^'Academy of Management Journal

1983, Vol. 26, No. 4, 587-595.

Job Satisfaction and

the Good Soldier:
The Relationship
Between Affect and
Employee ''Citizenship'^^
THOMAS S. BATEMAN
Texas A&M University
DENNIS W. ORGAN
Indiana University

A measure of a wide array of employee activities on

. thejob was completed by employees'supervisors at two
points in time; employees reported their own job satis-
faction via the Job Descriptive Index. Implications of rela-
tionships much higher than typically found in thejob sat-
isfaction-performance literature are discussed.

Apparently the dust has settled over what once was a controversial issue:
the satisfaction-performance linkage. Organizational psychologists (Lawler
& Porter, 1967) generally endorse the view that any covariance between
job satisfaction and job performance emerges only when satisfaction results
from performance-contingent rewards. Any notion that satisfaction
"causes" performance is regarded as naive folk wisdom, not supportable
by the empirical record.
Organ (1977) has cautioned that such a position might prematurely re-
ject something of value in lay psychology that endorsed the satisfaction-
causes-performance proposition. He suggested that a clue to the possible
reconcilability between the phenomenology of countless practitioners and
the noncorroborating empirical record might lie in the meaning of "per-
formance." Defined narrowly as quantity of output or quality of crafts-
manship—as perhaps operationalized in most of the formal research ad-
dressed to this issue—performance does not consistently or appreciably
follow from satisfaction in a direct functional relationship. But there are

'An earlier version of this paper was presented at the 42nd National Academy of Management meet-
ings, New York, 1982.

587
 588 Academy of Management Journal December

Other conceptions of performance that often may be more salient to the

practitioner.
Katz and Kahn (1966) have noted the many occasions in which organiza-
tional functioning depends on supra-role behavior—behavior that cannot
be prescribed or required in advance for a given job. These behaviors in-
clude any of those gestures (often taken for granted) that lubricate the social
machinery of the organization but that do not directly inhere in the usual
notion of task performance. Examples that come to mind include: helping
co-workers with a job related problem; accepting orders without a fuss;
tolerating temporary impositions without complaint; helping to keep the
work area clean and uncluttered; making timely and constructive statements
about the work unit or its head to outsiders; promoting a work climate
that is tolerable and minimizes the distractions created by interpersonal con-
flict; and protecting and conserving organizational resources. For lack of
a better term, the present authors shall refer to these acts as "citizenship"
behaviors.
Supervisors presumably value such behaviors, in part because they make
their own jobs easier and free their own time and energy for more substan-
tive tasks. One suspects that they value it all the more because they cannot
"require" such supra-role, citizenship behaviors, except perhaps to some
minimally acceptable or enforceable standards.
There are two distinct conceptual bases for thinking that such behavior
would be influenced by job satisfaction (or, at least, the affective state un-
derlying job attitudes). First of all, social exchange theory (Adams, 1965;
Blau, 1964) predicts that, given certain conditions, people seek to reciprocate
those who benefit them. To the extent that a person's satisfaction results
from the efforts of organizational officials and such efforts are interpreted
as volitional and nonmanipulative in intent, the person will seek to recip-
rocate those efforts. The person may not have the ability or opportunity
to reciprocate with greater work output or creative solutions to work prob-
lems. Citizenship behaviors of the sort described above are more likely to
be under the person's control and thus more likely to be a salient mode
of reciprocation.
A second basis for predicting this relationship derives from a series of
social psychological experiments (Rosenhan, Underwood, & Moore, 1974)
which strongly support the contention that prosocial gestures are most likely
to occur when a person experiences a generalized mood state characterized
by positive affect (Clark & Isen, 1982). To the extent that job satisfaction,
as conventionally measured, reflects this positive affective state, it is likely
that more satisfied persons display more of the prosocial, citizenship be-
haviors. As Rosenhan et al. (1974) phrased it, positive affect tends to de-
crease the psychological distance between self and others, and positive af-
fect tends to generalize from whatever caused it to other stimuli (notably
persons) in the temporal and social context.
In sum, it is predicted that there is a causal connection between prior
overall satisfaction and subsequent display of a host of citizenship behaviors
(as specified below). Moreover, a strong connection for satisfaction with
1983 Bateman and Organ 589

supervision in particular is predicted. The rationale for this prediction is

that the immediate supervisor represents the most direct source of variance
in events that arouse a felt need to reciprocate or that infiuence positive
affect. Also, citizenship behaviors, it is suspected, are more often seen as
"for" (i.e., benefiting) the supervisor than for any other single person rep-
resenting the organization.
Method

Subjects and Procedure

Data were obtained at two separate times from a sample of employees
in a major midwestern state university. A wide variety of jobs in non-
academic, administrative departments were surveyed, including data pro-
grammers, loan collectors, student counselors, fund-raisers, accountants,
and other professional, supervisory, and technical (nonclerical) positions.
The initial data collection was performed via questionnaire administra-
tion at the host departments. One of the researchers met with all members
of each department (groups ranged in size from 6 to 16) who completed
the measures of job satisfaction. Concurrently, departmental superiors who
had frequent contact with the subjects were completing a scale, rating each
subject's recent behavior at work. Subjects were not aware that they were
being evaluated; they also were assured of the anonymity of their responses.
Questionnaires were identified with a code number known only to the re-
searcher and the subject. The second administration employed the same
procedure and took place from five to seven weeks after the first collection.
Complete data (including subjects' and supervisors' ratings) were col-
lected for 82 employees at time 1. For various reasons, several individuals
were unavailable at time 2. Data were collected at both times for a total
sample of 77, which included subjects and their superiors. Of the sample,
29 (38 percent) were female and 48 (62 percent) were male. Their average
departmental tenure was 6.8 years.

Measures
Citizenship behavior was measured via the responses of each subject's
immediate superior on 30 7-point items. The items tapped a variety of be-
haviors such as compliance, altruism, dependability, housecleaning, com-
plaints, waste, cooperation, criticism of and arguing with others, and punc-
tuality.
A preliminary meeting with a small group of managers provided infor-
mation about any ambiguities or irrelevancies on the original items. Inter-
nal reliability coefficients for the final scale were a = .92 at time 1 and a = .94
at time 2, and the test-retest reliability was .80. Although individual items
on the scale appear to be behavior ally distinct, the psychometric proper-
ties of the scale indicate that it provided a composite criterion that is
 ^^ Academy of Management Journal December

indicative of an employee's overall evaluation by the supervisor in terms

of behavioral effectiveness.
Job satisfaction was measured with the Job Descriptive Index (JDI) de-
veloped by Smith, Kendall, and Hulin (1969). The JDI contains five scales
pertaining to employees' satisfaction with work, pay, promotions, co-
workers, and supervision. The instrument consists of 72 items—18 in each
of the work, supervision, and co-workers subscales, and 9 in the pay and
promotions subscales. All of the scales are reported to have corrected split-
half internal consistency coefficients exceeding .80, and test-retest reliabilities
averaging .57 (Schneider & Dachler, 1978).

Data Analysis

Subsequent to descriptive and correlational statistical analyses, the issue

of causal priorities was tested via cross-lagged regression analysis (Rogosa,
1980). Basic contextual information is afforded by six correlations between
the two variables x and >>—static correlations from time 1 and time 2 (rx,y,
an^ Txjyz), two autocorrelations (r^^^^ »nd Tyjyj) that indicate the test-retest
reliabilities or stabilities of the two variables, and two cross-lagged corre-
lations (rx,y2 and fyjxj) between the time 1 value of one variable and the
subsequent time 2 value of the other variable.
Causal analysis and tests of spuriousness typically are conducted by sta-
tistical corrections and comparisons for the magnitudes of the cross-lagged
correlations (Kenny, 1975). However, recent criticism of cross-lagged cor-
relation (Rogosa, 1980) alternatively suggests the use of structural regres-
sion models for the analysis of longitudinal panel data. This approach was
applied in the present study. For a pair of variables, x and jf, the causal
infiuence from ac to>' is represented by the regression parameter of the path
from X at time 1 to >' at time 2. In like manner, the causal inñuence from
J' to X is represented by the regression parameter of the path from a prior
y to a subsequent x. Thus, where: »

, and ' <

the parameters j8i and 71 represent the time-lagged influence of a variable

on itself and ^2 and 72 represent the time-lagged causal effects between x
and y. Under the usual assumptions governing regression analysis (linear-
ity, additivity, etc.), a nonzero value of a relevant parameter is indicative
of a significant causal effect.
Static and cross-lagged analyses were performed to test the relationships
between citizenship behavior and each of the separate facets of satisfac-
tion, as well as with a summated measure of overall satisfaction.
1983 Bateman and Organ 591

Table 1
Means and Standard Deviations at Time 1 and Time 2

Variable X SO X SD

Citizenship behavior 142.1 27.3 141.1 27.0

Overall satisfaction 143.5 28.3 138.0 29.4
Work 36.3 8.7 34.1 8.4
Co-workers 42.4 9.9 41.6 8.4
Supervision 42.0 10.6 41.2 10.1
Promotions 10.4 7.6 8.9 7.4
Pay 13.0 5.8 12.3 6.1

Table 2
Static Correlations (^i and h)
Between Facets of Job Satisfaction and Citizenship Behaviors
Job Satisfaction
Work Co-worker Supervision Promotions Pay Overall
t¡ Î2 t¡ t2 t] t2 tl t2 t¡ t2 t¡ t2

Citizenship
behaviors .09 .19* .24* .18 .46** .36* .37** .40** .16 .25*
*/><.O5
**p<.01

Results
Table 1 presents descriptive statistics for the study variables at both times
surveyed. Table 2 then shows the static correlations between citizenship
behaviors and the specific facets of satisfaction. There are indications that
each dimension of job satisfaction may be positively related to citizenship
behavior, with two facets—supervision and promotional opportunity—
reliably more important than pay, co-workers, and the work itself.
Subsequent cross-lagged analysis was conducted between citizenship be-
havior and each measure of satisfaction. The patterns of relationships were
virtually the same in all instances. As a summary example. Figure 1 dis-
plays the cross-lagged analysis surrounding the relationship between job
related citizenship behaviors and overall job satisfaction. The test-retest
reliabilities are fairly high for both variables. Both static correlations are
positive and strongly significant and are particularly substantial when com-
pared to most previous studies of the satisfaction-performance relation-
ship. Inspection of the cross-lagged statistics, however, failed to discriminate
a single causal direction. Both raw correlations are highly significant, the
relative magnitudes are in the predicted direction, and the predicted causal
correlation is slightly greater than the two static correlations. However,
the two cross-lagged correlations are fairly comparable to one another. Fur-
ther, both path coefficients (shown in parentheses) are positive, yet much
smaller than the correlations; they also are comparable in magnitude to
one another and are statistically insignificant.
 592 Academy of Management Journal December

Figure 1
Cross-Lagged Relationships Between Overall Satisfaction
and Citizenship Behaviors"
.71*
SATISFACTION] - -SATISFACTION
-.43* (.12)"
.41* .41«
-.39* (.11)
BEHAVIOR -BEHAVIOR2
.80*
K^ross-lagged patterns of relationships between citizenship behavior and specific facets of job sat-
isfaction reveal essentially the same results as overall satisfaction. These data are available from the
first author on request.
''Path coefficients are in parentheses.
•<001

Thus, evidence for the predicted direction of causality was not obtained.
However, the results do reliably suggest that job satisfaction is indeed strong-
ly and positively related to a "citizenship" dimension of role performance.

Discussion
The statistical relationships obtained here between general job satisfac-
tion and the aggregate measure of citizenship behaviors are considerably
stronger than those typically reported between satisfaction and "perfor-
mance." Of course, the sample size (77) limits the confidence that one could
attach to comparisons between correlations. However, when the 95 per-
cent confidence intervals are computed for the correlations involving over-
all satisfaction and satisfaction with supervision, the lower limits of these
intervals (. 15-.26) still exceed the r of. 14 from Vroom's (1964) review (the
upper limits of the confidence intervals range from .54 to .62).
The stronger relationship found here may be because the citizenship be-
haviors of interest here generally represent actions more under the voli-
tional control of workers than conventional productivity measures. Pro-
social gestures are less likely to be constrained by other situational forces,
and they pose very little in the way of ability requirements.
Consider, for example. Smith's (1977) study, which found that job at-
tendance on a given day was predicted by satisfaction much more strongly
in a location hit by a severe winter storm than in a different location expe-
riencing clement weather. In extremely bad weather, absence is somewhat
more defensible than usual; to attempt to show up for work becomes more
a matter of intent. This attenuation of the situation force "requiring" at-
tendance allows more variance of the behavior in question and increases
the likelihood that such variance can be attributed to "internal" (i.e., atti-
tudinal or dispositional) forces. It might be added that the act of strug-
gling through bad weather to report to work represents more of a prosocial,
citizenship gesture than does attendance on other days, and it probably is
more likely to be valued and appreciated by responsible officials.
1983 Bateman and Organ 593

A second reason for the unaccustomed strength of the attitude-behavior

relationship found here probably derives from the broad array of citizen-
ship gestures sampled by the behavioral measure. As Fisher (1980) has ap-
propriately pointed out: attitude-behavior linkages are attenuated if the
specificities of the constructs are not matched; narrowly defined attitude
measures are more predictive of their corresponding specific behaviors than
of general measures of behavior; and global measures of affect or mood
predict aggregative measures of broadly sampled behaviors better than does
any single behavior. Too often the interpretation of results from studies
of satisfaction and performance has not taken into account the imbalance
between generalized attitude measures and a measure of a very narrowly
conceived facet of job behavior (e.g., quantity and/or quality of task per-
formance). The results of this study demonstrate that gtwQxaX patterns of
employee behavior, rather than a single-act productivity criterion, are pre-
dictable from a general measure of satisfaction.
However, the strength of the static relationships found here did not pass
the test of causal inference; the cross-lagged differentials were not reliable.
Why was a causal connection between satisfaction and citizenship behaviors
not supported? One possible explanation concerns the time interval (six
weeks) chosen. Conceivably, this was too short an interval for a functional
relationship to manifest itself. The current state of conceptual and empirical
work in this area offered no clear cues as to what the appropriate interval
might be. Research in another areafleadership)indicates the appropriateness
of relatively short time lags (Sims &. Szilagyi, 1979), suggesting that the
six-week interval is not inherently weak methodologically. However, the
test-retest correlation of the time 1 and time 2 behavioral measures was
.80, approaching the limit set by the internal reliability of the measure itself.
With such stability, then, behavior at time 2 was best predicted by behavior
at time 1, with initial job satisfaction unable to add significantly to
prediction.
Kenny (1975) states that failure to find a significant cross-lagged differ-
ential means that one cannot reject the null hypothesis of spuriousness,
that is, cannot rule out the possibility that the variables are correlated be-
cause of their dependence on a common antecedent variable. In this case,
any such "spuriousness" does not originate from common-method variance.
The citizenship measure came from a respondent (the supervisor) indepen-
dent of those who reported their satisfaction; thus the "same-source" prob-
lem does not apply.
Supervisory behavior—especially that generally described as "suppor-
tive" or "considerate"—could represent a common cause. It has been well
established that considerate supervision affects job attitudes, even when
attitudes are indexed by nonself-report measures—for example, turnover
(Fleishman, 1973). Perhaps supportive supervision also elicits citizenship
behaviors, independently of its effects on job satisfaction.
Persondity, or some dimension of stable individual differences, also could
account for the covariance between mood and citizenship. The literature
 ^^ Academy of Management Journal December

does not offer either a consistent or a precise clue as to what dimension(s)

might be refiected here, but one could well imagine that certain tempera-
ments (e.g., low vs. high neuroticism) determine characteristic affective tone
as well as one's generalized orientation to prosocial gestures.
The strength of the relationships found in the present study—compared,
say, to the .14 reported by Vroom (1964) to be the mean satisfaction/job
performance correlation—coupled with the problem of stabilities that may
have precluded causal demonstrations, suggests future work with longer
time lags (and larger sample sizes, which v*dll enhance the power of the causal
tests and enable a more discriminatory factor analysis on the citizenship
scale). Additionally, ratings of behavior could be obtained from observers
other than the supervisor, including co-workers. If the relationship is more
convincingly shown to be spurious, it would seem that a more fully mapped
causal niodel is needed. Such a model should incorporate some identifi-
able environmental and personality variables in order to account adequate-
ly for the connections between satisfaction and citizenship behaviors. More-
over, one could better ascertain whether conventional measures of job at-
titudes, as used here, adequately operationalize the "felt need to recipro-
cate" invoked by exchange theory as a predictor of behavior.
The resolution of these issues would appear to be worthwhile in view
of the practical implications that would ensue. If both citizenship behaviors
and affect refiect enduring dispositions, then fairly drastic alterations of
the work environment would be needed to effect much change in these vari-
ables. If they result from situation-specific causes, then perhaps organiza-
tional officials can address them in a proactive posture. And if citizenship
behaviors are shown to be causal effects of satisfaction, it would appear
that job satisfaction is, in fact, more important to organizations than just
in its often-mentioned relationships with absenteeism and turnover. Citi-
zenship behaviors represent a multiple-observation criterion, which not only
is of potential value to managers and organizations but also can contribute
to less restrictive organizational research (by expanding the number of out-
come variables that are studied) and thereby provide for broader under-
standing of employee behavior.

References
Adams, J. A. Inequity in social exchange. In L. Berkowitz (Ed.), Advances in experimental social
psychology (Vol. 2). New York: Academic Press, 1965, 267-299.
Blau, P. Exchange and power in social life. New York: Wiley, Inc., 1964.
aark, M. S., & Isen, A. M. Toward understanding the relationship between feeling states and social
behavior. In A. H. Hastorf & A. M. Isen (Eds.), Cognitive social psychology. New York: Elsevier
North Holland, 1982, 73-108.
Fisher, C. On the dubious wisdom of expecting satisfaction to correlate with performance. Academy
of Management Review, 1980, 5, 607-612.
Fleishman, E. A. Twenty years of consideration and structure. In E. A. Fleishman & J. G. Hunt (Eds )
Current developments in the study of leadership. Carbondale, IU.: Southern Illinois University Press,'
1 " / J» 1~J / .
1983 Bateman and Organ 595

Katz, D., & Kahn, R. L. ne social psychology of organizations. New York: Wiley, 1966.
Kenny, D. A. Cross-lagged panel correlation: A test for spuriousness. Psychological Bulletin, 1975,
82, 887-903.
Lawler, E. E., Ill, & Porter, L. W. The effect of performance on job satisfaction. Industrial Rela-
tions, 1967, 7, 20-28.
Organ, D. W. A reappraisal and reinterpretation of the satisfaction-causes-performance hypothesis.
Academy of Management Review, 1977, 2, 46-53.
Rogosa, D. A critique of cross-lagged correlations. Psychological Bulletin, 1980, 88, 245-258.
Rosenhan, D. L., Underwood, B., & Moore, B. Affect moderates self-gratification and altruism. Journal
of Personality and Social Psychology, 1974, 30, 546-552.
Schneider, B., & Dachler, H. P. A note on the stability of the Job Descriptive Index. Journal of Ap-
plied Psychology, 1978, 63, 650-653.
Sims, H. P., & Szilagyi, A. D. Time-lags in leader reward research. Journal of Applied Psychology,
1979, 64, 66-71.
Smith, F. J. Work attitudes as predictors of attendance on a specific day. Journal of Applied Psychol-
ogy, 1977, 62, 16-19.
Smith, P. C , Kendall, L. M., & Hulin, C. L. The measurement of satisfaction in work and retire-
ment. Chicago, 111.: Rand McNally, 1969.
Vroom, V. R. Vf^ork and motivation. New York: Wiley, 1964.

ThomasS. Bateman is Assistant Professor of Organizational Be-

havior, Department of Management, College of Business Ad-
ministration, Texas A&M University.
Dennis W. Organ is Professor of Organizational Behavior, Grad-
uate School of Business Administration, Indiana University.
Copyright of Academy of Management Journal is the property of Academy of Management and its content may
not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written
permission. However, users may print, download, or email articles for individual use.