University of Southern California Dissertations and Theses

Projection on outgroups: complementary projection in aversive intergroup contexts

(USC Thesis Other)

Projection on outgroups: complementary projection in aversive intergroup contexts

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

PROJECTION ON OUTGROUPS:
COMPLEMENTARY PROJECTION IN AVERSIVE INTERGROUP CONTEXTS

by
Marija Spanovic

A Thesis Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
MASTER OF ARTS
(PSYCHOLOGY)

December 2008

Copyright 2008 Marija Spanovic

ii
Table of Contents
List of Tables iii
Abstract iv
Chapter 1: Introduction 1

Chapter 2: Pilot Study 1 9

Chapter 3: Experiment 1 14

Chapter 4: Pilot Study 2 32
Chapter 5: Experiment 2 34

Chapter 6: General Discussion 51

References 56
Appendix: Texts of the Newspaper Articles 60

iii
List of Tables
Table 1: List of Traits and their Opposites 10
Table 2: Relevant Complementary Pairs and their Correlations 12
Table 3: Irrelevant Complementary Pairs and their Correlations 13
Table 4: Number of Fearful and Brave Participants who Rated 24
the Partner as Non-aggressive or Aggressive

Table 5: Similarity Correlations of Self and Partner Traits 26
Table 6: Displaced Aggression Means and Standard Deviations for 28
Fearful and Brave Participants as a Function of Trigger and
Partner’s Aggressiveness

Table 7: Relevant Complementary Pairs and their Correlations 32
Table 8: Irrelevant Complementary Pairs and their Correlations 33
Table 9: Means of Outgroup Ratings on Stinginess and 42
Greediness across Conditions

Table 10: Similarity Correlations of Self and Outgroup Traits 46

iv
Abstract
Two experiments investigated complementary projection -- attributing the complement
(viz., the cause) of one’s own trait onto another (Bloom & Houston, 1975). In the
Triggered Displaced Aggression context, only among those who were initially provoked
and then triggered was their fearfulness associated with a perception of an outgroup
partner as aggressive (Experiment 1). Thus, these participants misattributed to their
partner a trait (viz., aggressive) that was complementary to their own trait (viz., fearful).
Among those who were not triggered, self-ratings and their ratings of their partner were
unrelated. Experiment 2 replicated these findings using a different pair of
complementary traits (stingy/greedy) and employing ratings of the outgroup category as
the dependent variable.

1
Chapter 1: Introduction
One’s perception of others depends on a variety of factors. Broadly, they can be
classified into self-variables, target variables, and situational variables. This paper
examines the interaction of the self and the situation on target perception. Research on
motivated perception has recognized that people tend to modify or even distort their
perceptions of others (e.g., Bruner and Goodman, 1947; Klein & Kunda, 1992;
McDonald & Hurt, 1997; Neuberg & Fiske, 1987). Furthermore, this distortion is often
influenced by the traits and feelings of the perceiver. For example, if people think that a
certain trait is highly characteristic of themselves and if it is important to them, that trait
will influence their perception of others (Markus, Smith & Moreland, 1985). Specifically,
those with masculine self-schemas processed schema-relevant behavior of targets
differently depending on the task. They divided a schema relevant film into larger units
of meaning than did a-schematics. However, when the task necessitated attention to
detail in the behavior of the target, schematics attended to smaller units of behavior,
whereas the performance of a-schematics did not change. This processing style also
allowed them to make more inferences about the personality and motivation of the target.
Thus, the self influenced the perception of the target. People often distort reality because
such distortions allow them to see what they want to see (Levine, Chein & Murphy,
1942).
Complementary Projection
Complementary projection is one, among the many ways in which perceptions
can become distorted. It refers to the justification of a negative self-attribute by

2
misperceiving its cause as being due to the characteristic of another -- the attribution of
the complement or cause of one’s own feelings or traits onto another person (Bloom &
Houston, 1975; Campbell, Miller, Lubetsky & O’Connell, 1964). Thus, young girls who
became frightened while playing a scary game on Halloween night judged photographs of
individuals to be more frightening after having played the game (Murray, 1933). In this
case the girls’ own feeling, presumably induced by playing the scary game, caused them
to misattribute to photos the trait (viz., frightening) that is the complement to their own
feeling (viz., frightened). The girls’ distorted perceptions, engendered by becoming
frightened, presumably led them to incorrectly judge that their fearfulness was caused by
other people’s frightening qualities. Thus, perceptions of others are distorted as people
strive to explain or justify their own internal state. Because one’s positive attributes
hardly require justification, the use of complementary projection should be constrained to
one’s negative attributes. In this sense, negative characteristics that are descriptive of
oneself affect one’s perception of others. When frightened, one sees others as aggressive
(Bloom & Houston, 1975; Feshbach & Singer, 1957).
Despite this sporadic evidence supporting the notion that a frightened state elicits
justifying attributions, the most thorough and extensive examination of the attribution of
self-justifying traits to others failed to provide a hint of evidence in support of the
complementary projection process. Campbell et al. (1964) not only examined college
students’ perceptions of peers in their residential unit, but also examined their perceptions
of photos of unknown persons. More importantly, they did not constrain their test of
complementary projection to the specific self-attribute of frightened. Instead, they

3
examined complementary projections that were based on self perceptions across a broad
array of 27 distinct attributes.
Self-justifying behavior is apparently a ubiquitous human tendency (Tavris &
Aronson, 2007) and may be linked to the equally ubiquitous tendency for self-
enhancement (Sedikedes, Gaertner, & Vevea, 2005). What can account for Campbell et
al.’s, (1964) failure not only to find any supportive evidence of complementary projection
with respect to the self-attribute of frightened, but also, their failure to find evidence of
the self-justifying attributions that characterize complementary projection on any of the
26 other complementary trait pairings that they examined? The key to an explanation
may lie in their failure to have elicited any acute motivational states. Specifically, in
contrast to other research on complementary projection, they did not expose the perceiver
to an aversive or threatening situation. Thus, chronic attributes of the perceiver (e.g.,
fearfulness) were not aroused and the perceivers had no motivation to justify them. Their
failure to find support for complementary projections with respect to any of the self-
attributes that they examined suggests that complementary projection only occurs when
one is exposed to a situation that motivates an explanation of one’s internal state. Thus
far, several studies (Feshbach & Singer, 1957; Murray, 1933) seem to suggest that fear-
arousing situations will cause perceivers to misattribute aggressiveness onto innocent
others. However, what remains unclear is whether perceivers project aggressiveness onto
others when they are in other types of aversive situations that are not specifically fear-
arousing. It seems likely that habitually timid or fearful people should become more
vigilant in other aversive or ego-threatening situations even if these situations do not

4
directly elicit fear. Furthermore, they may be prone to interpret ambiguous cues as more
fear inducing and threatening. Thus, because they feel afraid more frequently, they also
may be more frequently inclined to justify their fearfulness by projecting aggressiveness
onto others. And because fearfulness is such a common negative affect, this attributional
style may become sufficiently rehearsed to produce justifying projections for other types
of negative affect toward which they are vigilant.
Additionally, complementary projection is more likely to be directed to targets
who are relevant to the projected trait. Frightened participants rated pictures of males but
not females as more threatening (Hornberger, 1960). Males are relevant in this context
because they are considered to be more threatening than females (Hornberger, 1960).
This suggests that not only may the elicitation of an acute motivational state be a
necessary precursor of complementary projection, but so too may be the presence of an
appropriate or relevant target person.
Complementary Projection in an Intergroup Setting
Although there is some evidence suggesting that the acute state of ‘frightened’
augments the projection of ‘aggressive’ to appropriate interpersonal targets, the
phenomenon has not been studied in intergroup contexts. The wealth of literature
showing ethnocentric bias consistently confirms that outgroups are ‘appropriate’ targets
for negative depiction. Thus, in an intergroup context, those ingroup members who are
chronically frightened (and therefore particularly vulnerable to an experimental induction
of aversive affect) should display complementary projection even in the absence of fear-
inducing stimuli.

5
Groups, as well as individuals can justify their own negative attributes by
attributing their complements to the outgroup. Moreover, intergroup interactions can be
even more threatening than interpersonal interactions. Previous research has indeed
shown that people behave more competitively in intergroup as compared with
interpersonal contexts (e.g., Wildschut, Pinter, Vevea, Insko & Schopler, 2003). Thus,
when a group member’s own fearfulness is made salient by the intergroup context, she is
more likely to attribute aggressiveness to a threatening outgroup member, thereby
justifying her own negative attribute. At the same time, such justifying attribution makes
one’s own negative attribute less threatening to one’s self-concept by allowing one to see
it as being caused by an external source. Such benefits for one’s self-concept cannot be
achieved by other projective mechanisms, such as similarity or contrast projection – the
respective attributions of same and opposite traits to others.
Research on intergroup projection shows that people display contrast projection
when estimating the attitudes of outgroup members (Holtz & Miller, 2001; Mullen,
Dovidio, Johnson and Copper, 1992). When such instances of contrast projection are
found, however, either of two processes may be at work. One the one hand, the contrast
projection of an attitude opposite to one’s own to outgroup members may reflect the need
to differentiate own group from the outgroup in the service of group identity needs
(Tajfel & Turner, 1979). Alternatively, however, it may reflect complementary
projection -- a misperception that justifies one’s own prior category-based ethnocentric
hostility. I mistakenly see an outgroup member’s opposite attitude as the cause of my
dislike of her. Thus, in intergroup settings, when an ingroup member estimates an

6
outgroup’s attitudes one cannot easily differentiate contrast projection from
complementary projection.
When instead, one examines trait attribution, similarity projection is ubiquitous
(Chronbach 1955). Yet similarity projection too can mistakenly be confused with
complementary projection. Consider, for example, the case of perceiving oneself as
aggressive, while at the same time perceiving others as aggressive. At first glance, this
resembles similarity projection because the perceiver sees others as similar to self.
Alternatively, however, it is possible that the others’ aggressive behavior towards the
perceiver is seen as the cause of her own aggression. Consequently, one’s own trait
causes one to misattribute to others the trait that is complementary to one’s own trait
(viz., aggressive). Thus, in this example the projection of aggressiveness onto others
could be a sign of either similarity or complementary projection. As will be more fully
explained in the method section, such potential confounding between these two
explanatory mechanisms – complementary and similarity projection – can be
experimentally resolved by examining pairings of trait self-descriptions and trait
attributions that differentiate the two mechanisms. Campbell et al. (1964) were the first to
attempt to do so.
Although, as indicated, contrast projection of attitudes has frequently been found
in intergroup contexts, its application to trait attributions makes rather surprising
predictions. It implies that when one is thinking negatively of oneself or one’s ingroup
and thereby experiencing identity threat, one would attribute positive traits to the
outgroup, an outcome that seems totally unlikely.

7
The Current Research
Although several studies have demonstrated the occurrence of complementary
projection, the current research is the first to investigate its operation in an intergroup
context. Moreover, as complementary projection was expected only in a context wherein
intergroup relations are experienced as aversive, it was studied within the context of the
Trigger Displaced Aggression (TDA) paradigm (Miller, Pedersen, Earleywine, &
Pollock, 2003) – again, the first such application. The TDA paradigm suited the
purposes of the study well because it allowed manipulation of an aversive situation that
did not arouse fear directly, but would presumably have more influence on those who
habitually are more fearful. In the TDA paradigm participants are initially angered or
provoked and are then subsequently exposed to a second provocation, the trigger. Thus,
in Experiment 1, projective tendencies of angered participants who were subsequently
mildly criticized by an outgroup partner were compared to projective tendencies of
angered participants who were not subsequently criticized. Although aggression is
typically the dependent variable of interest in the TDA paradigm, projection was the main
focus of Experiment 1. This paradigm was used because it allowed examination of
whether the stable self-attribution of fearful, in combination with a provocation, would
provide impetus for the projection of aggressiveness. Even though fright is not
experimentally induced, the effect of a stable tendency to attribute fearfulness is
examined because fear is a frequent emotional response in intergroup contexts
(Jarymowicz & Bar-Tal, 2006; Mackie, Devos & Smith, 2000; Yzerbyt, Dumont,
Mathieu, Gordijn, Wigboldus, 2006). At the same time, the TDA paradigm also allowed

8
examination of the relationship between projection and aggression. Examining the
attributions of aggressiveness in situations wherein fear is not directly induced is
important because it may shed light on people’s tendency to form a stereotype of the
outgroup as aggressive and violent even when outgroups are not behaving in an
aggressive, fear-arousing manner, but instead, by their mere presence, make the fear-
arousing and other aversive aspects of intergroup relations salient.
Unlike Experiment 1 which examined projective tendencies towards a particular
outgroup member, Experiment 2 examined these tendencies towards the outgroup as a
category. Again, complementary projection was expected only when intergroup relations
were experienced as aversive, such as when the outgroup was perceived as threatening.
Additionally, Experiment 2 made a different set of complementary traits salient. Thus,
because the participants’ ingroup attribute of stinginess was made salient in Experiment
2, complementary projection onto the outgroup was expected for the stingy/greedy trait
pair. At the same time, to control for demand effects and to allow an equal opportunity
for similarity projection to be evidenced, the out group was described equally as both
stingy and greedy.

9
Chapter 2: Pilot Study 1
In both experiments, complementary projection was assessed via self and target
ratings on 27 traits. These traits formed 14 trait pairs that Campbell et al. (1964) judged
as complementary. To confirm the a priori assessment of complementary relationships
asserted by Campbell et al., a pilot study was undertaken wherein trained judges rated all
the trait pairs for complementarity. Because we expected complementary projection to
occur only on the traits that were salient in the experimental procedure, the judges also
rated the relevance of each trait and its complement to the experimental procedures.
Method
Complementarity Coding
The trait pairs are listed in Table 1. Fourteen pairs that were initially judged as
complementary were coded for complementarity. In addition to these 14 pairs, an
additional 14 pairs were formed by reversing the 14 complementary pairs. For example,
stingy-greedy was changed into greedy-stingy, because it seemed that some of these
“reversed” pairs could also be considered complementary. All similarity pairs (e.g.,
stingy-stingy) and all contrast pairs (e.g., stingy-generous) were also coded for
complementarity because it was suspected that some of these pairs confound their
respective projection mechanisms with complementary projection. A few traits occurred
on the list several times (see Table 1) and these repeats were excluded. This procedure
resulted in 79 pairs that were coded for complementarity by the author and another coder.
The coder was given the definition and an example of complementarity. The coders
judged if the pair was complementary or not (1 = yes, 0 = no). If they were not sure, they

10
could give a rating of “maybe.” When necessary, a third coder resolved the discrepancy
between the two coders.
Table 1
List of Traits and their Opposites

Trait Opposite
________________________________________________________________________
Stingy Generous
Greedy Temperate
Trusting Mistrusting
Honest Dishonest
Self-reliant Reliant on others
Unreliable Reliable
Fearful Brave
Aggressive Peaceful
Reserved Open
Trustworthy Non-trustworthy
Respectful Disrespectful
Respectable Disreputable
Anxious Calm
Approving Critical
Bossy Easygoing
Capable Incapable
Dependent Independent
Submissive Dominant
Powerful Powerless
Secretive Open
Prying Focused on self
Suspicious Trusting
Honest Scheming
Touchy Easygoing
Critical Approving
Unscrupulous Scrupulous
Gullible Suspicious
________________________________________________________________________

11
Relevance Coding
All of the pairs that were judged as complementary were then coded for how
relevant or salient they were likely to be in the context of the experimental manipulation
of Experiment 1. The coders were all familiar with the experimental manipulations
because they had acted as experimenters in a very similar experiment. They were
instructed to put themselves in the shoes of participants and try to imagine the thoughts
and feelings that participants experienced following the provocation and trigger within
the TDA paradigm. Specifically, six coders first rated the extent to which they would
experience the emotion or feeling characterized by each of the traits (listed in the first
column of Tables 2 and 3) had they been a participant in the experiment. Next, they rated
the extent to which projection of the complement to the partner was justified in the
situation. Both sets of the ratings were made on 4-point linear scales (1 = not at all, 2 = a
little, 3 = somewhat, 4 = highly).
Results and Discussion
Complementary Traits
Complementarity ratings of the two coders showed an adequate overlap (kappa =
.73). A third coder resolved the discrepancies of the cases wherein one coder rated the
pair as complementary, but the other coder failed to do so. This procedure yielded a total
of 24 complementary pairs (listed in Tables 2 and 3) that were then used in the analyses
of the two experiments.

12
Table 2

Relevant Complementary Pairs and Their Correlations
________________________________________________________________________
Self-trait Complement No Trigger Trigger
________________________________________________________________________

Most Relevant
________________________________________________________________________
critical (2.83) critical (3.33) .39 .18
touchy (2.60) critical (3.33) .20 -.04
anxious (2.60) critical (3.00) -.16 .19
fearful (2.00) aggressive (2.57) -.05 .50**
________________________________________________________________________

Note: The traits are listed from most to least relevant (they are ordered by complement
ratings; trait ratings are ordered from most to least relevant within a particular rating of
the complement). Relevance rating is listed in parentheses.
* Correlation is significant at the .05 level (2-tailed).
** Correlation is significant at the .01 level (2-tailed).

Relevant Traits
The relevance codes for self-trait were considered reliable (alpha = .74).
However, when rating the relevance of complementary traits, three of the coders were
judged unreliable because their codes decreased the value of Cronbach’s alpha. These
coders were discarded and the reliability coefficients that resulted when including only
the three remaining coders was judged adequate (alpha for complement = .80). This
allowed composite scores to be created for each respectively by averaging the ratings of
all coders. These composites are listed in the parentheses of Tables 2 and 3. The ratings
of the complement were more indicative of the relevance than the ratings of the self-trait.
As indicated previously, TDA paradigm should not necessarily or directly elicit fear. The
principle of complementary projection, however, proposes that if a person does in fact
feel fearful, their subsequent projection of aggressiveness is likely. This is exactly what

13
the judges rated when rating the relevance of the complement. Thus, those
complementary traits which obtained relevance scores that exceeded 2.5 were defined as
relevant. This procedure yielded four relevant pairs: critical-critical, touchy-critical,
anxious-critical, and fearful-aggressive (Table 2). The relevance ratings for these traits
was in the 2.67 to 3.33 range. The remaining 20 traits were judged as irrelevant and are
listed in Table 3. Their relevance ratings ranged from 1 to 2.33. Both sets of traits were
subsequently examined in the analyses of Experiment 1.
Table 3

Irrelevant Complementary Pairs and Their Correlations
________________________________________________________________________

Self-trait Complement No Trigger Trigger
_______________________________________________________________________
aggressive (2.33) aggressive (2.33) -.21 .30
dependent (1.33) capable (2.00) .41 (p = .05) .21
honest (2.83) honest (2.00) .11 .25
respectful (2.00) respectful (1.67) .07 -.35
respectful (2.17) respectable (1.67) .10 -.41
trusting (1.83) honest (1.67) -.03 .16
submissive (1.67) powerful (1.33) .25 -.33
self-reliant (2.67) unreliable (1.00) -.38 .28
reserved (2.00) reserved (1.00) .15 -.17
reserved (1.83) non-trustworthy (1.00) .13 .20
submissive (1.43) dominant (1.57) .18 .07
bossy (1.33) incapable (1.00) .17 .02
prying (1.33) secretive (1.00) .58** .42*
prying (1.17) prying (1.00) .05 -.18
secretive (1.17) secretive (1.00) -.06 .16
secretive (1.17) prying (1.00) .37 -.03
stingy (1.17) greedy (1.00) .01 -.24
unscrupulous (1.00) gullible (1.00) .15 -.14
non-trustworthy (1.00)non-trustworthy (1.00) .26 .08
unscrupulous (1.00) unscrupulous (1.00) .23 .01
________________________________________________________________________

* Correlation is significant at the .05 level or ** .01 level (2-tailed).

14
Chapter 3: Experiment 1
Initially, all participants were provoked by the experimenter’s harsh evaluation of
their poor performance on a difficult anagram task. Half of them then received a very
mild second provocation, the trigger. It consisted of a bogus outgroup member’s slightly
negative evaluation of an essay which they had been induced to write after the anagram
task. Then, all participants first rated themselves and then rated their partner on two
identical trait lists, thereby providing data for assessing the occurrence of similarity,
contrast, and complementary projection. Though all participants were provoked, only
those who received a trigger were expected to exhibit complementary projection.
Complementary projection was expected only on the specific traits that were
relevant to the displaced-aggression-eliciting context, whereas it was not expected on
traits experimentally judged as irrelevant to the intergroup interaction. No evidence of
similarity projection was expected for an outgroup partner, whereas contrast projection,
were it to occur, was only expected on positive self-traits that one might be motivated to
enhance by ascribing its negative opposite to the outgroup. As discussed earlier, contrast
projection was not expected on negative self-traits.
Specifically, the following hypotheses were tested. 1) In the trigger condition
fearful participants will evidence complementary projection by attributing aggressiveness
onto their partner. 2) In the trigger condition complementary projection will also occur
on other traits that coders rated as relevant, but it will not occur on irrelevant traits or in
the no trigger condition. Furthermore, in the trigger condition a positive correlation is
expected between the relevance ratings and the magnitudes of complementary

15
correlations, but not in the no trigger condition. 3) Although similarity and contrast
projection are generally not expected in the trigger condition, some instances of similarity
projection, are likely in the no trigger condition as well as in the trigger condition due to
the shared similar context within which both the participant and the bogus partner were
embedded. 4) Amongst triggered participants, those strongly identified with their ingroup
are more likely to exhibit complementary projection. 5) Finally, the influence of self
perceptions of fearfulness and perceptions of one’s partner as aggressive on aggression
toward one’s partner was examined.
Method
Participants and the Design
Participants were 61 undergraduates enrolled in the Psychology Department Subject
Pool at the University of Southern California. Thirty six were female and fourteen were
male. All participants were first provoked and then randomly assigned to a × 2 Trigger
(trigger vs. no trigger) × 2 Order of the dependent measures (projection first vs.
aggression first) × 2 (Target rated: Self, Outgroup) × 27 (Traits) mixed design. Trigger
and Order were between-factors, whereas the remaining factors were measured within
subjects. Most of the trait pairings were taken from Campbell et al. (1964). A self-trait
was included in the trait list only if it could be paired with another trait in the self list to
form a complementary pair. Trait pairs were considered to be complementary by first
assessing, for every possible pairing, whether each specific attribute, were it manifested
by another, would justify the response implicit in each of the specific self traits. As
previously indicated, however, when making these judgments, it is important to note that

16
although reliable correlations between same trait pairings ordinarily are interpreted as
evidence of similarity projection, they can also be viewed as indicating complementary
projection. In the case of hostile/hostile, for example, the tit for tat rule that governs most
social interaction (Axelrod, 1984) suggests that one is normatively justified in responding
to a hostile comment or act with hostile retaliation. Thus, only cross-trait pairs can be
diagnostic of complementary projection.
Procedure
Overview. Participants were told that the study was concerned with impression
formation in the business domain. They were also told that they would interact with a
partner who is participating in another room and that they would perform several tasks,
all of which were described as relevant in the business world. Participants first rated
themselves on a self-version of the projection measure, which consisted of 27 traits (see
Table 2 for the list of all traits). Next, they completed an anagram task that served as a
vehicle for provocation and then, they were asked to write a short essay on the topic of
abortion. Participants’ essays were exchanged with their bogus partner, who later
provided an evaluation of its quality. The evaluation of the participant’s essay, which was
always received after the participant had evaluated the bogus essay by the bogus partner,
was only mildly negative and served as the trigger manipulation. Participants then rated
their partner on the same 27 traits on which they had rated themselves and then
completed an aggression measure. The order of these last two measures was
counterbalanced. Finally, participants completed manipulation checks. They were also
probed for suspicion and debriefed.

17
Provocation procedure. Participants were told that one of the skills important in
the business world was one’s verbal ability. To measure their verbal ability they were
asked to complete 15 anagrams in four minutes. Eleven of the anagrams were difficult
(e.g., cconiftesa = confiscate) and they had to be solved while listening to loud (80
decibels) and distracting background music (Stravinsky’s Rites of Spring). After four
minutes, the research assistant interrupted the participant and took the anagram answer
sheet for scoring. A few minutes later, the experimenter entered with the score and
insulted the participant by stating in an irritated tone of voice: “You really got a lot of
these wrong. This data is useless to me. We should probably just start all over, but to be
perfectly honest with you, I don’t want to waste my time.” This provocation
manipulation successfully elicited aggression in several experiments (Bushman, Bonacci,
Pedersen, Vasquez & Miller, 2005; Denson, Pedersen & Miller, 2006; Vasquez, Denson,
Pedersen, Stenstrom, & Miller, 2005; Pedersen, Gonzales & Miller, 2000).
Manipulation of the partner’s group membership. Participants were then asked to
write either a pro-life or a pro-choice essay on the topic of abortion (Pedersen, Bushman,
Vasquez & Miller, unpublished manuscript). They were urged to write something that is
persuasive and that shows their ability to identify with a group and to effectively and
passionately communicate their argument and defend their group's position on it. This
skill was also described as being very important in the business world. Participants were
led to believe that both they and their partner were completing the task at the same time,
that their essays were to be exchanged, and that each of them would be evaluating the
other’s work. They also indicated how important the topic of abortion was to them and

18
how much they identified with their group (pro-choice or pro-life) on a scale from 1 (not
at all) to 10 (very much).
The bogus partner’s essay always expressed a viewpoint opposite to that of the
participant. For example, pro-choice participants always received an essay from a partner
who identified him or herself as pro-life. Thus, the partner always appeared as belonging
to the participant’s outgroup. This same procedure for manipulating partner’s group
membership was successfully used in previous research (Pedersen, Bushman, Vasquez &
Miller, unpublished manuscript). Partner’s gender and ethnicity always matched that of
the participant.
Trigger manipulation. After the participant completed this task, the experimenter
waited a few minutes before bringing the partner’s essay. Participants were given a few
minutes to evaluate the bogus essay and they were then told that their evaluations will be
exchanged. A few minutes later, they received the partner’s evaluation of the
organization, originality, writing style, clarity of expression, persuasiveness of
arguments, and overall quality of their own essay. Participants in the trigger condition
received respective ratings of 3, 3, 3, 4, 4 and 4 on 7-point linear scales with described
endpoints (1 = no good at all, 7 = extremely good), thereby indicating a slightly negative
evaluation. In the space designated for additional comments, the following statement was
provided: “I know that writing an essay from scratch is hard to do, but I would have
thought that a pro-lifer (pro-choicer) would have come up with better arguments.”
Participants in the no trigger condition received a rating of 4, 4, 4, 5, 5 and 5 and they
read the following comment: “I know that writing an essay from scratch is hard to do, but

19
I thought the other participant came up with fairly good arguments.” Though slightly
positive in an absolute sense, this evaluation was judged to be neutral because most
people exhibit a positivity bias and evaluate even totally unknown strangers in a slightly
positive manner (Lau, Sears & Centers, 1979; Rook, Sears, Kinder & Lau, 1978; Sears,
1983). Subjects received this evaluation after evaluating the partner’s essay.
Measures
Projection measure. Participants rated their self- and partner-traits on a 6-point
scale, with low scores indicating the presence of each trait and the high scores indicating
the presence of the opposite traits (e.g., 1 = fearful, 6 = brave and 1= aggressive, 6 =
peaceful ). This measure consisted of 27 traits.
Displaced aggression measure. Participants were told that the fictitious partner
was applying for a Research Assistant position with the Psychology Department and that
the Department was interested in the extent to which research participants recommended
him or her for the Research Assistant position. Specifically, they were asked to provide a
recommendation by rating him or her on the dimensions: intelligent, capable, friendly,
competent and likeable. Participants also indicated their agreement with the following
sentences: “Your partner should be hired as a research assistant,” “You like your
partner,” and “Your partner has a lot of potential.” Thus, this measure consisted of seven
items. The ratings were made on a 7-point linear scale with described endpoints (1 =
strongly agree, 7 = strongly disagree). This questionnaire measures ‘intent to harm’ in
that unfavorable ratings would deprive their partner of the coveted Research Assistant
Position. It has been used successfully as an aggression measure in previous research

20
(Pedersen, Gonzales & Miller, 2000). Such verbal measures assess aggression as reliably
as behavioral measures (Carlson, Marcus-Newhall & Miller, 1989; Giancola &
Chermack, 1996). Subjects then placed this form into a sealed envelope and dropped it in
a slotted box, thereby seemingly making it anonymous to the experimenter.
Manipulation checks. Participants then rated the emotions they felt as a result of
the anagram task on a modified version of the Mood Adjective Checklist (MACL;
Nowlis, 1965). They used a 7-point linear scale (1 = not at all, 7 = extremely so) to
describe their emotion on 28 emotion items (e.g., angry, cheerful, frustrated). Participants
also rated how happy, annoyed, complimented, irritated, pleased, angry and offended the
partner’s evaluation made them feel. The ratings were made on a 7-point linear scale
with described endpoints (1 = not at all, 7 = extremely so).
Results
Participants who were able to discern the true purpose of the provocation
manipulation and who did not believe that the partner was really there were judged as
suspicious. Eleven suspicious participants (6 in Trigger and 5 in No trigger condition)
were excluded from the analyses. A Chi-Square test indicated that there was no
differential attrition of participants across conditions, χ2 (1, N = 61) = .02. p = .88. The
following analyses included 50 participants. Comparable results were found for males
and females so the two groups were combined in the analyses below.
Provocation Manipulation Check
A composite score of negative affect items (angry, grouchy, frustrated, annoyed,
upset, offended and irritable) from MACL was formed (alpha = .91). A mean score of

21
this composite (M = 2.85, SD = 1.44) was comparable to the means obtained in other
research (e.g., Vasquez et al. (2005) found that the mean of negative affect composite in
the provocation condition was 3.10).
Trigger Manipulation Checks
After applying appropriate reverse coding to positive affects, a composite score of
participants’ affective reactions in response to the essay evaluation they received from
their partner’s was formed (α = .90). A 2 (trigger/no trigger) × 2 (order: projection
first/aggression first) ANOVA on the composite score yielded only a main effect of
trigger, F(1, 46) = 53.38, p < .01. Participants who were triggered responded more
negatively to their essay evaluation (M = 4.59, SD = .96), than those who were not
triggered (M = 2.67, SD = .88), thus indicating that the trigger manipulation was
effective.
Manipulation of Fearfulness
Consistent with the theorizing above, the TDA paradigm did not elicit
fearfulness. Participants’ ratings of fearfulness item from MACL in response to
provocation manipulation were examined. Specifically, triggered participants were not
more fearful (M = 1.50, SD = 1.10) than those who were not exposed to the trigger (M =
2.14, SD = 1.46), F(1, 42) = 2.28, p = .14. However, participants who thought of
themselves as fearful felt more fear (M = 2.33, SD = 1.45), in response to the
provocation manipulation than did participants who considered themselves brave (M =
1.55, SD = 1.18), F(1, 42) = 3.98, p = .05. This suggests that fearful participants became

22
more vigilant and even afraid in response to aversive situations that were designed to
elicit humiliation and frustration, but not fearfulness.
Aggression Measure
A 2 (trigger/no trigger) × 2 (order: projection first/aggression first) ANOVA was
conducted on the composite score of the aggression items. Only the main effect of
trigger was significant, F (1, 46) = 8.80, p < .01. Participants who received a trigger were
more aggressive (M = 3.65, SD = .84) than those who did not (M = 2.85, SD = 1.07).
Results for the Projection Analyses
Projection was assessed by correlating self-ratings with the partner ratings. The
correlations of identical traits from the two trait lists allowed the assessment of similarity
and contrast projection (+r = similarity, -r = contrast). The correlation of complementary
traits (e.g., correlation of self-fearful and partner-aggressive) allowed the assessment of
complementary projection.
Fearful-aggressive pair. The triggered participants who reported more
fearfulness rated their partner as more aggressive (r = .50, p < .01). As expected,
participants in the no trigger condition did not evidence complementary projection (r = -
.05, ns). A Z test for comparing two independent correlations indicated that the
correlation between self-fearful and partner-aggressive in the no trigger condition
differed reliably from that for self-fearful and partner-aggressive in the trigger condition,
p < .05. This pairing is examined more closely in the analyses below. Thus, as predicted
by Hypothesis 1, complementary projection appeared on the pair fearful/aggressive that
we a priori judged as the most relevant in the trigger condition. As described in the Pilot

23
Study 1, the coders confirmed our judgments and rated fearful/aggressive as relevant to
the experimental situation. Furthermore, complementary projection was evidenced only
when participants were triggered.
For the analyses below, self-fearful and partner-aggressive ratings were reverse
coded so that higher scores indicated more fearfulness and more aggressiveness. The
correlation between self-fearful and partner-aggressive was not solely due to the fact that
triggered participants were likely to see their partner as more aggressive (M = 4.42, SD =
1.06) than were non-triggered participants (M = 3.83, SD = 1.13), t(48) = 1.90, p = .06.
If the experimental situation made the partner appear more aggressive than in the no
trigger condition and if all participants who received the trigger considered themselves as
fearful, then it could be that the correlation was due to demand characteristics of the
experiment, and not due to complementary projection. In fact, the fearfulness scores
ranged from one to six. Because not all participants in the trigger condition considered
themselves fearful, it is safe to conclude that the correlation depended on participants’
tendency to justify their own trait. Indeed, as indicated, the trigger condition included
participants who considered themselves as brave (the opposite end of the fearful/brave
dimension). Specifically, eight triggered participants considered themselves as fearful
(i.e., they rated themselves above the scale midpoint), while considering the partner as
aggressive (i.e., they rated the partner above the scale midpoint). At the same time, 14
participants rated themselves as brave, while rating the partner as aggressive. These
numbers, as well as the numbers of the participants who fall into the remaining two cells,
can be seen in Table 4. Chi square analysis indicated that there were no significant

24
differences between the cells, χ
2
(1, N=26) = 2.10, ns. Furthermore, even among
triggered brave subjects, lower self-ratings of bravery were associated with higher ratings
of partner aggressiveness (r = .44, p < .07). Thus, consistent with the hypothesis, less
bravery (i.e., more fear) was associated with perceptions of greater partner
aggressiveness.
Table 4

.umber of Fearful and Brave Participants who Rated the Partner as .on-aggressive or
Aggressive

________________________________________________________________________

Trigger
________________________________________________________________________
Partner-Aggressive Partner-Non-aggressive

Self-Fearful 8 0

Self-Brave 14 4
________________________________________________________________________

Order of measurements. Examination of the effect of order of measurements
revealed that complementary projection only occurred when the projection measure
preceded the aggression measure. Under these circumstances, the more fearful
participants considered themselves, the more aggressiveness they attributed to their
partner (r = .56, p < .05). Complementary projection however did not occur for the
triggered participants who completed the aggression measure prior to completing the
projection measure (r = .34, p < .25). This lack of an effect could have been anticipated
here because subsequent measures often show diminished or little effect by comparison
with those positioned earlier (Lindsey & Anderson, 2000). Finally, in the no trigger

25
conditions no evidence of complementary projection was found either for those
participants who completed the projection measure first (r = -.16, p < .61) or those who
completed it second (r = .03, p < .93).
Relevance analyses. To evaluate Hypothesis 2, the correlations of the three
remaining complementary pairs that the coders rated as relevant were examined (Table
2). The correlation of critical-critical in the trigger condition was positive, but it was not
significant (r = .18, ns). However, this may be because this pair confounds
complementary projection with similarity projection. In other words, these correlations
may have reflected magnitudes of similarity projection, rather than complementary
projection. Complementary projection did not appear on touchy-critical. Correlations for
anxious-critical in the no trigger and the trigger condition were supportive of the
hypotheses, but neither was reliable.
The correlations of irrelevant pairs are listed in Table 3. In the trigger condition,
eight of the correlations were positive. The remainder was either negative or virtually
zero. Overall means are listed in Tables 2 and 3. In both no trigger and trigger
conditions, a positive correlation existed on prying-secretive pair (r = .58, p < .01 and r =
.42, p < .05 respectively).
The 24 correlations of all the complementary pairs in the trigger condition were
correlated with the judges’ ratings of relevance. These analyses provided only suggestive
support for the notion that stronger evidence of complementary projection might appear
on trait pairs judged as more relavant to the experimental setting (r = .27, p = .20). A t-
test comparing the correlations of the traits that were categorized as either relevant or

26
irrelevant showed a similar directional but unreliable pattern. The mean correlation of
relevant pairs in the trigger condition (M = .21, SD = .22) only marginally exceeded that
between irrelevant pairs in the trigger condition (M = .02, SD = .24), t(22) = 1.48, p = .15.
These results, though directionally consistent with Hypothesis 2, may have failed to
evidence reliability because of the small number of the traits being compared. Parallel
analyses of the no trigger condition showed no relationship between relevance ratings and
complementary correlations (r = -.08, p = .71). Similarly, a t-test showed that the
magnitude of relevant complementary trait correlations in the no trigger condition did not
exceed those judged as irrelevant, t(22) = -.27, p = ns.
Similarity and contrast projection. Consistent with Hypothesis 3, similarity
projection did not typically occur in the trigger condition. As depicted in Table 5,
participants exhibited similarity projection in the no trigger condition on the following
traits: stingy, trusting, suspicious, scheming, and approving. In the trigger condition,
participants exhibited similarity projection only for the traits anxious and bossy. No
instances of contrast projection were found (Table 5).
Table 5

Similarity Correlations of Self and Partner Traits
________________________________________________________________________

Similarity projection No trigger (N = 24) Trigger (N = 26)
________________________________________________________________________

Stingy .54** -.13
Greedy .28 .15
Bossy .05 .50*
Incapable .31 -.17
Submissive -.18 -.07
Powerful -.29 .15

27

Table 5 Continued
Trusting .50* -.05
Honest .11 .25
Self-reliant -.03 -.35
Unreliable .23 -.20
Secretive -.06 .16
Prying .05 -.18
Suspicious .78** .04
Scheming .42* .22
Unscrupulous .23 .01
Gullible .19 -.14
Anxious .30 .39*
Approving .48* -.06
Fearful .24 -.02
Aggressive -.21 .30
Respectful .07 -.35
Respectable .19 -.28
Touchy .14 .38
Approving .39 .18
Reserved .15 -.17
Non-trustworthy .26 .08
Dependent .05 -.02
________________________________________________________________________
* Correlation is significant at the .05 level or ** .01 level (2-tailed).

Complementary Projection and Identity Strength.
According to Hypothesis 4, strength of identification with one’s position on the
abortion was expected to moderate the magnitude of complementary projection on the
fearful-aggressive trait pair. Although the magnitude of the fearful-aggressive correlation
did not differ for high versus low identifiers, nonetheless the fact that this correlation was
reliable for high identifier (r = .52, p = .03), but not low identifiers (r = .37, p = .37) is
consistent with Hypothesis 4. Because there were only eight weakly identified subjects in
the trigger condition, the lack of complementary projection for weakly identified subjects
may reflect low power. There were no instances of either similarity or contrast projection
for either highly or weakly identified participants on the fearful and aggressive trait pair.

28
Aggression, One’s Fearfulness and Partner’s Aggression.
Addressing Hypothesis 5, a 2 (trigger/no trigger) × 2 (self-fearfulness:
fearful/brave) ×2 (partner aggressiveness (aggressive/non-aggressive) ANOVA
conducted on the composite score of the aggression items showed a main effect of partner
aggressiveness, F (1, 43) = 5.62, p =.02. Participants who rated the partner as aggressive
were more aggressive towards the partner (M = 3.53, SD = .96) than those who did not
think the partner was aggressive (M = 2.53, SD = .87). Despite the absence of reliable
simple effects, there was an interaction between trigger and self-fearfulness F (1, 43) =
3.94, p = .05. The aggression of brave participants in the no trigger condition
directionally exceeded that of fearful participants, t (22) = 1.59, p = .13. Nor was there a
reliable difference in the aggressiveness of fearful and brave participants in the trigger
condition, t(24) = .94, p = .36.
Table 6

Displaced Aggression Means and Standard Deviations for Fearful and Brave
Participants as a Function of Trigger and Partner’s Aggressiveness

________________________________________________________________________

Trigger No Trigger
____________________________________

Self-fearful – partner aggressive 3.50 (.68) 3.94 (1.86)

Self fearful – partner non-aggressive 3.29 (.59) 3.15 (.85)

Self brave – partner aggressive 4.33 (.72) 2.72 (.97)

Self brave – partner non-aggressive 3.47 (.89) 2.60 (1.06)
________________________________________________________________________

29
Discussion
As predicted, complementary projection occurred on the fearful-aggressive pair
only in the trigger condition. In the trigger condition participants’ fearfulness was
reliably related to a perception of greater partner aggressiveness. Thus, this experiment is
one of the first that shows that fearful participants attribute aggressiveness onto the
targets even in those aversive situations that do not directly elicit fear. When the order of
measurements was examined, on two out of the three most relevant pairs complementary
projection occurred when triggered participants completed the projection measure first.
The lack of effects when the projection measure follows the aggression measure is not
unusual, in that subsequent measures often show diminished or little effect by
comparison with those positioned earlier (Lindsey & Anderson, 2000).
Thus, Experiment 1 replicated Murray’s (1933) findings and extended
complementary projection into the domain of Triggered Displaced Aggression.
Additionally, both those participants who considered themselves fearful and those who
considered themselves brave exhibited complementary projection when triggered.
Specifically, even among brave participants, those who perceived themselves as less
brave (or more fearful) attributed more aggressiveness to the partner.
An examination of the remaining three pairs that the coders rated as relevant
revealed that although complementary projection directionally occurred on anxious-
critical and on critical-critical in the trigger condition, these relationships were not
reliable. Nor did evidence of complementary projection appear on the touchy/critical
pair. Complementary projection did not typically occur on irrelevant pairs or in the no

30
trigger condition. The only exception was prying-secretive. Furthermore, although
higher ratings of relevance were directionally associated with greater magnitudes of
complementary correlations in the trigger condition, this relationship was not significant.
As expected, relevance ratings were not related to the magnitudes of complementary
correlations in the no trigger condition. These analyses, therefore, offer only partial
support for Hypotheses 2.
As noted, there were several instances of similarity projection. According to
Mullen et al. (1992), similarity projection should only occur when both the participant
and the partner belong to the same social group. Somewhat consistent with this view,
similarity projection mostly occurred in the no trigger condition wherein the outgroup
member was likely seen as a partner in a cooperative enterprise who underwent the same
experiment at the same time. This may explain why similarity projection occurred even
though the partner was an outgroup member. Alternatively, it is possible that the
intergroup distinction was not very salient. Although their positions on the abortion issue
was the most relevant dimension of comparison, participants’ shared group memberships
on several other dimensions (ethnicity, gender and USC student) may have over-ridden
an attitude-induced attribution of outgroup status. Experiment 2 was designed to address
this problem by making the intergroup distinctions more salient. Participants in the
trigger condition exhibited similarity projection of bossiness and anxiousness to the
partner. This may have been due to the similarity of the contexts wherein participant and
the bogus partner found themselves.

31
Only the triggered participants who were strongly identified with the abortion
issue exhibited complementary projection. This is consistent with the notion that when
an intergroup context is more salient or important it is more likely to elicit
complementary projection. Complementary projection did not occur on any of the
irrelevant traits, their composite, or in the no trigger conditions.
This experiment replicates standard TDA findings (e.g., Pedersen et al., 2000).
Participants were more aggressive in the trigger condition. Furthermore, participants
who rated the partner as aggressive were more aggressive to the partner than were those
who did not think the partner was aggressive. In contrast, in the no trigger condition
those participants with low self-ratings on fearful exhibited less aggression compared to
participants in other cells.
Even though the significant correlation between self-fearful and partner-
aggressive did not appear to be due to the demand characteristics of the experimental
situation, a demand interpretation cannot be ruled out unequivocally. Experiment 2 was
designed with this in mind. To preclude demand interpretations, an outgroup needs to be
seen as possessing equal levels of the complementary trait in all conditions. Specifically,
if the complementary trait in question is greedy, then the outgroup should be made to be
seen as greedy to the same extent in all conditions. Moreover, both traits of the
complementary pair should be equally represented in each condition. For the pair stingy/
greedy, the outgroup should be perceived as possessing the same levels of both stinginess
and greediness. Because this was not possible with fearful/aggressive as traits,
Experiment 2 investigates a different pair of complementary traits, stingy/greedy.

32
Chapter 4: Pilot Study 2
Twenty four self-traits and their complements were coded for their relevance to
the experimental setting of Experiment 2 using coding procedures parallel to those used
for coding relevance in Experiment 1. Six coders read the bogus article that described
competition between LAS and business students. They were asked to put themselves in
the shoes of the LAS participants and rate the applicability of each trait and its
complement to the experimental situation. The reliability was judged as adequate (alpha
for trait = .63; alpha for complement = .66) and composite scores were formed. The traits
that had a complement composite score greater than 2.5 were judged as relevant. This
procedure yielded four pairs, listed in Table 7, that were considered relevant:
aggressive/aggressive, stingy/greedy, non-trustworthy/non-trustworthy, and self-
reliant/unreliable. Their relevance ratings ranged from 2.60 to 3.20. The remaining 21
pairs with the lowest relevance ratings on self-trait and on the complement were selected
as irrelevant and are listed in Table 8. Their relevance ratings ranged from 1 to 2.40.
Both sets of these traits were subsequently examined in the analyses of Experiment 2.
Table 7
Relevant Complementary Pairs and Their Correlations
________________________________________________________________________
Self-trait Complement Control Cooperation Competition
________________________________________________________________________
aggressive (2.40) aggressive (3.20) -.27 -.06 .38(.07)
stingy (2.0) greedy (3) .24 .23 .44*
non-trustworthy (1.80) non-trustworthy (2.60) .52* .22 -.03
self-reliant (1.80) unreliable (2.60) .11 .30 .51*
________________________________________________________________________

* Correlation is significant at the .05 level or ** .01 level (2-tailed).

33

Table 8

Irrelevant Complementary Pairs and Their Correlations
________________________________________________________________________

Self-trait Complement Control Cooperation Competition
_______________________________________________________________________
anxious (2.40) critical (2.40) .26 .09 .22
reserved (1.40) non-trustworthy (2.40) .22 -.37 .17
critical (2.80) critical (2.40) -.11 -.08 .37
touchy (2.00) critical (2.25) .02 .35 .41*
unscrupulous (1.00) unscrupulous (2.20) -.24 -.37 .17
secretive (1.40) secretive (2.20) .17 .05 -.09
submissive (1.70) powerful (2.20) .26 .23 -.15
submissive (1.70) dominant (2.20) -.16 .50* -.33
fearful (2.00) aggressive (2.20) -.07 .05 -.03
dependent (1.60) capable (1.80) .31 .15 -.20
reserved (1.40) reserved (1.80) .56* -.31 -.15
respectful (1.10) respectful (1.80) -.05 .26 .49*
honest (1.60) honest (1.60) .13 .32 -.05
prying (1.10) secretive (1.60) .06 -.25 .06
prying (1.10) prying (1.40) .02 .53* -.12
bossy (1) incapable (1.40) .13 .44 .19
secretive (1.20) prying (1.40) .33 .17 -.18
trusting (1.00) honest (1.40) .20 .26 .11
respectful (1.10) respectable (1.20) .11 -.05 .56**
unscrupulous (1) gullible (1) -.33 -.31 -.22
________________________________________________________________________

* Correlation is significant at the .05 level or ** .01 level (2-tailed).

34
Chapter 5: Experiment 2
Unlike Experiment 1 which examined projective tendencies towards a particular
outgroup member, Experiment 2 assessed these tendencies towards the entire outgroup.
This extension is important in that complementary projection -- justifying one’s own trait
by attributing its complement to the outgroup category-- may be a process that in part
explains stereotype formation. Experiment 2 abandons the Triggered Displaced
Aggression paradigm used in Study 1 and applies a direct aggression paradigm in which
the participant is provoked by an outgroup and then has an opportunity to aggress
towards that outgroup. Additionally, Experiment 2 examines the projection of a different
pair of traits (i.e., stingy and greedy). To make these traits salient, Experiment 2
manipulated the intergroup goal structure (cooperative, competitive, and neutral control)
while using alleged news articles to describe the outgroup as equally both stingy and
greedy. Imposing this equivalence precludes a demand interpretation if complementary
projection were to be obtained. Had the articles described the outgroup as only greedy,
one could not know whether stingy participants’ ratings of the outgroup as greedy were
due to the implicit demand or to their tendency to use complementary projection.
Liberal Arts participants first rated both themselves and the outgroup (viz.,
Business students) on two identical trait lists, thereby providing an opportunity to test
similarity, contrast, and complementary projection. Then, one of three intergroup goal
structures was imposed: competitive, cooperative, and no differentiation. Being in a
competitive goal structure with an outgroup is likely to elicit feelings of threat (Esses,
Jackson & Armstrong, 1998). In response to such threat, I expected people to exhibit

35
complementary projection when depicting that outgroup. Such complementary projection
was expected on the specific traits that were made salient within the intergroup context
(viz., stingy and greedy), whereas it was not expected on traits that were irrelevant to the
intergroup interaction.
By contrast, a cooperative goal structure is likely to diffuse category
differentiation (Gaertner, Dovidio, Banker, Houlette, Johnson & McGlynn, 2000). In the
absence of salient motivations that elicit other perceptual tendencies, assumed similarity
between self and others, appears to be the fallback propensity --whether with respect to
attitudes or other personal attributes such as personality traits (Chronbach, 1955).
Consequently, a cooperative goal structure is likely to elicit similarity projection. In the
third condition, wherein no intergroup differentiation is made salient, the many
similarities between Liberal Arts and Business students, such as age, school attended, the
shared goal of completing college, city of residence, etc., are likely to dominate attention.
Therefore, in that condition, too, similarity projection was expected.
Finally, as previously noted, motivated differentiation is a common intergroup
dynamic that is used in the face of threat to bolster ingroup solidarity and positive
ingroup regard (Tajfel & Turner, 1979). Consequently evidence of contrast projection
was only expected in the Competitive goal structure. And because such differentiation is
invoked in the service of maintaining a positive group identity, evidence of contrast
projection in the competitive condition was only expected with respect to positive self-
traits.

36
To summarize, the following hypotheses were tested in Experiment 2: 1) Only
participants in the Competitive intergroup goal structure will show evidence of
complementary projection by attributing the cause of their own stinginess (i.e.,
greediness) onto the partner; 2) In the trigger condition Complementary projection will
also occur on other trait pairs that the coders rated as relevant, but it will not occur on
irrelevant traits. Furthermore, judges’ relevance ratings were expected to correlate with
the magnitude of complementary correlations of the trigger condition; 3) Similarity
projection was expected in the Cooperative goal structure and in the control condition,
whereas contrast projection was expected in the Competitive goal structure.
Method
Participants and Design
Participants were 72 Liberal Arts students (LAS) enrolled in the Psychology
Department Subject Pool at the University of Southern California. There were 14 males
and 58 females. They were randomly assigned to a 3 (Type of intergroup goal structure:
Competitive, Cooperative, Control) × 2 (Target rated: Self, Business students) × 27
(Traits) mixed design. Type of intergroup goal structure was a between-factor, whereas
the other two factors were measured within subjects.
Overview of Procedure
Participants were informed that the study was investigating perceptions of group
members. Participants first rated themselves on 27 traits. Next, they completed an
identification induction questionnaire that allegedly probed for participants’ awareness of
the issues to be presented in the articles. In reality, the questionnaire helped strengthen

37
Competitive and Cooperative participants’ identity with their ingroup by making them
aware of the competition or cooperation between their ingroup and the outgroup.
Participants then read one of three newspaper articles that were allegedly taken from the
“Daily Trojan,” the campus newspaper. Competitive and Cooperative structures were
induced via bogus news articles (see Appendix) describing Business and LAS student
inter-relations (340 and 342 words). The bogus newspaper articles are described in the
next section. The Control condition used a neutral, non-intergroup article (342 words).
Participants then completed manipulation checks and rated Business students on the
outgroup version of the projection measure, which was comprised of the 27 self-rated
traits. Participants were then thanked for participating and fully debriefed.
Competitive intergroup goal structure. Participants in the Competitive goal
structure condition first completed an identification induction questionnaire designed to
create a conflict between LAS and Business students. These questions, for example,
asked if the participant was aware of the existence of an organization that provides
benefits to Business students at the expense of LAS students. Participants made their
responses on a 5-point Likert-type scale (1=not aware, 2= hardly, 3=somewhat, 4=very,
5=fully aware).
Participants in the Competitive goal structure then read about a conflict between
LAS and Business School students. The article, ostensibly extracted from the Daily
Trojan, described both student groups as being faced with shortages in computer labs and
facilities. Both groups were described as generously devoting their time to obtain more
computer labs and facilities, thus indicating that both groups were generous. LAS

38
students allegedly proposed creation of a fund that would aid in fighting these economic
difficulties. Business students were said to have rejected that same plan due to their
unwillingness to donate their own money. Thus, in addition to their generosity, the
article also implied that Business students are stingy. Finally, Business students were
described as believing that the Office of Academic Affairs should increase funding to
allocate more scholarships to business school students at the expense of LAS students,
thereby indicating that Business students are greedy as well. Furthermore, these trait
descriptions were ordered via Latin square design. The Latin square design varied the
order of trait descriptions within the article such that each trait appeared in each position
once. Specifically, one of the trait orderings was: generous, stingy, greedy, while the
remaining two were stingy, greedy, generous and greedy, generous, stingy.
Cooperative intergroup goal structure. Participants in the Cooperative goal
structure also completed an identification induction questionnaire designed to emphasize
the cooperation between LAS and business students. For example, participants answered
questions concerning whether they were aware that relations between LAS and business
students were friendly.
Participants in the Cooperative goal structure then read a news article about
cooperation between the two student groups. Again, both student groups were described
as generously devoting their time to improve the economic situation on campus. In this
news article, however, the two student groups allegedly agreed on how to improve the
economic situation. Additionally, unlike the Competitive news article, wherein only
Business students were stingy and greedy, in the Cooperative article both student groups

39
were described as similarly stingy and greedy. Again, these trait descriptions were
ordered via Latin square design.
Control condition. The induction questionnaire and the news article for the
control condition were concerned with parking structures on campus. No intergroup
differentiation was described
Measures
Projection measure. The projection measure consisted of two parts: a self-version
and outgroup version each consisting of the same 27 traits that were used in Experiment
1. Participants made their ratings on a 6-point linear scale with described endpoints.
Low scores indicated the presence of a trait, whereas high scores indicated the presence
of the opposite trait (e.g., 1 = stingy, 6 = generous and 1= greedy, 6 = temperate ).
Furthermore, some traits were relevant to the context (i.e., stingy and greedy), whereas
other traits were irrelevant (i.e., reserved and non-trustworthy).
Manipulation checks. After reading the article, participants rated their affective
reactions (e.g., threatened, anxious, vulnerable, calm) in response to the news article.
This measure consisted of 11 items that were both positively and negatively valenced.
Also, they rated the intensity of the conflict between LAS and Business students.
Participants reported the degree to which they found the views and values of the outgroup
threatening and the extent to which outgroup threatened the wellbeing of their group. All
of these ratings were made on a 9-point scale with described endpoints (1=not at all,
9=extremely).

40
Results
Those participants who did not think the articles were real were judged as
suspicious. Eight suspicious participants (3 in the Competitive and 5 in the Cooperative
condition) were excluded from the analyses. Fisher’s exact test indicated that there was
no differential attrition of participants across conditions (p = .26). Consequently, the
following analyses included 64 participants. Comparable results were found for males
and females so the two groups were combined in the analyses below.
Intergroup Goal Structure Manipulation Checks
A composite score of affective reactions towards the news article was formed in
order to assess the effectiveness of the manipulation of intergroup goal structures.
Reverse coding was applied to positive affects (calm and happy) and the internal
consistency was high (α = .85). A one-way ANOVA showed that participants reacted
more negatively to the news article in the Competition condition (M = 4.19, SD = 1.43)
by comparison with the Cooperation (M = 3.00, SD = 1.28) or Control conditions (M =
2.32, SD = .92), F (2, 61) = 13.59, p < .01. A Bonferroni post-hoc test confirmed that
participants reacted more negatively under Competition compared to Cooperation and
Control. The latter two conditions did not differ from each other.
A one-way ANOVA on the intensity of the conflict also revealed a significant
difference between conditions, F(2, 60) = 12.55, p < .01. A Bonferroni post-hoc test
showed that participants perceived intergroup conflict as more intense in the Competition
(M = 6.04, SD = 1.65) than in the Cooperation (M = 3.61, SD = 2.51) or the Control (M =
3.00, SD = 2.26) conditions. The latter two conditions did not differ from each other.

41
A one-way ANOVA on the composite score of two threat items (α =.84) showed a
similar pattern, F(2, 61) = 34.25, p < .01. A Bonferroni post-hoc test showed that
participants experienced more threat under Competition (M = 5.19, SD = 1.44 by
comparison with the Cooperation (M = 2.58, SD = 1.58) or Control conditions (M = 1.89,
SD = 1.25). Again, the latter two conditions did not differ from each other. Together
these analyses indicate a successful manipulation of intergroup goal structures.
Trait Manipulation Checks
In the following analyses, stingy and greedy ratings were recoded so that higher
ratings indicated more stinginess and more greediness. The news articles for Competition
and Cooperation were intended to describe equal levels of both stingy and greedy
behavior in the outgroup. A one-way ANOVA examined whether conditions differed
from one another on the ratings of outgroup stinginess and greediness. This analysis
revealed a significant difference between conditions for both the ratings of outgroup
stinginess (F(2, 61) = 5.96, p < .01) and outgroup greediness (F(2, 61) = 5.14, p < .01).
The Bonferroni post-hoc tests however, revealed that the difference was due to the
control condition. Table 9 shows that participants in the Control condition rated the
outgroup as less stingy and more generous and less greedy and more temperate than did
those in the Competition or Cooperation conditions. This is to be expected of Control
participants who did not receive any information that made the differentiation between
business and LAS students salient. Nor did they receive any information about the
outgroup or its stinginess and greediness. More importantly, the Competition and
Cooperation conditions did not differ from each other on the ratings of outgroup

42
stinginess (p > .05) or outgroup greediness (p > .05). This outcome is consistent with the
notion that the Competitive and Cooperative articles were equally successful in
portraying the outgroup as stingy and greedy. Furthermore, in all three conditions the
levels of stinginess did not differ from those for greediness (t[23] = 1.77, ns under
Competition, t[17] = 1.30; ns under Cooperation; t[21] = 1.40, ns in Control). Thus, in
each condition the outgroup was not seen as differing in their levels of stinginess and
greediness. The means of outgroup stinginess and outgroup greediness ratings for each
condition are displayed in Table 9.
Table 9

Means of Outgroup Ratings on Stinginess and Greediness across Conditions
________________________________________________________________________

Trait
___________________________________

Condition Stingy-generous Greedy-temperate
________________________________________________________________________
Control 3.23 3.64
Cooperation 4.22 4.56
Competition 4.58 4.88
________________________________________________________________________
Projection
The analyses of Experiment 2 parallel those of Experiment 1. Projection was
assessed by correlating self-ratings with the outgroup ratings. The correlations of
identical traits from the two trait lists allowed the assessment of similarity and contrast

43
projection (+r = similarity, -r = contrast). The correlation between complementary traits
(e.g., correlation of self-stingy and outgroup greedy) allowed the assessment of
complementary projection.
Stingy/greedy pair. Consistent with Hypothesis 1, complementary projection was
evidenced only under Competition, and on the traits linked to the nature of the articles.
Under Competition, the more one saw oneself as stingy, the more Business students were
perceived as greedy (r = .44, p <. 05). Furthermore, under Competition, seven stingy
subjects attributed greediness to the outgroup, whereas only one stingy subject failed to
attribute greediness to the outgroup. Generous subjects in the Competition condition
were equally likely to attribute greediness as they were to attribute temperateness. In
other words, eight generous subjects attributed greediness to the outgroup, whereas eight
of them attributed temperateness to the outgroup.
Correlational analyses were also conducted separately for stingy and generous
participants after splitting the stingy-generous variable at the median. The relationship
between self-stingy and outgroup-greedy was particularly strong for stingy participants in
the Competition condition. The more stingy participants perceived themselves as stingy,
the more they perceived the outgroup as greedy (r = .84, p <. 01). The same did not hold
for generous participants (r = .19, p = .48) under Competition.
Relevance analyses. To evaluate Hypothesis 2, complementary correlations of
other trait pairs judged to be relevant (Table 7) were also examined. Positive
relationships existed for the aggressive/aggressive pair (r = .38, p = .08) in the
Competition condition and on non-trustworthy/non-trustworthy pair (r = .52, p < .05) in

44
the Control condition. However, both of these pairs confound similarity with
complementary projection, leaving them as non-diagnostic with respect to the occurrence
of similarity or complementary projection. Under Competition, the more one saw oneself
as self-reliant, the more the outgroup was perceived as unreliable (r = .51, p <. 05).
There were a few instances of complementary projection on irrelevant traits in
the Competition condition (Table 8). Complementary projection occurred on the
touchy/critical pair of traits under Competition (r = .41, p < .05). This relationship may
have appeared because the relevance rating of 2.25 for this pair was rather high. A
significant relationship also emerged for respectful-respectful (r = .49, p < .05), but this
also could have reflected similarity rather than complementary projection.
Complementarity also occurred on respectful/reputable complementary pair (r = .56, p <
.01) under Competition. Examining the means of these traits, it seems that this
relationship was actually driven by the participants’ tendency to rate the outgroup as
disrespectful. Thus, this correlation could to be reflecting the relationship of
disrespectful/disreputable, instead of respectful/reputable. Its relevance to the
experimental context was not captured by the judges who merely rated the relevance of
respectful/reputable and did not rate the opposites.
In the Cooperation condition, significant relationships appeared only on the
submissive/dominant (r = .50, p < .05) and prying/prying (r = .53, p < .05) pairs. These
relationships, however, could have respectively reflected contrast and similarity
projection. In the Control condition, a significant correlation existed on

45
reserved/reserved (r = .56, p < .05), but this correlation could have also reflected
similarity, rather than complementary projection.
The 24 correlations of all the complementary pairs in the Competition condition
were correlated with the judges’ ratings of relevance. This analysis suggested that the
higher correlations occurred on the traits that the judges rated as more relevant (r = .35, p
< .1). A t-test comparing the magnitude of correlations for the traits that were categorized
as either relevant or irrelevant showed a similar pattern. The mean correlation of relevant
pairs in the Competition condition (M = .33, SD = .24) marginally exceeded that of the
irrelevant pairs in the Competition condition (M = .06, SD = .26), t(22) = 1.90, p = .07.
These results, though directionally consistent with Hypothesis 2, may have failed to attain
statistical significance because of the small number of the traits that are being compared.
Complementary correlations of the Cooperation and Control conditions failed to show
any relation with their relevance ratings (r = -.02, p = .94 and r = .04, p = .87
respectively), as did t-tests comparing relevant to non-relevant traits: Cooperation and
Control conditions (t(22) =.59, ns and t(22) = .29, ns respectively).
Similarity and contrast projection. With respect to Hypothesis 3, for the stingy-
greedy pair, similarity projection appeared only in the Control condition, wherein
category membership was not salient (e.g. r
stingy
= .48). Participants in the Cooperation
condition did not project. Similarity correlations for all of the traits can be found in
Table 11. Examining the rest of the traits revealed that instances of similarity projection
were found in all three conditions. As depicted in Table 10, participants in the Control
condition projected similarity to the outgroup on the following traits: stingy, suspicious,

46
reserved, and non-trustworthy. In the Cooperation condition, significant similarity
correlations were found for prying and touchy, whereas marginal similarity correlations
were found for bossy and stingy. Marginally significant similarity correlations were also
found under Competition on respectful, critical, gullible, aggressive, and respectable. As
discussed earlier, however, some of these similarity correlations found under Competition
may reflect complementary projection.
Participants in the Cooperation condition exhibited contrast projection when
judging the outgroup’s submissiveness (r = -.50, p < .05) and fearfulness (r = -.45, p =
.06). Under Competition only marginal evidence of contrast projection was found when
judging outgroup’s dishonesty (r = -.37, p = .08). No reliable instances of contrast
projection were found in the Control condition.
Table 10

Similarity Correlations of Self and Outgroup Traits
________________________________________________________________________

Similarity projection Control (N = 22) Cooperation (N = 18) Competition (N = 24)
________________________________________________________________________

Stingy .48* .41 (p = .10) .21
Greedy .20 .23 .17
Bossy -.27 .42 (p = .08) .19
Incapable .06 -.20 .14
Submissive .16 -.50* .33 (p = .12)
Powerful -.18 -.06 -.05
Trusting .20 -.21 .22
Honest .13 .32 -.05
Self-reliant -.16 -.28 -.13
Unreliable .23 -.11 .22
Secretive .17 .05 -.09
Prying .02 .53* -.12
Suspicious .49* .35 -.05

47

Table 10 Continued
Dishonest .33 (p = .13) .05 -.37 (p = .08)
Unscrupulous -.24 .37 .17
Gullible -.22 -.15 .39 (p = .06)
Anxious .06 .23 .12
Critical -.11 -.08 .37 (p = .08)
Fearful -.08 -.45 (p = .06) .08
Aggressive -.27 -.06 .38 (p = .07)
Respectful -.05 .26 .49*
Respectable .29 -.07 .39 (p = .06)
Touchy -.02 .51* .04
Approving .07 -.26 .50*
Reserved .56** -.31 -.15
Non-trustworthy .52* .22 -.03
Dependent .15 .07 .13
________________________________________________________________________

* Correlation is significant at the .05 level or ** .01 level (2-tailed).

Discussion
This study provides evidence for complementary projection in an intergroup
context wherein stingy and greedy behaviors were made salient. Among Competitive
participants, self rated stinginess was associated with perceptions of the outgroup as
greedy, suggesting that they justified their own stinginess by attributing its cause to the
outgroup. This implies that were there a measure that explicitly asked participants who
were in a competitive relationship with their outgroup to explain their own stinginess,
they would have indicated that it was a response to the outgroup’s greed. Note that, as
indicated previously, Business students were perceived as possessing equal levels of
stinginess and greediness in all conditions. This precludes a demand interpretation in that
the articles that participants read did not induce differential perceptions of the outgroup
as greedy, but not stingy. Thus, the obtained link between perceptions of outgroup
greediness and own stinginess, cannot be viewed as an artifactual consequence of the

48
substantive content of the article. Complementary projection may also have occurred in
the Competition condition on two other pairs that the coders considered relevant to the
experimental setting. Positive relationships existed for aggressive/aggressive pair in the
Competition condition and on non-trustworthy/non-trustworthy pair in the Control
condition. Note that both of these pairs confound similarity with complementary
projection, leaving them non-diagnostic with respect to whether they reflect similarity or
complementary projection. Finally, under Competition, the more one saw oneself as self-
reliant the more the outgroup was perceived as unreliable. The relevance analysis also
showed higher correlations in the Competition condition on complementary trait pairs
that the judges rated as more relevant.
There were a few instances of complementary projection on irrelevant traits in
the Competition condition (Table 8). Complementary projection occurred on the touchy/
critical pair, perhaps because the relevance rating of 2.25 was rather high. A significant
relationship also emerged for respectful/respectful, which can be viewed as reflecting
either similarity or complementary projection. Finally, complementarity also occurred on
respectful/reputable complementary pair under Competition. Upon closer examination, it
seemed that this relationship could have been driven by the participants’ tendency to rate
the outgroup as disrespectful. Thus, this correlation could be viewed instead as reflecting
the relationship of respectful/reputable. The judges may have failed to capture the
relevance of the disrespectful/ disreputable trait pair to the experimental context because
they were merely asked to rate the relevance of respectful/reputable, but not their
opposites.

49
In the Cooperation condition, relationships appeared only on the
submissive/dominant and prying/prying pairs. In the Control condition, a significant
correlation existed on reserved/reserved. These relationships, however, could have
reflected similarity and contrast projections, rather than complementary projection.
As expected, participants used complementary projection in the competitive
intergroup goal structure and mostly on the pairs that were considered relevant. Positive
relationships with the trait and the complement existed only on those traits that
confounded complementary projection with either similarity or contrast projection.
Consistent with Mullen et al. (1992), similarity projection occurred mainly in the control
condition and in the cooperative goal structure. Participants in the Competitive
intergroup goal structure also used similarity projection when judging the outgroup’s
tendency to be respectful and critical. Again, it is possible that this may have been, in
fact complementary projection rather than similarity projection. In other words,
participants may have felt the need to justify their own tendency to be respectful and
critical by attributing these traits to the outgroup. Contrast projection was evident in the
Cooperative goal structure. The more submissive the participants were, the more they
judged the outgroup as dominant. Similarly, the more they were fearful, the more they
judged the outgroup as brave. However, this last correlation was only marginal. In the
Competitive goal structure, the more one considered him or herself as honest, the more
one judged the outgroup as dishonest.
The Cooperative goal structure manipulation described both the ingroup and
the outgroup as equally stingy and greedy. This may be seen as problematic because in

50
the Competitive goal structure only the outgroup was portrayed as stingy and greedy,
whereas the ingroup’s levels of stinginess and greediness were not discussed. Thus, it is
possible that this apparent intergroup similarity in the Cooperative goal structure may
have reduced participants’ inclination to rely on complementary projection. If the
Cooperative goals structure was solely cooperative, without the additional induction of
intergroup similarity on the dimensions of stinginess and greediness, participants might
have used complementary projection even in the Cooperative goal structure. However,
two previous findings indicate that complementary projection would not have occurred in
the Cooperative goal structure, even if the Cooperative goal structure had not induced this
intergroup similarity. First, there were no instances of similarity projection on stingy and
greedy traits in the Cooperative goal structure. Second, according to Experiment 1 and
previous research (e.g., Murray, 1933), complementary projection typically occurs in
threatening or otherwise aversive contexts.

51
Chapter 6: General Discussion
The present study shows evidence for complementary projection in the domain
of traits. Murray (1933) found initial support for complementary projection among four
of the five girls who participated in his study. Feshbach and Singer (1957) induced fear
in their subjects by having them undergo electric shocks while observing a film of the
target person whom they later rated. They concluded that fearful participants used
complementary projection, having rated the target person as more aggressive than did the
participants in the control group. However, participants did not rate their own emotional
state. Hence, it is reasonable to suppose that the electric shocks induced anger,
aggression, or arousal that in turn caused more aggressive perceptions of the target
person, leaving it unclear whether the rating of the target person as aggressive was indeed
an instance of complementary or similarity projection. Despite these concerns about the
quality of their procedures and data, the present study not only supports Murray’s (1933),
Feshbach and Singer’s (1957) and Hornberger’s (1960) interpretations of their outcomes
and is the first to conclusively establish the phenomenon of complementary projection,
but further, extends complementary projection to the domains of intergroup relations and
traits, as opposed to feelings.
Maner et al. (2005) recently found evidence for a process that resembles
complementary projection. They argued that a self-protection motive that arises out of
fear will cause one to perceive more anger in the faces of others. They refer to this
process as functional projection because one’s motivational state (viz., self-protection)
causes one to perceive others as having emotions that are functionally related to their

52
motivational state (viz., anger). Maner et al., however, did not discuss past research on
complementary projection and its distinction from functional projection, leaving it
unclear at this point whether these two types of projection have discriminative construct
validity.
Another contribution of the current work is that, unlike all previous research on
complementary projection and functional projection, it did not directly induce a state of
fear in participants. Instead, Experiment 1showed its effects because dispositional
fearfulness was made salient via an aversive setting that was not designed to and did not
explicitly elicit fear. Specifically, after initial exposure to an insulting provocation,
participants were triggered by an outgroup member’s slightly negative evaluation of an
essay they had written. Although Experiment 1 lacked a control condition in which the
target person was not an outgroup member, the findings is important because it suggests
that outgroup membership per se may cue fear (Stephan & Stephan, 2000).
As expected, complementary projection was found under the trigger condition in
Experiment 1 and under competition in Experiment 2. There were a few exceptions to
that rule, but they may be attributable to the fact that some of the traits that the judges
rated as complementary are confounded with either similarity or contrast projection.
Thus, like Murray (1933) and Feshbach and Singer (1957), we find evidence that
complementary projection occurs in an aversive context, such as when participants are
provoked or threatened by the outgroup. Furthermore, complementary projection occurs
mostly on relevant traits. This is somewhat consistent with Hornberger (1960) who
found complementary projection only in judgments of males, but not females. When

53
fearfulness is made somewhat salient, the display of justifying trait attributions for it onto
others may require that those others possess characteristics that are consonant with the
projected negativity. Attributing aggressiveness to females makes little sense as females
are stereotypically perceived as non-aggressive. Similarly, attributing greediness onto an
outgroup that is clearly not greedy is unlikely. This distinction between relevant and
irrelevant targets is thus related to the distinction between relevant and irrelevant traits.
In other words, if the outgroup is not greedy, greed is then unlikely to be attributed to it
and would therefore be an irrelevant trait in a given context.
Limitations and Future Directions
One of the limitations of this study is that there were only a few participants who
perceived themselves as possessing an aversive trait (i.e., stingy and greedy). Thus,
distributions of the scores were somewhat problematic. Future studies should examine
complementary projection for participants who clearly possess the aversive trait and feel
badly about it. This could involve manipulating the possession of aversive traits. In such
a design, participants will be made to believe that they possess a certain negative trait.
This would extend the present findings and make the context more alike the real world
wherein one is often disparaged by others for one’s possession of aversive traits.
Furthermore, self-traits were assessed via subjects’ self-reports. Relying on peer-reports
and personality tests would also be informative.
A further limitation of Experiment 1 is the absence of a control condition wherein
the source of the trigger was not explicitly an outgroup member. Another potential
limitation is that both the provocation and trigger manipulation only involved

54
competency-based feedback. Future research should examine complementary projection
in TDA contexts that rely on different types of provocation and trigger manipulations.
Experiment 2, did not control for the amount or quality of contact participants had with
the outgroup. Specifically, complementary projection might be weaker for those
participants who have friends in the outgroup.
It is still questionable to what extent complementary projection serves a conscious
self-enhancement function, or instead, is automatic. Do frightened girls feel better about
their fearfulness after having projected its cause onto others? In other words, does the
display of complementary projection solely reflect a cognitive explanatory effort
designed to impose meaning onto a complex world, or instead, is it a mood regulation
process by means of which one reduces aversive internal states? Without knowledge of
its cause, fearfulness is irrational and threatening to the self-concept. Furthermore, the
person who projects can remove the aversive feeling by removing its cause. For
example, the girls in Murray (1933) who thought their fear was caused by the pictures
could have refused to look at the pictures. Thus, one should be highly motivated to seek
the cause of aversive feelings and satisfied when that cause is located. In fact, the
gratification of finding the cause is so potent that it overrides the motivation to be
accurate. As soon as any likely cause is identified, it is adopted as accurate and all other
search stops. In this sense, complementary projection seems automatic and possibly
subconscious.
The self-enhancement that results from complementary projection occurs for both
feelings and traits. One should feel better about possessing an aversive trait if the cause

55
of the trait is known to be externally caused. The thought of one’s stinginess should not
be that threatening when it is known that that stinginess was caused by another person’s
greed. Again, this process is likely to be fairly automatic, as one implicitly knows that
stinginess is an adaptive response to greediness. One may then automatically apply
greediness onto available others. Thus, the sub-consciousness of complementary
projection needs to be further examined, as does the distinction between the cognitive
provision of explanatory meaningfulness versus the self-regulatory alleviation of negative
mood.
Future research should also examine the links between complementary projection
and stereotyping. Cottrell and Neuberg (2005) found that certain types of intergroup
threat are associated with certain types of intergroup emotions. For example, threat to
safety was associated with fear, whereas threat to values and freedoms was associated
with disgust. It is possible that different types of intergroup threats and emotions are
responsible for the formation of different kinds of stereotypes and that the mechanism
responsible for the creation of stereotypes is complementary projection. Allport argued
that the purpose of stereotypes “is to justify (rationalize) our conduct in relation to that
category” (1954/1979, p. 191). As argued previously, complementary projection is one
mechanism of justification that among other things can contribute to the creation of
stereotypes.

56
References
Allport, G. W. (1954/1979). The nature of prejudice. Cambridge, MA: Perseus Books.
Axelrod, R. (1984). The evolution of cooperation. New York, NY: Basic Books.
Bloom, L. J., & Houston, B. K. (1975). An experimental investigation of the
effectiveness of complementary projection in reducing anxiety. Journal of Clinical
Psychology, 31, 525-529.

Brunner, J. S., & Goodman, C. C. (1947). Value and need as organizing factors in
perception. Journal of Abnormal and Social Psychology, 42, 33-44.

Bushman, B. J., Bonacci, A. M., Pedersen, W. C., Vasquez, E. A., & Miller, N. (2005).
Chewing on it can chew you up: Effects of rumination on triggered displaced
aggression. Journal of Personality and Social Psychology, 88, 969-983.

Campbell, D. T., Miller, N., Lubetsky, J., & O’Connell, E.J. (1964). Varieties of
projection in trait attribution. Psychological Monographs: General and Applied, 78,
1-33.

Cottrell, C. A. & Neuberg, S. L. (2005). Different emotional reactions to different
groups: A sociofunctional threat-based approach to “prejudice.” Journal of
Personality and Social Psychology, 88, 770-789.

Cronbach, L. J. (1955). Processes affecting scores on “understanding of others” and
“assumed similarity.” Psychological Bulletin, 52, 177-193.

Denson, T. F., Pedersen, W. C., & Miller, N. (2006). The displaced aggression
questionnaire. Journal of Personality and Social Psychology, 90, 1032-1051.

Esses, V. M., Jackson, L. M., & Armstrong, T. L. (1998). Intergroup competition and
attitudes toward immigrants and immigration: an instrumental model of group
conflict. Journal of Social Issues, 54, 699-724.

Feshbach, S. & Singer, R. (1957). The effects of fear arousal and suppression of fear
upon social perception. Journal of Abnormal Social Psychology, 55, 282-288.

Gaertner, S. L., Dovidio, J. F., Banker, B. S., Houlette, M., Johnson, K. M., & McGlynn,
E. A. (2000). Reducing intergroup conflict: From superordinate goals to
decategorization, recategorization, and mutual differentiation. Group Dynamics:
Theory, Research, and Practice, 4, 98-114.

57

Gross, S. R., Holtz, R., & Miller, N. (1995). Attitude certainty. In R. E. Petty & J. A.
Krosnick (Eds.), Attitude strength: Antecedents and consequences (pp. 215-246).
Mahwah, NJ: Erlbaum.

Gross, S. R., & Miller, N. (1997). The “Golden Section” and bias in perceptions of social
consensus. Personality and Social Psychology Review, 1, 241-271.

Hornberger, R. (1960). The projective effects of fear and sexual arousal on the ratings of
pictures. Journal of Clinical Psychology, 16, 328-331.

Holtz, R. (1997). Length of group membership, assumed similarity, and opinion
certainty: The dividend for veteran members. Journal of Applied Social
Psychology, 27, 539-555.

Holtz, R. & Miller, N. (2001). Intergroup competition, attitudinal projection, and
opinion certainty: capitalilzing on conflict. Group Processes and Intergroup
Relations, 4, 61-73.

Jarymowicz, M. & Bar-Tal, D. (2006). The dominance of fear over hope in the life of
In the life of individuals and collectives. European Journal of Social Psychology, 36,
367-392.

Klein, W. M., & Kunda, Z. (1992). Motivated person perception: Constructing
justifications for desired beliefs. Journal of Experimental Social Psychology, 82,
145-168.

Lindsey, J. J. & Anderson, C. A. (2000). From antecedent conditions to violent actions:
A general affective aggression model. Personality and Social Psychology Bulletin. 5.
533-547.

Lau, R. R., Sears, D. O. & Centers, R. (1979). The ‘positivity bias’ in evaluations of
public figures: Evidence against instrument artifacts. Public Opinion Quarterly, 43,
347-358.

Levine, R., Chein, I., & Murphy, G. (1942). The relation of the intensity of a need to the
amount of perceptual distortion: a preliminary report. Journal of Psychology:
Interdisciplinary and Applied. 13, 283-293.

Mackie, D. M., Devos, T. & Smith, E. R. (2000). Intergroup emotions: Explaining
offensive action tendencies in an intergroup context. Journal of Personality and
Social Psychology, 79, 602-616.

Maner, J. K., Kenrick, D. T., Becker, D. V., Robertson, T. E., Hofer, B., Neuberg, S. L.,
Delton, A. W., Butner, J., & Schaller, M. (2005). Functional projection: How

58

fundamental social motives can bias interpersonal perception. Journal of Personality
and Social Psychology, 88, 63-78.

Markus, H., Smith, J., & Moreland, R. L. (1985). Role of the self concept in the
perception of others. Journal of Personality and Social Psychology, 49, 1494-1512.

McDonald, H. E., & Hirt, E. R. (1997). When expectancy meets desire: Motivational
effects in reconstructive memory. Journal of Personality and Social Psychology, 72,
5-23.

Miller, N., Pedersen, W. C., Earleywine, M., & Pollock, V. E. (2003). A theoretical
model of triggered displaced aggression. Personality and Social Psychology Review,
7, 75-97.

Mullen, B., Dovidio, J. F., Johnson, C., & Copper, C. (1992). In-group – out-group
differences in social projection. Journal of Experimental Social Psychology, 28, 422-
440.

Murray, H. A. (1933). The effects of fear upon estimates of the maliciousness of other
personalities. Journal of Social Psychology, 4, 310-339.

Nowlis, V. (1965). Research with the Mood Adjective Check List. In S. S. Tompkins &
C. E. Izard (Eds.) Affect, cognition, and personality: Empirical studies. Oxford, UK:
Springer.

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

Pedersen, W. C., Bushman, B. J., Vasquez, E. A., & Miller, N. (Unpublished
manuscript). The moderating effect of target attributes on triggered displaced
aggression.

Pedersen, W. C., Gonzales, C., & Miller, N. (2000). The moderating effect of trivial
triggering provocation on displaced aggression. Journal of Personality and Social
Psychology, 78, 913-927.

Rook, K. S., Sears, D. O., Kinder, D. R. & Lau, R. R. (1978). The 'positivity bias' in
evaluations of public figures: Evidence against interpersonal artifacts. Political
Methodology, 5, 469-499.

Sears, D. O. (1983). The person-positivity bias. Journal of Personality and Social
Psychology, 44, 233-250.

59

Sedikides, C., Gaertner, L., & Vevea, J. L. (2005). Pancultural self-enhancement
reloaded: A meta-analytic reply to Heine (2005). Journal of Personality and Social
Psychology, 89, 539-551.
Stephan, W. G., & Stephan, C. W. (2000). An integrated threat theory of prejudice. In S.
Oskamp (Ed.), Reducing prejudice and discrimination: “The Claremont Symposium
on Applied Social Psychology (pp. 23-45). Mahwah, NJ: Lawrence Erlbaum
Associates Publishers.

Tajfel, H., & Turner, J. C. (1979). An integrative theory of intergroup conflict. In S.
Worchel & W. G. Austin (Eds.), The social psychology of intergroup relations (pp.
33-47). Monterey, CA: Brooks-Cole.

Tavris, C. & Aronson, E. (2007). Mistakes were made (but not by me): Why we justify
foolish beliefs, bad decisions, and hurtful acts. Orlando, FL: Harcourt.

Vasquez, E. A., Denson, T. F., Pedersen, W. C., Stenstrom, D. A., & Miller, N. (2005).
The moderating effect of trigger intensity on triggered displaced aggression. Journal
of Experimental Social Psychology, 41, 61- 67.

Wildschut, T., Pinter, B., Vevea, J. L., Insko, C. A., & Schopler, J. (2003). Beyond the
group mind: A quantitative review of the interindividual-intergroup discontinuity
effect. Psychological Bulletin, 129, 698-722.

Yzerbyt, V., Dumont, M., Mathieu, B., Gordijn, E. & Wigboldus, D. (2006). Social
comparison and group-based emotions. In S. Guimond (Ed.), Social comparison and
social psychology: Understanding cognitions, intergroup relations, and culture (pp.
228-246). New York, NY: Cambridge University Press.

60
Appendix
Texts of the Newspaper Articles
Competitive Intergroup Goal Structure
Business school students and College of Letters, Arts and Sciences (LAS)
students interact in a number of different ways. Most importantly, they are working
towards creating a university environment that will benefit all students. Business school
students, especially, have shown their generosity in devoting their time to obtain better
and newer facilities and more computer labs that will benefit both business school
students and LAS students.
Recent economic difficulties, however, have hindered their efforts in obtaining
new facilities and more computer labs. LAS Student Council has proposed a plan to
obtain money that will then be used for creation of new facilities and computer labs.
They suggested that willing LAS and business school students should volunteer a little of
their time by working in various positions offered throughout the University. The
proceeds of their work should go towards a fund that will later be used for the creation of
new facilities and computer labs. Business students however, disagree with this plan.
Representatives of the Global Business Student Association said that they believe that the
students should keep the money that they earn instead of devoting it to the fund.
Opinions of LAS and business school students differ in other ways as well. For
example, business school students believe that the Office of Academic Affairs should
increase the funding and allocate more scholarships to business school students. The
Global Business Student Association proposed that this could be done at the expense of

61
LAS students. Specifically, the Office of Academic Affairs could cut the funds allocated
to LAS students in order to obtain necessary funds. Business school students argue that
such a demand is reasonable considering the fact that business school students after
graduation contribute more to the economic well-being of society compared to LAS
students. LAS Student Council finds their beliefs unfounded and motivated by greed.
They argue that their fields of study benefit the society to the same extent if not more
than business school students. Specifically, LAS students point to their contributions to
the advancement of knowledge and science.

Cooperative Intergroup Goal Structure
Business school students and College of Letters, Arts and Sciences (LAS)
students interact in a number of different ways. Most importantly, they are working
towards creating a university environment that will benefit all students. Business school
students, especially, have shown their generosity in devoting their time to obtain better
and newer facilities and more computer labs that will benefit both business school
students and LAS students.
Recent economic difficulties, however, have hindered their efforts in obtaining
new facilities and more computer labs. Medical students have proposed a plan to obtain
money that will then be used for creation of new facilities and computer labs. They
suggested that willing medical, LAS and business school students should volunteer a
little of their time by working in various positions offered throughout the University. The
proceeds of their work should go towards a fund that will later be used for the creation of

62
new facilities and computer labs. LAS and business students however, disagree with the
plan proposed by the medical students. They believe that the students should keep the
money that they earn instead of devoting it to the fund.
On other matters as well, opinions of LAS and business school students are
similar. For example, business school students believe that the Office of Academic
Affairs should increase the funding and allocate more scholarships to LAS and business
school students. They propose that this could be done at the expense of students of
medicine. Specifically, the Office of Academic Affairs could cut the funds allocated to
medical students in order to obtain necessary funds. Business school students argue that
such a demand is reasonable considering the fact that LAS and business school students
after graduation contribute more to the economic well-being of society compared to
medical students. LAS students agree with such proposals and they do not seem to be
troubled by the apparent greed that motivates the proposals. LAS students argue that
students of medicine have received many benefits throughout the past that puts them at an
unnecessary advantage compared to other student groups.

Control .ewspaper Article
Business school students and College of Letters, Arts and Sciences (LAS)
students interact in a number of different ways. Most importantly, they are working
towards creating a university environment that will benefit all students. Business school
students, especially, have shown their generosity in devoting their time to obtain better

63
and newer facilities and more computer labs that will benefit both business school
students and LAS students.
Recent economic difficulties, however, have hindered their efforts in obtaining
new facilities and more computer labs. Medical students have proposed a plan to obtain
money that will then be used for creation of new facilities and computer labs. They
suggested that willing medical, LAS and business school students should volunteer a
little of their time by working in various positions offered throughout the University. The
proceeds of their work should go towards a fund that will later be used for the creation of
new facilities and computer labs. LAS and business students however, disagree with the
plan proposed by the medical students. They believe that the students should keep the
money that they earn instead of devoting it to the fund.
On other matters as well, opinions of LAS and business school students are
similar. For example, business school students believe that the Office of Academic
Affairs should increase the funding and allocate more scholarships to LAS and business
school students. They propose that this could be done at the expense of students of
medicine. Specifically, the Office of Academic Affairs could cut the funds allocated to
medical students in order to obtain necessary funds. Business school students argue that
such a demand is reasonable considering the fact that LAS and business school students
after graduation contribute more to the economic well-being of society compared to
medical students. LAS students agree with such proposals and they do not seem to be
troubled by the apparent greed that motivates the proposals. LAS students argue that

64
students of medicine have received many benefits throughout the past that puts them at an
unnecessary advantage compared to other student groups.

Abstract (if available)

Abstract Two experiments investigated complementary projection -- attributing the complement (viz., the cause) of one's own trait onto another (Bloom & Houston, 1975). In the Triggered Displaced Aggression context, only among those who were initially provoked and then triggered was their fearfulness associated with a perception of an outgroup partner as aggressive (Experiment 1). Thus, these participants misattributed to their partner a trait (viz., aggressive) that was complementary to their own trait (viz., fearful). Among those who were not triggered, self-ratings and their ratings of their partner were unrelated. Experiment 2 replicated these findings using a different pair of complementary traits (stingy/greedy) and employing ratings of the outgroup category as the dependent variable.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

The angry brain: neural correlates of interpersonal provocation, directed rumination, trait direct aggression, and trait displaced aggression

PDF

Children's triggered displaced aggression

PDF

Justification of selfishness: projection as a mechanism of justification

PDF

Frustration within the triggered displaced aggression paradigm

PDF

The moderating effect of triggering provocation on direct aggression

PDF

Implicit theory ideology about human attributes and aggression

PDF

Spouse aggression, depression, and physical health: a multivariate longitudinal study of midlife couples

PDF

Black sheep or devil protection?: the effects of cooperative and competitive intergroup goal structures on group members reactions to an offensive act when committed by an ingroup or an outgroup ...

PDF

You always say that: physical aggression perpetration, linguistic behavior, and conflict intensity in young adult dating couples

PDF

Mediators and moderators in the link between maternal psychological control and peer victimization for Hong Kong Chinese boys

PDF

Prior exposure to aggression and young adults' negative expectancies about romantic partner discussions

PDF

The effect of massed versus spaced arguments on persuasion

PDF

Family aggression, prosocial friends, and the risk of dating and friend victimization in late adolescence

PDF

Social support, constraints, and protective buffering in prostate cancer patients and their partners

PDF

Measuring narcissistic cognitions in response to relationship threat using articulated thoughts in simulated situations

PDF

Partners of breast cancer survivors: Research participation and understanding of survivors' concerns

PDF

The articulated thoughts of heterosexual male college students in reaction to anti-gay hate speech

PDF

Investigating the effects of personality factors on goal shielding

PDF

Marital conflict and parenting: moderating effects of ethnicity and parent sex

PDF

Evaluating a cultural process model of depression and suicidality in Latino adolescents

Asset Metadata

Creator Spanovic, Marija (author)

Core Title Projection on outgroups: complementary projection in aversive intergroup contexts

School College of Letters, Arts and Sciences

Degree Master of Arts

Degree Program Psychology

Publication Date 12/01/2008

Defense Date 10/27/2008

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

Tag displaced aggression,intergroup relations,OAI-PMH Harvest,projection

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Miller, Norman (committee chair), Margolin, Gayla (committee member), Read, Stephen J. (committee member)

Creator Email mspanovic@gmail.com,spanovic@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-m1845

Unique identifier UC1273519

Identifier etd-Spanovic-2499 (filename),usctheses-m40 (legacy collection record id),usctheses-c127-129325 (legacy record id),usctheses-m1845 (legacy record id)

Legacy Identifier etd-Spanovic-2499.pdf

Dmrecord 129325

Document Type Thesis

Rights Spanovic, Marija

Type texts

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

Repository Name Libraries, University of Southern California

Repository Location Los Angeles, California

Repository Email cisadmin@lib.usc.edu