Content uploaded by Craig A Anderson
Author content
All content in this area was uploaded by Craig A Anderson on Oct 24, 2016
Content may be subject to copyright.
9. Shortcomings in the attribution
process: On the origins and maintenance of
erroneous social assessments
Lee
Ross
and Craig A. Anderson -
Introduction to attribution theory and attributional errors
Attribution theory and intuitive psychology
Attribution theory, in its broadest sense, is concerned with the attempts of
ordinary people to understand the causes and implications of the events
they witness. It deals with the "naive psychology'; of people as they
interpret their own behavior and the actions of others. The current
ascendancy of attribution theory in social psychology thus culminates a
long struggle to upgrade that discipline's conception of man. No longer
the stimulus-reponse (S-R) automaton of radical behaviorism, promoted
beyond the rank of information processor and cognitive consistency
seeker, psychological man has at last been awarded a status equal to that of
the scientist who investigates him. For in the perspective of attribution
theory, people are intuitive psychologists who seek to explain behavior
and to draw inferences about actors and about their social environments.
To better understand the perceptions and actions of this intuitive
scientist we must explore his methods. First, like the academic psycholo-
gist, he is guided by a number of implicit assumptions about human
nature and human behavior -for example, that the pursuit of pleasure and
the avoidance of pain are ubiquitous and powerful human motives, or that
conformity to the wishes and expectations of one's peers is less exceptional
and less demanding of further interpretation than non-conformity. The
lay psychologist, like the professional one, also relies heavily upon data,
albeit data that rarely satisfy formal requirements regarding randomness
This chapter draws heavily, both in content and organization, from a contribution by the first
author to Advances in Experimental Social Psychology (1977). Permission for use of these
materials is gratefully acknowleJged.
130 CAUSALITY AND ATTRIBUTION Shortcomings in the attribution process 131
or representativeness. Sometimes these data result from first-hand experi-
ences; more often, they are the product of informal social communication,
mass media, or other indirect sources. The intuitive psychologist must
further adopt or develop techniques for coding, storing, and retrieving the
data. Finally, he must employ various strategies for summarizing,
analyzing, and interpreting the data -that s,s, he must employ rules,
heuristics, or schemas that permit him to form new inferences. The
intuitive scientist's ability to master his social environment, accordingly,
will depend upon the accuracy and adequacy of his hypotheses, evidence,
and analyses. Conversely, any systematic errors in existing theories, biases
in available data, or inadequacies in methods of analysis, yield serious
consequences -both for the lay psychologist and for the society that he
builds and perpetuates. These shortcomings, explored from the vantage
point of contemporary attribution theory, provide the focus of this chap-
ter.1
The broad outlines of attribution theory were first sketched by Heider
(1944, 1958) and developed in greater detail by Jones and Davis (1965),
Kelley (1967, 1971, 1973), and their associates (see Jones et aI., 1971;
Weiner, 1974). These theorists dealt with two closely related tasks
confronting the social observer. The first task is that of causal judgment:
The observer seeks to identify the cause, or set of causes, to which some
particular effect (Le., some action or outcome) may most reasonably be
attributed. The second task is that of social inference: The observer of an
episode forms inferences about the attributes or dispositions of the relevant
actors and about the attributes or properties of the situations to which they
have responded.
Causal judgment and social inference tasks have both been the subject of
intensive theoretical and empirical inquiry and, until recently, had consti-
tuted virtually the entire domain of attribution theory. Lately, however, a
third task of the intuitive psychologist has begun to receive some atten-
tion; that task is the prediction or estimation of outcomes and behavior. The
intuitive psychologist not only must seek explanations and make disposi-
tional inferences; he must also form expectations and make guesses about
actions and outcomes that are currently unknown or that will occur in the
future. For instance, when a presidential candidate promises to "ease the
burden of the average taxpayer," we consider possible causes for the
statement and implications about the candidate's personal dispositions.
(Did the promise simply reflect the demands of political expediency? Can
we conclude anything about the candidate's true convictions?) But we are
also likely to speculate about his subsequent behavior and his views on
related issues that have not yet been explored. (If elected, will he slash
property taxes? Does he favor curtailment of social welfare programs?) The
psychology of intuitive prediction, in short, is a natural extension of
attribution theory's domain.
Logically and psychologically, of course, the three attribution tasks are
interdependent. Explanations for an event, and inferences about the actors
and entities thaf figure in that event, are intimately related. And together
they provide the basis for speculation about the nature of events that are
currently unknown or are likely to unfold in the future. Each task,
however, offers unique possibilities (and unique problems of interpreta-
tion and methodology; see Ross, 1977, pp. 175-179) for revealing the
assumptions and strategies that underlie the intuitive scientist's perfor-
mance. It is worth noting that in recent years the use of estimations and
predictions as dependent variables in studies of lay inference has become
increasingly popular. One reason for this increased popularity is particu-
larly important. Unlike the causal judgments of dispositional inferences
that follow from a perceiver's analysis of an event, estimations or predic-
tions about new or unknown events can often be evaluated with respect to
their accuracy. That is, one can compare predictions and estimates about
events with actual observations or measurements. This permits assessment
both of the relative adequacy of the intuitive scientist's attributional
strategy and of the direction of specific errors or biases.
Logical attributional principles vs. self-serving biases
Contemporary attribution theory has pursued two distinct but comple-
mentary goals. One goal has been to demonstrate that social perceivers'
assessments and inferences generally follow the dictates of some logical or
rational model. The other goal has been to illustrate and explicate the
sources of bias or error that distort those generally veridical assessments
and inferences. We shall consider briefly the so-called logical, or rational,
schemata employed by the intuitive psychologist and then devote the
remainder of the chapter to the sources of error in his attempts at
understanding, predicting, and controlling the events that unfold around
him.
IFor a more thorough and systematic explanation of the layman/s(;entit;t parallel. the reader
is referred to Nisbett and Ross, 1980.
1ne "covariation" and "discounting" principles. Individuals must, for the most
part, share a common understanding of the social actions and outcomes
that affect them, for without such consensus, social interaction would be
chaotic, unpredictable, and beyond the control of the participants. Intro-
spection by attribution theorists, buttressed by some laboratory evidence,
has led to the postulation of a set of "rules" that may generally be
employed in the interpretation of behaviors and outcomes. These "com-
mon sense" rules or schemata are analogous, in some respects, to the more
formal rules and procedures that social scientists and statisticians follow in
their analysis and interpretation of data.
H. H. Kelley, E. E. Jones, and their associates have distinguished two
132 CAUSALITY AND ATTRIBUTION
cases in ~hich logical rules, or schemata, may be applied. In the multiple
observation case the attributer has access to behavioral data that might be
represented as rows or columns of an Actor x Object x Situation (or
Instance) response matrix. Typically, in this domain of research summary
statements are provided to the participants rather than actual responses.
Thus the potential attributer learns that "Most theatergoers like the new
Pinter play," or "Mary can't resist stray animals," or "The only television
program that Ann watches is Masterpiece Theatre." In the single observa-
tion case the attributer must deal with the behavior of a single actor on a
single occasion. For instance, he may see Sam comply with an experiment-
er's request to deliver a painful shock to a peer, or he may learn that
"Louie bet all his money on a long shot at Pimlico."
The logical rules or principles governing attributions in these two cases
are rather different (Kelley, 1967, 1971, 1973). In the multiple observation
case the attributer applies the "covariance principle"; that is, he assesses
the degree to which observed behaviors or outcomes occur in the
presence, but fail to occur in the absence, of each causal candidate under
consideration. Accordingly, the attributer concludes that the new Pinter
play is a good one to the extent that it is liked by a wide variety of
playgoers, that it is liked by individuals who praise few plays (e.g.,
"critics"), and that it is applauded as vigorously on the ninetieth day of its
run as on the ninth.
In the single observation case the attributer's assessment strategy
involves the application of the "discounting principle," by which the
social observer "discounts" the role of any causal candidate in explaining
an event to the extent that the other plausible causes or determinants can
be identified. This attributional principle can be restated in terms of social
inferences rather than causal attributions: To the extent that situational or
external factors constitute a "sufficient" explanation for an event, that
event is attributed to the situation and no inference logically can be made
(and, presumably, no inference empiricaHy is made) about the dispositions
of the actor. Conversely, to the extent that an act or outcome seems to
occur in spite of and not because of attendant situational forces, the
relevant event is attributed to the actor and a "correspondent inference"
(Jones & Davis, 1965) is made -that is, the attributer infers the existence
and influence of some trait, ability, intention, feeling, or other disposition
that could account for the actor's action or outcome. Thus, we resist the
conclusion that Louie's long-shot plunge at Pimlico was reflective of his
stable personal attributes to the extent that such factors as a hot tip, a
desperate financial crisis, or seven pre-wager martinis could be cited. On
the other hand, we judge Louie to be an inveterate long-shot player if we
learn that his wager occurred in the face of his wife's threat to leave him if
he ever lost his paycheck at the track again, his knowledge that he would
not be able to pay the rent if he lost, and a track expert's overheard remark
that the favorite in the race is "even better than the track odds suggest."
Shortcomings in the attribution process 133
Ii;
It is worth noting that the application of these two different principles
places rather different demands upon the intuitive scientist. The covar-
iance principle requires the attributer to apply rules that are essentially
logical or statistical in nature and demands no further insight about the
characteristics of the entities in question. Application of the discounting
principle, by contrast, demands considerable insight about the nature of
man and the impact of such situational forces as financial need, alcohol
consumption, and a spouse's threat of abandonment. In a sense, the
covariance principle can be applied by a mere "statistician," whereas the
discounting principle requires a "psychologist" able to assess the role of
various social pressures and situational forces and even to distinguish
intended acts and outcomes from unintended ones (cf. Jones & Davis,
1965).
Evidence concerning the systematic use of common-sense attributional
principles comes primarily from questionnaire studies in which subjects
read and interpret brief anecdotes about the responses of one or more
actors to specified objects or "entities" under specified circumstances (e.g.,
L. Z. McArthur, 1972, 1976). Occasional studies of narrower scope have
exposed the attributer to seemingly authentic responses, encounters, and
outcomes (e.g., Jones, Davis, & Gergen, 1961; Jones & DeCharms, 1957;
Jones & Harris, 1967; Strickland, 1958; Thibaut & Riecken, 1955). Such
research has demonstrated that attributers can, and generally do, make at
least some use of the hypothesized principles or rules of thumb. What the
methodologies employed to date have left ambiguous is the degree of the
layperson's accuracy and the magnitude and direction of his errors.
I
Self-serving motivational biases in attribution. In speculating about possible
distortions in an otherwise logical attribution system, theorists were quick
to postulate "ego-defensive" biases through which attributors maintained
or enhanced their general self-esteem or positive opinion of their specific
dispositions and abilities (Heider, 1958; Jones & Davis, 1965; Kelley, 1967).
Attempts to prove the existence of such a motivational bias have generally
involved demonstrations of asymmetry in the attribution of positive and.
negative outcomes -specifically, a tendency for actors to attribute "suc-
cesses" to their own efforts, abilities, or dispositions while attributing
"failure" to luck, task difficulty, or other external factors. Achievement
tasks (e.g., Davis & Davis, 1972; Feather, 1969; Fitch, 1970; Wolosin,
Sherman, & Till, 1973) and teaching performances (e.g., Beckman, 1970;
Freize & Weiner, 1971; Johnson, Feigenbaum, & Weiby, 1964) have
provided most of the evidence for this asymmetry. It has also been shown
that actors may give themselves more credit for success and less blame for
failure than do observers evaluating the same outcomes (Beckman, 1970;
Gross, 1966; Polefka, 1965).
C itics skeptical of broad motivational biases, however, have experi-
enced little difficulty in mounting challenges to such research (see Miller
134 CAUSALITY AND ATTRIBUTION
I
& Ross, 1975, also Nisbett & Ross, 1980, Chap. 10, for detailed discussions).
The most telling argument against research purporting to show motiva-
tional biases is the obvious distinction between subjects' private percep-
tions and their public judgments. One can easily create situations where a
person will publicly deny (or claim) responsibility for an event that he
privately accepts (or does not accept) as his responsibility. While these
public judgments may be self-serving in the sense of preserving one's
public image, they do not imply the operation of ego-defensive biases in
the sense of preserving one's private image (Miller, 1978).
Furthermore, asymmetries in the private attributions (were they avail-
able to researchers) of success and failure, and differences in the judg-
ments of actors and observers may reflect other non-motivational sources
of bias. As several researchers have noted, success, at least in test
situations, is likely to be anticipated and congruent with the actor's past
experience, whereas failure may be unanticipated and unusual. Similarly,
successful outcomes are intended and are the object of plans and actions by
the actor, whereas failures are unintended events that occur in spite of the
actor's plans and efforts. Observers, furthermore, rarely are fully aware of
the past experiences or present expectations and intentions of the actors
whose outcomes they witness.
Challenges to the existence of pervasive ego-defensive biase!> have been
empirical as well as. conceptual. Thus, in some studies subjects seem to
show "counterdefensive," or esteem-attenuating, biases. For example,
Ross, Bierbrauer, and Polly (1974), using an unusually authentic instruc-
tor-learner paradigm, found that instructors rated their own performances
and abilities as more important determinants of failure than of success.
Conversely, the instructors rated their learner's efforts and abilities as
more critical determinants of success than failure. In the same study these
seemingly counterdefensive attributional tendencies proved to be even
more pronounced among professional teachers than among inexpe-
rienced undergraduates, a result that contradicted the obvious derivation
from ego-defensiveness theory that those most directly threatened by the
failure experience would be most defensive.
Researchers who insist that self-serving motivational biases exist can, of
course, provide alternative interpretations of studies that seem to show no
motivational biases or counterdefensive biases (d. Bradley, 1978). Indeed,
in many respects the debate between proponents and skeptics has become
reminiscent of earlier and broader debates in learning theory and basic
perception in which the fruitlessness of the search for a "decisive"
experiment on the issue of motivational influences (i.e., one that could not
be interpreted by the "other side") became ever more apparent as data
multiplied and conceptual analysis sharpened.
One response to this state of affairs has been to abandon motivational
constructs temporarily and to concentrate upon those non-motivational
Shortcomings in the attribution process 135
factors (i.e., informational, perceptual, and cognitive factors) that
influence and potentially distort attributional judgments. Beyond the
existing conceptual difficulties, empirically mixed results, and historical
lessons that discourage the investigators who would search for encompass-
ing motivational biases, there are two additional reasons for the contem-
porary shift. First, there is a growing conviction that a fuller appreciation
of non-motivational influences might lead us to understand and anticipate
those circumstances in which attributions of responsibility are likely to
enhance the attributer's self-esteem and those in which such attributions
are likely to attentuate his self-esteem (d. Miller & Ross, 1975). Second,
there is the increasing recognition that accurate attributions generally are
apt to be more "self-serving" than inaccurate ones -that is, that distor-
tions of causal judgment are apt to leave the organism badly prepared for
the task of long-term survival, however pleasant the immediate conse-
quences of certain inaccurate perceptions and influences.
The rest of this chapter deals with a limited number of such non-
motivational biases (see Nisbett & Ross, 1980, for a more thorough review).
It also discusses a general phenomenon that increases the "costs" of such
biases -the tendency for erroneous impressions, judgments, and even
broader theories to survive in the face of logically powerful data that
contradict these beliefs. Let us recognize from the outset, however, that
the errors and biases dealt with are not inexplicable perversities on the
intuitive scientist's part. Typically, they reflect the operation of mecha-
nisms and strategies that serve the organism reasonably well in many
circumstances; otherwise they surely would not survive the learning
history of the individual or the evolutionary history of the species. These
errors and biases can fairly be regarded as "domain specific" failings of
inferential strategies and tactics that are at least cost efficient (and proba-
bly generally quite accurate as well) in the organism's overall experience.
Non-motivational attribution biases
The fundamental attribution error
The first identified (Heider, 1958) and most frequently cited non-motiva-
tion bias, one that we shall term the fundamental attribution error, is the
tendency for attributers to underestimate the impact of situational factors
and to overestimate the role of dispositional factors in controlling behav-
ior. As "intuitive" psychologists, we seem too often to be nativists, or
proponents of individual differences, and too seldom S -R behaviorists.
We too readily infer broad personal dispositions and expect consistency in
behavior or outcomes across widely disparate situations and contexts. We
jump to hasty conclusions upon witnessing the behavior of our peers,
overlooking the impact of relevant environmental forces and constraints.
136 CAUSALITY AND ATTRIBUTION
General e'l'idence for the fundamental attribution error, Beyond anecdotes and
appeals to experience, the evidence most frequently cited for this general
bias (e,g" Jones & Nisbett, 1971; Kelley, 1971) involves the attributer's
apparent willingness to draw "correspondent" personal inferences about
actors who have responded to very obvious situational pressures, For
instance, Jones and Harris (1967) found that listeners assumed some
correspondence between communicators' pro-Castro remarks and their
private opinions even when these listeners knew that the communicators
were obeying the experimenter's explicit request under no-choice condi-
tions,
A more direct type of evidence that observers may ignore or underesti-
mate situational forces has been provided by Bierbrauer (1973), who
studied subjects' impressions of the forces operating in the classic Milgram
(1963) situation, In Bierbrauer's study, participants witnessed a verbatim
reenactment of one subject's "obedience" to the point of delivering the
maximum shock to the supposed victim. Regardless of the type and
amount of delay before judging, regardless of whether they actually
played the role of a subject in the reenactment or merely observed,
Bierbrauer's participants showed the fundamental attribution error; that
is, they consistently and dramatically underestimated the degree to which
subjects in general would yield to those situational forces that compelled
obedience in Milgram's situation (see Figure 1). In other words, they
assumed that the particular subject's obedience reflected his distinguish-
ing personal dispositions rather than the potency of situational pressures
and constraints acting upon all subjects.
The special case of role-conferred advantages in self-presentation. The tendency
of social observers to underestimate the potency of situational forces and
constraints and to overestimate the role of individual dispositions has
figured heavily in the strategy, conceptual analyses, and even in the
professional debates of contemporary social psychology (see Nisbett &
Ross, 1980; Ross, 1977). Certain special cases of this fundamental attribu-
tion error help to focus our attention on mediating processes and more
specific failures of the intuitive psychologist. An experiment by Ross,
Amabile, and Steinmetz (1977), dealing with evaluations made about
actors who were role-advantaged or role-disadvantaged (by random
assignment), is a case in point. The particular roles dealt with by Ross et a1.
were those of questioner and contestant in a general-knowledge quiz
game. The questioner's role obliged the subject to compose a set of
challenging general-knowledge questions, to pose these questions to the
contestant, and to provide accurate feedback after each response by the
contestant. The contestant's role was restricted to answering, or attempt-
ing to answer, the relevant questions. Both of these participants (and, in a
subsequent reenactment, observers as well) were then required to rate the
questioner's and the contestant's general knowledge.
Shortcomings in the attribution process 137
tOOl I I
l0 NO DELAY PREDICTION
90 . DISTRACTED DELAY PREDICTION
4 UNDISTRACTED DElAY PREDICTION ....
a I- .OBTAINED RATES BY
8MilGRAM (19631
w
~70
w
0
UJ
:!:! 60
'"
0
IL 50
0
w
'"
;;! 40
z
w
~30
w
Q. /-
~t2.5"1.
.
,/300V
20
aa75 135 195 265
SHOCK LEVEL 315 375 450
VOLTS:
Figure l. Comparison of predicted and actual disobedience rates.
The arbitrary assignment and fulfillment of these roles, it should be
apparent, forced the participants and observers to deal with blatantly
non-representative or biased "samples" of the questioners' and contes-
tants' knowledge. The questioners' role encouraged them to display
esoteric knowledge and guaranteed that they would avoid areas of igno-
rance; the contestants were denied such advantages in self-presentation.
Indeed, there was virtually no ambiguity about the arbitrariness of the
role assignment or about the differing prerogatives associated with each
role, unlike many real-world situations in which social roles similarly
confer advantages and disadvantages in self-display. Nevertheless, the
unequal contest between questioners and contestants led to consistently
biased and erroneous impressions. The participants, in a sense, simply
failed to make adequate allowance for the situationally conferred advan-
tages and disadvantages of the relevant roles. Thus, contestants rated their
questioners as far superior to themselves, and uninvolved observers
dearly agreed (see Figure 2). The observers, armed with the knowledge
that they could no more answer the esoteric questions posed than could
the contestants, recognized that the contestants were not deficient in their
general knowledge. What the observers concluded, instead, was that the
questioners were truly outstanding in their general knowledge. Interest-
ingly, the questioners themselves were not misled by their encounter. An
appreciation of this fact shifts our focus from the general existence of the
fundamental attribution error, and the specific impact of social roles, to
138 CAUSALITY AND ATTRIBUTION
tOO
.I. 1
-iii QUESTIONER ,...
80
I""
13CONTEST";'T
§10
.....
~
0
Z
.. IiO
.....
«
II:
~SO
0
::>
0
z40
i=
«
II:
~30
w
:I;
0
Figure 2. Ratings of questioners' and contestants' general knowledge.
the particular "data samples" upon which the various participants relied
for their inferences. Unlike the contestants and the observers, the ques-
tioners were in no case forced to rely exclusively upon "biased" samples of
general knowledge. Presumably, they had a great deal of additional
evidence about the extent and limitations of their own general knowledge
and about the unrepresentativeness of the esoteric items of information
.they displayed in their questions; consequently, they rated both them-
selves and the contestants as "average."
Both the social and theoretical implications of the Ross, Amabile, and
Steinmetz demonstration should be clear. It prompts us to consider the
countless social contexts in which formal or informal roles constrain
interpersonal encounters and, in so doing, bias the impressions of the
participants -even to the point of seeming to justify the prerogatives and
limitations that are imposed by the advantaged and disadvantaged roles. It
also prompts us to sharpen our focus on one of the specific failings of the
intuitive scientist -his seeming insensitivity to the limited inferential
value of biased data samples (see also Hamill, Wilson, &: Nisbett, 1980;
Nisbett &:Ross, 1980, Chap. 4).
Salience ur availability biases
Perhaps the most energetically researched area of attributional bias has
been that involving the effects of attention and of the perceptual and
cognitive factors that mediate attention. Briefly stated, it appears that
whenever some aspect of the environment is made disproportionately
salient or "available" to the perceiver (d. Tversky &:Kahneman, 1973,11)
Shortcomings in the attribution process 139
that aspect is given more weight in causal attribution. Th~s, when an actor
is made visually salient because of a unique racial or sexual status within a
larger group (Taylor et aL, 1976), because of some striking feature of
appearance or dress (McArthur &:Post, 1977; McArthur &:Soloman, 1978),
because of an instructional set (Regan &:Totten, 1975), or even because of
seating arrangements or other determinants of visual perspective (e.g.,
Storms, 1973; Taylor &:Fiske, 1975), that actor is assigned disproportionate
responsibility for any outcome to which she' or he contributes. (See Taylor
&:Fiske, 1978, for a more complete review.) Indeed, a number of studies
derived from "objective self-awareness" theory (Duval &:Wicklund, 1972;
Wicklund, 1975) have shown that actors' perceptions of their own causal
roles can similarly be influenced by simple manipulations that direct their
attention toward or away from the self as a social object (e.g., Duval &:
Hensley, 1976; Ellis &:Holmes, 1979).
Recognition and understanding of how salience or availability factors
affect the attributional process may help us better to understand the bases
of many familiar attributional and inferential biases, perhaps even
subsuming them as special cases. The fundamental attribution error, for
instance, may importantly reflect the fact that actors are simply more
salient than environmental features and therefore are more likely to be
noticed in the attributer's initial search for causal candidates. Indeed,
when situational factors and constraints are made disproportionately
salient to the attributer, we might expect attributional errors that seem to
be opposite to the so-called fundamental error. Thus, a supervisor can be
led to incorrectly attribute a worker's trustworthy performance to an
external factor -that is, the supervisor's surveillance -when that external
factor is made highly salient (see Strickland, 1958). By the same token, an
actor's intrinsic interest in a given task can be undermined (e.g., Oed,
1971; Lepper&: Greene, 1975, 1978; Lepper, Greene, &:Nisbett, 1973) if that
actor is led to focus attention on an external incentive or constraint that
seemingly encourages, but is in fact not necessary to encourage, perfor-
mance in that task.
Consider also Jones and Nisbett's (1971) empirical generalization that
actors, in accounting for their behavior, are relatively more inclined to cite
situational factors and less inclined to dte dispositional factors than are
observers of such behavior. To the extent that actors and observers show
corresponding differences in their focus of attention -that is, actors attend
to relevant features of their environment while observers focus their
attention on the actors themselves -the Jones and Nisbett generalization
becomes a special case of the attention/attribution generalization. Indeed,
experimental evidence suggests that by manipulating actors' and observ-
ers' focus of attention, or perspective, their tendencies to cite situational
versus dispositional causes can likewise be manipulated (e.g., Storms,
1973; Taylor &:Fiske, 1975). .
Finally, let us consider an inferential shortcoming noted by that astute
140 CAUSALITY AND ATTRIBUTION
fictional detective/psychologist Sherlock Holmes -the tendency to over~
look the inferential value of non~occurrences. Such informative non-
occurrences are events or'actions that have not occurred in some context,
which by not occurring thereby contain potentially important information
(d. Ross, 1977). The special relevance of the relatively low cognitive
availability of non-occurrences should be clear. When one searches for or
considers causal candidates, non-occurrences are unlikely to be highly
salient or appropriately attended to; accordingly, one is not apt to give
them sufficient weight in accounting for observed actions and outcomes.
Jill is more likely to attribute Jack's anger to something she has "done"
than to something she has failed to do, simply because the former is apt to
be more salient to her than the latter. Indeed, assuming that the sins of
omission are apt to be less salient than the sins of commission, Jackis apt
to make the same error in accounting for his own anger.
The false consensus or egocmtric attribution bias
The final non~motivational, or "informational," bias to be considered in
this chapter relates to people's estimates of social consensus -the
perceived commonness or oddity of the various responses they witness.
Unlike the professional psychologist, who relies upon well-defined
sampling techniques and statistical procedures for making such estimates,
the layperson must rely upon intuitions and subjective impressions based
on limited access to relevant data. The possibilities for bias in such
estimates, and in the various social inferences or attributions that reflect
such estimates, are thus legion. The specific attributional bias that we shall
consider here concerns people's tendency to perceive a "false consensus" -
that is, to see their own behavioral choices and judgments as relatively
common and appropriate to existing circumstances while viewing alterna-
tive responses as uncommon, deviant, and inappropriate.
Refere~ces to "egocentric attribution" (Heider, 1958; Jones & Nisbett,
1971), to "attributive projection" (Holmes, 1968), and to specific findings
and phenomena related to false consensus biases have appeared sporadi-
cally in the social perception and attribution literatures (d. Katz & Allport,
1931; Kelley & Stahelski, 1970). Perhaps the most compelling evidence,
however, has been provided in a series of studies by Ross, Greene, and
House (1977).
In the first study reported, subjects read descriptions of hypothetical
conflict situations of the sort they might personally face and were to (a)
estimate the commonness of the two possible response alternatives; (b)
indicate the alternative they, personally, would follow; (c) assess the traits
of the "typical" individual who would follow each of the two specified
alternatives.
The estimates and ratings demonstrated the "false consensus" effect;
subjects estimated that the alternative they chose would be relatively more
I
Shortcomings in the attrib\.Jtion process 141
common than the unchosen alternative. An obvious corollary to the false
consensus proposition is that the intuitive psychologist judges those
responses that differ from his own to be more revealing of the actor's
stable dispositions than those responses that are similar to his own. The
Ross, Greene, and House (1977) data support this prediction; subjects made
relatively more confident and extreme predictions about the typical
person who would perform the subject's unchosen alternative than about
the typical person who would perform the subject's chosen alternative.
The term relative is critical in this formulation of the false consensus bias
and it requires some clarification. Obviously, the man who would walk a
tightrope between two skyscrapers, launch a revolution, or choose a life of
clerical celibacy recognizes that his choices would be shared by few of his
peers. The false consensus bias would, however, lead him to see his
personal choices as less deviant than they would seem to us who would
not walk tightropes, launch revolutions, or become celibate clerics. Simi-
larly, the present thesis concedes that for some response categories
virtually all raters' estimates may be biased in the same direction. The
incidence of infant abuse, for instance, might be underestimated by
abusing and non-abusing parents alike. The relative terms of the false
consensus hypothesis lead only to the prediction that abusing parents will
estimate child abuse to be more common and less revealing of personal
dispositions than will non-abusing parents.
In a final demonstration by Ross, Greene, and House (1977) the hypo-
thetical questionnaire methodology was abandoned and subjects were
confronted with a real and consequential conflict situation. Subjects were
asked to walk around campus for 30 minutes wearing a large sandwich-
board sign bearing the message "EATATJOE'S." The experimenter made it
clear to subjects that they could easily refuse to participate in the
sandwich-board study but that he would prefer them to participate and
thereby "learn something interesting while helping the research project."
Subjects were subsequently asked to make their own decision about taking
part in the study, to estimate the probable decisions of others, and to make
trait inferences about particular peers who agreed or refused to partici-
pate.
The results using this "rea}" conflict situation (Table 1) confirmed the
findings of earlier questionnaire studies dealing with hypothetical
responses. Overall, subjects who agreed to wear the sandwich-board sign
estimated that 62% of their peers would make the same choice. Subjects
who refused to wear the sign estimated that only 33% of their peers would
comply with the experimenter's request. Furthermore, as predicted,
compliant and non-compliant subjects disagreed sharply in the relative
strength of inferences they were willing to make about one peer who
agreed and one who refused to wear the sandwich board. Compliant
subjects made more confident and more extreme inferences about the
personal characteristics of the non-compliant peer; non-compliant subjects
made stronger infer~nces ahol1t thO'>r"m"I;~~. ~.A~-
jj
,~
~
s:
...
~
~
~
0
'S
o::s
..:::
....
..t:>
0
~
~
~
c:
~
...
I/)
~
s:
~
:}
... 0
~...
.-!
::l::lC:
::I bO
<I) \11""
... .. <I)
gO:;;
.1
..c:1I\1I
~ < ~ ~
v
c:
~
~
.S
....
'S ... 0
v....
'0 E~
..c: ::I ~~
<I) bO""
bO ... <II II>
c: ::10"
~ ~..c: ~
ii) <~~
-
';;
...
-
.....
0
<I)
II>
\11~
"I:~
\111:---
""0....
ee:x
:::: e.E
&J13~
1:
..
'"
I/)
....
s:
s:
0
E
E
0
'-'
'0'
~
-
tI
.s
1;;
....
-~
....
~
a::
.....
o~
"'~
II>---
\11...
"C:c:
\11 c: \11
-:a 0e
e e \11
:=e~
&J13~
ti
....
~
'"
::s
."
t;
."
s:
8
3!
~
...
~
....i
\11
:c
~
Cf)
.0
N
.....
...
0
N
.....
ex>
Cf)
N
>0
0.2
"'1O3:r:I
t~::ICf)
toli; I
fII Ii:" Ii:
0--0---
..c:C:..c:c:
~.~~,~
<I) II) !l II)
..
~:;; v:;;
\1I\1IQ/
:O~E~
::I ::I
CI'J CI'J
Shortcomings in the attribution process 143
co
-0
0
...
Some broad implications of the Ross, Greene, and House (1977) demon-
strations for our conception of the intuitive psychologist should be clear.
Lay estimates of deviance and normalcy, and the host of social inferences
and interpersonal responses that accompany such estimates, are systemati-
cally and egocentrically biased in accord with the layperson's own behav-
ioral choices. More generally, it is apparent that attributional analyses may
be distortE'd not only by errors in the intuitive psychologist's eventual
analysis of social data but also by earlier biases in sampling or estimating
such data,
Several non-motivational factors appear to playa role in producing false
consensus phenomena. Principal among these are (a) selective-exposure
and availability factors, and (b) factors pertaining to the resolution of
situational ambiguity.
Selective-exposure factors underlying false consensus are fairly straight-
forward. Obviously, we know and associate with people who share our
background, experiences, interests, values, and outlook. Such people do, in
disproportionate numbers, respond as we would in a wide variety of
circumstances. Indeed, our close association is determined, in part, by
feelings of general consensus, and we may be inclined to avoid those
whom we believe unlikely to share our judgments and responses. This
exposure to a biased sample of people and behavior does not demand that
we err in our estimates concerning the relevant populations, but it does
make such errors likely. More subtle and more cognitive in character are
the factors that increase our ability to recall, visualize, or imagine paradig-
matic instances of behavior. In a given situation the specific behaviors we
have chosen or would choose are likely to be more readily retrievable from
memory and more easily imagined than opposite behaviors. In Kahneman
and Tversky's (1973; 4) terms, the behavioral choices we favor may be
more cognitively "available," and we are apt to be misled by this ease or
difficulty of access in estimating the likelihood of relevant behavioral
options.
A second non-motivational source of the false consensus effect arises
from the intuitive psychologist's response to ambiguity -both about the
nature and magnitude of situational forces and about the meaning and
implications of various response alternatives. Attempts to resolve such
ambiguity involve interpretation, estimation, and guesswork, all of which
can exert a parallel effect on the attributor's own behavior choices and
upon his predictions and inferences about the choices of others. Thus,
subjects who anticipated and feared the ridicule of peers for wearing the
"EAT AT JOE'S" sign and who regarded the experimenter's wishes and
expectations as trivial were likely to refuse to wear the sign, to assume
similar refusals by their peers, and to draw strong inferences about the
traits of any subject who chose to wear the sign. Opposite priorities, of
course, would have produced opposite personal choices and opposite
social estimates and inferences.
t-..
~
Cf)
....
..:
Q/
E
>,
.0
'"
\11
V
c:
~
~
.S
Q/
e
t::
oW
)(
<1/
<1/
..
0
e
"
c:
fII
...
I:
Q/
1:1
\d
I:
8
<1/
..
0
e
t-..
>0
Cf)
Cf)
~ .....:.
-:a~
vO'>
...
,,---
.S 3:
.. ::I
Q/ 0
X.o
§~
c::ii
.. c:
<1/<1/
..c: \11
bO""
;a~
,,<I)
!l '"
0
~Di:
-::6
g.g
':"
~.£
III
\::\::
bO'"
.S e
oW 6
e ::I
CI'J
0"
S ~
::I
;s
fI\~
144 CAUSALITY AND ATTRlBUTION
In summary, the false consensus bias both reflects and creates distor-
tions in the attribution process. It results from non-random sampling and
retrieval of evidence and from idiosyncratic resolution of ambiguous
situational factors and forces. In turn, it biases judgments about deviance
and deviates, and, more generally, promotes variance and error in the
interpretation of social phenomena.
Belief perseverance in the face of empirical challenges
The intuitive psychologist's various shortcomings -those described in this
chapter and elsewhere (see Nisbett & Ross, 1980) -can lead him to hold
beliefs about himself, about other people, or even about the nature of the
social world, that are premature and in many cases erroneous. As long as
they remain private and are not acted upon, such beliefs may seem incon-
sequential - merely tentative in nature and adjustable to new input. A
gradually increasing body of theory and research, however, can now be
marshaled to suggest the contrary.
It appears that beliefs -from relatively narrow personal impressions to
broader social theories -are remarkably resilient in the face of empirical
challenges that seem logically devastating. Two paradigms illustrate this
resilience. The first involves the capacity of belief to survive and even be
strengthened by new data, which, from a normative standpoint, should
lead to the moderation of such beliefs. The second involves the survival of
beliefs after their original evidential bases have been negated.
Belief perseverance and polarization in the face of new data
Individuals, social factions, interest groups, and even nations often hold
differing beliefs about pressing social or political issues. Such divergences
in opinion are hardly surprising. Given the informal and often purely
intuitive basis on which such opinions are formulated, and given the role
that social communications (often highly biased ones) play in shaping our
beliefs, honest disagreements are inevitable. But what happens when the
holders of divergent viewpoints are allowed to examine relevant evidence
-especially when that evidence is relatively formal in nature and is
identical for all concerned parties?
An optimistic expectation is that the contending factions would narrow
the gap between their beliefs. This narrowing might consist of change
toward the position justified by the relevant evidence, if such evidence
were consistent and compelling; alternatively, it might consist of change
toward greater moderation or mutual tolerance, if the relevant evidence
were mixed or inconclusive. A less optimistic expectation is that the
contending factions would remain unmoved; that is, they would disregard
the new evidence and hold fast to their original positions. A recent
Shortcomings in the attribution process 145
experiment by Lord, Lepper, and Ross (1979) suggests an even more
disheartening result (disheartening, at least, for those who hope or expect
the objective data of the social scientist to dampen the fires of social
dispute).
Lord et a1. (1979) first selected subjects who either supported capital
punishment and believed it to be an effective deterrent (proponents) or
opposed capital punishment and believed it not to be a deterrent (oppo-
nents). The subjects were presented, in a counterbalanced design, with
two purportedly authentic empirical studies. One seemingly provided
empirical support for their position; the other seemingly opposed that
position. At strategic points in the reading of these two studies, the two
groups completed ratings dealing both with their evaluations of the two
studies and with their own changes in attitudes and beliefs. These ratings
dramatically revealed the capacity of theory-holders to interpret new
evidence in a manner that strengthens and sustains their theories. First,
both proponents and opponents of capital punishment consistently rated
the study that supported their beliefs as "more convincing" and "better
conducted" than the study that opposed those beliefs. Second, and in
contrast to any normative strategy imaginable for incorporating new
evidence relevant to one's beliefs, the net effect of reading the two studies
was to polarize further the beliefs of the death penalty opponents and
proponents. The manner in which this polarization occurred was particu-
larly illuminating (see Figure 3). Upon reading a brief statement of a result
that supported their own viewpoint, subjects' beliefs became considerably
more extreme; these ch.mges were maintained or enhanced when the
subjects considered details about the procedure and data. By contrast, upon
reading a brief result statement that opposed their own viewpoint,
subjects became only slightly less extreme; and upon reading the relevant
details concerning procedures and data the subjects tended to revert to the
beliefs th~y had held before ever learning of the study's existence. In fact,
many individual subjects who had read both the results summary and the
procedural details of a study that opposed their belief ultimately became
more convinced of the correctness of that belief! No such effects occurred
when the same results and procedures were read by subjects whose initial
views were supported.
Obviously, professional scientists frequently are guilty of the same
offense as intuitive ones. Again and again one sees contending factions
that are involved in scholarly disputes -whether they involve the origins
of the universe, the line of hominid ascent, or the existence of ego-
defensive attribution biases -draw support for their divergent views from
the same corpus of findings. Later in this chapter we shall consider the
processes underlying such phenomena in more detail and comment more
specifically on the normative status of the scientist's willingness to process
evidence in the light of his existing theories and expectations. First it is
necessary to consider a second general class of perseverance phenomena.
80 CIIAIUIE -UU11
I
MllIIE_OTD IPRO- DETERRENCE ANTI-DETERRENCE I
RESUl TI DETAILS RESULTS DETAILS
""Tal __IT III I
"RE .. fAVOR Of
I
IIIIrl
CAPITal_IT +1
+,
80 _5E .
-I
,-I
ANTI-DETERRENCE rRO-DETERRENCE
..IIE_ITO IRESUlTS DETAILS RESULTS DETAilS
CAPITal "-IT I I II
Z 3 4
TIME
146 CAUSALITY AND ATTRIBUTION
lili" 'AVOII If
CAPITal- /f~ ==~TS I+1
+1
+1.
Figure 3. Top panel: Attitude changes on capital punishment relative to start of
experiment as reported across time by subjects who received pro-deterrence study
first. Bottom panel: Attitude changes on capital punishment relative to start of
experiment as reported across time by subjects who received anti-deterrence study
first.
Belief perseverance after evidential discrediting
Sometimes, one's beliefs are threatened not by new data but rather by
challenges to tile "formative" evidence for such beliefs -that is, to the
information or analysis that led one to form the belief in the first place. At
an anecdotal level it is easy to cite such instances. Sally fails miserably in
her first attempts to learn how to skate and then finds out that her
borrowed skates were much too large to give her the needed ankle
support. Mary assumes that John's bouquet of flowers reflects thoughtful-
ness, a romantic nature, and a certain conventionalism, only to learn
subsequently that John's father owns a flower shop. Broader theories or
beliefs about the world similarly can be challenged. A baby-sitter decides,
on the basis of his experience with a single infant who cried all night, that
bottle feeding produces colicky babies, only to discover that the infant in
question was suffering from a high fever. Or a scientist discovers that a
classic experiment that figured in the emergence of a particular theory was
tainted with some severe artifact or by outright fraud.
The perseverance hypothesis, in its general formulation, suggests that
the social perceivers identified above would persist in their initial assess-
Shortcomings in the attribution process 147
ments to an unwarranted and inappropriate degree. But the terms of such
a contention are obviously too general and too vague to be testable. When,
precisely, can we infer that a social perceiver is "inappropriately" persist-
ing in an impression or belief whose basis has been undermined? To
explore the perseverance hypothesis experimentally, what dearly is
required is a paradigm that permits us to specify precisely how much
perseverance and how much change might be warranted.
One such paradigm is suggested by the dilemma of the social psycholo-
gist who has made use of deception in the course of an experiment and
then seeks to debrief the subjects who had been the target of such
deception. The psychologist reveals the totally contrived and inauthentic
nature of the information presented presuming that this debriefing will
thereby eliminate any effects such information might have exerted upon
the subjects' feelings or beliefs. Many professionals, however, have
expressed public concern that such experimental deceptions may do great
harm that is not fully undone by conventional debriefing procedures (e.g.,
Kelman, 1972; A.G. Miller, 1972; Orne, 1972; Silverman, 1965).
A series of experiments by Lepper, Ross, and their colleagues (see also
earlier studies by Walster et aI., 1967; Valins, 1974) have used the total
discounting, or debriefing, paradigm to explore the phenomenon of belief
perseverance in the face of evidential discrediting. We may begin discuss-
ing this work by outlining a pair of experiments by Ross, Lepper, and
Hubbard (1975) that dealt with subjects' postdebriefing impressions about
their own abilities at a particular task or about the abilities of a peer.
Post debriefing perseverance of personal impressions. The procedure employed
by Ross et ai. (1975) was quite straightforward. Subjects first received
continuous false feedback as they performed a novel discrimination task
(i.e., distinguishing authentic suicide notes from fictitious ones). In the
first experiment reported this procedure was used to manipulate the
subjects' perceptions of their own performance and ability. A second
experiment introduced observers, who formed social impressions as they
witnessed the false feedback manipulation. In both experiments after this
manipulation of first impressions had been completed, the experimenter
totally discredited the "evidence" upon which the actors' and/or observ-
ers' impressions had been based. Specifically, the actor (overheard in
Experiment 2 by the observer) received a standard debriefing session in
which he learned that his putative outcome had been predetermined and
that his feedback had been totally unrelated to actual performance. Before
dependent variable measures were introduced, in fact, every subject was
led to explicitly acknowledge his understanding of the nature and purpose
of the experimental deception.
Following this total discrediting of the original information, the
subjects completed a dependent variable questionnaire dealing with the
actors' performances and abilities. The evidence for postdebriefing
Observer's perceptions
Actor's own perceptions of actors
Success Failure tSuccess Failure
Estimated initial
number correct 18.33 12.83 5-910" 19.00 12.42 4.43"'"
Predicted future
number correct 18.33 14.25 4.23"0 19.08 14.50 2.68"
Rated ability
at task 5.00 3.83 2.65" 5.33 4.00 3.36"
148 CAUSALITY AND ATTRIBUTION
Table 2. Pas/debriefing perceptions of the actor's performance and ability
"1' < 5.5. "1' < 1.1. ""1' < 01.1.
Source: Summarized from Experiment 2 of Ross, Lepper, & Hubbard (1975).
impression perseverance was unmistakable for actors and observers alike.
On virtually every measure (i.e., objective estimates of the actor's just-
completed performance, estimates for performance on a future set of
discrimination problems, and subjective estimates of the actor's abilities)
the totally discredited initial outcome manipulation produced significant
"residual" effects upon actors' and observers' assessments (see Table 2).
Follow-up experiments have since shown that a variety of unfounded
personal impressions, once induced by experimental procedures, can
survive a variety of total discrediting procedures. For example, Jennings,
lepper, and Ross (1980) have demonstrated that subjects' impressions of
their ability at interpersonal persuasion (having them succeed or fail to
convince a confederate to donate blood) can persist after they have learned
that the initial outcome was totally inauthentic. Similarly, in two related
experiments Lepper, Ross, and Lau (1979) have shown that students'
erroneous impressions of their "logical problem solving abilities" (and
their academic choices in a follow-up measure two months later) perse-
vered even after they had learned that good or poor teaching procedures
provided a totally sufficient explanation for the successes or failures that
were the basis for such impressions.
Postdebriefi1lg persel'erance of discredited theories. A recent series of experi-
ments by Anderson, Lepper, and Ross (1980) have extended the domain of
perseverance demonstrations from personal impressions to broader beliefs
about the world. Anderson et al.'s studies first manipulated and then
attempted to undermine subjects' theories about the functional relation-
ship between two measured variables: the adequacy of firefighters' profes-
sional performances and their prior scores on a paper and pencil test of
risk preference. In one particularly pertinent variation, the formative
evidence consisteJ of only one pair of specific cases - i.e., one successful
Shortcomings in the attribution process 149
and one unsuccessful firefighter with appropriately discrepant scores in
their respective tests of risk-taking preferences. Interestingly, such mini-
mal data did suffice to produce strong theories, on the subjects' part, about
the probable relationship between the relevant measures. More important,
however, was the finding that such theories survived the revelations that
the cases in question had been totally fictitious and the different subjects
had, in fact, received opposite pairings of riskiness scores and job
outcomes. Indeed, when comparisons were made between subjects who
had been debriefed and those who had not been, it appeared that over 50%
of the initial effect of the "case history" information remained after
debriefing.
In summary, it is clear that beliefs can survive potent logical or
empirical challenges. They can survive and even be bolstered by evidence
that most uncommitted observers would agree logically demands some
weakening of such beliefs. They can even survive the total destruction of
their original evidential bases. While much work remains to be done in
specifying the precise limits and exploring inevitable exceptions to such
phenomena, it is clear that the costs of the layperson's attributional biases
and other inferential shortcomings are apt not to be corrected but instead
to be compounded by subsequent experience and deliberations. The
question that must at last be addressed, therefore, is how and lI.-'llydoes such
perseverance occur? That is, what cognitive mechanisms underlie the
unwarranted persistence of our impressions, beliefs, and broader social
theories?
Possible mechanisms underlying belief perseverance
Biased searcll, recollection, and assimilation of information. There can be little
doubt that our beliefs influence the processes by which we seek out, store,
and interpret relevant information. Indeed, without prior knowledge and
corresponding preconceptions, our understanding of everyday experience
would demand considerably more time and effort, and in all likelihood
that understanding would be greatly diminished. But an inevitable conse-
quence of our willingness to process evidence in the light of our prior
beliefs is the tendency to perceive more support for those beliefs than
actually exists in the evidence at hand.
Such "confirmation biases" (see Einhorn &: Hogarth, 1978; Hamilton,
1979; Hastie &:Kumar, 1979; Wason &:Johnson-Laird, 1972) have long been
noted by philosophers of science (e.g., Bacon, 1620/1960). Perhaps most
noteworthy is the theory holder's response to equivocal or ambiguous
data. As Lord et a!. (1979) have documented, potentially confirmatory
evidence is apt to be taken at face value while potentially disconfirmatory
evidence is subjected to highly critical and skeptical scrutiny. Thus, two
ISO CAUSALITY AND ATTRIBUTJON
consequences follow: First, any pattern of evidence processed in this
fashion, even evidence th~t is essentially random, will tend to bolster the
initial belief. Second, once evidence has been processed in this fashion it
gains the capacity to sustain.the prior belief when that belief is subjected to
new empirical disconfirmation or to attacks on its original evidential
basis.
The role of biased assimilation has been shown fairly convincingly, we
think, for the case where the theory holder is confronted with new data
(i.e., Lord et al., 1979). But the role of this mechanism in the discounting or
debriefing paradigm is perhaps less obvious, and we are forced to rely on
speculation rather than hard data. We suggest that the subject who forms
an initial.impression about himself, about another person, or about some
functional relationship is apt to search his memory and the immediate
situation for additional data relevant to that impression. Such data, then,
are apt to be recalled and regarded as pertinent or probative only to the
extent that they confirm the impression at hand. Thus a subject who has
succeeded or failed at a given task recalls similar successes or failures at
related tasks -and decides upon their relevance to the present case -on
the basis of the congruency of the relevant outcomes. Similarly, a subject
who has come to believe that variables X and Yare functionally related
will recall, and give credence to, cases that confirm rather than challenge
that presumed relationship. Once again, such biased searching, recollec-
tion, and assimilation not only bolster one's initial belief, they also
produce a pattern of biased evidence that remains highly available to
sustain the belief in question when its initial basis is attacked or even
destroyed. The critical assumption here is that people do not constantly
update or reevaluate the evidence relevant to their beliefs. They do not
commonly decide "now that my prior hypothesis has been undermined
somewhat I must go back and reassess all of the evidence that I ever
considered in the light of that hypothesis."
Tile formation of causal explanations. People do more than merely note
evidence relevant to their impressions or beliefs. They also engage in
causal analysis or explanation (Heider, 1958). That is, they try to account for
the characteristics of self or others, or for the functional relationships that
they have come to believe exist. Thus, the subject who believes herself a
superior or inferior discriminator of suicide notes in the Ross et al. (1975)
study might search for some aspect of her background that would account
for such a talent or deficiency. Similarly, the subject who is induced to
believe in a positive or negative relationship between firefighting ability
and risk preference will have little difficulty in postulating a logical basis
for either relationship. Once again, this process not only buttresses an
initial impression or belief, it is apt to sustain that impression or belief in
the face of subsequent challenges or attacks. ..
Evidence for the operation of this perseverance mechanism comes
Shortcomings in the attribution process 151
primarily from two debriefing studies demonstrating that when subjects
are explicitly required to formulate such explanations, prior to debriefing,
the magnitude of the perseverance effect is increased. In the Anderson et
al. (1980) study one group of subjects was explicitly instructed to explain
the positive or negative relationship suggested by the two firefighter
cases. As pre~icted, this manipulation greatly enhanced the 'relevant
perseverance effect. In fact, subjects who explained the basis for a positive
or for a negative relationship before being debriefed were only trivially
less certain of that relationship than subjects who received no debriefing.
Similar results were obtained by Ross, Lepper, Strack, and Steinmetz
(1977), who found that subjects induced to explain outcomes in the lives of
clinical patients (whose earlier case histories they had read) continued to
regard such outcomes as relatively likely even when they learned that the
explained events were inauthentic and had been contrived by the experi-
menter.
Behavioral confirmation or "self-fulfilling" hypotheses. The two research para-
digms used by Ross, Lepper, and their colleagues to investigate persever-
ance phenomena lack one element that may be critical to many everyday
situations. Specifically, subjects in those studies lacked the opportunity to
act upon their beliefs. Such actions are important partially because they
can increase the psychological costs or "dissonance" (Festinger, 1957)
involved in changing one's beliefs (d. Ashmore & Collins, 1968; Collins &
Hoyt, 1972; Hovland, Campbell, & Brock, 1957). Furthermore, such actions
create new data relevant to those beliefs. Not only may this new data be
pr~essed in a biased manner, but the data themselves may also 1?ebiased
in a direction that tends to confirm the relevant hypothesis.
The idea of self-confirming, or self-fulfilling, hypotheses is not a new
one to social scientists. The famous but controversial "Pygmalion" studies
by Rosenthal and Jacobson (1968), which dealt with the impact of teachers'
expectations upon the "blooming" of their students' abilities and perfor-
mances, is a case in point. However, a recent series of studies by Snyder
and his colleagues have considerably advanced our appreciation and
understanding of such phenomena by demonstrating the manner in
which subjects' expectations, or the hypotheses they are led to test, can
generate "objective support" for those expectations or hypotheses (e.g.,
Snyder & Swann, 1978a, 1978b; Snyder, Tanke, & Berscheid, 1977).
Concluding remarks: Beliefs do change!
Our foregoing discussion of phenomena and mechanisms should not
make the reader lose sight of the fact that beliefs about ourselves, our
political leaders, and even our scientific theories do change. In part such
change may simply be the result of brute force. Even if logical or empirical
challenges have less impact than might be warranted by normative
152 CAUSALITY AND ATTRIBUTION
standards (see Ross & Lepper, 1980) they may still get the job done. In part,
such change may reflect the fact that formal methods of hypothesis testing
sometimes are deliberately employed to protect us from the dangers of
informal ones. But we suspect there is more to the story, for there is
evidence that prior theories can sometimes be overcome without massive
amounts of disconfirming evidence or decisive well-controlled experi-
ments. Thus, the changes in outlook and belief that can be wrought by
vivid, concrete, first-hand experience (see Nisbett & Ross, 1980) and the
effectiveness of groups and leaders that accomplish dramatic political or
religious conversions offer inviting targets for future research.
'C'
I..
REFERENCES
Anderson, C. A., Lepper, M. R., and Ross, L. Theory perseverance: The
persistant influence of discredited information.
and Social Psycholo~y, in press, 1980.
Journal of Personalitv
Ashmore, R. D., and Collins, B. E. Studies in forced compliance X: Attitude
change and commitment to maintain publicly a counter-attitudinal position.
Psychological Reports, 1968, 22, 1229-1234.
Bacon, F. The new organon and related writings. New York: Liberal Arts
Press, 1960. (Originally published, 1620.)
Beckman, L. Effects of students' performance on teachers' and observers'
attributions of causality. Journal of Educational Psychology, 1970,
ih, 75-82.
Bierbrauer, G. Effect of set, perspective, and temporal factors in attribution.
Unpublished doctoral dissertation, Stanford University, 1973.
Bradley, G. W. Self-serving biases in the attribution process: A reexamination
of the fact or fiction question. Journal of Personality and Social
Psychology, 1978, ~, 56-71.
Collins, B. E., and Hoyt, M. F. Personal responsibility-for-consequences:
An integration and extension of the forced-compliance literature.
Journal of Experimental Social Psychology, 1972, i, 558-593.
Davis, W. L., and Davis, D. E. ~terna1-external control and attribution of
responsibility for success and failure. Journal of personalitz, 1972,
~, 123-136.
Deci, E. Effects of externally mediated rewards on intrinsic motivation.
Journal of Personality and$~~,-c:j.al~PsychQlon, 1971, ,g, lOS-US.
Duval, S., and Hensley, V. Extensions of objective self-awareness theory:
The focus of attention-causal attribution hypothesis. In J. H. Harvey,
W. J. Ickes, and R. F. Kidd (Eds.), New directions in attribution research,
Vol. 1, 1976.
....
Duval, S., and Wicklund, R. A. A theory of objective self awareness. New York:
Academic Press, 1972.
Einhorn, a. J., and Hogarth, R. M. Confidence in judgment: Persistence in the
illusion of validity. Psycholo~ical Review, 1978,~, 395-416.
Ellis, R. J., and HolJnes, J. G. Focus of attention in social interaction:
The impact on self evaluations. Unpublished manuscript, University of
Waterloo, 1979.
Feather, N. T. Attribution of responsibilitY. and valence of success and
failure in relation to initial confidence and task performance. Journal
of Personality and Social Psycholo~, 1969,11,129-144.
.~ ..--.-
Festinger, L. A Theo!I., of~Co~itive Dissonance. Stanford, CA: Stanford
t
University Press, 1957.
Fitch, G. Effects of self-esteem, perceived performance, and chance on causal
attributions. Journal of Personality and Social Psvcholo~, 1970, ti,
311-315.
Freize, I., and Weiner, B. Cue utilization and attributional judgments for
success and failure. Journal of Personalitv, 1971,11, 591-606.
Gross, A. Evaluation of the t:arget person in a social influence situation.
Unpublished doctoral dissertation, Stanford University, 1966.
Hamill, R., Wilson, T. D., and Nisbett, R. E. Ignoring sample bias: Inferences
about collectives from atypical cases. Unpublished manuscript, Univ. of
Mic:higan, 1980.
Hamilton, D. L. A cognitive attribut:ional analysis of stereotyping. In L.
Berkowit:z (Ed.), Advances in exPerimental social osvcholo~y, Vol. 12.
New York: Academic Press, 1979.
aastie, R., & Kumar, P. A. Person mcll1ory: Personality t:raits as organizing
principles in memory for behavior. Journal of Personality and Social
PsycholoiZ. 1979. lI. 25-38.
Heider, F. Social perception and phenomenal causality. Psychological Review,
1944, 51, 358-373.
Heider, F. The psychology of interpersonal relations. Wiley, 1958.
Holmes, D. S. Dimensions of projection. Psycho1o~ical Bulletin, 1968, ~,
248-268.
Hovland, C. I., Campbell, E. H., and Brock, T. The effects of "commitment"
on opinion change following communication. In C. I. Hovland et al.,
The order of presentation in persuation. New Haven: Yale University
Press, 1957.
Jennings, D. L., Lepper, M. R., and Ross, L. Persistence of impressions of
personal persuasiveness: Perseverance of erroneous self assessments out-
side the debriefing paradigm. Unpublished manuscript, Stanford Univ., 1980.
Johnson, T. J., Feigenbaum, R., and Weiby, M. Some determinants and conse-
quences of the teacher's perception of causation. Journal of E~eri-
mental Psycholo~, 1964,21, 237-246.
Jones, E. E., and Davis, K. E. From acts to dispositions: The attribution
process in perception. In L. Berkowitz (Ed.), A4vances in Ex"erimental
Social Psychology, vol. 2, Academic Press, 1965.
Jones, E. E., Davis, K. E., and Gergen, K. Role playing variations and 'their
informational value for person perception. Journal of Abnormal and
Social Psychology, 1961, ~, 302-310.
Jones, E. E., and DeCharms, R. Changes in social perception as a function of
the personal relevance of behavior. Sociomet.;:z., 1957, 12., 75-85.
Jones, E. E., and Harris, V. A. The attribution of attitudes. Journal of
Experimental Social Psvcholo~, 1967,1,1-24.
Jones, E. E., Kanouse, D. E., Kelley, H. H., Nisbett, R. E., Valins, S., &
Weiner, B. (Eds.). Attribution: PerceivinLth~__causes of behavior.
Morristown, N. J.: General Learning Press, 1971.
Jones, E. E., and Nisbett, R. E. !he actor and the observer: Divergent
perceptions of the causes of behavior. In E. E. Jones et al. (Eds.),
Attribution: Percei~ the causes of behavior. Morristown, N.J.:
General Learning Press, 1971.
Kahneman, D., and 'I'versky, A. On the psychology of prediction. Psychological
Review, 1973, 80,237-251.
-
Katz, D., and Allport, F. Student's attitudes. Syracuse: Craftsman Press,
1931.
Kelley, H. H. Attribution theory in social psychology. In D. Levine (Ed.),
Nebraska S~osium on Motivation. vol. 15. Lincoln: University of Nebraska
Press, 1967.
Kelley, H. H. Attribution in social interaction. In E. E. Jones et al.
(Eds..), ~ttribution: Perceiving the causes of behavior. New York:
General Learning Press, 1971.
Kelley. H. H. !he process of causal attribution. American Psychologist.
1973, ~, 107-128.
Kelley, H. H., and Stahelski, A. J. Social interaction basis of cooperators'
and competitors' beliefs about others. Journal of Personalitv and
So~ialPsycholo~, 1970, ~, 66-91.
Kelman, Ei. C. !he human use of human subjects: !he probl~ of deception
in social psychological experiments. In A. G. Miller (Eds.). !he SQcial
~sycholo~ of ~svcholo~ical research. New York: Macmillan, 1972.
Lepper, M. R., and Greene. D. Turning play into work: Effects of adult
surveillance and extrinsic rewards on children's intrinsic motivation.
Journal of ?ersonality and Social~!3vchol°.s.I.. 1975. 11. 479-486.
Lepper. M. R.. and Greene. D. (Eds.). The hidden cost~ of reward. ~.Lorristown .
NJ: Lawrence Erlbaum Associates. 1978.
Lepper, M. R., Greene, D., and Nisbett, R. E. Undermining children's intrinsic
interest with extrinsic reward: A test of the overjustification hypo-
thesis. Journal of Pers~na1ity and Social Psycholo~y, 1973, 1!, 129-137.
Lepper, M. R., Ross, L., and Lau, R. Persistence of inaccurate and discredited
personal impressions: A field demonstration of attributional perseverance.
Unpublished manuscript, Stanford University, 1979.
Lord, C., Lepper, M. R., and Ross, L. Biased assimilation and attitude polar-
ization: !he effects of prior theories on subsequently considered evi-
dence. Journal of_Pe;os()neil!tyand Social Psychology, 1979, 11., 2098-2110.
McArthur, L. Z. !he how and what of why: Some determinants and consequences
of causal attribution. Journal of Personality and Social Psycholo~y,
1972, ~, 171-193.
McArthur, L. Z. !he lesser influence of consensus than distinctiveness
information on causal attributions: A test of the person-thing hypothesis.
Journal of~fe1:'s~)Ua.1.llyand Social Psvcholo~, 1976, 11, 733-742.
McArthur, L. Z., and Post, D. Figural emphasis and person perception. Journal
of E~erimental Social Psycholo~, 1977.
McArthur, L. Z., and Solomon, L. K. Perceptions of an aggressive encounter
as a function of the victim's salience and the perceiver's arousal.
Journal of Person~litv--and Social Psychology, 1978, ~, 1278-1290.
Milgram, S. Behavioral study of obedience. Journal of Abnormal and Social
Psychology, 1963, iL, 371-378.
Miller, A. G. (Ed.), !he social psychology of psychological research. New York:
Macmillan, 1972.
Miller, D. T. What constitutes a self-serving attributional bias? Journal
of Personalitv and Social Psychology, 1978, ~, 1221-1223.
Miller, D. T., and Ross, M. Self-serving biases in the attribution of causality:
Fact or fiction? Psycho1o2ical Bulletin, 1975, ~, 213-225.
Nisbett, R. E., and Ross, L. Ruman inference: Strategies and shortcomings.
Englewood Cliffs, New Jersey: Prentice Hall, Inc., 1980.
Orne, M. T. Demand characteristics and their implications for real life:
The importance of quasi-controls. In A. G. Miller (Ed.), The social
psychology of psycholo~ical research. New York: l'f.acmillan, 1972.
Poleua, J. The perception and evaluation of responses to social influences.
Unpublished doctoral dissertation, Stanford University, 1965.
Regan, D. T., and Totten, J. Empathy and attribution: Turnmg 0bservers
into actors. Journal of Persgnal~ty and Social Psvcholo!il;Y, 1975, 11,
850-856.
Rosenthal, R., and Jacobson, L. ~lion in the classroom: Teacher e~ectation
and pu'Pils' intellectual development. New York: Rolt, R:1nehart, &
W:1nston, 1968.
Ross ,L. The intuitive psychologist and his shortcomings. In L. Berkowitz
(Ed.), Advances in Ex-oerimental Social Ps'Vchology, vol. 10, New York:
Academic Press, 1977.
Ross, L. D., Amabile, T. M., and Stembetz, J. L. Social roles, social
control, and biases in social-perception processes. Journal of personalitz
and Social Psychology, 1977, 12" 485-494.
Ross, L., Bierbrauer, G., and Polly, S. Attribution of educational outcomes
by professional and non-professional instructors. Journal of Personalitv
and Social Psychology, 1974, ~, 609-618.
Ross, L., Greene, D., and Rouse, P. The false consensus phenomenon: An
attributional bias in self perception and social perception processes.
Journal of Experimental Social Ps'Vcho1o~, 1977, li, 279-301.
Ross, L., Lepper, M. R.~ and Hubbard, M. Perseverance in self perception
and social perception: Biased attributional processes in the debriefing'
paradigm. Journal of Personalit'V and Social Psvcholo~7, 1975, ]1, 880-892.
Ross, L., Lepper, M. R., Strack, F., and Steinmetz, J. L. Social explanation
and social expectation: Th~ effects of real and hypothetical explanations
upon subjective likelihood. Journal of~P~rso~ality and Social psvch~lo~v,
1977,11,817-829.
Silverman, I. Motives underlying the behavior of the subject in the psychological
experiment. Paper presented at the meeting of the American Psychological
Association, Chicago, 1965.
Snyder, M., and Swann, W. B. Hypothesis-testing processes in social interaction
Journal of Personality and Social Psvcholo~v, 1978a, ]i, 1202-12i2.
Snyder, M., and Swann, W. B. Behavioral confirmation in social intera«tion:
From social perception t~ social reality. Journal of E~erimental Social
Psvcho1os;z., 1978b, 14, 148-162.
Snyder, M., Tanke, E. D., and Berscheid, E. Social perception and interpersonal
behavior: On the self-fulfilling nature of social stereotypes. Journal
of Perspnalitv and S~cial Psvchology, 1977, 35, 656-666.
Storms, H. D. Videotape and the attribution process: Reversing actors'
and observers' point of view. Journal of personalitz and Social
Psychology, 1973, 1L, 165-175.
Strickland, L. H. Surveillance and trust. Journal of personalitz., 1958,
l2., 200-215.
Taylor, S. E., and Fiske, S. T. Point of view and perceptions of causality.
Journal of Personality and Social Psycholo~y, 1975, 32, 429-445.
Taylor, S. E., and Fiske, S. T. Salience, attention and attribution: Top
of the head phenomena. In L. Berkowitz (Ed.), Advances in Experimental
Psychology, vol. 11. New York: Academic Press, 1978.
Taylor, S. E., Fiske, S. T., Close, M., Anderson, C., and Ruderman, A. Solo
status as a psychological variable: The power of being distinctiv.e.
Unpublished manuscript, Harvard University, 1979.
Thibaut, J. W., and Riecken, H. W~ Some determinants and consequences of the
perception of social causality. Journal ofPersonalitz., 1955, 24,
113-133.
Tversky, A., and Kahneman, D. Availability~ A heuristic for judging frequency
and probability. Cognitive Psychology, 1973,1, 207-232.
Valins , S. Persistent effects of information about internal reactions:
Ineffectiveness of debriefing. In H. London & R. E. Nisbett (Eds.),
Thou~ht and feelin~: Co~itive modification of feeling states.
Chicago: Aldine, 1974.
Walster, E., Berscheid, E., Abrahams, D., and Aronson, E. .
Effectiveness of
debriefing following deception experim~~ts. Journal of Personal~it!. and
Social Psycho~~!Z, 1967, i, 371-380.
Wason, P. C., and Johnson-Laird, P. N. Psvcholo~ of reasonin&: Structure
and content. London: Batsford, 1965.
':-1'einer, B. Achievement motivation and attribution theory. ~'!orristown, :LJ.:
General Learning Press, 1974.
Wicklund, R. A. Objective self-awareness. In L. Berkowitz (Ed.), Advances
in eXDerimental social psycholo~v, Vol. 9. New York: Academic Press, 1975.
~Jolosin, R. J., Sherman, S. J., and Till, A. Effects of cooperation and
competition on responsibility attribution after success and failure.
Journal of Experimental Social Psvcholo~, 1973, i, 220-235.