People come to know themselves in large part by observing others’ reactions to them when engaged in face-to-face social interaction. The popularity of and reliance on social media (e.g., Facebook) for social connection interferes with the self-knowledge process. This interference is compounded, further, by the tendency for people to pose as who they wish to be but are not (Wright, White, & Obst, 2018). Others’ reactions to them as a source of self-knowledge is based, then, on an idealized presentation, and social feedback on this idealized presentation can have little self-knowledge value.  

Reference:

Wright, E. J., White, K. M., & Obst, P. L. (2018). Facebook false self-presentation behaviors and negative mental health. Cyberpsychology, Behavior, and Social Networking, 21(1), 40–50.

Assignment:

Submit 3–5 pages, excluding title page and reference page:

• What implications does the pressure of presenting an ideal self on social media have in terms of how people use introspection, self-observation, and other people’s reactions to know themselves?
• If relying on social media for social connection fails to provide useful information for developing an authentic self, then how confident might a person be when meeting new people face-to-face?
• What information about how others perceive you might you need but do not get when relying on social media for social connection?
• Your arguments and conclusions must be supported by social psychology theory and research.

 In addition to the Learning Resources, search the Walden Library and/or Internet for peer-reviewed articles to support your Assignment. Use proper APA format and citations, including those in the Learning Resources.  

Aronson, E., Wilson, T. D., Akert, R. M., & Sommers, S. R. (Eds.). (2019). Social psychology (10th ed.). Boston, MA: Pearson.

• Chapter 5, “The Self: Understanding Ourselves in a Social Context” 

Journal of Personality and Social Psychology
1997, Vol. 73, No. 1,

31

-44

Copyright 1997 by the American Psychological Association, Inc.
0022-35i4/97/$3.00

Prejudice as Self-image Maintenance:
Affirming the Self Through Derogating Others

Steven Rsin
Williams College

Steven J. Spencer
Hope College

The authors argue that self-image maintenance processes play an important role in stereotyping and
prejudice. Three studies demonstrated that when individuals evaluated a member of a stereotyped
group, they were less likely to evaluate that person negatively if their self-images had been bolstered
through a self-affirmation procedure, and they were more likely to evaluate that person stereotypically
if their self-images had been threatened by negative feedback.’ Moreover, among those individuals
whose self-image had been threatened, derogating a stereotyped target mediated an increase in their
self-esteem. The authors suggest that stereotyping and prejudice may be a common means to maintain
one’s self-image, and they discuss the role of self-image-maintenance processes in the context of
motivational, sociocultural, and cognitive approaches to stereotyping and prejudice.

A most striking testament to the social nature of the human
psyche is the extent to which the self-concept—that which is
the very essence of one’s individuality—is integrally linked
with interpersonal dynamics. Since the earliest days of the for-
mal discipline of psychology, the significant influences of a
number of social factors on the self-concept have been recog-
nized. A central focus of sociocultural and social-cognitive
approaches to psychology has concerned the ways in which
individuals’ self-concepts are defined and refined by the people
around them. This is evident in early discussions of the social
nature of individuals’ self-concepts (Cooley, 1902; Mead, 1934)
and of social comparison theory (Festinger, 1954), and it contin-
ues to be evident in more recent work, such as that concerning
self-fulfilling prophecies (e.g., Eccles, Jacobs, & Harold, 1990;
Rosenthal & Jacobson, 1968; Snyder, 1984) and cultural influ-
ences (Abrams, 1994; Cameron & Lalonde, 1994; Cohen &
Nisbett, 1994; H. R. Markus & Kitayama, 1991, 1994; Triandis,
1989; Turner, Oakes, Haslam, & McGarty, 1994).

The converse focus—the self-concept’s influence on percep-
tions of and reactions toward others—has been recognized more
fully within the last two decades, through, for example, research
on self-schemas (H. Markus, 1977; H. Markus & Wurf, 1987),
self-verification (Swann, Stein-Seroussi, & Giesler, 1992), self-
discrepancies (Higgins, 1996; Higgins & Tykocinski, 1992),
and a host of self-serving biases in individuals’ perceptions,
judgments, and memories involving the self (e.g., Ditto & Lo-
pez, 1992; Greenwald, 1980; Klein & Kunda, 1992, 1993; Nis-

Steven Fein, Department of Psychology, Williams College; Steven J.
Spencer, Department of Psychology, Hope College. Steven J. Spencer is
now at Department of Psychology, University of Waterloo, Waterloo,
Ontario, Canada.

Correspondence concerning this article should be addressed to Steven
Fein, Department of Psychology, Bronfman Science Center, Williams
College, Williamstown, Massachusetts 01267, or to Steven J. Spencer,
Department of Psychology, University of Waterloo, Waterloo, Ontari

o

N2L 3G1 Canada. Electronic mail may be sent via the Internet to
steven.fein@williams.edu.

bett & Ross, 1980; Ross & Sicoly, 1979; Schlenker, Weigold, &
Hallam, 1990).

Particularly within the past decade, research has converged on
the role of self-image- and self-esteem-maintenance process

es

in people’s perceptions and reactions regarding others. These
approaches, whose roots can be seen in the earlier work of
James, Festinger, Heider, Sherif, Tajfel, and others, include re-
search on downward social comparison (Brown, Collins, &
Schmidt, 1988; Brown & Gallagher, 1992; Gibbons & Gerrard,
1991; Gibbons & McCoy, 1991; Taylor & Lobel, 1989; Wills,
1981, 1991; Wood & Taylor, 1991), self-evaluation maintenance
(Tesser, 1988; Tesser & Cornell, 1991), social identity (Ab-
rams & Hogg, 1988; Brewer, 1993; Crocker, Thompson,
McGraw, & Ingerman, 1987; Hogg & Abrams, 1988; Smith,
1993; Turner, 1982), terror management (Greenberg et al.,
1992), and self-affirmation (Liu & Steele, 1986; Steele, 1988;
S t e e l e & L i u , 1983).

This article examines the role of self-image-maintenance pro-
cesses in a particular set of reactions and perceptions: those
concerning prejudice and negative evaluations of others. More
specifically, we examine the thesis that many manifestations of
prejudice stem, in part, from the motivation to maintain a feeling
of self-worth and self-integrity. That is, self-image threat may
lead people to engage in prejudiced evaluations of others. These
negative evaluations can, and often do, make people feel better
about themselves. Prejudice, therefore, can be self-affirming.
By using available stereotypes to justify and act on prejudices,
individuals may be able to reclaim for themselves a feeling of
mastery and self-worth, often saving themselves from having to
confront the real sources of self-image threat.

Several self-image-maintenance processes are described or
implied in the existing literature, but the research reported in
this article focuses on one in particular: self-affirmation. Steele
and his colleagues (e.g., Steele, Spencer, & Lynch, 1993) have
argued that people seek to maintain “an image of self-integrity,
that is, overall moral and adaptive adequacy” (p. 885). If an
individual experiences a threat to this image, he or she attempts
to restore this image by reevaluating and reinterpreting experi-

31

32

FEIN AND SPENCER

ences and events in ways that reaffirm the self’s integrity and
value. Supported by research on self-affirmation effects in cog-
nitive dissonance, Steele et al. (1993) argued that when facing
a potential threat, even an important one, people have “the
option of leaving the threat unrationalized—that is, accepting
the threat without countering it or its implications—and af-
firming some other important aspect of the self that reinforces
one’s overall self-adequacy” (p. 885).

We argue that prejudice often serves a self-affirming function
for individuals, and providing people with other means of self-
affirmation should reduce their desire to make prejudiced evalu-
ations. The link between self-image threats and the use of preju-
dice should be weakened by providing people with the opportu-
nity to self-affirm, that is, by providing them with information
that restores their positive sense of self-integrity. This approach
is distinct from many of the classic approaches to stereotyping
and prejudice, such as frustration-aggression theory and scape-
goating (Dollard, Doob, Miller, Mowrer, & Sears, 1939;
Miller & Bugelski, 1948), social identity theory (Tajfel, 1982),
and downward social comparison theory (Wills, 1981). We
argue that this process of self-affirmation should reduce the
desire to make prejudiced evaluations even though it does not
release pent-up anger or aggression, as frustration-aggression
theory would require; enhance social identity, as social identity
theory would require; make self-other comparisons, as down-
ward social comparison theory would require; or confront the
threat itself in any way. Only a self-affirmational perspective
suggests that restoring a positive sense of self-integrity in this
way would result in the decrease of prejudiced evaluations. Of
course, this thesis shares many assumptions with these other
theoretical positions. Our approach, however, can be seen as
extending previous approaches by examining self-image mainte-
nance as both cause and effect of prejudiced evaluations and by
integrating these approaches with contemporary views of the
self.

Taken together, the studies reported in this article examined
both sides of this process: the roles of self-affirmation and self-
image threat in influencing the likelihood that individuals will
use stereotypes or prejudice and the role of prejudice in helping
individuals restore a positive sense of self.

Study 1

In Study 1, we examined the hypothesis that self-affirmation
should make participants less likely to evaluate another individ-
ual in ways that reflect their prejudice toward the individual’s
group. Participants in this study were asked to evaluate a target
person who apparently was either a member of a group for
which there was a readily available negative stereotype or a
member of some other outgroup for which there was not a strong
available stereotype. Before being exposed to this target person,
participants were either self-affirmed or not affirmed. That is,
half of the participants completed a task designed to affirm and
make salient an important aspect of their self-concepts, and the
other half completed a task designed not to affirm any important
aspects of their self-concepts.

We believe that many stereotypes and prejudices are such
readily available and cognitively justifiable means of self-en-
hancement that individuals often use their stereotypes and preju-

dices to self-enhance in the face of everyday vulnerabilities and
frustrations (e.g., see Wood & Taylor, 1991). That is, unless
other motives are activated, such as a goal of accurate perception
(Darley, Fleming, Hilton, & Swann, 1988; Neuberg & Fiske,
1987), accountability (Tetlock, 1983), or social desirability or
egalitarian motives (Dovidio & Gaertner, 1991; Monteith,
1993), people may find stereotyping and prejudice to be a reli-
able and effective way to protect their self-esteem in a frequently
threatening world. To the extent, then, that the use of stereotypes
and prejudice stems in part from self-image maintenance needs,
self-affirmation should make individuals less likely to resort to
this use. Study 1 was designed to test this hypothesis.

Method

Participants

Seventy-two introductory psychology students from the University of
Michigan participated in this experiment as partial fulfillment of a course
requirement.1

Procedure

The participants were told that they would participate in two experi-
ments in this session. The first experiment was portrayed as a study of
values. The second experiment was portrayed as an investigation of how
employees evaluate candidates in the hiring process.

Manipulation of self-affirmation. Half of the participants completed
a self-affirmation procedure, and half did not. This procedure was a
modified version of that used by Steele and Liu (1983; see also Steele,
1988; Tesser & Cornell, 1991) to affirm and make salient an important
part of individuals’ self-concepts. Participants were given a list of several
values (adapted from values characterized by the Allport-Vemon Study
of Values), including business/economics, art/music/theater, social
life/relationships, and science/pursuit of knowledge. Participants in the
self-affirmation condition were asked to circle the value that was most
important to them personally and then to write a few paragraphs ex-
plaining why this value was important to them.2 In contrast, participants
in the no-affirmation condition were asked to circle the value that was
least important to them personally and then to write a paragraph ex-
plaining why this value might be important to someone else. Steele and
his colleagues (e.g., Spencer & Steele. 1990; Steele, 1988) have found
that causing participants to think about a value that is personally very
important to them is an effective means of producing self-affirmation
and that, in the absence of self-image threat, it does not affect partici-
pants’ state self-esteem.

Evaluation task. For what we portrayed as the second experiment,
participants were placed in individual cubicles and were told that their
task was to evaluate an individual who had applied for a job as a
personnel manager at a particular organization. The participants were
given general information about the responsibilities of a personnel man-
ager at this hypothetical organization and were encouraged to try to
make an accurate assessment of the candidate’s suitability for the job.

All participants next examined information about a fictitious job can-

1 Although 72 people participated in the experiment, 18 were excluded
because they were Jewish, for reasons that are described in the Manipu-
lation of target’s ethnicity section. Thus, the data from 54 participants
were included in all analyses.

2 None of our participants wrote paragraphs concerning prejudice or
tolerance. Moreover, the effects of the manipulation were not related to
which value—business /economics, art/music/theater, social life/rela-
tionships, or science/pursuit of knowledge—the participants chose.

PREJUDICE AS SELF-IMAGE MAINTENANCE

didate who was about to graduate from their university. Participants
were given the candidate’s completed job application to examine. The
application contained questions about the candidate’s previous work
experience, academic and extracurricular skills and interests, and other
resume-type information. The completed application was constructed to
suggest that the applicant was fairly well qualified for the position but
was not necessarily a stellar candidate. Attached to the application was
a photograph of the candidate. All of the participants saw virtually the
same application and photograph; the variations are noted in the section
below. After examining this material, participants watched an 8-min
videotape presented as excerpts from the candidate’s job interview. All
participants saw the same videotape, which featured a fairly neutral
performance by the candidate—that is, her responses tended to be ade-
quate but not extremely positive or negative. After watching the excerpts,
participants completed a questionnaire about the candidate and her
qualifications.

Manipulation of target’s ethnicity. Although all participants saw the
same job interview excerpts, saw the same woman in the photograph
attached to the job application, and read the same information about her
work experiences, academic record, and other job-relevant information,
we included two minor variations in the photograph and three in the
application to suggest either that the candidate was Jewish or that she
was not Jewish (and probably was Italian).

We used this distinction for several reasons. At the time and place in
which this study was conducted, there was a very well known and
relatively freely discussed stereotype concerning the “Jewish American
princess” (JAP). There was a fairly sizable and salient minority of
students at this campus who were Jewish women from New “\brk City
and Long Island, New \brk, and these women were the targets of a
number of JAP jokes that spread across campus. In contrast to stereo-
types about African Americans, gay men and lesbians, and many other
groups, the JAP stereotype was one that many students were willing to
discuss quite candidly, with many of them openly endorsing it.3

Another factor that played a role in our decision to examine this form
of prejudice was that we were able to select a stimulus person who
could be considered representative of the Jewish American princess and
yet, with a few subtle manipulations, could just as easily be considered
representative of a non-Jewish group—one that also was an outgroup
to most participants but about which there was no strong negative stereo-
type or prejudice on this campus. This alternative categorization was of
an Italian American woman. Although also a minority on campus, this
group was not nearly as salient on campus, and as pilot testing confirmed,
there was no strong, consensual stereotype or prejudice on campus con-
cerning this group.

To manipulate the target’s ethnic background, we varied the following
elements of her application: her name (Julie Goldberg vs. Maria D’Agos-
tino), an extracurricular activity (volunteering for a Jewish or Catholi

c

organization), and her sorority (either of two sororities that shared
similar reputations in terms of status, but one of which consisted pre-
dominantly of Jewish women and one of which consisted predominantly
of non-Jewish women of European, but not Hispanic, descent). All the
other information on the application, including all of the job-relevant
information, was identical.

In both conditions, the photograph attached to the job application was
of the same woman (who was also featured in the videotape). We had
chosen a female undergraduate, unknown to the participants, who could
be seen either as fitting the prototypic image of a Jewish American
princess or as non-Jewish (and probably Italian). The photograph varied
slightly, however, so that ‘ ‘Julie” was wearing a necklace featuring the
Star of David and had her hair clipped up in back (in a clip that some
pilot test students referred to as a JAP clip), whereas “Maria” was
wearing a cross and had her hair down. Pilot testing suggested that our
manipulation was successful.

This woman appeared in the video wearing a sweater that covered

her necklace, and her hair was down but brushed in such a way that its
length seemed somewhere in between the styles depicted in the two
photographs. As indicated above, all participants saw the same 8-min
video.

Dependent measures. Participants rated the candidate in terms of
her overall personality and her qualifications for the job. Her personality
was assessed by the extent to which participants agreed (on a 7-point
scale) that each of the following traits described her: intelligent, insensi-
tive, trustworthy, arrogant, sincere, inconsiderate, friendly, self-centered,
down-to-earth, rude, creative, materialistic, motivated, cliquish, ambi-
tious, conceited, happy, vain, warm, superficial. Negative traits were
reverse scored. Her job qualifications were assessed by the extent to
which participants agreed (on a 7-point scale) with the following state-
ments; ” I feel this person would make an excellent candidate for the
position in question,” ” I would likely give this person serious consider-
ation for the position in question,” ” I would guess that this person is
in the top 20% of people interviewed,” and ” I felt favorably toward
this person.” Both scales showed good internal reliability (Cronbach’s
alphas of .93 and .91, respectively). Finally, participants were asked to
indicate their own and the target’s ethnicity and religion.

Results

Recall that our prediction was that when participants were
not self-affirmed, they would evaluate the target more negatively
when she was portrayed as Jewish than when she was portrayed
as Italian, whereas when participants were self-affirmed, this
difference would be reduced or eliminated.

The critical measure in this study was participants’ ratings
of the target’s personality across a variety of dimensions. These
ratings were subjected to a two-way analysis of variance (AN-
OVA). The ANOVA revealed that there was no significant main
effect for the manipulation of affirmation, F( 1, 50) = 1.8, p >
.15, but that there was a significant main effect for the manipula-
tion of the apparent ethnicity of the target, as the target was
rated more positively when she appeared to be Italian than when
she appeared to be Jewish, F ( l , 50) = 4.9, p < .05. Most importantly, this main effect was qualified by a significant inter- action, F( 1, 50) = 8.5, p < .01. As can be seen in Figure 1, and consistent with our predictions, not affirmed participants who evaluated the Jewish target were significantly more negative in their evaluations of the target's personality than were partici- pants in all other conditions, r(50) = 3.7, p < .001. None of the other conditions differed significantly from each other.4

1 One of the reasons for this may be that the stereotype is diffused
across two types of prejudice: anti-Semitism and sexism. That is, those
who endorse the stereotype are protected against being considered anti-
Semitic because they are not implicating Jewish men in their derogatory
comments or beliefs, and they are protected against being considered
sexist because they are not implicating most women. A second reason
may be that the targeted group is perceived as being relatively privileged,
and thus, disparaging them may not seem as harmful.

4 We used this planned comparison for both dependent measures in
this study, as well as a comparable planned comparison for each of the
dependent measures in Studies 2 and 3, because it was the most direct
test of our theoretically derived hypotheses (see, e.g.. Hays, 1981; Kep-
pel, 1973; Rosenthal & Rosnow, 1991; Winer, 1971). For each of these
measures, we also conducted the more conservative Newman-Keuls
post hoc comparisons. In each case, the Newman-Keuls comparisons
indicated the difference tested in the planned comparison to be signifi-
cant and revealed further that none of the other conditions differed
significantly from each other.

34 FEIN AND SPENCER

1001

9 0 ”

80″

Jewish Candidate

Italian Candidate

es

o
/

ta

• 5

U

•
Of

Self-Affirmed Not Self-Affirmed

Condition

Figure 1. Rating of candidate’s personality as a function of self-affir-
mation condition and ethnicity of the candidate. Higher numbers indicate
more favorable evaluations.

Ratings of how qualified the target was for the job in question
also were consistent with our predictions. A two-way ANOVA
revealed that participants who had not been affirmed tended to
rate the target more negatively than did participants who had
been affirmed, F ( l , 50) = 4.6, p < .05. The ANOVA revealed further that participants rated the candidate more positively when she was depicted as Italian than when she was depicted as Jewish, F ( l , 50) = 6.3, p < .05. These main effects were qualified, however, by a marginally significant interaction be- tween the two independent variables, F ( l , 50) = 3.0, p < .10. Consistent with our predictions, not affirmed participants evaluated the qualifications of the candidate more negatively when she was portrayed as Jewish (M = 14.9) rather than Italian {M — 20.6), whereas affirmed participants did not make this discrimination (Ms = 20.2 and 21.2, respectively). The planned comparison indicated that not affirmed participants who evalu- ated the Jewish target were significantly more negative in their evaluations of the target's job qualifications than were partici- pants in all other conditions, f(50) = 3.7, p < .001. None of the other conditions differed significantly from each other.

Discussion

The results of this study demonstrate that self-affirmation can
reduce the likelihood that individuals will derogate members of
stereotyped groups. In the absence of self-affirmation, partici-
pants’ evaluations of the job candidate were biased as a function
of her apparent ethnicity. That is, these participants evaluated
the target more negatively if she was a member of a stereotyped
group than if she was not. Among participants who had been
self-affirmed, however, this difference was eliminated.

These results, therefore, highlight the significant role played
by the self-concept in prejudice. More specifically, they support
the idea that thinking about a self-relevant value, even one com-
pletely unrelated to prejudice, can reduce the expression of
prejudice. Thinking about a self-relevant value has this effect
even though it need not release pent-up anger or aggression,
enhance social identity, or involve self-other comparisons, as
frustration-aggression theory, social identity theory, and down-
ward social-comparison theory would require.

In a replication of this study, we also examined the potential
mediating role of participants’ mood in this paradigm with an
independent sample of 71 participants. We measured partici-
pants’ mood using the Mehrabian and Russell (1974) mood
scale after the manipulation of self-affirmation but before the
participants evaluated the target. The mood scale consists of
three subscales, each consisting of six sets of bipolar adjectives.
These subscales measure pleasure (e.g., happy unhappy,
pleased-annoyed), arousal (e.g., stimulated-relaxed, excited-
calm), and dominance (e.g., controlled-controlling, influen-
tial-influenced). Consistent with the findings of Liu and Steele
(1986), the manipulation of self-affirmation had no significant
effects on any one or any combination of these subscales (all
Fs < 1). Moreover, participants' mood was unrelated to their evaluation of the target's personality, r ( 6 9 ) = —.120, ns, or of her qualifications for the job, r(69) = -04, ns. Replicating the results of Study 1, not affirmed participants who evaluated the Jewish target rated the target's personality significantly more negatively than did participants in all other cells, t{61) — 7.A, p < .01. Similarly, not affirmed participants who evaluated the Jewish target tended to rate the target's job qualifications more negatively than did participants in all other cells, /(67) = 1.8, p < .05.

The results of these studies suggest that at least part of the
negative evaluation of people who are stereotyped may result
from people trying to affirm their self-image. To the extent that
people’s self-images have been buffered by other means of self-
affirmation, they should be less drawn to such a strategy. In
the absence of such self-affirmation, however, stereotyping and
prejudice may provide a mechanism by which people protect
or bolster their self-esteem. Stereotyping and prejudice may be
reinforced, therefore, because they can make people feel better
about themselves.

Study 2

The results of Study 1 suggest that self-affirmation can play
an important role in reducing the effects of stereotyping or
prejudice on individuals’ evaluations of a member of a stereo-
typed group. In Study 2, we focused on the other side of this
self-image maintenance coin by examining whether a self-image
threat would exacerbate the effects of stereotyping or prejudice
on individuals’ evaluations of a member of a stereotyped group.

Study 2 differed from Study 1 in two other important ways,
thereby providing a better test of the generalizability of our
hypotheses. First, rather than varying the target’s apparent eth-
nicity, in Study 2 we manipulated the target’s apparent sexual
orientation. Thus, whereas the stereotyped group in Study 1 was
contrasted with a nonstereotyped group that was also a distinct
minority, the stereotyped group in Study 2 was contrasted with

PREJUDICE AS SELF-IMAGE MAINTENANCE 35

the nonstereotyped majority. Second, rather than measuring par-
ticipants’ general derogation of a target as a function of her
membership in a stereotyped group, Study 2 measured partici-
pants’ stereotyping of an individual as a function of his member-
ship in a stereotyped group.

More specifically, some participants in this study received
self-image-threatening information in the form of bogus nega-
tive feedback on an intelligence test; the other participants re-
ceived no such threat. Later, all participants evaluated a target
on a series of trait dimensions relevant to popular stereotypes
of gay men. The biographical information about the target was
manipulated so as to suggest to some of the participants that
the target may have been gay and to suggest to the other partici-
pants that he was straight (heterosexual). The hypothesis tested
in Study 2 was that participants should be more likely to exhibit
stereotyping of the (apparently) gay target if they had previously
received negative feedback on the intelligence test than if they
had not.

Method
Participants

Sixty-one male undergraduates from Williams College participated in
this experiment either for extra credit for their introductory psychology
course or for the chance to win money in a random drawing.

Procedure

Participants reported to the laboratory individually and completed the
tasks in individual rooms containing a desk and a Macintosh computer.
Participants first read a sheet of paper containing the cover story, which
stated that the study involved a series of different cognitive and social
judgment tasks. The first part of the study involved the manipulation of
self-relevant feedback (described below). After some filler tasks (e.g.,
a simple word-stem completion task) designed to preserve the integrity
of the cover story, participants completed the social judgment task, in
which the participants read information about a male target. The infor-
mation was designed to suggest either that the target was gay or that he
was straight. After rating the target on a series of dimensions, the partici-
pants were probed for any suspicions, debriefed thoroughly, and thanked
for their participation.

Manipulation of feedback. Half of the participants were assigned
randomly to the negative feedback condition, and the other half were
assigned to the neutral condition. To the former half, the experimenter
introduced the first set of tasks as ” a new form of intelligence test that
is given on the computer. It measures both verbal and reasoning abili-
ties.'” To the latter half, the experimenter explained that they had been
assigned to a control condition in which they were simply to read the
materials contained in a bogus test of intelligence. The experimenter
revealed to these participants that the participants in the treatment condi-
tion of the study would be told that the test was a real, valid measure of
intelligence. In other words, the experimenter told the neutral condition
participants the truth. These participants were instructed to refrain from
trying hard to answer the questions on the bogus test because many of
the questions had no correct answer and because the time limits were
unrealistically quick. The experimenter also told them that the computer
would present them with bogus scores at the conclusion of the test. To
assure the participants that these scores were indeed bogus, the experi-
menter told them what these scores would be. The experimenter ex-
plained that the participants in the treatment condition would be led to
believe that the scores were real. The purpose of having the participants
in the neutral condition learn this cover story and go through the test

was so that they would be exposed to the same test and specific items
as the participants in the negative feedback condition, but that the test
would have no relevance to their self-image.5

All subsequent instructions for the test were presented on the com-
puter. The instructions were presented in a professional-looking design
that introduced the intelligence test as ‘ ‘The Reasoning and Verbal Acuity
Battery.” The instructions explained that the test had been validated in
numerous studies throughout the United States and Canada. The test
consisted of five parts, each tapping different sets of intellectual skills.
The first four parts consisted of analogies, antonyms, sentence comple-
tions, and syllogisms. The fifth part was called a “verbal-nonverbal
matching test” and involved matching difficult vocabulary words to
various pictures; this was a modified version of the Ammons and Am-
mons (1962) Quick Test of Intelligence. The instructions to this battery
of tests explained that research had shown that this combination of
tasks was the ideal, most valid method to measure individuals’ general
intelligence.

To emphasize the relevance of these intellectual skills, each test within
the battery was introduced with an explanation of what it measured.
Many of the specific items in these tests were taken from advanced tests
used for admission to graduate school or law school. To make the tests
seem even more challenging (and thus to help to justify the bogus
feedback for the participants in the negative feedback condition), we
modified several of the items so that there was no correct answer among
the options given. Moreover, the time limits for each item were very
short (ranging from 10 to 20 s, depending on the test), and a clock
showing the seconds ticking away appeared on the screen for each item.

At the conclusion of this battery of tests, the computer program indi-
cated that it was calculating the scores. After 7 s a new screen appeared
that indicated the participant’s percentile rankings (relative to other col-
lege students tested in the United States and Canada) for each test. Each
participant received an identical set of scores: 51st percentile for the
analogies test, 54th for the antonyms, 56th for the sentence completions,
33rd for the syllogisms, and 38th for the verbal-nonverbal matching.
Given the prestige of the college in which this study was conducted and
the students’ previous scores on tests such as the Scholastic Achievement
Test, these scores are extremely disappointing to the students from this
population. (See Footnote 5.)

Manipulation of target’s apparent sexual orientation. After adminis-
tering a series of brief cognitive tasks designed to enhance the integrity
of the cover story, the experimenter introduced the “social judgment
tasks” by informing the participants that they would read some informa-
tion about an individual and make some judgments about him or her.

All participants read about a target named Greg, a 31-year-old strug-
gling actor living in the East Village in New \brk City. The information
summarized Greg’s ambitions and career struggles and listed some of
the many odd jobs that Greg had taken to pay the rent while he pursued
his dream. The information continued by detailing a recent event in
Greg’s life concerning landing ” a fairly large part in a serious and rather
controversial play directed by a young director.” Participants read that
Greg was excited about the play and, in particular, about working with
this young director. The director’s name was not mentioned, but gender
pronouns indicated that the director was a man. The participants read
that after the first week of rehearsals, Greg approached the director and
asked him whether he wanted to get ” a drink or something” with him

5 Consistent with the intent of the manipulation, pilot testing of 36
other participants from the same population revealed that the state self-
esteem (as measured by Heatherton & Polivy’s [ 1991 ] state self-esteem
scale) of participants in the neutral condition was not significantly lower
than that of participants who were not exposed to the test or cover story
(F < 1). In addition, the state self-esteem of these participants (in either condition) was significantly higher than that of pilot test participants who were led to believe the test was real (Fs > 6 ) .

FEIN AND SPENCER

after that night’s rehearsal so that they could talk about his role in some
more depth. The story continued for a few paragraphs, summarizing the
play’s opening and reviews, find it concluded with the information that
while continuing to act in the play, Greg was writing his own play and
had already gotten a commitment from the director to help him with it.

The information about Greg was identical across conditions with the
following exceptions. In the first sentence, the participants in the straight-
implied condition read that Greg “has been living with his girlfriend,
Anne, in a small apartment” for several years. Anne’s name was men-
tioned three more times in subsequent parts of the story about Greg,
and there was one additional reference to his “girlfriend.” For the gay-
implied condition, in the first sentence we replaced the word “girl-
friend” with “partner” and dropped reference to Anne. Neither the
partner’s name nor the partner’s gender was specified, and there were
no subsequent references to this partner.

Many of the details of the story about Greg (e.g., his living in the
East Village, his caring “for a very close and very ill friend for the last
2 months of his friend’s life,” and his relationship with the director)
were included to support the implication in the gay-implied condition
that Greg was gay. Because each piece of information by itself very
plausibly could describe a straight actor’s life, however, we believed
that the participants who were introduced immediately to references to
Greg’s girlfriend would not entertain the idea that Greg was gay.6

Dependent measures. Participants used an II-point scale ranging
from 0 (not at all) to 10 (extremely) to rate Greg’s personality on each
of 10 dimensions. Three of these (intelligent, funny, and boring) were
included as stereotype-irrelevant fillers. The stereotype-relevant traits
included sensitive, assertive/aggressive, considerate, feminine, strong,
creative, and passive (see Fein, Cross, & Spencer, 1995; Kite & Deaux,
1987). Assertive/aggressive and strong were reverse-coded so that for
each item, higher ratings indicated greater stereotyping. An index of this
set of seven traits showed moderate internal reliability (Cronbach’s a
= .77). It may be worth noting that these traits, when taken out of a
stereotyped context, are not necessarily negative and may indeed be
rather positive. But to the extent that participants perceived these traits
as more descriptive of a target if they thought that the target was gay
than if they thought he was straight, this would indicate stereotyping,
and the valence of these traits would be debatable.

In addition, participants used the same 11-point scale to indicate the
degree to which they would like Greg as a friend and the degree to
which their own personality was similar to Greg’s. These measures, of
course, were less ambiguous in terms of valence: Lower ratings on these
two measures clearly indicated more negative feelings toward the target.

Results

Recall that we predicted that if participants read information
about a target that implied that he was gay, they would be more
likely to evaluate this target consistently with the gay stereotype
if they had received threatening, negative feedback about their
performance on the intelligence test than if they had not received
any threatening feedback. If the information about the target
indicated that he was straight, however, the manipulation of
feedback should not have had a strong effect on participants’
evaluation of the target. The results supported these predictions.

Stereotyping

A two-way ANOV\ on the ratings of the target on the set of
seven stereotype-relevant trait dimensions revealed a significant
main effect for the manipulation of feedback, F(l, 57) = 11.3,
p < .001, indicating that participants who had received negative feedback on the intelligence test rated the target more stereotypi-

7 i

c

‘EL

L.
tl

C/5

Gay Implied

I Straight Implied

Neutral Negative

Feedback

Figure 2. Rating of target on .stereotype-relevant traits as a function
of feedback and implied sexual orientation of the target. Higher numbers
indicate greater stereotyping.

cally (i.e., gave higher ratings on the stereotype-consistent
items) than did participants who had not received any feedback.
In addition, the ANOVA revealed a significant effect for the
manipulation of the target’s apparent sexual orientation, F( I,
57) = 5.3, p < .03, indicating that participants who read infor- mation that implied that the target was gay rated him more stereotypically than if they read information suggesting that he was straight. Most importantly, the ANOVA revealed a signifi- cant interaction, F{\, 57) = 4.4, p < .05. As can be seen in Figure 2, and consistent with our predictions, participants who had received negative feedback and read information implying that the target was gay rated the target much more stereotypically than did participants in all other conditions, z(57) = 4.1, p < .001. None of the other conditions differed significantly from each other (see Footnote 4 ) .

Although the stereotype-irrelevant traits were used as filler
to make the participants less likely to be suspicious of the intent
of our questions, we did conduct an ANOVA on the ratings
concerning those traits. The independent variables did not have

6 An obvious question is why we did not simply state that Greg was
gay. Pilot testing of students from this campus revealed quite strongly
that many of the participants became suspicious of the purpose of the
study if they read that the target was gay. More than half of the partici-
pants told the experimenter that they suspected that the study concerned
their stereotypes about gay men. When we eliminated any explicit refer-
ences to Greg’s sexuality, our pilot test participants did not raise these
suspicions, although most of them did spontaneously entertain the
thought that Greg was gay.

PREJUDICE AS SELF-IMAGE MAINTENANCE 37

any significant effects on participants’ ratings of the target on
any or all of these traits (all Fs < 1).

Liking and Similarity

The measure of stereotyping yielded results consistent with
our predictions. But would self-esteem threat also make partici-
pants less willing to indicate that they would like the target as
a friend or that their own personality was similar to the target’s?
To address this question, we conducted an ANO\A on each of
these measures.

The ANOV& on participants’ ratings of the degree to which
they would like the target as a friend revealed a significant main
effect for the manipulation of feedback, F ( l , 57) = 5.7, p
< .03, indicating that participants who had received negative feedback on the intelligence test rated themselves as less in- clined to like the target (M - 5.81) than did participants in the neutral condition (M = 6.87). The main effect for the manipula- tion of the information about the target's apparent sexual orien- tation did not approach significance (F < 1), but the interaction between the two variables was significant, F{\, 57) = 4.1, p < .05. Participants who had received negative feedback on the intelligence test were significantly less inclined to like the target than were those who had not received the feedback, whether or not the target information suggested he was gay, but the interac- tion reflects the tendency for this difference to be greater in the gay-implied condition (Ms = 5.48 vs. 6.98) than in the straight- implied condition (Ms = 6.11 vs. 6.75).

The ANOV\ on participants’ ratings of how similar their own
personality was to the target’s revealed a significant main effect
for the manipulation of feedback, F ( l , 57) = 5.3, p < .03, reflecting the tendency for participants to rate their personality as less similar to the target's if they had received negative feed- back on the intelligence test (M = 4.16) than if they had received no feedback (M = 5.33). The manipulation of information about the target's sexual orientation did not have a significant effect (F < 1). More important, the independent variables produced a significant interaction, F ( ] , 57) = 4.1, p < .05. Consistent with our predictions, participants were particularly unlikely to rate their personality as similar to the target's if they had re- ceived negative feedback and read information implying that the target was gay (M = 3.94), t(57) = 2.3, p < .03. None of the other conditions differed significantly from each other.

Discussion

Consistent with our predictions, participants showed more
stereotyping in their evaluations of the target if they had pre-
viously received negative feedback about their own performance
on an intelligence test. In addition to resulting in greater stereo-
typing, the negative feedback led participants to psychologically
distance themselves from the target if they had reason to suspect
that he was gay, by rating themselves as less likely to be friends
with or be similar in personality to the target. If the information
about the target suggested he was straight, however, the negative
feedback had less effect on these measures.

These results support the hypothesis that self-esteem threat
can increase individuals’ likelihood of exhibiting stereotyping
or prejudice toward members of stereotyped groups. Using a

different stereotype, a different stereotype comparison condition
(i.e., a majority rather than alternative minority group condi-
tion), and different dependent measures from those used in
Study 1, Study 2 yielded results consistent with the hypothesis
that self-image-maintenance processes can play an important
moderating role in stereotyping or prejudice.

But does stereotyping or prejudice in response to self-image
threat restore an individual’s self-esteem? This question was
addressed in Study 3.

Study 3

Our view suggests that one motivation for stereotype- or prej-
udice-based evaluations is that these sorts of evaluations can
restore a threatened self-image. Study 3 provides the first com-
plete test of this hypothesis by examining both sides of this
process: the role of a threatened self-image in causing partici-
pants to derogate a member of a stereotyped group and the
role of this derogation in restoring participants’ threatened self-
image. Thus, an important goal of Study 3 was to provide the
first evidence that negative evaluation of a stereotyped target in
response to self-image threat mediates increase in self-esteem.

Participants in Study 3 took what they thought was an intelli-
gence test. Unlike in Study 2, all participants in Study 3 were
led to believe that the test was real. They received bogus positive
or negative feedback.7 After the feedback, all participants com-
pleted a questionnaire that measured their state self-esteem. In
an ostensibly unrelated experiment that followed, participants
evaluated a woman portrayed as Jewish or Italian, as in Study
1. Following this evaluation, participants again completed the
state self-esteem questionnaire so that we could monitor changes
in their self-esteem.

We predicted that ( a ) participants who received negative feed-
back would have lower state self-esteem than participants who
received positive feedback, (b) participants who received nega-
tive feedback and evaluated the Jewish target would rate the
target more negatively than would the participants in the other
conditions, ( c ) participants who received negative feedback and
evaluated the Jewish target would exhibit a greater increase in
state self-esteem than would participants in the other conditions,
and (d) this increase in state self-esteem would be mediated by
their evaluations of the target.

Method
Participants

One hundred twenty-six introductory psychology students from the
University of Michigan participated in this experiment for partial ful-
fillment of a course requirement.8

7 We believed that it would be difficult or impossible to provide
performance feedback that would be neutral for most participants, un-
less, as in Study 2, we did not lead the participants to believe that the
test was real. An average score was quite threatening to our participants,
and determining how much above average would be neutral for all
participants seemed impossible.

8 Although 126 participants participated in the experiment, 17 were
excluded because they were Jewish, 7 because they were foreign students
and, consequently, would have been less likely to be familiar with the
stereotype about Jewish American women, 4 because they misidentified
the target’s ethnicity, and 2 because they did not believe the false feed-

38 FEIN AND SPENCER

Procedure

Overview. Participants reported to the laboratory in pairs and were
told that they would be participating in two experiments: an intelligence
test and a social interaction. Participants first were given an intelligence
test and were given bogus feedback about their performance. They next
completed a measure of their state self-esteem (Heatherton & Polivy,
1991) and were asked to indicate their score on the intelligence test,
after which they were thanked for their participation, dismissed, and
sent to the “social evaluation” experiment, where they were met by a
different experimenter. The social evaluation experiment involved the
same procedure as that used in Study 1. That is, participants received
information about a job candidate who was depicted as either Jewish or
Italian. After evaluating this candidate’s personality and job qualifica-
tions by using the same measures as those used in Study I, participants
again completed the Heatherton and Polivy measure of state self-esteem,
after which they were asked to indicate their own and the target’s race
and ethnicity. Finally, they were probed for any suspicions, debriefed
thoroughly, and thanked for their participation.

Manipulation of feedback. When participants arrived for what was
portrayed as the first experiment, they were told that the study was
concerned with a new, improved form of intelligence test. The rationale
and instructions were similar to but briefer than those given to the
participants in Study 2. The intelligence test used in this study consisted
of a longer but less difficult version of one of the tests from the battery
of tests used in Study 2: the verbal-nonverbal matching test in which
participants tried to match difficult vocabulary words to various pictures.
This test was purported to be a very valid test of verbal and nonverbal
skills. The experimenter began by giving each participant a pencil and
a form commonly used for exams featuring multiple-choice questions
that are graded via a computer. The test consisted of three sets of 10
words each.

The test was designed to be difficult and ambiguous enough for stu-
dents to believe either positive or negative performance feedback. Some
of the words were difficult or obscure for the average student (e.g.,
capacious, celerity), and some were easier (forlorn, imminent), but all
had the feel of the kinds of vocabulary items that are included in college
entrance exams, and many were such that participants felt as if they
may have known what they meant but could not be sure. Moreover, the
match between words and pictures often was not obvious, particularly
given the fast pace of the test. Pretests and postexperiment interviews
confirmed that participants tended to be unsure of how they were doing
during the test and to believe the feedback that was given them.

At the completion of the test, the experimenter took the participants*
answers and went into an adjacent room. The door to this room was
left open, and the participants could hear what sounded like a Scantron
machine grading the tests. The experimenter returned each participant’s
answer form to him or her. The experimenter explained that a red mark
appeared next to each incorrect answer, that the first number on the
bottom of the form indicated the number of correct answers, and that
the second number indicated the participant’s pcrcentile ranking relative
to all the other students who had taken the test thus far.

The feedback was, of course, bogus. Half of the participants received
positive false feedback about their test performance (i.e., a high score
that ostensibly put them in the 93rd percentile for the university),
whereas the other half received negative false feedback (i.e., a low score
that ostensibly put them in the 47th percentile). Although the 47th
percentile is close to the median, pretesting had indicated that partici-
pants uniformly found this to be very negative feedback (see also Stein,
1994).

back about their performance on the intelligence test. Thus, the data
from 96 participants were included in all analyses.

Results and Discussion

Recall that we predicted that if participants had received
threatening, negative feedback about their performance on the
intelligence test, they would be more likely to derogate the target
as a function of her apparent ethnicity than if they had received
positive feedback about their performance. We also predicted
that derogating the stereotyped target would help restore threat-
ened participants’ self-esteem. The results were consistent with
these predictions.

Evaluations of the Target

Participants’ ratings of the target’s personality were subjected
to a two-way ANOVA, which revealed strong support for our
predictions. Two significant main effects emerged: Participants
who had received negative feedback about their performance on
the intelligence test rated the target’s personality more nega-
tively than did participants who had received positive feedback,
F ( l , 92) — 9.1, p < .05, and participants who were led to believe that the woman was Jewish rated her qualifications more negatively than did participants who were led to believe that the woman was Italian, F( 1, 92) — 5.2, p < .01. More importantly, these main effects were qualified by a significant interaction between the manipulations of feedback and ethnicity, F( 1, 92} = 7.1, p < .01. As can be seen in Figure 3, participants who had received positive feedback did not evaluate the personality of the target as a function of her apparent ethnicity, whereas participants who had received negative feedback evaluated the qualifications of the target much more negatively if she was portrayed as Jewish than if she was portrayed as Italian. The planned comparison indicated that participants who had re-

Jewish Candidate
Italian Candidate
o

•5
c

u

• t a –

es

Positive Negative

Feedback

Figure 3. Rating of candidate’s personality as a function of feedback
and ethnicity of the candidate. Higher numbers indicate more favorable
evaluations.

PREJUDICE AS SELF-IMAGE MAINTENANCE 39

ceived negative feedback and evaluated the Jewish target were
significantly more negative in their evaluations of the target’s
personality than were participants in all other conditions, f (92)
= 4.5, p < .001. None of the other conditions differed signifi- cantly from each other (see Footnote 4 ) .

The ANOVA of the ratings of the target’s job qualifications
yielded a similar pattern of results. The two main effects were
again significant: Participants who had received negative feed-
back about their performance on the intelligence test rated the
target’s qualifications more negatively than did participants who
had received positive feedback, F ( l , 92) = 3.7, p – .05, and
participants who were led to believe that the woman was Jewish
rated her qualifications more negatively than did participants
who were led to believe that the woman was Italian, F( 1, 92)
= 6.3, p < .05. Although the interaction was not significant for this measure, F ( l , 92) = 2.3, p < .12, the pattern of cell means was consistent with our predictions. Participants who had received positive feedback did not evaluate the target very differently as a function of her apparent ethnicity (MJeulsh = 18.8 vs. Mnaiian = 19.7), but participants who had received negative feedback evaluated the qualifications of the target much more negatively if she was portrayed as Jewish (M = 15.3) than if she was portrayed as Italian (M = 19.3). The planned comparison indicated that participants who had received negative feedback and evaluated the Jewish target were significantly more negative in their evaluations of the target's qualifications than were parti- cipants in all other conditions, z(92) = 3.4, p < .001. None of the other conditions differed significantly from each other.

These results, therefore, provide a conceptual replication of
those found in Study 2 and support the generalizability of the
findings by demonstrating them in the context of a different
stereotype, a different kind of nonstereotyped group, and differ-
ent dependent measures.

Self-Esteem

In Study 3 we measured participants’ state self-esteem at two
points: after the feedback manipulation and after they rated the
target. The theoretical range for this scale is 20 to 100, with
higher numbers indicating higher state self-esteem. As expected,
feedback had a significant effect on participants’ state self-
esteem. Participants who received the positive feedback felt bet-
ter about themselves (M = 77.5) than did those who received
the negative feedback (M = 72.9), F ( l , 94) = 4.4, p < .05.

The change in state self-esteem from this first measure to the
measure taken after participants evaluated the target was also
consistent with predictions. The ANOVA revealed a marginally
significant interaction between feedback and ethnicity, F( 1,92)
= 2.7, p – . 10. As can be seen in Figure 4, and consistent with
our predictions, participants who received negative feedback and
evaluated the Jewish target had a significantly greater increase in
state self-esteem than did participants in the other conditions,
7(92) = 2.3, p < .05. None of the other conditions differed significantly from each other on this measure.

These results suggest that the participants who received nega-
tive feedback and rated the Jewish woman restored their self-
esteem by engaging in negative evaluation of the stereotyped
target. We conducted a path analysis to test this reasoning
(Baron & Kenny, 1986). Figure 5 depicts the results of this

41

a
go

4-

3″

gj
SB

2-

1-

fl Jewish Candidate
EH Italian Candidate

Positive Negative
Feedback

Figure 4. Change in state self-esteem as a function of feedback and
ethnicity of the job candidate. Higher numbers indicate greater increase
in state self-esteem.

analysis. We allowed the planned interaction contrast to predict
change in participants’ self-esteem. This direct effect was sig-
nificant, 0 = .23, r(92) = 2.3, p < .05. Next we allowed the planned interaction contrast to predict participants' ratings of the target's personality. This path was significant as well, /? = .42, t(92) = 4.6, p < .01. Finally, we allowed the planned interaction contrast and participants' ratings of the target's per- sonality to predict participants' change in state self-esteem. The path from participants' ratings of the target was significant, 0 = .37, r(92) = 3.5, p < .01, but the direct effect of the planned interaction contrast on participants' change in self-esteem was no longer significant, /5 = .07, /(92) = 0.7, p > .40. Thus, this
path analysis suggests that the direct effect of the manipulations
in this experiment on participants’ change in state self-esteem
was mediated by their evaluations of the stereotyped target’s
personality. These analyses suggest that the negative feedback
led to increased derogation of the Jewish target, which in turn
led to increased state self-esteem, rather than suggesting that
positive feedback led to a reduced derogation of the Jewish
target.

Taken together, these results provide the first demonstration
that self-image threats, such as negative feedback, can lead to
negative evaluations of a stereotyped target and that these nega-
tive evaluations, in turn, can restore people’s threatened self-
images. Moreover, these findings support our hypothesis that
derogating a stereotyped target in response to self-image threat
mediates increase in self-esteem. These results, therefore,
strongly corroborate the idea that negative evaluations of a ste-
reotyped target may often result from an effort to affirm a
threatened self-image.

40 FEIN AND SPENCER

Planned
Interaction
Contrast

Planned
Interaction
Contrast

.23*

.07

^ ^ 4 2 * *

Change in
State
Self-Esteem

Change in
State
Self-Esteem

Evaluation of
the Target

Figure 5. Change in state self-esteem as mediated by negative evalua-
tions of the job candidate’s personality. *p < .05. **p < .01.

General Discussion

This set of three studies examined evaluations of a member
of a stereotyped group. Study I found that participants evaluated
an individual target person more negatively if they thought she
was a member of a stereotyped group than if they thought she
was a member of a nonstereotyped group, but this effect did
not occur if the participants’ self-images had been bolstered
through an affirmation procedure. Study 2 found that receiving
self-image-threatening information led participants to evaluate
an individual more stereotypically if he appeared to be a member
of a stereotyped group. Study 3 demonstrated that receiving
self-image-threatening information led participants to negatively
evaluate an individual if she appeared to be a member of a
stereotyped group, and these negative evaluations in turn were
particularly effective in restoring participants’ self-esteem.
Moreover, the degree to which these participants made negative
evaluations of the stereotyped target mediated the restoration of
their self-esteem. Taken together, this research suggests that a
threat to one’s positive self-image or a self-affirmation that pro-
vides a buffer against self-image threats can moderate negative
evaluations of a member of a stereotyped group and that these
biased evaluations can in turn affect one’s sense of self-worth.

Self-Affirmation and Negative Evaluations of Others

This set of studies highlights the role of self processes in
the perceptions of others. Information that threatens perceivers’
sense of self-worth leads to the need to restore a positive self-
image. Research by Steele and others (Steele, 1988; Steele &
Liu, 1983; Steele cl al., 1993) has shown that people can restore
a threatened self-image in a number of ways, including by draw-
ing on their own self-concept resources or by taking advantage
of affirmational opportunities available in the situation. Steeie
et al. (1993) have suggested, however, that it may be difficult
for people to spontaneously draw upon their self-concept re-
sources to affirm their self-image. Therefore, people will often
look to the situation to find opportunities to affirm their self-
image. The studies presented here demonstrate that stereotyping
or derogation of a member of a stereotyped group can provide
such situational opportunities to restore a threatened self-image.
Because it is likely that people often will encounter others in

situations where it is personally and socially acceptable to evalu-
ate them negatively, stereotyping and prejudice may be common
reactions to self-image threat. However, when perceivers en-
counter someone who is a member of a group for which they
do not have strong, accessible negative stereotypes, such as the
woman in Studies I and 3 who was Italian or the man in Study
2 who apparently was straight, stereotyping or derogation is
unlikely to be used as a sclf-affirmational strategy.

These studies also suggest that self-affirmation processes may
affect a wide range of phenomena. Most of the research on self-
affirmation theory has examined how self-affirmation affects
cognitive dissonance processes (Steele, 1988; Steele & Liu,
1983; Steele et al., 1993), but some research has suggested that
self-affirmation can also influence self-evaluation maintenance
(Tesser& Cornell, 1991), learned helplessness reactions (Liu &
Steele, 1986), and the academic performance of women and
minorities (Steele & Aronson, 1995). The present research,
in which self-affirmation affected stereotyping and prejudice,
provides further evidence that self-affirmation and self-image
maintenance processes have broad applicability to a wide range
of important phenomena.

Relations to Other Theories

Our approach emphasizes that stereotyping others is one of
several possible self-image-maintenance strategies (Steele,
1988; Tesser & Cornell, 1991). We argue that negatively evaluat-
ing others has the potential to restore a positive self-image.
Because these evaluations are part of a larger self-system that
seeks to maintain an overall image of the self as morally and
adaptively adequate, the state of the self-image—specifically,
the extent lo which it is threatened or affirmed—will influence
when people will engage in stereotyping and when that stereo-
typing will restore a positive self-image. This approach clearly
is related to other theories of stereotyping and prejudice, such
as frustration-aggression, social identity, and downward social
comparison. However, there are distinct theoretical differences
between our approach and these approaches. In addition, the
findings of the current studies support our approach and would
not be predicted by these other theories.

In contrast to frustration-aggression theory, which argues
that people may displace aggression by derogating others in
response to blocked goals and frustrations in their life, our ap-
proach emphasizes that threats to the self-image in particular,
rather than any source of frustration, lead to derogation of oth-
ers. The results of Study 1 highlight this difference. Consistent
with our predictions, we found that self-affirmation reduced
participants’ tendency to derogate a stereotyped targel. It is
unclear from frustration-aggression theory how a self-affirma-
tion procedure such as that used in Study I would reduce frus-
tration, unless frustration is defined more bruadly than it has
been in the past.

Social identity theory suggests that people favor their own
groups over other groups in an effort to boost their group’s
status, which in turn boosts their own self-esteem. Although our
approach would suggest that favoring one’s own group over
another group can restore one’s self-image, we argue that nega-
tively evaluating a stereotyped targel can restore one’s self-
image even if group evaluations and in-group-oul-group com-

PREJUDICE AS SELF-IMAGE MAINTENANCE 41

parisons are not made. In the current studies there is no evidence
that people are making in-group-out-group evaluations or com-
parisons. Both the threats and the affirmation were directed at
the self, rather than at the group, and the evaluations were always
of a single individual. Given that the self-affirmation manipula-
tion in Study 1 was irrelevant to participants’ group identity or
status, it is unclear how social identity theory could account
for the results of this study. Moreover, from a perspective that
emphasizes in-group-out-group differences, one might predict
that the negative feedback in Study 3 should have caused partici-
pants to derogate the Italian candidate because the Italian candi-
date could be considered an out-group member for most of
the participants. In addition, such derogation should have been
associated with a greater increase in self-esteem. The results do
not support this account.

Downward social comparison theory argues that people make
negative evaluations of others to bolster their self-esteem. A
more precisely defined conception of downward social compari-
son, however, might require that social comparisons involve
self-other distinctions. Our approach suggests that such self-
other distinctions might indeed restore one’s self-image, but
negative evaluations of stereotyped others that do not involve
self-other comparisons should also restore one’s self-image. In
the current studies there is no evidence that our participants
made self-other comparisons when evaluating the targets.
Moreover, even if participants made self-other distinctions,
downward social comparison theory would predict that the self-
image threats should have led to derogation of all other targets,
whether or not they appeared to be members of a stereotyped
group. The results of our studies do not support such a
prediction.

At a theoretical level, therefore, our approach is consistent in
many ways with other theories, such as frustration-aggression,
social identity, and downward social comparison theory, al-
though there are some important differences. In addition, only
our account can explain the set of results found in the current
studies.

At an empirical level, several studies have shown that self-
image threat can lead to negative evaluations of others (Brown &
Gallagher, 1992; Crocker et al., 1987; Gibbons & Gcrrard,
1991), and other studies have demonstrated that negative evalua-
tions of others can lead to increased self-esteem (Brickman &
Bulman, 1977; Taylor & Lobel, 1989; Wills, 1991; Wood &
Taylor, 1991). Our studies differ from these previous studies by
demonstrating that when people experience self-image threats,
their negative evaluations of stereotyped others can mediate an
increase in self-esteem. Furthermore, the current studies are the
first to show that thinking about a self-relevant value unrelated
to prejudice can lead to a reduction in stereotyping. Thus, the
findings of the current studies support our contention that stereo-
typic evaluations of others can serve a self-image-maintenance
function.

In our view, any negative evaluation of others—through
downward social comparisons, intergroup discrimination, or
stereotyping and prejudice—has the potential to serve a self-
image-maintenance function. Because of the prevalence, consen-
sual nature, and potential subtlety of negative stereotypes in
particular, stereotyping and prejudice may be an especially com-
mon and effective means of self-affirmation.

The Role of Motivation in Stereotyping and Prejudice

Major reviews of the stereotyping and prejudice literature
(e.g., Ashmore & Del Boca, 1981; Brewer & Kramer, 1985;
Hamilton & Trolier, 1986; Hilton & von Hippel, 1996; Snyder &
Miene, 1994; Stroebe & Insko, 1989) acknowledge the role of
motivational factors (which may be paired with or subsumed
under a personality or psychodynamic approach) as one of the
principal perspectives or approaches to the study of stereotyping
and prejudice, along with the sociocultural and cognitive ap-
proaches. Typically, however, relatively little empirical evidence
beyond research concerning psychodynamic-based constructs
and theories from the 1940s and 1950s or intergroup relations
and related phenomena (e.g., realistic group conflict and social
identity theory) is cited in support of this perspective. The pres-
ent research, along with recent examinations of the roles of
affect and emotion (Esses, Haddock, & Zanna, 1994; Forgas,
1995; Islam & Hewstone, 1993; Mackie & Hamilton, 1993)
and inhibition in stereotyping and prejudice (Bodenhausen &
Macrae, in press; Devine, 1989; Devine, Monteith, Zuwerink, &
Elliot, 1991; Monteith, 1993), examinations of the functions of
stereotyping and prejudice (Snyder & Miene, 1994), examina-
tions of the influence of desired beliefs on person perception
(Klein & Kunda, 1992), and examinations of the roles of self-
esteem and collective self-esteem in intergroup perceptions and
discrimination (Branscombe & Wann, 1994; Crocker et al.,
1987), reflects a burgeoning interest in processes that are rele-
vant to this underdeveloped motivational perspective.

The results of the studies reported in this article suggest that
prejudiced perceptions of members of stereotyped groups can,
under the appropriate conditions, help perceivers restore a posi-
tive self-image. Engaging in stereotyping and prejudice, there-
fore, can be an attractive way for many individuals to feel better
about themselves in the absence of more readily available means
of alleviating self-image threats or of affirming oneself. Given
the same sociocultural context, and given the same cues and
information and information-processing conditions, perceivers
who are motivated to restore a feeling of overall self-worth
should be more likely than other perceivers to seek out or take
advantage of stereotypes.

This is not to suggest, however, that sociocultural and cogni-
tive factors are not also critically important in the processes
examined in our studies. Rather, these studies reflect an interplay
of each of these factors. This is reflected in the interaction
between ethnicity or sexual orientation of the target and the
manipulation of self-affirmation (Study 1) or self-esteem threat
(Studies 2 and 3 ) . If the need to restore a positive overall sense
of self-worth influenced prejudice independently of social-cog-
nitive factors, then the manipulations of self-affirmation and
self-threat should have resulted simply in more positive (when
self-affirmed) or more negative (when the self was threatened
and not affirmed) evaluations of the target individual. Rather,
the manipulations of self-affirmation and self-threat significantly
influenced participants’ evaluations of the target only when they
thought she or he was a member of a group for which there was
a strong and negative stereotype, but not when they thought the
target was not a member of such a group. Furthermore, evaluat-
ing the target negatively was associated with greater self-esteem
boost in the former but not in the latter condition. Thus, the

42 FEIN AND SPENCER

presence of the stereotype, stemming from sociocultural and
cognitive factors, facilitated the process of derogating the target
person and restoring self-esteem.

Only after recognizing the interplay among sociocultural, cog-
nitive, and motivational factors can one adequately address the
question of why derogating any target would not make partici-
pants feel better about themselves. In other words, if a threat to
perceivers’ self-esteem makes them want to restore their self-
esteem, why not derogate an Italian woman if she is more avail-
able than a Jewish woman? Cognitive and sociocultural factors
provide an answer to this question. Within the culture in which
Studies 1 and 3 were conducted, there was a strong negative
stereotype of Jewish American women but not of Italian Ameri-
can women. The JAP stereotype provided participants with the
cognitive basis for perceiving the individual in a negative light.
Similarly, the gay man stereotype provided participants in Study
2 with the cognitive basis for perceiving the individual in a
stereotypical and negative light. Derogation would seem less
justifiable in the absence of the stereotype because participants’
judgments would not have been biased by the stereotype. Rather
than feel better about themselves, most individuals likely would
feel worse if they realized that they had disparaged another
person in order to restore their own sense of self-worth (e.g.,
Devine et al., 1991). Stereotypes, through social-cognitive pro-
cesses such as assimilation, illusory correlations, and schematic
processing, can therefore facilitate self-image maintenance, par-
ticularly to the extent that perceivers are not aware of this
influence.

The Nature of Stereotyping

Most of the stereotypes that we can think of are predomi-
nantly negative. Although they are very different from each
other, stereotypes about African Americans, people with disabil-
ities, Latinos, women. Native Americans, older people, gay men,
lesbians, and those low in social economic status are similar in
that they are primarily negative. The current analysis provides
a possible explanation for the predominantly negative character
of these stereotypes. Although there are undoubtedly other
mechanisms that create and perpetuate negative stereotypes
(e.g., illusory correlations, out-group homogeneity, in-group
bias, and social roles), our analysis suggests that stereotypes
may often take on a negative character because the negativity can
help restore people’s self-images. When people form stereotypes
about a group, they may be more likely to characterize the
group in negative terms because such characterizations allow
evaluations of the group that can be used for later self-affirma-
tion. Similarly, these stereotypes may be particularly resistant
to change because they can make perceivers feel better about
themselves.

This analysis emphasizes the important role that motivation
can play in stereotyping and prejudice. People may be more
likely to stereotype others or engage in prejudicial evaluations
to the extent that they are motivated to restore or enhance their
self-images. Thus, understanding people’s motivations may be
critical in determining whether they will stereotype others, how
they will stereotype others, and what form these stereotypes
will take. Stereotyping and prejudice are clearly an important
problem in our society. Our analysis suggests that a complete

understanding of these processes, and ways of mitigating them,
requires an understanding of the role of the self in people’s
perceptions of others.

References

Abrams, D. (1994). Social self-regulation. Personality and Social Psy-
chology Bulletin, 20, 473-483.

Abrams, D., & Hogg, M. A. (1988). Comments on the motivational
status of self-esteem in social identity and intergroup discrimination.
European Journal of Social Psychology, IS, 317-334.

Ammons, R. B., & Ammons, C. H. (1962). The quick test: Provisional
manual. Psychological Reports, II, 111-161.

Ashraore, R. D., & Del Boca, F. K. (1981). Conceptual approaches
to stereotypes and stereotyping. In D. L. Hamilton (Ed.), Cognitive
processes in stereotyping and intergroup behavior (pp. 1 – 3 5 ) . Hills-
dale, NJ: Erlbaum.

Baron, R. M , & Kenny, D. A. (1986). The moderator-mediator variable
distinction in social psychological research: Conceptual, strategic, and
statistical considerations. Journal of Personality and Social Psychol-
ogy, 51 1173-1182.

Bodenhausen, G. V, & Macrae, C. N. (in press). The self-regulation of
intergroup perception: Mechanisms and consequences of stereotype
suppression. In C. N. Macrae, M. Hewstone, & C. Stangor (Eds.),
Foundations of stereotypes and stereotyping. New York: Guilford
Press.

Branscnmbe, N. R., & Wann, D. L. (1994). Collective self-esteem con-
sequences of outgroup derogation when a valued social identity is on
trial. European Journal of Social Psychology, 24, 641-657.

Brewer, M. B. (1993). Social identity, distinctiveness, and in-group ho-
mogeneity. Social Cognition, 11, 150-164.

Brewer, M. B., & Kramer, R. M. (1985). The psychology of intergroup
attitudes and behavior. Annual Review of Psychology, 36, 219-243.

Brickman, P., & Bulman, R.J. (1977). Pleasure and pain in social
comparison. In J. M. Suls & R. L. Miller (Eds.), Social comparison
processes: Theoretical and empirical perspectives (pp. 149-186),
Washington, DC: Hemisphere.

Brown, J. D., Collins, R. L., & Schmidt, G. W. (1988). Self-esteem and
direct versus indirect forms of self-enhancement. Journal of Personal-
ity and Social Psychology, 55, 445-453.

Brown, J. D., & Gallagher, F. M. (1992). Coming to terms with failure:
Private self-enhancement and public self-effacement. Journal of Ex-
perimental Social Psychology, 28, 3 – 2 2 .

Cameron, J. E., & Lalonde, R. N. (1994). Self, ethnicity, and social
group memberships in two generations of Italian Canadians. Personal-
ity and Social Psychology Bulletin, 20, 514-520.

Cohen, D., & Nisbett, R.E. (1994). Self-protection and the culture of
honor: Explaining Southern violence. Personality and Social Psychol-
ogy Bulletin, 20, 551-567.

Cooley, C. H. (1902). Human nature and the social order. New York:
Schocken Books.

Crocker, J., Thompson, L. J., McGraw, K. M., & Ingerman, C. (1987).
Downward comparison, prejudice, and evaluations of others: Effects
of self-esteem and threat. Journal of Personality and Social Psychol-
ogy, 52, 907-916.

Darley. J. M., Fleming, J. H., Hilton, J. L., & Swann, W. B. (1988).
Dispelling negative expectancies: The impact of interaction goals and
target characteristics on the expectancy confirmation process. Journal
of Experimental Social Psychology, 24, 19-36.

Devine, P. G. (1989). Stereotypes and prejudice: Their controlled and
automatic components. Journal of Personality and Social Psychology,
56, 5-18.

Devine, P. G.. Monteith, M. J., Zuwerink, J. R., & Elliot. A. J. (1991).
Prejudice with and without compunction. Journal of Personality and
Social Psychology, 60, 817-830.

PREJUDICE AS SELF-IMAGE MAINTENANCE 43

Ditto, P. H., & Lopez, D. F. (1992). Motivated skepticism: Use of differ-
ential decision criteria for preferred and nonpreferred conclusions.
Journal of Personality and Social Psychology, 63, 568—584.

Dollard, J., Doob, L. W., Miller, N. E., Mowrer, O. H., & Seare, R. R.
(1939). Frustration and aggression. New Haven, CT: Yale University
Press.

Dovidio, J. R, & Gaertner, S. L. (1991). Changes in the expression and
assessment of racial prejudice. In H. J. Knopke, R. J. Norrell, & R. W.
Rogers (Eds.), Opening doors: Perspectives on race relations in con-
temporary America (pp. 119-148). Tuscaloosa: University of Ala-
bama Press.

Eccles, J. S., Jacobs, J. E., & Harold, R. D. (1990). Gender role stereo-
types, expectancy effects, and parents’ socialization of gender differ-
ences. Journal of Social Issues, 46, 183-201.

Esses, V. M., Haddock, G., & Zanna, M. P. (1994). The role of mood
in the expression of intergroup stereotypes. In M. P. Zanna & J. M.
Olson (Eds.), The psychology of prejudice: The Ontario Symposium
(Vol. 7, pp. 7 7 – 1 0 1 ) . Hillsdale, NJ: Erlbaum.

Fein, S., Cross, J. A., & Spencer, S. J. (1995, August). Self-esteem main-
tenance, stereotype consistency, and men’s prejudice toward gays.
Paper presented at the 103rd Annual Convention of the American
Psychological Association, New York.

Festinger, L. (1954). A theory of social comparison processes. Human
Relations, 7, 117-140.

Forgas, J. P. (1995). Mood and judgment: The affect infusion model
(AIM). Psychological Bulletin, 117, 3 9 – 6 6 .

Gibbons, F. X., & Gerrard, M. (1991). Downward social comparison
and coping with threat. In J. M. Suls & T. A. Wills (Eds.), Social
comparison: Theory and research (pp. 3 1 7 – 3 4 5 ) . Hillsdale, NJ:
Erlbaum.

Gibbons, F. X., & McCoy, S. B. (1991). Self-esteem, similarity, and
reaction to active versus passive downward comparison. Journal of
Personality and Social Psychology, 60, 414-424.

Greenberg, J., Solomon, S., Pyszczynski, T., Rosenblatt, A., Burling, J.,
Lyon, D., Simon, L., & Pinel, E. (1992). Why do people need self-
esteem? Converging evidence that self-esteem serves an anxiety-buff-
ering function. Journal of Personality and Social Psychology, 63,
913-922.

Greenwald, A. G. (1980). The totalitarian ego: Fabrication and revision
of personal history. American Psychologist, 35, 603-618.

Hamilton, D. L., & Trolier, T K. (1986). Stereotypes and stereotyping:
An overview of the cognitive approach. In J. F. Dovidio &. S.L.
Gaertner (Eds.), Prejudice, discrimination, and racism: Theory and
research (pp. 127-163). Orlando, FL: Academic Press.

Hays, W. L. (1981). Statistics (3rd ed.). New York: Holt, Rinehart &
Winston.

Heatherton, T. E, & Polivy, J. (1991). Development and validation of a
scale for measuring state self-esteem. Journal of Personality and
Social Psychology, 60, 895-910.

Higgins, E. T. (1996). Emotional experiences: The pains and pleasures
of distinct regulatory systems. In R. D. Kavanaugh, B, Zimmerberg, &
S. Fein (Eds.), Emotion: Interdisciplinary perspectives (pp. 2 0 3 –
241). Mahwah, NJ: Erlbaum.

Higgins, E. T, & Tykocinski, O. (1992). Self-discrepancies and bio-
graphical memory: Personality and cognition at the level of psycholog-
ical situation. Personality and Social Psychology Bulletin, 18, 5 2 7 –
535.

Hilton, J. L., & von Hippel, W. H. (1996). Stereotypes. Annual Review
of Psychology, 47, 237-271.

Hogg, M. A., & Abrams, D. (1988). Social identifications: A social
psychology of intergroup relations and group processes. London:
Routledge.

Islam, M. R., & Hewstone, M. (1993). Dimensions of contact as pre-
dictors of intergroup anxiety, perceived out-group variability, and out-

group attitude: An integrative model. Personality and Social Psychol-
ogy Bulletin. 19, 700-710.

Keppel, G. (1973). Design and analysis: A researchers handbook.
Englewood Cliffs, NJ: Prentice Hall.

Kite, M., & Deaux, K. (1987). Gender belief systems: Homosexuality
and the implicit inversion theory. Psychology of Women Quarterly,
11, 83-96.

Klein, W. M., & Kunda, Z. (1992). Motivated person perception: Con-
structing justifications for desired beliefs. Journal of Experimental
Social Psychology, 28, 145-168.

Klein, W. M., & Kunda, Z. (1993). Maintaining self-serving social com-
parisons: Biased reconstruction of one’s past behaviors. Personality
and Social Psychology Bulletin, 19, 732-739.

Liu, T. J., & Steele, C. M. (1986). Attributional analysis as self-affirma-
tion. Journal of Personality and Social Psychology, 51, 531 – 5 4 0 .

Mackie, D.M., & Hamilton, D . L . (Eds.). (1993). Affect, cognition,
and stereotyping: Interactive processes in group perception. San
Diego, CA: Academic Press.

Markus, H. (1977). Self-schemata and processing information about the
self. Journal of Personality and Social Psychology, 35, 6 3 – 7 8 .

Markus, H., & Wurf, E. (1987). The dynamic self-concept. Annual
Review of Psychology, 38, 299-337.

Markus, H. R., & Kitayama, S. (1991). Culture and the self: Implica-
tions for cognition, emotion, and motivation. Psychological Review,
98, 224-253.

Markus, H. R., & Kitayama, S. (1994). A collective fear of the collec-
tive: Implications for selves and theories of selves. Personality and
Social Psychology Bulletin, 20, 568-579.

Mead, G. H. (1934). Mind, self, and society. Chicago: University of
Chicago Press.

Mehrabian, A., & Russell, J. (1974). An approach to environmental
psychology. Cambridge, MA: MIT Press.

Miller, N. E., & Bugelski, R. (1948). The influence of frustrations im-
posed by the in-group on attitude expressed toward out-group. Jour-
nal of Psychology, 25, 437-442.

Monteith, M. J. (1993). Self-regulation of prejudiced responses: Impli-
cations for progress in prejudice-reduction efforts. Journal of Person-
ality and Social Psychology, 65, 469-485.

Neuberg, S. L., & Fiske, S. T. (1987). Motivational influences on im-
pression formation: Outcome dependency, accuracy-driven attention,
and individuating processes. Journal of Personality and Social Psy-
chology, 53, 431-444.

Nisbett, R. E., & Ross, L. (1980). Human inference: Strategies and
shortcomings of social judgment. Englewood Cliffs, NJ: Prentice Hall.

Rosenthal, R., & Jacobson, L. (1968). Pygmalion in the classroom:
Teacher expectation and pupils’ intellectual development. New York:
Holt, Rinehart & Winston.

Rosenthal, R., & Rosnow, R. L. (1991), Essentials of behavioral re-
search: Methods and data analysis (2nd ed.) New %rk: McGraw-
Hill.

Ross, M., & Sicoly, F. (1979). Egocentric biases in availability and
attribution. Journal of Personality and Social Psychology, 37, 3 2 2 –
336.

Schlenker, B. R., Weigold, M. F , & Hallam, J. R. (1990). Self-serving
attributions in social context; Effects of self-esteem and social pres-
sure. Journal of Personality and Social Psychology, 58, 855-863.

Smith, E. R. (1993). Social identity and social emotions: Toward new
conceptualizations of prejudice. In D. M. Mackie & D. L. Hamilton
(Eds.), Affect, cognition, and stereotyping: Interactive processes in
group perception (pp. 2 9 7 – 3 1 5 ) . San Diego, CA: Academic Press.

Snyder, M. (1984). When belief creates reality. In L. Berkowitz (Ed.),
Advances in experimental social psychology (Vol. 18, pp. 248—306).
New Irbrk: Academic Press.

Snyder, M., & Miene, P. (1994). On the functions of stereotypes and

44 FEIN AND SPENCER

prejudice. In M. P. Zanna & J. M. Olson (Eds.), The psychology of
prejudice: The Ontario Symposium (Vol. 7, pp. 3 3 – 5 4 ) . Hillsdale,
NJ: Erlbaum.

Spencer, S. J., & Steele, C M . (1990, May). The rule of self-esteem
fitnctioning in IQ estimation. Paper presented at the 62nd meeting of
the Midwestern Psychological Association, Chicago.

Steele, C M . (1988). The psychology of self-affirmation: Sustaining
the integrity of the self. In L. Berkowitz (Ed.), Advances in experi-
mental social psychology (Vol. 21, pp. 2 6 1 – 3 0 2 ) . New York: Aca-
demic Press.

Steele, C M., & Aronson, J. (1995). Stereotype threat and the intellec-
tual test performance of African Americans. Journal of Personality
and Social Psychology, 69, 7 9 7 – 8 1 1 .

Steele, C M., & Liu, T. J. (1983). Dissonance processes as self-affirma-
tion. Journal of Personality and Social Psychology, 45, 5 – 1 9 .

Steele, C M., Spencer, S. J., & Lynch, M. (1993). Self-image resilience
and dissonance: The role of affirmational resources. Journal of Per-
sonality and Social Psychology, 64, 885-896.

Stein, K. F. (1994). Complexity of the self-schema and responses to
discontinuing feedback. Cognitive Therapy and Research, 18, 1 6 1 –
178.

Stroebe, W., &lnsko, C A. (1989). Stereotype, prejudice, and discrimi-
nation: Changing conceptions in theory and research. In D. Bar-Tal,
C F. Graumann, A. W. Kruglanski, & W. Stroebe (Eds.), Stereotyping
and prejudice: Changing conceptions (pp. 3 – 3 4 ) . New %rk:
Springer-Verlag.

Swann, W. B., Stein-Seroussi, A., & Giesler, R. B. (1992). Why people
self-verify. Journal of Personality and Social Psychology, 62, 3 9 2 –
401.

Tajfel, H. (Ed.). (1982). Social identity and intergroup relations. Cam-
bridge, England: Cambridge University Press.

Taylor, S. E., & Lobel, M. (1989). Social comparison activity under
threat: Downward evaluation and upward contacts. Psychological Re-
view, 96, 569-575.

Tfesser, A. (1988). Toward a self-evaluation maintenance model of social
behavior. In L. Berkowitz (Ed.), Advances in experimental social
psychology (Vol. 21, pp. 181-227). New York: Academic Press.

Tesser, A., & Cornell, D. P. (1991). On the confluence of self processes.
Journal of Experimental Social Psychology, 27, 501-526.

Tetlock, P. E. (1983). Accountability and the perseverance of first im-
pressions. Social Psychology Quarterly, 46, 285-292.

Triandis, H. C. (1989). The self and social behavior in differing cultural
contexts. Psychological Review, 96, 506-520.

Turner, J. C (1982). Toward a cognitive redefinition of the social group.
In H. Tajfel (Ed.), Social identity and intergroup relations. Cam-
bridge, England: Cambridge University Press.

Turner, J. C , Oakes, P. J., Haslam, S. A., McGarty, C. (1994). Self
and collective: Cognition and social context. Personality and Social
Psychology Bulletin, 20, 454-463.

Wills, T. A. (1981). Downward comparison principles in social psychol-
ogy. Psychological Bulletin, 90, 245-271.

Wills, T. A. (1991). Similarity and self-esteem in downward compari-
son. In J. M. Suls <& T. A. Wills (Eds.), Social comparison: Theory and research (pp. 5 1 - 7 8 ) . Hillsdale, NJ: Erlbaum.

Winer, B. J. (1971). Statistical principles in experimental design (2nd
ed.). New York: McGraw-Hill.

Wood, J. V., & Taylor, K. L. (1991). Serving self-relevant goals through
social comparison. In J. M. Suls&T. A. Wills (Eds.), Social compari-
son: Theory and research (pp. 2 3 – 5 0 ) . Hillsdaie, NJ: Erlbaum.

Received December 21, 1995
Revision received November 1, 1996

Accepted November 8, 1996 •

O R I G I N A L P A P E R

Putting a price tag on others’ perceptions of us

Yohanes E. Riyanto1 • Jianlin Zhang2

Received: 26 August 2014 / Revised: 26 May 2015 / Accepted: 2 June 2015 /

Published online: 10 June 2015

� Economic Science Association 2015

Abstract Standard economic theories assume that people are self-interested and
their wellbeing solely dependent on their own material gains or losses. Unless they

have an impact on monetary payoffs, the perceptions of anonymous individuals are

irrelevant to people’s decision making. However, a large body of research in

sociology and social psychology demonstrates that self-identity is developed

through one’s understanding of how one is perceived by others. Using (Cooley’s,

Human nature and the social order, 1964) concept of the ‘‘looking-glass self’’ as a

framework, we evaluate experimentally whether or not people care about the

imputed judgment of anonymous others arising from their imagination of their

perceptions. We implemented variants of the Becker–DeGroot–Marschak mecha-

nism to elicit the monetary value attached to the perceptions by participants. In one

variant, only nonnegative bids were allowed, while in another, negative bids were

allowed. We show that in an environment in which the perceptions of others are

only conveyed to participants anonymously and privately, self-interested individuals

exhibited strong negative perception avoidance even though the perceptions have no

impact on their monetary payoff. The participants were willing to spend a

Electronic supplementary material The online version of this article (doi:10.1007/s10683-015-9450-
3) contains supplementary material, which is available to authorized users.

& Jianlin Zhang
zhangjianlin@sim.edu.sg

Yohanes E. Riyanto

yeriyanto@ntu.edu.sg

1

Division of Economics, School of Humanities and Social Sciences, Nanyang Technological

University, 14 Nanyang Drive HSS 04-83, Singapore 637332,

Singapore

2
Global Education, Singapore Institute of Management, 461 Clementi Road, Singapore 599491,

Singapore

123

Exp Econ (2016) 19:480–499

DOI 10.1007/s10683-015-9450-3

http://dx.doi.org/10.1007/s10683-015-9450-3

http://dx.doi.org/10.1007/s10683-015-9450-3

http://crossmark.crossref.org/dialog/?doi=10.1007/s10683-015-9450-3&domain=pdf

http://crossmark.crossref.org/dialog/?doi=10.1007/s10683-015-9450-3&domain=pdf

significant amount in order to avoid confirming the supposedly negative

perception.

Thus, for them, ignorance was truly bliss. We also show that, in the absence of the

audience effect, the fair-minded participants adopted a neutral attitude towards the

perception of them as fair.

Keywords Social image � The value of other’s perceptions � Pay to reveal �
Pay to conceal � Dictator game

JEL Classification C91 � D64 � Z13

1 Introduction

Standard theories in economics assume that people are self-interested and their

wellbeing is solely dependent on their own material gains or losses. Accordingly,

without exerting any impact on monetary payoffs, the perceptions of anonymous

individuals will be irrelevant to one’s decision making. Recent developments in

behavioral economics, however, show that it may not be the case. Recognizing that

people are social beings who seek to maintain a positive image in front of others

(Goffman 1959), concern for social image could drive people’s behaviors away

from homo economicus predictions toward behaviors that are seemingly altruistic,

as shown in a laboratory setting by Andreoni and Bernheim (2009) and in the field

by Ariely et al. (2009), DellaVigna et al. (2012) and Lacetera and Macis (2010).

Beyond a concern for maintaining a social image, others’ perceptions can be

crucial. Indeed, our self-identity is developed through our understanding of how

others perceive us (Cooley 1964). The sociology and social psychology literature

call this the ‘‘looking-glass self’’ metaphor, which comprises three sequential

components (Shaffer 2005). Firstly, we envision how we appear to others. Secondly,

we imagine what others are thinking about our appearance. Thirdly, we develop

feelings about ourselves such as pride or shame based on what we believe these

perceived judgments by others to be. Evidence from psychology studies has shown

that this socio-psychological process begins at an early age and develops throughout

one’s life (Beaman et al. 1979). Because the process shapes the way we see

ourselves, it also affects our judgment of others’ views of us and, consequently, our

relationships with others as well as a raft of socioeconomic decisions (Wallace and

Tice 2012; Yeung and Martin 2003).

Using this metaphor of the looking-glass self as a background, this paper

experimentally evaluates whether people inherently care about an imputed judgment

of anonymous others arising from an imagined perception of the self by others. Our

vehicle is a two-stage modified dictator game. In the first stage of the experiment,

each dictator was firstly asked to choose one of two dividing rules—(S$7, S$7) or

(S$10, S$2)—to split money between him/herself and the matched recipient. The

former rule is fair and welfare maximizing, whereas the latter benefits the dictator

more than the recipient and results in lower welfare overall. The choice of only

these two dividing rules sorted the dictators into two contrasting types: fair-minded

and selfish. The availability of the rules was common knowledge to both dictator

Putting a price tag on others’ perceptions of us 481

123

and the recipient before the game was played. On knowing the outcome of the

allocation decision, the recipients were asked to write a short message giving their

opinion of the dictator. However, the dictator was only told about the existence of

the written message after the allocation decision had been made.

This experimental design captures Cooley’s three sequential components. Firstly,

because the link between the dividing rule choice and the payoff outcome between

the parties was made clear before the dictator chose the rule, the process of

envisioning the recipient’s perception will have started to develop when the dictator

was just about to make the choice (Dana et al. 2007). This constitutes the first

component of the looking-glass self. Subsequent to the allocation decision, the

dictator was informed that the recipient would write a message about the dictator

commenting on his/her choice. This would induce the dictator to imagine how the

recipient might construct a judgment about his/her character consistent with the

dictator’s envisioned belief. This captures the second component. Finally, the

dictator’s perceptions of the recipient’s judgment would induce a feeling of pride or

shame in a manner consistent with the third component of the looking-glass self.

It should be noted that none of the dictators knew the detailed content of the

message before the subsequent bidding process started. Without actually seeing the

message, the dictators’ reflected appraisal would depend solely on their envisioned

belief about the recipient’s view of them. If the dictators cared about it sufficiently,

they would be willing to put a price tag on this socio-psychological factor.

To verify this, in stage two of the experiment, the dictators were asked to express

the maximum they were willing to pay to reveal or conceal the written message

from the recipient through a Becker–DeGroot–Marschak (BDM) bidding mecha-

nism (Becker et al. 1964; Kahneman et al. 1990; Rutström 1998; Shogren et al.

2001). For this purpose, we conducted two experimental treatments. In the pay-to-

conceal (PTC) treatment, by default, the message was privately passed on to the

dictator. The dictator was then given an opportunity to pay to prevent the message

from being passed to him/her. In the pay-to-reveal (PTR) treatment, by default, the

message was not given to the dictator. The dictator was required to bid to obtain it.

For both treatments, the minimum bid allowed is S$0.

When the minimum bid price is bounded by S$0, one problem that may arise is

that subjects might feel compelled to bid strictly more than S$0 in the BDM

procedure. Thus, the elicitation procedure itself might push the bid up from S$0. As

a robustness check, we conducted an additional treatment, which we labeled as the

modified pay-to-reveal (MPTR) treatment. Compared to the previous design, the

main difference is that negative prices were allowed in the BDM procedure.

Essentially, bidding a negative price in order to reveal the message would indicate

that the bidder required monetary compensation from the experimenter in order to

be willing to take the message. It will be seen that this is also equivalent to bidding a

positive amount to avoid the message
1

Canonical economic theories predict that no dictator would be willing to put any

price tag on the message written by the recipient. Consequently, all dictators would

1
We thank one of our anonymous referees for his/her excellent suggestion to conduct this additional

treatment.

482 Y. E. Riyanto, J. Zhang

123

bid zero in the stage-two decision. The theory of social image concern, however,

posits that people care about how they are perceived by others such that they take

actions to influence the formation of others’ perceptions of them. In our

experimental setup, however, the dictator’s image as perceived by the recipient

had been developed before the bidding started in stage two. Since the price

submitted in the bidding process would not alter this perceived image, we thus do

not know whether the recipients’ perception would still matter to the dictators.

Consequently, we also do not know if the dictators would be willing to sacrifice

personal benefits in order to conceal or reveal the written messages.

According to the looking-glass self theory, the dictators would seek to understand

themselves by inferring how they were being perceived by the recipients. In

particular, the knowledge that the recipients had expressed personal opinions about

them would remind the dictators to reconsider the latter’s positions. When the

dictators chose the self-interested dividing rule, this may have caused them to feel

shame or guilt about the possible psychological cost they may have inflicted on the

recipient. If the dictators have an incentive to lessen the negative impact on them,

they would attempt to act ignorant about the message. Such behavior would indicate

that when people envision and anticipate receiving a negative judgment about

themselves from others, it is in their best interests to hide behind a veil of ignorance

by avoiding confirmation of whether such a judgment of their character has been

vindicated.

If, instead, the fair dividing rule was chosen by the dictators, knowing that the

recipient could express an opinion about them would induce the dictators to imagine

that recipient having a happy or satisfied reaction. This would elicit a feeling of

pride in some of the dictators, and if pride has value for them, they would probably

want to know the content of the message to confirm whether the other person’s

judgment of their character had been vindicated. Our experimental results show that

in an environment in which perception of others is only conveyed to participants

anonymously and privately, self-interested dictators exhibited strong negative

perception avoidance even though the perception had no impact on their monetary

payoff. In both the PTC and the MPTR treatments, the dictators were willing to

spend significant monetary amounts in order to avoid confirming the imagined

negative perception of them. Thus, for them, ignorance is bliss. When the bid was

bounded below by S$0 in the PTR treatment, we found that the fair-minded

dictators were willing to spend a significant monetary amount in order to confirm

the imagined positive perception of themselves. When this lower bound was shifted

down to -S$10 in the MPTR treatment, interestingly, the willingness to pay

dropped significantly. This suggests that the fair-minded dictators adopted a neutral

attitude towards fairness perception. This is possibly because having acted fairly by

choosing the fair dividing rule, they may feel confident that the recipient would not

have any ill-feeling towards them, and so they may not feel important to confirm

whether the recipient’s judgment of their character had been vindicated.

Our results thus provide further laboratory evidence for the theories of social

image concern (Bernheim 1994; Ireland 1994; Bagwell and Bernheim 1996; Glazer

and Konrad 1996; Ellingsen and Johannesson 2008a; Tadelis 2007; Andreoni and

Bernheim 2009). One fundamental assumption that underpins these theories is that

Putting a price tag on others’ perceptions of us 483

123

people like to be seen as fair (Andreoni and Bernheim 2009). In our paper, the

observed price paid by the self-interested dictators in the PTC and the M

PTR

treatments is a reflection of dictators’ social image concern. That is, they do not

wish to be seen as unfair and selfish.

This research is closely related to two other studies that explore the use of

anticipated written feedback from a recipient as a nonmonetary sanction on selfish

behavior in the dictator game (Ellingsen and Johannesson 2008b; Xiao and Houser

2009). Both studies report a rise in giving from the dictator when the feedback from

recipients was anticipated before the allocation decision took place. One can argue

that the increase in the amount of giving to the recipients due to such anticipation is

equivalent to the price the dictators are willing to pay in order to either avoid getting

negative messages or to induce positive messages from the recipients. Unfortu-

nately, it is difficult to disentangle the two motives using the experimental design in

these studies. In contrast, in our design, the dictators are only told about the

existence of the written communication after the allocation decision has been made,

so they cannot respond strategically by altering the amount given to the recipients.

By evaluating the dictators’ bidding decisions, we can evaluate how important the

feeling of pride or guilt avoidance is as a motivator.

The rest of this paper is organized as follows. Section 2 discusses our

experimental design and procedures, and Sect. 3 presents our experimental results.

Section 4 discusses the findings from the post experimental survey, and discussion

of the results and conclusions follows in Sect. 5.

2 Experimental design and procedures

The participants in our experiment were undergraduate students at the Singapore

Institute of Management and came from various academic majors such as business,

information technology, arts and social sciences. The experiment was conducted

manually using paper and pen. In total, there were 424 participants.

Upon arriving at the experimental site, participants were randomly directed to

two separate rooms. Because the rooms were situated next to each other,

participants were fully aware that they were being divided into two separate

groups. Once all the participants had arrived at the experimental site, the

experimenter randomly assigned the role of dictators to all participants in one of

the two rooms and the role of recipients to all participants in the other. There were

equal numbers of participants in each room. Once all participants were in their

respective rooms, the experimenter then read out the experimental instructions to

them and checked that they all understood the game. Any question raised was

answered individually in private. The experiment only commenced once all the

participants had understood the game.

The experiment consisted of two stages. Before stage one began, the dictator was

informed that he/she would be asked to make two decisions and that one of the

decisions would be randomly chosen by the experimenter to calculate the dictator’s

earnings for the experiment. However, the dictator was also informed that the

484 Y. E. Riyanto, J. Zhang

123

format of the second decision would be revealed only after all the dictators had

completed the first decision.

At the beginning of stage one, each dictator was asked to choose one of two

dividing rules—(S$7, S$7) or (S$10, S$2)—to split the money (in Singapore

dollars) between themselves and the matched recipient. The dictator was also told

that the matched recipient had been informed about the game and the dividing rules

available to him/her. Once all the dictators understood the game, each then privately

indicated his/her chosen allocation decision on an allocation decision sheet (see

Fig. 1). Each dictator was then asked to insert the decision sheet into an envelope

and to drop the envelope into a box placed in a corner of the room. Both the decision

sheet and the envelope were marked with the dictator’s identity number. Except for

the experimenter, no one knew which identity number was matched with which

participant.

Once all the dictators had completed the allocation decision, the experimenter

collected the envelopes and placed them visibly in a corner of the room. Based on

the allocation decision made, payments to recipients, together with the dictators’

allocation decision sheets, were inserted into their respective envelopes. These

envelopes were then taken to the recipients’ room, and the dictators were told of

this.

The recipients were asked to pick one envelope randomly, verify the monetary

amount against the allocation decision made, and then take the money. Thereafter,

they were asked to write a short message stating their personal opinion of their

matched dictator on the reverse of the allocation decision sheet. Other than the

exclusion of threatening or foul language, no restriction was imposed on the content

of the message that the recipient could write. Once the recipient had written the

message, he/she then inserted it into the same envelope and dropped it back into an

open box for the experimenter to collect. The envelopes were then taken to the

dictators’ room. This completed stage one of the experiment.

In stage two, the dictator was told that his/her matched recipient had received

payment. Based on the allocation decision, the recipient had also written a short

message giving a personal opinion of the dictator. The second decision required the

dictator to express his/her maximum willingness to pay to reveal or to conceal the

written message from the recipient through a BDM mechanism.

A1

BeluRAeluR
I choose to hold 7 dollars for myself and pass 7
dollars to you.

I choose to hold 10 dollars for myself and pass 2
dollars to you.

I decide to choose Rule ________

Fig. 1 Allocation decision sheet

Putting a price tag on others’ perceptions of us 485

123

It should be noted that the experimental instructions for the dictators were split

into two parts. The first gave general instructions and explained how to make the

allocation decision. The second outlined the instructions for the BDM mechanism.

The latter was given to the dictators only after stage one of the experiment had been

completed. As such, the allocation decision was free from any influence based on

the communication from the recipient and/or the BDM mechanism.

Each dictator was given a S$10 endowment in order to participate in the bidding

process. We implemented three experimental treatments as outlined above: PTC,

PTR, and MPTR. In the PTC treatment, by default, the written message from the

recipient was given to the dictator. The dictator could bid to avoid receiving it. In

the PTR and the MPTR treatments, by default, the written message was not given to

the dictator. The dictator could bid to obtain it.

Through the BDM mechanism, we firstly asked the dictator to indicate on a

bidding decision sheet the maximum bid price he/she was willing to pay to either

conceal or reveal the message (see Fig. 2). The sheet was marked with the dictator’s

identity number. The permitted bids ranged from S$0 to S$10 with increments of

S$1 in the PTC and PTR treatments. In the MPTR treatment, the permitted bids

ranged from –S$10 to S$10. A detailed description of the bidding process was given

in the experimental instructions. Once all the dictators had decided on their bids,

they dropped the decision sheets into a box located in the corner of the room for the

experimenter to collect. Based on the identity number, the experimenter then

matched the bidding decision sheet with the corresponding dictator’s envelope.

Once all the dictators had submitted bids, the experimenter then drew a random

price ranging from either S$1 to S$10 in the PTC and PTR treatments, or from -S$9

to S$10 in the MPTR treatment, using a random number generator. If the bid

submitted by a dictator was larger than or equal to the random price, the bid was

considered to be successful and the dictator was allowed to either conceal or reveal

the message. The earnings from stage two were the difference between the S$

10

endowment and the random price.
2

If the submitted bid was less than the random

price, the bid was considered to be unsuccessful and the dictator’s earnings from

stage two were equal to the S$10 endowment.

Immediately after the bidding process had been completed, the experimenter

randomly chose one of the two decisions—allocation or bidding—to determine the

dictator’s final earnings. According to the decision made, a monetary payment to

2
Note that if the successful bid was below S$0 in the MPTR treatment, this essentially implies the

dictator would receive a payment from the experimenter in order to reveal the message.

A1

I decide to bid ________

Fig. 2 Bidding decision sheet

486 Y. E. Riyanto, J. Zhang

123

dictator was then inserted into his/her envelope. However, before the envelopes

were passed to the dictators, the written messages were removed from the

envelopes of those who had succeeded (failed) in the BDM mechanism in the

PTC (PTR and MPTR) treatments. All the envelopes were then placed inside a

box located in a corner of the room. The dictators were asked to collect their

envelopes based on their identity numbers. Before the dictators were allowed to

leave the room, each was given 2 min to read the message in private. Thereafter,

they took the monetary payment and were invited to participate in the post-

experimental survey.

We conducted nine experimental sessions with three sessions in each treatment.

Each session lasted for approximately 45 min. Including the S$5 attendance fee,

participants’ final earnings ranged from S$7 to S$19. Of the 424 participants, half

assumed the role of dictators, and the other half the role of recipients. About 82 %

of the participants assuming the role of dictator were Singaporean students, and the

remainder were from other Asian countries such as Malaysia and Indonesia. The

participants did not know each other before the experiment. None had previously

participated in a similar experiment.

3 Experimental results

Table 1 summarizes the descriptive statistics. Out of 62 dictators in the

PTC

treatment, 59.68 % chose (S$7, S$7) and 40.32 % of them chose (S$10, S$2). There

were 65 dictators in the PTR treatment, of whom 72.31 % chose (S$7, S$7) and

27.69 % chose (S$10, S$2). In the MPTR treatment, the corresponding proportions

were 55.29 and 44.71 %.

Figure 3 presents box plots of the bid prices submitted by the fair-minded and the

self-interested dictators, respectively, in the PTC and the PTR treatments. The boxes

and whiskers show the shape of the distribution, the central value, and the variability

of bids from each type of dictator in each treatment.

As shown in Fig. 3, a large majority of the self-interested dictators in the PTC

treatment submitted positive bids. The distribution function of the bid prices

submitted by these dictators appears to be symmetric, with the mean bid being

slightly more than S$3. This indicates that the self-interested dictators showed a

significant desire to conceal the message. However, the distribution function of the

bid prices submitted by the fair-minded dictators is skewed to the right, indicating

that the majority has less incentive to pay to conceal the written messages received

from their matched recipients.

In the PTC treatment specifically, 59.46 % of the fair-minded dictators bid zero

to conceal the message, whereas only 24 % of the self-interested dictators did so.

A Pearson Chi squared test shows that the association between the allocation

decision and the decision to either bid zero or a positive price was statistically

significant (v2(1) = 7.5746, p value = 0.006). Thus, the dictators’ choice to conceal
the message depended significantly on the type of dividing rule they had selected in

Putting a price tag on others’ perceptions of us 487

123

stage one. The mean bid from the self-interested dictators was S$3.28 and the

median was S$3. For the fair-minded dictators, the mean and median bids were,

respectively, S$1.27 and S$0. The two median bids were significantly different from

zero (Wilcoxon signed-rank: p value = 0.0001 and p value = 0.0000).

Comparatively, on average the self-interested dictators bid higher than the fair-

minded dictators by a factor of approximately 2.6. Both the mean and median bids

Table 1 Summary of the descriptive statistics

Data PTC treatment PTR treatment MPTR treatment

Number of allocators 62 65 85

Number of recipients 62 65 85

Total 124 130 170

Gender

Male 19 25 36

Female 43 40 49

Allocation

Rules (7, 7) (10, 2) (7, 7) (10, 2) (7, 7) (10, 2)

Observation 37 25 47 18 47 38

Bid

Mean 1.27 3.28 3.04 1.61 -0.5 -2.82

Median 0 3 2 1 0 -1.5

SD 1.90 2.74 3.09 2.03 3.36 3.17

0
2

4
6

8
10

B
id

A
m

ou
nt

RTPCTP
(10, 2) (7, 7) (10, 2) (7, 7)

Fig. 3 Box plots of the bid prices under PTC and PTR treatments

488 Y. E. Riyanto, J. Zhang

123

of the former were significantly higher than those of the latter (Mann–Whitney test:

z = -3.110, p value = 0.0019; k-sample median test: v2(1) = 9.3528, p value =
0.002). From the observed submitted bids, it is apparent that the self-interested dictators

exhibited a strong willingness to sacrifice their own monetary benefit in order to conceal

the expected negative messages from the matched recipients.

Studies in psychology demonstrate that the mere imagination of another person’s

emotional state may automatically trigger a representation of that state in the

observer’s mind (Preston and de Waal 2002). Experimental studies in economics

also show that a subject’s ability to show empathy toward others is positively

correlated with his/her prosocial behaviors (Ben-Ner et al. 2004; Ben-Ner and

Halldorsson 2010). Anticipation of receiving a message expressing anger and

disappointment toward the self-interested dictators may induce them to reconsider

the recipient’s position. This may cause them to feel shame or guilt about the

possible psychological cost they have inflicted on the recipient. Nevertheless, the

option to conceal the message gives them an opportunity to reduce the potential

damage from the feeling of shame or guilt, albeit at the expense of a lower monetary

payoff. In this regard, the price paid to conceal the expected message may be

interpreted as a measurement of the degree of shame or the guilt aversion exhibited

by the self-interested dictator.

Interestingly, subsequent to giving the recipient a fair amount, some of the fair-

minded dictators also showed some incentive to conceal the message. There are two

possible reasons for this. Firstly, these fair-minded dictators may want to delay the

resolution of the uncertainty about the expected positive message from the recipient.

If the fair-minded dictators savor the uncertainty about the expected good news,

concealing the message would be a rational decision because it allows them to tap

into this potentially utility-enhancing expectation indefinitely (Chew and Ho 1994;

Loewenstein 1987; Lovallo and Kahneman 2000). Secondly, this could also be

attributed to the experimenter demand effect (Zizzo 2010), something that is not

unique to this setting. However, given that the average bid submitted by the self-

interested dictators was significantly higher than that submitted by the fair-minded

dictators, the incentive of the former to bid was apparently not due merely to the

experimenter demand effect.

We now look at the PTR treatment. As shown in Fig. 3, the distribution function

of the bid prices submitted by the self-interested dictators is skewed to the right,

indicating that a majority of the self-interested dictators did not want to reveal the

written message received from their matched recipients. Also, the distribution

function of the bid prices submitted by the fair-minded dictators appears to be less

asymmetric than for the self-interested dictators. Despite the difference in their

shapes, the two distribution functions were not significantly different (two-sample

Kolmogorov–Smirnov test: p value = 0.259).

In the PTR treatment, 78.72 % of the dictators who chose (S$7, S$7) submitted

positive bids in an attempt to reveal the message. Interestingly, 66.67 % of the

dictators who chose (S$10, S$2) did the same. The Pearson Chi squared test shows

that the association between the allocation decision and the decision to either bid

zero or a positive price was not statistically significant (v2(1) = 1.0196,
p value = 0.313). The mean bid from the fair-minded dictators was S$3.04 and

Putting a price tag on others’ perceptions of us 489

123

that from the self-interested dictators was S$1.61. The median bids were S$2 and

S$1, respectively. Both median bids were statistically different from zero (Wilcoxon

signed-rank: p value = 0.000 and p value = 0.0008). In addition, the mean bid of

the fair-minded dictators was slightly, but significantly, higher than that of the self-

interested dictators (Mann–Whitney test: z = 1.854, p value = 0.0637). However,

the difference in the median bids was statistically significant at the 5 % level (k-

sample median test: v2(1) = 4.3858, p value = 0.036).
The expectation of receiving a negative message from the recipient did not

prevent some self-interested dictators from bidding to reveal it. One possible

explanation is that when the dictator believed that the recipient had expected him/

her to choose the self-interested rule, bidding to reveal the message would allow the

dictator to confirm this expectation should the bid be successful (Murnighan et al.

2001). However, the same explanation may also apply to the fair-minded dictator.

Given that the average bid submitted by the fair-minded dictators in the PTR

treatment was significantly higher than that submitted by the self-interested

dictators, the incentive of the former to bid was apparently not due solely to the

confirmation of expectations. Based on the data from the PTR treatment, the

submitted bids seem to suggest that fairness perception was sought among fair-

minded participants even in the absence of the audience effect. However, as shown

later in the MPTR treatment, this observation could also be an artifact of the S$0

lower bound in the BDM procedure.

Because all subjects participated in only one treatment and the permitted bids

were the same for the PTC and PTR treatments, the between-treatment comparison

in bid prices may also help to shed light on the importance of recipients’ views.

Conditional on choosing the self-interested dividing rule, the mean and median bids

from the dictators in the PTC treatment were significantly higher than those in the

PTR treatment (Mann–Whitney test: z = 2.077, p value = 0.0378; k-sample

median test: v2(1) = 7.0984, p value = 0.008). Thus, after making a self-interested
allocation, the dictators exhibited a stronger incentive to avoid the expected

negative message than to affirm it. Conditional on choosing the fair and welfare-

maximizing choice, the mean and median bids from the dictators in the PTR

treatment were significantly higher than those in the PTC treatment (Mann–Whitney

test: z = -3.210, p value = 0.0013; k-sample median test: v2(1) = 6.7730,
p value = 0.009). Thus, after choosing the fair and welfare-maximizing allocation

decision, the incentive to reveal the expected positive message was stronger than the

incentive to avoid it.

We also ran a series of ordinary least squares (OLS) and Tobit regressions. The

dependent variable was the bid price, and the independent variables were

Treatmenti, an indicator variable with the value of 1 if the subject was in the

PTR treatment and 0 in the PTC treatment; Allocationi, an indicator variable with

the value of 1 if the subject chose (7, 7) and 0 if the subject chose (10, 2); TreatAlloi,

an interactive dummy variable between Treatment and Allocation; and Malei, a

dummy variable with the value of 0 if the subject was female and 1 if the subject

was male. Table 2 summarizes the regression results.

As shown in Table 2, the treatment dummy was statistically significant in all the

regressions. That is, the dictators bid significantly higher in the PTC than in the PTR

490 Y. E. Riyanto, J. Zhang

123

treatment. The allocation dummy was also statistically significant in all regressions,

indicating that the self-interested dictators bid significantly higher than the fair-

minded dictators. However, the impact of gender on bid price was not statistically

significant.

We also performed a treatment effect analysis based on the OLS regression.

Since the impact of gender on bid was not statistically significant, our treatment

effect analysis was based on model (1). As structured by the model, the effect on the

bid of (1) being a fair-minded dictator in the PTR treatment is captured by

(b0 ? b1 ? b2 ? b3), (2) being a self-interested dictator in the PTR treatment is
captured by (b0 ? b1), (3) being a fair-minded dictator in the PTC treatment is
captured by (b0 ? b2), and (4) being a self-interested dictator in the PTC treatment
is captured by b0.

Conditional on being in the PTC treatment, b2 captures the difference in bids
between the self-interested and fair-minded dictators. Conditional on being in the

PTR treatment, (b2 ? b3) captures the difference in bids between the two groups.
Conditional on being a self-interested dictator, b1 captures the difference in bids
between the PTR and the PTC treatments. Finally, conditional on being a fair-

minded dictator, (b1 ? b3) captures the difference in the bid between treatments.
Table 3 summarizes the results of the treatment effect analysis based on the OLS

regression model (1) shown in Table 2. Conditional on the PTC treatment, the

difference in the bid between the self-interested and the fair-minded dictators,

captured by b2, was statistically significant at the 1 % level. Thus, the self-interested
dictators demonstrated a significantly stronger incentive to conceal the message

from the recipients than did the fair-minded dictators. Conditional on the PTR

Table 2 OLS and tobit regressions of the bid prices

Dep. var.: bid OLS Tobit

(1) (2) (3) (4)

Constant (b0) 3.28***
(0.5166)

3.2695***

(0.6512)

2.7559***

(0.7745)

2.7747***

(0.9956)

Treatment (b1) -1.6689**

(0.7984)

-1.6649**

(0.8152)

-2.0628*

(1.2139)

-2.0320*

(1.2350)

Allocation (b2) -2.0097***
(0.6687)

-2.0074***

(0.6771)

-3.4150***

(1.0564)

-3.3943***

(1.0672)

Treatment 9 allocation

(b3)
3.4411***

(0.9796)

3.4367***

(0.9976)

5.4501***

(1.5277)

5.4158***

(1.5482)

Male (b4) -0.0131
(0.4923)

-0.1034

(0.7634)

R-squared 0.1058 0.1058

Adj R-squared

Pseudo R-squared

0.0840 0.0765 0.0325 0.0325

Observations 127 127 127 127

Standard errors are in parentheses

* Significant at 10 % level; ** significant at 5 % level; *** significant at 1 % level

Putting a price tag on others’ perceptions of us 491

123

treatment, the difference in bids between the self-interested and the fair-minded

dictators, captured by (b2 ? b3), was statistically significant at the 5 % level. The
fair-minded dictators exhibited a significantly stronger motive to reveal the message

from the recipients than did the self-interested dictators.

Conditional on being a self-interested dictator, the difference in the bids

between the PTC and the PTR treatments, captured by b1, was statistically
significant at the 5 % level. Thus, the self-interested dictators demonstrated a

stronger incentive to conceal the messages written by the matched recipients than

to reveal them. Finally, conditional on being a fair-minded dictator, the difference

in the bids between the PTC and the PTR treatments, captured by (b1 ? b3), was
statistically significant at the 1 % level. Thus, the fair-minded dictators showed a

stronger incentive to reveal the messages written by the matched recipients than to

conceal

them.

We now look at the results from the MPTR treatment. In stage two, the dictators

were asked to submit 1 bid price from a selection of 21 ranging from -S$10 to S$10.

As S$0 is no longer the lower bound of the bid, this version of the BDM does not

have the problem of pushing the bid up from zero. Given that the dictator bids a

price B and the experimenter chooses a price R which is drawn randomly from

-S$9 to S$10, the probability that B C R will be 1
20

10 þ Bð Þ, and 1
20

10 � Bð Þ
otherwise. Consequently, the dictator’s expected payoff from the BDM procedure

can be expressed as follows:

E Pð Þ ¼
X

B � R

1

20
10 � Rð Þ þ

10 � Bð Þ
20

10

Note that the expected payoff under the new BDM procedure follows an inverted-U

shape even though the probability of being successful in the bidding process is still

Table 3 The differences in the bid across treatments and allocations

Allocation

(10, 2) (7, 7) Difference in Bid

Treatment

PTC b0 (b0 ?b2) b2
-2.010***

(0.668682)

PTR (b0 ? b1) (b0 ? b1 ? b2 ? b3) b2 ? b3
1.4314**

(0.7159)

Difference in Bid b1
-1.6689**

(0.7984)

(b1 ? b3)

1.7723***

(0.5677)

Standard errors are in parentheses
* Significant at 10 % level; ** significant at 5 % level; *** significant at 1 % level

492 Y. E. Riyanto, J. Zhang

123

monotonically increasing in the bid price. This is in contrast to the earlier BDM

procedure, where a tradeoff between a lower expected payoff and a higher proba-

bility of winning always applied to the entire range of bid prices. For both designs,

however, the payoff maximizing bid is S$0. The detailed payoffs corresponding to

every possible combination of bid and random price can be found in the experi-

mental instructions.

For a self-interested dictator, the optimum choices that maximize the monetary

payoffs from both decisions would involve firstly choosing the unfair rule in stage

one and secondly submitting a bid price of either -S$1 or S$0 in the BDM process.

If image matters to the self-interested dictator, he/she might want to bid low prices

so as to lower the probability of receiving the message. Between the prices of

-S$10 to -S$1, however, the expected payoff increases with the bid. The dictator

would therefore need to balance the tradeoff between a lower probability of

receiving the message and a higher expected payoff. While the self-interested

dictator might develop feelings of embarrassment and regret as a result of choosing

the unfair dividing rule, a fair-minded dictator who has given fairly and generously

would not. Consequently, we should expect the fair-minded dictators to refrain from

bidding highly negative prices.

Figure 4 presents box plots of the bid prices submitted by both types of dictators

in the MPTR treatment. As shown in Fig. 4, the distribution function of the bid

prices submitted by the self-interested dictators is heavily skewed to the left. Results

from the Shapiro–Wilk W test for normality reject the hypothesis that the

distribution of the bid prices was normal (z = 1.666; p \ 0.04781). Specifically,
66 % of the self-interested dictators bid negative prices, 29 % of them bid S$0, and

5 % of them bid positive prices. The mean bid was -S$2.82 and the median

-S$1.5. By bidding negative prices, the self-interested dictators demonstrated a

preference to voluntarily lower their expected earnings in order to avoid receiving

the message. These observations are consistent with our previous results obtained

Fig. 4 Box plots of the bid prices under MPTR treatment

Putting a price tag on others’ perceptions of us 493

123

from the PTC treatment, where it was observed that the self-interest dictators spent

S$3.28 on average in order to avoid receiving the message. Both sets of results thus

confirm our hypothesis that the self-interested dictators exhibit a preference to avoid

confronting a negative image of themselves even though it has no material impact

on them.

Figure 4 also shows that the distribution function of the bid prices submitted

by the fair-minded dictators was symmetric around -S$0.5. Results from the

Shapiro–Wilk W test for normality cannot reject the hypothesis that the

distribution of the bid prices was normal (z = 0.747; p \ 0.22750). Specifically,
40 % of the fair-minded dictators bid S$0 and 11 % bid -S$1. The rest chose to

bid either positive prices or less than -S$1. The mean bid was -$0.5 and the

median S$0.

The observation that the majority of the fair-minded dictators chose payoff-

maximizing bids indicates that they had adopted a neutral attitude towards the

message. Since they had given fairly and generously, they did not feel morally

deficient and, hence, had no particular urge to avoid the message. We also found

that about 20 % of the fair-minded dictators bid positive prices. This suggests that

positive image-seeking could be the reason behind some of these fair-minded

dictators’ bids. Another 30 % of the fair-minded dictators, however, bid prices

lower than -S$1, suggesting that another possible reason for them to conceal the

message could have been modesty. This result moderates substantially the previous

result under the PTR treatment, where we showed that the fair-minded dictators

wanted to confirm the supposedly nice message written by the recipient. It also

implies that the previous result under the PTR treatment could be caused by the

artifact of the BDM procedure we used in that treatment where the lower bound of

bid was S$0. Subjects might feel compelled to bid strictly positive amount in such a

setting.

It should be noted that both the mean and median bid prices submitted by the fair-

minded dictators were still higher than those submitted by the self-interested

dictators. The results from nonparametric tests show that the differences in the mean

and median bids of each type of dictators were statistically significant (Mann–

Whitney test: z = -2.875, p value = 0.0040; k-sample median test:

v2(1) = 5.4137, p value = 0.020). These observations are consistent with our
previous results from the PTC and PTR treatments.

We also examined the content of some of the messages sent by recipients to both

the selfish and fair-minded dictators. We can infer that the former were often

negative and the latter were generally positive. For example, one recipient said to a

selfish dictator ‘‘Wow! You are selfish. If you don’t feel you are selfish, then you are

worse than selfish.’’ Another said ‘‘Money is the root of all evil, or greed. (You)

should at least divide it equally to be fair. Oh well, GREEDY.’’ In contrast, the

messages sent to fair-minded dictators were generally pleasant, such as ‘‘You are a

good person,’’ ‘‘May God bless you,’’ and even ‘‘Marry me!’’
3

3
A better way of analyzing the messages would be to follow the procedure developed by Houser and

Xiao (2010).

494 Y. E. Riyanto, J. Zhang

123

4 Survey results

One plausible explanation for these results could be the correlation between pro-

sociality and the individual’s attitude towards the perceptions of others’. That is,

pro-social individuals may be inherently interested in knowing how others view

them than self-interested individuals. One way to confirm this would be to check

whether the two types of dictators exhibited significant differences in their attitudes

toward others’ perception based on their answers to the post experimental survey.

In the survey, the dictators were asked to indicate the level of importance of

others’ opinion or perception using a 5-point scale ranging from ‘‘Not at all

important’’ to ‘‘Very important.’’ If pro-sociality is indeed correlated with the desire

to know others’ perceptions, we should expect that the self-interested dictators to

give such perceptions a significantly lower importance than the fair-minded ones.

Based on the answers to the survey question, however, we found no such

correlation.
4

The mean reported scales for both types were not significantly different

from each other in the PTC (Mann–Whitney test: z = -0.723, p value = 0.4694),

the PTR (Mann–Whitney test: z = 1.400, p value = 0.1616) and the

MPTR

treatments (Mann–Whitney test: z = 1.091, p value = 0.2754). Overall, the

observed results do not seem to support the idea that pro-sociality is correlated

with the desire to know others’ perception of oneself.

In the survey, we also asked participants to choose a reason for submitting a bid

out of out possible reasons. They are; ‘‘Out of curiosity’’, ‘‘I just want to try my

luck’’, ‘‘My matched Player Y’s message is important to me’’, ‘‘My matched Player

Y’s message is not important to me’’, ‘‘I care about Player Y’s personal feeling’’,

and ‘‘I don’t care about Player Y’s personal feeling’’. We also gave an opportunity

for participants to write down their reasons if none of the given reasons explain their

bidding behavior. The following were observed from the answers to the survey

(Table 4).

On average, the most frequently cited reason was ‘‘Out of curiosity. I just want to

try my luck’’. This suggests that participants were concerned about the likelihood of

success of their bids. Conditional on this reason, the observed bid prices within a

treatment were different between the two types of dictators. In the PTC treatment,

the mean bid price submitted by the fair minded dictators was smaller than that of

the self-interested dictators. The pattern of the difference in bid prices was reversed

in the PTR treatment and the MPTR treatment. This observation suggests that,

beyond curiosity, the difference in bid prices could be caused by the self-interested

dictator’s motive to avoid confirming negative image.

Another commonly cited reason was ‘‘My matched Player Y’s message is not

important to me’’. This reason was cited by more than a quarter of self-interested

dictators. Not surprisingly, these self-interested dictators also submitted bids that

were close to the payoff maximizing bids. Both their allocation decision and bidding

decision are thus consistent with the reason cited. Notice that slightly less than a

4
The answers given to the survey questions were not incentivized. In addition, participation in the survey

was not compulsory, so the return rate was around 70 %. The results should therefore be interpreted with

some caution.

Putting a price tag on others’ perceptions of us 495

123

quarter of fair-minded dictators also cited this reason. Interestingly, the bids

submitted by these fair-minded dictators were also close to the payoff maximizing

bids. Even though both types of dictators cited the same reason and submitted

similar amount of bid prices, the motives behind the bids were probably different.

While ignorance was probably the motive behind the self-interested dictators’ bid,

the same motive probably does not apply to the fair-minded dictators. Dozens of

fair-minded dictators indicated in their written answers to the questionnaires that

they expected the recipient to write positive messages to them. Consequently, they

submitted a payoff maximizing bid price because receiving the message or not is no

longer important to them.

Also unsurprisingly, ‘‘I care about Player Y’s personal feeling’’ was almost

absent from the reason cited by the self-interested dictators. Likewise, ‘‘I don’t care

about Player Y’s personal feeling’’ was only cited by few fair-minded dictators.

Lastly, some participants preferred to write down the reasons behind their bids. The

reasons given by these participants largely show justifications for their allocation

decisions and the subsequent bidding decisions. For example, one fair-minded

participant who bid S$0 in the PTR treatment wrote: ‘‘I have chosen the one I feel

Table 4 Post-experimental survey results

Reasons (7, 7) (10, 2)

Mean SD n/N (%) Mean SD n/N

PTC

Out of curiosity 2.88 0.85 22 4.14 0.91 28

Message is important to me 1.17 0.48 16 1.50 0.50 8

Message is not important to me 0.00 0.00 14 1.57 0.87 28

Care about Y’s personal feeling 0.90 0.64 27 3.50 0.50 8

Don’t care about Y’s personal feeling – – 0 3.33 2.03 12

Other reasons 1.00 0.50 22 5.50 1.76 16

PTR

Out of curiosity 3.94 0.78 38 2.78 0.76 50

Message is important to me 7.00 0 2 – – 0

Message is not important to me 0.64 0.31 23 0.33 0.21 33

Care about Y’s personal feeling 4.63 1.22 17 – – 0

Don’t care about Y’s personal feeling 0.00 0.00 2 0.00 0.00 6

Other reasons 2.63 0.80 17 1.00 1.00 11

MPTR

Out of curiosity -0.22 12.00 23 -2.23 15.00 42

Message is important to me 2.00 2.00 8 -5.00 1.00 3

Message is not important to me -1.21 14.00 35 -0.88 11.00 2

Care about Y’s personal feeling 2.20 7.00 13 0.00 0.00 0

Don’t care about Y’s personal feeling -2.50 2.00 5 -5.75 4.00 13

Other reasons -1.00 9.00 18 -3.80 6.00 16

496 Y. E. Riyanto, J. Zhang

123

will benefit both players the most. Thus, there is a high probability that the message

will be a positive one. Thus, I feel little need for me to confirm the message.’’ One

self-interested dictator who bid $3 in the PTC treatment wrote: ‘‘Because I don’t

know who player Y is, my decision might change if I knew who I was deciding for.’’

5 Conclusions

Canonical economic theories predict that others’ perceptions of oneself would be

irrelevant if they have no material impact. The literature in sociology and social

psychology demonstrates that our self-identity is developed through our under-

standing of how others perceive us. Using the looking-glass self as a framework, we

reconstructed Cooley’s (1964) three sequential components inside an economic

laboratory. We also implemented the BDM procedure (Becker et al. 1964) to elicit

the monetary values that participants placed on how anonymous others perceived

them.

We showed that self-interested dictators exhibited strong negative perception

avoidance even in an environment in which all messages were conveyed to them

anonymously and privately. People develop negative self-views when they realize

that they have fallen short of their internal standards (Fejfar and Hoyle 2000;

Moskalenko and Heine 2003). Often, avoidance is instinctively seen by self-

perceived transgressors as a way to lessen the feelings of guilt resulting from being

reminded of such discrepancies (Duval and Silvia 2002; Csikszentmihalyi and

Figurski 1982). Hence, our results show that people may be willing to spend a

significant amount to avoid confirming what they imagine to be others’ negative

perception of themselves. Thus, ‘‘where ignorance is bliss, ‘tis folly to be wise’’

(Ode on a Distant Prospect of Eton College, Thomas Gray (1742)).

We also show that the fair-minded participants adopted a neutral attitude towards

perceptions of fairness in the absence of the audience effect. Surprisingly, the

anticipation of finding out about negative perceptions did not appear to deter some

self-interested dictators from bidding to reveal the messages. This indicated the

presence of heterogeneity among dictators in terms of the value of others’

perception.

Acknowledgments Financial support from the Nanyang Technological University and the Singapore
Institute of Management is gratefully acknowledged. We thank Francis Chin, Stephen Knowles, Maroš

Servátka, David de Meza, Jörn Rothe, Tatsuyoshi Saijo, Jason Shachat, and Songfa Zhong, as well as

participants at the 2013 World Meetings of the Economic Science Association (ESA) at the University of

Zurich (Switzerland), the 2013 Asia Meeting of the Econometric Society at the National University of

Singapore (Singapore), and the 2014 Asia–Pacific Meeting of the Economic Science Association (ESA) at

the University of Auckland (New Zealand).

References

Andreoni, J., & Bernheim, B. D. (2009). Social image and the 50-50 norm: A theoretical and

experimental analysis of audience effects. Econometrica, 77(5), 1607–1636.

Putting a price tag on others’ perceptions of us 497

123

Ariely, D., Bracha, A., & Meier, S. (2009). Doing good or doing well? Image motivation and monetary

incentives in behaving pro-socially. American Economic Review, 99(1), 544–555.

Bagwell, L. S., & Bernheim, B. D. (1996). Veblen effects in a theory of conspicuous consumption.

American Economic Review, 86(3), 349–373.

Beaman, L., Diener, E., & Svanum, S. (1979). Self-awareness and transgression in children: Two field

studies. Journal of Personality and Social Psychology, 37(10), 1835–1846.

Becker, G., DeGroot, M., & Marschak, J. (1964). Measuring utility by a single-response sequential

method. Behavioral Sciences, 9(3), 226–236.

Ben-Ner, A., & Halldorsson, F. (2010). Trusting and trustworthiness: What are they, how to measure

them, and what affects them? Journal of Economic Psychology, 31(1), 64–79.

Ben-Ner, A., Kong, F., & Putterman, L. (2004). Share and share alike? Gender-pairing, personality, and

cognitive ability as determinants of giving. Journal of Economic Psychology, 25(5), 581–589.

Bernheim, B. D. (1994). A theory of conformity. Journal of Political Economy, 102(5), 841–877.

Chew, S. H., & Ho, J. L. (1994). Hope: An empirical study of attitude toward the timing of uncertainty

resolution. Journal of Risk and Uncertainty, 8(3), 267–288.

Cooley, C. H. (1964). Human nature and the social order. New York: Schocken.

Csikszentmihalyi, M., & Figurski, T. (1982). Self-awareness and aversive experience in everyday life.

Journal of Personality, 50(1), 15–28.

Dana, J., Weber, R. A., & Kuang, J. X. (2007). Exploiting moral wiggle room: Experiments

demonstrating an illusory preference for fairness. Economic Theory, 33(1), 67–80.

DellaVigna, S., List, J., & Malmendier, U. (2012). Testing for altruism and social pressure in charitable

giving. Quarterly Journal of Economics, 127(1), 1–56.

Duval, T. S., & Silvia, P. J. (2002). Self-awareness, probability of improvement, and the self-serving bias.

Journal of Personality and Social Psychology, 82(1), 49–61.

Ellingsen, T., & Johannesson, M. (2008a). Pride and prejudice: The human side of incentive theory.

American Economic Review, 98(3), 990–1008.

Ellingsen, T., & Johannesson, M. (2008b). Anticipated verbal feedback induces altruistic behavior.

Evolution and Human Behavior, 29(2), 100–105.

Fejfar, M. C., & Hoyle, R. H. (2000). Effect of private self-awareness on negative affect and self-referent

attribution: A quantitative review. Personality and Social Psychology Review, 4(2), 132–142.

Glazer, A., & Konrad, K. A. (1996). A signaling explanation for charity. American Economic Review,

86(4), 1019–1028.

Goffman, E. (1959). The presentation of self in everyday life. New York: Doubleday.

Houser, D., & Xiao, E. (2010). Classification of natural language messages using a coordination game.

Experimental Economics, 14(1), 1–14.

Ireland, N. (1994). On limiting the market for status signals. Journal of Public Economics, 53(1), 91–110.

Kahneman, D., Knetsch, J., & Thaler, R. (1990). Experimental tests of the endowment effect and the

Coase theorem. Journal of Political Economy, 98(6), 1325–1348.

Lacetera, N., & Macis, M. (2010). Social image concerns and pro-social behavior: Field evidence from a

nonlinear incentive scheme. Journal of Economic Behavior & Organization, 76(2), 225–237.

Loewenstein, G. (1987). Anticipation and the valuation of delayed consumption. The Economics Journal,

97(387), 666–684.

Lovallo, D., & Kahneman, D. (2000). Living with uncertainty: Attractiveness and resolution timing.

Journal of Behavioral Decision Making, 13(2), 179–190.

Moskalenko, S., & Heine, S. J. (2003). Watching your troubles away: Television viewing as a stimulus

for subjective self-awareness. Personality and Social Psychology Bulletin, 29(1), 76–85.

Murnighan, J. K., Oesch, J. M., & Pillutla, M. (2001). Player types and self-impression management in

dictatorship games: Two experiments. Games and Economic Behavior, 37(2), 388–414.

Preston, S. D., & de Waal, F. B. (2002). Empathy: Its ultimate and proximate bases. Behavioral Brain

Science, 25(1), 20–72.

Rutström, E. (1998). Home-grown values and incentive compatible auction design. International Journal

of Game Theory, 27(3), 427–441.

Shaffer, L. (2005). From mirror self-recognition to the looking-glass self: Exploring the justification

hypothesis. Journal of Clinical Psychology, 61(1), 47–65.

Shogren, J., Cho, S., Koo, C., List, J., Park, C., Polo, P., & Wilhelmi, R. (2001). Auction mechanisms and

the measurement of WTP and WTA. Resource and Energy Economics, 23(2), 97–100.

Tadelis, S. (2007). The power of shame and the rationality of trust. Berkeley: Mimeo, Haas School of

Business, U.C.

498 Y. E. Riyanto, J. Zhang

123

Wallace, H. M., & Tice, D. M. (2012). Reflected appraisal through a 21st-century looking glass. In M.

R. Leary & J. P. Tangney (Eds.), Handbook of self and identity (Vol. 2, pp. 124–140). New York:

Guilford.

Xiao, E., & Houser, D. (2009). Avoiding the sharp tongue: Anticipated written messages promote fair

economic exchange. Journal of Economic Psychology, 30(3), 393–404.

Yeung, K. T., & Martin, J. L. (2003). The looking glass self: An empirical test and elaboration. Social

Forces, 81(3), 843–879.

Zizzo, D. J. (2010). Experimenter demand effects in economic experiments. Experimental Economics,

13(1), 75–98.

Putting a price tag on others’ perceptions of us 499

123

Reproduced with permission of the copyright owner. Further reproduction prohibited without
permission.

• c.10683_2015_Article_9450

Putting a price tag on others’ perceptions of us
Abstract
Introduction
Experimental design and procedures
Experimental results
Survey results
Conclusions
Acknowledgments
References