You are on page 1of 6

Journal of Applied Psychology

1976, Vol. 61, No. 1, 35-40

Perceived Situational Moderators of the Relationship Between


Subjective Role Ambiguity and Role Strain
Terry A. Beehr
Illinois State University
Previous research has focused on personality moderators of the relationship
between organizational stress and individual strain. This study was a search
for situational moderators of the relationship between one organizational stress,
role ambiguity, and four psychological strains: job dissatisfaction, life dis-
satisfaction, low self-esteem, and depressed mood. Three situational character-
istics were hypothesized to moderate the relationship by reducing its strength:
group cohesiveness, supervisor support, and autonomy. Group cohesiveness
moderated the relationship between role ambiguity and two of the role strains,
but the direction of its moderating influence was inconsistent. An explanation
was offered for this result. Supervisor support showed .a nonsignificant tendency
to reduce the strength of the relationship between role ambiguity and role
strain. Autonomy tended to moderate the relationship in the expected direc-
tion significantly and strongly.

Role stress is any aspect of role expecta- to test situational characteristics as moder-
tions that has adverse consequences (role ators of the relationship between role stress
strain) for individual role incumbents (Kahn and strain.
Si Quinn, 1970). Most research on role stress In this study, three situational character-
has focused on samples representative of istics were hypothesized as moderators of the
limited populations, for example, a single role relationship between role stress and role
or a small number of organizational roles strain. Group cohesiveness and supervisor
(French & Caplan, 1973; Lyons, 1971; Paul, support were proposed as moderators because
1974; Rizzo, House, & Lirtzman, 1970; Cap- of the argument by Kahn and Quinn (1970)
lan & Jones, Note 1). that psychological support in the presence of
In addition to the use of narrow samples, stress may reduce role strain. (Presumably,
most previous research has been limited to cohesive groups are more supportive than
the investigation of the direct, overall rela- noncohesive groups.) Autonomy also was
tionship between role stress and role strain. hypothesized as a moderator. If the role ex-
A few researchers have tested personality pectations that others send to the focal per-
characteristics (Johnson & Stinson, 197S; son are stressful, people with job autonomy
Kahn, Wolfe, Quinn, Snoek, & Rosenthal, may define their own role expectations, re-
1964; Lyons, 1971; Caplan Si Jones, Note 1) sulting in reduced role strain.
as moderator variables in a role stress model. This study added to the knowledge from
There has been no serious attempt, however, previous studies of role stress in two ways.
First, the variety of roles and of organizations
was greater, so that the results would be
This study is based on the author's doctoral dis- representative of phenomena that are widely
sertation, "Role Ambiguity as a Role Stress: Some dispersed in the world of work. Second, situa-
Moderating and Intervening Variables," which was
completed at The University of Michigan.
tional rather than personality moderators
The author wishes to thank Edward Bordin, were examined. Role ambiguity, that is, un-
Robert Cooke, Jack Drexler, Larry Michaelsen, certainty regarding what is expected on one's
Robert Quinn, and Stanley Seashore for their help job, was chosen for analysis because of the
during various stages of work on this article. long history of psychological research on am-
Requests for reprints should be sent to Terry A.
Beehr, Department of Psychology, De Garmo Hall, biguity, and because of the Kahn et al.
Illinois State University, Bloomington-Normal, Illi- (1964) rinding that role ambiguity is felt to
nois 61761. a stressful extent by about one third of the
35
36 TERRY A. BEEHR

American male wage and salary workers. Four Since the present study is seeking differences in the
psychological role strains were measured: job strengths of relationships, however, the results prob-
ably are conservative, that is, more significant dif-
dissatisfaction, life dissatisfaction low self- ferences might have been found if multiple sources
esteem, and depressed mood. of data had been used.
The specific hypotheses were that role am- Role ambiguity (If =1.92, SD = ,64) was mea-
biguity is more strongly related to the role sured by a four-item index:
strains (a) among people in noncohesive How true is this of your job?
groups than among people in cohesive groups, I can predict what others will expect of me
(b) among people with nonsupportive super- tomorrow.
visors than among people with supportive I am clear on what others expect of me on my
job.
supervisors, and (c) among people in non- On my job, whatever situation arises, there are
autonomous roles than among people in procedures for handling it.
autonomous roles. I get enough facts and information to work my
best.
METHOD
All items except the third (On my job . . .) were
Sample taken from the Quality of Employment Survey
The sample comprised 651 respondents employed (Quinn & Shepard, 1974). The response categories
35 hours or more per week by five midwestern work were "very true," "somewhat true," "a little true,"
organizations: a printing company (« = 173), a and "not at all true."
small research and development company (n = 21), Four psychological strains were measured. Pre-
two automotive supply companies (ns — 120 and sumably, psychological strains are more accurately
124), and four services departments of a hospital measured by interviews than physiological strains
( n — 2 1 3 ) . These organizations were chosen in part would be. The first, job dissatisfaction (M = 3.94,
on the criterion that they contain a diverse set of SD = .97), was measured by a four-item subset of
jobs, and there is evidence that the occupational Quinn and Shepard's (1974) global job satisfaction
census codes of the respondents roughly match the index:
codes of a national sample of workers (Beehr, 1974).
All supervisors were included in the sample, and All in all, how satisfied would you say you are
nonsupcrvisors were sampled systematically at rates with your job—very satisfied, somewhat satis-
varying from organization to organization (between fied, not too satisfied, or not at all satisfied?
25% and 100%). The response rate was 72.9%. In general, how well would you say that your job
measures up to the sort of job you wanted when
Fifty-one percent of the respondents were male, you took it? Would you say it is very much
79.8% were white, 68.5% were married, 73,0% had like, somewhat like, or not very much like the
finished at least high school, and the mean age was job you wanted when you took it?
35 years. Comparable information on the nonre- Knowing what you know now, if you had to de-
spondents was not available. cide all over again whether to take the job you
now have, what would you decide? Would you
Measures decide without any hesitation to take the same
job, would you have some second thoughts, or
Data were collected during a 90-minutc structured would you decide definitely not to take the
interview in the respondent's home conducted during same job?
the winter of 1972-1973 by interviewers from the If a good friend of yours told you that he/she
Survey Research Center. The data were collected was interested in working in a job like yours,
for another study (Cammann, Quinn, Beehr, & what would you tell him/her? Would you
Gupta, Note 2) unrelated to the present study. strongly recommend this job, would you have
The measures of all variables in the study were doubts about recommending it, or would you
subjective, that is, they were perceptions of the sub- strongly advise him/her against this sort of job?
jects. It could be argued, however, that the percep-
tions of the subject are important in stress research, The job dissatisfaction index had a possible range
because strain may be largely an individual's reac- of 1-5.
tion to his subjective environment. In addition, some The life dissatisfaction measure 01'/ = 4.8', SD =
of the environmental variables in this study have 1.25) consisted of a nine-item subset of Quinn and
a partially subjective nature. For example, regardless Shepard's (1974) life satisfaction index. Seven of
of a supervisor's objective behavior, it cannot be said these items were answered on a semantic differential
that a subordinate is given true psychological support scale. The seven pairs of anchors were "interesting-
unless the subordinate feels supported. boring," "enjoyable-miserable," "friendly-lonely,"
Since all of the data were gathered during a single "full-empty," "hopeful-discouraging," "rewarding-
interview, there is a possibility that relationships dissappointing," and "brings out the best in me-
could bo inflated by correlated method variance. doesn't give me much of a chance" in response to
SITUATIONAL MODERATORS 37

the question, "How do you feel about your present score of 1 on the group cohesiveness index. The re-
life in general?" The fixed-alternative questions were: sponse categories for the last four questions were
coded 2 through 5, and respondents answering those
In general, how satisfying do you find the ways questions received their mean score on those four
you're spending your life these days? Would items as their score on the group cohesiveness mea-
you call it completely satisfying, pretty satisfy- sure. Thus, a high score on the index meant that
ing, or not very satisfying? the respondent was part of a very cohesive group,
Taking all things together, how would you say and a low score indicated that he was part of no
things are these days? Would you say you're group at all or of a group with little cohesiveness.
very happy, pretty happy, or not too happy Supervisor support (M — 3.12, SD = .81) was mea-
these days? sured by three items:
The life dissatisfaction index had a possible range How true is this of your supervisor?
of 1-7. My supervisor keeps informed about the way
Low self-esteem (M = .99, SD = 1.05) was mea- his/her people think and feel about things.
sured by a three-item subset of Quinn and Shepard's My supervisor stands up to outsiders for the
(1974) 7-point semantic differential scale. The an- people he/she supervises.
chors were "successful-unsuccessful," "important- My supervisor takes a personal interest in those
not important," and "doing my best-not doing my he/she supervises.
best" in response to the question, "How do you see
yourself in your work?" These items were taken from Hemphill and Coombs
Depressed mood (jV/=1.8S, SD = .47) was mea- (1957), and the response categories were "very true,"
sured by a 10-item index taken from Quinn and "somewhat true," "a little true," and "not at all
Shepard (1974). true."
Autonomy (M = 2.97, SD — .75) was measured by
How often do you feel this way at work? four items:
I feel downhearted and blue.
I get tired for no reason. How true is this of your job?
I find myself restless and can't keep still. I have a lot of say over what happens on my
I find it easy to do the things I used to do. job.
My mind is as clear as it used to be. I have enough authority to do my best.
I feel hopeful about the future. My job allows me to make a lot of decisions on
I find it easy to make decisions. my own.
I arn more irritable than usual. I have enough freedom as to how I do my work.
I still enjoy the things I used to.
I feel that I am useful and needed. These items were taken from Quinn and Shepard
(1974), and the response categories were "very true,"
The response categories were "often," "sometimes," "somewhat true," "a little true," and "not at all
"rarely," and "never." true." Table 1 contains the correlations among the
Three situational characteristics, that is, group variables for the whole sample. Although all of the
cohesiveness, supervisor support, and autonomy, were variables were obtained from the same interview,
measured. The first, group cohesiveness (^ = 3.65, none of the correlations were strong enough to sug-
SD = 1.19), was measured by five items: gest that any two variables were measuring identical
constructs.
Is there any group of people at [name of re-
spondent's organization] that you think of as
your .coworkers—people whom you see just Analyses
about everyday and with whom you have to
work closely in order to do your job well? I The sample was divided at the median on each
situational variable, and Pearson product-moment
know it's hard to talk about a whole group of
people like this, but I'd like to get your general correlations between role ambiguity and each role
views about the people in this group. strain were computed within each half sample.
Hypotheses were directional, that is, the relationship
How ready are they to defend each other from
between role ambiguity and each role strain was
criticism by outsiders?
How well do you feel they help each other on predicted to be stronger in the less cohesive than in
the more cohesive group; stronger in the low-super-
the job?
visor-support than in the high-supervisor-support
How well do they get along together?
group; and stronger in the low-autonomy than in
How well do they stick together ?
the high-autonomy group. Therefore, one-tailed tests
The last three items were adapted from Seashore of the significance of the difference between correla-
(1954). The response categories for the last four tions were computed.
items were."great, couldn't be better," "pretty good,
but some could be better," "not very good, but RESULTS
some pretty good," and "not very good." Respon-
dents answering no to the first question were not The first column of Table 1 shows that role
asked the other four questions and were assigned a ambiguity is significantly related to the four
38 TERRY A. BEEHR

TABLE 1
CORRELATIONS OP ROLE AMBIGUITY, STRAINS, AND SITUATIONAL MODERATORS

Item 1 2 3 4 5 6 7 8

Role stress
1. Role ambiguity (.69)
Role strains
2. Job dissatisfaction .22** (.80)
3. Life dissatisfaction .08* .27** (.79)
4. Low self-esteem .19** .25** .28** (.68)
5. Depressed mood .19** .43** .39** .31* (.71)
Situational moderators
6. Group cohesiveness -.12** -.14** -.02 -.03 -.05 (.78)
7. Supervisor support -.30** -.37** -.17** -.23** -.32** .13** (.78)
8. Autonomy -.26** -.37** -.21** -.27** -.32** .13** .36** (.74)

Nole. N = 587 and represents the number of respondents having complete data for all variables in the table. Numbers in par-
entheses are reliability estimates obtained by applying the Spearman-Brown propheey formula to the median interitem correlation.
*/> < .05.
** t> < .01.

role strains for the sample as a whole. Job job dissatisfaction (p < .10) in the direction
dissatisfaction is the strain most strongly opposite from that predicted. A two-tailed
correlated with role ambiguity. Life dissatis- test was used for this comparison because the
faction, as one would expect, has the weakest direction had not been predicted correctly.
relationship to ambiguity in the work role. Although three of the four differences be-
Obviously, there are many sources of life tween correlations are in the predicted direc-
satisfaction other than those in the work role. tion for the analysis of supervisor support as
Table 2 shows that group cohesiveness a moderator, none of the differences are sta-
moderates the relationship between role am- tistically significant.
biguity and low self-esteem as predicted (p < All differences between correlations of role
.05). However, group cohesiveness moderates ambiguity and role strain are in the predicted
the relationship between role ambiguity and direction when the sample is divided on
autonomy (see Table 3). The differences for job
TABLE 2 dissatisfaction and depression are significant
at the .01 level and the .10 level.
GROUP COHESIVENESS AS A MODERATOR OF THE RELA-
TIONSHIP BETWEEN ROLE AMBIGUITY AND PSY-
CHOLOGICAL ROLE STRAIN DISCUSSION
Correlations with This study suggested that even if there is
role ambiguity a stress (role ambiguity), people in roles with
certain situational characteristics, especially
Low High Signif-
autonomy, do not suffer from it. This has
group group icance of important implications because situational
cohe- cohe- difference characteristics can be modified by organiza-
sive- sive- between tions.
Role strain ness ness ri and ri
There was only suggestive evidence that
Job dissatisfaction .16* .29* p < .10 people with supportive supervisors might not
Life dissatisfaction .05 .15* US feel some role strains even if their roles are
Low self-esteem .24* .08 p < .05 ambiguous. Even if psychological support has
Depressed mood .19* .10 ns the expected beneficial effect, perhaps the
\'otf. For ri, n = 386. for >'9, n = 246; jts represent the num-
supervisor is not the best source of that sup-
ber of respondents having complete data for all variables in the port. Previous laboratory work (Schachter,
table.
*p < ,01. 1959) has shown that people expecting an
SITUATIONAL MODERATORS 39

TABLE 3 for the ambiguity on themselves. The fact


AUTONOMY AS A MODERATOR OF THE RELATIONSHIP that both dissatisfaction measures, that is,
BETWEEN ROLE AMBIGUITY AND PSYCHOLOGICAL job dissatisfaction and life dissatisfaction,
STRAIN tend to be correlated more strongly with role
ambiguity in the high-cohesive group while
Correlations
low self-esteem and depressed mood tend to
with role be correlated more strongly with role am-
ambiguity biguity in the low-cohesive group adds con-
Signifi- fidence to this interpretation.
n ri cance of Autonomy is the strongest and most con-
Low High difference
auton- auton- between sistent moderator of the relationship between
Role strain omy omy r\ and r-> role ambiguity and role strain. Organizations
that wish to reduce the role strain associated
Job dissatisfaction .25* .05 p <0.1 with role ambiguity should increase the
Life dissatisfaction .10 .03 ns autonomy in their members' roles. Future re-
Low self-esteem .19* .09 P < .10
Depressed mood .23* -.02 p < .01 search should attempt to determine whether
job characteristics such as autonomy reduce
Note. For r i , n = 340. for J'2. n = 296; ns represent the num-
ber of respondents having complete data for all variables in the
the relationship between a large number of
table. role stresses and role strain. If they do, re-
* p < .01.
designing jobs would seem even more likely
to reduce the strain due to role stress.
electrical shock prefer to spend their waiting
time with others like themselves, especially REFERENCE NOTES
with others expecting shocks. This indicates 1. Caplan, R. D., & Jones, K. W. Technological de-
that peers, especially those in similarly stress- pendency, role stress, and strum. Paper presented
ful roles, may be the most beneficial source at the meeting of the American Psychological
Association, New Orleans, September 1974.
of psychological support.
Group cohesiveness has an effect on the 2. Gupta,Cammann, C., Quinn, R. P., Beehr, T, A., &
N. Effectiveness in work roles Report 1:
extent to which role ambiguity is related to Validating, quality oj employment indicators (PB-
strain, but the results are contradictory. For 241 907/SWS). Ann Arbor: University of Michi-
persons in cohesive groups, the relationship gan, Survey Research Center, April 1975.
between role ambiguity and low self-esteem is
weaker than it is for those in noncohesive REFERENCES
groups. Job dissatisfaction, however, is re- Beehr, T. A, Role ambiguity as a role stress: Some
lated more strongly to role ambiguity in co- moderating and intervening variables. Unpublished
hesive than in noncohesive groups. doctoral dissertation, University of Michigan, 1974.
A partial explanation of these results stems French, J. R. P., & Caplan, R. D. Organizational
stress and individual strain. In A. J. Marrow
from the assumption that people in cohesive (Ed.), The failure of success. New York:
groups are more likely to communicate with AMACOM, 1973.
coworkers than are people in noncohesive Hemphill, J. K., & Coons, A. E. Development of the
leader behavior description questionnaire. In R. M.
groups. People perceiving ambiguous roles can Stogdil! & A. E. Coons (Eds.), Leader behavior:
blame the ambiguity either on their roles or Its description and measurement (Research Mono-
on themselves, Discussing the work situation Graph No. 88). Columbus: Bureau of Business
with peers may result in social support for Research, Ohio State University, 19S7.
external placement of the blame, that is, on Johnson, T. W., & Stinson, J. E. Role ambiguity,
vole conflict, and satisfaction: Moderating effects
the role rather than on the self. Thus, while of individual differences. Journal of Applied Psy-
people in cohesive groups still are dissatisfied chology, 1975, 60, 329-333.
with ambiguous roles, they are less likely to Kahn, R. L., & Quinn, R. P. Role stress: A frame-
internalize the blame in the form of lowered work for analysis. In A. McLean (Ed,), Occupa-
tional mental health. New York: Rand-McNally,
self-esteem. People in noncohesive groups are 1970.
not as likely to have such discussions and, Kahn, R. L., Wolfe, D. M., Quinn, R. P., Snoek,
therefore, are more likely to place the blame J. D., & Rosenthal, R. A. Organizational stress;
40 TERRY A. EEEHR

Studies in role conflict and ambiguity. New York: Survey of Working Conditions. Ann Arbor, Mich.:
Wiley, 1964. The Institute for Social Research, 1974.
Lyons, T. F. Role clarity, need for clarity, satisfac- Rizzo, J. R., House, R. J., & Lirtzman, S. K. Role
tion, tension, and withdrawal. Organizational Be- conflict and ambiguity in complex organizations.
havior and Human Performance, 19V1, 6, 99-110. Administrative Science Quarterly, 1970, 15, 150-
Paul, R. J. Role clarity as a correlate of satisfaction, 163.
job related strain and propensity to leave—male Schachter, S. The psychology oj affiliation. Stanford,
vs. female. The Journal oj Management Studies, Calif.: Stanford University Press, 1959.
1974, 11, 233-245. Seashore, S. E. Group cohesiveness in the industrial
Quinn, R. P., & Shepard, L. J. The 1972-1973 work group. Ann Arbor: University of Michigan
Quality of Employment Survey: Descriptive sta- Press, 1954.
tistics with comparison data from the 1969-1970 (Received June 9, 1975)

You might also like