in and minimize the negative impact
on other desirable outcomes.
Conclusion
This meta-analytic review of the work design literature inte-
grated motivational, social, and work context characteristics. We
reviewed 677 articles, coded 259 empirical articles, and meta-
analyzed 6,333 correlations to examine 276 relationships. Our
results demonstrate that work design has a large impact on worker
attitudes and behaviors, explaining on average 43% of the variance
in these outcomes. Our results also suggest weaknesses in work
design theory and empirical research, indicating areas in need of
future research. Due to the importance and impact of work design,
we hope that our meta-analytic review helps stimulate future
research and reinvigorates the work design literature within the I/O
psychology and management domains.
References
References marked with an asterisk indicate studies included in the meta-
analysis.
*Abdel-Halim, A. A. (1978). Employee affective responses to organiza-
tional stress: Moderating effects of job characteristics. Personnel Psy-
chology, 31, 561–579.
*Abdel-Halim, A. A. (1979). Individual and interpersonal moderators of
employee reactions to job characteristics: A reexamination. Personnel
Psychology, 32, 121–137.
*Abdel-Halim, A. A. (1980). Effects of higher-order need strength on the
job-performance–job-satisfaction relationship. Personnel Psychology,
33, 335–347.
*Abdel-Halim, A. A. (1981). Effects of role stress job-design technology
interaction on employee work satisfaction. Academy of Management
Journal, 24, 260 –273.
Adler, P. S., & Kwon, S. (2002). Social capital: Prospects for a new
concept. Academy of Management Review, 27, 17– 40.
*Agho, A. O. (1993). The moderating effects of dispositional affectivity on
relationships between job characteristics and nurses’ job-satisfaction.
Research in Nursing & Health, 16, 451– 458.
*Agho, A. O., Mueller, C. W., & Price, J. L. (1993). Determinants of
1348
HUMPHREY, NAHRGANG, AND MORGESON
employee job-satisfaction: An empirical-test of a causal model. Human
Relations, 46, 1007–1027.
*Aldag, R. J., & Brief, A. P. (1975). Impact of individual differences on
employee affective responses to task characteristics. Journal of Business
Research, 3, 311–322.
Alderfer, C. P., & Smith, K. K. (1982). Studying intergroup relations
embedded in organizations. Administrative Science Quarterly, 27, 35–
65.
*Arnold, H. J., & House, R. J. (1980). Methodological and substantive
extensions to the job characteristics model of motivation. Organiza-
tional Behavior and Human Performance, 25, 161–183.
Arthur, W., Jr., Edwards, B. D., Bell, S. T., Villado, A. J., & Bennett, W.
(2005). Team task analysis: Identifying tasks and jobs that are team
based. Human Factors, 47, 654 – 669.
Arthur, W., Jr., Villado, A. J., & Bennett, W., Jr. (in press). Innovations in
team task analysis: Identifying task elements, tasks, and jobs that are
team-based. In W. Bennett, Jr. (Ed.), The future of job analysis. Mah-
wah, NJ: Erlbaum.
*Ashforth, B. E., Saks, A. M., & Lee, R. T. (1998). Socialization and
newcomer adjustment: The role of organizational context. Human Re-
lations, 51, 897–926.
*Axtell, C., Wall, T., Stride, C., Pepper, K., Clegg, C., Gardner, P., &
Bolden, R. (2002). Familiarity breeds content: The impact of exposure to
change on employee openness and well-being. Journal of Occupational
and Organizational Psychology, 75, 217–231.
*Baba, V. V., & Jamal, M. (1991). Routinization of job context and job
content as related to employees’ quality of working life: A study of
Canadian nurses. Journal of Organizational Behavior, 12, 379 –386.
Babbage, C. (1835). On the economy of machinery and manufactures.
London: Knight.
Bailey, D. E. (1998). Comparison of manufacturing performance of three
team structures in semiconductor plants. IEEE Transactions on Engi-
neering Management, 45, 20 –32.
*Baird, L. S. (1976). Relationship of performance to satisfaction in stim-
ulating and nonstimulating jobs. Journal of Applied Psychology, 61,
721–727.
*Baker, W. E. (1978). Certain dimensions of job content and their rela-
tionship to job satisfaction. Dissertation Abstracts International, 39(5),
2551B.
*Bakker, A. B., Demerouti, E., & Verbeke, W. (2004). Using the job
demands-resources model to predict burnout and performance. Human
Resource Management, 43, 83–104.
*Barnabe, C., & Burns, M. (1994). Teachers’ job characteristics and
motivation. Educational Research, 36, 171–185.
*Barnett, R. C., & Brennan, R. T. (1995). The relationship between job
experiences and psychological distress: A structural equation approach.
Journal of Organizational Behavior, 16, 259 –276.
Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable
distinction in social psychological research: Conceptual, strategic, and
statistical considerations. Journal of Personality & Social Psychology,
51, 1173–1182.
*Barrick, M. R., & Mount, M. K. (1993). Autonomy as a moderator of the
relationships between the big five personality dimensions and job-per-
formance. Journal of Applied Psychology, 78, 111–118.
*Basu, R., & Green, S. G. (1997). Leader-member exchange and transfor-
mational leadership: An empirical examination of innovative behaviors
in leader-member dyads. Journal of Applied Social Psychology, 27,
477– 499.
*Bateman, T. S., & Strasser, S. (1984). A longitudinal analysis of the
antecedents of organizational commitment. Academy of Management
Journal, 27, 95–112.
Baumeister, R. F., & Leary, M. R. (1995). The need to belong: Desire for
interpersonal attachments as a fundamental human motivation. Psycho-
logical Bulletin, 117, 497–529.
*Beaton, R. D., Murphy, S. A., Pike, K. C., & Corneil, W. (1997). Social
support and network conflict in firefighters and paramedics. Western
Journal of Nursing Research, 19, 297–313.
*Becherer, R. C., Morgan, F. W., & Richard, L. M. (1982). The job
characteristics of industrial salespersons: Relationship to motivation and
satisfaction. Journal of Marketing, 46(4), 125–135.
Bechtel, W. (1988). Philosophy of science: An overview for cognitive
science. Hillsdale, NJ: Erlbaum.
*Bechtold, S. E., Sims, H. P., & Szilagyi, A. D. (1981). Job scope
relationships: A three-wave longitudinal analysis. Journal of Occupa-
tional Behaviour, 2, 189 –202.
*Beehr, T. A. (1976). Perceived situational moderators of relationship
between subjective role ambiguity and role strain. Journal of Applied
Psychology, 61, 35– 40.
*Beehr, T. A., & Drexler, J. A. (1986). Social support, autonomy, and
hierarchical level as moderators of the role characteristics-outcome
relationship. Journal of Occupational Behaviour, 7, 207–214.
*Beehr, T. A., Jex, S. M., Stacy, B. A., & Murray, M. A. (2000). Work
stressors and coworker support as predictors of individual strain and job
performance. Journal of Organizational Behavior, 21, 391– 405.
*Bell, S. J., & Menguc, B. (2002). The employee-organization relationship,
organizational citizenship behaviors, and superior service quality. Jour-
nal of Retailing, 78, 131–146.
*Bennett, P., Lowe, R., Matthews, V., Dourali, M., & Tattersall, A. (2001).
Stress in nurses, coping, managerial support and work demand. Stress
and Health, 17, 55– 63.
Berman, S. L., Down, J., & Hill, C. W. L. (2002). Tacit knowledge as a
source of competitive advantage in the National Basketball Association.
Academy of Management Journal, 45, 13–31.
Biddle, B. J. (1979). Role theory: Expectations, identities, and behaviors.
New York: Academic Press.
*Birnbaum, P. H., Farh, J. L., & Wong, G. Y. Y. (1986). The job
characteristics model in Hong Kong. Journal of Applied Psychology, 71,
598 –605.
*Birnstihl, E. N. (1988). The influence of growth needs and psychological
states on the relationship of job dimensions and job performance of
service related educators. Dissertation Abstracts International, 49(4),
696A.
*Bischoff, S. J. (1995). Job dimensions and health risks: Extending
Karasek’s model of job strain. Dissertation Abstracts International,
55(11), 4792B.
*Block, L. K., & Stokes, G. S. (1989). Performance and satisfaction in
private versus nonprivate work settings. Environment and Behavior, 21,
277–297.
*Bottger, P. C., & Chew, I. K.-H. (1986). The job characteristics model and
growth satisfaction: Main effects of assimilation of work experience and
context satisfaction. Human Relations, 39, 575–594.
*Bowman, S. H. (1984). Relationships of work context perceptions and
individual and positional characteristics to work satisfaction and perfor-
mance. Dissertation Abstracts International, 44(8), 2587B.
*Bradley, J. R., & Cartwright, S. (2002). Social support, job stress, health,
and job satisfaction among nurses in the United Kingdom. International
Journal of Stress Management, 9, 163–182.
*Brass, D. J. (1980). Effects of relationships among task positions on job
characteristics, interpersonal variables, and employee satisfaction and
performance. Dissertation Abstracts International, 40(10), 5498A.
*Brass, D. J. (1981). Structural relationships, job characteristics, and
worker satisfaction and performance. Administrative Science Quarterly,
26, 331–348.
*Breaugh, J. A. (1985). The measurement of work autonomy. Human
Relations, 38, 551–570.
*Breaugh, J. A. (1998). The development of a new measure of global work
autonomy. Educational and Psychological Measurement, 58, 119 –128.
1349
WORK DESIGN META-ANALYTIC SUMMARY AND EXTENSION
*Breaugh, J. A. (1999). Further investigation of the work autonomy scales:
Two studies. Journal of Business and Psychology, 13, 357–373.
*Brief, A. P., & Aldag, R. J. (1975). Employee reactions to job character-
istics: A constructive replication. Journal of Applied Psychology, 60,
182–186.
*Brief, A. P., & Aldag, R. J. (1976). Correlates of role indices. Journal of
Applied Psychology, 61, 468 –472.
*Brief, A. P., & Aldag, R. J. (1978). The Job Characteristics Inventory: An
examination. Academy of Management Journal, 21, 659 – 670.
*Brief, A. P., Burke, M. J., Robinson, B. S., George, J. M., & Webster, J.
(1988). Should negative affectivity remain an unmeasured variable in the
study of job stress? Journal of Applied Psychology, 73, 193–198.
*Brief, A. P., Munro, J., & Aldag, R. J. (1976). Correctional employees’
reactions to job characteristics: Data based argument for job enlarge-
ment. Journal of Criminal Justice, 4, 223–230.
*Burr, R., & Cordery, J. L. (2001). Self-management efficacy as a mediator
of the relation between job design and employee motivation. Human
Performance, 14, 27– 44.
*Cahill, J., & Landsbergis, P. A. (1996). Job strain among post office
mailhandlers. International Journal of Health Services, 26, 731–750.
*Caldwell, D. F., & O’Reilly, C. A. (1982). Task perceptions and job
satisfaction: A question of causality. Journal of Applied Psychology, 67,
361–369.
Campion, M. A. (1988). Interdisciplinary approaches to job design: A
constructive replication with extensions. Journal of Applied Psychology,
73, 467– 481.
*Campion, M. A., & McClelland, C. L. (1991). Interdisciplinary exami-
nation of the costs and benefits of enlarged jobs: A job design quasi-
experiment. Journal of Applied Psychology, 76, 186 –198.
*Campion, M. A., & McClelland, C. L. (1993). Follow-up and extension of
the interdisciplinary costs and benefits of enlarged jobs. Journal of
Applied Psychology, 78, 339 –351.
Campion, M. A., Mumford, T. V., Morgeson, F. P., & Nahrgang, J. D.
(2005). Work redesign: Eight obstacles and opportunities. Human Re-
source Management, 44, 367–390.
Campion, M. A., & Thayer, P. W. (1985). Development and field evalu-
ation of an interdisciplinary measure of job design. Journal of Applied
Psychology, 70, 29 – 43.
*Cappelli, P., & Rogovsky, N. (1998). Employee involvement and orga-
nizational citizenship: Implications for labor law reform and “lean
production.” Industrial & Labor Relations Review, 51, 633– 653.
*Carayon, P., Yang, C. L., & Lim, S. Y. (1995). Examining the relationship
between job design and worker strain over time in a sample of office
workers. Ergonomics, 38, 1199 –1211.
*Carayon, P., & Zijlstra, F. (1999). Relationship between job control, work
pressure and strain: Studies in the USA and in The Netherlands. Work
and Stress, 13, 32– 48.
*Cellar, D. F., Kernan, M. C., & Barrett, G. V. (1985). Conventional
wisdom and ratings of job characteristics: Can observers be objective?
Journal of Management, 11, 131–138.
*Chalykoff, J., & Kochan, T. A. (1989). Computer-aided monitoring: Its
influence on employee job-satisfaction and turnover. Personnel Psychol-
ogy, 42, 807– 834.
*Champoux, J. E. (1980). A three sample test of some extensions to the job
characteristics model of work motivation. Academy of Management
Journal, 23, 466 – 478.
*Champoux, J. E. (1991). A multivariate test of the job characteristics
theory of work motivation. Journal of Organizational Behavior, 12,
431– 446.
*Champoux, J. E. (1992). A multivariate analysis of curvilinear relation-
ships among job scope, work context satisfactions, and affective out-
comes. Human Relations, 45, 87–111.
*Chen, Y. (1991). Effects of objective job stressors, job satisfaction,
negative mood, and negative affectivity on perceived job conditions.
Dissertation Abstracts International, 52(3), 1760B.
*Chu, C. I., Hsu, H. M., Price, J. L., & Lee, J. Y. (2003). Job satisfaction
of hospital nurses: An empirical test of a causal model in Taiwan.
International Nursing Review, 50, 176 –182.
*Cleave, S. (1993). A test of the job characteristics model with adminis-
trative positions in physical-education and sport. Journal of Sport Man-
agement, 7, 228 –242.
Cohen, S., & Wills, T. A. (1985). Stress, social support, and the buffering
hypothesis. Psychological Bulletin, 98, 310 –357.
*Cordery, J. L., & Sevastos, P. P. (1993). Responses to the original and
revised Job Diagnostic Survey: Is education a factor in responses to
negatively worded items? Journal of Applied Psychology, 78, 141–143.
*Coyle, Y. M., Aday, L. A., Battles, J. B., & Hynan, L. S.