Professional Documents
Culture Documents
Crandall Et Al. (2002)
Crandall Et Al. (2002)
The authors studied social norms and prejudice using M. Sherif and C. W. Sherif’s (1953) group norm
theory of attitudes. In 7 studies (N ⫽ 1, 504), social norms were measured and manipulated to examine
their effects on prejudice; both normatively proscribed and normatively prescribed forms of prejudice
were included. The public expression of prejudice toward 105 social groups was very highly correlated
with social approval of that expression. Participants closely adhere to social norms when expressing
prejudice, evaluating scenarios of discrimination, and reacting to hostile jokes. The authors reconcep-
tualized the source of motivation to suppress prejudice in terms of identifying with new reference groups
and adapting oneself to fit new norms. Suppression scales seem to measure patterns of concern about
group norms rather than personal commitments to reducing prejudice; high suppressors are strong norm
followers. Compared with low suppressors, high suppressors follow normative rules more closely and are
more strongly influenced by shifts in local social norms. There is much value in continuing the study of
normative influence and self-adaptation to social norms, particularly in terms of the group norm theory
of attitudes.
Most of the bigoted remarks I have heard and prejudice I have nomic and political repression, which goes too often unacknowl-
experienced came from people who were trying to be popular, not edged (Pettigrew, 1991). Although all of these contextual factors
despised. They were following what they believed to be acceptable are important to understanding individual prejudices, they often
behavior in their group or sub-group, not deviating from it. contaminate phenomena and limit the generality of data and the-
—Clarence Page (1999, p. B5) ory. Sherif (1948) argued that
Psychological research on prejudice has focused on three basic [It] is safe to assume that there cannot be separate psychologies of
areas: personality and attitude systems, cognitive dynamics, and prejudice in relation to this or that group, but that they are specific
social norms (Pettigrew, 1991). Each area paints a part of the cases of the general picture of prejudice. This general picture of
psychological picture of prejudice, and each area is incomplete by prejudice is, in turn, part and parcel of the psychology of social
attitudes and of identifications. (p. 64)
itself. In the last several decades, research on the cognitive dy-
namics has enjoyed a remarkable success in illuminating the basic In the present program of research, we use a social normative
cognitive processes of prejudice, and research in personality and approach to study how social norms affect the expression of
attitude systems has been the object of sustained attention over prejudices across a wide range of possible targets.
many years (Pettigrew, 1999). With few exceptions, research and
theory in the area of social norms have languished.
In this article, we revisit social norms and prejudice and try to Definition of Prejudice
illuminate the value of applying traditional theoretical perspectives We define prejudice as a negative evaluation of a group or of an
to develop broader understandings of psychological phenomena individual on the basis of group membership. This definition does
(e.g., Sherif, 1936, 1948). Research on the social psychology of not differentiate between prejudices that are based on an erstwhile
prejudice has largely been focused on individual prejudices (e.g., “accurate” perception of a group and does not depend on whether
racism, anti-Semitism, antifat prejudice). When studying a partic- such a prejudice can be justified according to some moral code.
ular prejudice, researchers study a phenomenon surrounded by a Unlike Allport (1954), we do not require that a prejudice be
context of social history, power relations, legal history, and eco- inaccurate, unjustified, or overgeneralized (a standard that Allport
acknowledged was virtually impossible to satisfy). Instead, we
focus on the processes that are common across all group-based
Christian S. Crandall, Amy Eshleman, and Laurie O’Brien, Department negative evaluations—whether some psychological authority de-
of Psychology, University of Kansas. termines such a negative evaluation is earned or not is irrelevant to
Amy Eshleman is now at the Department of Psychology, Wagner understanding the underlying psychological phenomena. Racial
College, Staten Island, New York.
prejudice has apparently declined over the decades. Angus Camp-
We thank Eileen DelCore, Jessica Stites, and Stacy Johnson for their
hard work in developing extensive lists of groups and Jennifer Ma and
bell (1947), writing just after World War II, described openly
Chuck Stangor for comments on an earlier version of this article. expressed prejudice as “old-fashioned” and socially unacceptable.
Correspondence concerning this article should be addressed to Christian People’s willingness to express prejudice has continued to de-
S. Crandall, Department of Psychology, 1145 Jayhawk Boulevard, Uni- crease (Dowden & Robinson, 1993); attitudes toward racial and
versity of Kansas, Lawrence, Kansas, 66045. E-mail: [email protected] ethnic minorities as measured in surveys are becoming increas-
359
360 CRANDALL, ESHLEMAN, AND O’BRIEN
ingly positive and less prejudiced (Case & Greeley, 1990; Fire- prejudice “are not the product of individual preferences acquired
baugh & Davis, 1988). Many sociologists and psychologists inter- over the lifetime of this or that individual. They are the products of
pret these trends as evidence that the broad normative climate had contact with members of a group; they are standardized and
turned against racial prejudice (Dovidio & Gaertner, 1986; become common property within a group” (p. 124). Individual
Rokeach & Ball-Rokeach, 1989). To what extent does the mea- ideologies and belief systems are based on the social norms of
sured decrease in expressions of prejudice reflect genuine changes groups with which a person identifies. Sherif argued that individ-
in attitude? To what extent does it reflect internalization of pre- ual points of view are the socially accepted norms of one’s
vailing social norms? Or, to what extent does it reflect conformity group(s) (Sherif, 1936).
without internalization? Although expressions have clearly
Attitudes toward members of other groups, as well as attitudes toward
changed, we cannot be sure whether genuine belief has fallen in one’s own group, are learned. But attitudes toward members of other
line with the prevailing antiprejudice norms or whether respon- groups are not determined so much by experiences while in contact
dents are merely cutting their conscience to fit the current fashion. with the groups in question as by contact with the attitudes toward
Although it may be encouraging that survey reports of prejudice these groups prevailing among the older members of the groups in
are on the decline, these reports may reflect conformity to social which they develop. (Sherif & Sherif, 1953, pp. 94 –95)
rules regarding appropriate behavior rather than personal values
and beliefs. Researchers must examine the (lack of) implications of This process develops through a mixture of internal and external
these declining reports of prejudice. In the case of racism, Weitz conflict, with its resolution based on the internalization of external
(1972) found that “Extremely favorable verbal attitudes were cou- norms. Sherif (1936) wrote that
pled with subtle signs of rejection of blacks” (p. 20). Crosby, A social value is first external to the individual. Being external to the
Bromley, and Saxe (1980) reviewed extensive evidence that racial individual, it has for each individual who first confronts it, an objec-
prejudice and discriminatory behavior appear low in prevalence tive reality . . . we come to realize acutely the reality of social values
when measured overtly but that prejudice and discrimination are when we violate them and thereupon find ourselves in an embarrass-
much more extensive when prejudice is measured unobtrusively. ing situation, or punished in varying degree according to the nature of
Hurwitz and Peffley (1992) noted, “While there has been a dra- the offense; and no less so when we find internal conflict, a war within
ourselves . . . this is, to a great extent, the social in him. (p. 125)
matic increase in support for the principles of equality and inte-
gration, this positive trend has clearly not been extended to support In the GNT, internalization—and conscious awareness of it—
for policies designed to implement these goals” (p. 396, emphasis comes from group identification with new groups with different
in original). Americans may claim to want to end inequality, but norms and values.
they are not generally willing to support affirmative action or
Increasingly with age comes an awareness of one’s own group and
busing programs, actions that would lead toward such a goal. The
other groups and their relative positions in the existing scheme of
evidence suggests that a substantial amount of the reduction in
social relationships. It is no coincidence that this increasing awareness
prejudice reports is based on conformity pressures and not on and increasingly consistent manifestation of a scale of prejudice
fundamental changes in people’s hearts and minds. occurs during the stage when the child is beginning to participate
Allport’s (1954) The Nature of Prejudice has served for actively in group activities—that is, when he can psychologically
nearly 50 years as a guide to prejudice researchers (Ruggiero & become a member of a group . . . [prejudices] become so much a part
Kelman, 1999). In recent decades, researchers have focused on the of himself— of his ego—that the individual usually becomes unaware
cognitive aspects of prejudice that Allport reviewed, but Allport of their derivations but considers them his own. (Sherif, 1948, p. 66)
focused even more on normative approaches. The best specified
In his classic study of the racial attitudes of children, Lasker
psychological theory of social norms during Allport’s time—and
(1929) found that racial attitudes did not come from actual contact
still useful in ours—is Sherif and Sherif ’s group norm theory
with racial groups. As Sherif (1936) described, racial attitudes are
(GNT; e.g., Sherif, 1936; Sherif & Sherif, 1953, 1964).
“a matter of adopting the established values of the group. . . . The
child’s attitude toward the Negro is the result not of contact with
Sherif and Sherif’s Group Norm Theory the Negro but of contact with the prevailing attitude toward Ne-
groes” (p. 140; see also Horowitz, 1936). In short, to be a good
Sherif and Sherif ’s (1953) GNT describes the development of
group member, one must adopt the prejudices that the group holds
prejudice-related norms within social groups and the pressures
and abstain from those prejudices that the group frowns upon.1
placed on individuals to conform to group norms. In GNT, atti-
tudes toward particular objects are formed by simply adopting the
Norms and Attitude Change
attitudes of a valued group.
The GNT of prejudice predicts that, because an attitude’s true
Social norms are formed in group situations and subsequently serve as starting place is the group rather than the individual, changing
standards for the individual’s perception and judgment when he is not
in the group situation. The individual’s major social attitudes are
formed in relation to group norms. (Sherif & Sherif, 1953, pp. 202– 1
This formulation is also a central aspect of a more recent approach:
203) self-categorization theory. As Terry, Hogg, and White (2000) described it,
a “norm is subjectively represented as a group prototype that describes and
According to this theory, attitudes, values, beliefs, and preju- prescribes beliefs, attitudes, feelings, and behaviors that minimize ingroup
dices are all acquired as part of the socialization process; “the differences and maximize outgroup differences . . . the process of self-
attitude of prejudice is a product of group membership” (Sherif, categorization also meant that there is an assimilation of self to the
1948, p. 66). According to Sherif (1936), attitudes, values, and prototype and thus the self is transformed” (p. 72).
SOCIAL NORMS, SUPPRESSION, AND PREJUDICE 361
group attitudes will be more efficient than changing individual ambiguous and do not overtly sanction prejudice, discrimination is
attitudes. Allport (1954) reviewed this research, concluding that significantly more prevalent. In the subtle and blatant prejudice
“in certain studies whole communities, whole housing projects, model (Meertens & Pettigrew, 1997; Pettigrew & Meertens, 1995),
whole factories, or whole school systems have been made the subtle prejudice is a combination of hostile prejudice and social
target of change. . . . New norms are created, and when this is norms that proscribe blatant expressions of prejudice; subtle prej-
accomplished, it is found that individual attitudes tend to conform udice arises from compliance with new egalitarian norms without
to the new group norm” (p. 40; see also Bolton, 1935; Marrow & the internalization of these norms. Regressive racism, aversive
French, 1945). racism, and subtle prejudice theories are all concerned with how
Several classic studies have been based on the demonstration of social norms can prevent expression of prejudice and how, under
the power of social norms to affect the expression of prejudice. ambiguous circumstances, social norms allow for prejudice to be
Minard (1952) found that when miners in the Pocahontas coal field expressed in a masked or “covered” fashion. Some models of
were working below the ground, integration between Black and norms and influence take this even further; in social identity theory
White miners was complete and relatively conflict-free. However, “norms and stereotypes are treated as being conceptually identical”
when the miners were above ground and off the job, segregation in (Hogg & Abrams, 1988, p. 185).
housing and social lives was almost total. Workers easily adopted
highly racist or highly egalitarian norms and moved between the Expanding the Domain of Prejudice
two settings easily—the inconsistency between their behavior in
the two settings was based on the inconsistent social norms be- Much of the research on prejudice has focused on a small
tween the two settings, not in the consistent way in which people number of target groups. Although negative evaluations of social
comfortably follow the prevailing social norms (Pettigrew, 1991). groups are widespread and various, social psychologists have
Pettigrew (1958) showed that racial prejudice among White South primarily studied prejudice on the basis of race, religion, sexual
Africans was due more to conformity to norms than to authoritar- orientation, or generic social categorization. In general, social
ian personality. In a country where discriminative racial policy was psychologists have tended to study target groups for which (a)
encoded in federal law and public policy, the tendency to conform these psychologists believe prejudice exists and (b) these psychol-
was associated more with racial prejudice than with child-rearing ogists personally feel this form of prejudice is inappropriate. Social
patterns or personality type. Pettigrew found much the same effect psychologists rarely study the prejudices they openly share.
in the Jim Crow American south of the 1950s (Pettigrew, 1959). We believe that this is unnecessarily limiting and suggest that
much can be learned about the structure of prejudice by refocusing
Norms and the Expression of Prejudice research on prejudice as a general process (and on social norms in
particular) rather than on studying particular prejudices (which are
Changing the norm about the expression of prejudice can have subject to a variety of historical and normative idiosyncrasies).
a strong effect on people’s tolerance for prejudice. Blanchard, Understanding prejudice requires that psychologists study preju-
Crandall, Brigham, and Vaughan (1994) found that a single con- dices that they abhor but also those that they endorse. One must
federate expressing antiracist views could dramatically reduce presume that a core of psychological states and processes are
tolerance for racist acts among experimental participants. Con- common to a White person’s anti-Black racism and an average
versely, when the same confederate expressed benign acceptance American’s attitude toward members of the American Nazi Party.
of racist acts, participants also recommended acceptance. The The difference between these two prejudices lies in an outside
manipulated social norm affected attitudes when measured pub- perceiver’s sense of the justification of the prejudices but not in the
licly and privately, suggesting that the single confederate effected psychological structure and experience of the persons holding
private acceptance (see Monteith, Deneen, & Tooman, 1996, for these prejudices.
similar effects on gay-related attitudes). The critical difference between these two prejudices is that the
Stangor, Sechrist, and Jost (2001) have shown that merely American Nazi Party is a normatively acceptable target of preju-
changing the apparent consensus about a group is enough to dice and Blacks are not. The justifications of the prejudice against
change prejudiced expression. When Stangor et al. led people to Nazis are widely accepted; the justifications of the prejudice
believe that their view of African Americans was more stereotypic against Blacks are not (for discussions of the justification of
than their peers, their participants reduced endorsement of African prejudice, see Crandall, 2000; Crandall & Eshleman, 2000). Nev-
American stereotypes, and when they were led to believe their ertheless, prejudices against both groups are accompanied by a
view was less stereotypic than their peers, they increased stereo- ready anger, stereotypes that can nullify individuating information,
typing. A high degree of consensus, indicating that the person’s and a willingness to discriminate against group members. The
belief was normatively appropriate, also led to greater resistance to study of how social norms affect the experience and expression of
persuasion. One’s perception of the stereotyping norms directly prejudice can help delineate these similarities and differences
affects willingness to endorse stereotypes (see also Sechrist & among prejudices.
Stangor, 2001).
Social norms are still important to theories of prejudice. In Outline
regressive racism (Rogers & Prentice-Dunn, 1981), egalitarian
social norms proscribe prejudice. By holding Whites’ prejudice In this article, we report the results of a research program on the
toward Blacks in check, prejudice is masked by egalitarian norms social appropriateness of prejudices and how the appropriateness
for appropriate interracial behavior. In aversive racism (Dovidio & determines which prejudices are publicly stated and which are
Gaertner, 1991; Gaertner & Dovidio, 1986), when social norms are suppressed. There are three sections of the article. In the first
362 CRANDALL, ESHLEMAN, AND O’BRIEN
section, a population of 105 prejudices is developed, and we first surveyed the literature and listed many targets of prejudice that had
compare how much of each prejudice people say they have with been used in previous research. In addition, we detailed several undergrad-
how normatively acceptable it is. In the second section, we look at uate students to make a list of groups that were discussed in newspapers,
how people evaluate overt forms of discrimination toward groups magazines, and on television over a 2-week period. Finally, we conducted
a series of discussions with graduate and undergraduate students, who
that differ in the normativeness of prejudice. In the third section,
nominated modern targets of prejudice. From this, we generated a list of
we consider individual differences in attempts to suppress preju-
over 500 potential targets of prejudice, which we winnowed down by
dice and the comparison of internal motivation to suppress preju- removing redundancies, striving to represent as wide as possible a range of
dice with conformity to perceived social norms. In this third prejudices. In a final committee meeting of Christian S. Crandall and Amy
section, we consider the hypothesis that a report of internal moti- Eshleman and three undergraduates, we selected a final list of 105 targets.
vation to suppress prejudice is a sign of the struggle to internalize The list includes people toward whom American social norms proscribe
group norms. prejudice (e.g., racial groups, the physically challenged) and people toward
whom American social norms prescribe prejudice (e.g., rapists, child
I. Measuring Social Norms About Expressing Prejudice abusers), and several groups somewhere in-between. The list of groups is
presented in Table 1.
Study 1: One Hundred Five Prejudices Acceptability of prejudice. With the list of groups in hand, we recruited
150 undergraduates (43% women; 91% White) from the introductory
In our first study, we developed a list of targets of prejudice and psychology pool at the University of Kansas, who received course credit.
ascertained the normative appropriateness for expressing prejudice Participation took place in groups of up to 20.
toward these groups. We then compared the normative appropri- To rate the normative acceptability of prejudice, each of 105 group
ateness for the expression of prejudice with the degree of prejudice names was printed on a small slip of paper. The slips were jumbled
expressed toward these groups. We predicted that the normative randomly and put into a large manila envelope. One manila envelope was
appropriateness of a given prejudice would substantially correlate given to each participant, and participants were told to sort the slips into
with the amount of prejudice people are willing to publicly report three piles, according to the labels on three No. 10 envelopes that were
on a questionnaire. placed in front of the participant. The three business-sized envelopes read
“Definitely OK to have negative feelings about this group,” “Maybe it’s
OK to have negative feelings about this group,” and “Definitely not OK to
Method
have negative feelings about this group.” The categories were later as-
Developing the list. The first step to broad sampling of prejudices was signed values of 2 for definitely OK, 1 for maybe OK, and 0 for definitely
to create a list of groups (see Crandall, Preisler, & Aussprung, 1992). We not OK. The verbal instructions made clear that participants were rating
Table 1
Mean Prejudice Acceptability Score for 105 Potential Prejudice Targets
1.967 Rapists 0.900 Guests on Ricki Lake 0.227 People with AIDS
1.967 Child abusers 0.887 Iraqi soldiers 0.227 High school cheerleaders
1.933 Child molesters 0.824 Politicians 0.220 Cat owners
1.932 Wife beaters 0.764 People who smell bad 0.215 Interracial couples
1.907 Terrorists 0.733 Gamblers 0.207 Auto mechanics
1.840 Racists 0.733 Feminists 0.196 People who put their kids in daycare
1.827 Members of the Ku Klux Klan 0.725 Rednecks 0.193 Ugly people
1.820 Drunk drivers 0.653 People who go to Kansas State University 0.167 People on Medicare
1.793 Members of the American Nazi Party 0.620 Welfare recipients 0.167 Accountants
1.780 Pregnant women who drink alcohol 0.607 Homosexuals who raise children 0.148 Canadians
1.779 Men who refuse to pay child support 0.600 People who smoke 0.147 Trash collectors
1.713 Negligent parents 0.573 Members of National Rifle Association 0.147 Local residents of Lawrence, KS (non-college)
1.644 People who cheat on their spouses 0.560 People who got a job due to Affirmative Action 0.147 Asian Americans
1.540 Kids who steal other kids’ lunch money 0.560 People who call the Psychic Hotline 0.141 Hispanics
1.533 Men who leave their families 0.553 Police officers 0.140 Whites
1.513 Gang members 0.547 IRS agents 0.140 Students who attend community college
1.487 Liars 0.541 British punks 0.133 Doctors
1.336 Male prostitutes 0.520 Gay soldiers 0.128 Catholics
1.307 Men who go to prostitutes 0.510 People with open sores 0.127 Spelling bee champions
1.253 People who cheat on exams 0.460 Lawyers 0.120 Manual laborers
1.235 Female prostitutes 0.453 Students who rarely study 0.120 Jews
1.227 Careless drivers 0.430 Country music fans 0.120 Business women
1.213 Drug users 0.427 Waif fashion models 0.120 Black Americans
1.188 People who litter 0.407 Mentally unstable people 0.114 Native Americans
1.141 People who cut in line 0.407 Hare Krishnas 0.114 Elderly people
1.134 Members of religious cults 0.367 Jehovah’s Witnesses 0.114 Dog owners
1.133 Illegal immigrants 0.356 FBI agents 0.113 Members of a bowling league
1.127 Juvenile delinquents 0.347 Homeless people 0.107 Librarians
1.080 People who sell marijuana 0.329 People who are illiterate 0.100 Male nurses
0.993 Lazy people 0.313 Traveling salesmen 0.093 Farmers
0.980 Ex-convicts 0.313 Environmentalists 0.081 Family men
0.980 Alcoholics 0.275 Rap music fans 0.053 Mentally retarded people
0.967 Porn stars 0.247 Beauty contestants 0.053 Deaf people
0.947 College teachers with poor English skills 0.235 White Southerners 0.047 Women who stay home to raise kids
0.927 Tele-evangelists 0.228 Fat people 0.047 Blind people
Note. The mean ratings reported above are based on the following scale: 2 ⫽ OK to feel negatively toward these people, 1 ⫽ Maybe OK to feel negatively toward these people,
and 0 ⫽ Not OK to feel negatively toward these people.
SOCIAL NORMS, SUPPRESSION, AND PREJUDICE 363
their perceptions of the predominant social norm and not their own specific more nearly normal distribution (␥2 ⫽ .32, p ⬎ .19). In subsequent
attitudes. statistical analyses, we use the square-root transformed version.
Expressions of prejudice. A separate group of 121 University of Kan- To what extent are reports of prejudice a function of perceived
sas undergraduate students was recruited in the same fashion as the first
normative acceptability? To test this, we reversed the feeling ther-
sample (in a later semester). This sample rated their own personal attitudes
mometer rating for each group (so that high scores corresponded to
toward the 105 groups on 0 –100 “feeling thermometer” scales, in 5-point
increments; 0 was labeled cold/not positive, 50 was labeled medium warm/ low liking) and calculated the mean ratings. We then treated the 105
average, and 100 was labeled hot/very positive. Target groups were pre- groups as individual observations and correlated the mean (square-
sented in alphabetical order. root transformed) normative acceptability scores with the mean feel-
ing thermometer ratings (with N ⫽ 105). The results are displayed in
Figure 1; the two variables correlate at r ⫽ .96.
Results and Discussion This is a high correlation. When two different scales correlate so
Table 1 is arranged in the descending order of the normative highly, it suggests that both scales measure the same thing. This is
acceptability of prejudice. Although Table 1 has some amusing usually understood as a flaw in methodology. However, in this
juxtapositions of particular prejudices, the groups at the extreme case, it is almost exactly the argument: People will report their own
ends of the normative acceptability of prejudice are mostly not prejudice according to how much it is socially acceptable. We
surprising; prejudice against rapists, child abusers, and thieves is hypothesize that expressed prejudice is a direct function of its
quite acceptable, whereas prejudice against racial groups and the social acceptability, and when collapsing across individuals, the fit
physically challenged is not. between how much people report and how much people feel is
The list in Table 1 ranges from near complete agreement that it appropriate to have is nearly perfect. Because ratings are collapsed
is OK to have negative feelings (e.g., toward rapists) to near over individuals, we have highly reliable estimates of prejudice
complete agreement that negative feelings are not acceptable (e.g., and normativeness, without any need to correct for correlated error
toward blind people). Between these extremes, the groups ranged or biases between samples or ratings.
across the entire spectrum of acceptability. These data are purely correlational, and we cannot determine the
The mean ratings collapsed across participants on the feeling causal ordering of social norms and prejudice. Nevertheless, the fit
thermometer were normally distributed, but the normative accept- between the two is so close that it is very difficult to imagine that
ability distribution was positively skewed (␥2 ⫽ .77, p ⬍ .001). A the two do not stem from the same source or that one (norms)
square-root transformation of the acceptability ratings resulted in a causes the other (reports of prejudice).
II. The Pervasiveness of Social Norms in Controlling Following each scenario were two items; the first inquired about the
Prejudice Expression personal acceptability of the act of discrimination, the second asked how
likely the participant would be to discriminate against the target in the same
Study 2: The Acceptability of Discrimination way. Response options ranged from 1 (very unacceptable or extremely
unlikely) to 7 (very acceptable or extremely likely).
If the high correlation between reports of prejudice and norms
about their expression is caused by social norms determining
prejudice reports, then the social norms should determine which Results and Discussion
kinds of discrimination are considered acceptable and which kinds
The hypothesis that the acceptability of discrimination would be
elicit a negative reaction. In Study 2, we predicted that the accept-
ability of discrimination would be closely related to the accept- predicted by the acceptability of prejudice was strongly supported
ability of prejudice. We examined the acceptability of discrimi- (see Table 2). To calculate how closely the acceptability of dis-
nating against a number of social groups in the areas of dating, crimination fit the normative appropriateness scores from Table 1,
housing, and employment. we collapsed across participants and calculated the correlation
between the (square-root transformed) scores in Table 1 and the
average acceptability scores displayed in Table 2. The correlations
Method
are displayed in the last row of the table. For each domain, the
Participants were 104 introductory psychology students (73% women, acceptability of discrimination was substantially and significantly
all born in the United States) from the University of Kansas, who received correlated with the normative appropriateness of prejudice.
partial course credit for participation. Participation occurred in groups of In the final column, the three domains are combined into a
up to seven. single “prejudice acceptability” measure; across the 10 target
Participants each read a series of three discrimination scenarios. The first groups, the acceptability of prejudice predicted the acceptability of
scenario depicted discrimination in the area of dating (Jamie refused to date
discrimination with r ⫽ .86. This correlation is somewhat lower
Chris because of Chris’ group membership). The second scenario depicted
discrimination in the area of housing (an apartment manager tells Terry that than .96, perhaps because we estimated the correlation with
an apartment is not available to rent because of Terry’s group member- only 10 groups rather than with the 105 groups in Study 1.
ship). The third scenario depicted discrimination in the area of employment Discrimination is acceptable to the extent that prejudice is; if it’s
(a business manager fails to consider Pat for a job because of Pat’s group OK to express prejudice against a group, then it’s OK to discrim-
membership). Each of the three scenarios presented a different, randomly inate against it.
selected target of discrimination. Ten target groups, representing different Discrimination was rated as more acceptable in dating than in
levels of acceptability of prejudice, were used across the three scenarios. In housing or employment (both p’s ⬍ .005), consistent with the
order of acceptability, these groups were racists, drug users, ex-convicts, finding that prejudice is more tolerated in intimate relationships
rednecks, welfare recipients, environmentalists, fat people, Hispanics,
(Anant, 1975; Pettigrew & Meertens, 1995; Triandis & Davis,
Black Americans, and Native Americans.
Each group was paired with each scenario across participants; and each
1965). In this study, as in the others, we cannot be sure what the
participant read scenarios that described discrimination against three of the source of the social norm is. The norms of rejection may be rooted
groups (one in dating, a second in housing, a third in employment). The in safety concerns, value threat, and social history, among many
scenarios were written to be ambiguous with regard to sex of the charac- others. Whatever their source, norms can be independently mea-
ters. (See Appendix A for complete text of scenarios.) sured, and their effects are detectable and considerable.
Table 2
Discrimination Acceptability Ratings for 10 Potential Prejudice Targets
Acceptability of discrimination
Note. Acceptability of prejudice scores are from Study 1. The acceptability of discrimination scores are the
mean scores across participants. The overall acceptability of discrimination score equals the mean acceptability
of discrimination over dating, housing, and employment.
* p ⬍ .025. ** p ⬍ .005.
SOCIAL NORMS, SUPPRESSION, AND PREJUDICE 365
Study 3: The Acceptability of Prejudiced-Based Humor funniest targets was uncorrelated with the normative appropriate-
ness of expressing prejudice (r ⫽ ⫺.12, N ⫽ 10, ns). Thus, we did
Jokes are a way of expressing prejudice and hostility. Several not find support for the second hypothesis. A significant effect of
studies have looked at the expression of prejudice through humor joke on funniness, F(9, 2097) ⫽ 24.69, p ⬍ .001, indicated that
and found that jokes are an indirect way of establishing hierarchy
some jokes were funnier than others.
(e.g., Burma, 1946; Cantor, 1976; La Fave & Mannell, 1976), in
Our third hypothesis was that offensiveness would make a joke
part by establishing local social norms that favor prejudice and
funnier when prejudice was acceptable to express, but that it would
discrimination (e.g., LaFrance & Woodzicka, 1998). We predicted
make a joke less funny when that prejudice was unacceptable. To
that people’s reactions to jokes that denigrate groups would reflect
test this, we calculated a correlation between the offensiveness and
the social norms measured in Table 1.
funniness ratings (collapsed across jokes) for the 10 target groups.
We developed three specific hypotheses. First, we predicted that
people would take offense at jokes to the extent that it is norma- These correlations were then treated as a new variable, which was
tively inappropriate to express prejudice toward the joke’s target in turn correlated with the Table 1 scores for each group (see
group. Second, we predicted that jokes that denigrate acceptable Figure 3). Figure 3 shows the support for the third hypothesis (r ⫽
targets of prejudice would be considered funnier than jokes that .87, N ⫽ 10, p ⬍ .001). When it is normatively appropriate to
denigrate inappropriate targets of prejudice. Third, we suggested express prejudice against a group, offensiveness and funniness are
that if the offensiveness of a joke is a way of expressing hostility, positively correlated (e.g., racists; r ⫽ .23, N ⫽ 239, p ⬍ .001).
then when the expression of prejudice is normatively appropriate, When it is not normatively appropriate to express prejudice, of-
offensiveness will contribute to the funniness of a joke. However, fensiveness and funniness were negatively correlated (e.g., men-
when the expression of prejudice is normatively condemned, of- tally retarded people; r ⫽ ⫺.17, N ⫽ 239, p ⬍ .01).
fensiveness will detract from the humor inherent in the joke. This experiment shows that the expression of prejudice and
hostility through humor follows normative lines. Although jokes
Method may serve as indirect expressions of prejudice, they are tolerated
only to the extent that the prejudice they express is normatively
Participants were 239 introductory psychology students (57% women, appropriate; when norms suppress expression of prejudice against
79% White) from the University of Kansas, who each read a selection of 10 a group, jokes were considered offensive.
different jokes, each joke targeting 1 of 10 different groups. After reading The funniness of jokes against targets was uncorrelated with
each joke, participants rated it on two qualities— offensiveness and hu-
normative appropriateness, which is inconsistent with our second
mor— on scales ranging from 1 (not at all offensive or not at all funny) to 7
(very offensive or very funny). Each person read all 10 jokes, and each hypothesis. It may be that our hypothesis is simply wrong, or it
group was paired with only 1 joke. may be that, even when the joke’s target should not be openly
As in Study 2, the 10 groups were selected to cover a range of normative derogated, the release of suppressed prejudice makes the joke
acceptability of prejudice. The groups chosen, in order of acceptability, funnier (see O’Brien, Eshleman, & Crandall, 1999), and these two
were racists, drug users, ex-convicts, rednecks, welfare recipients, people effects cancel each other out. Or it may be that the judgment of
who are illiterate, fat people, Hispanics, Black people, and people who are humor may bypass some of the conscious controls that determine
mentally retarded.
open reports of prejudice. At this point, one cannot be sure why the
The 10 jokes were selected from the on-line archives of the Usenet
funniness of the joke and social norms are empirically unrelated.
newsgroup, rec.humor.funny, and needed to be generic enough to apply to
the different targets. (Appendix B has the joke texts.) Supporting our third hypothesis, we found that offensiveness
Each of the 10 jokes was matched with each of the 10 target groups in 1 potentiates the normative effects. When the joke targets a norma-
of 10 different combinations. The order of the jokes was maintained across tively appropriate group, offensiveness enhances the humor. When
all participants, and the order of target groups varied across forms. the joke’s target is normatively inappropriate, offensiveness ne-
gates its humor.
Results and Discussion This experiment, and the two that precede it, suggest that the
expression of prejudice and the tolerance of discrimination closely
Our first hypothesis was that offensiveness would vary by group follow normative strictures. It is not in any way surprising that we
in accordance with Table 1 normativeness scores.2 On offensive- found the correlations we did. Not only are the findings common
ness, there was a significant main effect of target group, F(9, sensical and intuitive, but many theories predict the finding (e.g.,
2097) ⫽ 57.76, p ⬍ .0001. As predicted, the offensiveness of a Newcomb, 1943; Sherif, 1936; Turner, 1991).
joke targeting a specific group was closely related to Table 1 However, what is remarkable is the very high level of fit
scores, with r ⫽ ⫺.82 (N ⫽ 10, p ⬍ .01; see Figure 2). The less
between social norms and the various expressions of, and tolerance
normatively appropriate it was to express prejudice, the more
for, prejudice. In most of the cases demonstrated above, the fit of
offensive the joke was rated. There was also a significant effect of
individual expression–tolerance for discrimination to social norms
joke, F(9, 2097) ⫽ 14.49, p ⬍ .0001; some jokes (collapsed across
is quite close, ranging from r ⫽ .96 for prejudice and norms, to r ⫽
target groups) were rated more offensive than others (an unsur-
.86 between acceptance of discrimination and norms, to r ⫽ .82
prising result).
Our second hypothesis was that the funniness of the joke would
be affected by the normative appropriateness of expressing preju- 2
Because of the structure of the questionnaire, with only 10 different
dice. There was a significant main effect of target group, F(9, forms, we can analyze the effect of joke and the effect of target in separate
2097) ⫽ 6.60, p ⬍ .001—jokes are funnier when used against orthogonal one-way repeated measures analyses of variance (ANOVAs),
some groups than against others. However, which groups were the but we cannot calculate an interaction term.
366 CRANDALL, ESHLEMAN, AND O’BRIEN
between offensiveness of humor and norms. These appear to be However, in the case of prejudices, social norms are constantly
pervasive and strong effects. changing, and the individual who wishes to be in sympathy with
his or her reference group’s norms must be vigilant for changes in
III. Compliance and Internalization of Social Norms who is an acceptable target of prejudice, and who is not. Sherif
About Prejudice (1936) suggested that during times of social change and social
mobility
Knowledge about norms must be acquired. Social norms about
the expression of prejudice vary from group to group, and as a the strength of norms incorporated in the individuals becomes uncer-
person moves from one group to another, identification with the tain and liable to break down. Such a delicate, unstable situation is the
new group involves taking on that group’s frame of reference fertile soil for the rise of doubts concerning the existing norms, and a
(Cantril, 1941; Sherif & Sherif, 1964). There is a period during challenge to their authority. . . . The transition is not simply from the
which people learn what the norms are and begin the process of orderliness of one set of norms to chaos, but from one set of norms to
a new set of norms, perhaps through a stage of uncertainty, confu-
adaptation and internalization (Kelman, 1958, 1961). This process
sion . . . as a result of the strain and stress, of the confusion and
can be challenging, and may require some amount of cognitive and uncertainty and feelings of insecurity, apparent stability followed by
emotional work. The period of adaptation and identification is a fresh instability. The outcome is the final emergence or establishment
period when the norms are salient (Hogg & Abrams, 1988), and of a stable set of norms having the status of standards . . . this new set
people are aware of their deviance (Turner, 1991). of norms supplants the old ones and becomes the regulator of social
The first three studies in this article present conformity to norms life. (p. 86)
as a fait accompli. And, of course, the prevailing norm at any given
moment tends to represent the majority’s view. As Sherif and What is the process by which people adapt themselves to social
Sherif (1953) pointed out, “When we speak of a social-distance norms? In the case of prejudice or attitudes of any sort, the initial
scale or of prejudice over a time span, it can be seen that the response will be hesitancy, characterized by the suppression of
classifications and definitions which come to prevail are those of potentially inappropriate responses. This first suppression of prej-
the majority, and not those of minority groups” (p. 115). udice is externally motivated; it represents compliance (Kelman,
SOCIAL NORMS, SUPPRESSION, AND PREJUDICE 367
1958). However, when a person begins to identify with the group Suppression, Identification, and the Expression
and take on its norms, compliance turns to identification and of Prejudice
internalization.
We suggest that people’s adaptation to norms—from compli- Previous researchers have suggested that the suppression of
ance to identification and internalization—is not always accu- prejudice is motivated by egalitarian or humanitarian motives (e.g.,
rately reported. Identification and internalization are often sub- Fazio, Jackson, Dunton, & Williams, 1995; Katz & Hass, 1988).
This is particularly true of the suppression of prejudice that is
jectively experienced as individual processes of self-motivated
“internally motivated” (e.g., Monteith, Sherman, & Devine, 1998;
change, at least by members of individualistic societies (see
Plant & Devine, 1998). There is clear evidence that suppression of
Kim, Triandis, Kagitcibasi, Choi, & Yoon, 1994). As people
prejudice is motivated, and that failures to suppress when desired
feel pressure to adapt to norms that they wish to resist, they may
can lead to guilt and shame (Monteith, 1993, 1996; Monteith,
accurately report external pressure to change. However, when
Devine, & Zuwerink, 1993; Plant & Devine, 1998).
group identification and self-categorization lead to a feeling of
However, we suggest that the report of motivation to suppress
attraction to group norms (or to group prototypes; Turner, prejudice is not always motivated by some inner hunger for justice
Hogg, Oakes, Reicher, & Wetherell, 1987), people will report a or equality. Rather, we hypothesize that the suppression of preju-
willingness and desire to change, that is, when identification dice is often motivated by an attempt to conform to perceived
with a new group has begun, people will report an internal social norms regarding the appropriateness of expressing preju-
motivation to change. dice. As cultural norms become increasingly negative toward
The next set of studies focus on the process by which people straightforward prejudices toward ethnic, racial, and religious
come to follow the prejudice norms of their social groups. The groups (Dowden & Robinson, 1993), and as people mature, they
transition from conformity to acceptance, from suppression to become motivated and skilled at suppressing inappropriate forms
internationalization, requires substantial motivation, and we hy- of prejudice. We argue that when people report the motivation to
pothesize that people who report an internal motivation to suppress suppress prejudice, whether internal or external, what they are
their prejudices are in the midst of a struggle to internalize the primarily reporting is their awareness of pressures to change their
values and norms of the groups with which they identify. attitudes to fit a prevailing group norm.
368 CRANDALL, ESHLEMAN, AND O’BRIEN
There are two well-known and previously published scales ization begins to succeed, external pressure to conform will no
measuring the suppression of prejudice, from two different re- longer be salient, and so reports of external motivation will shrink,
search labs. These scales, published by Dunton and Fazio (1997) leaving only a high report of internal motivation. Ultimately,
and by Plant and Devine (1998) are both two-scale inventories that should the student once again fully obtain an internal representa-
measure motivation to suppress prejudice. The first scale in Dun- tion of group norms, very low reports of internal and external
ton and Fazio’s inventory measures motivation to suppress preju- motivation would be reported. (Of course, full internal represen-
dice; the second scale measures a desire to avoid open conflict. tation of group norms is only possible when these norms are
The first scale in Plant and Devine’s inventory measures internal themselves stable and unambivalent and when a person does not
motivation to suppress prejudice; the second scale measures a move among groups—a state very unlikely to occur in our
sense of pressure to conform to external expectations. Both scales example.)
have proved useful as predictors of how people will respond in In the next four studies, we consider this approach. First, we test
prejudice-relevant settings (e.g., Fazio et al., 1995; Plant & De- the notion that the quality of identification with the college peer
vine, 1998). Despite their face validity and usefulness in research, group is associated with different kinds of suppression of prejudice
in the next set of studies, we suggest another possible meaning for toward Blacks (Study 4). We then report our development of
the measured constructs. another internally motivated measure of suppression (Study 5), and
then use it along with Dunton and Fazio’s (1997) measure to
The Natural History of Prejudice Suppression investigate whether the report of the suppression of prejudice is
associated with (a) the overall levels of prejudice people report and
In his book, The Moral Judgment of the Child, Piaget (1965) (b) conforming to social norms about which prejudices are con-
argued that the development of moral reasoning follows two demned, and which are condoned (Study 6). Finally, in Study 7,
stages. First, the child’s moral reasoning is heteronomous; it is we create a situation that pits the internal motivation to suppress
“subject to another’s authority,” from outside or above. Moral prejudice directly against conformity to the group, and we predict
reasoning at this stage is characterized by a sense that external that, when the norm tolerates prejudice, people who score high in
rules are sacred, and if something brings about punishment, then internal motivation to suppress prejudice will respond with greater
it’s wrong. The moral rules may appear arbitrary, but they have the tolerance for discriminatory acts than will those who score low in
power of punishment and reward. Later moral reasoning is auton- internal motivation.
omous; it is “subject to one’s own law,” when the child internalizes
fundamental basic moral norms and makes them her or his own. Study 4: Identification With College and the Suppression
This process of internalization is accompanied by cognitive effort,
of Prejudice
and it requires cognitive and social abilities such as empathy,
perspective-taking, and logical and counterfactual reasoning. To test the idea that identification with college is associated with
The logic of Piaget’s (1965) argument can apply directly to reporting internally motivated attempts to suppress prejudice, we
acquisition of morally tinged norms, such as those that control the carried out a study in which students filled out the Plant and
expression of prejudice. When an individual first encounters moral Devine (1998) scales (which measure suppression of prejudice
norms, they appear arbitrary, and subject to the caprices of influ- toward Blacks), along with a questionnaire that we developed to
ential group leaders. But identification with the group leads to an measure identification with college. We predicted that (a) scores
increased desirability of the group’s point of view (Turner, 1991; on the Internal Motivation Scale (IMS) will increase as identifi-
Turner et al., 1987). Identification with the group begins the cation with college increases. However, this was qualified by the
processes of internalization. prediction that, (b) as identification proceeds to its highest level
We use Plant and Devine’s (1998) two-dimensional model (of and discrepancies between the group norm and individual attitudes
internal and external motivations to suppress prejudice) as a way diminish, the salience of self-motivated change will shrink, stu-
of illuminating the connections between Piaget’s (1965) model and dents will become well-adjusted to the new group’s norms, and so
that of the norms and influence models of Sherif and Sherif (1953; scores on the IMS will begin to decrease.
Sherif, 1936) and Kelman (1958, 1961). We suggest that, in the The perception of external pressure to suppress prejudice
earliest stages, a person will report very low levels of internal or against Blacks, without internalization of that attitude, should lead
external motivation to suppress, as they will be successfully to high scores on the External Motivation Scale (EMS). During the
adapted to their current groups (e.g., family, neighborhood, or very first days or weeks of being at college, one should expect low
school norms). Then, this person will encounter a group with EMS scores, which begin to rise quite soon, when students dis-
norms that do not tolerate a particular prejudice, such as moving cover their deviation from the college attitude norms. However,
from a rural homogeneous high school to a heterogeneous univer- because we carried out our identification study in the spring
sity with an antiracial prejudice climate. At first, the external semester, we assumed that the vast majority of our students were
sanctions for deviance will lead to a perception of pressure, and beyond this earliest stage. Thus, we predicted that (c) low levels of
this will be reflected in high external motivation to suppress college identification would be associated with high scores on the
prejudice. The perception of external pressure will occur only if EMS.
the new group is seen as attractive, and high external pressure
should be perceived in the early stages of identification. As the
Method
student begins to identify with the university, and his or her friends
within it, an elevated internal motivation will also grow, leading to Participants were 251 White American students (56% women) in a
a high internal and external motivation to suppress. As internal- psychology class, who were given credit for participation. Most partici-
SOCIAL NORMS, SUPPRESSION, AND PREJUDICE 369
pants were in their first year at the University of Kansas (88%); 81% of .21, p ⬍ .05), which was significant (Z ⫽ 2.59, p ⬍ .01). Because
participants were in their second semester. Because of the race-specific males scored lower on college identification and EMS, these data
nature of the IMS–EMS, we selected only participants who indicated are consistent with the view that men are in a fairly early stage of
“White” on an ethnicity question (80% of the original sample). adoption of the college’s antiprejudice norms.
A short questionnaire was passed out in class, one page of which
Because women are more “advanced” in their identification
contained questions about college identification and one page of which
with college, we expected a curvilinear relationship between IMS
contained a measure of suppression. The participants were given about 7
min to complete the questionnaire. The order of the questionnaires was and identification: Women at the high end of identification will
counterbalanced. report lower self-perceived motivation to change their attitudes. To
To measure suppression, we included Plant and Devine’s (1998) IMS test this, we regressed IMS on both a linear and quadratic college
and EMS scales, measured on 0 –9 scales. To measure college identifica- identification term. Both the linear term ( ⫽ 2.36, p ⬍ .01) and
tion, we asked a series of questions about identification with college, a the quadratic term ( ⫽ 2.21, p ⬍ .02) were significant; the pattern
sense that it has had an impact on them, and the sense that they identify of results is displayed in Figure 4, F(2, 134) ⫽ 4.86, p ⬍ .009,
more with their college selves than with their high school selves. The items Multiple R ⫽ .26. These results conform quite nicely to Hypoth-
had a 1–7 scale, and are displayed in Table 3. The six-item scale had an esis 2, that at the highest levels of group identification, internal-
␣ ⫽ .52, a principal components analysis showed that a single factor
ization be associated with lower IMS scores. A similar regression
accounted for 40.4% of the total variance.
calculated for men showed the linear effect ( ⫽ .24, p ⬍ .02), but
no quadratic effect ( p ⬎ .50), consistent with the interpretation of
Results men’s lower college identification scores indicating substantially
To score the College Identification Scale, items were reversed less internalization.
appropriately and averaged. The mean score was 5.57 College identification increased with length of tenure at the
(SD ⫽ 0.81). Women reported a significantly higher level of university (r ⫽ .21, N ⫽ 250, p ⬍ .001). Although IMS was not
college identification (M ⫽ 5.71) than did men (M ⫽ 5.36), correlated with tenure (r ⫽ ⫺.01, ns), EMS was significantly
t(249) ⫽ 3.54, p ⬍ .001, d ⫽ 44. This was not based on their negatively correlated with tenure (r ⫽ .16, N ⫽ 250, p ⬍ .01; r ⫽
different amounts of time at the university, t(160) ⫽ 1.16, p ⬎ .25. ⫺.19 for men, r ⫽ ⫺.14 for women). These data suggest that as
The average score on the IMS was 7.12 (SD ⫽ 1.58), and on the length of stay at the university increases, the perception of external
EMS it was 5.15 (SD ⫽ 1.64). Gender differences also emerged on pressure to conform on racial issues diminishes. Counterbalancing
these scales, with women scoring higher than men on both the IMS order of scales had no significant effect.
(M ⫽ 7.58 for women; M ⫽ 6.54 for men), t(246) ⫽ 5.51, p ⬍
.001, d ⫽ .66, and the EMS (M ⫽ 5.34 for women; M ⫽ 4.92 for Discussion
men), t(244) ⫽ 2.01, p ⬍ .05, d ⫽ .26. The sex difference was
more than twice as large for internal motivation as for external We found that internally motivated suppression was higher for
motivation, and the interaction was significant, F(1, 240) ⫽ 5.69, women than for men (see also Plant & Devine, 1998) and that
p ⬍ .02. Because of the gender difference on both scales, which identification with college as a reference group was also higher for
indicate that women are more identified with college and thus women. We suggest that it is the identification with a group that
should show a different pattern of suppression from the less- has a high antiracist norm that accounts for this covariation;
identified males, we analyzed the results both together and sepa- women were better identified, and they conformed better to the
rately by sex. prevailing norm.
Overall, college identification was moderately correlated with Furthermore, women showed the predicted curvilinear relation-
IMS (r ⫽ .25, p ⬍ .001). The correlation was slightly higher for ship between identification and internal motivation. We suggest
men (r ⫽ .24, N ⫽ 104, p ⬍ .02) than for women (r ⫽ .16, N ⫽ that the drop-off in IMS as identification reaches its highest levels
135, p ⬍ .06). These results support Hypothesis 1, although the suggests that these women feel less internal motivation to change,
relationship does not appear to be large. because they are well-adjusted to the group’s norms. The fact that
EMS was not directly correlated with college identification (r ⫽ IMS does not drop down to zero suggests that this process is not
.06), but an interesting sex difference emerged. For women, EMS complete (an eventuality we doubt is possible among the typical
was slightly negatively correlated with college identification (r ⫽ college sophomore).
⫺.12, p ⬍ .15), but for men, EMS was positively correlated (r ⫽ Men, however, showed only a linear relationship between iden-
tification and internally motivated suppression. We suggest that
this is because our cross-sectional method provides only a snap-
Table 3 shot of the group in time—not enough of the men identified
The College Identification Scale sufficiently with the group to provide the curvilinear relationship.
This picture of the men as “less advanced” is supported by the fact
1. I tend to identify more with my friends from high school than I do that identification was correlated with the report of external mo-
with people at KU. (R)
2. The best times of my life were in high school, not while in college. tivation for men, suggesting that they, as a group, are in transition
(R) from seeing external pressures to conform to the antiracist norm of
3. I’ve found that college is not really for me. (R) the group as undue and unwelcome pressure from the outside to
4. I expect that after 4 years, KU will have changed me. adopting the group norm as their own.
5. I didn’t want to come to college. (R)
6. College has been a time of change for me.
Overall, the subtle pattern of relationships among the variables
is consistent with all three hypotheses. Motives to suppress prej-
Note. R indicates item is reverse-scored. KU ⫽ University of Kansas. udice reflect the natural history of group identification. External
370 CRANDALL, ESHLEMAN, AND O’BRIEN
Figure 4. Intrinsic motivation to suppress prejudice as a function of identification with college. IMS ⫽ Internal
Motivation Scale.
motivation is associated with the early stages of adaptation to a that applies more broadly could prove quite useful. Although the
group and predates internal motivation to conform to group atti- Dunton and Fazio (1997) scale is not specific as to which prejudice
tudinal norms. Internal motivation to suppress prejudice is associ- is being suppressed (3 of the 17 items that do mention a target
ated with a later, self-motivated transformation of attitudes to fall group mention “Black[/s]”), the mixing of internal and external
in line with group norms. Our data suggest that women at the motivation in their Concern scale (see Plant & Devine, 1998)
University of Kansas have gone further in this process of adapta- suggests that there is still room for a nonspecific scale measuring
tion than have comparable men. By contrast, men in their second the perception of internal motivation to suppress prejudice. This is
semester of college are less identified as college students and not the focus of Study 5.
highly conforming to the antiracist norms of that group.
Although they report lower levels of internal and external mo-
Study 5: The Suppression of Prejudice Scale
tivation, the gender difference is more than twice as large for
internal as for external motivation, suggesting that they lag behind We developed a scale measuring individual differences on in-
women in the natural history of their identification with college ternally motivated inhibition. We wanted to tap individual differ-
peers and their values, and consequently the suppression of prej- ences in the desire to appear nonprejudiced, primarily to the self,
udice. Still, one must keep in mind that the phenomena we are and to a lesser extent to others as well. Our work on the scale
theorizing are developmental and longitudinal, these data are predated our learning of similar scales by Dunton and Fazio (1997)
cross-sectional. and Plant and Devine (1998), and so we did not sample from their
One limiting factor of the Plant and Devine (1998) scale is that initial item pool. This scale was designed to account for a desire to
it is specific to African American targets. A more general scale suppress prejudice against multiple outgroups.
might be useful to measure responses to other groups whose
prejudice varies with specific group norms, such as homosexuals,
political or religious minorities, and so on. If the suppression of
Method
many different prejudices results from shifting group norms and Four hundred fifty-five introductory psychology students (52% women,
identification with new groups and subsequent internalization, then 88% White) from the University of Kansas received course credit for
creating a scale to measure an internally motivated suppression completing our Suppression of Prejudice Scale (SPS; response options
SOCIAL NORMS, SUPPRESSION, AND PREJUDICE 371
ranged from 0 to 9); 288 of these participants (56% women, 90% White) Study 6: Suppression and the Normative Appropriateness
also completed McConahay, Hardee, and Batts’ (1981) Modern Racism of Prejudices
Scale (response options ranged from 1 to 5). Participation occurred in large
groups. What exactly do suppression scales measure? Certainly, people
who score high on these measures of suppression report less
prejudice toward traditionally studied targets of prejudice. Plant
Results and Discussion and Devine (1998) reported correlations with the IMS (internal
motivation) and the Modern Racism Scale (McConahay et al.,
Seven of the original 15 items loaded substantially onto a single, 1981) of ⫺.57, with the Anti-Black Scale (Katz & Hass, 1988) of
theoretically meaningful factor (eigenvalue ⫽ 2.64) accounting for ⫺.48, and with the Attitudes Toward Blacks Scale (Brigham,
38% of the variance, ␣ ⫽ .64 for the SPS (see Table 4). The item 1993, with higher scores representing more positive attitudes) of
content suggests that the SPS has face validity as a measure of .79. Dunton and Fazio (1997) observed a correlation between their
interest in avoiding appearing prejudiced to oneself or others. The Suppression scale and the Modern Racism Scale of ⫺.18. When
scale has a mean of 4.35 (SD ⫽ 1.00). On average, women we expand our study of prejudice to include the 105 social groups
(M ⫽ 4.54) reported more suppression than men (M ⫽ 4.10), examined in Study 1, will measures of the suppression of prejudice
t(276) ⫽ 3.73, p ⬍ .001, d ⫽ .44. The SPS correlated with the represent the constructs described by Dunton and Fazio (1997) and
Modern Racism Scale at r ⫽ ⫺.35 (N ⫽ 282, p ⬍ .001). Plant and Devine (1998)?
In an initial validity test, we examined the SPS’s relationships If high suppressors are people who suppress negative thoughts,
with the Modern Racism Scale and gender. Men routinely score feelings, and communications toward all outgroups, then we would
higher on the Modern Racism Scale than women (Biernat & expect Suppression scales to be negatively correlated with all 105
Crandall, 1999); in this case, Ms ⫽ 2.11 for men and 1.93 for outgroups. However, if the internal measures of generalized sup-
women, t(284) ⫽ 2.24, p ⬍ .03, d ⫽ .26. However, as we have pression are actually indicators of a struggle for the internalization
seen, men suppress less often than women. To what extent is the of group norms, then we would predict that suppressors will report
difference in racism merely a function of difference in suppres- lower levels of normatively inappropriate prejudice (e.g., toward
sion? To test this, we calculated an analysis of covariance racial, ethnic, and religious groups) but would report higher levels
(ANCOVA), comparing Modern Racism Scale scores and cova- of prejudice toward socially appropriate targets (criminals, social
deviants, people whose behavior is politically incorrect). If high
rying out SPS scores. SPS was a significant covariate, F(1,
suppression scores are an outward reflection of the inward struggle
275) ⫽ 31.07, p ⬍ .001, and the gender difference was no longer
to adapt to new social norms, then high suppression scores should
significant, F(1, 275) ⫽ 1.61, p ⬎ .20, d ⫽ .15. This suggests that
be associated with high levels of socially appropriate prejudice.
a significant portion of gender differences in racial prejudice may
We used both the SPS and the Dunton and Fazio (1997) Con-
be attributed to a difference in the amount of suppression of cern scale to measure suppression, which provides another oppor-
prejudice between the two genders, and not because of a difference tunity to test the validity of the SPS. Because we know from Study
in the genuine underlying attitudes. 1 what the normatively appropriate level of prejudice expression is
The SPS has face validity, modest internal consistency, and for the 105 target groups, we can also compare the fit-to-prejudice
explains a portion of the gender differences in reported racial norms for high and low suppressors.
prejudice. It is short, straightforward, and seems to measure what
it purports to measure. We further establish its validity in subse-
Method
quent studies in this article. In Studies 6 and 7, we return to the
hypothesis that suppression scales measure people’s concerns with Participants were 127 introductory psychology students (50% women,
the pressures to conform to group norms. 85% White) from the University of Kansas, who received partial course
credit for participation. Participation occurred in groups of up to seven.
Participants completed the feeling thermometer measure of prejudice of
105 groups (described in Study 1), an unrelated filler task, the SPS, and
Dunton and Fazio’s (1997) Concern scale (response options ranged from 1
Table 4 to 7 on the SPS and Concern scale).
The Suppression of Prejudice Scale (SPS)
positive correlation between measured suppression and a prejudice people who report high levels of internal motivation are people
when that prejudice is normatively appropriate. The feeling ther- who are very concerned with reporting attitudes that are consistent
mometer was once again reversed, so that higher scores represent with, or even extreme versions of, the prevailing group norm.
greater prejudice. Table 5 displays the pattern of correlations Comparing SPS and Concern scales. To compare the power of
(these correlations represent a sampling from the first and last the SPS and Concern scales for predicting prejudice reports, we
thirds of targets from Table 1). correlated both with reported prejudice and examined the relative
For traditional, normatively inappropriate targets of prejudice magnitude of the two sets of 105 correlations. We discarded the 28
(displayed in the right-hand half of Table 5), suppression is neg- relatively trivial correlations, when both suppression scales corre-
atively correlated with expressions of prejudice. However, a very lated less than ⫾.10 with the feeling thermometer score (see
different picture emerges from the left-hand half of Table 5. When Levine, 1977). Of the remaining 77 groups, the correlation be-
it is normatively appropriate to express a prejudice, we found that tween SPS and prejudice exceeded the correlation with the Con-
high suppressors report more prejudice. These data are substan- cern scale for 60 of them; the Concern scale exceeded the SPS 17
tially consistent with the internalization hypothesis; suppressors times (binomial p ⬍ .005).
are acute norm followers.
To create an overall statistical test of this hypothesis, we cor- Discussion
related the normative appropriateness scores from Table 1 (square-
root transformed) for each prejudice target, with that target’s When it is unacceptable to express prejudice, high suppressors
correlation between suppression and prejudice measured by feel- express less prejudice. But when is it acceptable to express prej-
ing thermometer. udice, high suppressors express more prejudice. High suppressors
A positive correlation would indicate that the more normatively are acute norm followers in expressing prejudice.
appropriate a prejudice is, the more high suppressors endorse it. Measures of motivated suppression correspond with adherence
When a prejudice is normatively inappropriate, high suppressors to social norms, and we suggest that high suppressors of prejudice
report less prejudice, when it’s normatively appropriate, high are hypersensitive to perceived cultural norms. This study suggests
suppressors report more prejudice. The correlation provides an that people who score high on suppression scales are not people
index of the extent to which people who score high in motivated who attempt to suppress all kinds of prejudice. Instead, suppres-
suppression are “super-conformers.” By contrast, if high suppres- sors are people who enhance appropriate prejudices and attenuate
sors are people who suppress all sorts of prejudice, it would not inappropriate prejudices. Dispositional suppression scales appear
vary across normative appropriateness, and one would expect no to measure how closely an individual follows social norms for
correlation at all. prejudice expression rather than measuring a motivation to avoid
For the 105 targets, the fit between normativeness of the prej- or deny all forms of prejudice. The motivation to suppress preju-
udice and the effect that suppression has on it is very good for SPS dice may only be a motivation to suppress certain, normatively
(r ⫽ .69) and for the Concern scale (r ⫽ .62, N ⫽ 105, both ps ⬍ unacceptable forms of prejudice. When only normatively inappro-
.001). These data are quite consistent with the hypothesis that priate prejudices are studied, suppression appears to be no different
Table 5
Selected Correlations Between Two Suppression Scales and Expression of Prejudice Among Normatively Acceptable and
Unacceptable Prejudices
Note. SPS ⫽ Suppression of Prejudice Scale; Concern ⫽ Concern scale from Dunton and Fazio (1997).
SOCIAL NORMS, SUPPRESSION, AND PREJUDICE 373
from norm following. It is only when socially appropriate preju- In many cases, it is very difficult for relatively new members to
dice is included that the difference between the two can be know which attitudes are normative. Norms can be remarkably
uncovered. malleable, and in certain circumstances the expression of prejudice
When the target of prejudice is normatively appropriate, people toward inappropriate targets is tolerated—in the locker room, at
with high levels of suppression actually report higher levels of the poker table, or among close friends from the neighborhood
prejudice. Study 6 points to the strength of thinking broadly about where certain prejudices are tolerated. How will high suppressors
prejudice to understand it as a general psychological process. It act when social norms point to tolerance of the vocal racial
was only by sampling from acceptable prejudices that we found prejudice? If local social norms are at odds with the overarching
people who report high levels of prejudice suppression are people societal norms for prejudice, will high suppressors follow their
who also report high levels of prejudice— of a certain kind. hearts or follow the herd?
These data help bring the modern work on the suppression of
prejudice and stereotyping (e.g., Bodenhausen & Macrae, 1998;
Study 7: Suppression and Adherence to Group Norms—
Monteith et al., 1998) in line with Sherif and Sherif’s (1953;
Sherif, 1936) group norm theory of attitudes. Having attitudes that
Conformity and Ambiguity
are discrepant from a valued group creates an internal pressure to We investigate this question in the seventh and final study of
change oneself to fit the group’s attitudes. Festinger, Schachter, this article. High and low prejudice suppressors were presented
and Back (1950) wrote, “If a person wants to stay in a group he with situations that suggested either condemning or condoning
[sic] will be susceptible to influence coming from the group, and racist speech and confidentially recommended responses to racist
he will be willing to conform to the rules which the group sets up” conduct. If suppression is based on egalitarian convictions, then
(p. 91). Although one might argue that adaptation to group norms social influence should play a small role. If suppression is based on
are in fact an external motivation, we suggest that the experience a desire to conform to an important reference group, high suppres-
of this adaptation is primarily in the form of inner strivings to sors will not only condemn racist speech more vigorously, but they
internalize a valued reference group’s attitude. Changing one’s will also condone racist speech more tolerantly than low suppres-
attitude to fit the group’s is the hallmark of group membership. sors, depending on the immediate norm of the situation.
“The presence or absence of change of attitude and the degree Fletcher Blanchard and his colleagues (Blanchard et al., 1991,
thereof is a function of the degree of the individual’s assimilation 1994) provided a means by which norms for the expression of
to the atmosphere of the new community” (Sherif, 1948, p. 67). prejudice may be experimentally manipulated. They found that a
Sherif and Sherif (1953) argued that the socialization, identifi- single confederate condemning or condoning racism could dramat-
cation, and internalization combine to create a feeling that external ically alter tolerance for racist acts. When the confederate con-
social pressures are experienced as an internal choice made with demned racist speech, so too did subjects. When the confederate
free will: condoned racist speech, subjects showed great tolerance for it. By
using a norm manipulation, we explored whether high suppressors
Concepts and ideas concerning social relations become, for the indi-
vidual member, “his own” concepts and ideas. Concerning as they do report less overall tolerance of racial prejudice, regardless of the
the characteristics and relations of individuals to other individuals and norm of the situation, or whether high suppressors pay more
groups, such concepts, ideas, and approaches to problem situations in attention to the manipulated social norm and exhibit heightened
every age become part of the individual’s personal identity. Once tolerance of prejudice when the social norm points in that direc-
learned—at the time of their development, or later, through books, or tion. We predicted that (a) when participants see others condemn-
from the lips of parents or teachers—they seem as “natural” to the ing or condoning prejudice, they will do likewise, and that (b) if
individual as breathing. (pp. 16 –17) high suppressors are more conformist on prejudice-relevant issues
than low suppressors, this conformity will be heightened among
And later, high suppressors.
In short, attitudes of prejudice, learned chiefly through contact with
the norms of social distance prevailing in the group . . . come to Method
constitute a part of the individual’s very self-identity, of his ego. . . . It
is small wonder that they come to be experienced as a “natural” part Participants were 58 psychology students (52% women, 98% White)
of oneself—almost as natural as one’s name. (p. 99) from the University of Kansas, who received course credit. They were
selected from a group (N ⫽ 471; 42% men, 87% White) who volunteered
Following Sherif and Sherif (1953), we suggest that the distinc- for a mass testing session including multiple questionnaires from different
tion between internal and external motivation is a very sensible researchers. During this mass testing session, participants completed the
one, but that it is not exactly what it appears to be on the surface. SPS, Dunton and Fazio’s (1997) Concern scale, and Plant and Devine’s
On one hand, external motivation measures (e.g., EMS; Plant & (1998) IMS. Response options for the three scales ranged from 0 to 9, with
Devine, 1998) reflect the perception of an external, somewhat higher numbers indicating greater suppression.
hostile environment, inhabited by people whose values you do not Individuals scoring in the upper third (high suppressors) and lower third
(low suppressors) of a composite scale created from scores on all three
exactly share and with whom you do not identify, but whose social
suppression scales (overall ␣ ⫽ .85, with each scale representing a single
power you recognize. On the other hand, internal motivation is item) were solicited for participation. Participants were randomly assigned
reported when the values of these people are admired, shared, and to one of three conditions: condemning racism, condoning racism, or a
more intimate contact is desirable. However, both kinds of moti- control condition. On average, 11 weeks passed between participation in
vation to suppress prejudice come from the discrepancy between the mass testing session and the experimental session (range ⫽ 8 –13
the person and his or her social situation. weeks).
374 CRANDALL, ESHLEMAN, AND O’BRIEN
In all three conditions, participants responded to Blanchard et al.’s greater understanding of prejudice phenomena; and recommends
(1991) Acceptance of Racist Conduct scale (ARC; see Table 6). The courses of action for the alteration of prejudice. It is essential to
response options for this scale ranged from 1 to 9, with higher numbers recognize the value of multiple perspectives for the broadest
indicating greater acceptance of racist conduct, and appeared in a petition- understanding of prejudice. We assert that the explicit study of
like form of Blanchard et al.’s scale, with each question appearing on a new
social norms is one essential perspective. Social norms can be
page. In the condemn racism condition, “four previous participants” had all
described, independently measured, and are influential in deter-
marked responses highly condemning racist conduct (M ⫽ 1.25). In the
condone racism condition, the 4 participants had all marked responses mining the expression of prejudice and the acceptability of
highly condoning racist conduct (M ⫽ 8.75). In the control condition, discrimination.
participants were exposed to an unmarked petition. Participants were asked Social norms are remarkably strong predictors of expressed
to fill in their responses in the space provided. After completing the scale, prejudice—to ask about the norms regarding prejudice is, in prac-
participants were debriefed. tical terms, the same as asking people how they personally feel.
Study 1 showed that people adhere firmly to social norms when
Results and Discussion asked to express prejudice; the difference between a correlation of
.96 and unity is well within the range of measurement error.
The first hypothesis is clearly borne out; participants in the The acceptability of discriminatory acts and expressions of
condemn racism condition were more condemning of racist con- hostility also closely follow social norms. The tolerance of dis-
duct (ARC; M ⫽ 2.75), whereas participants in the condone racism crimination in employment, housing, and dating closely fits nor-
condition were more willing to condone racist conduct (M ⫽ 6.04), mative expectations. When jokes provide a means by which groups
with participants in the control condition falling in-between can be denigrated, norms determine which jokes will offend and
(M ⫽ 4.04), F(2, 55) ⫽ 29.78, p ⬍ .001, ⫽ .79. which will delight. The story told by the first three studies is that
To test the second hypothesis, we analyzed the ARC with an social norms are strong predictors of expressions of prejudice-
ANCOVA, with condition as an independent variable and suppres- related attitudes. Knowledge of social norms allows us to predict
sion as a covariate. The proper test of the hypothesis is the expressions of prejudice, reactions to discrimination, and reactions
interaction between condition and covariate—the test for equal to expressions of hostility.
slopes. The correlations between suppression and ARC differed These data can be easily predicted, in broad strokes, from many
significantly by condition, F(2, 53) ⫽ 4.60, p ⬍ .015. In Table 7, different social psychological theories, in addition to plain com-
we report the correlations between the various suppression scales mon sense. People reflect the social milieu in which they live, and
and the ARC scale. In the control and condemn conditions, the prejudices are simply part of that world. What is most compelling
correlation between suppression of prejudice and tolerance for is just how closely the norms apply— individual tolerances solidly
racist conduct is negative and strong (mean rs ⫽ ⫺.65 and ⫺.47, reflect group tolerances.
respectively). However, in the condone condition, the correlation Much has been made of the apparent decline of certain forms of
between suppression and tolerance completely reverses (mean r ⫽ prejudice in recent decades (e.g., Whites’ anti-Black racism; Case
.34). In each case, the reversal is statistically significant. When & Greeley, 1990; Dowden & Robinson, 1993; Firebaugh & Davis,
faced with a normative situation in which peers establish a norm of 1988). But when one uses unobtrusive measures (Crosby et al.,
tolerance for racist speech, the high suppressors are more influ- 1980) or assesses attitudes in normatively ambiguous situations
enced by the manipulation than the low suppressors, and they (Gaertner & Dovidio, 1986), personal attitudes do not seem quite
express substantial tolerance of prejudice. so transformed. In most prejudice research, prejudice is measured
by self-report, and thus is a mixture of attitudes and display rules.
General Discussion The current research is in this long tradition, and because what gets
Normative theory provides rich descriptions of prejudice phe- expressed sets the normative context for future stereotyping and
nomena; suggests strong predictions across cultures, forms of prejudice (Schaller & Conway, 1999; Sechrist & Stangor, 2001), it
prejudice, and historical contexts; proposes explanations for is a sensible strategy. However, if we make a distinction between
the genuine– heartfelt–true attitude and overt expressions, we must
consider the possibility that it may not be hearts and minds that
change, but rather the normative acceptability of certain preju-
Table 6
dices. The public unwillingness to express prejudice may be more
Blanchard, Lilly, and Vaughn’s (1991) Acceptance of Racist
determined by normative influence than by personal attitudes. In
Conduct Scale
the long run, according to the GNT of attitudes, external norms
1. Any student who commits a flagrant act of racism should be expelled become internal attitudes.
(R). One must not lose sight of the fact that a great many prejudices
2. Schools have been making too big an issue of racist incidents, were widely socially acceptable. It is sometimes expressed that one
thereby causing divisiveness on campus. form of group-based antipathy or another is “the last acceptable
3. We need to have more diversity programs at the University of Kansas
(R). prejudice.” Our data show that many groups are the targets of
4. Whenever a White person does something mean to a Black person, unalloyed, unabashed negative affect (e.g., gang members, male
everyone seems to assume it’s racism. and female prostitutes, members of the Ku Klux Klan, illegal
5. Investigators should be allowed to do everything they can to conduct immigrants, members of religious cults, ex-convicts, and porn
an official inquiry of a racist act and everyone should cooperate
completely (R).
stars). The reader may object to our characterization of certain
kinds of group-based antipathy as prejudice per se, as many of
Note. R indicates item is reverse-scored. these targets scoring high in prejudice acceptability (Table 1) are
SOCIAL NORMS, SUPPRESSION, AND PREJUDICE 375
Table 7
Correlations Between Acceptance of Racist Conduct and the Various Suppression Scales by
Experimental Condition
Suppression of Dunton & Fazio’s (1997) Plant & Devine’s (1998) Overall
Condition N Prejudice Scale Concern scale IMS combined
Note. For all four versions of the scales, condemn and condone correlations are different according to Fisher’s
z test (all ps ⬍ .05). IMS ⫽ Internal Motivation Scale.
people who certainly deserve this hostility. However, we argue We propose a developmental model of internalization of group
that deservingness is not a necessary or even useful disqualifier as norms. When a person breaks a social norm as a new member of
a part of a definition of prejudice, in part because deservingness is a group, the initial response will be either a rejection of the group
itself under social normative control. Furthermore, the same psy- or, if the group is valued, a hesitancy characterized by the sup-
chological processes seem to apply in both cases. pression of potentially inappropriate responses. This initial sup-
The study of multiple targets of prejudice permits a broader pression is externally motivated compliance (Kelman, 1958), and
picture of prejudice. Including multiple forms of prejudice will it can be captured with measures of externally motivated behavior
also allow us to avoid idiosyncrasies associated with a particular (e.g., the EMS). Compliance will be followed by identification and
form of prejudice (or a particular researcher’s agenda). When we internalization as the individual comes to identify with the group
limit the scope of our targets, we also limit our perspective of the and take on its norms. These changes can be captured with mea-
phenomena. If we are to understand prejudice as a general psy- sures of internally motivated behavior (e.g., the IMS).
chological phenomenon, we must study it as a general phenome- When only normatively inappropriate targets of prejudice are
non. When we do not temper our biases in the study of prejudice, included in social psychological studies, the struggle for internal-
our data might only reflect a limited scope of a complex mosaic. ization may be taken for an individual commitment to reducing
Psychologists will profit from studying the prejudices they uphold prejudice. In an individualistic society, people may interpret their
as well as those they abhor; we must study normative forms of strivings to become a good member as changes in personal atti-
prejudice as well as counter-normative forms. tudes. On first entering a group, an individual will accurately
We have defined prejudice as a negative evaluation of a group identify external pressures from the group to conform. The
or of an individual on the basis of group membership. We propose changes that result from identification with the group and the
that all forms of prejudice are essentially alike in psychological struggle toward internalization of the group’s norms may be ex-
structure and experience, regardless of how various forms of perienced as internal attitudes and commitments. We suggest that
prejudice may differ in external justification. People justify their the basis of the motivation for suppression is a concern with one’s
prejudices to themselves (Crandall & Eshleman, 2000), whether an relationships to group norms.
external audience approves of this justification is not a psycholog- When suppression scales measure attitudes toward normatively
ical issue. This definition allows us to examine White anti-Black inappropriate forms of prejudice (e.g., the SPS; Dunton & Fazio,
racism simultaneously with an average American’s attitude toward 1997; Plant & Devine, 1998), individuals of various levels of
members of the American Nazi Party. Although a social historian, identification with their groups may appear to differ in their
politician, or college lecturer may keep these attitudes quite sep- commitments to reducing prejudice. The present studies suggest
arate (which makes terrific sense to us), a social psychologist that suppression scales measure awareness of one’s group’s norms,
should recognize when they rely on the same psychological and one’s attitude toward them, more than they measure personal
mechanisms. attitudes toward prejudice (see Piaget, 1965; Turner, 1991; Turner
An important advantage to using normative theory to understand et al., 1987). When the study of prejudice is expanded to include
prejudice is how it highlights the struggle for internalization. As a normatively appropriate prejudice (either as acceptable groups in
person moves from one social group to another, the process of Study 6 or when the norm itself is manipulated, as in Study 7), we
identification with the new group involves taking on the new find that suppression scales seem to measure one’s patterns of
group’s frame of reference (Cantril, 1941; Sherif & Sherif, 1964). concerns about our relationships with the group’s norms better
The person learns the new group’s norms and begins to adapt to than one’s commitments to reducing prejudice.
and internalize them (Kelman, 1958, 1961; Sherif, 1948). During Plant and Devine (1998) found that people high in internal
this struggle for internalization, norms will be salient (Hogg & motivation to suppress prejudice “reported the campus standards to
Abrams, 1988), and the individual will be aware of any deviance be somewhat less prejudiced than their low IMS counterparts” (p.
from the norms (Turner, 1991). It is this deviance from a reference 821), despite the fact that all participants perceived their campus to
group’s norms, we argue, that provokes internal and external be strongly nonprejudiced. They argued that this was due to dif-
motivations to suppress prejudice. We have used normative theory ferential contact; high internal suppressors “are more likely to
to conceptualize the roots of suppression, a phenomenon widely spend time with other low-prejudice people” (p. 821), and subse-
studied and reported (Dunton & Fazio, 1997; Fazio et al., 1995; quently judge the campus as less prejudiced. We suggest a slightly
Gaertner & Dovidio, 1986; Plant & Devine, 1998; Sears, 1998). different interpretation—those high in internal motivation to sup-
376 CRANDALL, ESHLEMAN, AND O’BRIEN
press are actually more ambivalent, more concerned, anxious, Brigham, J. C. (1993). College students’ racial attitudes. Journal of Applied
apprehensive, motivated, and delicate in their sense that they fit the Social Psychology, 23, 1933–1967.
social norms. Their experience in the initial discovery that racial Burma, J. (1946). Humor as a technique in race conflict. American Socio-
and ethnic prejudice is not tolerated on campus is taken to heart, logical Review, 11, 710 –715.
and their identification with the group leads to a vigorous and Campbell, A. A. (1947). Factors associated with attitudes toward Jews. In
T. Newcomb & E. Hartley (Eds.), Readings in social psychology (pp.
powerful self-directed attempt at conformity, which extremitizes
518 –527). New York: Holt.
the perception of social norms (see Turner, 1991).
Cantor, J. R. (1976). What is funny to whom? Journal of Communica-
It is important to point out that, although we suggest a new
tion, 26, 164 –172.
approach to the source of motivations to suppress prejudice, our Cantril, H. (1941). The psychology of social movements. New York:
account is largely consistent with Plant and Devine’s (1998) ap- Chapman & Hall.
proach. We endorse the internal– external distinction, and we agree Case, C. E., & Greeley, A. M. (1990). Attitudes toward racial equality.
that the correlates and consequences of motivation and suppression Humboldt Journal of Social Relations, 16, 67–94.
failures should look much as Plant and Devine have shown. Our Crandall, C. S. (2000). Ideology and lay theories of stigma: The justifica-
description of the process speaks to one important factor in chang- tion of stigmatization. In T. Heatherton, J. Hull, R. Kleck, & M. Hebl
ing reported motivations—identifying with new groups and inter- (Eds.), The social psychology of stigma (pp. 126 –150). New York:
nalizing their attitude norms. However, we expect that violations Guilford.
of internal motivations would lead to guilt and self-criticism, and Crandall, C. S., & Eshleman, A. (2000). The justification-suppression
that violations of external motivations would lead to feelings of model of prejudice: A general model of the expression and experience of
external threat, anger, and frustration, or that people high in prejudice. Unpublished manuscript, University of Kansas.
Crandall, C. S., Preisler, J. J., & Aussprung, J. (1992). Measuring life event
external motivation will adapt readily to external audiences while
stress in the lives of college students: Undergraduate Stress Question-
those primarily high in internal motivation will not—as Plant and
naire (USQ). Journal of Behavioral Medicine, 15, 627– 662.
Devine (1998) have shown. We trace the source of the feelings of
Crosby, F., Bromley, S., & Saxe, L. (1980). Recent unobtrusive studies of
internal and external motivation to suppress, but we do not ques- Black and White discrimination and prejudice: A literature review.
tion the presence of these motives. Psychological Bulletin, 87, 546 –563.
Normative theory may have fallen out of vogue in recent de- Dovidio, J. F., & Gaertner, S. L. (1986). Prejudice, discrimination, and
cades, but these findings demonstrate the value of this traditional racism: Historical trends and contemporary approaches. In J. F. Dovidio
theoretical perspective. Social norms are powerful predictors of & S. L. Gaertner (Eds.), Prejudice, discrimination, and racism (pp.
attitudes and behaviors, and prejudice and discrimination are no 1–34). New York: Academic Press.
exception. One advantage of the normative approach is it allows Dovidio, J. F., & Gaertner, S. L. (1991). Changes in the expression and
for a pragmatic optimism. Rather than facing the daunting task of assessment of racial prejudice. In H. J. Knopke, R. J. Norrell, & R. W.
changing the ingrained attitudes of millions of individuals, a norms Rogers (Eds.), Opening doors: Perspectives on race relations in con-
approach suggests that changing the normative climate can be an temporary America (pp. 119 –148). Tuscaloosa: University of Alabama
efficient and effective approach to attitude change (see Bem, Press.
Dowden, S., & Robinson, J. P. (1993). Age and cohort differences in
1970). The more desirable the group, the more people will wish to
American racial attitudes: The generational replacement hypothesis re-
follow its lead. And when people identify with attractive groups
visited. In P. M. Sniderman, P. E. Tetlock, & E. G. Carmines (Eds.),
that condemn a prejudice, they are likely to win the struggle for Prejudice, politics, and the American dilemma (pp. 86 –103). Stanford,
internalization. CA: Stanford University Press.
Dunton, B. C., & Fazio, R. H. (1997). An individual difference measure of
motivation to control prejudiced reactions. Personality and Social Psy-
References chology Bulletin, 23, 316 –326.
Fazio, R. H., Jackson, J. R., Dunton, B. C., & Williams, C. J. (1995).
Allport, G. W. (1954). The nature of prejudice. Cambridge, MA: Addison-
Variability in automatic activation as an unobtrusive measure of racial
Wesley.
attitudes: A bona fide pipeline? Journal of Personality and Social
Anant, S. S. (1975). The changing intercaste attitudes in North India: A
Psychology, 69, 1013–1027.
follow-up after four years. European Journal of Social Psychology, 5,
49 –59. Festinger, L., Schachter, S., & Back, K. (1950). Social pressures in
Bem, D. (1970). Beliefs, attitudes, and human affairs. Belmont, CA: informal groups. New York: Harper.
Brooks/Cole. Firebaugh, G., & Davis, K. E. (1988). Trends in anti-Black prejudice,
Biernat, M., & Crandall, C. S. (1999). Racial attitudes. In J. Robinson, P. 1972–1984: Region and cohort effects. American Journal of Sociol-
Shaver, & L. Wrightsman (Eds.), Measures of political attitudes (2nd ogy, 94, 251–272.
ed.; pp. 297– 411). New York: Academic Press. Gaertner, S. L., & Dovidio, J. F. (1986). The aversive form of racism. In
Blanchard, F. A., Crandall, C. S., Brigham, J. C., & Vaughn, L. A. (1994). J. F. Dovidio & S. L. Gaertner (Eds.), Prejudice, discrimination, and
Condemning and condoning racism: A social context approach to inter- racism (pp. 61– 89). New York: Academic Press.
racial settings. Journal of Applied Psychology, 79, 993–997. Hogg, M. A., & Abrams, D. (1988). Social identifications. London:
Blanchard, F. A., Lilly, T., & Vaughn, L. A. (1991). Reducing the expres- Routledge.
sion of racial prejudice. Psychological Science, 2, 101–105. Horowitz, E. L. (1936). The development of attitude toward the Negro.
Bodenhausen, G. V., & Macrae, C. N. (1998). Stereotype activation and Archives of Psychology, 194, 47.
inhibition. In R. S. Wyer (Ed.), Advances in social cognition (Vol. 11; Hurwitz, J., & Peffley, M. (1992). Traditional versus social values as
pp. 1–52). Mahwah, NJ: Erlbaum. antecedents of racial stereotyping and policy conservatism. Political
Bolton, E. B. (1935). Effects of knowledge upon attitudes towards the Behavior, 14, 395– 421.
Negro. Journal of Social Psychology, 6, 68 – 89. Katz, I., & Hass, R. G. (1988). Racial ambivalence and American value
SOCIAL NORMS, SUPPRESSION, AND PREJUDICE 377
conflict: Correlational and priming studies of dual cognitive structures. Pettigrew, T. F. (1991). Normative theory in intergroup relations: Explain-
Journal of Personality and Social Psychology, 55, 893–905. ing both harmony and conflict. Psychology and Developing Societies, 3,
Kelman, H. C. (1958). Compliance, identification and internalization: 3–16.
Three processes of attitude change. Journal of Conflict Resolution, 2, Pettigrew, T. F. (1999). Gordon Willard Allport: A tribute. Journal of
51– 60. Social Issues, 55, 415– 427.
Kelman, H. C. (1961). Processes of opinion change. Public Opinion Pettigrew, T. F., & Meertens, R. W. (1995). Subtle and blatant prejudice in
Quarterly, 25, 57–78. Western Europe. European Journal of Social Psychology, 25, 57–75.
Kim, U., Triandis, H. C., Kagitcibasi, C., Choi, S.-C., & Yoon, G. (1994). Piaget, J. (1965). The moral judgment of the child. New York: Free Press.
Individualism and collectivism: Theory, method, and applications. Plant, E. A., & Devine, P. G. (1998). Internal and external motivation to
Thousand Oaks, CA: Sage. respond without prejudice. Journal of Personality and Social Psychol-
La Fave, L., & Mannell, R. (1976). Does ethnic humor serve prejudice? ogy, 75, 811– 832.
Journal of Communication, 26, 116 –123. Rogers, R. W., & Prentice-Dunn, S. (1981). Deindividuation and anger-
LaFrance, M., & Woodzicka, J. A. (1998). No laughing matter: Women’s mediated interracial aggression: Unmasking regressive racism. Journal
verbal and nonverbal reactions to sexist humor. In J. K. Swim & C. of Personality and Social Psychology, 41, 63–73.
Stangor (Eds.), Prejudice: The target’s perspective (pp. 61– 80). San Rokeach, M., & Ball-Rokeach, S. J. (1989). Stability and change in
Diego, CA: Academic Press. American value priorities, 1968 –1981. American Psychologist, 44, 775–
Lasker, B. (1929). Race attitudes in children. New York: Henry Holt. 784.
Levine, M. S. (1977). Canonical analysis and factor comparison. Beverly Ruggiero, K. M., & Kelman, H. C. (1999). Introduction to the issue.
Hills, CA: Sage. Journal of Social Issues, 55, 405– 414.
Marrow, A., & French, J. R. P., Jr. (1945). Changing a stereotype in Schaller, M., & Conway, L. G. (1999). Influence of impression-
industry. Journal of Social Issues, 1, 33–37. management goals on the emerging contents of group stereotypes:
McConahay, J. B., Hardee, B. B., & Batts, V. (1981). Has racism declined Support for a social-evolutionary process. Personality and Social Psy-
in America? It depends on who is asking and what is asked. Journal of chology Bulletin, 25, 819 – 833.
Conflict Resolution, 25, 563–579. Sears, D. O. (1998). Racism and politics in the United States. In J. L.
Meertens, R. W., & Pettigrew, T. F. (1997). Is subtle prejudice really Eberhardt & S. T. Fiske (Eds.), Confronting racism: The problem and
prejudice? Public Opinion Quarterly, 61, 54 –71. the response (pp. 76 –100). Thousand Oaks, CA: Sage.
Minard, R. (1952). Race relations in the Pocahontas coal field. Journal of Sechrist, G. B., & Stangor, C. (2001). Perceived consensus influences
Social Issues, 8, 29 – 44. intergroup behavior and stereotype accessibility. Journal of Personality
Monteith, M. J. (1993). Self-regulation of prejudiced responses: Implica- and Social Psychology, 80, 645– 654.
tions for progress in prejudice-reduction efforts. Journal of Personality Sherif, M. (1936). The psychology of social norms. New York: Harper.
and Social Psychology, 65, 469 – 485. Sherif, M. (1948). The necessity of considering current issues as part and
Monteith, M. J. (1996). Affective reactions to prejudice-related discrepant parcel of persistent major problems: Illustrated by the problem of prej-
responses: The impact of standard salience. Personality and Social udice. International Journal of Opinion and Attitude Research, 2, 63–
Psychology Bulletin, 22, 48 –59. 68.
Monteith, M. J., Deneen, N. E., & Tooman, G. D. (1996). The effect of Sherif, M., & Sherif, C. W. (1953). Groups in harmony and tension. New
social norm activation on the expression of opinions concerning gay men York: Harper.
and Blacks. Basic and Applied Social Psychology, 18, 267–288. Sherif, M., & Sherif, C. W. (1964). Reference groups. New York: Harper
Monteith, M. J., Devine, P. G., & Zuwerink, J. R. (1993). Self-directed & Row.
versus other-directed affect as a consequence of prejudice-related dis- Stangor, C., Sechrist, G., & Jost, J. T. (2001). Changing racial beliefs by
crepancies. Journal of Personality and Social Psychology, 64, 198 –210. providing consensus information. Personality and Social Psychology
Monteith, M. J., Sherman, J. W., & Devine, P. G. (1998). Suppression as Bulletin, 27, 486 – 496.
a stereotype control strategy. Personality and Social Psychology Re- Terry, D. J., Hogg, M. A., & White, K. M. (2000). Attitude– behavior
view, 2, 63– 82. relations: Social identity and group membership. In D. J. Terry & M. A.
Newcomb, T. M. (1943). Personality & social change. New York: Dryden. Hogg (Eds.), Attitudes, behavior, and social context: The role of norms
O’Brien, L. T., Eshleman, A. K., & Crandall, C. S. (1999, May). Preju- and group membership (pp. 67–93). Mahwah, NJ: Erlbaum.
diced jokes and social norms: What’s offensive and what’s funny? Paper Triandis, H. C., & Davis, E. E. (1965). Race and belief as determinants of
presented at the 70th Annual Meeting of the Midwestern Psychological behavioral intentions. Journal of Personality and Social Psychology, 2,
Association, Chicago, IL. 715–725.
Page, C. (1999, September 6). Racism is social, not just mental. Kansas, Turner, J. C. (1991). Social influence. Pacific Grove, CA: Brooks/Cole.
City Star, p. B5. Turner, J. C., Hogg, M. A., Oakes, P. J., Reicher, S. D., & Wetherell, M. S.
Pettigrew, T. F. (1958). Personality and sociocultural factors in intergroup (1987). Rediscovering the social group: A self-categorization theory.
attitudes: A cross-national comparison. Journal of Conflict Resolu- New York: Blackwell.
tion, 2, 29 – 42. Weitz, S. (1972). Attitude, voice, and behavior: A repressed affect model
Pettigrew, T. F. (1959). Regional differences in anti-Negro prejudice. of interracial interaction. Journal of Personality and Social Psychol-
Journal of Abnormal and Social Psychology, 59, 28 –36. ogy, 24, 14 –21.
(Appendixes follow)
378 CRANDALL, ESHLEMAN, AND O’BRIEN
Appendix A
Discrimination Scenarios
Dating the manager’s assistant. Terry called the apartment manager the next day
to set up an appointment to sign the lease. While the manager was
Jamie is a first year KU student who enjoys going to concerts, dating, searching for the appointment book, the manager’s assistant mentioned that
and hanging out with friends at coffee shops. Jamie’s friend Dana recently Terry is (Hispanic). Because Terry is (Hispanic), the manager told Terry
suggested that Jamie should meet Dana’s friend Chris, who lives upstairs that the apartment was no longer available.
from Jamie in the residence hall. Jamie agreed to go out on a blind date
with Chris, and Dana started making plans with Chris about where they
should go on their date. A few days later, Jamie discovered that Chris is Employment
(African American) and called off the date because Chris is (African
American). Pat saw an advertisement for a data entry position at a local business.
The next day Pat stopped by the business, filled out an application, and
Housing chatted with the receptionist for several minutes. The receptionist gave
Pat’s application to the business manager in charge of hiring and men-
Terry has been looking forward to renting an apartment in Lawrence. tioned that Pat is (fat). The business manager filed away Pat’s application
Terry found a perfect place on the bus route, near shopping, and close because Pat is (fat), without any intention of seriously considering Pat for
enough to campus to walk on nice days. Terry was shown the apartment by the job.
Appendix B
Jokes
1. Q. Why did the racist die in a pie-eating contest? 5. Q. Why did the woman on welfare climb up to the roof of the bar?
A. A cow stepped on his head. A. She heard that the drinks were on the house.
2. Q. How do you save a drowning drug user? 6. Q. Why don’t fat people have elevator jobs?
A. Throw him a rock. A. They don’t know the route.
3. A ventriloquist in a nightclub is telling a series of jokes about 7. Q. What are the worst six years in a Black person’s life?
ex-convicts. He plays straight man while the little dummy on his knee A. Third grade.
poses the riddles and answers the questions. After many of these, an 8. Q. What do UFOs and smart Hispanics have in common?
ex-convict in the audience gets up and shouts, “Will you knock it off? A. You keep hearing about them, but never see any.
I’m just sick and tired of all these idiotic ex-convict jokes that try to 9. Q. What is the definition of gross ignorance?
make us look stupid. Just stop it, because we ex-convicts are proud of A. 144 illiterate people.
our contributions to civilization and many great intellects have 10. Two guys are sitting in a bar swapping jokes about mentally retarded
been ex-convicts.” The ventriloquist says, “I’m really sorry to cause people. A mentally retarded person comes in, and after listening for a bit
such offense. Will you accept my apology?” The ex-convict exclaims, contributes one of his own: “How do you get a mentally retarded girl
“It’s not you I’m talking to, you fool, it’s that little bastard on your pregnant?” he asks. The two guys are stumped. “I dunno,” they reply. The
knee!” mentally retarded person responds, “Gee, and you say the mentally re-
4. Two rednecks were driving a semi-truck down a road when they came tarded are stupid.”
to a viaduct. The sign said 10 feet 0 inches, so they got out to measure their
truck. Unfortunately, the truck was just over 12 feet high. They didn’t Received December 29, 2000
know what to do, when finally one of them looked in both directions and Revision received July 20, 2001
said, “I don’t see any cops, let’s go for it.” Accepted July 24, 2001 䡲