Egocentric Biases in Availability and...

Journal of Personality and Social Psychology1979, Vol. 37, No. 3, 322-336

Egocentric Biases in Availability and Attribution

Michael Ross and Fiore SicolyUniversity of Waterloo, Waterloo, Canada

Five experiments were conducted to assess biases in availability of informationin memory and attributions of responsibility for the actions and decisions thatoccurred during a previous group interaction. The subject populations sampledincluded naturally occurring discussion groups, married couples, basketballteams, and groups assembled in the laboratory. The data provided consistentevidence for egocentric biases in availability and attribution: One's own con-tributions to a joint product were more readily available, that is, more fre-quently and easily recalled; individuals accepted more responsibility for a groupproduct than other participants attributed to them. In addition, statements at-tributed to the self were recalled more accurately and the availability bias wasattenuated, though not eliminated, when the group product was negatively eval-uated (Experiment 2). Finally, when another participant's contributions weremade more available to the individual via a selective retrieval process, theindividual allocated correspondingly more responsibility for the group decisionsto the coparticipant (Experiment 5). The determinants and pervasiveness ofthe egocentric biases are considered.

One instance of a phenomenon examinedin the present experiments is familiar to al-most anyone who has conducted joint re-search. Consider the following: You haveworked on a research project with anotherperson, and the question arises as to whoshould be "first author" (i.e., who contributedmore to the final product?). Often, it seemsthat both of you feel entirely justified inclaiming that honor. Moreover, since you areconvinced that your view of reality must beshared by your colleague (there being onlyone reality), you assume that the other personis attempting to take advantage of you. Some-times such concerns are settled or preventedby the use of arbitrary decision rules, for ex-ample, the rule of "alphabetical priority"—

This research was supported by a Canada Councilgrant to the first author. We are grateful to the fol-lowing people for their comments on earlier versionsof this article: Dick Bootzin, Leslie McArthur,Hildy Ross, Lee Ross, Shelley Taylor, Amos Tversky,and Mark Zanna. We thank Vicky Vetere for col-lecting the data in Experiment S.

Requests for reprints should be sent to MichaelRoss, Department of Psychology, University ofWaterloo, Waterloo, Ontario, Canada N2L 3G1.

a favorite gambit of those whose surnamesbegin with letters in the first part of thealphabet.

We suggest, then, that individuals tend toaccept more responsibility for a joint productthan other contributors attribute to them. Itis further proposed that this is a pervasivephenomenon when responsibility for a jointventure is allocated by the participants. Inmany common endeavors, however, the par-ticipants are unaware of their divergent views,since there is no need to assign "authorship";consequently, the ubiquity of the phenomenonis not readily apparent. The purpose of thecurrent research was to assess whether theseegocentric perceptions do occur in a varietyof settings and to examine associated psy-chological processes.

In exploring the bases of such differentialperceptions, we are not so naive as tosuggest that intentional self-aggrandisementnever occurs. Nonetheless, it is likely thatperceptions can be at variance in the absenceof deliberate deceit; it is from this perspec-tive that we approach the issue.

To allocate responsibility for a joint en-deavor, well-intentioned participants presum-

Copyright 1979 by the American Psychological Association, Inc. 0022-3514/79/3703-0322$00.75

322

EGOCENTRIC BIASES AND ATTRIBUTION 323

ably attempt to recall the contributions eachmade to the final product. Some aspects ofthe interaction may be recalled more readily,or be more available, than others, however.In addition, the features that are recalledeasily may not be a random subset of thewhole. Specifically^ a person may recall agreater proportion of his or her own contribu-tions than would other participants.

An egocentric bias in availability of in-formation in memory, in turn, could producebiased attributions of responsibility for ajoint product. As Tversky and Kahneman(1973) have demonstrated, people use avail-ability, that is, "the ease with which relevantinstances come to mind" (p. 209), as a basisfor estimating frequency. Thus, if self-gen-erated inputs were indeed more available, in-dividuals would be likely to claim more re-sponsibility for a joint product than otherparticipants would attribute to them.

There are at least four processes that maybe operating to increase the availability ofone's own contributions: (a) selective en-coding and storage of information, (b) dif-ferential retrieval, (c) informational dispar-ities, and (d) motivational influences.

Selective Encoding and Storage

For a number of reasons, the availabilityof the person's own inputs may be facilitatedby differential encoding and storage of self-generated responses. First, individuals' ownthoughts (about what they are going to saynext, daydreams, etc.) or actions may dis-tract their attention from the contributionsof others. Second, individuals may rehearseor repeat their own ideas or actions; for ex-ample, they might think out their positionbefore verbalizing and defending it. Conse-quently, their own inputs may receive more"study time," and degree of retention isstrongly related to study time (Carver, 1972).Third, individuals' contributions are likelyto fit more readily into their own cognitiveschema, that is, their unique conception ofthe problem based on past experience, values,and so forth. Contributions that fit into suchpreexisting schemata are more likely to beretained (Bartlett, 1932; Bruner, 1961).

Differential Retrieval

The availability bias could also be pro-duced by the selective retrieval of informa-tion from memory. In allocating responsibil-ity for a joint outcome, the essential questionfrom each participant's point of view maybe, "How much did / contribute?" Partici-pants may, therefore, attempt to recall prin-cipally their own contributions and inappro-priately use the information so retrieved toestimate their relative contributions, a judg-ment that cannot properly be made withouta consideration of the inputs of others as well.

Informational Disparities

There are likely to be differences in theinformation available to the contributors thatcould promote egocentric recall. Individualshave greater access to their own internalstates, thoughts, and strategies than do ob-servers. Moreover, participants in a com-mon endeavor may differ in their knowledgeof the frequency and significance of eachother's independent contributions. For ex-ample, faculty supervisors may be less awarethan their student colleagues of the amount oftime, effort, or ingenuity that students investin running subjects, performing data analyses,and writing preliminary drafts of a paper. Onthe other hand, supervisors are more cog-nizant of the amount and of the importanceof the thought, reading, and so on that theyput into the study before the students' in-volvement begins.

Motivational Influences

Motivational factors may also mediate anegocentric bias in availability. One's sense ofself-esteem may be enhanced by focusing on,or weighting more heavily, one's own inputs.Similarly, a concern for personal efficacy orcontrol (see deCharms, 1968; White, 1959)could lead individuals to dwell on their owncontributions to a joint product.

The preceding discussion outlines a num-ber of processes that may be operating torender one's own inputs more available (andmore likely to be recalled) than the contribu-tions of others. Consequently, it may be dif-

324 MICHAEL ROSS AND FIORE SICOLV

ficult to imagine a disconfirmation of thehypothesis that memories and attributions areegocentric. As Greenwald (Note 1) has ob-served, however, the egocentric character ofmemory "is not a necessary truth. It is pos-sible, for example, to conceive of an organiza-tion of past experience that is more like thatof some reference work, such as a historytext, or the index of a thesaurus" (p. 4). Inaddition, we were unable to find publisheddata directly supportive of the hypothesizedbias in availability. Finally, recent develop-ments in the actor-observer literature seeminconsistent with the hypothesis that mem-ories and attributions are egocentric. Jonesand Nisbett (1971) speculated that actorsare disposed to locate the cause of their be-havior in the environment, whereas observersattribute the same behavior to stable traitspossessed by the actors. Though a variety ofexplanations were advanced to account forthis effect (Jones & Nisbett, 1971), the re-cent emphasis has been on perceptual in-formation processing (Storms, 1973; Taylor& Fiske, 1975). The actor's visual receptorsare aimed toward the environment; an ob-server may focus directly on the actor. Thus,divergent aspects of the situation are salientto actors and observers, a disparity that is re-flected in their causal attributions. This pro-posal seems to contradict the thesis that ac-tors in an interaction are largely self-absorbed.

Two studies offer suggestive evidence forthe present hypothesis. Rogers, Kuiper, andKirker (1977) showed that trait adjectiveswere recalled more readily when subjectshad been required to make a judgment aboutself-relevance (to decide whether each traitwas descriptive of them) rather than abouta number of other dimensions (e.g., syno-nymity judgments). These data imply thatself-relevance increases availability; however,Rogers et al. did not contrast recall of ad-jectives relevant to the self with recall ofadjectives relevant to other people—a com-parison that would be more pertinent to thecurrent discussion. Greenwald and Albert(1968) found that individuals recalled theirown arguments on an attitude issue more ac-curately than the written arguments of othersubjects. Since the arguments of self and

other were always on opposite sides of theissue, the Greenwald and Albert finding couldconceivably reflect increased familiarity with,and memory for, arguments consistent withone's own attitude position rather than en-hanced memory for self-generated statements(although the evidence for attitude-biasedlearning is equivocal, e.g., Greenwald & Saku-mura, 1967; Malpass, 1969).

We conducted a pilot study to determinewhether we could obtain support for thehypothesized bias in availability. Students inan undergraduate seminar were asked toestimate the number of minutes each mem-ber of the seminar had spoken during theimmediately preceding class period. An addi-tional 26 subjects were obtained from natu-rally occurring two-person groups approachedin cafeterias and lounges. The participantsin these groups were asked to estimate thepercentage of the total time each person hadspoken during the current interaction.

It was assumed that subjects would basetheir time estimates on those portions of theconversation they could recall readily. Thus,if there is a bias in the direction of betterrecall of one's own statements, individuals'estimates of the amount of time they them-selves spoke should exceed the average speak-ing time attributed to them by the othermember(s) of the group.

The results were consistent with this rea-soning. For seven of the eight students in theundergraduate seminar, assessments of theirown discussion time exceeded the averagetime estimate attributed to them by the otherparticipants (p < .05, sign test). Similarly,in 10 of the 13 dyads, estimates of one's owndiscussion time exceeded that provided by theother participant (p < .05, sign test). Themagnitude of the bias was highly significantover the 13 dyads, F(l, 12) = 14.85, p<.005; on the average, participants estimatedthat they spoke 59% of the time. These dataprovide preliminary, albeit indirect, evidencefor the hypothesized availability bias ineveryday situations.

The principle objectives of the current re-search were (a) to assess the ocurrence ofegocentric biases in availability and attribu-tions of responsibility in different settings; (b)


to examine factors that were hypothesizedto influence these biases; and (c) to offerpreliminary evidence of a relation between abias in availability and a bias in attributionsof responsibility. Experiment 1 assessed theoccurrence of egocentric biases in availabilityand allocations of responsibility in a naturalsetting and examined the relation betweenthe two biases. Next, a laboratory experi-ment was conducted to address the issue ofwhether the quality of the group's perform-ance affects the availability bias: Is the tend-ency for one's own inputs to be more avail-able reduced substantially when the group'sperformance is poor, as a motivational in-terpretation would suggest? Experiment 3further examined the effects of success andfailure in a natural setting. The experimentalmanipulations in Experiments 4 and 5 weredesigned to influence availability, and changesin attributions of responsibility were as-sessed. The manipulation in Experiment 4induced differential encoding; the manipula-tion in Experiment 5 varied the retrievalcues provided to the subjects.

Experiment 1

In this experiment, we wished to examineegocentric biases in naturally occurring, con-tinuing relationships. Married couples ap-peared to represent an ideal target group.Spouses engage in many joint endeavors ofvarying importance. This circumstance wouldappear to be rife with possibilities for ego-centric biases.

Accordingly, the first experiment was con-ducted (a) to determine if egocentric biasesin allocations of responsibility occur in mari-tal relationships; (b) to replicate, using adifferent dependent measure, the egocentricbias in availability obtained in the pretest;and (c) to correlate the bias in availabilitywith the bias in responsibility. If the bias inresponsibility is caused by a bias in avail-ability, the two sets of data should be related.

Method

Subjects

The subjects were 37 married couples living instudent residences. Twenty of the couples had

children. The subjects were recruited by two femaleresearch assistants who knocked on doors in theresidences and briefly described the experiment. Ifthe couple were willing to participate, an appoint-ment was made. The study was conducted in thecouple's apartment; each couple was paid $5 forparticipating.

Procedure

A questionnaire was developed on the basis ofextensive preliminary interviews with six marriedcouples. In the experiment proper, the questionnairewas completed individually by the husband andwife; their anonymity was assured. The first pagesof the questionnaire required subjects to estimatethe extent of their responsibility for each of 20activities relevant to married couples by putting aslash through a 150-mm straight line, the endpointsof which were labeled "primarily wife" and "pri-marily husband." x The twenty activities were mak-ing breakfast, cleaning dishes, cleaning house, shop-ping for groceries, caring for your children, planningjoint leisure activities, deciding how money shouldbe spent, deciding where to live, choosing friends,making important decisions that affect the two ofyou, causing arguments that occur between the twoof you, resolving conflicts that occur between thetwo of you, making the house messy, washing theclothes, keeping in touch with relatives, demonstrat-ing affection for spouse, taking out the garbage, ir-ritating spouse, waiting for spouse, deciding whether,to have children.

Subjects were next asked to record briefly ex-amples of the contributions they or their spousesmade to each activity. Their written records weresubsequently examined to assess if the person's owninputs were generally more "available." That is,did the examples reported by subjects tend to focusmore on their own behaviors than on their spouses'?A rater, blind to the experimental hypothesis, re-corded the number of discrete examples subjectsprovided of their own and of their spouses' con-tributions. A second rater coded one third of thedata; the reliability (Pearson product-moment cor-relation) was .81.

Results

The responses of both spouses to each ofthe responsibility questions were summed,

1 In the preliminary interviews, we used percentageestimates. We found that subjects were able to re-member the percentages they recorded and thatpostquestionnaire comparisons of percentages pro-vided a strong source of conflict between the spouses.The use of the 150-mm scales circumvented thesedifficulties; subjects were not inclined to converttheir slashes into exact percentages that could thenbe disputed.

326 MICHAEL ROSS AND FIORE SICOLY

so that the total included the amount that thewife viewed as her contribution and theamount that the husband viewed as his con-tribution. Since the response scale was ISOmm long, there were ISO "units of responsi-bility" to be allocated. A sum of greater thanISO would indicate an egocentric bias inperceived contribution, in that at least oneof the spouses was overestimating his or herresponsibility for that activity. To assess thedegree of over- or underestimation thatspouses revealed for each activity, ISO wassubtracted from each couple's total. A com-posite score was derived for the couple, aver-aging over the 20 activities (or 19, when thecouple had no children).

An analysis of variance, using the coupleas the unit of analysis, revealed that thecomposite scores were significantly greaterthan zero, M = 4.67, F(l, 35) = 12.89, p< .001, indicating an egocentric bias in per-ceived contributions. Twenty-seven of the 37couples showed some degree of overestima-tion (p < .025, sign test). Moreover, on theaverage, overestimation occurred on 16 of the20 items on the questionnaire, including nega-tive items—for example, causing argumentsthat occur between the two of you, F(l, 32)= 20.38, p < .001. Although the magnitudeof the overestimation was relatively small,on the average, note that subjects tended touse a restricted range of the scale. Most re-sponses were slightly above or slightly belowthe halfway mark on the scale. None of theitems showed a significant underestimationeffect.

The second set of items on the question-naire required subjects to record examplesof their own and of their spouses' contribu-tions to each activity. A mean differencescore was obtained over the 20 activities(averaging over husband and wife), with thenumber of examples of spouses' contributionssubtracted from the number of examples ofown contributions. A test of the grand meanwas highly significant, F(l, 35) = 36.0, p< .001; as expected, subjects provided moreexamples of their own (M = 10.9,) than oftheir spouses' (M = 8.1) inputs. The cor-relation between this self-other differencescore and the initial measure of perceived

responsibility was determined. As hypothe-sized, the greater the tendency to recall self-relevant behaviors, the greater was the over-estimation in perceived responsibility, r(3S)= .50, p < .01.

The number of words contained in eachbehavioral example reported by the subjectswas also assessed to provide a measure ofelaboration or richness of recall. The meannumber of words per example did not differas a function of whether the behavior wasreported to be emitted by self (M = 10.0)or spouse (M = 10.1), F< 1. Further, thismeasure was uncorrelated with the measureof perceived responsibility, r(35) = —.15, ns.

In summary, both the measure of responsi-bility and the measure reflecting the avail-ability of relevant behaviors showed the hy-pothesized egocentric biases. Moreover, therewas a significant correlation between the mag-nitude of the bias in availability and the mag-nitude of the bias in responsibility. Thisfinding is consistent with the hypothesis thategocentric biases in attributions of responsi-bility are mediated by biases in availability.Finally, the amount of behavior recalledseemed to be the important factor, ratherthan the richness of the recall.

Experiment 2

The data from Experiment 1 indicate thategocentric biases in availability and attribu-tions of responsibility occur in ongoing rela-tionships. The remaining experiments weredesigned to demonstrate the prevalence ofthese phenomena, and to, investigate someof the factors that were expected to influencetheir magnitude.

The major purpose of Experiment 2 wasto evaluate the self-esteem interpretation ofthe availability bias. If the availability biasis caused primarily by the motivation to en-hance self-esteem, recall of a joint endeavorshould facilitate an acceptance of personalresponsibility after success and a denial ofpersonal responsibility after failure. Conse-quently, the self-esteem interpretation im-plies that self-generated inputs should bemore available after success than after fail-ure. The evidence from past research that


people accept more responsibility for a suc-cess than for a failure is consistent with thisreasoning (e.g., Luginbuhl, Crowe, & Kahan,197S; Sicoly & Ross, 1977; Wortman, Cos-tanzo, &Witt, 1973).

In Experiment 2, subjects learned severaldays after participating in a problem-solvingtask that their group had performed eitherwell or poorly. It was hypothesized thatsubjects would recall a greater proportion oftheir own statements when the group productwas positively evaluated. Because we movedto the laboratory for this experiment, it waspossible to tape record the group's initial in-teraction. This recording provided a "realitybase" against which to compare the subse-quent recall of subjects.

Method

Subjects

The subjects were 37 males and 7 females selectedfrom lists of students living at the university. Sub-jects were paid $5 each. All of the subjects par-ticipated in both sessions.

Procedure

The experiment was conducted in two sessionsseparated by a 3- or 4-day interval. Subjects re-ported for the first session in groups of two. Theywere told that the purpose of the study was to de-termine whether groups exhibit more social aware-ness than individuals. They were given 10 minutesto read a case study of Paula, a psychologicallytroubled person (selected from Goldstein & Palmer,1975). Each subject in the dyad was provided withdifferent portions of the case study. The subjectswere next asked questions designed to assess theirpsychological understanding of Paula's difficulties.They were told to discuss each question and arriveat a joint response, taking into account the differ-ent information that each group member broughtwith him or her to the case. Subjects were toldthat their discussions were being tape recorded.The experimenter informed them that she wouldlisten to the tapes following the session to evaluatethe group's answers.

Subjects returned individually for the second ses-sion. In a random one half of the dyads, subjectswere led to believe that their group had performedpoorly relative to other groups in the experiment(third from the worst). In the remaining dyads,subjects were informed that their group had per-formed relatively well (third from the best). Sub-jects were then told, "Write down as much as you

can recall of your group's discussion of Paula.You will only have a short time to do this, so itis unlikely that you will be able to report all oreven most of what was said. It is, therefore, im-portant that you put things down in the order thatthey come to you. . . . If you remember the idea,but not the exact comment, rephrase it in your ownwords." Subjects were told not to record who saideach statement. They were simply to write what wassaid.

Subjects were asked to stop writing at the end of8 minutes and to go back over their responses to in-dicate who said each statement during the discussion.Finally, they were asked whether, in their opinion,each statement improved or lowered their group'sscore, or whether they were uncertain. Subjectswere debriefed at the end of the second session.

An observer who was blind to the subjects' treat-ment conditions contrasted subjects' recall withtheir original comments on the tape to assess ac-curacy. A statement was judged to be accurate ifit represented an idea that the subject expressedduring the interaction, even though the actual wordsused during the discussion might have differed fromthe words recalled by the subject. A second raterscored a random one third of the tapes, and agree-ment was 93%.

Results and Discussion

Availability

The proportion of statements that subjectsattributed to themselves was calculated foreach member of the dyad. The average pro-portion for each dyad served as the unit ofanalysis. In 21 of the 22 dyads, the subjectsattributed the majority of the statementsthat they recalled to themselves (p < .001,sign test). The average proportion of sub-jects' own statements was .70 in the successcondition and .60 in the failure condition.Each of these proportions was significantlygreater than a .SO or chance expectancy,i(10) = 9.09, p < .001, and <(10) = 3.22, p< .01, respectively. Thus, in bdth the suc-cess and failure conditions, subjects attributedsignificantly more of the recalled statementsto themselves than would be expected bychance. Nevertheless, as hypothesized, sub-jects attributed a greater proportion of therecalled statements to themselves after asuccess than after a failure, F(l, 20) = 7.10,p < .025. The total number of statementsrecalled (adding over statements attributedto self and the other person) did not differ


significantly as a function of the group's per-formance (F = 1.57).

Accuracy

Subjects' recall was compared with thetaped record in a 2 X 2 between-within anal-ysis of variance (Success vs. Failure X Selfvs. Partner), with the dyad as the unit ofanalysis. Subjects recalled a higher percent-age of their own actual statements (M —5.6%) than of their partner's actual state-ments (M = 2.6%), F(l, 19) = 18.37, p <.001.-' Although the means seem low, notethat subjects were given only an 8-minute re-call period. The group's performance level didnot affect the percentage of actual statementsrecalled (main effect and interaction Fs < 1).

We also compared the accuracy of thestatements subjects attributed to themselveswith the accuracy of the statements they at-tributed to their partners. Sixty-nine per-cent of the statements that subjects attributedto self were accurate reflections of self-gen-erated comments; 56% of the statementsthat subjects attributed to their partnerswere accurate. The difference between thesetwo percentages was significant, F(l, 19) =7.06, p < .025. The group's performancelevel did not significantly affect the accuracyof the attributed statements (success-failmain effect F = 1.14, interaction F < I ) .

Most of the errors that subjects made wereof two types: They recalled material fromthe case history that had not been mentionedin the discussion; they reported inferencesand conclusions that were not contained inthe case history or in the discussion. In onlya few instances (approximately 2% of theerrors) did subjects take credit for statementsmade by their partners.

Evaluations

Finally, subjects' evaluations of the state-ments were transcribed onto a 3-point scale:+ 1 (improved the group's score), 0 (uncer-tain), and —1 (lowered the group's score).Two scores were obtained for each subject:the average rating of comments attributedto self and the average rating of statements

attributed to the other person. An analysis ofvariance, with the dyad as the unit of analy-sis, revealed a main effect for success-fail,F(l, 20) - 14.56, p < .005, and a Success-Fail X Self-Other interaction, F(l, 20) =5.19, p< .05.

The success-fail main effect indicated thatstatements were evaluated more positivelyfollowing success (M = .75) than followingfailure (M=.41) . The interaction revealedthat whereas subjects' evaluations of theirown comments were marginally lower in thefailure condition than in the success condi-tion (M difference = .18, t = 1.85, p < .10),their evaluations of the other person's com-ments were significantly lower in the failurecondition than in the success condition (Mdifference = .50, t = 5.14, p < .01).

In summary, the present study providedsome evidence for the self-esteem maintenancehypothesis. Subjects attributed a higher pro-portion of the recalled comments to them-selves after success than after failure; sub-jects' evaluations of the recalled statementssuggested an attempt to shift the blame forfailure onto their partners. On the otherhand, contrary to the self-esteem interpreta-tion, recall was egocentric even in the failurecondition.

Note that the strong egocentricity ob-tained on the recall measure and the increasedaccuracy of self-generated statements mayreflect, in part, the fact that subjects initiallyread different aspects of the case history.Since they subsequently presented this ma-terial to the other person in responding tothe questions, subjects' own contributionsmay have received more "study time." Never-theless, this differential is ecologically valid.A person's inputs are often derived from hisor her previous history and experiences.

Experiment 3

In Experiment 3 we examined the effectsof success and failure in a more natural set-ting. We had the players on 12 intercollegiatebasketball teams individually complete a

- The tapes from one of the failure groups werelost; this group is omitted from the analysis.


questionnaire in which they were asked torecall an important turning point in theirlast game and to assess why their team hadwon or lost.

It is a leap to go from the self-other com-parisons that we have considered in the pre-vious studies to own team-other team com-parisons. There are, however, a number ofreasons to expect that the actions of one'sown team should be more available to theattributor than the actions of the other team:I know the names of my teammates, andtherefore, I have a ready means of organizingthe storage and retrieval of data relevant tothem; our success in future games againstother opponents depends more on our own of-fensive and defensive abilities than on theabilities of the opposing team. Consequently,I may attend more closely to the actions ofmy teammates, which would enhance encodingand storage. Also, there are informational dis-parities: The strategies of my own team aremore salient than are the strategies of theopposing team (Tversky & Kahneman, 1973).

If the initiatives of one's own team aredifferentially available, players should recalla turning point in terms of the actions oftheir team and attribute responsibility forthe game outcome to their team. On the basisof the data from Experiment 2, it may be ex-pected that these tendencies will be strongerafter a win than after a loss.

Method

Subjects

Seventy-four female and 84 male intercollegiatebasketball players participated in the study. Theteam managers were contacted by telephone; allagreed, following discussions with their players, tohave their teams participate in the study.

Procedure

The questionnaires were administered after sixgames in which the teams participating in thestudy played each other. Thus, for the three malegames chosen, three of the six male teams in thestudy were competing against the other three maleteams. Similarly, the three female games selectedincluded all six of the female teams. The question-naires were administered at the first team practicefollowing the target game (1 or 2 days after thegame), except in one case where, because of the

teams' schedules of play, it was necessary to col-lect data immediately after the game (two femaleteams). The questionnaires were completed indi-vidually, and the respondents' anonymity was as-sured. The relevant questions, from the current per-spective, were the following:

1. Please describe briefly one important turningpoint in the last game and indicate in whichperiod it occurred.

2. Our team won/lost our last game because. . . .

The responses to the first question were examinedto determine if the turning point was describedas precipitated by one's own team, both teams, orthe other team. Responses to the second questionwere examined to assess the number of reasons forthe win or loss that related to the actions of eitherone's own or the opposing team. The data werecoded by a person who was unaware of the experi-mental hypotheses. A second observer independentlycoded the responses from 50% of the subjects.There was 100% agreement for both questions.

Results

There were no significant sex differences onthe two dependent measures; the results are,therefore, reported collapsed across gender.Since team members' responses cannot beviewed as independent, responses were aver-aged, and the team served as the unit ofanalysis.

A preliminary examination of the "turningpoint" data revealed that even within a team,the players were recalling quite differentevents. Nevertheless, 119 players recalled aturning point that they described as pre-cipitated by the actions of their own team;13 players recalled a turning point that theyviewed as caused by both teams; 16 playersrecalled a turning point seen to be initiatedby the actions of the opposing team (theremaining 10 players did not answer thequestion). Subjects described such events asa strong defense during the last 2 minutes ofthe game, a defensive steal, a shift in offen-sive strategies, and so on.

The percentage of players who recalled aturning point caused by their teammates wasderived for each team. These 12 scores weresubmitted to an analysis that compared themto a chance expectancy of 50%. The ob-tained distribution was significantly differentfrom chance, F(\, 11) = 30.25, p< .001,with a mean of 80.25%. As hypothesized,


most reports emphasized the actions of theplayers' own team.

The percentage of players who recalleda turning point caused by their teammateswas examined in relation to the team's per-formance. The average percentage was higheron the losing team than on the winning teamin five of the six games (p < .11, sign test).The mean difference between the percentageson losing (M = 88.5) and winning (M =72.) teams was nonsignificant (F < 1).

The players' explanations for their team'swin or loss were also examined. Of the 158participants, only 14 provided any reasons thatinvolved the actions of the opposing team.On the average, subjects reported 1.79 rea-sons for the win or loss that involved theirown team and .09 reasons that involved theopposing team, F(\, 11) = 272.91, p < .001.Finally, the tendency to ascribe more rea-sons to one's own team was nonsignificantlygreater after a loss (M — 1.73) than aftera win (M = 1.65), F < 1.

Discussion

The responses to the turning point ques-tion indicate that the performances of sub-jects' teammates were more available thanthose of opposing team members. Further,subjects ascribed responsibility for the gameoutcome to the actions or inactions of theirteammates rather than to those of membersof the opposing team. Thus, biases in avail-ability and judgments of responsibility canoccur at the group level. Rather and Hes-kowitz (1977) provide another example ofgroup egocentrism: "CBS (news) became asolid Number One after the Apollo moonshotin 1968. If you are a CBS person, you tendto say our coverage of the lunar landingtipped us over. If you are a NBC person, youtend to cite the break-up of the Huntley-Brinkley team as the key factor" (p. 307).

Contrary to the data from Experiment 2,the availability bias in Experiment 3 was asstrong after failure as after success. Thereare differences between the studies that maycontribute to this discrepancy. The "ego-centric" availability and attributions in thebasketball experiment were team rather than

self-oriented; as a result, responsibility forfailure was more diffused, and subjects' self-esteem was threatened less directly. Also, un-like the group in the laboratory study, thebasketball team had a future: The playerscould enhance their control over subsequentgame outcomes by locating causality withintheir own team. Finally, and perhaps mostimportant, unlike the laboratory group, theteam also had a past. Team members recalledaspects of their behavior that changed and at-tributed the game outcomes to these varia-tions (e.g., we win because of discipline andhustle; we lose because of a lack of disciplineand hustle). What players seemed to ignore,however, was that the opposing teams mightcontribute to these fluctuations.

It seems likely that a tendency to perceiveboth teams as responsible for the game out-come might increase with the magnitude ofthe win or loss (assuming that large wins orlosses are atypical). As Kelley (1973) noted,multiple causes are necessary to explain ex-treme outcomes. Although no such tendencywas observed in the current study, there weretoo few data points (games) to provide anaccurate determination.

Experiment 4

In the final two experiments, we examinedthe hypothesized relation between the bias inavailability and the bias in attributions ofresponsibility more directly by introducingmanipulations that should affect availabilityand measuring changes in attribution. InExperiment 4, subjects were required torecord either their own comments (self-focuscondition) or those of the other person (part-ner-focus condition) during a problem-solv-ing session. At a second session, subjectswere shown their notes and asked to assessthe extent to which either they or their part-ners had been responsible for various aspectsof the decision-making process. It was as-sumed that the partner-focus condition wouldenhance encoding and retrieval of the part-ner's contributions. Thus, the partner's inputswould be more available when assessmentsof responsibility were made, and subjectsshould assign their partner more responsi-


bility for group decision-making in the part-ner-focus condition than in the self-focuscondition.

Method

Subjects

The subjects were 40 males recruited from theintroductory psychology subject pool.

Procedure

Subjects were scheduled in pairs for the first ses-sion. In a few introductory comments, the experi-menter described the difficulty of preventing peoplefrom smoking. Subjects were then told they wereparticipating in a pilot project to assess the ef-ficacy of "brainstorming techniques" as a means ofproviding possible solutions to this problem. Theywere further informed that solutions generated dur-ing their discussion would be sent to the Committeefor the Prevention of Cigarette Smoking (a govern-ment committee). Subjects were told to follow afour-step sequence: define the problem, generate asmany solutions as possible, discuss the pros and consof each proposed solution, and finally, select a pre-ferred solution and explain the reasons for thischoice.

Subjects in the self-focus condition were askedto keep a record of their own contributions to thediscussion. Subjects in the partner-focus conditionwere asked to keep a record of only the other per-son's inputs: "This will leave you free to think anddevelop your ideas because your partner will bedoing the writing." Subjects were given about 45minutes for discussion.

Subjects returned individually for a second session2-3 days later. Each subject was asked to look overthe notes he had taken during the previous session,"in order to refresh your recollection of the dis-cussion."

Subjects completed the dependent measures afterthey had reviewed their notes of Session 1. Theprincipal dependent variable required subjects toindicate who tended to control the course and con-tent of the discussion during the first session. Theywere asked to assess this overall, and also with re-spect to the various stages of the discussion, onISO-mm scales with endpoints labeled "the otherperson" and "me."

Results

A 2 X 2 analysis of variance was performedon the data. Self- versus partner focus was abetween-subjects factor. Since the dyad wastreated as an experimental unit, the responsemade by each member of the pair was con-sidered to be a repeated measure.

The focus manipulation had no reliableimpact on attributions of responsibility (all^s < 1). Once again, however, there wasstrong evidence of an egocentric bias in al-locations of responsibility. Subjects reportedthat they had exerted more control over thecourse and content of each segment of thediscussion than their partners ascribed tothem—solutions stage: 85 versus 68, F(l, 18)= 7.32, p < .025; evaluation stage: 79 ver-sus 70, F(l, 18) = 4.13, p < .07; final pro-posal stage: 86 versus 67, F(l, 18) = 11.31,p < .01; overall discussion: 89 versus 72,F(l, 18) = 9.21, p < .01. Note that for eachitem, A's self-attributions were, on the aver-age, beyond the midpoint of the 150-mmscale, indicating a perceived contribution ofgreater than 50%; on the other hand, thepartner viewed A's contributions as beingless than 50% in each instance.

Discussion

The results revealed strong egocentricbiases in individuals' attributions of responsi-bility for segments of the problem-solvingtask. The focus manipulation had surprisinglylittle effect, however. What we had viewed tobe a sledgehammer manipulation turned outto be ineffective.

Why did the attention manipulation haveso little impact? One possibility is that sub-jects may have found the written records tobe relatively uninformative and relied moreon their memories than on the notes in re-sponding to the questionnaire (hence, thestrong egocentric biases). The notes may haveappeared inadequate, in part because theywere very brief, usually about one page, rela-tive to the length of the interaction (45 min-utes). Moreover, much of what subjects wrotemay have seemed irrelevant to the final deci-sions made by the group. In short, we maynot have succeeded in focusing subjects' at-tention on what, from their perspective, werethe important aspects of the interaction. Toobtain this information they were, perhaps,only too willing to rely on their memories.

Finally, note that recent research on therelation between attention and recall in in-terpersonal perception settings has yielded


inconsistent results (Taylor & Fiske, 1978).This situation stands in marked contrast tothe strong relation evident in the cognitiveliterature (Cofer, 1977; Loftus & Loftus,1976). Unlike the constrained experimentalsettings utilized in cognitive research, how-ever, the present study and those reviewedby Taylor and Fiske incorporate rich socialenvironments. Consequently, the manipula-tion of attention is relatively gross; it is lesscertain that the individual is attending tothose aspects of the situation that are rele-vant to the dependent measures.

Experiment 5

In Experiment 5, we again attempted tovary the individual's focus of attention so asto affect availability. In this experiment, how-ever, we employed a manipulation designedto promote selective retrieval of informationdirectly relevant to attributions of responsi-bility.

In our initial analysis, we suggested thategocentric attributions of responsibility couldbe produced by the selective retrieval of in-formation from memory and that retrievalmight be guided by the kinds of questionsthat individuals ask themselves. Experiment5 was conducted to test this hypothesis. Sub-jects were induced to engage in differing re-trieval by variations in the form in whichquestions were posed. Graduate students werestimulated to think about either their owncontributions to their BA theses or the con-tributions of their supervisors. The amountof responsibility for the thesis that subjectsallocated to either self or supervisor was thenassessed. It was hypothesized that subjectswould accept less responsibility for the re-search effort in the supervisor-focus than inthe self-focus condition.

Method

Subjects

The subjects were 17 female and 12 male psy-chology graduate students. Most had completedeither 1 or 2 years of graduate school. All of thesestudents had conducted experiments that served astheir BA theses in their final undergraduate year.

Procedure

The subjects were approached individually intheir offices and asked to complete a brief ques-tionnaire on supervisor-student relations. None re-fused to participate. The two forms of the ques-tionnaire were randomly distributed to the sub-jects; they were assured that their responses wouldbe anonymous and confidential.

One form of the questionnaire asked the subjectsto indicate their own contribution to each of anumber of activities related to their BA theses. Thequestions were as follows: (a) "I suggestedpercent of the methodology that was finally em-ployed in the study." (b) "I provided per-cent of the interpretation of results." (c) "I initiated

percent of the thesis-relevant discussionswith my supervisor." (d) "During thesis-related dis-cussions I tended to control the course and contentof the discussion percent of the time, (e) "Allthings considered, I was responsible for per-cent of the entire research effort." (f) "How wouldyou evaluate your thesis relative to others done inthe department?"

The second form of the questionnaire was identi-cal to the above, except that the word / (self-focuscondition) was replaced with my supervisor (super-visor-focus condition) on Questions 1-5. Subjectswere asked to fill in the blanks in response to thefirst five questions and to put a slash through aISO-mm line, with endpoints labeled "inferior" and"superior," in response to Question 6.

Results and Discussion

For purposes of the analyses, it was as-sumed that the supervisor's and the student'scontribution to each item would add up to100%. Though the experiment was intro-duced as a study of supervisor-student rela-tions, it is possible that the students mayhave considered in their estimates the inputsof other individuals (e.g., fellow students).Nevertheless, the current procedure providesa conservative test of the experimental hy-pothesis. For example, if a subject responded20% to an item in the "I" version of thequestionnaire, it was assumed that his or hersupervisor contributed 80%. Yet the super-visor may have contributed only 60%, withan unspecified person providing the re-mainder. By possibly overestimating the su-pervisor's contribution, however, we are bias-ing the data against the experimental hy-pothesis: The "I" version was expected toreduce the percentage of responsibility al-located to the supervisor.


Subjects' responses to the first five ques-tions on the "I" form of the questionnairewere subtracted from 100, so that highernumbers would reflect greater contributionsby the supervisor in both conditions. Ques-tion 5 dealt with overall responsibility for theresearch effort. As anticipated, subjects allo-cated more responsibility to the supervisor inthe supervisor-focus (M = 33.3%) than inthe self-focus (M = 16.5%) condition, F(l,27) = 9.05, p < .01. The first four questionswere concerned with different aspects of thethesis, and the average response revealed asimilar result: supervisor-focus M — 33.34;self-focus M = 21.82; F(l ,27) = 5.34, p <,05. Finally, subjects tended to evaluate theirthesis more positively in the self-focus condi-tion than in the supervisor-focus condition:112.6 versus 94.6, F(l, 27) = 3.59, p < .10.

The contrasting wording of the questionshad the anticipated impact on subjects' allo-cations of responsibility. The supervisor ver-sion of the questionnaire presumably causedsubjects to recall a greater proportion of theirsupervisors' contributions than did the "I"form of the questionnaire. This differentialavailability was then reflected in the alloca-tions of responsibility. Note, however, thatthe questions were not entirely successful incontrolling subjects' retrieval. The supervisorwas allocated only one third of the responsi-bility for the thesis in the supervisor-focuscondition.

In light of the present data, the basketballplayers' attributions of responsibility for thegame outcome in Experiment 3 need to bereexamined. Recall that the players wereasked to complete the sentence, "Our teamwon/lost our last game because. . . ." Thisquestion yielded a highly significant ego-centric bias. With hindsight, it is evidentthat the form of the question—"Our team. . . our last game"—may have prompted sub-jects to focus on the actions of their ownteams, even though the wording does notpreclude references to the opposing team. The"turning point" question in Experiment 3was more neutrally worded and is not sus-ceptible to this alternative interpretation.

The leading questions in these studies ema-nate from an external source; many of ourretrieval queries are self-initiated, however,

and our recall may well be biased by theform in which we pose retrieval questions toourselves. For example, basketball playersare probably more likely to think in termsof "Why did we win or lose?" than in termsof a neutrally phrased "Which team was re-sponsible for the game outcome?"

General Discussion

The five studies employed different sub-ject populations, tasks, and dependent mea-sures. As hypothesized, the egocentric biasesin availability and attribution appear to berobust and pervasive.

Determinants of the Availability Bias

Several processes were hypothesized tocontribute to the increased availability ofself-generated inputs. It is possible to con-sider how well each accounts for the existingdata. Selective encoding and storage cannothave contributed to the effects of successversus failure on availability in Experiment2 or of supervisor- versus self-focus in Experi-ment 5 (since these manipulations occurredlong after encoding and storage took place).Informational disparities should not havecontributed to the pretest results (subjects'time estimates were based solely on the pre-ceding discussion), to the tendency to at-tribute a higher proportion of the recalledstatements to oneself in the success as com-pared to the failure condition in Experiment2, or to the effects of supervisor- versus self-focus in Experiment 5 (since neither perform-ance level, as operationalized here, nor focuscould affect the information initially avail-able to the subjects). Two motivational pro-cesses were posited. Self-esteem maintenancedoes not seem pertinent to the results ob-tained from the two-person groups in the pre-test. Nor does it account for (a) the over-recall of self-generated inputs in the failurecondition of Experiment 2 and (b) the find-ing that players on losing basketball teamsrecalled the turning point of the game interms of the actions of their teammates. Thecontrol motivation hypothesis fares somewhatbetter. Although focusing on one's own inputsin failure situations may lower self-esteem, it


does permit one to perceive personal controlover the activity. Hence, efficacy motivationcould account for these results. Nevertheless,a desire for personal efficacy does not appearto explain all of the data. The two-persongroups in the pretest seem to reveal a rela-tively "pure" information-processing effect:It is unlikely that people would feel a needto report that they dominated casual conver-sations. Also, the effect of supervisor- versusself-focus in Experiment 5 appears to bemediated by differential retrieval. Efficacyconsiderations may have induced the subjectsto report that they were major contributorsto their theses; nonetheless, motivational con-cerns do not dictate that focusing on thesupervisor's contributions will reduce one'sneed to assume responsibility.

In summary, selective encoding and stor-age, informational disparities, and motiva-tional influences do not appear to be neces-sary determinants of the egocentric bias inavailability. The one remaining process thatwas posited, selective retrieval, is not pre-cluded by any of the current data; further,it receives direct support from the findings inExperiments 2 and S.

Nevertheless, it seems premature to elimi-nate any of the hypothesized processes assufficient causes of the availability bias. Thetendency of spouses to recall their own con-tributions in Experiment 1 may reflect infor-mational disparities; the desire to maintainself-esteem may have contributed to theeffect of performance level in Experiment 2;basketball players' responses to the turningpoint question in Experiment 3 may well havebeen influenced by selective encoding and bycontrol motivation.

We suspect that, like many cognitive phe-nomena (cf. Erdelyi, 1974; Erdelyi & Gold-berg, in press; McGuire, 1973), biases inavailability are multidetermined in real life.Multidetermination may seem an unsatisfy-ing resolution; however, it is one that socialpsychologists shall probably confront increas-ingly as they begin to study cognitive phe-nomena in situ. Researchers in other sciencesface parallel complexities. For example, simi-lar cancers appear to have different etiolo-gies, depending, among other factors, on the

environment in which the patient lives (Good-field, 1976).

The Link Between Availability andAttributions of Responsibility

The focus of the present research has beenon demonstrating that the hypothesized biasesin availability and attribution exist and arerelatively ubiquitous. It was also hypothe-sized, however, that the egocentric bias inattributions of responsibility would be medi-ated by the bias in availability. Although thedata are suggestive, we have no definitiveevidence that the bias in availability causesthe bias in responsibility. The strongestaffirmative evidence is that the two biaseswere significantly correlated in the marriagestudy and that a manipulation designed toinduce selective retrieval influenced attribu-tions of responsibility (Experiment 5). Inopposition, it might be contended that thecovariation between the two biases is sus-ceptible to a number of alternative causalinterpretations and that there is no directevidence that the retrieval manipulation inExperiment 5 affected availability. Conceiva-bly, the attributions of responsibility in Ex-periment S were mediated by some otherfactor not yet identified. Further evidencewill be required to establish whether the biasin responsibility is caused by the bias inavailability. The present results suggest sev-eral additional considerations, however, con-cerning the determinants and pervasivenessof the biases.

Pervasiveness o) the Egocentric Biases

The egocentric biases obtained in the cur-rent studies may seem inconsistent with Jonesand Nisbett's (1971) proposal that actorslocate causality for their actions primarilywithin their environment. There are a num-ber of differences between the two paradigmsthat might account for the discrepancy. Mostimportant, Jones and Nisbett were concernedwith interpretation, whereas we focused onrecall and judgments of responsibility. Actorscould presumably overestimate their contri-butions to a joint product and, at the sametime, locate the cause of their behavior with-


in the environment. For example, supposethat a wife who reports that she does 80% ofthe cleaning is asked why she cleans (the cen-tral question for Jones and Nisbett). Shemay respond that the house is dirty, an en-vironmental attribution. Conversely, her hus-band, who perhaps accepts 30% of the re-sponsibility for cleaning, may answer thesame question by pointing out that his wifehas a fetish for cleanliness, a trait attribu-tion.

Thus, the current data do not speak di-rectly to the Jones and Nisbett hypothesis.Nevertheless, our data do seem to contradictrelated evidence that the responses of actorsare more salient and available to observersthan to actors themselves (Storms, 1973;Taylor & Fiske, 1975). The critical variablemay be the extent to which the observer de-parts from a passive role and interacts withthe actor. When, as in the present research,individuals undertake complex social interac-tions, they alternate between the roles ofspeaker (actor) and listener (observer), yetmuch of their attention may be directed atplanning and executing their own responses.Although they do not attend to themselvesperceptually, they may be cognitively self-focused; therefore, self-generated inputs arelikely to be more available in recall. On theother hand, passive observers may concentrateon other persons in their environment. Also,observers may be less self-absorbed whentheir own responses require little attention,as, for example, when they enact well-prac-ticed behaviors (Langer, 1978, has specu-lated that a wide range of social behaviorsrequire minimal thought).

These instances notwithstanding, the pres-ent research demonstrates the prevalence ofself-centered biases in availability and judg-ments of responsibility. In everyday life,these egocentric tendencies may be over-looked when joint endeavors do not requireexplicit allocations of responsibility. If allo-cations are stated distinctly, however, thereis a potential for dissension, and individualsare unlikely to realize that their differencesin judgment could arise from honest evalua-tions of information that is differentiallyavailable.

Reference Note

1. Greenwald, A. G. The tolitarian ego: Fabrica-tion and revision of personal history. Unpub-lished manuscript, 1978.

References

Bartlett, F. C. Remembering, Cambridge, England:Cambridge University Press, 1932.

Bruner, J. S. The act of discovery. Harvard Educa-tional Review, 1961, 31, 21-32.

Carver, R. P. A critical review of mathagenic be-haviors and the effect of questions upon the re-tention of prose materials. Journal of Reading Be-havior, 1972, 4, 93-119.

Cofer, C. N. On the constructive theory of memory.In I. A. Uzgiris & F. Weizmann (Eds.), Thestructuring of experience. New York: PlenumPress, 1977.

deCharms, R. C. Personal causation: The internalaffective determinants of behavior. New York:Academic Press, 1968.

Erdelyi, M. H. A new look at the new look: Per-ceptual defense and vigilance. Psychological Re-view, 1974, 81, 1-25.

Erdelyi, M. H., & Goldberg, B. Let's not sweeprepression under the rug: Towards a cognitivepsychology of repression. In J. F. Kihlstrom &F. J. Evans (Eds.), Functional disorders of mem-ory. Hillsdale, N.J.: Erlbaum, in press.

Goldstein, M. J., & Palmer, J. O. The experience ofanxiety. New York: Oxford University Press, 1975.

Goodfield, J. The siege of cancer. New York: Dell,1976.

Greenwald, A. G., & Albert, R. D. Acceptance andrecall of improvised arguments. Journal of Person-ality and Social Psychology, 1968, 8, 31-34.

Greenwald, A. G., & Sakumura, J. S. Attitude andselective learning: Where are the phenomena ofyesteryear? Journal of Personality and SocialPsychology, 1967, 7, 387-397.

Jones, E. E., & Nisbett, R. E. The actor and theobserver: Divergent perceptions of the causes ofbehavior. Morristown, N.J.: General LearningPress, 1971.

Kelley, H. H. The process of causal attribution.American Psychologist, 1973, 28, 107-128.

Langer, E. J. Rethinking the role of thought in so-cial interaction. In J. H. Harvey, W. J. Ickes, &R. F. Kidd (Eds.), New directions in attributionresearch (Vol. 2). Potomac, Md.: Erlbaum, 1978.

Loftus, G. R., & Loftus, E. F. Human memory:The processing of information. Hillsdale, N.J.:Erlbaum, 1976.

Luginbuhl, J. E. R., Crowe, D. H., & Kahan, J. P.Causal attribution for success and failure. Journalof Personality and Social Psychology, 1975, 31,86-93.

Malpass, R. S. Effects of attitude on learning andmemory: The influence of instruction-induced sets.Journal of Experimental Social Psychology, 1969,5, 441-453.


McGuire, W. J. The yin and yang of progress insocial psychology: Seven Koan. Journal of Per-sonality and Social Psychology, 1973, 26, 446-456.

Rather, D., & Heskowitz, M. The camera neverblinks. New York: Ballantine Books, 1977.

Rogers, T. B., Kuiper, N. A., & Kirker, W. S. Self-reference and the encoding of personal informa-tion. Journal of Personality and Social Psychology,1977, 35, 677-688.

Sicoly, F., & Ross, M. The facilitation of ego-biased attributions by means of self-serving ob-server feedback. Journal of Personality and So-cial Psychology, 1977, 35, 734-741.

Storms, M. D. Videotape and the attribution pro-cess: Reversing actors' and observers' points ofview. Journal of Personality and Social Psychol-ogy, 1973, 27, 165-175.

Taylor, S. E., & Fiske, S. T. Point of view andperceptions of causality. Journal of Personalityand Social Psychology, 1975, 32, 439-445.

Taylor, S. E., & Fiske, S. T. Salience, attention, andattribution: Top of the head phenomena. In L.Berkowitz (Ed.), Advances in experimental socialpsychology (Vol. 11). New York: Academic Press,1978.

Tversky, A., & Kahneman, D. Availability: A heur-isitc for judging frequency and probability. Cog-nitive Psychology, 1973, 5, 207-232.

White, R. W. Motivation reconsidered: The conceptof competence. Psychological Review, 1959, 66,297-333.

Wortman, C. B., Costanzo, P. R., & Witt, T. R.Effect of anticipated performance on the attribu-tion of causality to self and others. Journal ofPersonality and Social Psychology, 1973, 27, 372-381.

Received February 20, 1978

Egocentric Biases in Availability and...

Documents

Transcript of Egocentric Biases in Availability and...