You are on page 1of 6

Journal of Applied Psychology Copyright 1992 by the American Psychological Association, Inc.

1992, Vol. 77, No. 5,623-628 0021-9010/92/53.00

The Meaning of Occupational Stress Items to Survey Respondents


Steve M. Jex, Terry A. Beehr, and Cathlyn K. Roberts
Central Michigan University

This study tested the effect of using the word stress in the measurement of self-reported occupa-
tional stressors and strains. Employees from two organizations responded to a questionnaire that
included specific occupational stressors, strains, and 16 items in which the word stress was used.
Survey respondents tended to interpret the word stress to refer both to employees' strains or
reactions to the work environment and to job stressors or elements of the environment itself.
Implications of these findings for occupational stress research are discussed.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Work-stress research has been characterized by a lack of 1980). Thus, as recommended by McGrath (1976), the word
agreement about important terms and constructs (Beehr & stress is used to label an area of study, as is done in other topical
Bhagat, 1985; Beehr& Newman, 1978; Ivancevich & Matteson, areas in industrial and organizational psychology (e.g., leader-
1980; Schuler, 1980). When Selye (1976) began using the term ship, job design). In this context, the term stressor is often used
stress to mean a syndrome of human responses, his English was to refer to job or organizational conditions, and the term strain
not yet very good, leading him to misuse the term. He noted is used to refer to the individual's response to these conditions,
that he should have chosen the word strain to be consistent with as is done in this article.
fields such as physics and engineering. It is probably no exagger- To investigate the degree to which there is commonality in
ation to say there has been confusion ever since with respect to usage of the word stress among researchers, we manually
the term stress. Work-stress researchers have typically denned searched the 1985-1989 issues of six major journals in organiza-
stress in one of three ways: as a stimulus, a response, or a stimu- tional behavior. Each article in which the word stress or stress-
lus-response relationship. ful appeared was assigned to one of four categories: stimulus,
A stimulus definition of stress refers to a job stressor, which is response, stimulus-response, and usage unclear. Articles were
any environmental event in the workplace requiring some type first examined by Terry A. Beehr and Cathlyn K. Roberts, and
of adaptive response. Stimulus definitions were derived from disagreements were then classified by consensus. Of the 51
physics and were borrowed by organizational psychologists stress articles found, 21 (41 %) used the word to refer to stimulus
from the field of engineering. In this frame of reference, stress conditions, 11 (22%) referred to responses, and 13 (25%) implied
means any outside force on an object (person), and strain is a stimulus-response conditions. For 7 (14%) articles, the usage
potentially harmful effect of the force on that object (e.g., Kahn was unclear. There was a slight tendency for researchers to use a
& Quinn, 1970; Lazarus, 1966). Using this definition, a wide stimulus definition of stress, but the variation in usage was
variety of working conditions have been studied as examples of considerable. Most researchers clarified their definitions
stresses, including role conflict, role ambiguity, role overload within a single article, however.
(e.g., Beehr, Walsh, & Taber, 1976; Kahn, Wolf, Quinn, Snoek, & Even though researchers often defined the word stress pre-
Rosenthal, 1964), the danger of accidents, and the pace of work cisely within a single article, the variation in usage is a problem
(e.g, Frese, 1985). for two reasons. First, it simply makes it more difficult to inte-
In contrast, a response definition of stress is associated with
grate research findings. Furthermore, because the advance-
what was referred to earlier as strain. Stress is an individual's ment of any field is based on the accumulation of research
response to work-related environmental stressors. Selye (1976)
findings, this could be one of the reasons why psychologists'
refers to stress as the reaction of the organism, which can be
understanding of occupational stress has progressed slowly. Al-
psychological, physiological, or behavioral.
though occupational stress is important, a common meaning
Finally, stress may be defined within a stimulus-response
for the word stress has proved elusive (Ilgen, 1990; Kasl, 1984).
approach. Researchers who refer to stress in this way refer to
A second, at least as serious, problem with the varied use of
the interaction between environmental stimuli (job stressors)
and individual responses (strains; Beehr & Franz, 1987; Schuler, the word stress relates to its use by the general public. Most
organizational psychology research on work stress has used
self-reports of organizational members to measure some or
even all variables in the job-stress process. Self-report mea-
We wish to thank Maria Tinsley and Dave Gudanowski for their
assistance with data analysis. We would also like to thank Paul Spector sures of stressors ask people to describe their work situation,
and three anonymous reviewers for their helpful comments on an ear- whereas self-report measures of strains ask people to describe
lier version of this article. their reactions to these stressors. Some researchers, however,
Correspondence concerning this article should be addressed to use the word stress or its close cognates (e.g, stressor, stressful)
Steve M. Jex, Department of Psychology, Sloan Hall, Central Michi- in questionnaire items. Table 1 presents examples of such items,
gan University, Mt. Pleasant, Michigan 48859. along with the constructs they are purported to measure. As
623
624 S. JEX, T. BEEHR, AND C. ROBERTS

Table 1
A Summary of Stress Items and Their Sources
Total no.
Source Sample item of items Construct measured
Schaubroeck, Cotton, & Jennings My job (e.g., the type of work, the amount of responsibility, etc.) 3 Job tension
(1989) causes me a great deal of stress and anxiety.
Relations with people I work with (e.g., co-workers, supervisors,
subordinates) cause me a great deal of stress and anxiety.
General aspects of the organization I work for (e.g., policies and
procedures, general working conditions) tend to cause me a
great deal of anxiety and stress.
Hendrix, Ovalle, & Troxler (1985) I feel a great deal of stress and anxiety in the performance of my 10 Perceived stress
job.
My life away from home is extremely tense and stressful.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Hendrix(1987) To what extent do you have a great deal of stress on the job? 5 Job stress
This document is copyrighted by the American Psychological Association or one of its allied publishers.

To what extent is your job stressful as a whole?


To what extent do you have job-related stress?
To what extent does your organization cause you stress?
To what extent is your life away from the job stressful?
Motowidlo, Packard, & Manning I feel a great deal of stress because of my job. 6 Stress
(1986) Very few stressful things happen to me at work.
My job is extremely stressful.
Stefry& Jones (1988) Working with people in my job is stressful. 21 Job stress
Fimian(1984) Subjects (teachers) were presented with 42 job conditions with 42 Unknown
the following instructions:
Rate the degree to which items relate to individual concepts of
teacher stress. . . .
Sykes& Eden (1985) Subjects (medical students) were presented with 4 general areas 25 Unknown
of concern (stress of medical school, maintenance tasks,
adapting to a new culture and environment, and interpersonal
stresses) with the following instructions:
Different people experience different degrees of stress. Given
your capacities, time limitations, and conflicting pressures,
how stressful is each of the following tasks and conditions for
you?

can be seen, some of these items clearly refer to individuals' thereby risking spuriously high correlations between stressors
responses or strains (e.g., "I feel a great deal of stress because of and strains.
my job"; Motowidlo, Packard, & Manning, 1986), and others
appear to focus on stressors or stimuli (e.g., "how stressful is Current Study
each of the following tasks and conditions for you?"; Sykes &
Eden, 1985). Unfortunately, many items are more ambiguous The current study was designed to determine the meaning of
and can be interpreted by respondents as referring to either the word stress to survey respondents. To accomplish this ob-
stressors or strains. An example is, "To what extent do you have jective, we administered 16 stress items to survey respondents.
job-related stress?" (Hendrix, 1987). Researchers' use of these Work-related stressors (role ambiguity, role conflict, perceived
items has been based on the assumption that respondents inter- workload, and interpersonal conflict) and psychological
pret the word stress in the intended manner. strains (job dissatisfaction, anxiety, frustration, depression, and
If respondents have interpreted the word stress in a manner turnover intent) were also measured with self-reports. Canoni-
consistent with researchers' interpretation (e.g., as a stressor), cal correlations compared the variance shared by stress items
the validity of the studies' conclusions probably have not been and stressors with the variance shared by stress items and
affected. If, however, respondents have interpreted the word strains. Judgments about respondents' interpretations of the
stress to mean something different, this could cause serious word stress were then based on the degree to which stress items
problems. There has been little previous research on the ways shared variance with either stressors or strains.
in which survey respondents interpret the word stress. Chen
(1989), however, modified Peters and O'Connor's (1980) frus- Method
tration scale by substituting the word stress for frustration. The
correlation between the modified scale and the original frus- Subjects
tration scale was .77, suggesting that respondents tended to Two independent samples contained a total of 245 subjects; 131 were
interpret stress in terms of their affective reactions (frustration). city and county employees and 114 were hospital employees in Michi-
Research using stress words in self-report measures of stressors gan. The average age of all employees was 40 (with a range of 18 to 67
might be contaminating the measurement of the theoretically years), 77% were women, and the average organizational tenure was 90
causal variable (stressor) with the outcome variable (strain), months (range, 1 to 312 months). The samples were combined for analy-
OCCUPATIONAL STRESS ITEMS 625

sis because they were similar in terms of age, gender composition, and within 48 hr. Questionnaires not completed within 48 hr were col-
organizational tenure. In addition, all analyses were run separately for lected one week after the distribution date. Of the 345 questionnaires
the two samples; the results were quite similar. distributed, 131 usable questionnaires were returned (a response rate
of 38%). This low response rate was most likely due to absences and to
employees' being occupied with job responsibilities.
Measures For the hospital sample, various times were set aside during two days
Stressors. Stressors included role ambiguity, role conflict, per- for Steve M. Jex to administer the questionnaire to employees on all
ceived workload, and interpersonal conflict. Role ambiguity was mea- shifts. Employees at each of these sessions were briefed about the pur-
sured with a four-item index taken from Beehr et al. (1976). Role con- pose of the study (although none of the respondents in either sample
flict was measured with Rizzo, House, and Lirtzman's (1970) index were told that interpretation of the word stress was being examined)
(eight items). Perceived workload was measured with a seven-item in- and told that responses would be confidential. Of the approximately
dex developed by Spector, Dwyer, and Jex (1988), which combines 500 employees in the hospital, 113 (23%) provided usable data. This
items from Arsenault and Dolan (1983), Caplan (1971), Mayes, Gan- low response rate was largely due to the inability of many employees to
ster, Sime, and Tharp (1984), and Payne and Fletcher (1983). Interper- get away from their worksites to attend the data collection sessions.
sonal conflict on the job was measured with a 4-item index from Spec- Despite the low response rate, the sample appeared to be quite repre-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

tor (1987). sentative of the hospital employee population in terms of demo-


This document is copyrighted by the American Psychological Association or one of its allied publishers.

Psychological strain. Psychological strains included overall job dis- graphics and job titles.
satisfaction, anxiety, frustration, depression, and turnover intent.
Overall job dissatisfaction was assessed with the 3-item overall job
satisfaction index of the Michigan Organizational Assessment Scale Results
(Cammann, Fichman, Jenkins, & Klesh, 1979). A modified version of
the 10-item state index of Spielberger's (1979) State-Trait Personality Results are presented in three parts. First, descriptive statis-
Inventory was used to measure anxiety. Specifically, subjects were tics and reliabilities are presented for all variables. Second, an
asked how they had generally felt at work during the past 30 days. analysis of the correlations is presented. Third, a summary of
Frustration was measured with Peters and O'Connor's (1980) 3-item the canonical correlation analysis is presented.
index assessing overall frustration with one's job. Depression was mea-
sured with a 10-item index taken from Quinn and Shepard (1974),
which was based on a 20-item scale originally developed by Zung
(1965). Descriptive Statistics
Turnover intentions were measured with a three-item scale devel-
oped by Beehr and O'Driscoll (1990). One of these items reflected the Descriptive statistics (sample sizes, means, standard devia-
degree to which subjects were thinking about quitting their jobs, and tions, and reliabilities) for all variables are presented, along
the other two reflected subjects' plans to seek alternative employment. with correlations, in Table 2. The observed ranges for most
Stress items. To maximize the relevance of the present study's re- variables were close to the possible range; thus, restriction of
sults to previous research, we derived 16 items containing the word range did not appear to be a problem. Coefficient alphas for the
stress (or its close cognates, e.g., stressful) directly from the items pre- stressor and strain scales were reasonably high, with most being
sented in Table 1. We chose items that were not specific to any one close to .70 or higher. Because the randomly formed stress item
occupation (e.g., teachers) or type of organization (e.g., the military). It scales were highly correlated (52 to .71), a composite measure
must be noted, however, that these items were taken out of the original consisting of all 16 items was formed. The coefficient alpha for
context in which they were used. The items were randomly placed into this composite was .90.
groups of four, and these groups were randomly situated among the
stressor and strain measures. The first group of items preceded the
workload scale; the second group followed the ambiguity measure; the
third group was placed between two personality scales not included in Correlations
the analysis for the present study; and the fourth scale followed the
turnover intent measure. The purpose of this arrangement was to avoid As can be seen in Table 2, intercorrelations among Stressors
having item order imply a specific meaning to respondents. There has were all significant, ranging from .21 to .49. Psychological
been little research on the effects of item order in employee surveys strains were also all significantly intercorrelated, ranging from
(Roberson & Sundstrom, 1990). Spector and Michaels (1983) found no .22 to .56.
order effect, but Roberson and Sundstrom (1990) found an effect for All except one of the correlations between Stressors and psy-
ordering based on importance of the information to employees. The chological strains were significant. The strongest relations were
issue of the ambiguity of the items has not been addressed previously, those between workload and frustration (.56) and between role
however, and we therefore deemed it wise to be concerned about item ambiguity and anxiety (.52). Stressors were strongly related to
order. Respondents were asked to indicate their agreement with each
item on a 7-point scale ranging from 1 (strongly disagree) to 7 (strongly the stress item composite (.35 to .60), with the strongest correla-
agree). tion involving role conflict. Psychological strains were also
strongly related to the stress item composite (.39 to .72), with
the strongest correlation involving anxiety. In general, the mag-
Procedure nitude of these correlations was only slightly stronger than that
For the county and city employee sample, questionnaires were dis- of the correlations between Stressors and the stress items. Thus,
tributed at the work site by Cathlyn K. Roberts. A cover letter stated the zero-order correlations provide some information as to re-
that responses would be confidential and that participation was volun- spondents' interpretation of the stress items. To shed more light
tary. Employees were instructed to keep completed questionnaires at on this issue, we undertook a summary analysis of the individ-
their worksite so that they could be picked up by Cathlyn K. Roberts ual stress items.
626 S. JEX, T. BEEHR, AND C. ROBERTS

Table 2
Descriptive Statistics and Intercorrelations Among All Variables
Variable N M SD 1 10
Stressors
1. Role ambiguity 237 3.2 1.3 (.77)
2. Role conflict 236 3.0 0.8 .46 (.87)
3. Perceived workload 233 3.4 2.1 .21 .49 (.81)
4. Interpersonal conflict 242 2.1 0.6 .30 .45 .36 (.72)
Strains
5. Satisfaction 243 4.8 1.1 -.25 -.29 -.12 -.20 (.87)
6. Anxiety 238 2.0 0.6 .35 .52 .41 .43 -.38 (.86)
7. Frustration 239 3.8 1.1 .24 .39 .56 .35 -.24 .56 (.67)
8. Depression 231 1.8 0.4 .36 .32 .19 .28 -.39 .55 .35 (.80)
9. Turnover intent 240 2.8 1.6 .32 .44 .14 .22 -.55 .37 .22 .36 (.87)
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Stress items
This document is copyrighted by the American Psychological Association or one of its allied publishers.

10. Composite 234 4.4 1.4 .35 .60 .48 .48 -.44 .72 .59 .49 .39 (.90)
Note. Reliabilities (pt) are presented on the diagonal. Correlations larger than . 12 are significant beyond the .05 level.

Canonical Correlation Analysis items dealt with general aspects of the organization as a source
of stress, the degree to which stressful things happen at work,
Three canonical correlation analyses were computed. Stress opportunities for growth, getting respect from one's supervisor,
items were first correlated with strains and then with stressors. and earning enough money to maintain one's lifestyle. All
Finally, stressors were correlated with strains. This procedure strains except anxiety were correlated with this root. The stan-
used the individual items rather than the composite, and it dardized canonical coefficients indicated that the shared vari-
provided summary statistics about overall relationships that ance between stress items and strains was due primarily to
could be compared (i.e., relationships of stress items with frustration and turnover intent (-.62 and —.58, respectively).
strains, stress items with stressors, and stressors with strains). The redundancy index for stress items was .02, indicating
The objective of this analysis was to compare the variance that 2% of the variance of the second strain root was accounted
shared between stress items (as a set) and strains with the vari- for by the stress items. The redundancy index for the strain
ance shared between stress items and stressors. If stress items scales was .07, indicating that 7% of the second stress-item root
tend to be interpreted primarily as either strains or stressors, was accounted for by strains. In total, strains accounted for 40%
they should share more variance with one of these sets of mea- of the variance in the stress items.
sures than with the other. Furthermore, the variance shared Stress items and stressors. In the canonical analysis relating
should be substantially more than that between stressors and stress items to stressors, two significant (p < .05) canonical
strains. roots were extracted. The canonical correlations were .72 and
Stress items and strains. In the canonical analysis relating .57, respectively.
the set of stress items to the set of strains, three significant (p < The first root for stress items was composed of all items ex-
.05) canonical roots were extracted from each set. The canoni- cept one. All of the stressor measures were strongly correlated
cal correlations for the first three pairs of roots were .84, .55, with the first root. The standardized coefficients indicated that
and .37, respectively. Because the variance accounted for in shared variance between stress items and stressors was due to
each set of variables by the third root was quite small, only the all stressors except role ambiguity. The redundancy index for
first two were interpreted. stress items was .20, whereas the redundancy index for the
To interpret these results, we examined component loadings stressor scales was .27.
(correlations), standardized canonical coefficients, and redun- The second root for stress items consisted of a total of three
dancy indices for the first two roots extracted from both sets of items. These dealt with relations with co-workers as a source of
variables. Using a criterion of component loadings greater than stress, one's job in general as a source of stress, and opportuni-
or equal to .30 for interpretation (Tabachnick & Fidell, 1983), ties for growth as a source of stress. All of the stressors except
the first root for stress items was composed of all items except role conflict correlated strongly with this root, although the
two. All of the strain measures correlated strongly with the first strongest correlation was clearly with perceived workload. The
root. The standardized coefficients indicated that frustration standardized canonical coefficients indicated that the shared
and anxiety contributed most to the shared variance between variance between stress items and stressors was due primarily
stress items and strains (—.44 and —.54, respectively). to perceived workload (.98). The redundancy index for stress
The redundancy index for stress items was .26, indicating items was .02, and the redundancy index for stressors was .06.
that 26% of the variance of the strain root was accounted for by Stressors accounted for a total of 33% of the variance in stress
the stress items. The redundancy index for the strain scales was items.
.33, indicating that 33% of the stress-item root was accounted Stressors and strains. In the canonical analysis relating
for by strains. stressors and strains, three significant (p < .05) roots were ex-
The second root for stress items consisted of five items. These tracted. The canonical correlations were .70, .42, and .22, re-
OCCUPATIONAL STRESS ITEMS 627

spectively. Because the variance accounted for by the third root pressure is a source of stress."). However, attributions for emo-
was quite small, only the first two were interpreted. tional states may not be very accurate and can be manipulated
The first root for stressors consisted of all four stressors. Simi- rather easily (Nisbett & Schachter, 1966). Thus, these items may
larly, all five strains were correlated with this root. The standard- present respondents with an attribution task that they are inca-
ized canonical coefficients indicated that the shared variance pable of performing accurately.
between stressors and strains was due to all stressors, although If the word stress has an affective component, as seems likely,
the contributions of role conflict and workload were strongest this is consistent with the stress appraisal process proposed by
(.49 and .41, respectively). The redundancy index for stressors Lazarus (e.g. Lazarus, 1966; Lazarus & Launier, 1978). A stress-
was 22%, and the redundancy index for strains was 20%. ful situation in Lazarus's theory includes the person's evalua-
The second root for stressors consisted of all stressors except tion of the situation as threatening, and the word stress may be
interpersonal conflict. All strains except anxiety and depres- a term many people associate with threat. If so, rather than
sion were correlated with this root. The standardized canonical being an objective measure of environment, a stressor measure
coefficients indicated that the shared variance between stress- using the word stress probably taps a postappraisal evaluation
ors and strains was due primarily to workload (.99). The redun- by the employee.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

dancy index for stressors was 3%, as was the redundancy index The use of the word stress probably has not been prevalent
This document is copyrighted by the American Psychological Association or one of its allied publishers.

for strains. Stressors accounted for a total of 25% of the variance enough in the stressor and strain measures in the occupational
in strain measures. stress literature as yet to have led to serious biases in the magni-
Summary. The canonical correlation analysis revealed tude of correlations, but as found in the literature review, this
that, as a set, stress items shared slightly more variance with practice seems to have become more common in recent years.
strains than with stressors. However, the shared variance be- In addition, the field of occupational stress is multidisciplinary,
tween stress items and stressors was larger than that between including not only industrial and organizational psychology
stressors and strains. Thus, important variance was also shared but also medicine, engineering, and clinical psychology, among
between stress items and stressors. Stressors accounted for 25% others (Beehr & Franz, 1987). The degree to which the word
of the variance in the strain measures. Strain measures ac- stress has been used in survey items in these other disciplines is
counted for a total of 40% of the variance in stress items, and uncertain, and thus the possible bias in results could be greater
stressors accounted for 33%. than would be imagined based on the industrial and organiza-
tional research reviewed here. It is therefore a methodological
Discussion problem of concern in occupational stress research.
The most obvious conclusion is that the use of the word stress
The purpose of this study was to ascertain the meaning of the in survey items should be avoided because of potential con-
word stress to survey respondents when this word is included in founding and other difficulties. There are many scales available
survey items. To this end, two types of evidence are offered. (with acceptable psychometric properties) that measure both
First, the intercorrelations among stressors, strains, and a com- job-related stressors and employees' strains without using the
posite stress-item scale were examined. This analysis showed word stress. Because the validity of occupational stress research
that the stress scale correlated strongly with both stressors and is so heavily dependent on measurement integrity (Jex & Beehr,
strains. The strongest correlations appeared to be with strains, 1991), this represents a very simple way to eliminate a poten-
particularly anxiety. tially serious problem and improve the overall quality of occu-
Second, the canonical correlation analysis showed that stress pational stress research.
items, as a set, shared more variance with strains than stressors.
Both canonical correlations and redundancy indices showed References
that, although the variance shared between stress items and
strains was somewhat greater than that shared with stressors, it Arsenault, A., & Dolan, S. (1983). The role of personality, occupation,
was clear that stress items were associated with both stressors and organization in understanding the relationship between job
and strains. If researchers use stress items in either stressor or stress, performance, and absenteeism. Journal of Occupational Psy-
chology, 56, 227-240.
strain indexes, there is the potential for confounding of the Beehr, T. A., & Bhagat, R. S. (1985). Introduction to human stress and
measures. Including such stress items in both stressor and cognition in organizations. In T. A. Beehr & R. S. Bhagat (Eds.),
strain scales will, of course, lead to the most serious confound- Human stress and cognition in organizations: An integrated perspec-
ing problems. tive (pp. 3-19). New York: Wiley.
One issue essentially boils down to how confident one can be Beehr, T. A., & Franz, T. (1987). The current debate about the meaning
that the researcher's interpretation of such stress items matches of job stress. In J. M. Ivancevich & D. C. Ganster (Eds.), Job stress:
the survey respondents'. One should not be too optimistic, be- From theory to suggestion (pp. 5-18). New York: Haworth Press.
cause in most of the 1985-1989 research articles we reviewed, Beehr, T. A., & Newman, J. E. (1978). Job stress, employee health, and
we found that researchers were more likely to use the word organizational effectiveness: A facet analysis, model, and literature
stress to mean stressor. Only 22% used the word stress to refer review. Personnel Psychology, 31, 665-699.
Beehr, T. A., & O'Driscoll, M. P. (1990, May). Employee uncertainty as a
to responses or strains. Even if researchers' and respondents' factor in occupational stress. Paper presented at the 62nd Annual
interpretations were similar, stress items may be problematic Meeting of the Midwestern Psychological Association, Chicago.
for other reasons. For example, many of the items in this study Beehr, T. A., Walsh, J. T, & Taber, T. D. (1976). Relationship of stress to
required respondents to determine which aspects of the work- individually and organizationally valued states: Higher order needs
place are a source of stress (e.g., "Working under constant time as a moderator. Journal of Applied Psychology, 61, 41 -47.
628 S. JEX, T. BEEHR, AND C. ROBERTS

Cammann, C, Fichman, M., Jenkins, D., & Klesh, J. (1979). The Michi- tional stress: Its causes and consequences for job performance. Jour-
gan Organizational Assessment Questionnaire. Unpublished manu- nal of Applied Psychology, 71, 618-629.
script, University of Michigan, Ann Arbor. Nisbett, R. E., & Schachter, S. (1966). The cognitive manipulation of
Caplan, R. D. (1971). Organizational stress and individual strain: A pain. Journal of Experimental Social Psychology, 2, 227-236.
psychosocial study of risk factors in coronary heart disease among Payne, R., & Fletcher, B. C. (1983). Job demands, supports, and con-
administrators, engineers, and scientists (Doctoral dissertation, Uni- straints among schoolteachers. Journal of Vocational Behavior, 22,
versity of Michigan, Institute for Social Research, 1971). (University 136-147.
Microfilms No. 72-14822) Peters, L. H., & O'Connor, E. J. (1980). Situational constraints and
Chen, P. Y. (1989). Relationships of frustratorswith affective, behavioral, employee affective reactions: The influences of a frequently over-
and physical reactions: The effects of potential moderators. Unpub- looked construct. Academy of Management Review, 5, 391-397.
lished master's thesis, University of South Florida, Tampa. Quinn, R. P., & Shepard, L. J. (1974). Quality of Employment Survey.
Fimian, M. J. (1984). The development of an instrument to measure Ann Arbor: University of Michigan, Institute for Social Research.
occupational stress in teachers: The Teacher Stress Inventory. Jour- Rizzo, J. R., House, R. J., & Lirtzman, S. I. (1970). Role conflict and
nal of Occupational Psychology, 57, 277-293. role ambiguity in complex organizations. Administrative Science
Frese, M. (1985). Stress at work and psychosomatic complaints: A Quarterly, 15,150-163.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

causal interpretation. Journal of Applied Psychology, 70, 314-328. Roberson, M. X, & Sundstrom, E. (1990). Questionnaire design, re-
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Hendrix, W H. (1987). Organizational and Individual Assessment Sur- turn rates, and response favorableness in an employee attitude ques-
vey. Clemson, SC: Clemson University. tionnaire. Journal of Applied Psychology, 75, 354-357.
Hendrix, W H., Ovalle, N. K., & Troxler, R. G. (1985). Behavioral and Schaubroeck, J., Cotton, J. L., & Jennings, K. R. (1989). Antecedents
physiological consequences of stress and its antecedent factors. Jour- and consequences of role stress: A covariance structure analysis.
nal of Applied Psychology, 70,188-201. Journal of Organizational Behavior, 10, 35-58.
Ilgen, D. R. (1990). Health issues at work: Opportunities for industrial/ Schuler, R. S. (1980). Definition and conceptualization of stress in
organizational psychologists. American Psychologist, 45, 273-283. organizations. Occupational Behavior and Human Performance, 25,
Ivancevich, J. M., & Matteson, M. T. (1980). Stress and work. Glenview, 184-215.
IL: Scott, Foresman. Selye, H. (1976). The stress of life (2nd ed.). New York: McGraw-Hill.
Jex, S. M., & Beehr, T. A. (1991). Emerging theoretical and methodolog- Specter, P. E. (1987). Interactive effects of perceived control and job
ical issues in the study of work-related stress. In G. R. Ferris & K. W stressors on affective reactions and health outcomes for clerical
Rowland (Eds.), Research in personnel and human resources manage- workers. Work & Stress, 1, 155-162.
ment (pp. 311-365). Greenwich, CT: JAI Press. Spector, P. E., Dwyer, D. J., & Jex, S. M. (1988). Relations of job stress-
Kahn, R. L., & Quinn, R. P. (1970). A framework for analysis. In A. ors to affective, health, and performance outcomes: A comparison
McLean (Ed.), Occupational mental health (pp. 50-115). New York: of multiple data sources. Journal of Applied Psychology, 73,11-19.
Rand McNally. Spector, P. E., & Michaels, C. E. (1983). A note on item order as an
Kahn, R. L., Wolfe, D. M., Quinn, R. P., Snoek, D., Jr., & Rosenthal, artifact in organizational surveys. Journal of Occupational Psychol-
R. A. (1964). Organizational stress: Studies in role conflict and ambi- ogy, 56, 35-36.
guity. New York: Wiley. Spielberger, C. D. (1979). Preliminary manual for the State-Trait Person-
Kasl, S. V (1984). Stress and health. Annual Review of Public Health, 5, ality Inventory (STPI). Unpublished manuscript, University of South
319-342. Florida, Tampa.
Lazarus, R. S. (1966). Psychological stress and the coping process. New Steffy, B. D., & Jones, J. W (1988). Workplace stress and indicators of
York: McGraw-Hill. coronary disease risk. Academy of Management Journal, 31, 686-
Lazarus, R. S., & Launier, R. (1978). Psychological stress and the cop- 698.
ing process. In L. A. Pervin & M. Lewis (Eds.), Internal and external Sykes, I. J., & Eden, D. (1985). Transitional stress, social support, and
determinants of behavior (pp. 287-327). New York: Plenum Press. psychological strain. Journal of Occupational Behavior, 6, 293-298.
Tabachnick, B. G., & Fidell, L. S. (1983). Using multivariate statistics.
Mayes, B. T, Ganster, D. C., Sime, W E., & Tharp, G. D. (1984, August).
Philadelphia, PA: Harper & Row.
A multivariate test of a general model of job stress. Paper presented at
Zung, W R. K. (1965). A self-rating depression scale. Archives of Gen-
the 44th Annual Meeting of the Academy of Management, Boston,
eral Psychiatry, 12, 63-70.
MA.
McGrath, J. E. (1976). Stress and behavior in organizations. In M. D.
Dunnette (Ed.), Handbook of industrial and organizational psychol- Received June 12,1991
ogy (pp. 1351-1395). Chicago: Rand McNally. Revision received January 21,1992
Motowidlo, S. J., Packard, J. S., & Manning, M. R. (1986). Occupa- Accepted January 28,1992 •

You might also like