University of Southern California Dissertations and Theses

Effect of affective reactions by an outgroup on preferences for crossed categorization discussion partners

(USC Thesis Other)

Effect of affective reactions by an outgroup on preferences for crossed categorization discussion partners

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content INFORMATION TO USERS
This manuscript has been reproduced from the microfilm master. UMI
films the text directly from the original or copy submitted. Thus, some
thesis and dissertation copies are in typewriter free, while others may be
from any type of computer printer.
The quality of this reproduction is dependent upon the quality of the
copy submitted. Broken or indistinct print, colored or poor quality
illustrations and photographs, print bleedthrough, substandard margins,
and improper alignment can adversely afreet reproduction.
In the unlikely event that the author did not send UMI a complete
manuscript and there are missing pages, these will be noted. Also, if
unauthorized copyright material had to be removed, a note will indicate
the deletion.
Oversize materials (e.g., maps, drawings, charts) are reproduced by
sectioning the original, beginning at the upper left-hand comer and
continuing from left to right in equal sections with small overlaps. Each
original is also photographed in one exposure and is included in reduced
form at the back of the book.
Photographs included in the original manuscript have been reproduced
xerographically in this copy. Higher quality 6” x 9” black and white
photographic prints are available for any photographs or illustrations
appearing in this copy for an additional charge. Contact UMI directly to
order.
UMI
A Bell & Howell Information Company
300 North Zed) Road, Ann Arbor MI 48106-1346 USA
313/761-4700 800/521-0600
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
EFFECT OF AFFECTIVE REACTIONS BY AN OUTGROUP O N PREFERENCES
FOR CROSSED CATEGORIZATION DISCUSSION PARTNERS
Copyright 1998
by
Nurcan Ensari
A Thesis Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment o f the
Requirements for the Degree
MASTER of ARTS
(Psychology)
May 1998
Nurcan Ensari
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
UMI Number: 1391081
UMI Microform 1391081
Copyright 1998, by UMI Company. All rights reserved.
This microform edition is protected against unauthorized
copying under Title 17, United States Code.
UMI
300 North Zeeb Road
Ann Arbor, MI 48103
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
U N IV ERSITY O F S O U T H E R N C A U F O R N IA
T H E G R A D U A T E SCHOOL.
U N IV E R S IT Y PA R K
t-O S A N G E L E S . C A L IF O R N IA 8 0 0 0 7
This thesis, written by
H scu-1 ..
under the direction of hS^T.— Thesis Committee,
and approved by a ll its members, has been pre
sented to and accepted by the Dean of The
Graduate School, in partial fulfillm ent of the
requirements for the degree of
Master of Arts
Dtam
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Acknowledgments
I would like to express my sincere gratitude to my committee chair Dr. Norman
Miller for his everlasting encouragement and support at every stage o f the growth o f this
study. I would also like to express my sincere gratitude to Dr. Steve Read and Dr. Shelley
Duvall for the time they have devoted to reading and commenting on my thesis. I really
appreciate their support and guidance.
I am especially grateful to Dr. Ercan Alp for his support in conducting Study 3,
and Jared Kenworthy for his assistance in the data collection for Study 1. A debt o f
gratitude is owed to my friends and colleagues, Lynn Urban, Darren Urada and Eric
Vanman for their helpful feedback provided at various stages o f this thesis. I also
gratefully acknowledge the financial support o f the National Science Foundation Grant
SBR 931 9752 (Norman Miller, Primary Investigator). Last, but not least, I thank my
loving husband, Semsi for his usual unfailing encouragement and support, and my family,
to whom this thesis is dedicated. To them, I owe special thanks.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
iii
Table of Contents
A cknow ledgm ents....................................................................................................................................ii
List o f T ables.............................................................................................................................................iv
Abstract.........................................................................................................................................................v
Introduction..................................................................................................................................................1
Experiment 1............................................................................................................................................. 10
Method............................................................................................................................................. 10
R esults............................................................................................................................................. 18
D iscussion ......................................................................................................................................22
Experiment 2 .............................................................................................................................................23
Method.............................................................................................................................................26
R esults.............................................................................................................................................30
D iscussion......................................................................................................................................33
Experiment 3 .............................................................................................................................................35
Method.............................................................................................................................................35
Results and D iscussion...............................................................................................................38
General D iscussion.................................................................................................................................46
References..................................................................................................................................................54
Footnotes....................................................................................................................................................59
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
List of Tables
Table 1: Mean Preferences for Four Types o f Target as a Function o f Compliment from
Outgroup Members.................................................................................................................................65
Table 2: Mean Preferences for Four Types o f Target as a Function o f Insult from
Outgroup Members and R elevancy...................................................................................................66
Table 3: Mean Preferences for Four Types o f Target as a Function o f Insult and
Compliment from Outgroup M embers............................................................................................. 67
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
V
Abstract
Assessed the effect o f positive and negative affective arousal from an outgroup
source on preferences for the four targets o f crossed categorization paradigm (ingroup-
ingroup (ii), ingroup-outgroup (io), outgroup-ingroup (Oi), and outgroup-outgroup (oo))
as discussion partners. Study 1 used complimentary interviews from members o f an
outgroup to induce positive affective arousal. Study 2 exposed participants to insulting
interviews from members o f an outgroup. Preferences for the crossed category discussion
partner who matched the outgroup source o f arousal on one dimension o f categorization
(Oi) were increased by compliments, and decreased by insults. Both experiments showed
an Additivity pattern in the control condition. Experiment 3 cross culturally replicated
these effects within a single design.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect o f Affective 1
Effect o f Affective Reactions by an Outgroup on Preferences for
Crossed Categorization Discussion Partners
Two salient and mutually exclusive social categories, such as gender and age,
when combined, create a situation in which people either share both categories — a double
ingroup (ii), share one o f the two categories — a crossed categorization (io and oi), or
share none — a double outgroup (oo) with one another. For example, female/young people
and female/old people belong to the same group with respect to gender but to different
groups with respect to age. Deschamps and D oise (1978) called this situation a “crossed
categorization”. When cues for crossed categorization have been made salient, it has been
argued that the perception of group differentiation on one dimension w ill be canceled out
by the simultaneous perception o f group similarity on the other dimension (Vanbeselaere,
1996), and as a result, a conflict between the accentuation both o f differences and
similarities inside each o f the categories and between the opposing categories decreases
the extent of intergroup differentiation (Deschamps & D oise, 1978). Thus, making cues
o f crossed categorization salient has been view ed as a strategy that reduces intergroup
discrimination (Brewer, Ho, Lee & Miller, 1987; Brown & Turner, 1979; Deschamps &
Doise, 1978; Hewstone, Islam, & Judd, 1993; Vanbeselaere, 1987, 1991). However, the
conditions necessary for crossed categorization to reduce intergroup bias have not been
fully explored. One plausible factor which researchers have identified to play a role in
determining the crossed categorization situation and have important consequences for
targets in the crossed categorization paradigm is the positivity or negativity o f one's
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect o f Affective 2
affective state (Miller, Urban & Vanman, in press; Urada & Miller, 1997; Urban &
Miller, 1997). This paper attempts to explore the effects o f affective valence on crossed
categorization in order to provide a better understanding o f intergroup cognition that
occur in the crossed categorization situation, and thereby enable more appropriate
application o f the crossed categorization principle for the purpose o f reducing intergroup
conflict.
Six different patterns o f results have been obtained in past research on crossed
categorization (Miller et al., in press; Urban & Miller, 1997). In the Equivalence pattern
(Deschamps & Doise, 1978) all four target groups are evaluated equally, (ii=io=oi=oo). In
the Additivity pattern (e.g., Hagendoom & Henke, 1991) the double ingroup is evaluated
most positively, the crossed groups are evaluated as equally intermediate, and the double
outgroup is evaluated most negatively, (ii>io=oi>oo). In the Conjunction Similarity
pattern (e.g., Brown & Turner, 1979) the double ingroup and the crossed targets are
evaluated equally positively, but the double outgroup is evaluated negatively,
(ii=io=oi>oo). In the Conjunction Dissimilarity pattern (e.g., Eurich-Fulcher & Schofield,
1995) the double ingroup is evaluated positively, but the crossed and double outgroup
targets are evaluated equally negatively, (ii>io=oi=oo). In the Category Dominance
pattern (e.g., Stangor, Lynch, Duan, & Glass, 1992) one o f the two category dimensions
is dominant and evaluations o f the crossed targets depend on which dimension is
dominant, (Ii=Io>Oi=Oo) or (iI=oI>iO =oO ).‘ In the Hierarchical Acceptance pattern 2
(e.g., Triandis & Triandis, 1960), ingroup and outgroup members are differentially
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect o f Affective 3
evaluated on the dominant dimension, whereas on the nondominant dimension
ingroup/outgroup status is ignored unless the target is an ingroup member on the
dominant dimension, (Ii>Io>Oi=Oo) or (iI>oI>iO=oO). Thus, when a crossed target has
ingroup status on the dominant dimension, that Io (or oi) target is not evaluated as
favorably as the double ingroup, whereas an Oi (or iO) target is not differentiated from
the double outgroup.
Clearly, the variation in the patterns o f results obtained in prior research calls for
explanation. In an attempt to account for this inconsistency and identify the circumstances
under which particular patterns o f outcomes can be expected to occur, Urban and Miller
(1997) examined the explanatory role of a number of theoretically derived moderator
variables within the context o f a meta analysis. Their analyses showed that affective
valence o f the experimental setting, opportunity for personalized contact, cognitive
overload, and the discrepant importance o f the two social category dimensions are major
factors relevant for predicting patterns o f evaluation across the four crossed
categorization targets. Moreover, with the exception of discrepant importance, each
provided independent or unique explanation. In the studies comprising the meta analysis,
however, none specifically manipulated these variables. Instead, the assessment of their
explanatory role was based on between-study variation, as assessed by judges who rated
aspects o f the experimental procedures. For instance, variation in affective valence rested
on the overall positivity/negativity o f the experimental setting, as affected by the nature
o f the task, and the characteristics o f the reward or goal structure. Therefore, although
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect o f Affective 4
affect strongly moderated the mean effect sizes reflecting the magnitude o f evaluative
difference in comparisons between the targets o f the crossed categorization paradigm, this
evidence for its explanatory role is only correlational. And because the magnitude o f
variation in affect across studies was in an absolute sense extremely small, distinct
theoretical predictions (Miller et al., in press) for negative and positive affect could not be
evaluated. Consequently, the purpose o f this article is to confirm this correlational
evidence experimentally, and examine the distinct effects of positive and negative affect
on the pattern o f evaluations o f the four target persons o f the crossed-categorization
paradigm.
Within the crossed categorization paradigm, the source o f affective arousal can
vary. It can arise from aspects o f the setting, such as pleasant versus unpleasant music or
oders; others in the setting, such as the experimenter or a confederate; or a person clearly
identifiable as an ingroup or outgroup member (Miller et al., in press). Bodenhausen
(1993) has distinguished integral affect from incidental affect. Incidental affect is affect
that stems from a source that is unrelated to the intergroup context, whereas integral
affect arises directly from the intergroup features o f a setting. In the present set o f studies,
instead o f manipulating general affective arousal, w e focused on the effect o f integral
affect, that is, affect induced by a particular specifiable outgroup category source.
The presence o f a positive affective state reduces systematic and careful
processing o f available information (Bodenhausen, Kramer, & Susser, 1994; Worth &
M ackie, 1987), and increases the extent to which people form superordinate group
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect of Affective 5
representations (Dovidio, Gaertner, Isen, & Lowrance, 1995), which in turn enhances out
group evaluations and reduces intergroup bias. Moreover, positive affect can be induced
by positive feedback about one’s ingroup (Queller, M ackie, & Neddermeyer, 1994).
When social science students read an interview with an engineering student about the
contribution o f social science students to his own work, they reported feeling happier
when the engineering student described it as positive than when he described it as
negative. Consequently, we expect in general that when an experimental induction o f a
strongly positive emotional state (happiness) is introduced into the crossed categorization
paradigm it w ill shift evaluations closer to an Equivalence pattern. In addition, the
priming function o f an antecedent positive emotional state w ill direct attention toward the
positively valenced stimuli which is associated with ingroup-defining features o f targets.
Because each o f the crossed targets possess an ingroup cue, when these processes are
combined with the priming and directing functions of mood (Schwartz & Clore, 1996),
two crossed categories will be assimilated to the double ingroup in comparison with the
double outgroup. This outcome corresponds to the Conjunction Similarity Pattern (Miller
et al., in press). However, when positive affective arousal takes the form o f a compliment
from an outgroup source, it will increase the tendency to accept new members o f that
particular outgroup category, and reduce intergroup hostility. Accordingly, within the
context o f the crossed categorization paradigm, one o f the crossed category will become
more salient and dominant as a result of compliments from the members o f that category,
therefore an evaluation o f a crossed category target who is a member o f this particular
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect of Affective 6
outgroup category (Oi) should be assimilated toward that of the double ingroup (ii) (i.e.,
ii = Oi). At the same time targets who represent the other crossed category (io) or a
double outgroup (oo) respectively, should be evaluated more negatively than the crossed
target whose counterparts were the source of compliments (i.e., Oi > io > 0 0 ). These
effects produce a pattern that combines features of the Conjunctive Similarity and
Dominance Patterns (viz.:ii=Oi>io>oo). Because people generally react more positively
to others as a direct function o f the number of shared group memberships (Tajfel &
Turner, 1979) we expected our control condition in which there is no induction o f affect,
to produce a pattern o f evaluations that corresponds to an Additivity Pattern (viz.:
ii>oi=io>oo). We tested these predictions in Experiment 1.
By contrast, negative affective arousal has a different effect. Previous studies
suggest that a negative affective state (such as anger) induces people to react negatively
toward others, but particularly if they are outgroup members (Miller et al., in press). In
addition, it threatens one’s social identity, and invokes identity protection mechanisms
such as rejecting that outgroup (Miller et al., in press). Therefore, when introduced into
the crossed categorization paradigm, we expect that the priming and directing effect o f
negative affect will make the outgroup characteristic o f the crossed targets salient and in
turn increase the negativity o f evaluations o f both crossed category targets as well as the
double outgroup target. Accordingly, the crossed targets are evaluated similarly to the
double outgroup, and this w ill produce a Conjunction Dissimilarity pattern of evaluations
across the four targets (Miller et al., in press). However, when the negative affect takes
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect o f Affective 7
the form o f an insult from an outgroup source, it makes that specific outgroup category
dimension more dominant over other dimensions. As shown by the literature on displaced
aggression (meta-analysis: Miller & Marcus-Newhall, in press) when people experience
an aversive event in circumstances that precludes subsequent retaliation against its
source, they aggress more toward another target than do those who have not previously
been provoked. Likewise, insult from an outgroup source increases the category salience
and hostility towards others who are members o f that particular outgroup category
(Lewicki, 1986, p.200). Thus, when negative affective arousal is linked to a particular
outgroup, subsequent exposure to other members of that social category w ill trigger
negative reactions. Hence, a member o f the double outgroup (oo) and a member of that
crossed category who is not one o f the specific outgroup members who provide insults,
but just a member o f this particular outgroup category (Oi) will be evaluated more
negatively by comparison with members o f the io and ii category combinations.
Therefore, insult from an outgroup source should produce a pattern that corresponds to a
combination o f the Conjunctive Dissimilarity and Dominance patterns (viz.:
ii>io>O i=oo) in which the double ingroup target is evaluated most positively, and crossed
targets that share an outgroup cue with source o f the insult are rated as negatively as
double outgroup targets. Again, in the absence o f insult we expected participants’
evaluations to show an Additive pattern (viz.: ii>io=oi>oo). We tested these predictions
in Experiment 2.
Anticipated elevation and reduction in preference for the crossed target who is the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect of Affective 8
source o f the affect manipulation (Oi) were also expected to cause differential preferences
for the other crossed target (io) in the compliment and insult conditions relative to the
control condition. According to the contrast principle, if two things are presented one
after another, and the second item is fairly different from the first, then people tend to see
it as more different that actually it is (Cialdini, 1984). Similar judgmental contrast effect
was found by Berkowitz and Knurek (1969) in an aggression experiment. W hen the
participant was frustrated initially by an experimenter, and he was required to judge
som eone who was clearly different from his frustrater, than comparison o f the second
person with the first led him to exaggerate the difference between the two, and rate the
second person more favorable than he otherwise would have done (Berkowitz and
Knurek, 1969). So, on the basis o f prior findings regarding the judgmental contrast effect
found for individuals, we expect the same effect in intergroup level for individual
members o f category as well. When people are exposed to a set o f complimentary (or
insulting) interviews with members of an outgroup, and if they are asked to evaluate a
member o f that particular outgroup (Oi) and a member o f another outgroup (io), they w ill
be more unfavorable (or favorable) toward the second target than when they are not
exposed to any positive or negative affective stimuli, i.e., in the neutral condition.
Specifically, as a result o f contrast effect produced by comparison with the Oi target, the
preference for the io target in the compliment (or insult) condition was expected to be
lower (or higher) than that expressed in the neutral condition.
For the purposes o f this set of studies, it is important to clarify two aspects o f our
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect o f A ffective 9
design. First, Oi (the source of positive or negative affect) is not one of the specific
outgroup members who provide compliments (or insults), but just happened to be a
member o f that particular outgroup. In other words, instead o f one o f the outgroup
members who compliment (or insult) participant's ingroup, sim ply a member o f that
outgroup w ill be presented as a possible discussion partner. This procedure w ill allow us
to determine whether the predicted increased (or decreased) preference for Oi as a
discussion partner indicate generalization of positive (or negative) affect connected with a
specific outgroup to the categorization dimension that define that outgroup.
Second, instead o f crossing two dimensions to create four targets o f crossed
categorization paradigm, in this set o f studies, idiographic selection precedure was
applied. That is, as explained below, four important social categories formed the basis for
selection o f two dimensions for each target. For example, if the four idiographically
selected dimensions were age, major, university affiliation and citizenship, then for each
partner description two o f those four dimensions were crossed randomly. As a result, for
instance, when UCLA students (outgroup) compliment or insult USC students (ingroup),
the targets presented as possible discussion partners might be young adult/LAS major (ii),
US citizen/older adult (io), UCLA student/US citizen (Oi-source o f compliments or
insults), and non-US citizen/business major (oo). This idiographic selection precedure
enabled us to reduce between-condition variance in importance, and between target
variation in importance within each participant, and therefore increase the sensitivity of
our experimental designs.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect of Affective 10
In Experiment 3 we manipulated integral positive and negative affect within the
context o f a single study and attempted to replicate the predicted effects cross-culturally.
Experiment 1
In Experiment 1, positive affect was created by exposing participants to
compliments. The source of affect was one of the outgroup categories in the crossed
category situation. That is, a compliment came from outgroup members in the outgroup-
ingroup crossed category (Oi). Eight different category dimensions (age, year in school,
university attended, political affiliation, graduate vs. undergraduate student, honors vs.
non-honors student, citizenship and major in school) previously assessed by pilot test as
being o f moderate psychological importance for the participant population, formed the
basis for idiographic selection o f two equally important category dimensions for each
participant. The dependent measure was preference for the four types of targets (ii; io; Oi
or oi, depending on the affect condition; & oo) as a discussion partner in a subsequent
stage o f the experiment. As previously discussed, we predicted an Additive Pattern (viz.:
ii>io=oi>oo) for the control condition whereas we expected a combination o f the
Conjunctive Similarity and Dominance patterns (viz.: ii=Oi>io>oo) in the positive affect
condition.
Method
Participants
Forty nine undergraduate students enrolled in the Introductory Psychology course
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect o f A ffective 11
at the University o f Southern California participated in order to receive partial credit
toward their course grade. Data from two students who were suspicious about the cover
story were discarded.
Design
The design was a 2 (compliment from one outgroup category/ no compliment) X
4 (type o f target: ii, io, Oi, oo) factorial with the first factor manipulated between subjects
and the second manipulated within subjects. Participants were randomly assigned to
compliment versus no-compliment conditions prior to the experimental session. The
order o f the four target descriptions was counterbalanced using a Latin-square design to
assure that each target description appeared an equal number of times at each ordinal
position. Each target description consisted o f two pieces o f information created by a
random combination selected from among the four distinct social categories identified by
the participant as most important.3
Materials
Three questionnaires were used to implement the study. A Participation
Information Form was used to assess the participants’ ingroups and outgroups and
thereby select for each individual participant four appropriate target descriptions. It also
provided a basis for selecting the category for the complimentary outgroup. The form
asked participants to categorize themselves along the dimensions of age (young adult/old
adult), political beliefs (liberal/conservative), nationality (US citizen/non-US citizen),
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect of Affective 12
year in school (freshmen/sophomore/junior/senior), major (business/LAS), honors/not
honors student, university affiliation (USC/UCLA), graduate/ undergraduate student.
After the participants categorized themselves on each dimension, they were asked to
indicate how important each of the groups were in terms o f “your sense, of who you are”
and which categories were “most important to your self identity”. For each participant,
the category experimentally chosen to be the source o f compliments was selected
randomly from among those four categories that each participant had individually
identified (on the basis of these two questions) as the more important ones among the
larger set o f eight. This procedure was designed to reduce between-subject variation in
the importance levels o f the complimentary outgroups that were selected for each
participant and to reduce extraneous within-subject variation across targets.
The eight social category dimensions listed above had been judged as
approximately equal in importance among 20 dimensions in an independent survey o f
students from the Introduction Psychology course at the University o f Southern
California (Urban, 1995). The mean ratings, on a 7-point scale, and their standard
deviations respectively were; age, M = 4.20, SD =1.65, political belief, M=3.43, SD =1.51,
nationality, M=4-22, SD=1.65. year in school, M =3.37, SD =1.61. major, M =3.82,
SD =1.74, honor-not honors student, M =3.39, SD =1.61. university affiliation, M =3.36,
SD =1.74, and graduate/undergraduate, M =3.84, SD =1.68. The mean rating of importance
o f the 20 group dimensions ranged from 2.42 to 5.01.
To manipulate positive or neutral affect, participants read a single news article
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect of Affective 13
containing either complimentary remarks from the outgroup about their ingroup, or in the
control condition, remarks about Disneyland. The latter were o f low to moderate affective
intensity and balanced for pro/con affect. For each participant in the experimental
condition, one o f eight alleged newspaper articles (interviews with an outgroup category
o f the participant, namely, old adults, liberals or conservatives, non-US citizens, seniors,
business majors, honors students, UCLA students, or graduate students) was used to
induce positive affect. These complimentary articles consisted o f approximately 25
interviews with specific outgroup members o f a category randomly selected from among
the four dimensions that had been idiographically identified by each participant as more
important from among the larger set of eight on the Participation Information Form. Each
complimentary article allegedly had been photocopied from the university newspaper
(Daily Trojan). In actuality, the interviews were created by the experimenter and copied
onto a page containing the format and logo o f the university newspaper. In each article
the interviewed outgroup members o f the selected category consistently stated positive,
complimentary remarks about the participant’s ingroup. The content o f compliments
focused most often on performance and appearance. For instance, some o f the ideas
contained in the alleged interviews with old adults were: young people today adopt a
more adult attitude that warrants respect from old people; young adults have a purpose in
life; young adults make an effort to create a significant impact on society’s view about
important social issues or provide alternative views.
A form titled “The Description o f Other Participants”, which contained spaces for
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect o f Affective 14
up to four pieces o f information per target in order to facilitate acceptance o f the cover
story, was used to create descriptions o f the four targets o f the crossed categorization
paradigm. For each o f the four targets, it contained two hand-written pieces o f descriptive
information concerning the category identity of each target on each of two dimensions.
This information was allegedly provided by the other participants in the experimental
session and copied onto the form by the experimental assistant. The key dependent
measure was a seven-point Likert type scale that asked the participant to indicate how
strongly he or she wished to be a partner with each described target in the subsequent
discussion task. To facilitate acceptance o f this aspect o f the cover story they were also
asked to rank the four targets (1-4) in order o f their preference for each.
Procedure
General cover story. Upon arrival, the participant was told that four others were
participating at the same time. To convince the participant that others were indeed in
other experimental rooms, 5 student ID numbers were listed on the door with instructions
for each to report to the experimental room number next to his/her student ID number. (In
actuality, only one o f the numbers was real; four were for bogus participants). Real
participants were always assigned to the same room.
Then, the experimenter said that she needed to check to see if the others had
arrived. Leaving the door open and remaining within earshot o f the participant, she
entered an adjoining room to engage in a short conversation on the phone, supposedly
with another experimenter. She then returned to the experimental room and informed the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect o f Affective 15
participant that they needed to wait for one additional participant. During that time, she
had the participant sign in and briefly engaged him/her in a conversation. After about a
minute, she went back into the adjoining room and engaged in another brief conversation
on the phone. Following this alleged telephone conversation, the participant was
informed that the other four participants in other rooms were ready to begin. The
participant was told that two o f the other participants were recruited through the normal
subject pool, and two were recruited from newspaper advertisements; that the study was
concerned with memory ability and with the effects o f cooperation on a subsequent group
discussion; that later in the session they would be given a chance to choose a partner for
their discussion task; and that he/she would only have a limited amount o f information
about others because part o f this study was concerned with how the amount o f
information one receives about his/her partner before meeting him/her w ill affect the
future effectiveness o f the discussion. This latter requirement o f limited information was
given as the reason that the participants initially were separated into different rooms.
Next, the participant was asked to draw a number which ranged allegedly from 0
to 4 from a box filled with numbered slips o f paper. The number drawn purportedly
would determine the number o f pieces o f information that the participant would be given
about each one o f the other four participants. Actually, all o f the slips o f paper were
marked with the number 2, which corresponded to the two indicators o f social category
membership that defined the four combinations o f crossed and convergent cues for target
ingroup/outgroup identity. He/she was told that each participant would receive the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect o f Affective 16
amount o f information for each target determined by their own random drawing. Next,
the Participant Information Form asked each participant to categorize himself/herself
along the eight dimensions. The participant was told that the bogus participants were also
being given the same form at the same time by other experimenters, that the information
on these forms was the only information that was given to the participants, and that the
information about the other participants would be exchanged later.
Compliment/no-compliment conditions. After the participant completed the
Participant Information Form, the experimenter explained the memory aspect o f the
experiment. The participant was given the specially prepared page allegedly photocopied
from the university newspaper and told that his/her task was to read the interview article
carefully and remember its content as thoroughly as possible, because later he/she would
be asked to recall its content. To ensure that the experimenter was blind to the conditions,
an assistant who was present in the next room during the experiment gave the article to
the experimenter in a closed file.
In the compliment condition, this page consisted o f complimentary interviews
with one o f the outgroup categories o f the participant, as described earlier. As previously
indicated, the outgroup category that was the source o f the compliment was selected
randomly from among the four important categories that the participant had listed in the
Participant Information Form. In the no-compliment condition, the participant was given
the neutral article.
Before the participant started to read the article, the experimenter explained that
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect o f Affective 17
she needed to get the amount o f information determined by each participant’s drawing,
copy the relevant information, and distribute it. She then left the room. At this point, four
target descriptions o f the bogus participants (ii,io,oi,oo) were prepared according to the
information produced by the participant on his/her Participant Information Form. The
experimental assistant randomly selected two social category memberships for each target
description from among the four most important categories and wrote these four target
descriptions on the Descriptions o f Other Participants Form by hand. She then gave it to
the experimenter in a closed file to ensure that the experimenter remained blind to the
conditions. The source o f compliments was the outgroup in the Oi description. Thus, the
distinction between the descriptions o f io and oi targets was that the outgroup in the oi
combination (designated as Oi) was the source o f compliments, whereas the outgroup in
the io crossed category was one o f the other three important social categories among the
four.
Manipulation check and other dependent measures. After the participant read the
interviews, he/she was told that memory questions about the article would be asked after
he/she had indicated his/her relative preference among the four other participants as
partner for the upcoming discussion task. To assess the efficacy o f the manipulation of
positive affect, participants completed a questionnaire that contained o f 14 adjectives (7
positive and 7 negative adjectives) with a 4-point scale for expressing feelings and
opinions about the article. Lower scores indicated that the adjective was a better
description of the participant’s feelings, whereas higher scores indicated that that
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect o f Affective 18
adjective did not apply to their feelings. Then, he/she was given the four target
descriptions (ii; io; Oi or oi, depending on the affect condition; and oo) and the Partner
Selection Form asking how much he/she wished to be a partner with each described target
when participating with him/her in the discussion task. As previously indicated, the
participant was also asked to rank the targets (1-4) in order o f reference. After completing
the dependent measures, the participant was probed for suspicion, fully debriefed, and
excused.
Results
Manipulation Check
A M ANOVA revealed a significant interaction between the type of article
(compliments vs. neutral) and the valence of the adjectives (negative vs. positive)
(F(l,34)=6.89, p<.05), indicating more extreme ratings as a function o f the valence o f the
adjectives for the complimentary articles than for the neutral article. For participants
given the complimentary articles, the endorsement o f positive adjectives (M =2.11)
exceeded that o f negative adjectives (M=3.41), (t (18)=4.25, pc.OOl), whereas for those
given the neutral article the endorsement of negative (M=3.08) versus positive adjectives
(M =2.89) did not differ (t(16)=.63, g>.05). Thus, participants reported that positive
adjectives were better descriptions o f their feelings about complimentary articles than
negative adjectives. In analyses o f specific ratings o f the 14 adjectives, the
complimentary articles were reliably seen as more kind (t (34)= -4.92, p<.001), friendly (t
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect o f Affective 19
(34)= -4.02, g<.001), warmer (t (34)= -2.71, £<.01), and less cold (t (34)=2.67, £<.05),
and angry (t (34)=2.62, £<.05) than the neutral article. Additionally, the lack o f an effect
o f category dimension (F(5,13)= 1.70, £>.05) suggests consistency in the level of positive
affect induced by the different complimentary articles used in this experiment. In sum,
participants experienced complimentary articles as more positive than the neutral article.
Partner Evaluations
As previously indicated, two o f the 49 participants were excluded from all
analyses because o f suspicion. Tests o f normality were performed prior to examination o f
mean target evaluations o f the four targets as a function o f the type o f article,
complimentary vs. neutral. They indicated that the distributions for each o f the four
targets (ii,io,oi,oo) were normal (£<.05) within each affect condition, with the single
exception that the distribution o f the oo target in the no-compliment condition was only
marginally normal (£=.08). Therefore the use o f A NO VA procedures were deemed
appropriate for the subsequent analyses.4
A MANOVA revealed a significant interaction between the type o f article
(compliment vs. neutral) and the type o f target (F_(3,135)= 4.98, £< .01), as well as a main
effect for type o f target (F(3,135)=39.92, £ < .0 0 1), but no main effect for the type o f
article (F j4,45)= 1.94, £ > .0 5 ).5 M ost important and as predicted, the preferences for the
crossed target (Oi) who was a member o f the category that provided compliments
(M=4.79) exceeded that for the corresponding crossed target (oi) in the neutral condition
(M=4-22), (t(45)=2.04, £ < .0 5 ).6Corresponding to this effect, (see Table 1) the greater
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect of Affective 20
preference for the Oi (M=4-79) than the io discussion partner (M =3.92) that was found in
the compliment condition (£<.01), was not exhibited in the neutral condition (M =4.22 for
the oi, and M=4-65 for the io; £>.05). Furthermore, as anticipated, the greater preference
for the ii (M =6.00) than the oi (M=4.22) partner that was found in the neutral condition
(g<.001), was not found in the compliment condition (M =5.17 for the ii, and M =4.79 for
the Oi, £>.05). That is, the simple interaction o f affect condition with ii versus oi targets
was significant, F( 1,45)= 10.52, £<.01. As predicted by the contrast effect, the preferences
for the other crossed target (io) in the compliment condition (M=3.92) was reliable lower
that that expressed in the neutral condition (M =4.65). (t(45)=-2.04, £<.05). Pairwise
comparisons among targets (see Table I) also indicated that the preferences for the ii
target were highest among all targets in the control condition (M=6.00). and those for the
oo target were equivalently lowest among all targets in both conditions (in the
compliment M= 3.29, and in the control condition M =3.35).
More generally, the anticipated additive pattern o f preference for targets found in
the control condition (ii>io=oi>oo), was altered in accord with our prediction in the
compliment condition (ii=Oi>io>oo), showing a pattern in which the crossed target
whose outgroup category was the source o f compliments was evaluated as favorably as
the double ingroup target. In sum, positively valenced affective arousal that was induced
by compliments from outgroup members resulted in an increased preference for new
members o f that specific outgroup category as discussion partners. As predicted, the
obtained pattern in the compliment condition (ii=Oi>io>oo) conformed to a pattern that is
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect o f Affective 21
a combination of the Category Dominance and the Conjunction Similarity patterns, as
shown by analyses which used the contrast weights o f 2, 2 ,0,-4, (F( 1, 69)= 51.19,
P < .001). Moreover there was no residual variance, F(l,69)=2.32, £>.05). By contrast, the
pattern o f preferences for targets were additive in the control condition (ii>io=oi>oo), as
shown by application o f the contrast weights o f 2 ,0 ,0 , -2, (F (l, 66)=70.34, £<.001).
Again, there was no residual variance, F (l,66)=3.03, £ > .0 5 .7
Mediation Analyses
Given that integral positive affect (compliments from an outgroup source)
influenced participants’ preferences for discussion partners, we performed mediational
analyses to further examine its effects. Because the Oi and oi targets were the key targets
for our analyses, the regression analyses examined the effects o f the independent variable
(the type o f article: complimentary vs. neutral) on the differential preference for these
specific crossed category discussion partners. Three regression equations assessed
mediation: First, the mediator was regressed on the independent variable; second the
dependent variable was regressed on the independent variable to test for direct effects;
finally the dependent variable was regressed simultaneously on both the independent
variable and on the mediator to examine the link between the mediator and the dependent
measure (Baron & Kenny, 1986).
A mood score for each participant was obtained by taking the average o f negative
and positive adjective ratings, after reverse coding the negative adjective ratings (lower
scores indicate positive mood). The regression o f mood on the type o f article yielded a
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect o f A ffective 22
significant effect, F (l, 34)=9.61, p<.01, (R:= .22), confirming previous analyses o f the
manipulation check data which had indicated that participants who read the
complimentary articles were in a more positive mood than those who read the neutral
article. Consistent with the previously reported ANO VA results, regressing the preference
for the Oi versus the oi targets on the type o f article yielded a significant effect, F (1,45) =
4.16, p<.05, (R: = .09), confirming again the greater preference for the oi crossed target
in the compliment condition than in the neutral condition. Finally, when controlling for
the type o f article, mood had a significant effect on the preference ratings, F (2, 33) =
11.54, pc.OOl, (R2 = .41). With mood simultaneously in the equation, however, the direct
effect o f type o f article on the preference ratings no longer remained significant,
F(l,34)=3.73, p>.05, (its standardized beta coefficient decreased from .38 to .11). Thus,
the regression analyses provide evidence that m ood mediated the effect of the type o f
article on the preference for the oi targets. *
Discussion
By complimenting the participant’s ingroup, outgroup members in Experiment 1
performed a positive act that was likely to be seen as inconsistent with prior expectations
about their attitudes. According to the ultimate attribution error (Hewstone, 1990;
Pettigrew, 1979), when a positive act is performed by an outgroup member, such
behavior is attributed to luck, or aspects o f the situation. The increased preference that
was shown toward a new outgroup member o f the particular outgroup category that was
the source o f compliments not only reflects a generalization o f increased acceptance, but
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect o f A ffective 23
also suggests that under some circumstances people may make internal attributions for
the positive actions o f an outgroup. That is, to put this another way, why was preference
for them increased if their positive behavior w as merely situational and ephemeral? It is
important to note, however, that this result is contextually embedded in a situation o f high
self-relevance in that the crossed targets who were the source o f compliments were
ingroup members on the second social category dimension.
Experiment 2
In Experiment 2 we investigated the effect o f linking negatively valenced
affective arousal (insult) to an outgroup category o f a crossed categorization target. We
expected that by contrast with compliments, insults from outgroup members o f a category
w ill heighten antagonism toward a new member o f this outgroup category and decrease
preference for him or her as a discussion partner.
In this experiment, we also investigated the effect o f the relevance o f the outgroup
category that is the source o f affective arousal on crossed categorization. In a meta-
analytic study, Mullen et al. (1992) examined ingroup bias as a function o f salience,
relevance, and status. The obtained interaction between item relevance and ingroup status
indicated that higher status groups showed increased ingroup bias on more relevant
attributions, as compared to low status groups. Further, Wilder (1984) suggested that
ingroup and outgroup homogeneity and similarity were strongest on items relevant to the
categorization criterion. Correspondingly, in the context o f political affiliation, Kelly
(1989) showed that the more relevant the dimension to the categorization criterion, the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect of Affective 24
greater the perceived homogeneity o f ingroups and o f outgroups.
In Experiment 1, when participants were exposed to compliments from members
o f an outgroup, it focused their attention on that particular dimension o f social
categorization and thereby made it more relevant. The design of Experiment 1, however,
did not allow us to examine the potential effects o f relevancy. In Experiment 2 we added
this factor to the design. Thus, for Experiment 2, a relevant group refers to a category
dimension on which outgroup members have provided insults. The effect o f such insult
from members o f an outgroup category might call attention to negative characteristics as
a general quality of all outgroups. That is, it might prime the negative intrinsic affect
generally associated with outgroups. Consequently, in addition to affecting preferences
for other members o f the category that provided the insult, the effect o f insults from one
outgroup might also extend to other irrelevant outgroups that have not provided insults,
albeit perhaps to a weaker degree. For example, when old adults insult young adults, age
becomes a relevant social category dimension. By contrast, when the ingroup, crossed, or
outgroup status of a target is defined by dimensions of categorization other than age, such
as graduate/undergraduate student and liberal/conservative political ideology, those
dimensions are less relevant. Consequently, the differences in evaluative bias expressed
toward the four targets may not be as large as they are when one of the defining category
attributes o f the four targets has been made relevant by having read insults from members
o f the outgroup on that category dimension. But in comparison to a neutral affect
condition, the priming function of the insult might increase rejection o f targets whose
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect o f Affective 25
outgroup status is irrelevant with respect to the source o f insults. Therefore, in the
relevant condition, relevant groups were subsequently made available among choices for
a discussion partner. By contrast, in the irrelevant condition, with the single exception
that the outgroup category in the Oi description was of the same category as the source of
insult (the relevant group), the target descriptions only referred to irrelevant dimensions.
W e expected stronger differential evaluative bias to be evidenced toward those
crossed categorization targets whose defining category attributes had been made relevant
by the insult manipulation. Specifically, we hypothesized that when the social category
dimension that cues ingroup status for the Ii and Io targets is the same social category
dimension that defines the source o f the insults one has just read (i.e., relevant group), the
preference for this double ingroup and crossed target will be higher than when they are
not. That is, participants in the relevant condition will prefer these targets as discussion
partners more than will participants in the irrelevant condition (for whom these are ii and
io targets respectively). Similarly, when one o f the social category dimensions that
defines the double outgroup status of the oo target is the same dimension that defines the
source o f insults as having outgroup status, the preference for such targets (Oo) as
discussion partners will be reduced by comparison with oo targets who possess outgroup
status on dimensions that differ from that o f an outgroup category whose members had
been the source o f insults.
For example, reading an article containing insults from old adults (outgroup)
increases a young adult’s preference for another young adult as a partner, by comparison
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect o f Affective 26
with persons defined as ingroup members on other dimensions o f social categorization
(e.g., freshmen, US citizen, or USC major). That is, being insulted by old adults may
emphasize one’s self identity as a young adult, and in turn accentuate one’s preference for
ingroup members o f the same category dimension, which in this case is age. Similarly,
after having been insulted by old adults, if one o f the social category memberships o f an
oo target is old adult, a participant’s preference for this Oo target as a discussion partner
may be discem ibly reduced by comparison with oo targets from whom one had not been
insulted.
Method
Participants
Participants were 63 students enrolled in University o f Southern California. They
were recruited from the departmental subject pool and received partial credit toward their
course grade for their participation. Data from three students were excluded from all
analyses because they indicated some suspicion about the cover story and the purpose of
the experiment.
Design
The design was a 3 (insult from an outgroup category-relevant other targets /
insult from an outgroup category-irrelevant other targets / no insult) X 4 (type o f target:
ii, io, Oi, oo) factorial with the first factor manipulated between subjects and the second
manipulated within subjects. When the information for defining the ingroup, crossed, or
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect o f Affective 27
outgroup status of the four targets indicated category membership on a same category
dimension as that o f the insulting outgroup members (Oi) in the article, the experimental
condition was insult-relevant. When the ingroup or outgroup information in the three
descriptions (ii, io, oo) referred to social category dimensions other than that o f the
outgroup members who provided insults in the news article, the experimental condition
was insult-irrelevant. It is important to note that in the no-insult condition, there is no
possible meaningful manipulation of relevancy because the information provided by the
neutral article had no affective valence.
Participants were randomly assigned to insult-relevant, insult-irrelevant, and no
insult conditions prior to the experimental session. A s in the first experiment, the order o f
the four target descriptions was counterbalanced using a Latin-square design, and within
the constraints imposed by the relevancy manipulation, the two pieces o f information
comprising the target descriptions were created by a random combination selected from
among the four distinct social categories identified by the participant as most important.
Materials
The materials paralleled those used in Experiment 1. The Participation
Information Form asked participants to categorize them selves along the 8 dimensions.
The Description o f Other Participants Form, consisted o f four target descriptions, and The
Partner Selection Form, measured the participants’ preferences for their partners in the
discussion task. The articles that were used to manipulate integral negative affect linked
to a particular outgroup category differed from those used in the first experiment in that
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect o f Affective 28
instead o f containing compliments from members o f an outgroup, they contained insults.
Paralleling the procedure o f Experiment 1, eight insulting articles were created. They
allegedly had been photocopied from the university newspaper (D aily Trojan). Examples
of the information contained in the insulting interviews with old adults are: young adults
never make a sufficient effort to create a significant impact on society’s views about
important social issues or to provide alternative views; young people need to adopt a
more adult attitude that would enable them to gain respect from old people; young adults’
activities have a negative effect on the image that other countries have o f our own society.
The neutral article was identical to that used in the first experiment.
Procedure
General cover story. The procedure and cover story used in the first study were
used again in this experiment. The only procedural distinction between the second and the
first study was the addition o f the relevancy manipulation and the change in affective
content o f the articles (insulting instead of complimentary).
Manipulation o f insult. After the participant completed the Participant Information
Form, he/she was given an article, instructed to read it carefully, and be prepared to recall
its content later. In the insult condition, this page consisted of several insulting interviews
with one o f the participant’s outgroup categories selected randomly among the four
important categories as indicated in the Participant Information Form. As indicated, in the
control condition, the participant was given the same neutral article containing no insult
as was used in the first study.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect of Affective 29
Manipulation of relevancy. In the insult-relevant condition, one piece o f category
membership information in each target description was from the same category
dimension on which outgroup member had provided insults. For example, if a participant
read interviews in which old adults insulted young adults, then young/old adult was
relevant, but all other categories were irrelevant. In this case, the ii was described as a
young adult and as a member o f one o f the other ingroup categories, the io target w as
described as a young adult and as a member o f one o f the other outgroup categories, the
Oi target was described as an old adult and as a member o f one of the other ingroup
categories, and finally the oo target was described as an old adult and as a member o f one
o f the other outgroup categories. The non-relevant dimensions were randomly selected
from the three important dimensions that remained after one o f the important dimensions
was selected for the insult dimension. That is, in the insult-irrelevant condition, the social
category dimensions in all target descriptions except that o f the Oi, differed from the
insult category. For example, if the insult category was age (selected randomly from
among the four important categories), then young/old adult was considered as relevant,
and the 3 other dimensions were considered as irrelevant categories with respect to age
dimension. Hence, the ii, io and the oo targets were created by the random combination of
two descriptions from three irrelevant categories, whereas Oi was described as an old
adult and as one o f the ingroup category among the 3 irrelevant dimensions.
Dependent measures. After the participant read the article, he/she was asked to
indicate his/her relative preference among the four other participants described on the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect o f Affective 30
Description o f Other Participants Form as partner for the discussion task. He/she also
ranked the targets (1-4) in order o f preference on the same form. At the end o f the
experiment, the participant was probed for the suspicion about the credibility o f the cover
story, fully debriefed and excused.
Results
Manipulation Check
In Experiment 1, all participants in the experiment filled out the mood
questionnaire before they completed the dependent measures. It is conceivable that in
Experiment 1 the obtained interaction between the experimental manipulation o f mood
and the category information that differentiated the four targets was contingent on the
presence o f the mood manipulation check because it functioned to maintain or crystallize
the induced mood or because it invoked demand effects. In order to mle out such rival
interpretations o f the effects, for Experiment 2 the impact o f the experimental
manipulation o f affect was assessed in a separate experiment by distributing the eight
insulting articles to 83 students and the neutral article to 18 students on the university
campus. Each student was asked to read the articles and complete the same questionnaire
used in the first study, indicating their feelings about each article on a 4-point scale.
Again, lower scores on the negative adjectives indicated more negative feelings about the
article, whereas lower scores on the positive adjectives indicated more positive feelings
about the article.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect o f Affective 31
A M ANOVA revealed a significant interaction between the type o f article
(insulting vs. neutral) and the valence o f the adjectives (F (l,99)= 8.84, gc.O l), indicating
more extreme ratings as a function o f the valence o f the adjectives (positive/negative) for
the insulting articles than for the neutral article. As anticipated, for insulting articles, the
endorsement o f negative adjectives (M=2.03) exceeded that o f positive adjectives
(M =3.54), (t(82)=l 1.46, g < .0 0 1), whereas no differential endorsement o f negative
(M =2.64), versus positive (M=3.20) items was found for the neutral article (t( 17)= 1.82,
£>.05). Thus, participants reported that negative adjectives were better descriptions o f
their feelings about insulting articles than positive adjectives. Analysis o f the individual
ratings on the 14 adjectives as a function of type o f article showed that insulting articles
were experienced as colder (t(99)= -2.10, £<.05), and more irritating (t(99)= -2.48,
£<.05), annoying (t(99)= -3.45, £<.01), and insulting (t(99)= -2.26, £<.05), but also, less
enjoyable (t(99)= 2.26, £<.05), and fair (t(99)= 2.10, £<.05). Finally, the category
dimension that was selected as the source of the insults had no significant effect on mood,
(F(7,75)=.82, £>.05). Thus, the results were consistent over the eight different social
categories that were sources o f insult. In sum, insulting articles were less positive and
more negative than the neutral article.
Partner Evaluations
Tests o f normality were performed prior to examination o f mean preferences for
the four targets as a function of the type o f article (insulting vs. neutral) and the relevancy
o f the target category information to the insulting category (relevant vs. irrelevant
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect o f Affective 32
condition). They indicated that the target ratings were distributed normally (£ < .0 1) within
both conditions, with the single exception that the distribution for the ii target in the no
insult condition was only marginally normal (£=.20). Therefore the use o f ANO VA
procedures was deemed appropriate for the subsequent analyses.9
A M ANOVA revealed a significant interaction between the type o f condition
(insult-relevant vs. insult-irrelevant vs. no-insult) and the type o f target, (F (3 ,1 7 1)=2.73,
£<.05), and a main effect o f the type o f target, (F(3,171)=47.26, £<.001), but no main
effect for the type o f condition, (F(2,57)=1.14, £>.05). Whereas reliable simple
interactions were found between the insult-relevant vs. no-insult conditions and the type
o f target, (F (3,114)=3.55, £<.05), and between the insult-irrelevant vs. no-insult
conditions and the type o f target, (F(3,l 14)=3.68, £<.05), there was no hint o f a simple
interaction between the insult-relevant vs. insult-irrelevant conditions and the type of
target, (F (3,l 14)=.70, £>.05). Thus, these analyses, in combination with the pairwise tests
between the relevancy/irrelevancy conditions, shown in Table 2, make it clear that in no
instance did relevancy have an effect on participant’s preferences for targets as partners.
Consequently, we collapsed across this factor in subsequent analyses. 1 0
As predicted and shown in the pairwise comparisons presented in Table 2,
preferences for the Oi targets (the source of the insult) were lower in the insult condition
(M= 3.63) than for the oi targets in the neutral condition (M= 4.60), t(58)=-2.71, £< .01.1 1
Hence, as expected, when insults were received from a specific outgroup, participants
tended to reject others from that same outgroup category as discussion partners. In accord
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect of Affective 33
with this effect, the greater preference for the io partner (M=5.23) than the Oi partner
(M =3.63) found in the insult condition (£<.001), was not evidenced in the neutral
condition (M =4.50 and M=4.60. respectively, £>.05). Additionally, as expected, the
evaluations o f the Oi target (M=3.63) in the insult condition did not differ from those o f
the oo target (M=3.20, £>.05). By contrast, preference for the oi partner (M=4.60)
exceeded those for the oo partner (M =2.60) in the neutral condition, £<.001. As predicted
by the contrast effect, the evaluations o f the io target in the insult condition (M =5.23) was
reliable higher than in the neutral condition (M=4.50), (t(58)=-2.34, £<.05). Finally,
pairwise comparisons showed greatest preference for the ii partner over all other targets,
both in the insult (M=5.73) and the control conditions (M=5.40).
Again, we applied contrasts weights to test whether the patterns o f preference
across the four targets matched our predictions. As expected, the pattern o f outcome
across the target evaluations in the insult condition (ii>io>Oi=oo) supported a pattern that
combines features of the Conjunction Dissimilarity and the Dominance patterns, as
shown by application o f the contrast weights o f 4, 0, -2, -2, (F (l,l 17)=93.82, £<.001).
The residual variance was not significant, F( 1,117)=2.37, £>.05). Replicating the pattern
found in the control condition of Experiment 1, w e confirmed an Additivity pattern
(ii>io=oi>oo) in the control condition, as shown by analyses that used the contrast
weights of 2, 0 ,0 , -2, (F( 1,57)=39.40, £<.001). Again, there was no residual variation,
(F (l,5 7 )= 3 .0 9 ,£ > .0 5 ).1 2
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect of Affective 34
Discussion
In Experiment 1, compliments from the Oi members increased preference for a
new member o f this outgroup category as a discussion partner, whereas in Experiment 2,
insult from the same group decreased the preferences for a new member o f this outgroup
category as a discussion partner. The obtained pattern conformed to that expected from a
combination o f the Conjunction Dissimilarity and Dominance patterns, as induced by
insult. Thus, these data suggest that the relative preferences for the convergent and
crossed targets as discussion partners were strongly affected by insults from the outgroup
members o f a category dimension. In addition, they support the overall theoretical idea
that affective arousal is indeed a major factor for predicting patterns o f outcome across
the four targets in the crossed categorization paradigm (Miller et al., in press).
In this study, w e also examined the effect o f the relevancy of the dimensions that
defined ingroup/outgroup status of targets other than the source o f insult. We had
anticipated stronger differential preference in response to cues for ingroup versus
outgroup status when these cues were relevant to the category dimension that was the
source o f the insults. However, the data show no hint o f such effects. Perhaps were the
relevancy manipulation linked directly to the impending discussion task, relevancy might
have had greater impact. Specifically, if the elderly was the category that was the source
o f insults, the discussion might allegedly concern a topic such as the solution to social
security financing.
Finally, it is reassuring that the affect manipulation in this study showed its effect
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect o f Affective 35
even when we changed the experimental design so as to separate the manipulation check
from the experiment. Although the studies induced opposing valences oi' affect, the fact
that the results o f Experiment 2 conformed to our prediction as strongly as those o f
Experiment 1 suggests that the outcomes o f Experiment 1 were not a result o f demand
effects created by the presence o f the manipulation check.
Experiment 3
In the previous experiments, we examined the effects o f positive (Experiment 1)
and negative (Experiment 2) integral affect on preferences for the ii, io, oi, and oo targets
as potential discussion partners by making members o f an outgroup category the source
of affective arousal. Experiment 3 was designed to compare positive and negative
affective arousal within a single design, and to examine whether the support found for our
predictions is evidenced in another culture. In accord with the first two studies, we
predicted that people w ill prefer a member o f a particular outgroup as a discussion partner
when they have previously been exposed to compliments from members o f that outgroup,
as compared to those with no such exposure. We also expected those exposed to insults
from members o f an outgroup to reject a member of that outgroup as a discussion partner
more than those who have not been insulted by members of that outgroup.
Finally, because the design o f Experiment 2 assessed manipulation effectiveness
in an independent study and thereby precluded mediation analyses, we reverted to a
design like that o f Experiment 1 in which there were manipulation check data for each
experimental participant. This enabled us to examine mediation effects for insults as well
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect o f Affective 36
as compliments, thereby filling the gap imposed by the design o f Experiment 2.
Method
Participants
Participants were 60 undergraduate students at Bosphorus University in Istanbul
who participated to gain partial credit toward their course grade in an introductory
psychology course. 1 3 The data from 4 of the 60 participants were excluded from all
analyses because they were suspicious about the cover story and the purpose o f the
experiment.
Design
The design was a 3 (compliment from an outgroup category vs. insult from an
outgroup category vs. control) X 4 (type of target: ii, io, oi, oo) factorial with the first
factor manipulated between subjects and the second manipulated within subjects.
Participants were randomly assigned to compliment, insult, or neutral conditions prior to
the experiment, and the order o f the descriptions was counterbalanced as before.
Materials
All three questionnaires (The Participation Informadon Form, The Description of
Other Participants and The Partner Selection Form) were translated into Turkish for the
Turkish sample, and then, back-translated and inspected by the first author to ensure
equivalence. We adjusted the social categories in the Participation Information Form used
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect o f A ffective 37
in the previous experiments to make them more appropriate for Turkish sample. In this
experiment, participants categorized themselves along the dimensions of age (young
adult/ old adult), political beliefs (right oriented/ left oriented), socioeconomic level (low
incom e/ middle income/ high income), place o f birth (Istanbul/ Anatolia/ other), major
(business / LAS/ engineering), university affiliation (Bosphorus/ Istanbul University/
Istanbul Technical University), favorite sport team (Fenerbahce/ Galatasaray/ Besiktas/
none/ other) on this form. The importance ratings o f these categories are presented in the
results section.
Paralleling the previous designs, we used one o f eight complimentary, eight
insulting, or one neutral article for each participant. All articles were created by the
experimenter and allegedly copied onto a page containing the same format and logo as
that o f the university newspaper (Bosphorus). The contents o f the complimentary and
insulting articles were similar to those used in the first and the second experiments. The
neutral article contained neutrally valenced student opinions about university life.
Procedure
Other than the exceptions already indicated, the general cover story and the
experimental procedure o f the previous experiments were used in this experiment without
any modification. As in the previous experiments, after the participant filled out the
questionnaire asking to categorize himself/herself on 8 dimensions, he/she received the
insulting, complimentary, or neutral article. As in Experiment 1, the experimental
manipulations were assessed after each participant read the article. Each participant
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect o f Affective 38
indicated his/her feelings about it and then received the four types of target descriptions
created by the experimenter. Next the participant completed the dependent measure
assessing relative preferences among the four other participants as partners for the
discussion task. At the end o f each session each participant was probed for suspicion
about the believability o f the article and the general cover story and then debriefed.
Results and Discussion
Manipulation Check
The manipulation check measure contained the same 14 adjectives used in the
previous experiments to assess the effect of the articles on mood. Lower scores for the
positive adjectives indicated that the participant found the article more positive, whereas
lower scores for the negative adjectives indicated that the participant found the article
more negative.
A M ANOVA revealed a significant interaction between the type o f article
(complimentary vs. insulting vs. neutral) and the valence o f the adjectives (negative vs.
positive) (F(2,50)=45.69, p < .00l), a main effect o f the type o f article (F(2,50)=6.05,
gc.O l), and a main effect o f the valence (F(l,50)=25.35, pc.OOl). These results indicated
more extreme ratings for complimentary and insulting articles as a function o f the valence
o f the adjectives than for the neutral article.
The endorsement o f positive adjectives (M =2.39) exceeded that o f negative
adjectives (M =3.66) for complimentary articles, t(20)=8.96, pc.OOl, but the reverse was
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect o f A ffective 39
true for insulting articles (M=2.15 for the negative adjectives; M =3.17 for the positive
adjectives, t(14) =-4.01, £<.001). Thus, complimentary articles were evaluated more
positively, whereas insulting articles were evaluated more negatively. An One-Way
ANOVA indicated that the mean positive adjective ratings were also significantly
different across the three conditions, F(2, 51)= 14.65, £<.00 1. The post hoc tests showed
that the insulting interviews (M =3.17) were less positive than both complimentary
interviews (M =2 .18) and the neutral interview (M =2.39). £<.05. Similarly, the mean
negative adjective ratings were significantly different across the three conditions,
(F(2,50)=60.06, gc.OO 1). The post hoc tests showed that insulting interviews (M =2.15)
were more negative than both the complimentary (M =3.63) and the neutral interviews
(M =3.22), £<.05. One-Way A N O VA ’s revealed that the mean ratings o f the 12 adjectives
(snobbish, (F(2,52)= 12.71, £<.001), enjoyable, (F(2,52)=5.81, £<.01), cold,
(F(2,51)= 10.91, £<.001), irritating (F(2,52)= 10.04, £<.001), angry (F(2,52)=45.17,
£<.001), unbiased (F(2,52)=7.26, £<.01), friendliness (F(2,51)= 17.68, £<.001), warm
(F(2,51)= 17.42,_£<.001), annoying (F(2,52)=22.19,_£<.001), fair (F(2,52)=5.57, £<.01),
insulting (F(2,51)=26.67, £<.001), and prejudiced (F(2,52)= 12.06, £<.001)) differed
across the three types o f articles. Comparison of the evaluations o f the complimentary
and neutral articles showed that the complimentary articles were more enjoyable than the
neutral article; similarly, comparisons o f the evaluations o f the insulting and neutral
articles showed that insulting articles were viewed as colder, more snobbish, irritating,
annoying, insulting, angry, and prejudiced, but also less unbiased, friendly, warm, fair
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect o f Affective 40
than the neutral article. In addition, the post hoc tests showed that the insulting articles
were evaluated as colder, more snobbish, irritating, angry, annoying, insulting, and
prejudiced, but also as less enjoyable, unbiased, friendly, warm, and fair than the
complimentary articles.
Partner evaluations. Tests o f normality were performed prior to examination o f
mean preferences for the four targets as a function o f the type o f article (complimentary,
insulting vs. neutral). They indicated normal distributions for all four targets (£<.05)
across all 3 conditions, with the exceptions that the io distribution in the control condition
(£=.2), and the oi distribution in the compliment (£=.2) and control conditions (£=.17)
were not significantly normal. Since, in general, normality was attained and the analyses
o f trimmed means of all targets paralleled the results of analyses o f untrimmed means,
untrimmed ANOVA procedures are reported.
A M ANOVA revealed a significant interaction between the type o f target and the
three types o f articles (F (6,159) =2.30, £<.05), and a main effect of the type o f target
(F(3, 159) =40.82, £<.001), but no main effect o f the type o f article (F(2,53)=.37, £>.05).
1 4 Separate analyses of preferences for the ii and oo targets respectively revealed no effect
for the between-subject manipulation o f affect (the compliment, insult, and neutral
conditions, £>.05) (see Table 3 ).1 5 This was not true of preferences for the Oi targets,
namely the crossed targets who were members o f the outgroup categories that provided
compliments and insults, and the corresponding oi target in the control condition.
Newman-Keuls tests indicated that preference for the Oi target in the compliment
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect of Affective 41
condition (M =4.79) exceeded that expressed in the neutral condition (M =4.25. £<.05). On
the other hand, by comparison with the compliment condition, preference for the Oi target
was reliably lower in the insult condition (M =3.35) than in the compliment condition
(M =4.79, g< .01). Also, the preference for the Oi target was significantly lower in the
insult condition (M =3.35) than in the control condition (M =4.25. p<.05). In sum, when
the compliment came from a specific outgroup member, participants tended to accept
others from that same outgroup category, whereas in the case o f insult from a specific
outgroup member, they tended to reject others from that same outgroup category as
discussion partners. '‘Further, Newman-Keuls tests indicated greater preferences for the
io target in the insult condition (M=4-71) than in the control condition (M =4.20. p<.05).
Therefore, expected elevation in preference for the io target in the insult condition by
comparison with that elicited in the control condition was confirmed the contrast effect
once more.
The pairwise comparison of targets revealed effects that generally supported our
predictions and in no instance contradicted them. The preference for the io (M=4.71) over
the Oi partner (M =3.35) that was found in the insult condition (£<.01), was reversed in
the compliment condition (M=3.95 for the io, and M=4-79 for the Oi, p<.05). In the
neutral condition there was no hint o f differential preference between the crossed targets
(£>.05). Preference for the Oi target (M=4.79) exceeded that for the oo targets (M =3.21)
in the compliment condition (£<.01), and in the neutral conditions (M =4.25 for the oi,
and M =2.95 for the oo, £<.01). In the insult condition, preference for the Oi target as
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect o f A ffective 42
partner (M =3.35) did not differ from that for the oo target (M =3.12, £>.05). Finally, in
the compliment condition preference for the io target (M =3.95) exceeded that for the oo
target (M =3.21. £<.05), and preference for the ii partner was highest among all targets in
all three conditions.
Overall, these data suggest that the relative preference for the four types of target
as discussion partners were predictably affected by the manipulation o f affective arousal
(compliment and insult) from an outgroup source. Consistent with the outcomes o f
Experiments 1 and 2, the results supported the Additivity pattern for the neutral
condition, as shown by application o f the contrast weights o f 2, 0, 0, -2, F (l, 57)=51.43,
£<.001. There was no residual variation, F (l,57)= .59, £>.05. The results for the insult
condition supported a pattern that corresponds to a combination o f the Conjunction
Dissimilarity and the Dominance pattern (ii>io>O i=oo), as shown by application of the
contrast weights of 4, 0, -2, -2, F (l, 48)=69.18, £<.001. Again the residual variation was
not significant, F (l,48)=1.55, £>.05. Although the results for the compliment condition
support a pattern that is a combination o f the Conjunction Similarity and the Category
Dominance patterns (ii=Oi>io>oo), shown by the application o f the contrast weights o f
2, 2, 0, -4 (F (l, 54)=23.61, £<.001), there was a significant residual variation
F(l,54)=6.55, £<.05. As previously indicated, the pairwise comparisons showed that
preference for the ii target reliably exceeded that for the Oi target, thereby suggesting a
pattern that combines the Additive and the Category Dominance patterns (ii>O i>io>oo).
Indeed contrast analysis o f the weights (3, 1,-1, -3) was highly reliable, F( 1,54)=30.00,
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect o f Affective 43
£<.001, and left no residual variation, F(L54)=.14, £>.05. These results suggest cultural
differences with respect to the ii target. When we examine patterns with the ii target
excluded, by the applying contrast weights o f 2 ,0 , -2 to Oi, io and oo targets, the patterns
in Study 1 (Oi>io>oo, F (l,46)=28.72, £<.001) and Study 3 (O i>io>oo, F (l,36)=9.67,
g c .0 1) are identical. In both analyses, there was no residual variation (F( 1,46)=.27, £>.05,
and F (l,36)=.016, £>.05. respectively). Extending this approach, we also excluded the ii
ratings from the comparison o f the insult conditions in Study 2 and Study 3, by applying
contrasts weights of 2, -1, -1, to io, Oi, and oo. The results supported our previous
findings showing that the pattern in the insult condition o f Study 3 (io>Oi=oo) replicates
that of Study 2 (io>Oi=oo), (F (l,78)=56.16, £<.00 1, and F (l,32)=21.5, £<.001,
respectively). Again in both analyses residual variation was not significant (F( 1,78)= 1.92,
p>.05, and F(l,32)=.42, p>.05, respectively).
In sum, the results with Istanbul college students show that insults and
compliments from members o f a specific outgroup differentially affect preferences for
crossed category targets who are members o f these respective outgroups. They replicate
the results found with Los Angeles students in Experiments 1 and 2. Within the context of
this overall consistency, however, preference for the ii target appears to differ across
cultures. In a subsequent section we directly examine cross cultural effects. First, in
parallel with our analyses o f Experiments 1 and 2, we present mediational analyses.
Mediation analyses. W e performed sequential simultaneous regression analyses to
test whether mood mediated the effects o f type o f article (insulting vs. complimentary) on
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect o f A ffective 44
the preferences for the Oi target (Baron & Kenny, 1986). A mood score was obtained by
taking the average o f negative and positive adjective ratings, after reverse coding the
positive adjective ratings (higher scores indicate more positive mood).
First, the regression o f mood ratings on the type o f article yielded a significant
effect, F( 1, 34)=70.51, pc.001, (R: = .67), confirming the manipulation check analyses
which showed that that participants who were exposed to insulting interviews were in a
more negative mood than those who read the complimentary articles. Second, in accord
with the ANOVA results we have previously reported, regressing the preference ratings
of the Oi on the type o f article confirmed a reliable effect, F (1,34) = 12.33, g c.O l, (R: =
.25), indicating less preference for the Oi in the insult condition than in the compliment
condition. Finally, mediation was tested by regressing the preference ratings o f the Oi on
the mood. When controlling for type o f article, mood had a significant effect on the
preference ratings, F (2,30) = 5.94, pc.O l, (R: = .28). The direct effect of type o f article
on the preference ratings, however, no longer attained significance with mood in the
equation, F(l,31)=3.74, p>.05 (its standardized beta coefficient decreased from .51 to
.27). These analyses show that mood mediated the effect o f the type of article on the
preference for the Oi target.
Cross Cultural Effects
To assess the effect of culture, MANOVA was applied to the data from all three
studies by creating a 2 (culture: American vs. Turkish sample) X 3 (type o f article:
complimentary vs. insulting vs. neutral) X 4 (type o f target: ii, io, oi, oo) factorial design,
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect o f Affective 45
in which the neutral conditions o f Experiment 1 and 2 were pooled and in which the
compliment and insult conditions o f the two US studies were collapsed to parallel the
design o f the Turkish study. The results revealed no interactions other than the previously
reported (implicitly between Studies I and 2, and explicitly within Study 3) interaction
between type o f article and type of target (F (6,471 )=7.96, gc.OO I), and no main effect of
culture (F(1,157)=.27, p>.05). Despite the absence o f significant interactions that many
mistakenly (Hall & Rosenthal, 1995; Rosenthal & Rosnow, 1985, 1991) believe
necessary in order to justify further statistical exploration o f their source, we performed
additional analyses because theory suggests a specific a priori hypothesis which received
some indirect support in our analyses o f the patterns o f effect across the targets o f the
Turkish data. Triandis (1989) has posited that there is a difference in behavior toward
ingroups in collectivist and individualist cultures. Collectivists perceive survival as
possible only with the support o f their ingroups (Vassiliou & Vassiliou, 1974). They are
more cooperative toward and show stronger harmony within ingroups (Triandis,
Bontempo, Villareal, Asai & Lucca, 1988). On the other hand, in individualist cultures,
people define ingroups in terms o f others who are similar with respect to social class,
race, beliefs, attitudes and values (Triandis, 1989). The basic unit o f society is the
individual, which allows people to more readily abandon one ingroup to replace it or
invest their identity more heavily in another (Triandis et al., 1988). The cross-cultural
comparison o f the insult, compliment, and neutral conditions showed that the preferences
for the four types of target in US did not differ from those in Turkey in any o f the three
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect o f Affective 46
conditions (£>.05), with the single exception that when exposed to compliments from
outgroup, Turkish students preferred ii targets (M =5.79) to a greater degree than did
American students (M =5.17). F (l,41)=4.00, £=.05. When the data within Turkish and
American samples is pooled across affect conditions, only the preference for the ii target
in Turkey exceeded that found in US, (t(42)=2.01, £<.01). 1 7 Additional cross cultural
comparisons o f the compliment and insult conditions showed that when exposed to
compliments from outgroup members, the preference o f Turkish students for the Oi target
(M =4.79) did not differ from that o f the American students (M=4.79). Likewise, when
exposed to insults, their preference for the Oi target (M =3.63) did not differ from the
preference o f American people (M=3.35, £>.05)
General Discussion
This set o f studies was designed to explore the effect o f positive and negative
integral affect from an outgroup source on preferences for the four targets (ii,io,oi,oo) o f
the crossed categorization paradigm. W e induced positive or negative affect from an
outgroup source by exposing participants to a set o f complimentary and insulting
interviews with members o f a single outgroup within the context o f a newspaper article.
W e expected compliments from members o f an outgroup to increase preference for a
member o f that particular outgroup as a discussion partner. When participants were
exposed to insults from an outgroup w e expected the opposite direction o f effect, namely,
a decreased preference for a member o f that outgroup as a discussion partner. In
Experiment 3, which was conducted in Turkey, we examined the effects o f manipulated
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect o f Affective 47
positive and negative integral affect within a single study.
Our major predictions were confirmed across studies and across cultures.
Compliments from members o f a specific outgroup increased the preference for a
crossed-category (Oi) member o f that outgroup as a potential discussion partner, by
comparison with a potential crossed-category discussion partner who had not been
associated with complimentary remarks (oi). It is important to note that these results were
obtained for a crossed target who was not one o f the specific outgroup persons who had
provided compliments, but instead was simply a member o f that outgroup category.
Likewise, insults from members o f an outgroup category decreased participants’
preference for a crossed target who was a member o f an outgroup that had been the
source o f insults (Oi) by comparison with a crossed target who had not been associated
with derogatory remarks about one o f the participant’s ingroup identities (oi).
Mediational analyses that used manipulation-check data on the m ood induction
showed that mood, and not some other aspect o f the news article used to manipulate
mood, mediated the obtained effect. We could not perform mediational analyses in
Experiment 2 because w e altered the experimental design o f that study by assessing the
effectiveness o f the mood manipulation in an independent manipulation check study.
Although this enabled us to rule out demand effects as a rival interpretation o f the
outcome, it precluded the application o f mediational analyses. However, in Experiment 3
w e reverted to the design o f Experiment 1 and assessed participants moods after they had
read the complimentary, insulting, or neutral news article. This enabled us to confirm that
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect o f Affective 48
subjectively experienced differences in mood (and not some other inadvertent difference
between the complimentary and insulting news articles) explained the effects o f the m ood
induction on preferences for discussion partner.
In these experiments we used an idiographic design in which the social category
dimensions that were used to create the four combinations of targets for the crossed-
categorization paradigm were selected individually by each participant to be of
moderately high importance. This not only enabled us to reduce between-condition
variance in importance, but also, within each participant, to reduce between target
variation in importance. While this procedure could only increase the sensitivity o f our
experimental designs, it does constrain the generality o f our results to relatively important
social category dimensions. Whether the reliable generalization effects concerning
preferences for crossed-category targets that we induced by experimental manipulation of
positive and negative integral affect w ill be found when the social category dimensions
are o f low importance remains to be seen. Certainly additional research needs to exam ine
the interaction between manipulated affect and the importance of the category
dimensions.
In this context, it is instructive to think again about our attempt to manipulate the
relevancy: viz. the match between (a), that outgroup category dimension that was
experimentally associated with the affective induction, and (b), the social category
dimensions that defined the status o f the other targets available as potential discussion
partners (ii, io, & oo). Specifically, category relevance is related to the considerations w e
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect o f Affective 49
have raised about the potential impact of category importance. That is, if, as w e have
argued, the association of affect with a particular outgroup category increases the
dominance o f that category, it should simultaneously make that category more important.
Therefore, other targets whose status as ingroup or outgroup members is defined (in part)
by ingroup or outgroup membership on the dimension that was the source o f the
manipulated affect should be viewed as more important than are targets defined by other
(irrelevant) dimensions. Indeed, in our meta-analysis o f the crossed categorization
literature (Urban & Miller, 1997), differential importance o f the two category dimensions
o f the crossed-categorization paradigm was associated with dominance effects in the
relative preference for the io versus the oi targets. Consequently, it is noteworthy that our
attempt to manipulate relevance in Experiment 2 by specifically selecting categories that
were linked or unlinked to the affect manipulation failed to have any discernible effect.
Admittedly, we had no manipulation check on the effectiveness o f the relevancy
manipulation. Although the participants’ importance ratings o f the social category
dimensions could have been used for this purpose, they were obtained prior to the
induction o f affect and thus cannot speak on the issue. And perhaps, as we suggested in
our discussion o f Experiment 2, the relevancy manipulation could only have been
expected to be effective if the discussion topic had been selected so as to be perceived as
related to the relevant category dimension. By contrast, however, category dimensions
that participants judge to be generally important are ones that have been salient to them in
countless episodes o f life, and not just for a single short discussion required by an
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect o f Affective 50
experiment.
We also tested the pattern o f preferences across the four targets within the context
o f our data sets. Overall, the obtained patterns were congruent with our predictions,
showing support for a combination o f Conjunction Similarity and Dominance patterns in
the compliment condition (ii=Oi>io>oo), Conjunction Dissimilarity and Dominance
patterns in the insult condition (ii>io>O i=oo), and an Additive pattern (ii>io=oi>oo) in
the absence o f any manipulation o f affective arousal. Although the pattern o f results in
Experiment 3 also provided some cross-cultural support for the pattern elicited by the
compliment condition in the US, a pattern that combines features o f the Additive and
Dominance patterns (ii>Oi>io>oo) provided a better fit for the Turkish data in that this
latter pattern left no significant residual variance. The slight variation obtained between
the effects o f compliments on the patterns across targets in the Turkish and US data may
reflect greater preference for the double ingroup target in Turkey.
In all three studies, the affect manipulations induced anticipated differentiation in
the preferences for the crossed target who was not the source o f the affect manipulation
(io). In Study 1, preferences for this target in the control condition reliably exceeded
those expressed in the compliment condition, but in Study 3, too, there was a directional
hint of this effect. By contrast, in Study 2 and Study 3, preferences for the io targets in the
insult condition exceeded those expressed in the control condition. To examine this effect
more systematically, a 2 (manipulation o f affect: affect vs. control) X 2 (condition:
compliment vs. insult) X 2 (culture: US vs. Turkey) factorial ANOVA analysis was
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect o f A ffective 51
applied to the pooled data o f the io ratings. The result revealed an interaction between
affect and condition, F( 1, 175)=7.20, £<.01 (as well as the expected main effect o f
condition, F(l,175)=5.36, £<.05). Thus, preference for the io target decreased in the
compliment conditions o f Study 1 and Study 3 (M =3.93), whereas it increased in the
insult conditions o f Study 2 and Study 3 (M=5.07) as compared to control conditions
(M=4.44 in Study 1 and Study 3, and M=4-35 in Study 2 and Study 3). Therefore, the
substantial elevation or reduction in preference for the Oi that was produced by the
compliment or insult relative to the control condition also caused differential preferences
in an opposing direction for the other crossed target by comparison with that elicited in
the control condition. This variation in preferences for the io target was expected by the
judgmental contrast effect (Berkowitz and Knurek, 1969) that is produced by comparison
with the complimenting or insulting Oi target. In this sense, it emphasizes the strong
contribution that affective processes can impose on social comparisons.
Among the social categories used in the present set of studies, being a member of
some categories is more advantageous (e.g., being US citizen, young adult, honors
student or graduate student), than membership in others (e.g., being non-US citizen, old
adult, non-honors student or undergraduate student). Messick and Mackie (1989) claimed
that people who belong to a disadvantageous category will emphasize their simultaneous
membership in an advantageous category in order to improve their social identity.
Vanbeselaere (1996) found that membership in an overlapping advantageous category
reduced the perceived ingroup inferiority o f disadvantaged groups. These findings
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect o f Affective 52
suggest that if one o f one’s ingroup memberships is advantaged, preferences for targets
who share ingroup status with the participant on that advantaged category might be
expected to elicit stronger preference as a discussion partner. By contrast, outgroup
members who are advantaged might be expected to elicit stronger rejection. When w e
examined this factor, however, we found no reliable effects. For both Study 1 and Study
2, we divided all target descriptions into two groups: descriptions which contain at least
one advantaged group (honors student, US citizen, graduate student, or young adult), or at
least one disadvantaged group (non-honors student, non-US citizen, undergraduate
student or old adult). For Study 3 advantaged groups were considered as young adults,
Bosphorus University, high income, birth place (Istanbul), whereas disadvantaged groups
were old adults, Istanbul and ITU University, low income, birth place (other than
Istanbul). The remaining descriptions were coded as neutral for all three studies. For each
target (ii, io, oi, oo), One-way ANO VA’s were applied to the pooled data. The results o f
the preferences for four targets revealed no significant main effect of advantaged
categories; for ii, F(2,160)=2.44, £>.05; for io, F(2,160)=2.61, p>.05; for oi,
F (2,160)= 1.09, p>.05; and for oo, F(2,160)=.81, £>.05. Perhaps, as we speculated with
respect to category relevance, were the discussion task made relevant to the advantageous
and disadvantageous categories, the expected social comparison distortions might appear.
The present work aimed to investigate the effects o f positive and negative integral
affect in the form of compliments and insults from outgroup members. Beyond this,
future research needs to examine the effects o f incidental affect, as well as distinct types
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect o f A ffective 53
o f positive and negative integral affect (e.g., happiness, anger, jealously, anxiety or
disgust). The fact that specific emotional states vary in terms o f their implications for
approach and avoidance responses (Smith, 1993) suggests interesting interactions
between the specific emotion that is aroused and reactions to the targets o f the crossed
categorization paradigm. (Schwarz, 1990).
In conclusion, the theoretical view that affect is a major factor for predicting
patterns o f outcome across crossed categorization targets (Miller et al., in press) was
supported. Our results are important because they experimentally establish what was only
a correlational relationship in our meta-analytic findings. At the same time, by
experimentally substantiating the mediational role o f affect, they buttress the important
theoretical role that meta-analytic procedures can play. That is, our findings confirm that
within social psychological literatures, the use o f meta-analysis in combination with
judges ratings o f the between-study variation in key moderator variables, can provide
strong evidence (albeit correlational) in support o f theoretically derived relationships that
were never examined in the primary literature (e.g., Miller & Carlson, 1990; M iller &
Pollack, 1994, 1995; Mullen, Salas & Miller, 1991).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect o f Affective 54
References
Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable
distinction in social psychological research: Conceptual, strategic, and statistical
considerations. Journal o f Personality and Social Psychology. 51 . 1173-1182.
Bodenhausen, G. V. (1993). Emotions, arousal, and stereotypic judgments: A
heuristic model o f affect and stereotyping. In D. M. Mackie & D. L. Hamilton (Eds.).
Affect, cognition, and stereotyping: Interactive processes in group perception (pp. 13-37).
San Diego, CA: Academic Press.
Bodenhausen, G. V., Kramer, G. P., & Susser, K. (1994). Happiness and
stereotypic thinking in social judgment. Journal o f Personality and Social Psychology. 18.
335-345.
Brewer, M. B., Ho, H. K., Lee, J. Y., & Miller, N. (1987). Social identity and
social distance among Hong Kong schoolchildren. Personality and Social Psychology
Bulletin. 13. 156-165.
Brown, R. J., & Turner, J. C. (1979). The criss-cross categorization effect in
intergroup discrimination. British Journal of Social and Clinical Psychology. 18. 371-
383.
Deschamps, J. C„ & Doise, W. (1978). Crossed category memberships in
intergroup relations. In H. Tajfel (Ed.), Differentiation between social groups (pp. 141-
158). Cambridge: Cambridge University Press.
Dovidio, J. F., Gaertner, S. L., Isen, A. M., Lowrance, R. (1995). Group
representations and intergroup bias: Positive affect, similarity, and group size. Personality
and Social Psychology Bulletin. 21. 856-865.
Eurich-Fulcher, R., & Schofield, J. W. (1995). Correlated versus uncorrelated
social categorization: The effect o f intergroup bias. Personality and Social Psychology
Bulletin. 21. 149-159.
Hagendoom, L., & Henke, R. (1991). The effect o f multiple category membership
on intergroup evaluations in a north Indian context: Class, caste, and religion. British
Journal o f Social Psychology. 30. 247-260.
Hall, J. A. & Rosenthal, R. (1995). Interpreting and evaluating meta-analysis.
Evaluation & the Health Professions. 18. 393-407.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect of Affective 55
Hewstone, M. (1990). The ‘ultimate attribution error’? A review o f the literature
on intergroup causal attribution. European Journal o f Social Psychology. 20. 311-335.
Hewstone, M., Islam, M. R., & Judd, C. M. (1993). Models o f crossed
categorization and intergroup relations. Journal o f Personality and Social Psychology, 64.
779-793.
Johnson, D. W. & Johnson, R. T. (1989). Cooperation and Competition. Edina;
MN: Interaction Book Co.
Kelly, C. (1989). Political identity and perceived intragroup homogeneity. British
Journal o f Social Psychology. 28. 239-250.
Lewicki, P. (1986). Nonconscious social information processing. Orlando, FL:
Academic Press.
M essick, D. M., & Mackie, D. M. (1989). Intergroup relations. Annual Review o f
Psychology, 40 , 45-81.
Miller, N. & Carlson, M. (1990). Valid theory-testing meta-analyses further
question the negative state relief model o f helping. Psychological Bulletin. 107. 215-225.
Miller, N. & Marcus-Newhall, A. (in press). A conceptual analysis of displaced
aggression. In R. Ben-Ari & Y. Rich (Eds.), Understanding and enhancing education for
diverse students: A multi-national perspective. Ramat Gan, Israel: Bar Ilan University
Press.
Miller, N. & Pollack, V. E. (1994). Meta-analytic synthesis for theory
development. In H. Cooper & L. V. Hedges (Eds.). The handbook o f research synthesis
(pp. 457-483). N. Y.: Russell Sage.
Miller, N. & Pollack, V. E. (1995). Use o f meta-analysis for testing theory.
Evaluation and the Health Professions. 18. 370-392.
Miller, N., Urban, L. M., & Vanman, E. J. (1998). A theoretical analysis of
crossed social categorization effects. In C. Sedikides, J. Schopler, & C. Insko (Eds.),
Intergroup cognition and intergroup behavior. Hillsdale, New Jersey: Lawrence Erlbaum.
Mullen, B., Brown, R, Smith, C. (1992). Intergroup bias as a function o f salience,
relevance, and status: An integration. European Journal o f Social Psychology. 22. 103-
122.
Mullen, B., Salas, E., & Miller, N. (1991). Using meta-analysis to test theoretical
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect o f Affective 56
hypotheses in social psychology. Personality and Social Psychology Bulletin. 17. 258-
264.
Pettigrew, T. F. (1979). The ultimate attribution error: Extending Allport’s
cognitive analysis o f prejudice. Personality and Social Psychology Bulletin. 5. 461-476.
Queller, S., Mackie, D. M., & Neddermeyer, K. D. (1994). Perceptions o f group
variability: An investigation o f the effects o f integrally induced affect. Unpublished
manuscript, University of California, Santa Barbara.
Rosenthal, R. & Rosnow, R. L. (1985). Contrast analysis: Focused comparisons in
the analysis o f variance. N. Y.: Cambridge University Press.
Rosenthal, R. & Rosnow, R. L. (1991). Essentials o f behavioral research:
Methods and data analysis (2nd ed.). N. Y. : McGraw-Hill.
Schwartz. N. (1990). Feelings as information: Informational and motivational
functions o f affective states. In E. T. Higgins and R. M. Sorrentino (Eds.), Handbook of
motivation and cognition (pp. 527-561). N ew York: Guildford.
Schwartz, N. & Clore, G. L. (1996). Feelings and phenomenal experiences. In E.
T. Higgins and A. W. Kruglanski (Eds.), Social Psychology: Handbook o f Basic
Principles (pp. 433-465). N.Y.:Guilford.
Smith, E. R. (1993). Social identity and social emotions: Toward new
conceptualizations o f prejudice. In D. M. Mackie & D. L. Hamilton (Eds.), Affect,
cognition, and stereotyping: Interactive processes in group perception (pp. 297-315). San
Diego, CA: Academic Press.
Stangor, C., Lynch, L„ Duan, C., & Glass, B. (1992). Categorization o f
individuals on the basis o f multiple social features. Journal o f Personality and Social
Psychology. 62. 207-218.
Stephan, W. G., & Stephan, C. W. (1989). Antecedents o f intergroup anxiety in
Asian-Americans and Hispanic-Americans. International Journal o f Intercultural
Relations. 13. 203-219.
Tajfel, H., & Turner, J. C. (1979). An integrative theory o f intergroup conflict. In
S. Worchel & W. G. Austin (Eds.), The social psychology o f intergroup relations.
Monterey, CA: Brooks-Cole.
Triandis, H. C. (1989). The self and social behavior in differing cultural contexts.
Psychological Review. 96. 506-520.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect of Affective 57
Triandis, H. C., & Triandis, L. M. (1960). Race, social class, religion, and
nationality as determinants o f social distance. Journal of Abnormal and Social
Psychology. 61. 110-118.
Triandis, H. C., Bontempo, R., Villareal, M. J., Asai, M., & Lucca, N. (1988).
Individualism and collectivism: Cross-cultural perspectives on self-ingroup relationships.
Journal o f Personality and Social Psychology. 54 . 323-338.
Urada, D. I. (1996). The effect o f positive affective arousal on the crossed
categorization effects. Masters Abstracts International, 34(5) 2083 (University
Microfilms N o. 1379594).
Urban, L. M. (1995). Judgments of persons from crossed social categories:
Psychological importance of categories and insult. Unpublished master’s thesis,
University o f Southern California, Los Angeles.
Urban, L. M., & Miller, N. M. (1997). A meta-analvsis o f crossed categorization
effects. Manuscript in press.
Vanbeselaere, N. (1987). The effects o f dichotomous and crossed social
categorizations upon intergroup discrimination. European Journal o f Social Psychology.
I i 143-156.
Vanbeselaere, N. (1991). The different effects of simple and crossed
categorizations: A result o f the category differentiation process or o f differential category
salience? In W . Stroebe & M. Hewstone (Eds.), European R eview o f Social Psychology.
Vol. 2 (pp. 247-278). Chichester, England: W iley.
Vanbeselaere, N. (1996). The impact o f differentially valued overlapping
categorizations upon the differentiation between positively, negatively, and neutrally
evaluated social groups. European Journal of Social Psychology. 26. 75-96.
Vassiliou, V. G., Vassiliou, G. (1974). Variations o f the group process across
cultures. International Journal o f Group Psychotherapy. 24. 55-65.
Watson, D. & Tellegen, A. (1985). Toward a consensual structure o f mood.
Psychological Bulletin. 98121. 219-235.
W ilcox, R. R. (1996). Statistics for the Social Sciences. San D iego, CA:
Academic Press.
Wilder, D. A. (1984). Predictions of belief homogeneity and similarity following
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect of Affective 58
social categorization. British Journal o f Social Psychology. 23. 323-333.
Wilder, D. A. (1996). Integral and incidental affect as triggers o f stereotyping. In
Sorrentino & E. T. Higgins (Eds.), Handbook o f motivation and cognition. Vol. 3 (pp.
397-422). New York: Guilford Press.
Worth, L. T., & Mackie, D. M. (1987). Cognitive mediation o f positive mood in
persuasion. Social Cognition. 5. 76-94.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect of Affective 59
Footnotes
1 The use o f a capital letter signifies the crossed category that is made dominant.
Associating a category with or linking it to positive or negative affect w ill make it
dominant.
: Previously (Brewer et al., 1987) we did not differentiate between two distinct
types o f Hierarchical patterns. However, Hewstone et al. (1993) drew attention to such a
distinction. We label this pattern Hierarchical Acceptance because it reflects differential
acceptance between the double ingroup target and that crossed target who has ingroup
status on the dominant dimension. In the Hierarchical Rejection pattern (termed
Hierarchical Derogation by Hewstone et al. (1993) because their response measure was
not behavioral) if there is outgroup status on the dominant dimension, then ingroup-
outgroup differentiation is greater on the nondominant dimension (whereas status on the
nondominant dimension is irrelevant when there is ingroup status on the dominant
dimension).
3 Darren Urada developed this idiographic procedure for controlling importance.
We thank him for help in implementing it in this set o f studies and for use o f the three
questionnaires described in the next section (Urada, 1996).
4 Additionally, analyses o f trimmed means (W ilcox, 1996) o f all targets yielded
results identical with the analyses o f means. Therefore, only the analyses o f the latter are
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect o f Affective 60
reported.
5 The mean ratings o f the importance o f the eight social categories (age, political
beliefs, nationality, year in school, major, honors/not honors student, university
affiliation, graduate/ undergraduate student), made on a 5- point scale on which the high
scores indicated greater importance, were all within one scale point o f the midpoint:
M a g e = 3 - 7 5 , M p o l i t i c - 3 - 2 0 , M n a tio n a lity = ^ - ^ ^ ■ > ^ i y e a r in s c h o o l = 2 - 7 5 , h 4 m a j o r = 2 - 9 6 , h 4 h o n o r s = 2 - 5 8 ,
M u n i v e r s i t y = 3 - 1 1 , M g r a d u a t e = 2 . 4 9 . Thus, in an absolute sense the eight social categories were
not judged as very different from each other in importance. Nevertheless, we applied
ANCO VA to confirm that our outcome did not reflect a com plex artifacual interaction
with importance differences. It yielded an interaction between type o f article and type o f
target ( F ( 3 , 1 0 5 ) = 6 . 9 3 , pc.OO 1) that paralleled the form o f the interaction obtained in the
M ANOVA with respect to preferences for the four targets. We therefore concluded that
whatever the differences in importance, they did not contribute to our obtained effects
concerning the interaction between affect and target.
6 We applied Newman-Keuls method to test the difference between all possible
pairs o f means within each condition, and t-test to compare preferences for each target
across two conditions.
7 We examined the correlations between the rankings o f four targets and the
preferences for targets. Correlation coefficients between ii, io, Oi, and oo target rankings
and ratings in the compliment condition were, -.80, -.47, -.66, and -.50, respectively, and
in the control condition were, -.67, -.49, -.63 and -.50, respectively. Therefore, the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect o f Affective 61
rankings were consistent with the ratings o f the targets reported in the text. Additionally,
analyses o f ranks o f four targets yielded outcomes that are consistent with the outcomes
o f the preference ratings and therefore are not reported in detail.
g
Because some have argued that the structure o f mood is bi-dimensional (e.g.,
Watson & Tellegen, 1985) and because compliments from the outgroup members were
used to induce positive affect, w e also tested whether positive m ood per se mediated the
effects o f the manipulation on the dependent measure. Positive m ood ratings were
obtained by taking only the average o f the 7 positive adjectives used for the mood
manipulation check and ignoring the effects o f the manipulation on the negative
adjectives in the manipulation check measure. The results o f these analyses showed that
positive mood, by itself, mediates the effect o f the type o f article on the preference for the
Oi target. In the first step o f sequential simultaneous regression analyses, the regression
o f positive mood on the type o f article yielded a significant effect, F( 1, 34)= 14.09, p<.01,
(R“ = .29), indicating that participants who read the complimentary articles were in a
more positive mood than those w ho read the neutral article. The second step (regression
o f the dependent measure on the predictor variable) is identical to the one in the previous
mediation analysis. In the final step, regression o f the preference ratings o f the Oi on the
positive mood with the type o f article showed that the direct effect for the preference
ratings disappeared with the positive mood in the equation, and the beta coefficient for
this effect decreased from .38 to .02, F (2, 33) = 13.88, pc.001, (R2 = .46).
9 As in the first study, the analyses o f trimmed means (W ilcox, 1996) o f all targets
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect o f Affective 62
yielded results that were identical to those obtained with the analyses o f the untrimmed
means. Therefore, only the analyses o f the untrimmed means are presented.
1 0 The mean ratings o f the importance o f the eight social category dimensions,
which were obtained from the five-point scales o f the Participant Information Form (with
high scores indicating greater importance) show substantial convergence around the scale
midpoint. The means, respectively, were: M-c„=3.88. Mr |j T ^=3.10. Mnationaiity=3-54, Myear
in school- 3.02, ^Iniajoi— 3.23, ^ ilhon ors 2.65. h^niversity— 3.23, ^Igraduate 3.12. Nevertheless, tO
assure that the results o f the reported M A N O V A were not an artifact o f a confound
between the manipulated variables and differences in the importance o f the category
dimensions, we applied ANCOVA, with importance ratings as the covariate. In accord
with the MANOVA results, the AN C O V A yielded the predicted form o f interaction
between type of article and type o f target (F(3,126)=3.81, £<.05).
1 1 As in the Study 1, we applied Newman-Keuls method to test the difference
between all possible pairs o f means within each condition, and t-test to compare
preferences for each target across two conditions.
1 ^
" The results o f the target rankings were consistent with the results o f the target
ratings. Correlational analyses showed that in the insult condition showed reliable
correlations between ii (r=-.68), io (r=-.76), Oi (r=-.55) and oo (r=-.60) target rankings
and ratings (p<.05). In the control condition the correlations were -.75, -.69, -.86 and -.41,
respectively (p<.05).
1 3 According to Triandis et al. (1988) college students are expected to be more
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect o f A ffective 63
individualistic than the average member of the population in most cultures, because they
tend to be from the upper socio-econom ic levels, from urban areas, and they are educated.
Therefore, because we sampled college students in all three studies, the potential effects
o f culture as a relevant between-study factor is at least partially controlled. For this
reason, we did not derive differential predictions for Experiment 3.
1 4 The mean ratings o f the importance o f the eight social categories (age, political
beliefs, socioeconomic level, place o f birth, year in school, major, university affiliation,
favorite sport team), made on a 5- point scale on which the high scores indicated greater
importance, were all within one scale point o f the midpoint: M ^ = 3 .4 5 , Mr |inr=3.07.
M s e s = 2 .8 7 , Mbinhplace= 2 - 4 1,
Myearinschooi=2.66, Mmajor=3.04, Muniversity=3.68, Msport=2.54, indicating that participants
judged the 8 social category dimensions similarly in terms of importance. Nevertheless,
we applied ANCOVA to confirm that our outcome did not reflect a com plex artifacual
interaction with importance differences. The results indicated an interaction between the
type o f article and the type of target (F (3,105)= 6.93, g< .001) that paralleled the form o f
the interaction obtained in the M ANOVA with respect to preferences for the four targets.
Thus, as in the previous studies, the differences in importance did not contribute to our
obtained effects concerning the interaction between affect and target.
1 5 As in Study 1 and Study 2, we applied Newman-Keuls method to test the
difference between all possible pairs o f means within each condition. Because there are
three conditions in Study 3, we also preferred Newman-Keuls method for multiple
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect o f Affective 64
comparisons o f preferences for each target across the three conditions.
1 6 Additionally, analyses o f ranks o f four targets yielded outcom es that are
consistent with the outcomes o f analyses o f the preference ratings. Correlation between
the preferences for targets and the target rankings were reliable in all conditions (insult
condition for ii, r= -.53; for io, r= -.48; for Oi, r=-.55; and for oo. r=-.50; compliment
condition,, for ii, r= -.66; for io, r= -.48, for Oi, r =-.53, and for oo, r =-.82,; and control
condition, for ii r= -.47, for io, r= -.85, for Oi, r= -.46, and for oo, r =-.67, p<.05).
1 7 Because the distribution o f the ii ratings o f the pooled data was not normal and
outliers were detected, the analysis o f the trimmed means (%30) as suggested by W ilcox
(1996) was applied to the data in order to increase the power.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect o f Affective 65
Table 1.
Mean Preferences for Four Types of Target as a Function o f Compliment from Outgroup
Members.
Condition
Four Types of Target
ingroup-ingroup ingroup-outgroup outgroup-ingroup
(source of
compliment)
outgroup-
outgroup
Compliment Mean 5.17b 3.92 d
4.79 b 3.29 e
(N=24) SD
.87 1.18 .88 1.08
Control Mean
6.00 a 4.65 be 4.22 c
3.35 e
(N=23) SD
.95 1.30 1.04 1.30
Note: Seven-point rating scales were used. Higher values indicate more preference for
target. Means that do not share a common subscripts within a row and a column reliably
(p<.05) differ from each other.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect o f Affective 66
Table 2.
Mean Preferences for Four Types of Target as a Function o f Insult from Outgroup
Members and Relevancy.
Condition
Four Types o f Target
ingroup-ingroup ingroup-outgroup outgroup-ingroup
(source of insult)
outgroup-
outgroup
Insult- Relevant Mean 5.65a 5.50ab
3.75d 3 .15d
(N=20) SD
1.31 1.19 1.02 1.18
Insult- Irrelevant Mean 5.80a 4.95bc
3-50d 3.25d
(N=20) SD
1.01 1.15 1.32 1.33
Control Mean 5.40 a 4.50 c 4.60 q 2.60 d
(N=20) SD
1.54 1.00 1.57 .94
Note: Seven-point rating scales were used. Higher values indicate more preference for
target. Means that do not share a common subscripts within a row and a column reliably
(p<.05) differ from each other.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Effect of Affective 67
Table 3.
Mean Preferences for Four Types of Target as a Function of Insult and Compliment from
Outgroup Members.
Condition
Four Types of Target
ingroup-ingroup ingroup-outgroup outgroup-ingroup
(source o f insult
and compliment)
outgroup-
outgroup
Com plim ent Mean 5.79a 3.95c 4.79b 3.21 d
(n=19) SD
1.18 1.58 1.03 1.93
Control Mean 5.95a 4.20c 4.25c 2.95d
(n=20) SD
1.05 1.77 1.10 1.19
Insult Mean 5.7 la
4.71b 3.35d 3.12d
(n=17) SD
.99 .99 1.32 1.41
Note: Seven-point rating scales were used. Higher values indicate more preference for
target. Means that do not share a common subscripts within a row and a column reliably
(p<.05) differ from each other.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
V *
IMAGE EVALUATION
TEST TARGET ( Q A - 3 )
/
✓
y .
1 . 0 E
9 ^
|32
1 3 .6
l.l
1 M
12.2
2.0
1 . 8
1.25
1.4
1 . 6
150mm
IM/4GE . Inc
1653 East Main Street
Rochester. NY 14609 USA
Phone: 716/482-0300
Fax: 716/288-5989
O 1993. Applied Image. Inc.. All Rights Reserved
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Alcohol's impact on triggered displaced aggression

PDF

Displaced aggression as a function of target power

PDF

Evidence for spontaneous situational inferences

PDF

Asymmetries in the bidirectional associative strengths between events in cue competition for causes and effects

PDF

Applying the theory of reasoned action to condom use: The effect of immediate consequences on intention-behavior consistency

PDF

Effects of threat and self-focus on consensual bias in majorities

PDF

Effects of personal resource sufficiency on perceived difficulty and desirability of earthquake preparedness

PDF

Affection and conflict in family relationships

PDF

The Effect Of Positive Affective Arousal On Crossed Categorization Effects

PDF

Effects of bilateral stimulation and stimulus redundancy on performance in processing nonword letter trigrams

PDF

An evaluation of a brief HIV /AIDS prevention intervention using normative feedback to promote risk reduction among sexually active college students

PDF

Effects of a corneal anesthetic on the extinction of the classically conditioned response in the rabbit

PDF

The out-group must not be so bad after all: Generalization of the effects of disclosure, typicality and salience on intergroup attributions

PDF

How casual attribution influences behavioral actions taken by self-aware individuals

PDF

Disgust as negative affect: A crossed categorization study

PDF

Intragroup evaluations, attitude source, and in-group member derogation

PDF

Adolescents' social attitudes: Genes and culture?

PDF

Aging and the use of context and frequency information in ambiguity resolution

PDF

Depressed children and the social and behavioral attributes of their best friend

PDF

Ideological threat, social identity, and consensus estimation

Asset Metadata

Creator Ensari, Nurcan (author)

Core Title Effect of affective reactions by an outgroup on preferences for crossed categorization discussion partners

School Graduate School

Degree Master of Arts

Degree Program Psychology

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag OAI-PMH Harvest,psychology, social

Language English

Contributor Digitized by ProQuest (provenance)

Advisor Miller, Norman (committee chair), Duval, Shelley (committee member), Read, Stephen (committee member)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c16-21234

Unique identifier UC11341122

Identifier 1391081.pdf (filename),usctheses-c16-21234 (legacy record id)

Legacy Identifier 1391081.pdf

Dmrecord 21234

Document Type Thesis

Rights Ensari, Nurcan

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the au...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA