ORGANIZATIONAL BEHAVIOR AND HUMAN PERFORMANCE 32, 160-177 (1983)

Organizational Determinants of Job Stress

DONALD E P A R K E R
The University of Michigan

AND

THOMAS A . D E C O T I I S

University of South Carolina

A model of job stress that focuses on organizational and job-related stress

is presented. Job stress is conceived of as a first-level outcome of the orga-
nization and job; it is a feeling of discomfort that is separate and distinct from
second-level outcomes or consequences of job stress. The second-level out-
comes may include varying levels of satisfaction, organizational commitment,
motivation, and performance. A partial test of the model examines relation-
ships between hypothesized stressors and experienced job stress. Survey data
obtained from 367 managers of a large restaurant chain were used with the
results generally supporting the model. Factor analysis supported the concept
that job stress is multidimensional. Two distinct dimensions of job stress were
identified: time stress and anxiety. Both job stress dimensions were signifi-
cantly related to each of the model's five organizational stressor categories,
but not all of the independent variables within the categories were significantly
related to job stress. Moreover, the specific stressors associated with each
dimension of job stress proved to be substantially different.

Interest in the phenomenon of work-related stress has increased mark-

edly during the last few years, as indicated by several reviews of the
literature, a number of books, and a rash of public seminars promoting
different approaches to stress control (Brief, Schuler, & Van Sell, 1981;
Cooper & Marshall, 1976; Kahn, 1981; Levi, 1981; Moss, 1981). Despite
this widespread interest, there have been relatively few reports of em-
pirical investigations of stress in work organizations. The organizational
literature on stress is dominated instead by concept papers with few tests
of the concepts presented.
Nevertheless, the available evidence and common sense suggest that
job stress contributes to health-related problems among workers and to
Our thanks to Professors Angelo DeNisi, Bruce Meglino, and John Ivancevich for their
reviews of an earlier draft of this paper. Our thanks also to B. Ruth Montgomery for her
assistance with the data analysis and her many helpful comments. The Graduate School of
Business Administration, The University of Michigan, provided computer support for the
analysis of these data. Requests for reprints should be sent to Dr. Donald E Parker, Grad-
uate School of Business Administration, University of Michigan, Ann Arbor, MI 48109.
160
0030-5073/83 $3.00
Copyright ~) 1983 by Academic Press, Inc'
All rizhts of reoroduction in any form reserved.
 D E T E R M I N A N T S OF JOB STRESS 161

organizational problems such as employee dissatisfaction, alienation, low

productivity, absenteeism, and turnover (Beehr & Newman, 1978; Schuler,
1980) at an estimated annual cost of between 10 and 20 billion dollars
(Jick & Payne, 1980). In view of the human and monetary costs of these
problems, there is a need for more empirical studies of stress phenomena
in work organizations. However, at least three factors make such studies
unusually complex, and difficult to interpret and generalize: notably, a
lack of conceptual clarity on the meaning of stress, the choice of an
appropriate research perspective, and methodological problems inherent
in the study of stress phenomena.

Conceptual Problems
A concept is a word that expresses an abstract generalization derived
from particular understandings of observable phenomena. Thus, for ex-
ample, the concept of aggression has been defined as " a number of par-
ticular actions having the similar characteristic of hurting people or ob-
jects" (Kerlinger, 1967, p. 4). One of the advantages of such a precise
definition is that it provides clear and common direction to research on
the concept and, thereby, facilitates comparisons across studies. Unfor-
tunately, the concept of stress lacks precision in that it has been both
broadly and narrowly defined, and treated as a stimulus, a response, an
environmental characteristic, an individual attribute, and an interaction
between an individual and his or her environment (Beehr & Newman,
1978; Katz & Kahn, 1978; Levi, 1981). Thus, for example, while one
researcher might label a physiological dysfunction as stress (Ivancevich
& Matteson, 1980a) another would call it a consequence of stress (Schuler,
1980). Given these differences in treatment, it is not surprising that there
is no concensus on the concept of stress. It is only a slight exaggeration
to say that it is whatever a given researcher says it is. Most often, the
concept is discussed as though it were unidimensional, less often as mul-
tidimensional, and least often as both multidimensional and variable, with
a potential for variation in the level of intensity associated with each
dimension (Schuler, 1980).
In sum, despite several excellent reviews of the literature (Beehr &
Newman, 1978; Cooper & Marshall, 1976; Ivancevich & Matteson, 1980a,
1980b; and Schuler, 1980), industrial and organizational psychologists have
not agreed on the meaning and process of stress in work organizations.
In recognition, Ivancevich and Matteson have labeled stress as "the most
imprecise [term] in the scientific dictionary" (1980a, p. 5). As will be
discussed, it is our view that job stress is a particular individual's aware-
ness or feeling of personal dysfunction as a result of perceived conditions
or happenings in the work setting.
162 P A R K E R A N D DE COTIIS

Problems of Research Perspective

Stress has been studied from the three perspectives of individual dif-
ferences, environmental factors, and some admixture of the two. Appro-
priately, the choice of one perspective over another has typically been
determined by the research question(s) to be answered. For example,
medical researchers who are interested in the physiological outcomes of
stress treat the individual as the unit of analysis and focus on personal
characteristics such as heredity, age, and personality traits. In contrast,
researchers with an organizational perspective typically ignore individual
differences and concentrate on organization-based sources of stress such
as job content and the quality of supervision, while the integrated ap-
proach focuses on both individual differences and environmental factors
(cf. Ivancevich & Matteson, 1980b).
Surprisingly, industrial and organizational psychologists have largely
ignored the theoretical and practical implications of the choice of a par-
ticular perspective even though it determines the variables included in a
study and, therefore, the conclusions reached and solutions recom-
mended. For example, it can be reasonably argued that many of the
currently popular stress-coping seminars owe their popularity to the im-
plicit assumption that individuals can choose their response (e.g., stress)
to stimuli or change the nature of the stimuli by acting on their environ-
ment. This assumption raises an interesting question: "What if the indi-
vidual can't?" Similarly, the organizational perspective leads to discus-
sions of how an organization can mediate stress by changing its processes
(stressors) or, indeed, to the more fundamental decision of whether stress
is a problem it chooses to address (Ivancevich & Matteson, 1980b).
At best, the available evidence suggests that individual differences have
a moderating effect on reactions to potentially stressful situations (French
& Caplan, 1972; Holland, 1973; Ivancevich & Matteson, 1980a; Kahn,
1981; Rosenman, Brand, Jenkins, Friedman, Straus, & Wurm, 1975). It
has been reported, for example, that certain personality characteristics,
such as those comprising the "Type A" behavior pattern, may affect
individual responses to potential stressors (Sales, 1969). It does not nec-
essarily follow, however, that individual differences are the most appro-
priate perspective from which to study stress in work organizations. Rel-
ative to other potential moderators, individual differences may, in fact,
account for little variance in stress reactions to a particular situation. In
addition, it may not be necessary or even useful to assess individual
differences directly in order to determine whether the modal response to
a situation is one that could be labeled stress--particularly if, as suggested
by French and Caplan (1972), French (1974), and Holland (1973), people
seek jobs that are congruent with their personal characteristics. If so,
 D E T E R M I N A N T S OF JOB STRESS 163

there may be too little variance among incumbents of certain jobs for
individual differences to be a significant predictor of stress reactions for
that occupation. This explanation may account for the finding that few
personal characteristics are associated with stress in what are generally
recognized to be highly stressful occupations (e.g., police officer, air traffic
controller). Thus in work settings, it may be that the organizational per-
spective deserves more theoretical and empirical attention than it cur-
rently receives. This may be particularly true when the investigation is
intraoccupational rather than across occupations, with the latter's higher
probability of greater variance in individual differences.

Methodological Problems
One problem that makes the use of existing research and future em-
pirical investigations difficult is the multicollinearity inherent in the large
number of intercorrelated variables typically associated with job stress.
The options for dealing with this problem are (a) ignore it; (b) delete one
or more of the offending variables; or (c) transform the variables into
combinations that are uncorrelated (Green, 1978). While none of the op-
tions is completely satisfactory, the first option is most often selected in
stress research, perhaps on the premise that we live in a multicollinear
world. A more practicable approach may simply be for the researcher to
point out where multicollinearity is thought to be a problem so that the
reader can knowledgeably interpret the results.
Another potential problem in the study of job stress is the need to
measure both stressors and stress through the unique perceptual lens of
the individual. That is, if the concept of stress is defined in terms of the
personal reaction of a particular individual to stimuli in his or her envi-
ronment, then self-report measures should be common in stress research.
In short, stress is in the " e y e of the beholder." Although this orientation
to stress has inherent in it the problems of monomethod research, there
appears to be no fully acceptable alternative that does not compromise
the precision of the concept.
Summary
The above discussion suggests a need to specify a carefully delimited
concept of stress as a basis for research that will result in a theory of
stress in work organizations. The possibility exists that stress phenomena
are too complex to support a theory specifying only a small set of vari-
ables that cause the content, process, and consequences of stress. Thus,
for the present at least, it will probably be necessary to settle for partial
tests of complex models and to tolerate some unwanted, but inescapable,
ambiguity resulting from problems such as incomplete abstractions, com-
promise operationalizations, and multicollinearity. The current literature
164 P A R K E R A N D DE COTIIS

is a more-than-adequate starting point for the conceptualization and testing

of such stress models.

A MODIFIED JOB STRESS CONCEPT

Beehr and Newman (1978), Cooper and Marshall (1976), Ivancevich
and Matteson (1980a), Katz and Kahn (1978), and Schuler (1980) all point
out that individuals in organizations are subject to conditions and occur-
rences that may result in psychological and/or physiological deviations
from normal functioning. These deviations are seen as resulting from
anticipated or missed opportunities, constraints on goal-directed be-
havior, or demands leading to important but uncertain outcomes (Schuler,
1980). Hence, these researchers would define stress as a deviation from
normal psychological or physiological functioning caused by exigencies
in the individual's immediate environment. It may be that the psycholog-
ical-physiological dysfunction conceptualization of job stress is too in-
clusive, and is problematic in that it does not distinguish whether the
individual is aware of the dysfunction. It is our view that physiological
dysfunctions are more likely to be the consequences of stress and not its
content, and, further, that job stress in order to have consequences of
any kind must be at least at some level of awareness. This conceptual-
ization anchors the notion of "deviations from normal" in the experience
of the individual; that is, what is normal for one individual's functioning
may well be abnormal for another.
In contrast, Margolis and Kroes (1974) suggest that individuals manifest
stress in various ways, including short-term psychological states and
physiological reactions, long-term or chronic psychological responses and
physical difficulties, and work performance decrements. However, to treat
all such deviations as stress without regard to time or duration may only
contaminate the concept, make it unduly complex, and allow any mal-
adaptive state to be defined as stress. It seems more reasonable to view
short-term psychological states (e.g., anxiety, tension) as representing a
different level of outcome than long-term deviations, such as impaired
health or reduced productivity. In our view then, stress is a transient
feeling relative to more lasting deviations; the latter second-level out-
comes are best viewed (when they are stress-related) as c o n s e q u e n c e s of
job stress rather than as stress per se. They occur, if at all, later in time
and usually in response to intense and recurring job stress. This distinc-
tion between stress as a first-level outcome, the potential for its recur-
rence, and second-level outcomes is consistent with Selye's (1976) view
that stress is additive. Kanner, Lazarus, and their associates (Kanner,
Coyne, Schaefer, & Lazarus, 1981; Lazarus, 1981) also suggest a similar
concept in their empirical study of the psychological and physiological
effects of the day-to-day "hassles" people experience. In general, these
 D E T E R M I N A N T S OF JOB STRESS 165

views suggest that the concept of stress be limited to the felt, manifestly
uncomfortable, response of an individual to environmental stimuli thereby
precluding its being treated as a characteristic of the environment or an
attribute of the individual.
Our discussion suggests that the feeling we call job stress may be found
in the presence or absence of more lasting deviations from normal func-
tioning. When the stressor is removed without delay, or when the indi-
vidual is successful in coping with it, the feeling of stress may dissipate
quickly without resulting in any second-level outcome. If stress is intense
and/or prolonged, however, second-level outcomes are more likely to
occur. Thus, whether job stress is short term and inconsequential or leads
to more lasting second-level outcomes depends upon its duration, inten-
sity, the number of operative stressors, and the individual's ability to
dissipate the feeling of stress. As noted, identical objective conditions
may cause intense stress and, perhaps, lasting second-level outcomes in
one individual, and little or no job stress in another. It may be that job
stress, like cognitive dissonance (Festinger, 1957), can be dissipated ei-
ther cognitively or behaviorally, or simply endured. Consequently, a va-
riety of outcomes, or the absence of any perceptible outcome, may result
from equivalent levels of job stress. If stress is intense or if it continues
over a prolonged period, the likelihood of second-level outcomes is in-
creased.
Thus, we use the term job stress to describe the feeling of a person
who is required to deviate from normal or self-desired functioning in the
work place as the result of opportunities, constraints, or demands relating
to potentially important work-related outcomes. Our use of the term feeling
is intentional, in that it implies a "subjective awareness of our own emo-
tional state" (Gaylin, 1979, p. 1). In addition, this feeling is both uncom-
fortable and undesirable to the individual. This conceptualization of job
stress as being both uncomfortable and undesirable helps to distinguish
it from the positive motivational feelings of arousal that result from the
challenge of a difficult but attainable goal. Hence, our concept of stress
is limited to an emotional response to stimuli that may have dysfunctional
psychological or physiological consequences.

An Organizational Model of Stress

Figure 1 presents our model of job stress. As shown in Fig. 1, the
stressors are grouped into six categories: (1) characteristics and condi-
tions of the job itself, (2) conditions associated with the organization's
structure, climate, and information flow, (3) role-related factors, (4) re-
lationships at work, (5) perceived career development, and (6) external
commitments and responsibilities. The typology of stressors shown in
Fig. 1 is similar to those discussed by Cooper and Marshall (1976) and
166 PARKER AND DE COTIIS

Stressor First-levelOutcome Second-levelOutcome

Work I t s e l f [

Organizational
Characteristics Organizational
Commitment

I Rolein OrganizationI Job SatisfactionI

JOB STRESS
~ AvoidanceBehaviorI

i C r er Oove'o m nt IJobPerformanceI

External Commitments
and Responsibilities

FIG. 1. A model of job stress.

Ivancevich and Matteson (1980b). However, our model differs in that it

is primarily organizational and does not treat individual differences as a
moderator of the stressor-stress relationship.
The model shows two levels of outcomes: first- and second-level. Only
the first-level outcome is referred to as stress. Second-level outcomes are
viewed as individual and organizational consequences which are affected
by stress, and quite likely by other variables as well. These second-level
outcomes might include decrements in organizational commitment, sat-
isfaction, motivation, and job performance, and avoidance behavior.
THE PRESENT STUDY
The present study was intended to be a partial test of the model de-
scribed above. Specifically, we set out to identify the nature of the rela-
tionships between several potential stressors, selected on the basis of a
review of the stress literature and a knowledge of the target population,
and the first-level outcome, job stress. We also intended to examine more
closely the dimensionality of job stress, to determine the extent to which
the dimensions are differentially affected by different stressors. For ex-
ample, our experience with managers and familiarity with the popular
management literature suggest the existence of a time pressure dimension
 D E T E R M I N A N T S OF JOB STRESS 167

(e.g., "crisis management" and "too much to do, too little time"), while
our reading of the empirical and psychoanalytic literature on the content
of emotion suggests a distinction between time pressure as stress and the
discomfort or feeling of stress. Moreover, if stress is multidimensional,
it is not unreasonable to expect the separate dimensions to have patterns
of stressors that differentially contribute to explained variance in stress.
However, beyond our anecdotal dimensionalization of stress, we have no
a priori expectations with respect to either the dimensions of stress or
patterns of stressors.

METHOD
Sample
Participants in the study were 367 managers employed by a major res-
taurant chain who constitute 61% of the total managerial employees. Their
jobs ranged from entry-level trainees through regional managers. The
sample was 97.5% male and over 99% white. The mean age, tenure, time
in present job, and income of the participants were 28.55 years (SD =
4.38 years), 2.67 years (SD = 2.10 years), 1.11 years (SD = 1.20 years),
and $13,229 (SD = $7,744), respectively.

Measures
The data were gathered on a voluntary basis through a mail survey
questionnaire. The questionnaire was designed to assess managers' per-
ceptions of several aspects of the organization, some of which were eval-
uated as being possible stressors; namely, its structure, human resource
processes, and organizational climate. Four aspects of perceived struc-
ture were assessed: formalization, centralization, role ambiguity, and role
conflict. The latter two scales were short forms of the measures devel-
oped by Rizzo, House, and Lirtzman (1970). Nine organizational pro-
cesses were assessed: three dimensions of leadership (i.e., interpersonal
style, task orientation, and closeness), compensation, training, career
management, planning, communication, and decision making. The seven
dimensions of organizational climate assessed were autonomy, trust, co-
hesiveness, support, recognition, innovation, and fairness. Stress was
defined by 15 items based on our review of the literature and earlier
interviews with managers in the organization. Responses to individual
items were measured using a 4-point Likert-type scale. Scales were sub-
sequently developed to measure various stressor dimensions; these scales
consisted of four to seven individual items. The internal consistency of
the scales was determined by computing corrected item-total score cor-
relations and coefficient alphas for each scale (Nunnally, 1978). For all
scales, the corrected item-total score coefficients ranged from a low of
.26 to a high of .84, while coefficient alpha ranged from .81 to .97. The
168 PARKER AND DE COTIIS

TABLE 1
VARIABLES INCLUDED WITHIN EACH STRESSOR CATEGORY

1. Aspects of job ° 4. Relationships at work

Autonomy b Trust
Stability Support from boss
Pay-performance limitations Cohesiveness
Task variety
Emphasis on achieving
Actual base salary*
Hours worked per week*
2. Structure, climate, information flow 5. Aspects of role
Recognition of good performance Innovativeness encouraged
Fairness Role conflict
Formalization Role ambiguity
Centralization Boss's encouragement/support
Quality of decision making Boss's task orientation
Communication adequacy Closeness of supervision
Communication openness Tenure in job*
Emphasis: profit vs people Tenure in company*
Concern for individuals Organization level*
Corporate management out of touch* Supply support
3. Career development 6. Extraorganizational variables
Training quality Age*
Basis of promotions Sex*
Performance feedback Number of dependents*
Emphasis on individual development Years of education*

o Measures marked with an asterisk are single items and, typically, objective measures;
all other variables are scales.
b The scales are 4-point scales.

score of an individual on a given scale was defined as his or her mean

response to all items in the scale with negative items reversed. 1 Using
the literature as a guide (Beehr & Newman, 1978; Brief et al., 1981;
Cooper & Marshall, 1976; Ivancevich & Matteson, 1980b; Schuler, 1980),
the scales were classified into the six categories of stressors shown in
Fig. 1. The content of each stressor category is shown in Table 1.

Stress D i m e n s i o n s
The dimensionality of job stress was identified through principal com-
ponents analysis, with the solution rotated to the varimax criterion. The
criterion of item retention was a factor loading of not less than .50 on the
defining component; with one exception, this resulted in a factor loading
of less than .35 on all other components. Two components, accounting

Additional information about the items and scales used in this study is available from
the authors.
 DETERMINANTS OF JOB STRESS 169

for 77.5% of the variance, were extracted from the 15 stress items. Using
the loadings criterion specified above, all but 2 of the 15 items were
retained. Table 2 presents the results of the principal components anal-
ysis.
The content of the items comprising the first component is closely
associated with feelings of being under substantial time pressure, and is
named time stress. The second component is dominated by items having
to do with job-related feelings of anxiety. This dimension of job stress is
referred to as anxiety. The means for time stress and anxiety are 2.47
and 1.93, respectively; the respective standard deviations are .682 and
.649. Cronbach's alpha for the components treated as scales are .86 and
.74, respectively. The corrected item-total score correlations for time
stress range from .61 to .79, and for anxiety, from .61 to .75. The cor-
relation between the factors used as scales was .54, indicating consider-
able nonoverlapping variance in the dimensions.

TABLE 2
PRINCIPAL COMPONENTS ANALYSIS OF JOB STRESS ITEMS

Component
I II

Item
I have felt fidgety or nervous as a result of my job .11 .76
Working here makes it hard to spend enough time with .74 .12
my family
My job gets to me more than it should .29 .70
I spend so much time at work, I can't see the forest for .57 .48
the trees
There are lots of times when my job drives me right up .32 .68
the wall
Working here leaves little time for other activities .77 .20
Sometimes when I think about my job I get a tight feeling .23 .67
in my chest
I frequently get the feeling I am married to the company .65 .3O
I have too much work and too little time to do it in .63 .28
I feel guilty when I take time off from job .05 .51
I sometimes dread the telephone ringing at home because .57 .16
the call might be job-related
I feel like I never have a day off .73 .08
Too many people at my level in the company get burned .66 .15
out by job demands
Percent Variance 28.7 48.8
Variables not included
I don't have enough time to develop my people .49 .45
This is a relaxed place to work .48 .30
170 PARKER AND DE COTIIS

Stressor-Stress Relationships
Assessment of the stressor-stress relationships consisted of: (1) within-
category analysis of each stressor for each dimension of stress; (2) de-
termination of the variance explained in each dimension of stress by all
stressor categories; and (3) summarizing the relationship b e t w e e n all
stressors and each of the dimensions of stress.

Within-Category Analysis
The first step was to compute the multiple correlation between the
variables defining each of the stressor categories and the dimensions of
stress. The partial regression coefficients and significance levels of each
of the variables within categories were then analyzed to determine the
contributions of each stressor to the total regression. Finally, both for-
ward and backward multiple regressions were performed as a basis for
assessing multicollinearity among the independent variables. 2 Table 3 pre-
sents the results of the within-category analysis.
As can be seen from the results presented in Table 3, most of the
categories of stressors made significant contributions to the variance ex-
plained in each dimension of stress. The RZ's, corrected for the number
of variables and sample size, ranged from .37 to .05 for time stress, and
from .25 to .03 for anxiety. However, as expected, the extraorganizational
stressors were relatively weak. Within categories, not all of the variables
made a significant contribution to the variance explained in each of the
dependent variables. Table 3 also presents the findings for the variables
within each of the stressor categories.
The job. As a set, the variables in this category showed the second
highest relationship with time stress (R 2 = .35) and the third highest with
anxiety (R 2 = . 17). With respect to time stress, five of the eight variables
in the category were found to have a significant relationship. The strongest
single relationship (a partial of .36) was with an objective measure of
hours w o r k e d per week. Autonomy, stability, perceived limitations on the
relationship between performance and pay, and the perceived basis for
compensation decisions also contributed significantly to the variance ex-
plained in time stress. With respect to anxiety, only three of the eight
independent variables relating to the job itself contributed to the variance
explained.
Structure~climate~information. As shown in Table 3, the scales and items
in this category yielded the highest corrected R 2 with time stress (R 2 =
2Both forward and backward multiple regressions were performed in order to assess
collinearity among the independent variables. The extent of this condition was determined
by examining changes in the partial regression coefficients and significance levels as the
independent variables were entered into or removed from the equation. Multicollinearityis
discussed at those points in the text where it appears to be a problem.
 TABLE 3
SUMMARYOF RELATIONSHIPSBETWEEN STRESSOR CATEGORIESAND THE DIMENSIONSOF
JOB STRESS

Stress

Time Anxiety

Stressor category R 2a Partial R2 Partial

Job .35*** .17"**

Autonomy - . 14"* NS
Stability -.24"** -.27"**
Pay-performance limit - . 13" NS
Compensation basis - . 16"* - . 13"
Task variety NS NS
Emphasis on achievement NS NS
Base salary NS NS
Hours worked per week .36*** .15"*
Structure, climate, information .37*** .24***
Recognition NS NS
Fairness NS NS
Formalization NS - . 11"
Centralization NS NS
Decision making NS NS
Communication adequacy NS NS
Communication openness - . 12' NS
Emphasis: profit vs people NS NS
Concern for individual - . 15"* - . 14'*
Corporate management out of touch .20** .16"*
Role .19"** .25***
Encouragement of innovation NS NS
Role conflict NS .25***
Role ambiguity NS NS
Task orientation, boss NS NS
Closeness of supervision .13" .13"
Tenure, present job NS NS
Supply support problems .14" .13"
Career .22*** .15"**
Training quality - . 11" - . 19"*
Promotion basis - . 11" - . 19'*
Performance NS NS
Emphasis on individual development -.30*** NS
Interpersonal relationships at work .12"** .10"**
Trust NS NS
Support from boss -.14" NS
Cohesiveness - . 15"* NS
Personal/extraorganizational .05" .03"
Age - . 13" NS
Sex NS NS
Number of dependents NS NS
Years of education .19"* .19"*

aR 2 is corrected for sample size and number of variables.

* p < .05.
** p < .01.
*** p < .001.
171
172 PARKER AND DE COTIIS

.37) and the second highest (.24) with anxiety. Of the 11 scales and items
in this category, 3 displayed significant relationships with time stress.
These variables involve feelings that top management is out of touch with
problems at the respondent's level, the concern shown for individuals in
the organization, and the openness of communications.
The feeling of anxiety showed significant relationships with two of the
same stressors as time, plus one additional variable. As in the instance
of time, the feeling that top management was out of touch with the re-
spondent's problems showed the strongest relationship to anxiety, and
beliefs about concern for individuals was the second strongest. However,
unlike time, anxiety was negatively related to the degree to which pro-
cedures and practices are formalized in the organization.
Role in the organization. This is the only category of stressors that
was found to be more strongly related to anxiety (R 2 = .25) than to time
stress (R 2 = . 19). As shown in Table 3, only two of the variables in this
category were found to have significant relationships with time stress,
both positive. The first involves problems experienced with a supply sup-
port unit of the organization, while the second deals with the extent to
which the respondent's superior is perceived to snoop around the work
place, to be present most of the time, and the like. Of the three variables
that are significantly associated with anxiety, the two just named---supply
support problems and closeness of supervision--were also related to time
stress. In both cases the partial regression coefficient was almost iden-
tical. The strongest relationship by far, however, was between role con-
flict and anxiety.
Career development opportunities. Three of the four variables shown
in Table 3 were significantly related to time stress: the emphasis placed
on individual development; the extent to which promotions are based on
merit; and the quality of training received in preparation for greater re-
sponsibility. The fourth variable, performance feedback, was not signifi-
cantly related to time stress.
Two of the four career development measures were significantly related
to anxiety: training quality and perceived basis for promotions. Both of
these variables were found to have negative relationships, yielding iden-
tical partial regression coefficients of - . 19. The two remaining variables
did not attain significance. However, this result may be due in part to
multicollinearity between the scales relating to performance feedback and
emphasis on individual development. Specifically, when the latter vari-
able was removed using backward stepwise regression, the former be-
came significant (p < .05) with a partial o f . 11, without yielding a signif-
icant increase in R 2.
Relationships. The category of hypothetical stressors having to do with
 DETERMINANTS OF JOB STRESS 173

interpersonal relationships on the job consists of three variables: trust,

perceived support from one's boss, and cohesiveness. An additional item
was the statement: "There is little "back-biting" between employees at
my level." As shown in Table 3, the interpersonal relationship category
yielded significant R2's with both time stress (R 2 = . 12) and anxiety (R 2
= . 10). Two variables, cohesiveness and support, yielded significant par-
tial regression coefficients with time. Interestingly, although the four in-
dependent variables together had a corrected R 2 o f . 10 (p < .001), none
of the four yielded a significant partial. The most likely explanation for
this result is multicollinearity between trust and support. When the former
is removed through backward stepwise regression, the latter becomes
significant (p < .001), and the partial regression coefficient increases from
- . 0 8 to - . 3 1 .
Extraorganization. This category of stressors consisted of four vari-
ables: age, sex, number of dependents, and years of education. Both
regressions were significant, but the corrected R2's were small and the
significance levels were reduced. Specifically, the corrected R2's were .05
(p < .01) and .03 (p < .05) for time stress and anxiety, respectively. Two
partial regression coefficients were found to be significant when regressed
on time stress: a positive relationship with years of education and a neg-
ative relationship with age. In the instance of anxiety, only one variable,
education, proved significant, with the relationship being positive.

Total Variance Explained

Within-stress dimensions. In order to determine the total explained
variance in each dimension and to identify the dominant stressors, each
of the dimensions of stress was regressed on the independent variables
without regard to stressor category. The corrected R2's for time stress
and anxiety were .43 and .29, respectively. When compared to the results
shown in Table 3, these results indicate that the total regression accounts
for significantly more variance than any single category of stressors.
Seven variables were consistently related to time stress in all the
regression analyses, including the total regression. In declining order of
partial regression coefficient, they are: hours worked per week, percep-
tion that top management is out of touch, concern for employees, close-
ness of supervision, perception of the basis of compensation, organiza-
tional level of respondent, and emphasis on individual development. While
trust and task orientation of the superior appeared significant, their effect
was muted by multicollinearity. From the perspective of the model shown
in Fig. 1, it is interesting to note that with the exception of extraorgani-
zational stressors, each of the stressor categories is represented among
these nine variables. Moreover, in all but one instance, where the pres-
174 PARKER AND DE COTIIS

ence of multicollinearity was noted, the variables that were significant in

the assessment of total variance in time stress were also significant on a
within-category basis.
The results with respect to anxiety are somewhat less clear. Specifi-
cally, only two variables (role conflict and acceptance of innovation) were
significant when all of the independent variables were forced into the
regression, suggesting the presence of multicollinearity. Tenure with the
company approached significance (p < .06), and several other variables
achieved significance at various stages of both the forward and backward
stepwise regressions but, because of multicollinearity, they were not con-
sistently significant. These include centralization, communication ade-
quacy, education level, degree to which innovation is encouraged, and
quality of training. Five of the seven measures that proved significant
were from two stressor categories: two were role-related variables, and
three were related to structure, climate, and information. Interestingly,
level of education again showed a significant negative relationship with
anxiety in the forward and backward regressions, although it was not
significant in the total regression.
Canonical analysis. An important question concerned how much of the
variance in job stress could be accounted for using self-report measures
of the status of organizational variables, without recourse to individual
differences. In order to answer this question, a canonical correlation anal-
ysis was calculated using all of the supposed stressors as independent
variables and the two job stress dimensions as dependent variables. The
redundancy statistics for the two canonical variates were summed to
determine the amount of variance explained in job stress. The two re-
dundancy statistics were .51 and .18, totaling 69% explained variance.

DISCUSSION
A limited concept of job stress that emphasizes an individual's subjec-
tive awareness of dysfunction was presented. The results of a partial test
suggest that there are at least two dimensions of this feeling but do not
rule out the possibility of additional dimensions. Indeed, one priority task
for future research is the investigation of other possible dimensions of
stress.
A number of the potential stressors were found to be associated with
one or both of the dimensions of job stress, indicating some commonality
of the determinants, but with differences in the patterns of determination.
Of interest is the finding that a number of variables that were expected
to be related to stress were not. For example, it is somewhat surprising
that variables such as emphasis on achievement, fairness, decision making,
and feedback were not predictive of these stress dimensions. In a few
 DETERMINANTS OF JOB STRESS 175

instances, this result may well be due to the problem of multicollinearity

among the independent variables, and in others due to inadequate oper-
ationalizations.
The results also indicate that some of the stressors are related to one
form of job stress, but not both. A number, such as the belief that upper-
level management is out of touch with day-to-day managerial problems,
opinions about the quality of company training programs, and beliefs
about whether rewards are based on merit, apparently affect both di-
mensions of stress. However, in addition to the shared stressors, each
dimension of stress has associated with it a unique pattern of stressors.
For example, as shown in Table 3, time stress is uniquely determined by
autonomy (work itself), the perception that there is a limit on the rela-
tionship between pay and performance (work itself), the openness of
communication (structure, climate, information), support from the boss
(relationships), and cohesiveness (relationships). In contrast, anxiety is
uniquely determined by formalization (structure, climate, information)
and role conflict (role). It may be that stress has its start in a constellation
of common stressors, and is then differentiated by the effects of unique
stressors. These findings must be regarded as tentative, however. The
meaning of these and other stressors thus merits further thought and
investigation.
In terms of explained variance, the results of the current study are
encouraging and suggest the importance of organizational sources of stress.
If the importance of the organization as a source of stress is supported
by future research, then a means of stress reduction not ordinarily used
(i.e., the organization itself) should be considered. Put somewhat differ-
ently, the results indicate that much of the effort now directed toward
educating individuals to understand and cope with job stress may be
misplaced. Instead, it may be that those who design and manage orga-
nizations are in a much better position to assess the causes of stress and,
where appropriate, to remove or moderate them. Indeed, if the present
results are replicated elsewhere, the individual change approach with its
focus on sensitizing people to the existence of stress may, in fact, be
harmful, especially when the individual has no means of removing the
source(s) of stress. This reasoning raises the intriguing question of whether
the individualized approach to stress may induce rather than reduce stress.
Obviously, more research is needed. Specifically, we believe four areas
of research are indicated by the findings reported here. First, efforts
should be made to determine whether there are other dimensions of job
stress. Second, the generalizability of the dimensions identified in the
present study should be tested. Such tests should concentrate especially
on different industries, age groups, and occupations. Third, research is
176 PARKER AND DE COTIIS

needed in order to learn more about the kinds of stressors identified in

the present study. Questions that merit consideration include the unique-
ness of stressors, given a particular dimension of stress, and whether
some stressors are situation specific. Specificity, for example, is sug-
gested by our finding that problems with supply sources were relatively
strongly related to both dimensions of stress in this restaurant industry
setting. It seems reasonable to assume that some stressors may be unique
to a particular type of setting while others may be found in many kinds
of organizations, but the issue needs to be investigated.
Finally, the model we have presented indicates that experienced job
stress will sometimes, but not always, lead to organizationally and indi-
vidually relevant second-level outcomes such as reduced job performance
and voluntary turnover. Our conceptualization of stress phenomena (Fig.
1) suggests that the relationship between stress and second-level out-
comes will depend upon the intensity of the stress, its duration, the number
of operative stressors, and alternatives the individual sees as being avail-
able to him or her. Future research could profitably be directed toward
identifying the nature and strength of the relationship between job stress
and its possible consequences.

REFERENCES
Beehr, T. A., & Newman, J. E. Job stress, employee health, and organizational effective-
ness: A facet analysis, model and literature review. Personnel Psychology, 1978, 31,
665-699.
Brief, A. P., Schuler, R. S., & Van Sell, M. Managing job stress. Boston: Little, Brown,
1981.
Cooper, C. L., & Marshall, J. Occupational sources of stress: A review of the literature
relating to coronary heart disease and mental ill health. Journal of Occupational Psy-
chology, 1976, 49, 11-28.
Festinger, L. A theory of cognitive dissonance. Evanston, II1.: Row, Peterson, 1957.
French, J. R. P. Person-role fit. In A. McLean (Ed.), Occupational stress. Springfield, II1.:
Thomas, 1974.
French, J. R. P., & Caplan, R. D. Organizational stress and individual strain. In A. J. Marrow
(Ed.), The failure of success. New York: American Management Association, 1972.
Gaylin, W. Feelings: Our vital signs. New York: Harper & Row, 1979.
Green, P. E. Analyzing multivariate data. Hinsdale, II1.: The Dryden Press, 1978.
Holland, J. L. Making vocational choices. Englewood Cliffs, N.J.: Prentice-Hall, 1973.
Ivancevich, J. M., & Matteson, M. T. Stress and work. Glenview, II1.: Scott, Foresman,
1980. (a)
Ivancevich, J. M., & Matteson, M. T. Optimizing human resources: A case for preventive
health and stress management. Organizational Dynamics, Fall 1980, 5-25. (b)
Jick, T. D., & Payne, R. Stress at work. Exchange, 1980, 3, 50-56.
Kahn, R. L., Wolfe, D. M., Quinn, R. P., Snoek, J. R., & Rosenthal, R. A. Organizational
stress: Studies in role conflict and ambiguity. New York: Wiley, 1964.
Kahn, R. L., Work and health. New York: Wiley, 1981.
Kanner, A. D., Coyne, J. C., Schaefer, C., & Lazarus, R. S. Comparison of two modes of
 DETERMINANTS OF JOB STRESS 177

stress measurement: Daily hassles and uplifts versus major life events. Journal of
Behavioral Medicine, 1981, 4(1), 1-39.
Katz, D., & Kahn, R. L. The social psychology of organizations. New York: Wiley, 1978.
2nd ed.
Kerlinger, F. N. Foundations of behavioral research. New York: Holt, Rinehart & Winston,
1967.
Lazarus, R. S. Little hassles can be hazardous to health. Psychology Today, 1981, 15(7),
58-62.
Levi, L. Preventing work stress. Reading, Mass.: Addison-Wesley, 1981.
Margolis, B. K., & Kroes, W. H. Occupational stress and strain. In A. McClean (Ed.),
Occupational stress. Springfield, II1.: Thomas, 1974. Pp. 15-20.
Moss, L. Management stress. Reading, Mass.: Addison-Wesley, 1981.
Nunnally, J. C. Psychometric theory. New York: McGraw-Hill, 1978.2nd ed.
Rizzo, J., House, R. J., & Lirtzman, S. I. Role conflict and ambiguity in complex organi-
zations. Administrative Science Quarterly, 1970, 15, 150-163.
Rosenman, R. H., Brand, R. J., Jenkins, C. D., Friedman, M., Straus, R., & Wurm, J.
Coronary heart disease in the western collaborative group study. Journal of the Amer-
ican Medical Association, 1975, 233(8), 872-877.
Sales, S. Organizational role as a risk in coronary disease. Administrative Science Quar-
terly, 1969, 14, 325-336.
Schuler, R. S. Definition and conceptualization of stress in organizations. Organizational
Behavior and Human Performance, 1980, 25, 184-215.
Selye, H. The stress of life. New York: McGraw-Hill, 1976.2nd ed.
RECEIVED: March 25, 1982