Académique Documents
Professionnel Documents
Culture Documents
Abstract
1. The present research was facilitated by grants from the National Institute of
Mental Health (1R08 31910) to Robert Arkin and from the National Science Foun-
dation (BNS78-08834) to Harris Cooper, Principal Investigator. Funds for the on-
line computer search were provided by the Center for Research in Social Behavior.
Special thanks are extended to Miron Zuckerman for critical comments on an earlier
draft and for providing statistical information on several of the included studies.
Requests for reprints should be sent to Robert Arkin, Department of Psychology,
210 McAlester Hall, University of Missouri, Columbia, MO 65211.
Journal of Personality 48:4, December, 1980. Copyright© 1980 by Duke University
Press.
436 Arkin et al.
subjects to assess both their own and the target person's contri-
bution to the influence attempt.
Review Procedures
Though perfect retrieval procedures are, of course, not available,
it is important to state how relevant studies were located, acknowl-
edging the possibility that a biased sample of studies could be
reviewed. Tests of the hypothesis that people assume more per-
sonal responsibility for success than failure on interpersonal influ-
ence tasks were located through (a) Zuckerman (1979); (b) informal
channels or personal involvement; (c) the Psychological Abstracts
between the years 1970 and 1978; and (d) an on-line computer
search. The computer search uncovered over 340 investigations
possessing the keywords: (a) self-serving bias; (b) ego-serving bias;
(c) egocentric bias; (d) teacher-student; (e) client-therapist; (f) in-
terpersonal influence; and (g) a crossing of the word attribution
with the words success, failure, performance, outcome, and ac-
tor's. Titles were judged for potential relevance by the investi-
gators. Ultimately 59 studies were examined directly in the Uni-
versity library.
2. Although not originally reported because of space limitations, the main effect
hypothesis test of Arkin et al. (1979) was retrieved.
438
IB
_^ ^ ^ to tn ^ ^
CO <D o CD 5. Q. '5. Q. '5. 5. Q.
c o O o o
o
O
to CO to ? CO CO (0 (0 (d (Q (Q (0 (0
.c B B Q) 0) 0} Qy 03 Q) O
2 B 2<n B iB oCO
o o c
o o ® o o o ooooo o
Q. o o
a.
o 2* to
onsi biii
onsi bili
onsi bili
onsi bili
onsi bili
a k_
a IB. a>
ructi
JZ 03 Qy Qy A) tf) 03 03
;her
CD
CO
o E EEEEE E
Q.
c in m $ o
a
1 a. a.$ O O O O O O O
o o o o o oo
J pei
J pel
J pel
J pei
ted r
IB
tion
TJ "5 "5 3 '3 '3 '5 3 3 3
O O O o OO O
C g 3 3 3
q XI 0 o o "5 B o o
Ital
ttri
ttri
ttri
ttri
ttri
CO o
ttri
to CO to CO CO
c •£ c c c c c
o
o o oo 00 o <o o CO CO in o ^ p p
CO |v! 1..^ O CD O O5
1 sg » 1 s a> r v oo m o> in CO
I
o CM CO o CO in o) o CO o o>
1 o> CJ> CO 8 CM iv. oq o oq p Tt
1 o 01 ' ' d ^ ' C3 i'
o T
S 1
CM y-
so O)
p
^-
CO
in
r-;
(o
IV.
O)
•*
13.
18.
18.
1 00
V
00 »-^ o i T- CD »- i n
^- -I- 1 - y CO V
O in m in in
I 1
o in
0) § 1§ 8 8 8 S
(OOT-ooooinoicM coooSoooooio)
ID
•5 m CO i n i n o O) CO o CO ^ 1^
bO h.. h. 1^ IV. 00 IV. tv. CD I-- t ^ 1^
O> O> O> O) O) Oi O) O> O) O) O)
a
u
c
o
o
0)
E
1a
CO
u. ~ CO
^ X
ther{tpist
o>
<
TJ ^ c c to to ?=»
in et
«. CO CO LU ^
c .S a5 S .2
CO O O 1 c £o
O ID d)
<E 2S
I T CD CD CD CD
I'- < < u! 5^
o
439
0)
o
a.
o
.2
.o ffl .E o — ~ o -J2 <»
o
I o o o ^ o o o o
I c c c c c c c c
o o o o o o o o
3 3
r I
CO O O
CM O O
O •>- 1- O O
r: in C3 o in
<p in p h^ in
in CO T- c\i • * T-; ,-
'- V '- •* V
•© d> •D
Q.2 £ :?
13
0)
C
XI
B g » i
o
U
440 Arkin et al.
tailed) and; (c) two studies (Beckman, 1970; 1973) reporting nearly
equivalent but opposite effects and no useable statistics were each
entered as exact nullfindings.^These conventions either have no
appreciable effects or tend to support the null hypothesis.
Four statistics were generated to describe the literature (cf. Coo-
per, 1979). Each was calculated for the entire set of studies ap-
pearing in Table 1 and also for just those cited in the earlier review.
An overall probability (Zma) was generated for the set of studies
without regard to experimental paradigm. This probability was cal-
culated by (a) finding the normal deviate (Z) associated with each
test's p level, and (b) dividing their sum by the square root of the
number of studies involved (cf. Mosteller & Bush, 1954). For the
earlier review's 13 tests, the Zma was nonsignificant (1.24, p < .22,
two-tailed). However, the entire body of 23 tests revealed a highly
significant Zma (4.81, p < .0001). A fail-safe N (Nfs.os) of 174 was
found for the 23-study run. This statistic is the number of null-
summing studies needed to raise the series' probability above
p < .05. No fail-safe N was calculated for the earlier review since
this run did not reach significance.^
A measure of effect size, d, was calculated for each study. The
d index describes, in standard deviation units, the distance be-
tween the success and failure means.^ For the previously reviewed
studies a d index of .15 was found. A d index of this size indicates
that the average member of a success condition took more personal
responsibility for the outcome than 56% (= U^; see Cohen, 1977)
of the people in failure conditions. For the total run of 23 studies,
d equaled .38, meaning average success subjects attributed out-
comes more to self-causation than 64% of the failure subjects.
Separate analyses of the three research paradigms revealed the
3. The two studies by Beckman (1970; 1973) were eliminated (assigned Z scores
of zero) because the chi square tests reported in each (all other studies used analysis
of variance) failed to include the appropriate comparisons. In both cases all com-
parisons reported were between actor-observer, or self-other, rather than within
self-attribution alone, rendering interpretation problematic (Arkin, Gabrenya, &
McGarvey, 1978). Additionally, only a subset of the relevant conditions were in-
cluded in the comparisons. Since (a) inspection of the means of subjects' self-attri-
butions revealed that they were in opposing directions (with one study appearing
to support the self-serving bias hypothesis and one study apparently uncovering a
reversal of this effect), and (b) the mean differences were roughly comparable in
magnitude, the entry of each as null findings seemed a proper representation.
4. No test employing a weighting of studies by number of subjects was performed.
The 22 investigations contained samples of a fairly similar size (see Table 1).
5. The d indices were estimated using a slight modification (i.e., d = 2f ^ ^ o f
the Friedman (1968) formula. The d index assumes an equal standard deviation
within the success and failure groups.
The self-serving bias 441
6. The exact value available for the disconfirming effect found in Ames (1975)
did not appear in the abstract of Ames (1973) used in the earlier review. The exact
value was used in the teacher-student and overall analyses anyway. Also, two dis-
sertations (Mayhew, 1972; Quirk, 1967) were not included in the meta-analysis,
though their results can be viewed as supporting the self-serving bias. The elimi-
nation of the Quirk and Mayhew dissertations was based upon their use of evalu-
ations of ratings (Zuckerman, 1979, p. 254) as dependent measures rather than the
more appropriate attributional ratings.
442 Arkin et al.
P d
Author level index Interacting variable(s)
Teacher-student paradigm
Beers (1975) — — perceived seif-effort
Ross etal. (1974) .001 .91 roie piayer/real teacher
Beckman (1976) — — participant/observer
Client-therapist paradigm
Weary (1980) .04 .55 pubiicity
Harvey et al. (1974) .02 .72 expected outcome
Arkin et ai. (1979) .03 .55 anticipated evaluation
and self-monitoring
Arkin (Exp. 1) et al. (1980) .02 .57 anticipated evaluation
and sociai anxiety
Federoff & Harvey (1976) .001 1.42 objective seif-awareness
Misceiianeous paradigms
Schopier & Layton (1972) .01 .87 ability of target person
Arkin(Exp. 2)etal. (1980) .01 .96 social anxiety
Arkin(Exp. 2)etai. (1980) .05 .48 measurement technique
Mynatt & Sherman (1975) .05 .74 group/alone
Krussell (1974) .001 1.14 sex
Krussell (1974) .001 .73 fear of success
7. Some authors have contended that the self-serving attribution bias might only
appear in experiments involving comparisons of increasing success with constant
failure. This possibility was tested. The six studies which compared increasing
success with constant failure (Beckman, 1970; Beers, 1975; Johnson, Feigenbaum,
& Weiby, 1964; Rizley, 1978; Schopler & Layton, 1972) produced evidence of a
self-serving bias (Zma = 2.45, p < .02; Nfs.os = 6; J = .27, U, = 60%). However,
the 17 remaining experiments comparing constant success with constant failure
revealed even stronger bias-supporting evidence (Zma = 4.24, p < .0001; NFS m =
96; 3 = .42, 1/3 = 66%).
The self-serving bias 443
References
Ames, R. E., Jr. Teachers' attributions of responsibility. Dissertation Abstracts
International, 1973, 33, 6167A.
Ames, R. Teachers' attributions of responsibility: Some unexpected nondefensive
effects. Journal of Educational Psychology, 1975, 67, 668-676.
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, 23-35.
Arkin, R. M., Gabrenya, W. K., Jr., Appelman, A. S., & Gochran, S. T. Self-presen-
tation, self-monitoring, and the self-serving bias in causal attribution. Person-
ality and Social Psychology Bulletin, 1979, 5, 73-76.
Arkin, R. M., Gabrenya, W. K., & McGarvey, B. The role of social perspective in
perceiving the causes of success and failure. Journal of Personality, 1978, 46,
762-777.
Arkin, R. M., Gleason, J. M., & Johnston, S. Effect of perceived choice, expected
outcome, and observed outcome of an action on the causal attributions of actors.
Journal of Experimental Social Psychology, 1976, 12, 151-158.
Beckman, L. Effects of students' performance on teachers' and observers' attribu-
tions of causality. Journal of Educational Psychology, 1970, 6 1 , 76-82.
Beckman, L. Teachers' and observers' perceptions of causality for a child's perfor-
mance. Journal of Educational Psychology, 1973, 65, 198-204.
Beckman, L. Gausal attributions of teachers and parents regarding children's per-
formance. Psychology in the Schools, 1976, 13, 212-218.
Beers, S. E. The role of effort in participants' and observers' attributions of causality
for student performance. Dissertation Abstracts International, 1976, 36, 5858B.
Bradley, G. Self-serving biases in the attribution process: A re-examination of the
fact or fiction question. Journal of Personality and Social Psychology, 1978, 36,
56-71.
Brandt, G. S., Hayden, M. E., & Brophy, J. E. Teachers' attitudes and ascription of
causality. Journal of Educational Psychology, 1975, 67, 677-682.
Gohen, J. Statistical power analysis for the behavioral sciences. Rev. Edition. New
York: Academic Press, 1977.
Gooper, H. Statistically combining independent studies: A meta-analysis of sex
differences in conformity research. Journal of Personality and Social Psychol-
ogy, 1979, 37, 131-146.
Gooper, H., & Rosenthal, R. Statistical vs. traditional procedures for summarizing
research findings. Psychological Bulletin, 1980, 87, 442-449.
Federoff, N. A., & Harvey, J. H. Focus of attention, self-esteem, and the attribution
of causality. Journal of Besedrch in Personality, 1976, 10, 336-345.
Friedman, H. Magnitude of experimental effects and a table for its rapid estimation.
Psychological Bulletin, 1968, 70, 245-251.
Gould, R., & Sigall, H. The effects of empathy and outcome on attribution: An
examination of the divergent perspectives hypothesis. Journal of Experimental
Social Psychology, 1977, 13, 480^89.
Harvey, J. H., Arkin, R. M., Gleason, J. M., & Johnston, S. Effect of expected and
observed outcome of an action on the differential causal attributions of actor and
observer. Journal of Personality, 1974, 42, 62-77.
Johnson, T. J., Feigenbaum, R., & Weiby, M. Some determinants and consequences
of the teacher's perception of causation. Journal of Educational Psychology,
1964, 55, 237-246.
Jones, E. E., & Nisbett, R. E. The actor and the observer: Divergent perceptions
of the causes of behavior. New York: General Learning Press, 1971.
448 Arkin et ai.