Professional Documents
Culture Documents
JEFF GREENBERG
University of Kansas
TOM PYSZCZYNSKI
AND
SHELDON SOLOMON
Skidmore College
An experiment was conducted to provide empirical support for the notion that
asymmetrical causal attributions for favorable and unfavorable outcomes result
from a self-serving attributional bias that occurs independently of self-presen-
tational concerns. Subjects did either well or poorly on an ego-involving test for
which their performance, attributions, and evaluations of the test were either
public or private. A pattern of self-serving responses for subjects’ attributions
and evaluations of the test was found in the private conditions, thus providing
evidence of the influence of outcome favorability on individuals’ perceptions of
causality. Theoretical and practical implications of these finding are discussed
and suggestions for future research are offered.
The authors wish to thank Jack Brehm, Ray Petty, and Lynne Steinberg for their helpful
comments on an earlier version of this paper. Requests for reprints should be sent to the
first author at the Department of Psychology. University of Kansas, Lawrence, KS 66045.
56
0022-1031/82/010056-12$02.00/O
Copyright @ 1982 by Academic Press, Inc.
All rights of reproduction in any form reserved.
SELF-SERVING PRIVATE ATTRIBUTIONS 57
Procedure
Upon arriving for the study, subjects were told that they would be participating in one
of a series of studies investigating students’ evaluations of standardized personality and
intelligence tests. They were told that in this particular study they would be given a sample
of a standardized intelligence test, known as the “Culture Fair Test of g,” and then be
’ Two of these eight participants finished the test early, looked at the back of the answer
sheets and saw that all the responses for a given problem were marked either correct or
incorrect; three participants finished the test early and while searching for the correct
answers to problems they had answered incorrectly found out that all the choices for a
given problem were marked either correct or incorrect; the other three suspicious partic-
ipants chose more than one answer as they proceeded through the problems and so also
found out that all the choices for a given problem were marked either correct or incorrect.
No other participants expressed any suspicion with regard to the veracity of the perfor-
mance feedback. Although the suspicion rate was rather high, the major determinant of
suspicion was quite simply whether or not a participant found out that all the responses
for a given problem were marked either correct or incorrect. Individuals who did find this
out were quite willing to express their knowledge; individuals who did not find out tended
to evince obvious surprise when the false nature of the feedback was revealed and even
after the deception had been revealed reported no prior doubts about the performance
feedback. Thus we believe that the participants we included in our analyses accepted the
feedback as veridical.
SELF-SERVING PRIVATE ATTRIBUTIONS 59
asked to give their reactions to an evaluation of the test. It was explained that in order
to standardize testing conditions, all instructions for the test would be written (the use of
written instructions allowed the experimenter to remain blind to conditions). The exper-
imenter then handed subjects a booklet containing the test, an answer sheet, instructions
for the test, a test evaluation questionnaire, and a Psychology Department Survey of
students’ reactions to various aspects of psychology studies. Subjects were then given 10
min to read through the instructions.
The written instructions gave a brief description of the background and use of the test
in order to insure that subjects would become actively involved in taking the test. Spe-
cifically, subjects read that (1) the test was designed by the Institute for Personality and
Ability Testing and had been used extensively in business and educational settings; (2)
because the test utilized abstract symbols rather than words, it measured pure intelligence
and reasoning abilities; (3) these abilities are basic to complex problem solving and creative
thinking; and (4) the test had been found to be an excellent predictor of both academic
and financial success.
The actual test used in the study consisted of 20 multiple-choice problems for the Test
of “g”: Culture Fair (Cattell & Catell, 1960) that required subjects to perceive a relationship
among abstract symbols. Each of four pages consisted of an example followed by five
problems. Subjects were to use the relationship demonstrated in the example at the top
of each page to answer the questions on that page. Although subjects were told that they
would have 10 min to work on the test, they were allowed as much time as necessary to
complete all of the problems. The longest time required by any subject was 14 min.
Publicity Manipulation
The manipulation of performance publicity, or the experimenter’s knowledge of subjects’
scores on the test, was made possible through the use of self-scoring answer sheets that
gave subjects immediate feedback regarding their performance. Subjects responded to the
test questions by peeling off one of five dots from an answer sheet corresponding to one
of five response choice in the test booklet. Subjects were told that if they answered an
item correctly, they would find a “c” beneath the dot. If they answered incorrectly,
however, they would find an “x” beneath the dot. In fact, all responses for each question
were keyed either correct or incorrect, and all subjects consequently found that they had
answered 12 of the 20 questions “correctly.“’
In the private performance condition, subjects were told that the experimenter was not
interested in their performance on the test, and that he would therefore not collect their
answer sheets or record their scores. Subjects in this condition were explicitly instructed
not to put their names on their answer sheets and to fold them and put them somewhere
out of the experimenter’s sight when they were finished with the test. Subjects in the
public performance condition were told to put their names on their answer sheets and
questionnaires, which would subsequently be collected and scored by the experimenter.
Furthermore, in the private performance condition the test evaluation questionnaire (which
contained most of the dependent measures) was accompanied by a cover sheet that asked
subjects to count the number of problems they answered correctly and assured them that
their responses would be anonymous. In the public performance condition, subjects were
asked to write down the number of problems they answered correctly and to write their
names on the test evaluation questionnaire.
’ The test questions were sufficiently ambiguous to allow this type of false feedback.
Additionally, subjects who did well on the test experienced the same pattern of correct
and incorrect responses as the participants who did poorly. A pattern of self-serving
attributions would therefore rule out a plausible nonmotivational explanation for these
findings in similar studies, which is that perceived response-outcome covariation is more
likely for increasing success than for constant failure.
60 GREENBERG, PYSZCZYNSKI, AND SOLOMON
Outcome Manipulation
The manipulation of performance outcome was contained in the cover sheet for the test
evaluation questionnaire. Subjects read that the average performance for college students
was either 8 (success) or 16 (failure) correct responses for 20 problems. Since all subjects
obtained 12 correct answers on the test, their scores were 4 points above or below average.
The test evaluation questionnaire contained all but one of the primary dependent mea-
sures for the study. Subjects were asked to indicate on 9-point scales (1 = not at all, 9
= extremely) the extent to which their ability, the amount of effort they put into taking
the test, the ease or difficulty of the test, and luck were responsible for their performance.
Subjects were also asked to indicate how difficult, fair, and valid they perceived the test
to be, and how clear the instructions for the test were.
After completing this questionnaire, subjects filled out a Department of Psychology
Survey of students’ reactions to tests. This contained checks on the performance publicity
and outcome manipulations. In order to maintain the experimenter’s ignorance of subjects’
performance in the private condition, subjects were told that this questionnaire was for
the purpose of obtaining information on students’ feelings about psychology experiments
(and tests they may entail) and that the experimenter would not see their responses. To
reinforce this point, subjects were instructed to fold their departmental questionnaires
when they had finished them and to put them in a large psychology department box which
contained other such questionnaires. Only by placing a small unique mark on the back of
all of each subject’s forms were we able to match the latter questionnaire with the other
materials.
As a check on the publicity manipulation, subjects were asked whether or not the
experimenter would have knowledge of their performance on the test. This was a dichot-
omous measure that subjects responded to by circling either yes or no. The performance
outcome manipulation was checked by a question that asked subjects to indicate on a 9-
point scale (1 = not at all well, 9 = extremely well) how well they did on the test. This
questionnaire also contained questions concerning subjects’ affective reactions to the study,
i.e., how enjoyable and interesting the study was for them; and the final dependent measure,
how important it was for them to do well on the test. These questions. and the questions
concerning the clarity of the test instructions, and the fairness and the validity of the test.
were included because they, as well as the typical attribution measures, may be subject
to distortions resulting from self-serving motives.
At this point the experiment was concluded. Subjects were completely debriefed and
thanked for their participation.
RESULTS
Manipulation Checks
Performance publicity. Subjects responded yes or no to the question,
“Will the experimenter have knowledge of your performance on the
test?” Twenty-one of twenty-three subjects answered yes in the public
conditions, and 17 of 23 subjects answered no in the private conditions.
In order to provide a more stringent examination of public and private
attributions, subsequent analyses were performed using only those sub-
jects who responded appropriately to this question. However, it should
be noted that the identical pattern of significant main effects would be
obtained if all of the subjects were included in the analyses.
Performance outcome. Subjects responded to a question about how
well they did on the test on a 9-point scale (1 = not at all well, 9 =
SELF-SERVING PRIVATE ATTRIBUTIONS 61
Private Public
2
Unfavorable Favorable Unfavorable Favorable
outcome outcome outcome outcome 3
Question (N = 9) (N = 8) (N = 10) (N = 11) F(1, 34) MS, p < E
6
To what extent was your ability responsible for
your performance on the test? 4.33 7.50 5.90 6.82 12.18 2.96 .0014 3
To what extent was the amount of effort you R
put into taking the test responsible for your
performance? 4.56 7.25 5.90 6.82 9.35 3.07 .0044 2z
To what extent was the ease or dimculty of the
test responsible for your performance? 6.00 6.63 6.30 6.73 0.81 3.24 .3738 E
To what extent was luck responsible for your
%
performance on the test? 4.33 2.25 3.60 2.54 5.13 4.31 .0300 tJ
How difficult was the test? 6.11 6.00 6.50 6.18 0.20 2.09 .6564
8.25 6.91 4.60 “0
How clear were the instructions for the test? 5.33 6.00 6.61 .0147
How fair was the test? 4.89 7.13 5.70 6.55 4.99 4.14 .0322 5
How valid do you believe the test to be? 2.44 5.63 4.10 5.91 14.98 3.88 .0005
If tests were used, how important was it for
you to do well? 2.56 6.63 5.90 6.00 6.77 5.32 .0137
--
’ F values are reported for the performance outcome main effect for each measure.
SELF-SERVING PRIVATE ATTRIBUTIONS 63
means for positive and negative outcomes within the public and private
conditions found significant (p < .05) differences between favorable and
unfavorable outcomes on all of the measures in the private condition,
but only for the validity of the test in the public condition. This suggests
that subjects’ judgments were more self-serving in the private than in
the public conditions. Significant interactions reflecting this tendency
were found for ability attributions (F(1, 34) = 4.00, p < .02), and for
the importance of doing well on the test (F(1, 34) = 6.93, p < 0.02).
DISCUSSION
The present investigation demonstrated asymmetrical causal attribu-
tions for favorable and unfavorable outcomes even when there was no
public knowledge of participants’ performance. These findings are par-
ticularly important because they cannot be explained by a self-presen-
tational motive and thus, provide empirical support for the notion that
there is a self-serving bias in individuals’ perceptions of causality.
One important question concerns whether a motivational model or an
information-processing model best accounts for the results of this study.
While this experiment certainly does not provide a definitive test between
these two types of explanation, certain aspects of our procedure reduce
the likelihood of an information-processing explanation of these findings.
In the study private favorable and unfavorable outcome experiences were
identical; the favorability was varied merely by providing normative
information after performance that suggested either a performance above
average or a performance below average. Therefore the most plausible
of Miller and Ross’ (1975) nonmotivational explanations of self-serving
attributions, that response-outcome covariation is more likely for in-
creasing success than for constant failure, is untenable. As Bradley (1978)
and Zuckerman (1979) have persuasively argued, the nonmotivational
explanation that success is expected and that expected events are at-
tributed to internal factors lacks both empirical support and theoretical
rationale. Miller and Ross also suggested that since favorable outcomes
are intended and individuals are most likely to perceive contingency
between their actions and intended outcome, such outcomes are partic-
ularly likely to be attributed to internal factors. However, this idea may
not pertain to our results since, in the absence of normative data or a
criterion for success prior to performance, all of the participants must
have intended simply to answer as many of the 20 problems correctly
as possible; thus, all participants perceived attaining the outcome they
intended, i.e., answering the problems correctly, for 12 out of the 20
problems. It is possible, however, that after participants found that they
had done relatively well, when confronted by the attribution questions,
they retrospectively focused on the problems they had answered cor-
rectly more so than did participants who found out they had done rather
poorly. This recall bias, combined with the Miller and Ross notion that
64 GREENBERG, PYSZCZYNSKI, AND SOLOMON
lem, which may be the first step toward remedying it. A similar type of
exacerbation cycle may also inhibit improvement in achievement settings.
Besides blocking an individual’s path to future improvement, asym-
metrical private attributions may also seriously undermine social under-
standing. In social relationships such as marriages, each partner may
make causal attributions that are not expressed to each other concerning
a variety of problems (e.g., why things aren’t going well sexually or
socially, why the children are spoiled, why they never have any money,
etc.). The present findings suggest that the partners will make quite
different private attributions for these problems. Eventually the partner’s
perceptions of the relationship may become so divergent that an insur-
mountable rift develops by the time the couple begins talking openly
about the causes of their problems. This process could inhibit under-
standing between team members, co-workers, and ethnic groups as well.
Given the implications of this asymmetry in private self-referent at-
tribution, it is important that we progress toward understanding the
causes of this phenomenon. Although researchers have examined some
personality variables that may affect the occurrence of the self-serving
bias, for example, self-esteem (Fitch, 1970; Levine & Uleman, 1979) and
depression (Kuiper, 1978; Rizley, 1978), these studies involved public
performances and attributions. Hopefully future research that is careful
in distinguishing between public and private attributions will further our
understanding of the determinants and consequences of this subtle mode
of self-aggrandizement.
REFERENCES
Abramson, L. Y., Seligman, M., & Teasdale, J. Learned helplessness in humans: Critique
and reformulation. Journal of Abnormal Psychology, 1978, 87, 49-74.
Adler, A. In H. L. Ansbacher & R. R. Ansbacher (Eds. and annot.) The individual
psychology of Alfred Adler. New York: Basic Books, 1956.
Allport, G. W. Personality: A psychological interpretation. New York: Holt, 1937.
Arkin, R. M., Appelman, A. J., & Burger, J. M. Social anxiety, self-presentation, and the
self-serving bias in causal attribution. Journal of Personality and Social Psychology,
1980, 38, 23-25.
Beck, A. T. Depression. New York: Hoeber, 1967.
Beckman, L. Effects of students’ performance on teachers’ and observers’ attributions of
causality. Journal of Educational Psychology, 1970, 61, 78-82.
Bowerman, W. R. Subjective competence: The structure, process, and function of self-
referent causal attributions. Journal of the Theory of Social Behavior, 1978, 8, 45-75.
Bradley, G. W. Self-serving biases in the attributional process: A reexamination of the
fact or fiction question. Journal of Personality and Social Psychology, 1978,36,56-71.
Cattell, R. B., & Cattell, A. K. S. Test of “g”: Culture Fair. Champaign: The Institute
for Personality and Ability Testing, 1960.
Dweck, C. S., Goetz, T. E., & Straus, N. L. Sex differences in learned helplessness. IV.
An experimental and naturalistic study of failure generalization and its mediators.
Journal of Personality and Social Psychology, 1980, 38, 441-452.
Feather, N. T.. & Simon, J. G. Attribution of responsibility and valence of outcome in
SELF-SERVING PRIVATE ATTRIBUTIONS 67
relation to initial confidence and success and failure of self and other. Journal of
Personality and Social Psychology, 1971, 18, 173-188.
Fitch, G. Effects of self-esteem, perceived performance and choice on causal attributions.
Journal of Personality and Social Psychology, 1970, 16, 3 11-3 15.
Heider, F. The psychology of interpersonal relations. New York: Wiley, 1958.
House, W. C. Effects of knowledge that attributions will be observed by others. Journal
of Research in Personality, 1980, 14, 528-545.
Jones, E. E., & Davis, K. E. From acts to dispositions: The attribution process in person
perceptions. In L. Berkowitz (Ed.), Advances in experimental social psychology, Vol.
2. New York: Academic Press, 1965.
Kelley, H. H. Attribution theory in social psychology. In D. Levine (Ed.), Nebraska
Symposium on Motivation, Lincoln: University of Nebraska Press, 1967.
Kuiper, N. A. Depression and causal attributions for success and failure. Journal of
Personality and Social Psychology, 1978, 36, 236-243.
Levine, R., & Uleman, J. S. Perceived locus of control, chronic self-esteem, and attri-
butions to success and failure. Personality and Social Psychology Bultetin, 1979, 5,
69-72.
Miller, D. T. Ego involvement and attributions for success and failure. Journal of Per-
sonality and Social Psychology, 1976, 34, 901-906.
Miller, D. T. What constitutes a self-serving attributional bias? A reply to Bradley. Journal
of Personality and Social Psychology, 1978, 36, 1221-1223.
Miller, D. T., & Ross, M. Self-serving biases in the attribution of causality: Fact or fiction?
Psychological Bulletin, 1975, 82, 213-225.
ROSS, L., Bierbrauer, G., & Polly, S. Attribution of educational outcomes by professional
and nonprofessional instructors. Journal of Personality and Social Psychology, 1974,
29, 609-618.
Rizley, R. Depression and distortion in the attribution of causality. Journal of Abnormal
Psychology, 1978, 87, 32-48.
Snyder, M. L., Stephan, W. G., & Rosentield, D. Egotism and attribution. Journai of
Personality and Social Psychology, 1976, 33, 435-441.
Weary, G. Self-serving attributional biases: Perceptual or response distortions? Journal
of Personality and Social Psychology, 1979, 37, 1418-1420.
Weary, G. Examination of affect and egotism as mediators of bias in causal attributions.
Journal of Personality and Social Psychology, 1980, 38, 348-357.
Zuckerman, M. Attribution of success and failure revisited, or: The motivational bias is
alive and well in attribution theory. Journal of Personality, 1979, 47, 245-287.
REFERENCE NOTE
1. Tetlock, P. E. The social consequences of defensive and counterdefensive attributions.
Paper presented at the meeting of the American Psychological Association, New York,
September 1979.