Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Social identification and the treatment of in-group deviants: the black sheep effect or the devil protection effect
(USC Thesis Other)
Social identification and the treatment of in-group deviants: the black sheep effect or the devil protection effect
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
SOCIAL IDENTIFICATION AND THE TREATMENT OF IN-GROUP
DEVIANTS: THE BLACK SHEEP EFFECT OR THE DEVIL PROTECTION
EFFECT
by
Joy Stratton
A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(PSYCHOLOGY)
August 2007
Copyright 2007 Joy Stratton
ii
Table of Contents
List of Tables iii
Abstract iv
Introduction 1
Social Identity Theory 3
The Treatment of In-group Deviants 3
Subjective Group Dynamics 4
Identification and the Treatment of In-group Deviants 7
Predictions 13
Chapter 1: Study 1 15
Method 15
Results and Discussion 19
Chapter 2: Study 2 27
Method 28
Results and Discussion 32
Chapter 3: Study 3 38
Method 39
Results and Discussion 43
Chapter 4: General Discussion 47
References 52
Appendices 59
Appendix A 59
Appendix B 60
Appendix C 61
Appendix D 62
Appendix E 65
Appendix F 66
Appendix G 67
Appendix H 69
iii
List of Tables
Table 1: Mean trait evaluations of targets as a function of target
behavior and participant’s level of identification with the
in-group (Experiment 1) 21
Table 2: Mean negative mood ratings of participants as a function
of target behavior and participant’s level of identification
with the in-group (Experiment 1) 23
Table 3: Mean trait evaluations of targets as a function of context
of target evaluations and participant’s level of identification
with the in-group (Experiment 2) 34
Table 4: Mean trait evaluations of targets as a function of participant’s
type of identification with the in-group (Experiment 3) 45
iv
Abstract
Social identity theory (Tajfel, 1979; Tajfel & Turner, 1986) in conjunction with
the subjective group dynamics model (Marques, Paez, & Abrams, 1998; Marques,
Abrams, Paez, & Hogg, 2001) predicts that the negative and/ or deviant behavior
of an in-group member may result in more negative evaluations of that person
relative to a similarly-described out-group member (the black sheep effect) when
certain conditions are met. These same theories also imply that people may also
‘protect’ deviant in-group members over out-group members (the devil protection
effect) in other situations. In study 1, I hypothesized that participants’ overall
level of identification with the in-group will moderate judgments of in-group and
out-group deviants, such that high identifiers would be more likely than low
identifiers to demonstrate the black sheep effect. However, this hypothesis was
not supported. Study 2 assessed whether the context of target evaluations (public
or private) moderates the relationship between social identification and the black
sheep effect. Context of target evaluations was found to moderate deviant target
evaluations such that deviant in-group targets were rated more positively than
deviant out-group targets when participants thought their evaluations would be
public (the devil protection effect). No differential evaluations were found when
participants thought their evaluations would be private. Study 3 investigated
whether the nature of a person’s identification with an in-group differentially
predicts the black sheep effect or the devil protection effect. Partial support was
found for this hypothesis.
1
Introduction
As social creatures, humans seek out affiliation with others. They naturally
form groups to enhance their ability to meet certain needs, both physical and
emotional. Describing how groups relate to each other and deciphering what sort
of processes mark both inter- and intra-group interactions has been the subject of
much investigation by psychologists. The goal of the current studies was to
increase understanding of one group process: the exaggerated response to an in-
group deviate.
All things being equal, the normal response to an in-group member is
favoritism over a similar out-group member. However, in-group deviates have
been known to provoke extreme reactions in both directions relative to similarly
behaving out-group members. Andres Escobar is an example of an in-group
deviate who provoked a much harsher response from his countrymen than out-
group members who committed the same act. Ten days after he scored a point
against his own team during the 1994 World Cup game against the United States,
Andres, a soccer player for Columbia’s national team, was gunned down by hard-
core fans of the Columbian soccer team. In contrast, O.J. Simpson provides an
example of an in-group deviant who was excessively favored by his fellow in-
group members relative to similarly behaving out-group members. Even though it
is very likely that he murdered two people, many African Americans defended and
supported him, applauding his ‘not-guilty’ verdict. At times, the same person may
even educe both types of extreme response: derogation and protection. After it
2
came to light that Evangelical leader Ted Haggard had regularly committed “sexual
immoralities” with a male prostitute, some evangelical leaders derided him whilst
others ‘forgave’ him. In the current studies, I sought to explore why under some
conditions deviant in-group members are evaluated and/ or treated more negatively
in comparison to similarly behaving out-group members (the black sheep effect)
whereas under other conditions, deviant in-group members are protected by the in-
group (the devil protection effect).
Study 1 focused on determining who among in-group members are mostly
likely to derogate in-group deviants and why. Specifically, group member’s
identification with the in-group was expected to moderate target evaluations, such
that high identifiers exhibit the black sheep effect whereas low identifiers do not.
Perceived threat to one’s social identity was expected to mediate differential target
evaluations. Study 2 sought to explore under what conditions the black sheep
effect is expected to be reversed. In particular, I explored whether highly public
target evaluations will result in a devil protection effect, in which the deviant in-
group member is protected by high identifiers. Study 3 explored whether different
kinds of identification with the in-group will produce the devil protection or the
black sheep effect. Group identification based on one’s personal connection with
the group was expected to produce protection of in-group deviants relative to out-
group deviants. In contrast, group identification based on the positivity of the
group’s image was expected to produce the black sheep effect.
3
I organized the introduction of this dissertation in the following manner:
first, I briefly review social identity theory and its implications for group target
evaluations. Then, I initiate discussion of the treatment of in-group deviants by
describing the black sheep effect. Finally, I discuss how a person’s level of
identification with their in-group is expected to interact with other factors to
produce the black sheep effect under some circumstances and the devil protection
effect in others.
Social Identity Theory
According to social identity theory (e.g. Tajfel, 1978, Tajfel & Turner,
1986), group categorization in and of itself stimulates comparison between the in-
group and out-group, and there is a motivational tendency for people to resolve
these comparisons in ways that maintain, enhance, or defend their self esteem.
Thus, one manner in which a positive self concept can be maintained or achieved is
through identification with a positively perceived group. In-group bias boosts the
comparative positive distinctiveness; thus, when one affiliates with a group that can
be positively compared with relevant out-groups, positive social identity is
achieved or maintained.
The Treatment of In-group Deviants
Social identity theory also implies that identification with an in-group can
have a negative impact on one’s self identity (i.e. when others have reason to
evaluate or perceive the in-group negatively (Lewin, 1948)). Negative and/ or
deviant behavior of other in-group members may provide one such reason why in-
4
group identification negatively impacts one’s self identity. According to Cohen
and Garcia (2005), conclusions about the entire group are often drawn from the
behavior of individual group members. Thus, deviant or negative behavior on the
part of one in-group member poses a threat to the entire group’s image, which in
turn, according to social identity theory, threatens the self concept of individual in-
group members. When the social identity provided by an in-group is
unsatisfactory, individuals will be motivated to either leave their existing group or
restore its positive image (Tajfel & Turner, 2001).
Subjective Group Dynamics
Besides providing a source of identity for group members, groups serve
other functions. Decades ago, Festinger (1950) hypothesized that individuals
selectively affiliate with similar others in order to avoid uncertainty about things
that cannot be otherwise proven (i.e. attitudes and opinions) and to achieve goals
that cannot be accomplished individually. This idea forms the basis of the
subjective group dynamics model which posits that the group can serve to uphold
members’ beliefs on the correctness of their opinions on relevant issues and
confirm their positive expectations about themselves and other group members
(e.g. Marques, Abrams, & Serodio, 2001; Abrams, et al., 2004). Deviant group
members undermine the subjective validity of the group’s social reality. The in-
group deviant may also threaten in-group distinctiveness, along with its positive
image; therefore such deviants are met with hostility, negative evaluations, and
attempts to bring them into line or ostracize them.
5
The black sheep effect
Marques and his colleagues (2001) argue that the derogation of deviant in-
group members, termed the black sheep effect, is an attempt to restore one’s
positive social identity. Traditionally, in-group membership is associated with
favorable consequences rather than mistreatment or punishment, even when the
category distinctiveness on which the group is based is irrelevant to the situation at
hand or ascribed randomly (e.g. Tajfel, et al., 1971; Billig & Tajfel, 1973; Brewer,
1979; Brewer & Brown, 1998). However, deviant or unlikeable in-group members
often receive poorer evaluations than similar out-group members (e.g. Marques,
Yzerbyt, & Leyens, 1988; Marques, Abrams, & Serodio, 2001; Marques, Robalo,
& Rocha, 1992; Marques & Yzerbyt, 1988; Matthews & Dietz-Uhler, 1998; Coull
et al, 2001). By derogating in-group deviants, deviants are punished for failure to
conform to group standards and are therefore motivated to ‘shape up or ship out’.
This idea is substantiated by research which found that the black sheep effect was
greater in situations where in-group norms were undermined in a social context
than when these norms were validated (Marques et al., 2001) and when the deviant
behavior violated group specific norms (norms specifically relevant to the in-group
membership in question) as opposed to general norms (Marques, Yzerbyt,&
Leyens, 1988). For the perceiver, derogation may be a means of psychologically
excising the deviant from the in-group so that when the perceiver makes mental
comparisons between his/her in-group and relevant out-groups, positive
distinctiveness can again be achieved and positive social identity restored. Thus,
6
the purpose of derogating an offending group member is to alleviate a threatened
social identity by motivating offenders to stop whatever it is that they are doing
which is damaging the in-group’s image, and/ or by psychologically redrawing the
boundaries of the in-group in a manner that excludes the deviant.
The devil protection effect
Under certain circumstances, the traditional options for dealing with social
identity threat (leaving the group or restoring its positive image by denigrating the
deviant in-group member) may be less viable. When a lot of public attention is
focused on group members, derogating an in-group member may weaken the
group’s power or set a standard to outsiders sanctioning ill treatment of the group’s
members. If the deviant is a group leader or has relatively high status, lower status
group members may not have the power to denigrate the deviant. Likewise, if one
is has a strong ideological connection with the group, derogating any in-group
member may prove more damaging to one’s social identity then justifying the
misbehavior in some way. In these cases, we may see instead a reversal of the
black sheep effect: deviant in-group members may be up-graded relative to
similarly deviant out-group members. For example, if the public is watching,
derogating the offending in-group member could further damage the group’s image
and power. If group members instead gloss over the misdeed and show favor to the
deviant, they are essentially sending a message to outsiders that the group is united
and strong. Thus, the devil protection effect may stem from motivations to
minimize further damage to group’s image and group members’ social identity.
7
Identification and the Treatment of In-group Deviants
At any given time, a single individual may belong to any number of groups.
However, only a few of these groups are likely to determine self definition and
therefore affect one’s social self concept. Thus, the extent to which individuals are
affected by group characteristics and group processes will differ, depending on the
extent to which that individuals identifies with the group in question (Ellemers,
Spears, & Doosje, 2002). In order to feel social identity threat, “individuals must
have internalized their group membership as an aspect of their self-concept,” (Tajel
& Turner, 2001, p. 101). High identifiers generally feel stronger emotions
associated with their group membership than low identifiers and are therefore more
sensitive to threats to the positivity of their social identity (Roccas, Klar, &
Liviatan, 2004). As a result, they are more likely to engage in defense mechanisms
aimed at restoring a positive social identity than low identifiers. Thus, overall
strength of group identification is a likely moderator of differential evaluations of
group targets. Indeed, correlational research on the black sheep effect indicates that
high identifiers are more likely than low identifiers to engage in the black sheep
effect (e.g. Branscombe, et al., 1993; Eidelman & Biernat, 2003, Coull, et al., 2001)
and more likely to react to threats of intergroup distinctiveness (Jetten, Spears, &
Postmes, 2004). But up to this point, no one has actually manipulated the strength
of group identification in studies on the treatment of in-group deviants.
Identification with the in-group may also produce an increase in the devil
protection effect, when certain conditions are met. Besides being more sensitive to
8
threats to their group’s image (Roccas, Klar, & Liviatan, 2004), high identifiers are
also more likely to show loyalty to the group than low identifiers. In other words,
they are more likely to remain in a group than low identifiers, even when the group
is at risk for being viewed negatively by others or when there are attractive reasons
for leaving the group (Van Vugt & Hart, 2004; Doosje, Spears, & Ellemers, 2002).
They are also more likely to sacrifice their own resources in favor of the group’s
greater good (Brewer & Kramer, 1986; De Cremer & Van Vugt, 1999; Kramer &
Brewer, 1984; Van Vugt, 2001). Thus, they may be more motivated to stick to the
group no matter what and demonstrate group unanimity when outsiders are
watching. In contrast, low identifiers tend to be more instrumental. Doosje,
Spears, and Ellemers (2002) found that low identifiers only expressed solidarity
with their low status group to the extent that improvement of the group status
constituted a likely prospect and concluded that “low identifiers commit themselves
to the group only when their individual goals and outcomes corresponded with
those of the group” (p. 68). Thus, when others are watching low identifiers may be
more likely to distance themselves from anything that damages their personal
image whether it be the group itself or group deviants.
Publicity of target evaluation
The context of target evaluations may produce a reversal of the black sheep
effect. Research on the intergroup sensitivity effect (ISE) indicates that the private
criticism of in-group members by other in-group members is generally better
received (the speaker is perceived more positively) than when the same criticism
9
comes from an out-group member (Hornsey, Oppes, & Svensson, 2002). However,
this effect disappears when the criticism of in-group members occurs in front of
out-groupers (Elder, Sutton, & Douglas, 2005). One possible explanation for this
reversal is that there are norms governing intergroup criticism such that “group
members should collaborate to foster a desired public image of their group,” (Elder,
Sutton, & Douglas, 2005, p. 241). When in-group members criticize other in-group
members in the presence of out-groupers, they are both deviating from an in-group
norm themselves and possibly adding further damage to the group’s image by
showing it to lack cohesiveness.
Displays of the black sheep effect are likely to be determined interactively
by the level of in-group identification and by the context in which the targets are
rated. In the face of outside attention, derision, increased out-group hostility and
potential attack, people may decide it is more important to maintain the image of a
united front rather than let the in-group seem disunited by derogating a deviant in-
group member. This is especially likely when perceivers are high identifiers.
Ellmers, Spears, and Doosje (1997) found that under high category salience or
threat, high-identifiers exhibited more commitment to the in-group and less desire
to leave their group for a higher status one. In essence, high identifiers were more
motivated to ‘stick together’ than low identifiers. This motivation is expected to be
particularly strong when high identifiers have an out-group audience as they “seem
prepared to incur the social costs that might be associated with standing up for their
group in public (Ellmers, et al., 2000, p. 70). For example, Postmes, et al. (1999)
10
found that when high identifiers made public evaluations they tended to ‘stand by
their group’ and show ‘a united front’ even in the face of potential aversive
consequences. In contrast, low identifiers tended to be more opportunistic,
changing their public evaluations in such a way as to appear more favorable to a
particular audience. Under a public context, low identifiers may be more likely to
engage in the black sheep effect, because they may feel a stronger need to distance
themselves from the deviant in-group member than to protect the image of the in-
group.
Type of identification
Recently, some researchers have called for distinctions to be made between
different dimensions of social identification based on empirical data that suggests
that different types of social identity predict different group processes (Ashmore,
Daeux, & McLaughlin-Volpe, 2004; Roccas, Klar, Liviatan; 2004). For example,
Ellemers, Kortekaas, and Ouwerker (1999) measured self-categorization,
commitment, and group self esteem and found that they could be distinguished as
separate factors on a principle components analysis and were differentially related
to displays of in-group favoritism. Other research supports the idea that the type of
identification that a person feels with their in-group is also likely to impact
judgments of deviant group members. Roccas, Klar and Liviatan (2004) found that
different aspects of identification with one’s national group (attachment to and
glorification of the group) produced different levels of collective guilt and opposite
types of reactions to in- and out-group members who committed various types of
11
moral violations. They termed these phenomena the paradox of group
identification. They had participants read about an in-group or out-group member
who committed a moral violation against an out-group member and rate the amount
of disgust they felt, how much they felt the act violated basic moral values and the
appropriate amount of imprisonment the perpetrator should receive for his/her
action (these items were combined to create a moral outrage scale). Attachment to
the in-group, defined as importance to the individual’s self concept and
commitment to the in-group, was positively related to moral outrage toward in-
group perpetrators. In contrast, glorification (perceptions that the in-group is better
and more worthy than other groups and idealization) was negatively related to
moral outrage toward in-group perpetrators. The authors attributed both of these
actions as attempts to defend the moral character of the group (i.e. maintain or
boost social identity) though one method involved denouncing the in-group deviant
and the other involved minimizing the blame assigned to in-group offenders.
Different dimensions of identification also can be distinguished on the basis
of particular needs they serve. Some people may identify with a group because
they share the same ideology as other group members and/ or feel very connected
with the group. For the purposes of this paper, we will call these people
“connected identifiers”. In contrast, others may identify with a group because it is
viewed positivity by others “positivity identifiers”. Persons with this latter type of
group identity are more likely to see their in-group as powerful, as a source of
rewards, and as a vehicle for them to attain their own goals. They are also likely to
12
use the group opportunistically to enhance their image with others. For example,
when the group is viewed positively, they may be more likely to draw associations
between themselves and their group and “bask in its reflected glory”, while at the
same time ‘blasting’ (derogating) rival groups (Cialdini et al., 1976; Cialdini &
Richardson, 1980). However, when the group is viewed negatively, persons who
identity with a group because of its positivity may be more likely than connected
identifiers to conceal connections between themselves and the group.
Connected identification and positivity identification are likely to provoke
different types of reactions to social identity threat. In particular, identification
based on one’s tight personal connection with the group may be more likely to
provoke the protection of the in-group deviant (the devil protection effect) because
there is greater commitment to the group for its own sake. Distancing oneself from
the group is tantamount to denying the values and ideas which the group stands for,
so instead, connected identifiers may choose to protect deviants by concealing their
deviant action and/or or to increase their use of exonerating cognitions that
minimize the importance or negativity of the deviant’s actions.
In contrast, positivity identification may be more likely to prompt the black
sheep effect. Persons with high positivity identity are keenly aware of how their
own image is intertwined with that of the group and are more concerned with
protecting their positive social identity than protecting the group itself. They are
therefore more likely to minimize the similarities between themselves and in-group
deviants by distancing themselves from the deviant (i.e. by derogating them) and
13
by concealing connections between themselves and the group. They may also be
more likely to see deviants as an obstacle between themselves and their goals and
are therefore motivated to punish the offenders.
Predictions
The purpose of study 1 is twofold; to test the hypotheses that the strength of
a participant’s identification with the in-group moderates the black sheep effect and
that perceptions of social identity threat mediate this moderation. For this
experiment, I used the typical black sheep paradigm and held certain contextual
features constant: private target evaluations, salient in-group norms, and no
intergroup threat/competition. I predicted that exposure to deviant in-group
members would trigger greater feelings of social identity threat (measured in terms
of affect) than exposure to similarly deviant out-group members. This, however, is
only expected to be true for individuals who identify strongly with the group
dimension, as levels of identification with the in-group are expected to moderate
the black sheep effect. Smaller differences in perceived threat are expected for
participants who do not identify strongly with the group (and therefore have no
reason to feel threatened by the potential aspersion that a deviant member may cast
on the group).
Replicating findings in the black sheep effect paradigm, I expected
evaluations of deviant in-group members to be more negative than evaluations of
similar out-group members, whereas evaluations of non-deviant or positive in-
group members would be more positive than evaluations of similar out-group
14
members. Again, I expected that these results would be moderated by social
identification. Only participants experimentally induced to feel strongly identified
with the group are likely to feel personally threatened when the group’s image is
threatened by a deviant. They are therefore expected to derogate deviant in-group
members in order to try and restore their positive social identity. Individuals who
are induced to feel low levels of identification with the in-group are not expected to
exhibit the black sheep effect.
15
Chapter 1: Study 1
Method
Design and participants. The study consisted of a 2 (target group
membership: in-group member, out-group member) x 2 (target behavior: deviant,
non-deviant) x 2 (social identification of participant: high, low) between-subjects
factorial design. The key dimension of intergroup differentiation was college
students versus non-college students. Materials were identical to those used in a
previous study at the University of Southern California which replicated the black
sheep effect (Stratton, Vasquez, & Miller, in progress). The valence of target
behavior was manipulated by describing a target who followed or transgressed
college student norms. The negative (deviant) target was a person who filed noise
complaints about college students early on weekend nights. Positive (non-deviant)
targets were described as opposing a noise ordinance that is unpopular with college
students.
115 undergraduates from the various colleges in southern California (e.g
University of Southern California, Los Angeles City College, West LA College
participated in the study for course credit. 65.2% of the sample was female. Ages
ranged from 18 to 63 with a mean age of 25.35.
Procedure. Participants were run individually (9%) and in classrooms
(91%).
1
A packet of study materials was created for each of the 8 experimental
1
Setting of experiment (individually vs. in classrooms) was analyzed as a factor in the main analysis,
but has no interaction effects with the study manipulations.
16
conditions. The packets were then organized according to a random table and
handed out in that order to participants.
For the cover story, participants were told that as part of a study on “person
perception” they would be asked to read an article and evaluate the person
described within it. Verbally the procedure was described and all participants were
asked to sign a consent form before they were handed the experimental packets.
On the first page of the experimental packets was a ‘personal information sheet’ the
purpose of which was to obtain general demographic information about the
participant.
Level of overall group identification. The next task consisted of a measure
designed to boost overall identification with the in-group for half of the
participants. In the high social identification conditions, participants were asked to
“please list ten issues or characteristics on which you and other college students
differ from non-college students.” They were then asked to pick one of these issues
and write a short essay explaining why their group’s position or attitude was better.
This manipulation was designed to induce depersonalization and increase in-group
identification (Turner & Oakes, 1989).
2
Participants in the low identification
conditions received a filler task unrelated to group membership. The task asked
them to indicate whether a stove or toaster is better, list ten similarities and
2
In two studies, Turner & Oakes (1989) found that the act of overemphasizing the similarities among
group members and exaggerating the differences between groups increases group identification. Thus,
participants were asked to write an essay that overemphasized the similarities among college students and the
differences between them and non-college students. This manipulation was found to work in a separate
(unpublished) study conducted with a population of students at the University of Southern California for the
group memberships of Republican and Democrat.
17
differences between the two inanimate objects, and write a short essay about why
their choice was the better product. Afterwards, strength of group identification
was measured by Luhtanen and Crocker’s (1992) ‘identity’ subscale of their
collective self esteem measure (see Appendix A), which has been used before in the
black sheep paradigm as a measure of group identification (Biernat et al., 1999).
Presentation of target information. After completing the group
identification measure, participants read an article describing two individuals. The
positive version of the article described a proposed noise ordinance near the college
that would make even small weekend dinner parties against the law. Both a
resident of the area and a neighbor were described as speaking out against the
ordinance. The negative version of the article described a scenario in which a local
college student (the resident) hosted a small dinner party on a weekend night. His
neighbor filed a noise complaint with the police against him. Half of the
participants read that the neighbor was a fellow college student (in-group target),
whereas the other half read that the neighbor was a non-college student (out-group
member).
3
After the participants read the article, they immediately evaluated the
neighbor and then filed out the social identity threat measures, which included a
combination of the Watson, Clark, and Tellegen (1988) PANAS state affect scale,
3
Results of Stratton, Miller, & Lickel’s (under review) meta-analysis on the black sheep effect
suggest that group-specific norm violations are critical for producing the black sheep effect; thus, the target was
described as violated a group-specific norm, turning in a college student for engaging in normal college student
behavior.
18
Luhtanen and Crocker’s (1992) ‘public self esteem’ subscale, and Lickel, et al.’s
(2005) self-image threat items.
Manipulation checks. Participants were asked to identify the group
membership of the neighbor. Only participants who correctly identified the group
membership of the neighbor were included in the analyses (93% of participants).
Dependent measures. For the target evaluations, participants were asked
to evaluate target members on a set of personality traits (e.g. intelligence,
competent, trustworthiness, sincerity, informativeness, sensitivity, etc.). To
measure social identity threat, participants were administered the PANAS mood
scale and the public collective self esteem from Luthanen & Crocker’s (1992)
collective self esteem scale (see Appendices B & C) combined with Lickel, et
al.’s (2005) self-image threat items.
I also included measures of perceived out-group homogeneity and a
measure of linguistic bias (see Appendix D) in order to assess how perceptions of
out-group homogeneity and the use of abstract versus concrete terms in describing
the deviant target affect target evaluations. Past research has shown that high
identifiers are more likely to perceive the in-group as well as the out-group to be
relatively homogeneous whereas low identifiers only emphasize group
homogeneity when the group prospects are good (Doosje, Spears, & Ellemers,
2002). Thus, I expected that high identifiers would perceive more homogeneity
among out-group members than low identifiers when faced with a deviant group
target.
19
Research on the linguistic bias effect shows that positive behavior of in-
group members tend to be described in relatively abstract terms, whereas the same
behavior shown by an out-group member is describe in relatively concrete terms
(Maass, 1999). This may be because concrete language implies low likelihood of
repetition and that less dispositional inferences should be drawn from that one
event, whereas abstract language implies a high likelihood of repetition and
dispositional inference. Describing the positive behaviors of in-group members
abstractly implies that this sort of behavior is typical of group members and will
occur more often. In contrast, using concrete language in describing positive out-
group behaviors indicates that this behavior is atypical and unlikely to reoccur.
These differential uses of abstract and concrete terms in describing the behavior
of two targets that differ only in group membership reflect in-group bias.
Negative behaviors of group members also tend to be described in ways that favor
the in-group over the out-group. Describing the negative behavior of in-group
members in concrete terms implies that this sort of behavior is a one-time offense
that should not reflect on other group members. The negative behavior of out-
group members, however, is more likely to be described in abstract terms,
implying that you can generalize from that person to the group and that that
behavior is likely to be repeated.
Results and Discussion
Manipulation checks. Participants rated the extent to which the target
possessed a series of personality traits from 1 to 9 (1=not at all; 9 = very much).
20
19 items were combined to create the target evaluation index ( = .91). Negative
items were reversed scored before being combined into the index. An
independent samples t-test revealed that the negative (deviant) target was indeed
seen as more negative (M = 4.03) than the positive (non-deviant) target (M =
6.26), t(95) = 9.95, p < .001.
I used the identity subscale from Luhtanen and Crocker’s (1992) collective
self esteem scale (7-pt scale) to check and see whether the identification
manipulation was successful ( = .78). It was not. No differences were found in
the level of identification between participants who wrote the essay designed to
boost their identification as college students and those who did not. The average
level of participant’s identification as college students was significantly higher
than the scale midpoint (M=5.33), t(106) = 14.56, p < .001 indicating that
participants on average were moderately identified with their group membership
as college students.
Because I found no differences in level of identification as a consequence
of my manipulations, I preformed a median split on the identification index to
classify participants as low or high identifiers, resulting in a group of high
identifiers (M = 6.12) and a group of relatively low identifiers (M = 4.56).
Target evaluations. A 2 (target group membership) x 2 (target valence) x 2
(participant identification) between-subjects ANOVA conducted on the overall
trait evaluation index indicated a main effect of target valence F(1, 89) = 104.67,
p < .001,
2
= .54. As predicted, the positive (non-deviant) target was rated more
21
positively (M = 6.29) than negative (deviant) target (M = 4.03). A marginal
interaction of target group membership and target valence F(1, 89) = 3.50, p =
.065,
2
= .038 was also found. The deviant target was rated similarly regardless
of target group membership; surprisingly, out-group normative targets were rated
more positively than in-group normative targets, t(46)= 2.42, p = .019 (see Table
1 for means). No other main effects or interactions were found.
Table 1: Mean trait evaluations of targets as a function of target behavior and
participant’s level of identification with the in-group (Experiment 1)
Target group membership
Target Behavior
Participants level of
identification with in-group In-group Out-group
High Identification
Mean
Std. Dev
N
4.23
1.19
14
3.94
1.33
10
Low Identification
Mean
Std. Dev
N
3.95
.86
13
3.96
1.38
12
Deviant Target
Identification combined
Mean
Std. Dev
N
4.09
1.03
27
3.95
1.33
22
High Identification
Mean
Std. Dev
N
5.77
.77
16
7.08
1.26
9
Low Identification
Mean
Std. Dev
N
6.26
1.06
8
6.39
1.02
15
Non-deviant Target
Identification combined
Mean
Std. Dev
N
5.93
.88
24
6.65
1.14
24
Note: higher scores indicate more positive trait ratings
22
My hypothesis that highly identified group members would derogate a
deviant in-group member more than a similarly deviant out-group member was
not supported. Perhaps this is due to the nature of the population used.
Community college students may not feel as identified with their community
college membership as students of 4-year institutions. Unlike the students at most
4-year college institutions, community college students are more likely to have
dependants and hold jobs that reduce the salience and importance of their group
membership as a college student. They may also not draw much of a distinction
between college students and non-college students. Thus, they may be less likely
to evaluate in-group members and out-group members differently.
An alternative explanation may have to do with the fact that the deviant
behavior used in this case was filing a noise complaint against student residents of
a local apartment building. This sample of community college students was made
up largely of commuter students. Very few of them were likely to live in the area
of the school or next to other students. Therefore, they may not have strong
norms against persons who file a noise complaint against student residents and
would not rate deviant in-group and out-group targets differently. Indeed, they
may view noisy college students negatively, siding with the out-group.
It is surprising that non-deviant out-group members were evaluated more
positively than in-group members. In this situation, a non-college student
publicly offering support to college students and against a local noise ordinance
may have positively violated expectations about non-college students.
23
Expectancy violations for salient group memberships tend to be associated with
more extreme affect-related evaluations (Kernahan, Bartholow, & Bettencourt,
2000). Consequently, the positive violation of expectations about non-college
students may have resulted in the more favorable trait evaluations for this target
than the in-group target.
Table 2: Mean negative mood ratings of participants as a function of target
behavior and participant’s level of identification with the in-group (Experiment 1)
Target group membership
Target Behavior
Participants level of
identification with in-group In-group Out-group
High Identification
Mean
Std. Dev
N
2.07
.73
14
1.55
.60
11
Low Identification
Mean
Std. Dev
N
2.00
.94
14
1.87
.94
13
Deviant Target
Identification combined
Mean
Std. Dev
N
2.03
.82
28
1.73
.80
24
High Identification
Mean
Std. Dev
N
1.93
1.00
16
1.65
.64
10
Low Identification
Mean
Std. Dev
N
1.34
.32
9
2.27
1.35
15
Non-deviant Target
Identification combined
Mean
Std. Dev
N
1.72
.86
25
2.02
1.14
25
Note: higher scores indicate more negative moods
24
Participants were presented with two measures of social identity threat: 1) a
mood scale and 2) the public collective self esteem from Luthanen & Crocker’s
(1992) collective self esteem scale combined with Lickel, et al.’s (2005) self-image
threat items. All 12 negative mood items were combined to create an index of
negative mood ( = .88). A 2 (target group membership) x 2 (target valence) x 2
(target behavior) ANOVA conducted on this index revealed a marginal interaction
between target group membership and target behavior valence, F(1, 94) = 3.06, p
=.08,
2
= .032 and a reliable interaction between target group membership and
participant identification, F (1,94) = 4.71, p = .033,
2
= .048 (see Table 2 for
means). However, independent samples t-tests found no differences between target
group memberships in either the positive or deviant group target conditions when
participant level of identification was collapsed over.
The second index of social identity threat created by combining Luhtahen
and Crocker’s 4-item public self esteem scale and Lickel, et al.’s (2005) three self
image threat items ( = .73). A 2 (target group membership) x 2 (target valence) x
2 (target behavior) ANOVA conducted on this index revealed no main effects or
interactions.
It is possible that no differences were found on either of these social identity
threat items because the act of evaluating the target vitiated any feelings of social
identity threat. It is also possible that this community college sample did not react
to the target manipulations because the deviant behavior was not a strong enough
25
manipulation for a non-residential college sample or because the intergroup
dimension was not important/meaningful for them.
Out-group homogeneity. A 2 (target group membership) x 2 (target
behavior valence) x 2 (participant identification) between-subjects ANOVA on
the out-group homogeneity item revealed a marginal interaction between group
membership and target behavior valence, F(1,94) = 3.19, p = .077,
2
= .033.
Independent samples t-tests revealed that when the target behaved in a normative
manner, presentation of an out-group target boosted perceptions that the out-
group (non-college students) as highly similar, t(47) = 1.98, p =.05 (M = 4.42 and
M = 3.30 respectively). There was no effect of target group membership on
perceptions of out-group homogeneity when the target behaved in a deviant
manner.
Linguistic in-group bias. Participants were asked to write an open-
response description of the target individual. Four coders read each essay and
coded the descriptions for their use of abstract and concrete terms according to
Maass’ (1999) suggestions. The sentence was coded as 1 if the description was
concrete and used descriptive action verbs. Sentences with the most abstract
terms (adjectives delineating general dispositions) were coded as a 4. A mean
score of language abstraction was calculated for each target description and the
ratings of the four coders were combined ( = .99). Research on the linguistic
bias effect shows that positive behavior of in-group members tends to be
described in relatively abstract terms, whereas the same behavior shown by an
26
out-group member is described in relatively concrete terms (Maass, 1999).
However, an independent samples t-test revealed that the behavior of the non-
deviant out-group target was more likely to be labeled in more abstract terms (M
= 2.07) than the behavior of the non-deviant in-group target (M = 1.57), t(46) =
1.98, p = .057. This indicates that participants attributed the non-deviant behavior
of out-group targets to enduring characteristics and/or psychological states
(Maass, 1999). This mirrors the overly positive trait evaluations given to the non-
deviant target over the deviant target, which can perhaps be explained by
expectancy violation theory (Kernahan, Bartholow, & Bettencourt, 2000). No
differences in terms were found between deviant in-group and out-group targets.
Ancillary analyses. Since the social identity theory measures did not work,
I explored the hypotheses that perceptions of out-group homogeneity or that the
attribution of non-deviant behavior to enduring characteristics of the target
(denoted by variation in description of the target’s behavior) mediate the more
positive evaluations seen in the non-deviant / out-group target cell compared to
the non-deviant / in-group target condition (Baron & Kenny, 1986). However,
inspection of a correlation matrix indicated that neither of these variables were
correlated with target trait evaluations.
27
Chapter 2: Study 2
In Study 1, I hypothesized that overall high identifiers would be more
threatened (measured by the social identity threat scales) by the behavior of deviant
or negative in-group members than low identifiers and therefore more likely to
derogate offending in-group members relative to similar out-group members. This
hypothesis was not supported, possibly due to artifacts in the sample. Study 2
sought to examine whether certain circumstances can lead to a reversal of this
effect. The results of an unpublished meta-analysis (Stratton, Miller, & Lickel,
under review) indicate that the amount of publicity surrounding the target’s deviant
behavior is negatively correlated with black sheep effect sizes. In addition, many
of the highly public conditions showed a reversal of the black sheep effect (i.e.
protection of the in-group deviant). In other words, participants showed a devil
protection effect. Perhaps participants are concerned that their own behavior will
be public, if the target has already drawn a lot of notice from others. When
participants’ evaluation of in-group targets is likely to become public, especially to
potential out-groupers, I predicted that high identifiers would be less likely than
low identifiers to exhibit the black sheep effect. In fact, I expected that they would
show the devil protection effect in public. When their target evaluations are public,
participants may decide it is more important to maintain the image of a united
group front rather than let the in-group seem disunited by derogating any in-group
member regardless of their behavior. In a public context, high identifiers may be
especially motivated to ‘stick to the group’ and defend its interests even when
28
doing so could result in a negative evaluation of the group or themselves (Ellmers
et al., 1997). The motivation to maintain a public united front may therefore
outweigh the motivation to reduce social identity threat through in-group member
derogation. In contrast, low identifiers tend to be more opportunistic in their public
evaluations, making attributions and evaluations that portray themselves in the best
possible light with respect to their specific audience (Postmes, et al., 1999). Thus,
they may be more likely to publicly distance themselves from any negative target
through negative evaluations of those targets, but especially from negative targets
who may be perceived as more similar to themselves. Study 2 was constructed to
test these hypotheses.
Method
Design and participants. 107 undergraduates from the University of
Southern California (18.3%) and California State University, Northridge (81.7%)
participated in the study for course credit and were included in the final analysis
4
(19 participants from the original set were excluded because they were suspicious
about the experimental manipulations, failed to follow instructions, or because they
were outliers).
The study consisted of a 2 (target group membership: in-group member,
out-group member) x 2 (strength of overall social identification of participant: high,
4
Source of participants (University of Southern California vs. California State University,
Northridge) was analyzed as a factor in the main analysis, but had no reliable interaction effects with the study
manipulations.
29
low) x 2 (evaluation context: public, private) between-subjects factorial design.
The key dimension of intergroup differentiation was college student versus non-
college student. Since the treatment of group deviants is mainly a prediction that
negative or deviant target members will be evaluated differentially depending on
their group membership, the target behavior was negative and constant across all
conditions.
Procedure. The procedure used for this study was similar to that of study 1.
The first author met with each participant individually to conduct the study. They
were first informed of the purpose and procedure of the study and asked to sign a
consent form. After filling out some demographic information about themselves,
half of the participants were given a task designed to boost their overall
identification with the in-group (see Study 1). Participants in the low identification
conditions received a filler task unrelated to group membership. All participants
then completed the ‘identity’ subscale of Luhtanen and Crocker’s (1992) collective
self esteem measure as a group identification manipulation check.
Publicity manipulation. Upon presentation of the target information,
participants in the high publicity conditions were told that their target evaluations
would be collected and published in special edition piece of the local school
newspaper. In these conditions, participants were told that they would have to
indicate and discuss their responses publicly with other participants in the
experiment (in-group as well as out-group members) at the end of the experimental
session (see publicity manipulation in Ellemers, et al., 2000).
30
Participants in the low publicity condition were told that their target
evaluations would be anonymous. After reading the article and filling out the
target evaluation measures (which contained no identifying information), they
were asked to seal their responses in an unmarked envelope and place it in a
basket full of other unmarked envelopes. To increase the perception of
anonymity, participants in this condition were told that their responses would be
coded by a research assistant and will not be seen by the experimenter or anyone
else.
All participants were presented with the negative article from Study 1
followed by the target evaluation measures (as in Study 1, half of the participants
read about an in-group target and half about an out-group target). Upon
completion of these tasks, all participants were then given a questionnaire
containing additional manipulation checks and several items designed to assess
the motivations underlying target evaluations. Finally, the experimenter returned
to collect final questionnaire, probe the participants for suspicion, and debrief
them.
Final manipulation checks. On the final questionnaire, participants were
asked to indicate “how likely is it that other persons would be aware of what
comments you made about the target?” on a 7-pt. linear scale (1 = no one else, 7
= lots of other people). All participants were also asked to identify the group
membership of the target (college student vs. non-college student).
31
Dependent measures. The target evaluation questionnaire was the same
as study 1. In the final questionnaire, participants were also asked several
questions regarding their motivations for evaluating the target in the way that
they did (see Appendix D). For example, they were asked “when you made
ratings about the target, to what extent were you concerned with presenting a
united front to outsiders,” to what extent would you like to hide the behavior of
the target from other people,” “would you prefer that the target be expelled from
college,” “how concerned were you that the target might be mistakeningly seen
as representative of you,” and “how important is the target’s act to you?.”
I also included measures of perceived out-group homogeneity and a
measure of linguistic bias (see Appendix D) in order to assess how perceptions of
out-group homogeneity and the use of abstract versus concrete terms in
describing the deviant target affect target evaluations. Past research has shown
that high identifiers are more likely to perceive the in-group as well as the out-
group as relatively homogeneous whereas low identifiers only emphasize group
homogeneity when the group prospects are good (Doosje, Spears, & Ellemers,
2002). Thus, when faced with a deviant in-group target, I expected high
identifiers to perceive more homogeneity between out-group members than low
identifiers.
In regards to the linguistic bias measure, I expected high identifiers to use
different words when describing in-group deviants in a public versus private
context. In a public context, high identifiers are expected to protect the in-group
32
deviants and one manner in which they may do so is by minimizing the
negativity or importance of the deviant ‘member’ or ‘act’. They should therefore
be more likely to use concrete terms when describing in-group deviants in a
public context versus a private context because concrete language implies low
likelihood of repetition and a lower likelihood that dispositional factors cause the
event (Maass, 1999).
Results and Discussion
Manipulation checks. Participants were asked to rate on a 7-pt scale
“how many other persons will know about the evaluations you made today about
the persons in the article” (1 = no one else, 7 = lots of other people). Participants
assigned to the high publicity conditions rated this item higher (M = 6.13) than
participants in the low publicity condition (M = 1.56), t(106) = 21.34, p < .001.
I used the same measures of social identification as study 1.
Paradoxically, participants assigned to the low identification conditions reported
feeling prouder to be a college student (M = 6.74) than participants assigned in
the high identification conditions (M = 6.47) t(110) = 2.39, p = .019. No
differences were found in the level of identification between participants on the
other items or the composite index. The overall average level of participant’s
identification measured by the 7-item composite ( = .77) was significantly
higher than the scale midpoint (M = 5.21), t(109) = 13.52, p < .001 indicating
that participants on average were moderately identified with their group
membership as college students.
33
Because I found little difference in the levels of identification as a
consequence of my manipulations, I preformed a median split on the
identification index to classify participants as low or high identifiers, resulting in
a group of relatively low identifiers (M = 3.2) and high identifiers (M = 4.52).
Target evaluations. Participants rated the extent to which the target
possessed a series of personality traits from 1 to 9 (1 = not at all; 9 = very much).
19 items were combined to create the target evaluation index ( = .66). Negative
items were reversed scored before being combined into the index. A 2 (target
group membership: in-group member, out-group member) x 2 (strength of overall
social identification of participant: high, low) x 2 (evaluation context: public,
private) between-subjects ANOVA conducted on this index indicated a main
effect for group F(1,103) = 4.16, p = .044,
2
= .039. In-group targets were rated
more positively (M = 4.19) than out-group targets (M = 3.92). There was also an
interaction between publicity and participant identification, F(1, 103) = 4.64, p =
.034,
2
= .043 (see Table 3 for means); when participants thought that their
evaluations would remain anonymous, low identifiers rated targets marginally
more positively (M = 4.20) than high identifiers (M = 3.88), t(56) = 1.78, p = .08.
I initially hypothesized that high identifiers would be less likely than low
identifiers to exhibit a black sheep effect when target evaluations are believed to
be public. In fact, I theorized that high identifiers would be likely to show a devil
protection effect (i.e. ‘protect’ in-group deviants with more positive evaluations)
when they think their evaluations will be made public, in order to avoid further
34
weakening the group, whereas low identifiers would distance themselves from
the in-group deviant by exhibiting the black sheep effect. My hypotheses were
partially supported.
Table 3: Mean trait evaluations of targets as a function of context of target
evaluations and participant’s level of identification with the in-group (Experiment
2)
Target group membership
Context of target
evaluations
Participants level of
identification with in-group In-group Out-group
High Identification
Mean
Std. Dev
N
4.44
.79
13
3.99
.85
12
Low Identification
Mean
Std. Dev
N
4.12
.73
14
3.85
.71
11
High publicity
Identification combined
Mean
Std. Dev
N
4.27
.76
27
3.87
.72
26
High Identification
Mean
Std. Dev
N
3.92
.69
15
3.82
.71
11
Low Identification
Mean
Std. Dev
N
4.30
.83
17
4.05
.46
15
Low publicity
Identification combined
Mean
Std. Dev
N
4.13
.78
32
3.97
.58
26
Note: higher scores indicate more positive trait ratings
I found that participants were likely to favor deviant in-group members
(M = 4.27) over deviant out-group members (M = 3.87) regardless of their
35
strength of identification with the in-group when they thought their evaluation
would be made public, t(51) = 1.96, p = .055. In contrast, participants did not
differentially evaluate the deviant in-group and out-group target when they
thought their evaluations would remain anonymous.
Motivation Analyses. To investigate the motivations behind differential
target evaluations in different evaluation contexts, I analyzed the motivation
questions in the final questionnaire with a series of independent samples t-tests. In
the public context, high identifiers were expected to report wanting to project a
united group front and hide the behavior of in-group deviants more than low
identifiers. Though the means were in this direction (M = 3.59 and 3.00
respectively), this difference was not reliable. In contrast, low identifiers were
expected to be more concerned that the target might be mistakenly seen as
representative of themselves and might see the target’s act as more important than
high identifiers, however independent samples t-tests revealed no differences on
these items. My hypotheses in these regards were not supported. Perhaps the
motivations between target evaluations are largely an unconscious process and
therefore cannot be measured explicitly. Alternatively, participants may wish to
be perceived as impartial person evaluations and may not want to admit that their
evaluations of others are affected by group image concerns.
Since I found that participants protect in-group deviants in a public
evaluation context regardless of their level of identification with the in-group, I
conducted a series of independent samples t-tests on the motivation items to
36
account for this finding. The only item which showed any difference was the
group expulsion item. Participants were less likely to recommend that a deviant
in-group target be expelled from their group than the similar deviant out-group
target, though this tendency was unrelated to target evaluations.
Linguistic in-group bias. The linguistic bias measure was coded as
described in study 1 ( = .98). A 2 (target group membership: in-group member,
out-group member) x 2 (strength of overall social identification of participant:
high, low) x 2 (evaluation context: public, private) between-subjects ANOVA
conducted on this index, indicated no main effects or interactions. I expected an
independent samples t-test to show that high identifiers who make target
evaluations in a public context are more likely to use concrete language than low
identifiers when describing the deviant in-group member, implying that the
behavior has a low likelihood of repetition and that dispositional inferences should
not be made on the basis of this target’s deviant act (Maass, 1999). However, no
such difference was observed, t(38) = .63, p = .53. No differences in descriptive
terms were found between deviant in-group and out-group targets in either of the
publicity conditions.
Out-group homogeneity. In regards to the homogeneity measure, out-
groups are typically seen as more homogenous than in-groups; however,
highlighting a deviant member of either group is likely to boost perceptions of its
heterogeneity. A 2 (target group membership: in-group member, out-group
member) x 2 (strength of overall social identification of participant: high, low) x 2
37
(evaluation context: public, private) between-subjects ANOVA conducted on this
item, indicated no main effects or interactions. High identifiers, who in a public
context are expected to be more concerned with projecting a uniform group image,
may actually reduce relative perceptions of in-group heterogeneity and see the out-
group as more variable. However, an independent samples t-test found no
differences between the out-group homogeneity estimates of high identifiers in the
low publicity conditions to those of the high identifiers in the high publicity
conditions.
Because I found that participants protect in-group deviants in a public
evaluation context regardless of their level of identification with the in-group, I
conducted an independent samples t-test to determine if the public evaluation of a
deviant in-group target rather than a deviant out-group target changed perceptions
of the out-group’s homogeneity. There were no differences in ratings of out-group
homogeneity between these conditions.
38
Chapter 3: Study 3
In Study 2, I hypothesized that a public or private context of target
evaluations would produce opposite judgments of in-group deviants relative to out-
group deviants. Under private conditions, high identifiers were predicted to show
the black sheep effect. In contrast, under public conditions high identifiers were
expected to exhibit the devil protection effect. I found partial support for these
hypotheses. In the public evaluation context, both high and low group identifiers
protected the in-group deviant by evaluating him/her more positively than the out-
group deviant. No such devil protection effect was found in the private target
evaluations conditions.
These findings support the idea that the same target can elicit different types
of responses depending on the context on target evaluations. However, real world
phenomena suggest that even when the evaluation context is held constant, the
same target can evoke disparate responses from other in-group members. In the
case of Ted Haggard, some evangelical Christians openly derided him and called
for his expulsion from the religious organization, whilst others publicly announced
their forgiveness of him and their support and love for him. In this case, it may be a
characteristic of the perceiver and not the evaluation context that moderates the
type of extreme response directed at the deviate.
Study 3 is aimed at exploring whether the type of identification one has
with the in-group moderates evaluative judgments of in-group and out-group
deviants. I predict that persons who identify highly with a group because they feel
39
a strong connection with the group and the ideology associated with the group
(connected identifiers) are more likely to ‘stick together’ and show solidarity when
threats to the group’s identity exist by protecting the in-group deviant (even if the
group threat stems from the actions of that in-group member). In contrast, persons
who identify highly with a group because of its positive image (positivity
identifiers) are expected to show the black sheep effect and derogate the in-group
deviant in relation to the out-group deviant.
Method
Design and participants. The study consisted of a 2 (target group
membership: in-group member, out-group member) x 2 (type of identification of
participant: personal connection to group identification, positivity identification)
between-subjects factorial design. This time the key dimension of intergroup
differentiation was conservative versus liberal. The deviant behavior was showing
disloyalty to the one’s own political party.
62 participants completed the measures over the internet in exchange for a
chance at a lottery prize. Three participants were excluded because they failed to
correctly remember the group membership of the target. Forty-six of the
participants were female, 12 were male (1 participant refused to report gender).
Participants were between the ages of 18 and 70 with a mean age of 31.28. 69.5%
of the sample identified themselves as Liberal with the rest identifying themselves
as Conservative.
40
Procedure. Participants were recruited through listserve emails asking
them to complete the study for a chance at a lottery prize and requesting that they
send the link on to anyone else they knew might be interested. The email contained
a link to the internet study. The first webpage described the purpose and procedure
of the study and had participants clink on the link below to indicate that they were
over 18 and had consented to participate. Next they filled about a series of
demographic questions about themselves, which included an item assessing their
political affiliation (conservative or liberal). I deliberately gave participants only
two choices for political affiliation (though participants could skip this question, in
which case they were excluded from analyses). This means that even people with
weak political affiliation would have to choose one of these categories; however,
that should make the manipulation of type of identification even stronger for these
participants.
Type of group identification. Type of group identification (connected
identification versus positivity identification) was manipulated in a bogus pipeline
procedure similar to Ellemers, Spears, and Doosje’s (1997) strength of social
identity manipulation. Participants were asked to indicate on a 9-pt scale (1 = not
at all, 9 = very much) to what extend they agree with a number of general
statements that indirectly refer to group members (see Appendix E). Participants
were led to believe that their type of group identification would be determined by
their answers to the questionnaire combined with general demographic
information about themselves (their gender, age, place of birth). I deliberately left
41
the exact manner of calculating type of group identification unspecified.
Participants in the personal connection to group identification conditions were
told, “Your answers indicate that you are a CONNECTED identifier. You are
highly committed to the groups of which you are part. You are attracted to
membership in different groups because of what the group stands for when it
comes to equality, individualism, justice, etc. in relation to your own personal
beliefs. Most people who are highly connected identifiers join groups that are
central their self identities. You spend time considering what you stand for and
what the group stands for before you join. Once you join a group you tend to stay
with it long term. Your group memberships define who you are and you feel a
strong connection with almost all the groups you are apart of.” Participants in the
positivity identification conditions were informed that they have high levels of
positivity identification with the social groups of which they are part, “Your
answers indicate that you are an INSTRUMENTAL identifier. You are attracted
to membership in different groups based on what the group can do for you and
how your membership within it can advance your personal and professional
goals. Most people who have high instrumental identification see groups as
vehicles for many people to achieve the same goals, like increasing job security or
getting an edge on the competition. You spend time considering the group’s
image, how it is perceived by others, and what it stands for before you join. You
are constantly aware of the image the group is projecting to others. Almost all of
the groups you belong to are viewed positively by yourself and others.”
42
Presentation of target information. After being informed of their type of
group identification, participants were presented with article which described a
target who was disloyal to his political affiliation (see Appendix G). Half of the
participants were told that the target was an in-group member (according to the
political affiliation they had indicated in the first page of the study), whereas the
other half of participants was told that the target was an out-group member. After
reading the article, participants filled out the dependent measures and the rest of
the manipulation checks. Finally, they were thanked for their participation and
shown a debriefing statement.
Manipulation checks and dependent measures. After they evaluated the
article, participants were asked to complete the party affiliation scale in which
they restated their political affiliation (conservative or liberal) and filled out items
from Luhtanen and Crocker’s (1992) collective self esteem scale designed to
assess whether they indeed felt high connection identification with their in-group
or high positivity identification (see Appendix F).
5
As in studies 1 and 2,
participants were asked to identify the group affiliation of the target. The
dependent measures were evaluations of target members on a set of personality
traits (e.g. intelligence, competent, trustworthiness, sincerity, informativeness,
sensitivity, etc.).
5
Items for the connection identification type were taken from the identification subscale that assesses
the centrality of the group to the person’s identity. Items for the positivity identification type were taken from
the public self esteem scale that assesses one’s perceptions of how positively other people evaluate one’s social
groups.
43
Results and Discussion
Manipulation checks. Under the guise of a memory quiz, participants
were asked to answer yes or no to a series of questions about the target person.
One of these items was a question asking about the group membership
(conservative or liberal) of the target. All but three participants answered this
item correctly. As mentioned above, the three participants who failed to
remember the group membership of the target were excluded from analysis.
As a manipulation check on type of group identification, participants were
asked to rate their agreement with a series of statements about their connection
with their political group. The personal connection to group identification was
measured by an item taken from Luhtanen and Crocker’s (1992) importance to
identity subscale, “Overall, my group membership as a [Liberal/ Conservative]
has very little to do with how I feel about myself.” Persons in the connection
identification conditions reported greater agreement with this item (indicating a
greater connection to the group identity) than persons in the positivity
identification conditions, t(56) = 2.29, p = .026. An item from Luhtanen and
Crocker’s (1992) public self esteem subscale was used to check strength of
positivity identification, “Most people consider Liberals, on the average, to be
more effective than other social groups.” Persons in the positivity identification
conditions reported marginally greater agreement with this item (indicating that
the group has a higher positive image) than persons in the connection
identification conditions, t(56) = 1.76, p = .08.
44
Trait evaluation. Participants rated the extent to which the target
possessed a series of personality traits from 1 to 9 (1=not at all; 9 = very much).
19 items were combined to create the target evaluation index ( = .68). Negative
items were reversed scored before being combined into the index. A 2 (target
group membership: in-group member, out-group member) x 2 (type of
identification of participant: personal connection to group identification, positivity
identification) between-subjects ANOVA conducted on this index indicated a
reliable main effect for target group membership, F(1,53)=6.091, p = .017,
2
=
.10, indicating that in-group members were evaluated more positively than out-
group members. More important, however, there was an interaction between
target group membership and type of participant identification, F(1, 53)=5.15, p =
.027,
2
= .089. As predicted, when participants were induced to feel a connected
identification with their in-group they showed in-group bias by evaluating the
deviant in-group target more positively (M = 5.65) than the similarly deviant out-
group target (M = 4.25), t(26) = 3.25, p = .003. However, my second hypothesis,
that participants induced to have a positivity identification would exhibit a black
sheep effect, was not supported. There were no differences between evaluations
of the deviant in-group and out-group targets (see Table 4) in the positivity
identification conditions.
These results support the notion that persons who feel a tight connection to
a group are likely to exhibit the devil protection effect when faced with a deviant
in-group target. This suggests that when their positive social identities are
45
threatened by the negative behavior of in-group members, these types of identifiers
choose to protect the deviant and exhibit group solidity rather than distance
themselves from the group or group members.
Table 4: Mean trait evaluations of targets as a function of participant’s type of
identification with the in-group (Experiment 3)
Target group membership
Participant’s type of
identification with in-group In-group Out-group
Connected Identification
M
SD
n
5.65
1.18
17
4.25
.98
11
Positivity Identification
M
SD
n
5.14
1.38
14
5.08
.71
15
Note: higher scores indicate more positive trait ratings
Linguistic in-group bias. The linguistic bias measure was coded as
described in study 1 ( = .95). A 2 (target group membership) x 2 (type of
participant identification) ANOVA on this composite revealed only a main effect
for target group membership, F(1,40) = 5.63, p = .02,
2
= .12. The deviant out-
group target was described more abstractly (M = 2.41) than the similarly deviant
in-group target (M = 3.24). This coincides with findings by Maass (1999), who
theorizes that the negative behavior of out-group members is described more
abstractly than the negative behavior of in-group members in order to imply to
others that out-group members are prone to undesirable behavior whereas the
undesirable behavior of an in-group member is an isolated event unrelated to the
general disposition of in-group members.
46
Out-group homogeneity. Out-groups are typically seen as more
homogenous than in-group; however, I predicted that highlighting a deviant
member of either group would boost perceptions of its heterogeneity. This
difference was not found. A 2 (target group membership) x 2 (type of participant
identification) ANOVA performed on this item revealed no main effects or
interaction.
Ancillary analyses. I next explored the hypothesis that the attribution of
deviant behavior to enduring characteristics of the target (denoted by variation in
concrete versus abstract description of the target) mediate the more positive
evaluations seen in the connected identifier / out-group target cell compared to the
connected identifier / in-group target condition (Baron & Kenny, 1986). The
connected identifier / in-group target condition was associated with higher
evaluations of the target, = .54, p = .003 and was also associated with a
decreased attribution of deviant behavior to enduring characteristics of the target,
= -.52, p = .009. Decreased attribution of the deviant behavior to enduring
characteristics of the target was associated with higher trait evaluations, = -.45,
p = .029. When controlling for attribution of deviant behavior to enduring
characteristics of the target, target group membership was no longer associated
with higher evaluations of the target, = .33, p = .14. However, neither did
attribution of deviant behavior to enduring characteristics of the target remain a
signification predictor of target evaluations, = .27, p = .23; indicating the
mediational implication of the preceding analyses to be untenable.
47
Chapter 4: General Discussion
In Study 1, I hypothesized that highly identified group members would
derogate deviant in-group members more than similarly deviant out-group
members. However, likely due to confounds in the strength of implementation of
participant’s level of identification and the specific dimension of intergroup
differentiation (i.e. college students versus non-college students), this hypothesis
was not supported. Specifically, community college students may not readily
identify as college students or see college students versus non-college students as
a valid intergroup differentiation. They may perceive little or no differences
between the two groups and may, in fact, see neither as a ‘real’ (entitative) group.
Thus, they would be unlikely to differentiate between in-group and out-group
targets when rating them on a set of personality traits. Future research might do
well to pretest the population to be studied on their perceptions of the entitativity
(i.e. groupiness) for various groups so that more relevant groups can be utilized in
these types of investigations.
An intriguing and important question arises from the finding that non-
deviant out-group members were evaluated more positively than non-deviant in-
group members in study 1. Expectancy violation theory (Kernahan, Bartholow, &
Bettencourt, 2000) may account for this observation of greater favorability
towards the out-group target than in-group target. In this situation, a non-college
student publicly offering support to college students by railing against a local
noise ordinance that college students ordinarily oppose may have positively
48
violated expectations about non-college students. Expectancy violation theory
posits that expectancy violations provoke more extreme trait evaluations in the
direction of the valence of the violation. Thus, the positive expectancy violation
of the non-deviant out-group target in this study resulted in more positive ratings
of that target than the non-deviant in-group target. However, future research is
necessary to support this tenet.
Study 2 found that both high and low identifiers ‘protected’ in-group
deviants with more positive evaluations when they thought that their evaluations
would be made public. No evaluation difference was observed between targets
for private evaluations. Thus, this indicates that group members are highly
sensitive to the evaluation context when they are evaluating group deviants. This
phenomenon may stem from concerns that negative evaluations of in-group
members may hurt the image of the group among outsiders and therefore could
negatively affect the power of the group. This has implication for real world
situations. With modern technology that can globally broadcast human folly and
group members’ reactions to it, we as researchers would do well understand why
some group members may choose to support in-group deviants even when that
person’s behavior jeopardizes the group’s positive image.
Study 3 suggests that deviant group members provoke different reactions
from perceivers depending on the type of identification the perceiver feels
towards the in-group. Connected identifiers exhibit a devil protection effect.
They protect in-group deviants with more positive evaluations than similarly
49
deviant out-group members. In contrast, persons who identify with a group for
instrumental concerns do not differentially evaluate deviant in-group and out-
group members. This suggests that when their positive social identities are
threatened by the negative behavior of in-group members, connected identifiers
choose to protect the deviant and exhibit group solidity rather than distance
themselves from the group or group members. Instead of derogating the in-group
deviant, they may use exonerating cognitions which minimize the importance or
negativity of the deviant behavior of in-group members. Though these
suppositions were not tested in the present experimental paradigm, the use of
exonerating cognitions to justify deviant in-group member behavior would be a
good avenue for future research. Understanding why certain group members
chose to protect group deviants rather than ostracize them has important
connotations for understanding intragroup relations.
Suggestions for Future Research
Studies 1 and 2 were limited by the ineffectual experimental manipulation
of identification. It is difficult to manipulate, even in the short duration of an
experimental session, how much identification a person feels with a group which
they have long standing feelings, experiences, and opinions. However, the bogus
feedback manipulation employed in Study 3 seems to have been effective in
manipulating the type of identification felt towards the in-group. Future research
would do well to use this manipulation when attempting to boost or reduce
participant’s overall level of social identification. As in Study 3, participants
50
could be administered a bogus identification questionnaire and then, in accord
with their experimental conditions, be given feedback about their overall strength
of identification with the in-group
Study 1 pointed out that in this experimental paradigm it is also important
to make sure that the group memberships employed are perceived to be actual
groups by the study population and that the deviant behavior is perceived by the
population in question to be a group specific norm violation. This is most likely
to be accomplished by extensively pretesting each sample. After generating a list
of potential groups, the research team should ask a subset of the population of
interest to rate the degree to which each group is perceived to be a distinct group.
A separate subset of the research population could then be given the list of groups
perceived as distinct groups by the earlier population subset (denoted by a high
mean ‘groupiness’ score) and asked to generate a list of in-group norms for each
group on the list. Based on the results of these two pretests, a future research
team could create a target description which possesses both a group membership
that is well recognized by the population and a behavior that is perceived to be a
strong group-specific norm violation; two conditions essential for replicating the
black sheep paradigm.
51
Conclusions
In sum, the present research offers support to the idea that evaluation
context and type of group identification affect perceptions and evaluations of in-
group and out-group deviants. Both findings have implications to real world
situations and therefore warrant future research.
52
References
Abrams, D., Marques, J. M., de Moura, G. R., Hutchison, P., & Bown, N. J.
(2004). The maintenance of entitativity: A subjective group dynamics
approach. In V. Yverbyt, C. M. Judd, & O. Corneille (Eds.) The
Psychology of Group Perception: Perceived Variability, Entitativity, and
Essentialism (pp. 361-380). New York, New York: Psychology Press.
Anderson, C. A., Deuser, W., E., & DeNeve K. M. (1995). Hot temperatures,
hostile affect, hostile cognition, and arousal: Test of a general model of
affective aggression. Personality and Social Psychology Bulletin, 21, 434-
448.
Ashmore, R. D., Deaux, K., & McLaughlin-Volpe, T. (2004). An organizing
framework for collective identity: Articulation and significance of
multidimensionality. Psychological Bulletin, 130, 80-114.
Billig, M., & Tajfel, H. (1973). Social categorization and similar in intergroup
behavior. European Journal of Social Psychology, 3, 27-52.
Branscombe, N. R., Wann, D. L., Noel, J. G., & Coleman, J. (1993). In-group or
out-group extremity: importance of the threatened social identity.
Personality and Social Psychology Bulletin, 19, 381-388.
Brewer, M. B. (1998). In-group bias in the minimal intergroup situation: A
cognitive-motivational analysis. Psychological Bulletin, 86, 307-324.
Brewer, M. B., & Brown, R. J. (1998). Intergroup relations. In D. T. Gilbert, S.
Fiske, & L. Gardner (Eds.), The Handbook of Social Psychology, Vol. 2 (4
th
Ed) (pp. 554-594). New York: McGraw-Hill.
Buckingham, J. T., & Alicke, M. D. (2002). The influence of individual versus
aggregate social comparison and the presence of others on self-evaluations.
Journal of Personality and Social Psychology, 83, 1117-1130.
Cialdini, R. B., Borden, R. J., Thorne, A., Walker, M. R., Freeman, S., & Sloan, L.
R. (1976). Basking in reflected glory: Three (football) field studies.
Journal of Personality and Social Psychology, 34, 366-375.
Cialdini, R. B., & Richardson, K. D. (1980). Two indirect tactics of image
management: Basking and Blasting. Journal of Personality and Social
Psychology, 39, 406-415.
53
Cohen, G. L., & Garcia, J. (2005). “I am us”: Negative stereotypes as collective
threats. Journal of Personality and Social Psychology, 89, 566-582.
Coull, A., Yzerbyt, V. Y., Castabim E., Paladino, M., & Leemans, V. (2001).
Protecting the ingroup: Motivated allocation of cognitive resources in the
presence of threatening ingroup members. Group Processes & Intergroup
Relations, 4, 327-339.
De Kramer, D., & Van Vugt, M. (1999). Social identification effects in social
dilemmas: a transformation of motives. European Journal of Social
Psychology, 29, 871-893.
Doosje, B., Spears, R., & Ellemers. (2002). Social identity as both cause and
effect: The development of group identification in responses to anticipated
and actual changes in the intergroup status hierarchy. British Journal of
Social Psychology, 41, 57-46.
Eidelman, S., & Biernat, M. (2003). Derogating black sheep: Individual or group
protection? Journal of Experimental Social Psychology, 39, 602-609.
Elder, T. J., Sutton, R. M., & Douglas, K. M. (2005). Keeping it to ourselves:
Effects of audience size and composition on reaction to criticisms of the
ingroup. Group Processes and Intergroup Relations, 8, 231-244.
Ellemers, N., Kortekaas, P., & Ouwerkerk, J. W. (1999). Self-categorisation,
commitment to the group and group self-esteem as related by distinct
aspects of social identity. European Journal of Social Psychology, 29, 371-
389.
Ellemers, N., Spears, R., & Doosje, B. (2002). Self and social identity. Annual
Review of Psychology, 53, 161-186.
Ellmers N., Spears, R., & Doosje, B. (1997). Sticking together or falling apart: In-
group identification as a psychological determination of group commitment
versus individual mobility. Journal of Personality and Social Psychology,
72, 617-626.
Ellmers, N., van Dyke, C., Hinkle, S., & Jacobs, A. (2000). Intergroup
differentiation in social context: Identity needs versus audience constraints.
Social Psychology Quarterly, 63, 60-74.
Festinger, L. (1950). Informal social communication. Psychological Review, 57,
271-282.
54
Haslam, N., Rothschild, L., & Ernst, D. (2000). Essentialist beliefs about social
categories. British Journal of Social Psychology, 39, 113-127.
Hirt, E. R., Zillman, D., Erickson, G. A., & Kennedy, C. (1992). Cost of benefits
of allegiance: Changes in fans’ self-ascribed competencies after team
victory versus defeat. Journal of Personality and Social Psychology, 63,
724-738.
Hornsey, M. J., Jetten, J. (2003). Not being what you claim to be: Impostors as
sources of group threat. European Journal of Social Psychology, 33, 639-
657.
Hornsey, M. J., Oppes, T., & Svensson, A. (2002). ‘It’s OK if we say it, but you
can’t’: Responses to intergroup and intragroup criticism. European Journal
of Social Psychology, 32, 293-307.
Jackson, L. A., Sullivan, L. A., Hodge, C. N. (1993). Stereotype effects on
attributions, predictions, and evaluations: No two social judgments are quite
alike. Journal of Personality and Social Psychology, 65, 69-84.
Jetten, J., Schmitt, M. T., Branscombe, N. R., & McKimmie, B. M. (2005).
Suppressing the negative effect of devaluation on group identification: The
role of intergroup differentiation and intra group respect. Journal of
Experimental Social Psychology, 41, 208-215.
Jetten, J., Spears, R., Postmes, T. (2004). Intergroup distinctiveness and
differentiation: A meta-analytic integration. Journal of Personality and
Social Psychology, 86, 862-879.
Kenworthy, J. B., & Miller, N. (2002). Attributional biases about the origins of
attitudes: Externality, emotionality and rationality. Journal of Personality
and Social Psychology, 82, 693-707.
Kernahan, C., Bartholow, B. D., & Bettencourt, B. A. (2000). Effects of category-
based expectancy violation on affect-related evaluations: Toward a
comprehensive model. Basic and Applied Social Psychology, 22, 85-100.
Kramer, R. M., & Brewer, M. B. (1984). Effects of group identity on resource use
in a simulated commons dilemma. Interpersonal Relations and Group
Processes, 46, 1044-1057.
55
Lemyre, L. & Smith, P. M. (1985). Intergroup discrimination and self-esteem in
the minimal group paradigm. Journal of Personality and Social
Psychology, 49, 660-670.
Lewin, K. (1948). Self-hatred among Jews. In K. Lewin (Ed). Resolving Social
Conflicts (pp. 186-200). New York, NYC: Harper & Brothers.
Lickel, B., Schmader, T., Curtis, M., Scarnier, M., & Ames, D. R. (2005).
Vicarious shame and guilt. Group Processes & Intergroup Relations, 8,
145-157.
Long, K., & Spears, R. (1997). The self-esteem hypothesis revisited. In R.
Spears, P.J. Oakes, N. Ellemers, & S. A. Haslam (Eds). (1997). The Social
Psychology of Stereotyping and Group Life. (pp. 296-317). Malden, MA:
Blackwell Publishing.
Luhtanen, R., & Crocker, J. (1992). A collective self-esteem scale: Self-evaluation
of one’s social identity. Personality and Social Psychology Bulletin, 18,
302-318.
Maass, A. (1999). Linguistic intergroup bias: Stereotype perpetuation through
language. In M. P. Zanna, (Ed). (1999). Advances in Experimental Social
Psychology, Vol. 31. (pp. 79-121). New York, New York: Academic Press,
Inc.
Marques, J. M., Abrams, D., & Serodio, R. G. (2001). Being better by being right:
Subjective Group Dynamics and derogations of in-group deviants when
generic norms are undermined. Journal of Personality and Social
Psychology, 81, 436-447.
Marques, J. M, Abrams, D., Paez, D., & Hogg, M. (2000). Social categorization,
social identification, and rejection of deviant group members. In M. A.
Hogg & R. S. Tindale (Eds.), Blackwell handbook of social psychology:
Group processes(pp. 400–424). Oxford: Blackwell.
Marques, J. M., Robalo, E. M., & Rocha, S. A. (1992). In-group bias and the
“black sheep” effect: Assessing the impact of social identification and
perceived variability on group judgments. European Journal of Social
Psychology, 22, 331-352.
56
Marques, J. M., Paez, D., & Abrams, D. (1998). In S. Worchel, J. F. Morales, D.
Paez, and J. C. Deschamps (Eds.), Social Identity: International
Perspectives (pp. 125-141). London, England: Sage Publications.
Marques, J. M., Yzerbyt, V. Y. (1988). The black sheep effect: Judgmental
extremity in inter- and intra-group situations. European Journal of Social
Psychology, 18, 287-292.
Marques, J. M., Yzerbyt, V. Y., & Leyens, J-P. (1988). The “black sheep” effect:
Extremity of judgments towards in-group members as a function of group
identification. European Journal of Social Psychology, 18, 1-16.
Matthews, D., & Dietz-Uhler, B. (1998). The black sheep effect: How positive
and negative advertisements affect voters’ perceptions of the sponsor of the
advertisement. Journal of Applied Psychology, 29, 1903-1915.
McGregor, I., & Marigold, D. C. (2003). Defensive zeal and the uncertain self:
What makes you so sure? Journal of Personality and Social Psychology, 85,
838-852.
MacKinnon, D. P., Lockwood, C. M., Hoffman, J. M., West, S. G., & Sheets, V.
(2002). A comparison of methods to test mediation and other intervening
variable effects. Psychological Methods, 7, 83-104.
Mikulincer, M., Gillath, O., Halevy, V., Avihou, N., Avidan, S., & Eshkoli, N.
(2001). Attachment theory and reaction to others’ needs: Evidence that
activation of the sense of attachment security promotes empathic responses.
Journal of Personality and Social Psychology, 81, 1205-1224.
Nummenmaa, L., Niemi, P. (2004). Inducing affective states with success-failure
manipulations: A meta-analysis. Emotion, 4, 207-214.
Oakes, P. J. & Turner, J. C. (1980). Social categorization and intergroup behavior:
Does minimal intergroup discrimination make social identity more positive.
European Journal of Social Psychology, 10, 295-301.
Postmes, T., Spears, R., Branscombe, N. R., & Young, H. (1999). Comparative
processes in personal and group judgments: Resolving the discrepancy.
Journal of Personality and Social Psychology 76, 320-338.
57
Roccas, S., Klar, Y., & Liviatan, I. (2004). Exonerating cognitions, group
identification, and personal values as predictors of collective guilt among
Jewish-Israelis. In N. R. Branscombe & B. Doosje (Eds.), Collective Guilt:
International Perspectives (pp. 130-147). Cambridge: Cambridge
University Press.
Sachdev, I., & Bourhis, R. (1987). Status differentials and intergroup behavior.
European Journal of Social Psychology, 17, 277-293.
Stratton, J., Lickel, B., & Miller, N. The black sheep effect: A meta analysis.
Manuscript under review.
Stratton, J., Vasquez, E., & Miller, N. Subtyping and subgrouping within the black
sheep effect. Manuscript in preparation.
Tajfel, H. (1979). Differentiation between social groups: Studies in the social
psychology of intergroup relations. London Academic Press.
Tajfel, H., Billig, M. G., Bundy, R. P., & Flament, C. (1971). Social
categorization and intergroup behavior. European Journal of Social
Psychology, 1, 149-178.
Tajfel, H., & Turner, J. C. (1979). An integrative theory of intergroup conflict. In
W. G. Austin & S. Worchel (Eds.), The Social Psychology of Intergroup
Relations (pp. 33-47). Monterey, CA: Brooks/Cole.
Tajfel, H., & Turner, J. (1986). An integrative theory of intergroup conflict. In S.
Worchel & W. G. Austin (Eds.), Psychology of Intergroup Relations (pp. 7-
24). Chicago, IL: Nelson-Hall Publishers.
Turner, J. C., & Oakes, P. J. (1989). Self-Categorization theory and social
influence. In P. B. Paulus (Ed.), Psychology of Group Influence (2nd ed.,
pp. 223- 275). Hillsdale NJ: Erlbaum.
Twenge, J. M., Catanese, K. R., & Baumeister, R. F. (2002). Social exclusion
causes self-defeating behavior. Journal of Personality and Social
Psychology, 83, 606-615.
Van Vugt, M. (2001). Community identification moderating the impact of
financial incentives in a natural social dilemma: Water conservation.
Personality and Social Psychology Bulletin, 27, 1440-1449.
58
Van Vugt, M., & Hart. C. M. (2004). Social identity as social glue: The origins of
group loyalty. Journal of Personality and Social Psychology, 86, 585-598.
Watson, D., Clark, L. A., & Tellegen, A. (1988). Development and validation of
brief measures of positive and negative affect: The PANAS scales. Journal
of Personality and Social Psychology, 54, 1063-1070.
Weber, R., & Crocker, J. (1983). Cognitive process in the revision of stereotypic
beliefs. Journal of Personality and Social Psychology, 45, 961-977.
59
Appendix A
Adapted version of Luhtanen and Crocker’s (1992) ‘identity’ subscale of their
Collective Self Esteem Scale plus 3 items from Lickel et al. (2005) research on
vicarious shame and guilt (items denoted below with an asterisk).
INSTRUCTIONS: Please read each statement and indicated your level of
agreement or disagreement using the following 7-point scale:
Strongly
Agree
Disagree Disagree
Somewhat
Neutral Agree
Somewhat
Agree Strongly
Agree
1 2 3 4 5 6 7
1. I am proud to be a college student.*
2. Overall, my status as a college student has very little to do with how I feel about
myself.
3. Being a college student is an important reflection of who I am.
4. Being a college student is unimportant to my sense of what kind of a person I
am.
5. In general, being a college student is an important part of my self image.
6. Being a college student is one of the good things about me.*
7. Although I’m a college student, it hardly affects how I feel about myself.*
60
Appendix B
Adapted version of Luhtanen and Crocker’s (1992) ‘public self esteem’ subscale of
their Collective Self Esteem Scale
INSTRUCTIONS: Please read each statement and indicated your level of
agreement or disagreement using the following 7-point scale:
Strongly
Agree
Disagree Disagree
Somewhat
Neutral Agree
Somewhat
Agree Strongly
Agree
1 2 3 4 5 6 7
1. I am afraid that this resident’s behavior could be viewed as indicating something
about the person I am.
2. I feel that people could make judgments about the type of person I am based on
this resident’s actions.
3. I feel that the resident reflects poorly on me.
4. In general, others think that college students are unworthy.
5. Overall, others view college students as if they are a joke.
6. I feel like this resident’s behavior confirms negative conceptions that other
people might have about the reputation of college students.
7. I am afraid that the behavior of the neighbor could be viewed as
indicating something about the person I am.
8. I feel that people could make judgments about the type of person I am based on
the policies of the neighbor.
9. I feel that the behavior of the neighbor reflects poorly on me.
10. I feel that the behavior of the neighbor confirms negative conceptions that
other people might have about the reputation of college students.
11. In general, others respect college students.
61
Appendix C
Adapted version of Watson, Clark, and Tellgen’s (1988) PANAS mood scale
INSTRUCTIONS: Please read each item below and mark the appropriate answer in
the space next to that work. Indicate to what extent you feel this way, right now,
that is, at the present moment. Use the following scale to record your answers:
1 2 3 4 5 6 7
very
slightly or
not at all
moderately extremely
_____ frustrated _____ irritable
_____ distressed _____ humiliated
_____ excited _____ ashamed
_____ upset _____ inspired
_____ strong _____ nervous
_____ guilty _____ annoyed
_____ scared _____ attentive
_____ hostile _____ surprised
_____ enthusiastic _____ active
_____ proud _____ afraid
62
Appendix D
Target Evaluation Motivation Scale
Participant Questionnaire 3
INSTRUCTIONS: Please read each statement and indicated your answer by circling
the appropriate number that describes your response to the candidate described in
the advertisement.
1. Would you prefer that the neighbor be expelled from/ not allowed into college?
1 2 3 4 5 6 7
Let the target
remain in the
group
Neutral Expel target
2. Would you prefer that the resident be expelled from/ not allowed into college?
1 2 3 4 5 6 7
Let the target
remain in the
group
Neutral Expel
target
3. How concerned were you that the neighbor might be mistakeningly seen as representative
of you?
1 2 3 4 5 6 7
Not at all
concerned
Somewhat
unconcerned
Neutral Somewhat
Concerned
Very
Concerned
How did you make your evaluations about the neighbor?
How did you make your evaluations about the resident?
63
4. How concerned were you that the resident might be mistakeningly seen as representative
of you?
1 2 3 4 5 6 7
Not at all
concerned
Somewhat
unconcerned
Neutral Somewhat
Concerned
Very
Concerned
5. How important is the neighbor behavior as described in this article in the grand scheme of
things?
1 2 3 4 5 6 7
Not at all
Important
Somewhat
unimportant
Neutral Somewhat
important
Very
Important
6. How important is the resident’s behavior as described in this article in the grand scheme of
things?
1 2 3 4 5 6 7
Not at all
Important
Somewhat
unimportant
Neutral Somewhat
important
Very
Important
7. I worry that people will draw conclusions about me, based on what they think about other
college students.
1 2 3 4 5 6 7
Strongly
Disagree
Disagree Disagree
Somewhat
Neutral Agree
Somewhat
Agree Strongly
Agree
8. College students should stick together regardless of what everyone else thinks.
1 2 3 4 5 6 7
Strongly
Disagree
Disagree Disagree
Somewhat
Neutral Agree
Somewhat
Agree Strongly
Agree
9. The behavior of one individual should not be allowed to represent all college students.
1 2 3 4 5 6 7
Strongly
Disagree
Disagree Disagree
Somewhat
Neutral Agree
Somewhat
Agree Strongly
Agree
10. I was afraid that the neighbor’s behavior would be viewed as indicating something about
the person I am.
1 2 3 4 5 6 7
Strongly
Disagree
Disagree Disagree
Somewhat
Neutral Agree
Somewhat
Agree Strongly
Agree
64
11. I was afraid that the resident’s behavior would be viewed as indicating something about
the person I am.
1 2 3 4 5 6 7
Strongly
Disagree
Disagree Disagree
Somewhat
Neutral Agree
Somewhat
Agree Strongly
Agree
INSTRUCTIONS: In the space provided, please describe the article you read.
The neighbor
__________________________________________________________________
__________________________________________________________________
__________________________________________________________________
__________________________________________________________________
______________________________________________________________
Some groups contain members who are very similar to one another: they may have
many things in common. Members of these groups are relatively uniform. Other
groups contain members who differ greatly from one another, and they don’t share
many characteristics.
Non-College students are:
1 2 3 4 5 6 7 8 9
Diverse / Uniform/
Differing Similar
65
Appendix E
Bogus Pipeline Identification Scale
INSTRUCTIONS: Please read each statement and indicated your level of
agreement or disagreement using the following 7-point scale:
Strongly
Agree
Disagree Disagree
Somewhat
Neutral Agree
Somewhat
Agree Strongly
Agree
1 2 3 4 5 6 7
1. Relationships with other people are very important to me.
2. Friendships are the most important treasure in life.
3. Family relationships are more important than work goals
4. Getting ahead in life takes priority over my personal relationships.
5. I often chose to go out and party with friends than study even when I have a test
or work project due the next day.
6. I have many people that I would call my best friends.
7. I have more acquaintances than friends.
8. You should never mix business and pleasure.
9. Sometimes, I feel isolated from everyone else.
10. Sometimes, I feel like it is hard to meet new people.
11. When I feel lonely, I will often immerse myself in work.
12. I belong to many clubs and social organizations.
13. I prefer going to a movie over going dancing.
14. I have no problem walking up to a stranger and becoming friends with them.
15. I have a large circle of friends.
66
Appendix F
Luhtanen and Crocker’s (1992) Collective Self Esteem Scale
Thinking about Liberals/ Conservatives, please answer the following
questions using the scale below.
Strongly
Agree
Disagree Disagree
Somewhat
Neutral Agree
Somewhat
Agree Strongly
Agree
1 2 3 4 5 6 7
1. I am a worthy member of this group.
2. I often regret that I belong to this group.
3. Overall, my group is considered good by others.
4. Overall, my membership in this group has very little to do with how I feel about
myself.
5. I feel I don’t have much to offer the group.
6. In general, I’m glad to be a member of this group.
7. Most people consider my group, on the average, to be more effective than other
social groups.
8. The group is an important reflection of who I am.
9. I am a cooperative participant in this group.
10. Overall, I often feel that the group of which I am a member is not worthwhile.
11. In general, others respect this group that I am a member of.
12. The group is unimportant to my sense of what kind of person I am.
13. I often feel I’m a useless member of this group.
14. I feel good about belonging to this group.
15. In general, others think that the group is unworthy.
16. In general, belonging to this group is an important part of my self image.
67
Appendix G
Articles Describing Either an In-group or Out-group Target
Liberal/ in-group conditions:
Donald Johnson describes himself as a liberal politically. In a
public interview for a local newspaper, he made these comments about
politics. “Everybody knows I am liberal because I talk a lot about my
political leanings and I always preface every conversation by saying I am
liberal. Aside from my political values, however, I think I have little in
common with other liberals. Liberals tend to be from privileged, middle-
class backgrounds and we are often too comfortable with our own cushy
lifestyles to better themselves as human beings. As a group, we are less
likely to vote and less likely to fight for our beliefs. It would be no surprise
to me if liberal issues started to lose ground to conservative issues because
of our behavior.
Liberal/ out-group conditions:
Donald Johnson describes himself as a conservative politically. In a
public interview for a local newspaper, he made these comments about
politics. “Everybody knows I am conservative because I talk a lot about my
political leanings and I always preface every conversation by saying I am a
conservative. Besides my political values, I think I have little in common
with liberals. Liberals tend to be from privileged, middle-class
backgrounds and are often too comfortable with their own cushy lifestyles
to better themselves as human beings. As a group, they are less likely to
vote and less likely to fight for their beliefs. It would be no surprise to me
if liberal issues started to lose ground to conservative issues because of their
behavior.
Conservative/ in-group conditions:
Donald Johnson describes himself as a conservative politically. In a
public interview for a local newspaper, he made these comments about
politics. “Everybody knows I am conservative because I talk a lot about my
political leanings and I always preface every conversation by saying I am a
conservative. Aside from my political values, however, I think I have little
in common with other conservatives. Conservatives tend to be from
privileged, middle-class backgrounds and we are often too comfortable with
our own cushy lifestyles to better themselves as human beings. As a group,
we are less likely to vote and less likely to fight for our beliefs. It would be
no surprise to me if conservative issues started to lose ground to liberal
issues because of our behavior.
68
Conservative/ out-group conditions:
Donald Johnson describes himself as a liberal politically. In a
public interview for a local newspaper, he made these comments about
politics. “Everybody knows I am liberal because I talk a lot about my
political leanings and I always preface every conversation by saying I am a
liberal. Besides my political values, I think I have little in common with
conservatives. Conservatives tend to be from privileged, middle-class
backgrounds and are often too comfortable with their own cushy lifestyles
to better themselves as human beings. As a group, they are less likely to
vote and less likely to fight for their beliefs. It would be no surprise to me
if conservative issues started to lose ground to liberal issues because of
their behavior.
69
Appendix H
Target Trait Evaluation Form
Using the following scale,
Not at all 1 2 3 4 5 6 7 8 9 very much
To what degree do you think the neighbor:
____ Emotionally sensitive ____ Honest
____ Cold ____ Unapproachable
____ Responsible ____ Likeable
____ Trustworthy ____ Warm
____ Malicious ____ Rude
____ Intelligent ____ Confident
____ Kind ____ Assertive
____ Annoying ____ Self Righteous
____ Friendly ____ Arrogant
____ Uninteresting
In your opinion, the neighbor described in the article is:
a good person 1 2 3 4 5 6 7 8 9 a bad person
Memory Quiz
Which of the following describes the neighbor (Check ALL that apply):
Skier Loner Cat lover
Adult
Robber
Menlo house resident
Laptop thief
Child
Race car driver
Filers of complaint
Roommate
Neighbor of college
students
Dog hater
Ski patrol member
Police officer
Astronomer
Noise intolerant
college student
bicyclist
Non-college
student
Against noise
ordnance
Abstract (if available)
Abstract
Social identity theory (Tajfel, 1979
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Black sheep or devil protection?: the effects of cooperative and competitive intergroup goal structures on group members reactions to an offensive act when committed by an ingroup or an outgroup ...
PDF
The role of framing in health social comparisons by multiple sclerosis patients
PDF
The social groups approach to quitting smoking: An examination of smoking cessation online and offline through the influence of social norms, social identification, social capital and social support
PDF
The effect of massed versus spaced arguments on persuasion
PDF
Neural correlates of learning and reversal of approach versus withdraw responses to in- and out-group individuals
PDF
The association between peer victimization and school functioning in Latino children
PDF
Social status, perceived social reputations, and perceived dyadic relationships in early adolescence
PDF
Neurobiological correlates of skilled and disabled reading
PDF
Multiple informant-rated psychopathy in young adolescents: a multitrait-multimethod investigation of the Antisocial Process Screening Device
PDF
Social support, constraints, and protective buffering in prostate cancer patients and their partners
PDF
Actual and perceived social reinforcements of weight-related cognitions and behaviors in adolescent peer groups
PDF
Significance testings in regression analyses
PDF
Cueing compliance and conferring context: testing the group-as-cue and group-as-conveyor models of social influence.
PDF
The effect of psychological skills and affective intervention programs on exercise adherence
PDF
Attachment, maltreatment and autonomic nervous system responsivity as predictors of adolescent anxiety and depression
PDF
Community violence exposure and children's subsequent rejection within the school peer group: the mediating roles of emotion dysregulation and depressive symptoms
PDF
The social meaning of sharing and geocoding: features and social processes in online communities
PDF
Case studies of national cultural protection in the era of globalization
PDF
Effects of racial exclusion from media
PDF
Deconstructing the psychological components of emotional decision making and their relation to the suicide continuum
Asset Metadata
Creator
Stratton, Joy
(author)
Core Title
Social identification and the treatment of in-group deviants: the black sheep effect or the devil protection effect
School
College of Letters, Arts and Sciences
Degree
Doctor of Philosophy
Degree Program
Psychology
Publication Date
07/21/2007
Defense Date
05/23/2007
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
black sheep effect,deviant,devil worship effect,OAI-PMH Harvest,social identification
Language
English
Advisor
Miller, Norman (
committee chair
), Hollingshead, Andrea (
committee member
), Lickel, Brian (
committee member
), Manis, Franklin R. (
committee member
)
Creator Email
athena.3@att.net
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-m638
Unique identifier
UC1468870
Identifier
etd-Stratton-20070721 (filename),usctheses-m40 (legacy collection record id),usctheses-c127-517169 (legacy record id),usctheses-m638 (legacy record id)
Legacy Identifier
etd-Stratton-20070721.pdf
Dmrecord
517169
Document Type
Dissertation
Rights
Stratton, Joy
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Repository Name
Libraries, University of Southern California
Repository Location
Los Angeles, California
Repository Email
cisadmin@lib.usc.edu
Tags
black sheep effect
deviant
devil worship effect
social identification