Narcissism, Sexual Refusal, and Aggression: Testing a Narcissistic Reactance Model of Sexual Coercion

Abstract

Laboratory analog studies investigated the theory that narcissism and reactance contribute to causing rape. In Study 1, narcissism correlated positively with rape-supportive beliefs and negatively with empathy for rape victims. In Study 2, narcissists reported more enjoyment than other men of film depictions that presented consensual, affectionate activity followed by rape (but not in response to either affection or rape alone). In Study 3, narcissists were more punitive than other men toward a female confederate who refused to read a sexually arousing passage aloud to them.

Full text

PERSONALITY PROCESSES AND INDIVIDUAL DIFFERENCES
Narcissism, Sexual Refusal, and Aggression: Testing a Narcissistic
Reactance Model of Sexual Coercion
Brad J. Bushman and Angelica M. Bonacci
Iowa State University
Mirjam van Dijk
Utrecht University
Roy F. Baumeister
Case Western Reserve University
Laboratory analog studies investigated the theory that narcissism and reactance contribute to causing
rape. In Study 1, narcissism correlated positively with rape-supportive beliefs and negatively with
empathy for rape victims. In Study 2, narcissists reported more enjoyment than other men of film
depictions that presented consensual, affectionate activity followed by rape (but not in response to either
affection or rape alone). In Study 3, narcissists were more punitive than other men toward a female
confederate who refused to read a sexually arousing passage aloud to them.
Rape and sexual coercion are widely recognized as a serious
social problem and a source of traumatic suffering for many
unfortunate individuals. The causes of rape, however, remain
poorly understood for a combination of reasons, including ideo-
logical and dogmatic commitments, outdated theoretical frame-
works, widely discrepant definitions, and formidable obstacles
(both ethical and pragmatic) to collecting data or conducting
simulation studies. Felson (2002) has argued convincingly that
prevailing theories about rape are at best weakly supported and at
worst strongly contradicted by the existing evidence. Hence, both
new theorizing and new efforts at collecting data seem urgently
needed.
The purpose of this investigation was to conduct some empirical
tests of a recent theory that sexual coercion might stem from a
combination of narcissistic tendencies and reactance. Specifically,
the core idea is that narcissism constitutes a personality trait that
may foster tendencies toward sexual coercion, especially given the
narcissistic propensity for self-serving interpretations, low empa-
thy toward others, and inflated sense of entitlement. Meanwhile,
some men (especially narcissists) may exhibit reactance when their
sexual desires are rejected, and the reactance may foster an in-
crease in sexual desire, attempts to take what has been denied, and
a willingness to aggress against the person who thwarted them—
responses that in concert may contribute to sexual coercion.
The narcissistic reactance theory of rape and sexual coercion
was put forth by Baumeister, Catanese, and Wallace (2002). How-
ever, that work was limited to theoretical analyses and interpretive
review of existing findings, few of which provided direct, unam-
biguous evidence that narcissism plays a role in the mistreatment
of women in sexual or romantic contexts. Baumeister et al. (2002)
concluded that although the narcissistic reactance theory provided
a good fit to existing evidence, conducting some direct empirical
tests that include a measure of narcissism was a crucial next step.
Definitions of rape and sexual coercion have been controversial,
especially in light of efforts by some theorists to deny that rape is
sexually motivated. Further complications have been introduced
by the legal system’s efforts at precise definitions, because even
the criterion of force or coercion is not uniformly applied. (Thus,
most states define consensual sex as rape if one person is under the
age of 18 and the other is 18 or older.) Yet another complication
is that perpetrator and victim may have radically different experi-
ences as to whether the act was sexual. Rather than seeking to
resolve all these definition problems, we acknowledge that our
focus is on the psychology of male perpetrators. Rape and sexual
coercion consist of using aggressive force to make a woman
Brad J. Bushman and Angelica M. Bonacci, Department of Psychology,
Iowa State University; Mirjam van Dijk, Department of Psychology,
Utrecht University, Utrecht, the Netherlands; Roy F. Baumeister, Depart-
ment of Psychology, Case Western Reserve University.
Roy F. Baumeister is now at the Department of Psychology, Florida
State University.
This work was supported by Templeton Foundation Grant 5039. Roy F.
Baumeister gratefully acknowledges the support of the Center for Ad-
vanced Study of the Behavioral Sciences, where he held the Fritz Redlich
Fellowship while working on this article. We thank Kathleen Vohs for her
comments and assistance.
Correspondence concerning this article should be addressed to Brad J.
Bushman, who is now at the Research Center for Group Dynamics,
Institute for Social Research, University of Michigan, 426 Thompson
Street, Ann Arbor, Michigan 48106. E-mail: bbushman@umich.edu
Journal of Personality and Social Psychology, 2003, Vol. 84, No. 5, 1027–1040
Copyright 2003 by the American Psychological Association, Inc. 0022-3514/03/$12.00 DOI: 10.1037/0022-3514.84.5.1027
1027

engage in sexual activity that the man desires but she actively
refuses. In this definition, the man is not necessarily seeking to
harm the woman, but he may be willing to harm her in order to get
his way. Moreover, as we shall explain, her continued refusal of
his wishes may eventually cause him to act in aggressive, punitive
ways toward her, in which case he would be seeking to hurt her,
but we assume that this is a frustrated last resort and he would
prefer having sex with her rather than hurting her.
Narcissism and Aggression
Narcissism is a personality trait defined by an unusually high
degree of self-love, as exemplified by the character from Greek
mythology who fell in love with his own reflection in the water.
Research has indicated the need for some refinements of the Greek
myth, however. In particular, the mythological character was so
wrapped up in himself that he was indifferent to the attentions and
affections of others, whereas empirical studies indicate that mod-
ern narcissists are preoccupied if not downright obsessed with
garnering the admiration of others (see Morf & Rhodewalt, 2001,
for review). The term narcissism is linguistically related to the
word narcotic, implying perhaps that people sometimes become
addicted to loving themselves (see Baumeister & Vohs, 2001).
According to the fourth edition of the Diagnostic and Statistical
Manual of Mental Disorders (American Psychiatric Association,
1994), narcissism is characterized by an exaggerated sense of
self-importance and uniqueness, an unreasonable sense of entitle-
ment, a craving for admiration, exploitative tendencies toward
others, deficient empathy, and arrogance. Narcissists are strongly
motivated to sustain their own and others’ perception of them as
superior beings.
Narcissism has been associated with aggression in empirical
studies. Bushman and Baumeister (1998) found that identical
remarks of insulting criticism elicited more severe and aggressive
retaliation from high narcissists than from other participants. Nar-
cissists provide some of the best evidence that threatened egotism
is an important cause of aggression (Baumeister, Smart, & Boden,
1996).
There are multiple reasons for predicting that narcissists would
be more likely than other men to engage in sexual coercion, in
addition to their propensity for aggressive retaliation (see
Baumeister et al., 2002). First, their inflated sense of entitlement
may make them think that women owe them sexual favors. Sec-
ond, their low empathy entails that they would not be deterred by
concern over the victim’s suffering. Ironically, narcissists are
capable of empathy but simply do not bother to use it when it is not
in their interests to do so. Third, their tendency to maintain inflated
views of self by means of cognitive distortions might help them
rationalize away any borderline objectionable behaviors, such as if
they could convince themselves that their coercion victims had
really desired the sex or had expressed some form of consent. Last,
their concern with getting others to admire them could lead them
to seek out sexual conquests in order to have something to boast
about to their peers, and in fact, studies of coercive men have
suggested that peer pressure and boasting are sometimes important
contributing factors (Kanin, 1985; Lisak & Roth, 1988).
Reactance and Sexual Refusals
Reactance is defined as negative responses to loss of freedom
(or threats of loss). When people lose a desired option, they
respond by increasing their desire for that option, by trying to do
what is now forbidden, or by aggressing against the person who
deprived them of the option (J. W. Brehm, 1966, 1972; S. S.
Brehm & Brehm, 1981; Wicklund, 1974).
Although some theorists such as Brownmiller (1975) argued
forcefully that rape has nothing to do with sex or sexual motiva-
tions, the weight of evidence has suggested that sexual motivations
are prominent factors from the rapist’s perspective, even though to
be sure rape is not a sexual experience for most victims (Muehlen-
hard, Danoff-Burg, & Powch, 1996). Felson (2002) and Palmer
(1988) have revealed both conceptual and empirical fallacies in the
argument that rape is not sex. The present investigation assumed
that sexual motivations (in the male aggressor) play a role in
leading to rape and coercion.
If one assumes that sex is a factor, then reactance can readily
come into play. A man may desire sex with a particular woman,
but she may refuse his advances. The potential for such conflict is
inherent in many heterosexual encounters, insofar as men gener-
ally desire sex earlier in the relationship, with more possible
partners, with less commitment, and otherwise more often than
women (see Baumeister, Catanese & Vohs, 2001, for review). The
woman’s refusal may lead to reactance, especially if the man had
anticipated sex with her. All three of the main consequences of
reactance (i.e., increased desire, attempt to exercise the forbidden
option, and aggression toward the source of the prohibition) would
contribute to male aggression toward a woman who has refused his
sexual advances.
We reiterate that we do not see this line of work as justifying
rape or reducing the culpability of the offender. Sexual coercion is
always immoral and odious. Our effort is simply to understand
how otherwise seemingly normal and decent men might engage in
this disgraceful behavior.
The reactance and narcissism components of the theory may
seem independent, but there are several overlaps. Narcissists have
an inflated sense of entitlement, so they should be more prone to
reactance, because they are more likely than others to believe they
deserve things that they are not getting. Moreover, empirical
studies have shown that reactance and narcissism are positively
correlated (e.g., Frank et al., 1998; Joubert, 1992), such that
narcissists have more reactance than others.
Present Investigation
The present investigation used three different methods to test the
prediction that narcissism would constitute a risk factor for sexual
coercion. Study 1 used questionnaire measures to establish
whether narcissists would exhibit attitudes that have been linked to
sexual coercion, specifically endorsement of rape myths and low
empathy toward rape victims. Study 2 assessed enjoyment of a
videotape depiction of a rape episode, as a function of narcissism
and contextual factors (including an ambiguous suggestion of
victim encouragement). Study 3 investigated whether male partic-
ipants reacted punitively to a female confederate who did or did
not refuse to provide sexual stimulation that had been anticipated.
Although recent findings indicate that both males and females
engage in sexually coercive behaviors (e.g., Anderson &
1028
BUSHMAN, BONACCI, VAN DIJK, AND BAUMEISTER

Struckman-Johnson, 1998), coercion of women by men is gener-
ally regarded as the more severe social and criminal problem, and
female victims of male coercion are much more likely to report
enduring distress and trauma than are male victims of female
coercion. Therefore, our investigation focuses exclusively on male
perpetrators and female victims, even though we do not intend to
minimize or condone victimization of males by females or same-
gender sexual coercion. Also, the present investigation relied on
laboratory and experimental evidence, which may have advantages
in terms of precise control but also possible limitations in terms of
generalizability, especially given the ethical and pragmatic diffi-
culty of simulating sexual coercion in the laboratory.
Study 1
Study 1 was a preliminary investigation designed to see whether
narcissism might correlate with attitudes that have been linked to
rape. We measured narcissism using the standard, widely used
instrument developed by Raskin and Terry (1988). This study
seemed a necessary first step in view of the relative scarcity of data
linking narcissism to sexual coercion.
Two measures of rape-relevant attitudes were administered. The
first was the Rape Empathy Scale (Deitz, Blackwell, Daley, &
Bently, 1982). This measure was designed to distinguish individ-
uals who are sensitive and sympathetic to the plight of rape victims
from those who would be more prone to blame the victim and
exonerate the perpetrators. Narcissists tend to blame others rather
than themselves for conflicts and problems (e.g., Patrick, 1990;
Sankowsky, 1995). If narcissism is indeed a risk factor for sexual
coercion, we reasoned that narcissistic males would show less
empathy for rape victims and would instead be inclined to shift
responsibility away from the perpetrators onto the victims.
The other measure we used was the Rape Myth Acceptance
Scale (Burt, 1980). It consists of attitudes that could be used to
rationalize sexual coercion, such as that ambiguous female behav-
iors constitute sexual encouragement that justify persistent sexual
advances by males and that sexual coercion can be justified under
some circumstances. As reviewed by Felson (2002), some studies
have found that men who score high on this scale are more likely
to engage in sexual coercion, although other studies have failed to
find a link. In any case, such self-serving interpretations and
rationalizations seemed potentially consistent with the narcissistic
version of sexual aggression, and so we predicted that narcissists
would score higher than other men on acceptance of rape myths.
Method
Participants
Participants were 403 college men enrolled in introductory psychology
courses. They received extra course credit in exchange for their voluntary
participation. Their mean age was 20.5 years (SD ⫽ 2.5 years). None of the
participants in Study 1 participated in Studies 2 or 3.
Individual Difference Measures
Narcissism. Narcissism was measured by the Narcissistic Personality
Inventory (Raskin & Terry, 1988). For each of the 40 forced-choice dyads
on the scale, participants choose either the narcissistic response (e.g., “If I
ruled the world it would be a better place.”) or the nonnarcissistic response
(e.g., “The thought of ruling the world frightens the hell out of me.”).
The 40 items are summed together. High scores indicate high narcissism.
The Narcissistic Personality Inventory has good psychometric properties.
The alpha coefficient for the full scale is .83 (Raskin & Terry, 1988).
Rape empathy. Rape empathy was measured by the Rape Empathy
Scale (Deitz et al., 1982). For each of the 20 forced-choice dyads on the
scale, participants choose either an empathic response toward rape victims
(e.g., “In general, I feel that rape is an act that is not provoked by the rape
victim.”) or a nonempathic response toward rape victims (e.g., “In general,
I feel that rape is an act that is provoked by the rape victim.”). The 19 items
are summed together.
1
High scores indicate high empathy toward rape
victims. The Rape Empathy Scale has good psychometric properties.
Item-total correlations for the 19 items range from .18 to .75. The 19-item
scale is also internally consistent, with alpha coefficients ranging from .82
to .89 (Deitz et al., 1982).
Rape myth beliefs. Belief in rape myths was assessed by the Rape
Myth Acceptance Scale (Burt, 1980). This scale measures the extent to
which respondents agree with views that have been deemed as conducive
to rape, such as that when women say no to sex, they really mean yes.
Sample items include “If a girl engages in necking or petting and she lets
things get out of hand, it is her own fault if her partner forces sex on her”
and “A woman who goes to the home or apartment of a man on their first
date implies that she is willing to have sex.” Responses to the 19 items on
the scale are summed together. High scores indicate high belief in these
rape-conducive views. The Rape Myth Acceptance Scale has good psy-
chometric properties. The item-total correlations for the 19 items range
from .27 to .61. The alpha coefficient for the full scale is .88 (Burt, 1980).
Procedure
Participants completed the Narcissistic Personality Inventory (Raskin &
Terry, 1988), the Rape Empathy Scale (Deitz et al., 1982), and the Rape
Myth Acceptance Scale (Burt, 1980) as part of a battery of questionnaires
given in mass-testing sessions.
Results
The alpha coefficients for the Narcissistic Personality Inventory
(Raskin & Terry, 1988), the Rape Empathy Scale (Deitz et al.,
1982), and the Rape Myth Acceptance Scale (Burt, 1980) were .81,
.93, and .82, respectively. All three scales were internally consis-
tent. The average scores for the scales were 17.92
(SD ⫽ 6.49), 5.66 (SD ⫽ 5.64), and 47.42 (SD ⫽ 11.85),
respectively.
As expected, narcissism was negatively correlated with empathy
toward rape victims and was positively correlated with belief in
rape myths (rs ⫽⫺.13 and .11, ps ⬍ .05). Empathy toward rape
victims was also negatively correlated with belief in rape myths
(r ⫽⫺.32, p ⬍ .0001).
Discussion
Study 1 was essentially a pilot study designed to show that
narcissism could indeed predict some of the undesirable traits that
1
In the standard version of the scale, after selecting the response they
prefer most, participants rate how strongly they prefer the statement they
chose over the other statement with the response using a scale ranging
from 1 (no preference)to7(strong preference). However, the alpha
coefficient for the scale scored this way was only .58. Thus, we decided to
simply sum the number of rape empathic responses instead, which pro-
duced an alpha coefficient of .93.
1029
NARCISSISM AND SEXUAL COERCION

have been put forward as conducive to sexual coercion. The
findings were consistent with the view that sexually coercive men
may have narcissistic tendencies. Narcissism was linked to low
empathy toward rape victims, suggesting that a narcissistic male
who was tempted to coerce a woman would not likely be deterred
by feelings of compassion or sympathy, or indeed even under-
standing, of how unpleasant the victimization might be for her. In
a similar fashion, narcissists were more prone than other males to
express beliefs in the so-called rape myths. These myths tend to
blame the rape victim for her victimization and suggest that
perpetrators of sexual coercion were likely misled or encouraged
by the victim’s actions.
Study 1 was hardly a full test of the narcissistic reactance theory
of rape. It does however lend some plausibility to it. Narcissists did
exhibit a pattern of attitudes that could be conducive to sexual
coercion. Although significant, the results were weak. The weak-
ness is perhaps understandable given that the measures involved
general attitudes rather than specific behaviors and that their
relationship to actual sexual coercion is itself rather weak. Alter-
natively, the small size of the effect could be an indication that
narcissism is only weakly (if at all) related to sexual coercion. Still,
the narcissistic reactance theory could have been falsified if there
had been no relationship or if narcissism had correlated in the
opposite direction with the rape-relevant attitude scales, and so in
that sense the theory did survive a preliminary test. In any case, it
was necessary to devise more rigorous empirical tests.
Study 2
Study 2 examined reactions to a film depiction of rape. Past
work has found that sexually coercive men respond more favor-
ably than other men to videotape and audiotape depictions of rape
(e.g., Bernat, Calhoun, Adams, 1999; Hall, Shondrick, & Hirsch-
man, 1993; Malamuth, 1989). We reasoned that if narcissism is a
risk factor for rape, then narcissistic men would enjoy a rape film
more than would other men.
As argued by Baumeister et al. (2002), the evidence does not
really justify the conclusion that sexually coercive men actually
prefer depictions of coercive sex. If anything, most of them show
a slight preference for depictions of consensual sex. Their enjoy-
ment of consensual sex is comparable with that of other men.
Thus, the difference is best characterized by saying that sexually
coercive men enjoy depictions of sex regardless of how the woman
is responding, whereas noncoercive men are strongly put off by
depictions of forcible sex. The depiction of force or coercion spoils
any enjoyment for most men but does not seem to bother sexually
coercive men. Possibly they have means of rationalizing or ignor-
ing any unpleasant aspect of using force to obtain sex. A meta-
analysis by Hall et al. (1993) confirmed that only studies with
small samples of highly deviant or pathological (e.g., incarcerated)
sex offenders showed even a trend toward preferring rape depic-
tions over depictions of consensual sex. Larger and less patholog-
ical samples generally showed that even sexually aggressive men
preferred consensual sex depictions over rape.
The present study was therefore set up with different versions of
the rape film. These varied as to whether they offered the viewer
some way to rationalize or downplay the coercive aspect. Some
viewers saw only the rape scene. Others saw the rape scene after
they watched the male and female actors engage in mutually
consensual physical activities including dancing, hugging, and
kissing. The latter version was particularly relevant to many date
rapes (which form the vast majority of incidents of sexual coer-
cion; e.g., Laumann, Gagnon, Michael, & Michaels, 1994). For
example, in Kanin’s (1985) study of self-confessed date rapists,
every single instance of date rape was preceded by mutually
consensual activity, usually oral sex or necking. Converging evi-
dence from victim reports was provided by O’Sullivan, Byers, and
Finkelman (1998), who found that the majority of rape victims
reported that they had had consensual intercourse with the rapist
on a previous occasion.
We predicted that the scenes involving consenting activity prior
to the rape would be especially acceptable, even appealing, to the
narcissistic men for two reasons. First, as already stated, our main
hypothesis was that narcissism is a risk factor for rape, and so
narcissists should respond like sexually coercive men—which
would include seeing consensual affection as possibly justifying
the use of coercion later. Second, one of the defining features of
narcissism is an inflated sense of entitlement, and so they should
be more prone than other men to think that a man is entitled to sex
if a woman has encouraged and aroused him through kissing,
hugging, and other physical contact.
If the predictions were confirmed, and narcissists showed
greater enjoyment of rape when combined with affection (but not
otherwise), an alternative explanation might be suggested. Specif-
ically, it would be possible that the scene of affectionate, consen-
sual sex appealed to narcissists more than to other men, indepen-
dent of the rape. To test this possibility, we included a third
condition, in which participants saw only the scene of consensual
affection, without the rape. The narcissism theory of rape would
predict that narcissists would show significantly higher enjoyment
than other men only when they viewed both the consensual affec-
tion and the rape, insofar as narcissists would be more prone than
other men to see the consensual affection as justifying the use of
force later.
The attempt to simulate sexual coercion by having participants
watch a film rests on the assumption that viewers identify with the
perpetrator in the film, at least to some extent. Rather than leave
this assumption implicit, we both manipulated and measured it.
That is, we asked all participants to report the extent to which they
identified with the male actor and with the female actor. We
predicted that narcissistic men would find it more difficult than
other men to identify with the female victim, consistent with the
view that they are not empathic when it is not in their own
interests. Furthermore, participants were explicitly instructed to
identify with either the man or the woman in the film. Enjoyment
of the film should be facilitated by identifying with the male
character (especially among narcissistic men) and inhibited by
identifying with the female character.
The manipulation of film content is also relevant to the role of
reactance. When a woman dances with a man and then hugs and
kisses him (in the consensual affection scenes), she may give rise
to hopes or expectations that there will be sexual activity. Al-
though she has every moral and legal right to refuse any further
sexual activity, some men may respond negatively to her refusal,
especially if they believe that she has given him legitimate reason
to expect sex. The removal of an expected benefit is an important
cause of reactance, which could cause him to use aggression in the
attempt to reclaim the option that he feels has been unfairly taken
1030
BUSHMAN, BONACCI, VAN DIJK, AND BAUMEISTER

away from him (J. W. Brehm, 1966). Viewers who identify with
the male character in such a scene may sympathize with his
disappointment, and some of them may be less inclined to con-
demn him for coercing her.
Method
Participants
Participants were 300 men enrolled in introductory psychology courses
at Iowa State University. Participants received extra course credit in
exchange for their voluntary participation. None of the participants in
Study 2 participated in Studies 1 or 3.
Participants were randomly selected from a larger group of participants
who had completed the Narcissistic Personality Inventory (Raskin & Terry,
1988) as part of a battery of tests administered in mass-testing sessions.
The coefficient alpha for the scale was .84, and the average narcissism
score was 16.31 (SD ⫽ 6.86).
Videotapes
We used film segments from two different commercially available,
R-rated films—The Accused (Kaplan, 1988) and Higher Learning (Single-
ton, 1995)—to increase the generalizability of the results (Wells & Wind-
schitl, 1999). Both segments showed a man raping a woman, who ex-
pressed pain and disgust during the rape. Both segments also showed
mutually consenting affectionate activity between the man and the woman
prior to the rape.
Three different versions of the segments were recorded for each film.
One version showed only the rape scene. A second version showed the
woman engaging in mutually consenting affectionate activities with the
man prior to the rape scene (e.g., kissing, dancing, and hugging). A third
version showed only the mutually consenting affectionate activities. Each
participant watched only one version of one film.
Procedure
Participants were scheduled by telephone. They were told that the study
was being conducted to evaluate clips from commercially available,
R-rated films. They were told that the film clips might contain violence,
sex, or both. The experimenter who scheduled participants and the exper-
imenter who conducted the experiment were both blind to participants’
narcissism scores.
Participants were randomly assigned to watch one of the three versions
of the film (i.e., rape only, rape and affection, affection only). Within each
film version, half of the participants were told to identify with the male
actor in the film, whereas the other half were told to identify with the
female actor.
Immediately after viewing the film segment, participants rated how
enjoyable, entertaining, sexually arousing, and violent they thought the
film was. They also rated how much they identified with the male and
female actors in the film segment. All ratings were made on a 10-point
scale ranging from 1 (strongly disagree)to10(strongly agree). Partici-
pants also indicated whether they had seen the film from which the
segment had been taken. As a manipulation check, participants were asked
to indicate which actor the experimenter told them to identify with (i.e.,
male actor, female actor, don’t remember). Finally, participants were fully
debriefed.
Results
Preliminary Analyses
Identification manipulation check. Over 95% of participants
correctly recalled which actor the experimenter told them to iden-
tify with.
Film differences. We used two different films (i.e., The Ac-
cused, Higher Learning) to increase the generalizability of the
results (Wells & Windschitl, 1999). No significant differences
were found between the two films on any of the rating dimensions
(i.e., enjoyable, entertaining, sexually arousing, violent). The
random-effects variance estimates for film exemplar were also
quite small, ranging from 0.00 to 0.03 (M ⫽ 0.02). None of the
maximum likelihood random-effects variance estimates signifi-
cantly differed from zero. Thus, the data were collapsed across
film exemplar for subsequent analyses.
Analysis Strategy
The dependent measures were analyzed by means of 3 (film
version: rape only, rape and affection, affection only) ⫻ 2 (identify
with male vs. female character) ⫻ 2 (high vs. low narcissists)
analyses of covariance. Whether the participants had seen the film
from which the segment was taken was used as a covariate in the
analyses (19% of those shown The Accused had seen it before;
19% of those shown High Learning had seen it before). The means
were therefore adjusted to reflect whether the participant had seen
the movie before. A median split was used to identify high versus
low narcissists. Regression analyses, treating narcissism as a con-
tinuous variable, yielded the same pattern of results.
We expected narcissists and other men to differ in their ratings
of the film showing affectionate activity (e.g., dancing, hugging,
kissing), followed by the rape. We did not expect narcissists and
other men to differ in their ratings of the film showing only
affectionate activity or in their ratings of the film showing only the
rape. To test this hypothesis, we performed simple effects analyses
between narcissists and other men for each version of the film.
Enjoyment of Film
Participants rated how much they enjoyed the film. Main effects
were obtained for film version, F(2, 287) ⫽ 40.26, p ⬍ .0001, and
narcissism, F(1, 287) ⫽ 6.67, p ⬍ .05. That is, viewers generally
liked the nonrape versions more than the rape versions and liked
the affectionate versions more than the nonaffectionate versions.
Plus, narcissists enjoyed the films in general better than
nonnarcissists.
We performed three simple effects tests to determine whether
narcissists and nonnarcissists differed in how much they enjoyed
each version of the film. As Figure 1 shows, narcissists enjoyed the
film more than other men when it showed both consensual affec-
tion and rape, t(287) ⫽ 3.04, p ⬍ .005, d ⫽ 0.36. Narcissists did
not differ from others in how enjoyable they found the film to be
when it depicted only the rape, t(287) ⫽ 0.73, ns. Hardly any of the
men enjoyed this version of the film. Furthermore, and crucially,
narcissists did not differ from other men in how enjoyable they
found the film to be when it depicted only the mutually consenting
affectionate activities between the man and woman (i.e., dancing,
hugging, kissing), t(287) ⫽ 0.73, ns. Thus, as predicted, narcissism
only enhanced enjoyment of the film that contained both consen-
sual affection and rape.
As expected, participants who were asked to identify with the
male actor enjoyed the film more than did those who were asked
to identify with the female actor (Ms ⫽ 2.43 and 2.12, respective-
ly), F(1, 287) ⫽ 5.78, p ⬍ .02, d ⫽ 0.28. Men who had not seen
1031
NARCISSISM AND SEXUAL COERCION

the film before enjoyed it more than did those who had seen it
before (Ms ⫽ 2.42 and 1.70, respectively), F(1, 287) ⫽ 16.47, p ⬍
.0001, d ⫽ 0.48. No other effects were significant.
Entertainment Ratings
Participants also rated how entertaining the film was. Main
effects were obtained for film version, F(2, 287) ⫽ 26.89, p ⬍
.0001, and narcissism, F(1, 287) ⫽ 10.09, p ⬍ .005. As with
enjoyment, entertainment was higher in response to affection and
lower in response to rape, and narcissists reported higher enter-
tainment than other men overall.
We performed three simple effects tests to determine whether
narcissists and nonnarcissists differed in how entertaining they
thought each version of the film was. As Figure 2 shows, narcis-
sists thought the film was more entertaining than others did when
it showed consensual activity between the man and the woman
prior to the rape, t(287) ⫽ 3.05, p ⬍ .005, d ⫽ 0.36. Narcissists did
not differ from others in how entertaining they found the film to be
when it depicted only the rape, t(287) ⫽ 1.39, ns, nor when it
depicted only the mutually consenting affectionate activities be-
tween the man and woman, t(287) ⫽ 1.09, ns. Men who had not
seen the film previously thought it was more entertaining than did
those who had seen it before (Ms ⫽ 2.53 and 1.91, respectively),
F(1, 287) ⫽ 10.81, p ⬍ .0001, d ⫽ 0.39. No other effects were
significant.
Sexual Arousing Ratings
Participants also rated how sexually arousing the film was. Main
effects were obtained for film version, F(2, 287) ⫽ 4.77, p ⬍ .01,
and narcissism, F(1, 287) ⫽ 4.20, p ⬍ .05. As with enjoyment and
entertainment, arousal was higher in response to affection and
lower in response to rape, and narcissists reported higher arousal
than other men overall.
We performed three simple effects tests to determine whether
narcissists and nonnarcissists differed in how sexually arousing
each version of the film was. As Figure 3 shows, narcissists
thought the film was more sexually arousing than others did when
it showed consensual, affectionate activity followed by the rape,
t(287) ⫽ 2.32, p ⬍ .05, d ⫽ 0.27. Narcissists did not differ from
others in how sexually arousing they found the film to be when it
depicted only the rape, t(287) ⫽ 0.73, ns, nor when it depicted only
the mutually consenting affectionate activities between the man
and woman, t(287) ⫽ 0.52, ns. Men who had not seen the film
previously thought it was more sexually arousing than did those
who had seen it before (Ms ⫽ 2.63 and 2.19, respectively), F(1,
287) ⫽ 4.95, p ⬍ .05, d ⫽ 0.26. No other effects were significant.
Violence Ratings
Participants also rated how violent the film was. A main effect
was obtained for film version, F(2, 287) ⫽ 225.47, p ⬍ .0001. Not
surprisingly, the film versions that depicted the rape were judged
to be more violent than the version depicting only consenting
activity. The three simple effects comparing narcissists and other
Figure 1. Interaction between narcissism and entitlement on film enjoy-
ment ratings.
Figure 2. Interaction between narcissism and entitlement on film enter-
taining ratings.
Figure 3. Interaction between narcissism and entitlement on film sexu-
ally arousing ratings.
1032
BUSHMAN, BONACCI, VAN DIJK, AND BAUMEISTER

men were nonsignificant for all three film versions,
ts(287) ⫽ 1.21, 0.33, and 0.52 for the rape only, rape and affection,
and affection only film versions, respectively.
There was a significant interaction between narcissism and
which actor participants were told to identify with on violence
ratings, F(1, 287) ⫽ 4.52, p ⬍ .05. This two-way interaction,
however, depended on the version of the film that participants
watched, as indicated by a nearly significant three-way interaction,
F(2, 287) ⫽ 2.33, p ⬍ .10. The mutually consenting activity only,
mutually consenting activity followed by rape, and rape only
versions of the film are depicted in Figures 4A, 4B, and 4C.
For the film that depicted only consensual affection (and no
rape), there was no difference in how violent narcissists and
nonnarcissists judged the film to be regardless of whether they
were asked to identify with female or male actor, ts(287) ⫽ 0.01
and 0.73, respectively, ns (see Figure 4A). The mutually consent-
ing activity tape was judged as nonviolent by all participants, and
identification made no difference.
Regarding the film that depicted both consensual, affectionate
activity and rape, identification did seem to matter. When asked to
identify with the female actor, narcissists judged this film to be less
violent than did others, although this difference was not quite
significant, t(287) ⫽ 1.88, p ⬍ .10, d ⫽ 0.22 (see Figure 4B).
When asked to identify with the male actor, narcissists and non-
narcissists did not differ in how violent they judged that same film
(consensual affection plus rape) to be, t(287) ⫽⫺1.44, ns (see
Figure 4B).
Identification also mattered when participants saw only the rape
segment. When asked to identify with the female actor, narcissists
judged the rape to be less violent than did others, t(287) ⫽ 2.19,
p ⬍ .05, d ⫽ 0.26 (see Figure 4C). When asked to identify with the
male actor, narcissists and nonnarcissists did not differ in how
violent they judged the rape to be, t(287) ⫽⫺0.42, ns (see Figure
4C). No other effects were significant.
Identification With Male and Female Actors
Regardless of whom they were told to identify with, narcissists
identified more with the male actor than did nonnarcissists
(Ms ⫽ 2.62 and 2.16, respectively), F(1, 287) ⫽ 5.50, p ⬍ .05,
d ⫽ 0.28. Thus, apparently narcissism promoted a tendency to see
the film actor as similar to oneself.
Not surprisingly, participants identified more with the male
actor if they were told to identify with him than if they were told
to identify with the female actor (Ms ⫽ 2.85 and 1.94, respective-
ly), F(1, 287) ⫽ 22.09, p ⬍ .0001, d ⫽ 0.55. This finding indicates
that the identification manipulation was effective. There also was
a significant effect for film version, F(2, 287) ⫽ 12.00, p ⬍ .0001.
Participants were more prone to identify with the male actor in the
film version depicting only consenting activity than in the film
versions depicting rape. No other effects were significant on the
measure of identification with the male actor.
Figure 4. (A) Interaction between narcissism and which actor partici-
pants were told to identify with on violence ratings for the film depicting
only consenting activity. (B) Interaction between narcissism and which
actor participants were told to identify with on violence ratings for the film
depicting consenting activity prior to the rape. (C) Interaction between
narcissism and which actor participants were told to identify with on
violence ratings for the film depicting only the rape.
1033
NARCISSISM AND SEXUAL COERCION

Likewise, participants identified more with the female actor if
they were told to identify with her than if they were told to identify
with the male actor (Ms ⫽ 3.68 and 2.51, respectively), F(1,
287) ⫽ 24.43, p ⬍ .0005, d ⫽ 0.58. This finding indicates that
identification manipulation was effective. No other effects were
significant.
Discussion
The findings of Experiment 2 provided further evidence that
sexual coercion may be more acceptable to narcissistic men than to
other men. Film depictions of rape were rated as more enjoyable,
entertaining, and sexually arousing by narcissists than by other men.
These results do not indicate that narcissists actually like or
enjoy films of rape. Among participants who saw only the rape
scene, narcissists did not show any more enjoyment than nonnar-
cissists. Indeed, our entire sample gave low ratings to the rape-only
film. Nor was there a significant difference in enjoyment when
participants saw only the consensual, affectionate activity, which
rules out the alternative explanation that narcissists simply respond
more favorably than other men to depictions of consensual sexual
activity.
The difference was mainly found when participants saw the rape
occurring after some depiction of mutually consensual, affection-
ate activity. Narcissists gave more positive ratings than other men
to that film. The most likely explanation appears to be that nar-
cissists and nonnarcissists enjoy watching sexually oriented affec-
tionate activity, but when this is followed by coercive sex, the
nonnarcissistic male’s pleasure is spoiled—whereas the narcis-
sist’s enjoyment remains significant. Coercive sexuality, when
preceded by consensual affection, is thus still positive and enjoy-
able to the narcissist.
Additional findings supported the hypothesis that low empathy
toward rape victims may mediate the responses of narcissists. To
rape victims, rape is essentially an act of violence, even though to
perpetrators it may be primarily a sexual act (e.g., Baumeister &
Tice, 2000; Brownmiller, 1975; Felson, 2002). The narcissists
seemed less able than other men to see the act from the woman’s
perspective. First, they reported identifying more with the male
character, regardless of whom they were instructed to identify
with. Second, when participants were instructed to identify with
the female character (which should have facilitated perception of
the rape as an act of violence), narcissists gave the scene lower
ratings on violence than other men.
The results are also consistent with a reactance analysis. Reac-
tance was most relevant to the version in which the couple first
was depicted engaging in romantic, affectionate activity, which
might plausibly lead the man to expect sex (and therefore feel
reactance when she refused). The film was rated as more enjoyable
and more entertaining in that condition, and narcissists in partic-
ular seemed to enjoy it. They did not respond positively to rape
depicted without the initial, consensual activity.
In sum, these results provide some support for the narcissistic
reactance theory. Depictions of rape were perceived more favor-
ably by narcissists than by other men, and their higher enjoyment
was most pronounced when the male character appeared to receive
first encouragement and then refusal.
Study 3
In Study 3, we sought to experimentally assess how high and
low narcissists reacted when a woman denied them something
sexual in nature. For obvious reasons, we used analog procedures
that fell short of full-fledged date rape. That is, we sought to create
a situation that was analogous in theoretical terms to events that
might occur in a date rape situation. Male participants believed
they were involved in a study assessing perceptions of porno-
graphic material. We told them a female actor would read them a
piece of pornographic prose (thereby leading them to expect a
form of sexual stimulation), which they would later evaluate. The
female actor stopped reading the sexual prose at the same point for
all participants. For some participants, however, the female was
very reluctant to read the prose, and eventually she refused to read
any more, leading the male participants to believe that they had
been denied access to the sexual material. For other participants,
the female actor read the entire passage without stopping and
without prodding from the experimenter.
Participants then decided how much money the female actor
should receive for the session, whether she should be rehired for
future studies, and whether they would like to participate in an-
other study with her. Empathic concern for the confederate was
manipulated by saying the female actor either needed the money or
did not need it. We predicted that high narcissists, compared with
low narcissists, would give the female actor less money, would be
less likely to endorse rehiring her, and would be less willing to
participate in another study with her, especially if she denied them
something sexual (i.e., stopped reading the sexual prose). The
narcissistic reactance theory proposes that male narcissists do not
have empathic concern for female victims. Therefore, we predicted
that high narcissists would be unaffected by the confederate’s need
for money, whereas low narcissists would give the female actor
more money and be more willing to rehire her when she needed the
money than when she did not need it.
Method
Participants
Participants were 120 males enrolled in introductory psychology
courses. Participants were randomly selected from a group of 512 men who
had completed the Narcissistic Personality Inventory (Raskin & Terry,
1988) as part of a battery of tests given in mass-testing sessions. The alpha
coefficient for the Narcissistic Personality Inventory was .74, and the
average narcissism score was 18.75 (SD ⫽ 6.32). The mass-testing and
experimental sessions were about 2 months apart. Students received extra
course credit in exchange for their voluntary participation. None of the
participants in Study 3 participated in Studies 1 or 2.
Procedure
Participants were told that the study was concerned with men’s percep-
tions of sexually explicit and nonexplicit material. Participants were told
that they would read one piece of prose and have another piece read to
them. All participants were told that a female actor would first read them
a sexually explicit passage. They were told that because of the sensitive
nature of the material, they would not see or meet the female actor.
The male experimenter seated the participant in an individual room with
a video camera that was turned on but was not actually taping the session.
Each participant was told that the video camera was needed to give the
1034
BUSHMAN, BONACCI, VAN DIJK, AND BAUMEISTER

female actor feedback about how her audience was responding to the
passage. Participants wrote a brief paragraph about themselves that the
female actor would read, supposedly to learn more about them. In reality,
there was no female actor present during the experiment. A tape recording
of the female actor reading the prose, and any dialog between the actor and
experimenter, was played over an intercom system.
While writing their self-descriptions, participants heard the female actor
enter the lab. They also overheard, through the intercom, a conversation
between the female actor and the experimenter about her payment. The
experimenter said that she would not be paid until the end of the semester.
In the high need for money condition, participants overheard the female
actor reply “Are you sure it couldn’t be any sooner? I really need the
money. I’m behind on my tuition payments because my student loan fell
through.” In the low need for money condition, participants overhead the
female actor reply, “That’s fine. I don’t really need the money anyway. I’m
just doing this for fun.”
The experimenter then returned to the participant and apologized for
“accidentally” leaving the intercom turned on. He said the female actor
would begin reading the prose selection after looking over the participant’s
self-description.
After a few minutes, the female actor began reading the sexually explicit
prose, a passage from Linda Howard’s (2000) romance novel titled A Game
of Chance. The selection depicted a sexual encounter between two con-
senting adults, but was not pornographic. However, during experimental
instructions, the participants were told the piece would become porno-
graphic. In the reluctant female condition, the female actor was clearly
uncomfortable while reading the prose selection. Although she began
reading calmly and seductively (e.g., low tone and slow rate), a third of the
way through the passage, she read nervously and hesitantly (e.g., stutter,
frequent pauses, and rushed speech). She then stated she was uncomfort-
able reading and wished to stop. After prodding by the experimenter to
continue reading, she continued reading, but hesitantly. Two thirds of the
way through the passage, she again refused to continue, saying she was
uncomfortable. The experimenter again prodded her to continue, arguing
that she had read the study for all of the other men in the study. We thought
the feedback about the female actor reading the passage to all other men,
but not to the participants personally, would be a big blow to narcissists’
egos. The female actor continued to read the selection, but was clearly
flustered. She then vehemently refused to continue reading, saying the
selection was too pornographic after this point. The experimenter again
reminded her she had not had problems reading the entire passage to other
men. The female actor responded that there was “something” about the
participant that made her uncomfortable, and she again refused to continue.
The latter statement was also expected to be a blow to narcissists’ egos. In
the willing female condition, the female actor read the entire selection
calmly and seductively without any protest. The stopping point in the
willing female condition was the same stopping point in the reluctant
female condition. However, this time we let participants believe this was
simply the end of the passage. Once the female actor finished reading, the
experimenter thanked her and said she was free to leave.
After the female actor finished reading, the experimenter returned to the
participant. In the reluctant female condition, the experimenter explained
that the female actor refused to continue reading, even though she still had
several minutes of reading left. Participants in the willing female condition
were simply told that she had finished the selection.
The experimenter then explained that the female actor was hired by the
Psychology Department and was being paid for her time. He told partici-
pants that the Psychology Department was interested in their feedback
about how much money she should receive, and if she should be rehired for
future studies. The experimenter assured participants their responses would
be completely confidential. Participants indicated how much money, from
$0 to $20, they felt the female actor should receive for the session.
Participants also were asked to answer the following questions: “If I were
part of the hiring committee, I would rehire the female actor I heard” and
“I would like to participate in another study with the female actor I heard.”
Responses were made on a scale ranging from 1 (strongly disagree)to10
(strongly agree).
There was some concern that hearing the sexually explicit passage may
cause some degree of arousal. As a means to dissipate this arousal, all
participants read and evaluated a nonsexual piece of prose: a selection from
Robert Fulghum’s (1988) All I Really Needed to Know I Learned in
Kindergarten. We told participants that because they heard the female actor
read the sexual prose, they would read the nonsexual prose themselves.
After the dependent variables were collected, participants responded to
some manipulation check items. As a check on the need for money
manipulation, participants were asked how much the female actor needed
the money on a scale ranging from 1 (not at all)to10(a lot). As a check
on the female refusal manipulation, participants rated whether the female
actor was eager and willing to read the sexual prose on a scale ranging
from 1 (not at all)to10(extremely).
2
Finally, participants were probed for
suspicion and thoroughly debriefed.
Results
Data Analysis Strategy
The data were analyzed using a 2 (high vs. low need for
money) ⫻ 2 (reluctant vs. willing female) ⫻ 2 (high vs. low
narcissism) factorial design. A median split was used to classify
participants into high-and low-narcissism categories. Treating nar-
cissism as a continuous variable in regression analyses yielded the
same pattern of results.
Manipulation Checks
Perceived need for money. As a check on the manipulation of
the confederate’s perceived need, participants were asked how
much the female actor needed the money on a scale ranging from 1
(not at all)to10(a lot). As expected, participants in the high-
empathy condition thought she needed the money more than did
participants in the low-empathy condition (Ms ⫽ 5.08 and 4.25,
respectively), F(1, 116) ⫽ 4.28, p ⬍ .05, d ⫽ 0.38. Thus, the
manipulation of monetary need was effective.
Female refusal. As a check on the manipulation of the con-
federate’s willingness versus refusal to read the sexual passage,
participants rated whether the female actor was eager and willing
to read the sexual prose on a scale ranging from 1 (not at all)to10
(extremely). As expected, the female actor was rated as more eager
and willing to read the sexual prose in the willing female condition
than in the reluctant female condition (Ms ⫽ 5.64 and 3.32,
respectively), F(1, 118) ⫽ 24.97, p ⬍ .0001, d ⫽ 0.91. Thus, the
female reluctant versus willing manipulation was effective.
2
Participants in Study 3 also rated how sexually arousing the sexually
explicit passage was, using a 10-point scale ranging from 1 (not at all)
to 10 (extremely). The overall mean for sexually arousing ratings was 4.19
(SD ⫽ 2.44), indicating intermediate arousal. None of the manipulated
variables (i.e., money need, female willingness) or measured variables (i.e.,
narcissism) influenced ratings of how sexually arousing the sexually ex-
plicit passage was. This is not surprising given that the sexually explicit
passage described mutually consenting activity between a man and a
woman. In Study 2, narcissists and nonnarcissists did not differ in their
reactions to the videotape depicting mutually consenting activity. In addi-
tion, sexual arousing ratings did not influence any of the dependent
variables. We therefore report the simpler analyses in this article.
1035
NARCISSISM AND SEXUAL COERCION

Attractiveness Measure
Desire to be in another study with the female actor. There was
a significant interaction between narcissism and the female’s will-
ingness to read the sexual prose on participants’ interest in being
in another study with the female actor, F(1, 112) ⫽ 4.50, p ⬍ .05.
As can be seen in Figure 5, high narcissists expressed less desire
to participate with the female actor again when she was reluctant
to read the sexual passage, t(112) ⫽ 2.44, p ⬍ .05, d ⫽ 0.46. Thus,
narcissists were less attracted to the female actor when she denied
them sexual stimulation. In the willing female condition, where
participants were not denied something sexual, there was no dif-
ference in how much high and low narcissists wanted to participate
with the female actor again, t(112) ⫽ 0.50, ns. No other effects
were significant.
Aggressive Retaliation Measures
Amount of cash the female actor should receive. Overall, high
narcissists gave the female actor less money than did low narcis-
sists (Ms ⫽ $7.83 and $10.75, respectively), F(1, 112) ⫽ 8.52, p ⬍
.005, d ⫽ 0.55. There also was a significant interaction between
narcissism and the female’s willingness to read the sexual prose,
F(1, 112) ⫽ 9.56, p ⬍ .005. As can be seen in Figure 6, high
narcissists gave the female actor less money than did low narcis-
sists when she was reluctant to read the sexual passage,
t(112) ⫽ 4.12, p ⬍ .0001, d ⫽ 0.78. Thus, narcissists became
aggressive in response to being denied sexual stimulation. In the
willing female condition, where participants were not denied
something sexual, there was no difference in how much money
high and low narcissists gave the female actor, t(112) ⫽ 0.13, ns.
No other effects were significant.
Evaluation of whether the female actor should be rehired.
There was a significant three-way interaction between narcissism,
empathy, and the female’s willingness to read the sexual prose,
F(1, 110) ⫽ 8.91, p ⬍ .005. To interpret the three-way interaction,
we examined the two-way interaction between narcissism and
empathy separately for reluctant female and willing female con-
ditions (see Figures 7A and 7B). In the reluctant female condition,
the two-way interaction between narcissism and empathy was
significant, F(1, 112) ⫽ 5.91, p ⬍ .05. As can be seen in Figure
7A, high narcissists were less likely than low narcissists to rec-
ommend rehiring the female actor if she really needed the money,
t(112) ⫽ 2.52, p ⬍ .05, d ⫽ 0.48. The hiring recommendations of
high and low narcissists did not differ when the female actor did
not need the money, t(112) ⫽ 0.98, ns. The two-way interaction
between narcissism and empathy was nonsignificant in the willing
female condition, F(1, 112) ⫽ 3.15, ns (see Figure 7B).
Discussion
The third study sought to test the narcissistic reactance hypoth-
esis by having a female confederate (actually a tape-recorded
confederate) refuse to read a sexually arousing passage aloud to
the narcissist, even though she had allegedly been willing to read
it to other, prior participants. In effect, she refused to provide
sexual stimulation to the participant. According to reactance the-
ory, this refusal should lead to aggression toward the woman, and
we predicted these reactions would be strongest among narcissistic
men.
The main measure of aggression was the amount of money that
the participant recommended paying the confederate. Participants
in the reluctant female condition had some grounds for recom-
mending relatively low amounts, because the woman had refused
to do her assigned job of reading the passage. Still, narcissists
appear to have reacted more negatively than other men. They
recommended lower payments. This does not appear to reflect a
general stinginess, because in the control condition (where the
woman read the passage without complaint) narcissists and non-
narcissists recommended almost identical average payments.
When the woman refused, however, the narcissists retaliated much
more strongly than others, as indicated by their withholding pay-
ment. The same pattern of results was obtained when participants
Figure 5. Interaction between narcissism and the female’s willingness to
read the sexual prose on ratings of how much participants wanted to be in
another study with the female actor.
Figure 6. Interaction between narcissism and the female’s willingness to
read the sexual prose on money allocated to the female actor.
1036
BUSHMAN, BONACCI, VAN DIJK, AND BAUMEISTER

were asked if they wanted to participate in another study with the
female actor.
Similar findings were obtained on a second measure of aggres-
sion, namely the recommendation as to whether the confederate
should be rehired. The most negative recommendations for rehir-
ing the confederate came when she refused to read the passage to
a narcissist and she had indicated that she really needed the money
(and therefore would be most hurt by not getting the job). Narcis-
sists were thus again punitive in response to being denied sexual
stimulation.
Although the responses of narcissists indicate punitive aggres-
sion, our procedure did not offer them the opportunity for specif-
ically sexual aggression. When she refused to provide the narcis-
sist with the sexual stimulation he anticipated, he responded by
reducing her pay and impairing her chances to get a job she
wanted. We did not measure whether he engaged in rape or sexual
coercion. In our view, the finding that narcissists respond to sexual
disappointments with aggression indicates that a hostile, punitive
aggression is involved and therefore increases the interest value of
the findings. (After all, if narcissists had responded to the sexual
disappointment with sexual coercion, one might interpret that
pattern as merely a persistence at pursuing the sexual goals,
without any hostile or aggressive attitudes.) It seems likely that
sexual aggression will follow the same patterns as aggression
generally, although if there were some theoretical reason to expect
that sexual disappointment would fail to produce sexual aggression
(even though it produced nonsexual aggression), then further re-
search would be warranted to replicate the present findings using
specifically sexual measures of aggression. For the present article,
our finding of nonsexual aggression supports the theory that nar-
cissistic males respond to sexual refusals with reactance, including
its aggressive aspect.
Withholding pay and voting against rehiring the woman may not
constitute sexual coercion per se, but it is more directly analogous
to what occurs in many instances of sexual harassment. Men who
harass female coworkers sometimes make job-related rewards,
including pay and sometimes continued employment, contingent
on sexual favors, and when the woman fails to fulfill the man’s
sexual wishes, he punishes her by impairing her prospects for
monetary success at her career. Hence, readers who are hesitant
about generalizing from Study 3⬘s results to rape may prefer to link
them to harassment.
We also note that the design of this study did not include
nonsexual stimuli. Our goal was to study reactions to being de-
prived of sexual stimulation, because we were testing a theory
about sexual coercion. However, it is plausible in principle that
participants would be offended had the confederate refused to
continue reading them a passage about statistical methods, global
warming, political economics, or any randomly chosen topic (or
perhaps at least some topic that held a promise of being enjoyable
and entertaining, such as sports). We suspect, however, that sexual
stimulation is of special importance to the undergraduate males
that formed our study population and that they would be especially
disappointed to miss out on this. In any case, the relevance of our
findings to sexual coercion would not be diminished if parallel
patterns were sometimes found in nonsexual contexts.
Consistent with the first two studies, there was evidence of low
empathy. On both the payment and the rehiring recommendations,
some differences emerged as a function of whether the woman said
she really needed the money. Her expression of need appears to
have elicited some sympathy from the nonnarcissistic men. Al-
though they may have been disappointed or even offended when
she refused to read the sexual passage, they were more generous
toward her when she indicated that she really needed the money.
Narcissists were however unmoved by her expression of need.
These results suggest that the reactions of narcissists (as compared
with other men) revolved mainly around their own wishes and
feelings and were relatively insensitive to the woman’s wishes and
feelings.
General Discussion
The main findings from this work can be summarized as fol-
lows. The first study showed that narcissism correlated positively
with rape myth acceptance and negatively with empathy toward
Figure 7. (A) Interaction between narcissism and empathy on hiring
evaluations in the female reluctant condition. (B) Interaction between
narcissism and empathy on hiring evaluations in the female willing con-
dition.
1037
NARCISSISM AND SEXUAL COERCION

rape victims. That is, narcissistic males were more likely than
other men to hold rape-conducive beliefs, such as the belief that if
a woman says no to sex she may actually mean yes. They were also
more likely than other men to think that rape victims often share
responsibility for the rape, and so by extension, perpetrators of
sexual coercion should not bear as much blame. Thus, in situations
in which a narcissistic man (as compared with other men) wanted
sex and a woman refused, he might be tempted to continue pushing
for sex (possibly including by force) and might find it easy to
justify his own use of coercive force as something she deserved or
caused.
The second study involved watching filmed depictions of rape.
Narcissists responded more favorably (or less unfavorably) than
other men to some of these scenes. Narcissists rated these scenes
as more entertaining, more enjoyable, and more sexually arousing.
More precisely, the relatively higher enjoyment of narcissists was
most pronounced when the rape scene was preceded by depictions
of consensual affectionate activity. This fits the view that narcis-
sists are less prone than other men to condemn sexual coercion if
the woman has seemingly encouraged the man.
The third study approached the issue by having male partici-
pants experience a sexual refusal from a female confederate.
Narcissists appear to have reacted more negatively than other men
to this refusal: They recommended lower levels of pay for her and
were less favorable toward rehiring her. (In the control condition,
where there was no refusal, narcissists did not differ from other
participants.) Thus, narcissists were more punitive than other men
toward a woman who refused them some sexual stimulation that
they had anticipated.
All three studies provided some evidence that narcissistic men
feel less empathy than other men toward a woman who may be a
victim of aggression or sexual coercion. Narcissists scored lower
on a questionnaire measure of empathy toward rape victims in
Study 1. In Study 2, they were less prone to identify with the
female than with the male character (as compared with other
viewers), regardless of instructions. Moreover, when instructed to
identify with the female victim, they apparently were less success-
ful in perceiving the rape the way a victim would feel, insofar as
they gave it lower ratings on violence than other viewers. They
were no different from other viewers when instructed to identify
with the male character. These results fit the pattern of selective
empathy: Narcissists can be empathic under some circumstances,
but apparently they do not bother trying to empathize with victims
of sexual assault. Last, narcissists also exhibited low empathy in
Study 3 in that they were less sympathetic and responsive to the
female confederate’s (overheard) claim that she needed money.
Indeed, when the woman refused to provide the narcissist with
sexual stimulation, he seemed especially punitive by voting against
rehiring her for the job when she said she most needed the money.
Her expression of need appears to have elicited some sympathetic
treatment from the nonnarcissistic men, whereas the narcissists
were unmoved by her plight and, if anything, seemed to use her
need as a way to punish her for the sexual refusal.
The present results also support the view that reactance can
contribute to sexual coercion. According to the theory, reactance
should arise when a man feels that he may justifiably anticipate
sexual stimulation but the woman then refuses it. In Study 2, the
positive reactions of narcissists (compared with other men) oc-
curred only when the rape scene followed depictions of consen-
sual, affectionate activity, and it seems likely that the consensual
activity may have encouraged the men (including, by identifica-
tion, the viewers) to expect that sexual intercourse would occur. In
Study 3, the participant was led to expect that the woman would
read a sexually explicit passage to him, but then the woman
refused. Moreover, her refusal was presented as somewhat arbi-
trary and specific to him (which should generate maximum reac-
tance). When the same woman read the same amount of material
without creating reactance (by refusing), participants treated her
more favorably, and narcissism made no difference.
Although the present studies were designed to test the narcis-
sistic reactance theory, the findings could be interpreted in other
contexts as well. Most notably, Malamuth (1996) has proposed a
confluence model of rape and sexual aggression, which holds that
the trait of hostile masculinity will combine with a preference for
impersonal sex so as to cause sexual aggression. Although hostile
masculinity and impersonal sex may sound quite different from
narcissism and reactance, there are some possible similarities. In
particular, Malamuth (1996) proposed that men who feel hurt,
rejected, and otherwise mistreated by women are more prone to
develop the hostile masculinity syndrome (which itself is charac-
terized by the desire to control women and an insecure but hostile
attitude toward them). Narcissists may be especially prone to
follow the path from feeling rejected, hurt, or mistreated into
becoming hostile and aggressive (e.g., Bushman & Baumeister,
1998). Malamuth, Heavey, and Linz (1993, p. 87) have noted that
some measures of hostile masculinity are highly correlated with
narcissism. Meanwhile, impersonal sex in Malamuth’s (1996)
model is understood as an enjoyment of casual, uncommitted,
game-playing sex. Because many women are averse to uncommit-
ted or casual sex (e.g., Oliver & Hyde, 1993), men who seek and
desire it may find their wishes thwarted, and they might well
respond with reactance—especially if they believed they were
going to have that kind of sex or that the woman encouraged and
then rejected them. As we proposed, narcissists may be especially
likely to have such feelings of inflated entitlements and expecta-
tions. If one equates rejected narcissists with Malamuth’s (1996)
notion of hostile masculinity, then the punitive reactions observed
in the present Study 3 seem quite compatible with the confluence
model. It is a greater stretch to interpret Study 2’s findings (espe-
cially the higher levels of entertainment and sexual arousal re-
ported by narcissists in response to affection, as compared with
how they responded to depictions of force) as supporting the
confluence model, but they do not clearly contradict it either.
Our findings are less compatible with certain other theories.
Brownmiller (1975) proposed that all men support and benefit
from rape, and the negative reactions of our participants to the rape
depictions contradict that view. Brownmiller’s approach was up-
dated by Hall and Barongan (1997), who proposed that rape in
America is a result of normal white male socialization and that
ethnic minority or feminist socialization is an effective antidote to
rape. (They conceded that their theory conflicts with FBI data
showing that some ethnic minorities commit rape at higher rates
than White males.) Again, though, most participants in our studies
were the products of the mainstream socialization decried by Hall
and Barongan, and yet only some became aggressive in these
sexual situations and only under certain, limited circumstances.
Our data suggest that narcissism, rather than white male socializa-
tion, is a more promising trait on which to pin a predisposition
1038
BUSHMAN, BONACCI, VAN DIJK, AND BAUMEISTER

toward rape. On the positive side, however, Hall and Barongan did
propose factors such as low empathy and indifference to coercion
as factors that predispose toward rape, and our findings are con-
sistent with those points.
The limitations of the present investigation are readily apparent.
It is not feasible to have research participants engage in actual
rape, of course. We have tried to test causal hypotheses by using
procedures that were in some ways analogous to real-world sexual
coercion, including measuring how people reacted to film depic-
tions of rape (Study 2) and by leading men to expect but then be
refused some sexual stimulation by a female confederate (Study 3).
Skeptics might certainly claim that these experiences differ in
important ways from sexual coercion outside the laboratory. Still,
the review by Baumeister et al. (2002) used real-world findings to
construct their theory of narcissistic reactance, and so that theory
does fit what is known about actual rape. Data on nonlaboratory,
actual rape has ample drawbacks and limitations as well, which is
why it seems necessary to look for convergence between labora-
tory simulations and experiences in reality. Moreover, the present
investigation used three quite different methods, and the conver-
gence of findings is therefore encouraging.
Another limitation is that not all acts of rape and sexual coercion
follow the same pattern or arise from the same roots. We think it
highly plausible that there is more than one correct theory about
the causes of rape. The present findings should not therefore be
taken as ruling out or discrediting alternative theories. They indi-
cate only that narcissism and reactance may be worthy of further
study as causal factors in many instances of rape and sexual
coercion.
Sexual coercion is an unconscionable abuse of another person
that exploits another’s body for one’s own sexual gratification.
When women refuse sexual advances, most men respect that
refusal, but a minority press ahead and use force. To explain these
unusual, shameful, but important instances, it seems necessary to
invoke both individual predisposing traits and situational factors.
The present results suggest that narcissistic men may be more
prone than others to engage in sexual coercion, especially in
circumstances in which they can rationalize their behavior as
having been encouraged by the woman or in which they feel that
the woman offended them by refusing them something they both
wanted and anticipated.
References
American Psychiatric Association. (1994). Diagnostic and statistical man-
ual of mental disorders (4th ed). Washington, DC: Author.
Anderson, P. B., & Struckman-Johnson, C. (Eds.). (1998). Sexually ag-
gressive women: Current perspectives and controversies. New York:
Guilford Press.
Baumeister, R. F., Catanese, K. R., & Vohs, K. D. (2001). Is there a gender
difference in strength of sex drive? Theoretical views, conceptual dis-
tinctions, and a review of relevant evidence. Personality and Social
Psychology Review, 5, 242–273.
Baumeister, R. F., Catanese, K. R., & Wallace, H. M. (2002). Conquest by
force: A narcissistic reactance theory of rape and sexual coercion.
Review of General Psychology, 6, 92–135.
Baumeister, R. F., Smart, L., & Boden, J. M. (1996). Relation of threatened
egotism to violence and aggression: The dark side of high self-esteem.
Psychological Review, 103, 5–33.
Baumeister, R. F., & Tice, D. M. (2000). The social dimension of sex. New
York: Allyn & Bacon.
Baumeister, R. F., & Vohs, K. D. (2001). Narcissism as addiction to
esteem. Psychological Inquiry, 12, 206–210.
Bernat, J. A, Calhoun, K. S., & Adams, H. E. (1999). Sexually aggressive
and nonaggressive men: Sexual arousal and judgments in response to
acquaintance rape and consensual analogues. Journal of Abnormal Psy-
chology, 108, 662–673.
Brehm, J. W. (1966). A theory of psychological reactance. New York:
Academic Press.
Brehm, J. W. (1972). Responses to loss of freedom: A theory of psycho-
logical reactance. Morristown, NJ: General Learning Press.
Brehm, S. S., & Brehm, J. W. (1981). Psychological reactance. New York:
Wiley.
Brownmiller, S. (1975). Against our will: Men, women, and rape. New
York: Simon & Schuster.
Burt, M. R. (1980). Cultural myths and supports for rape. Journal of
Personality and Social Psychology, 38, 217–230.
Bushman, B. J., & Baumeister, R. F. (1998). Threatened egotism, narcis-
sism, self-esteem, and direct and displaced aggression: Does self-love or
self-hate lead to violence? Journal of Personality and Social Psychol-
ogy, 75, 219–229.
Deitz, S. R., Blackwell, K. T., Daley, P. C., & Bently, B. J. (1982).
Measurement of empathy toward rape victims and rapists. Journal of
Personality and Social Psychology, 43, 372–384.
Felson, R. (2002). Violence and gender reexamined. Washington, DC:
American Psychological Association.
Frank, S. J., Jackson-Walker, S., Marks, M., Van-Egeren, L. A., Loop, K.,
& Olson, K. (1998). From the laboratory to the hospital, adults to
adolescents, and disorders to personality: The case of psychological
reactance. Journal of Clinical Psychology, 54, 361–381.
Fulghum, R. (1988). All I really needed to know I learned in kindergarten.
New York: Ivy Books.
Hall, G. C. N., & Barongan, C. (1997). Prevention of sexual aggression:
Sociocultural risk and protective factors. American Psychologist, 52,
5–14.
Hall, G. C. N., Shondrick, D. D., & Hirschman, R. (1993). The role of
sexual arousal in sexually aggressive behavior: A meta-analysis. Journal
of Consulting and Clinical Psychology, 61, 1091–1095.
Howard, L. (2000). A game of chance. New York: Silhouette Books.
Joubert, C. E. (1992). Antecedents of narcissism and psychological reac-
tance as indicated by college students’ retrospective reports of their
parents’ behaviors. Psychological Reports, 70(3, Pt. 2), 1111–1115.
Kanin, E. J. (1985). Date rapists: Differential sexual socialization and
relative deprivation. Archives of Sexual Behavior, 14, 219–231.
Kaplan, J. (Director). (1988). The accused [Motion picture]. United States:
Paramount Pictures.
Laumann, E. O., Gagnon, J. H., Michael, R. T., & Michaels, S. (1994). The
social organization of sexuality: Sexual practices in the United States.
Chicago: University of Chicago Press.
Lisak, D., & Roth, S. (1988). Motivational factors in nonincarcerated
sexually aggressive men. Journal of Personality and Social Psychology,
55, 795–802.
Malamuth, N. M. (1989). The Attraction to Sexual Aggression scale: II.
Journal of Sex Research, 26, 324–354.
Malamuth, N. M. (1996). The confluence model of sexual aggression:
Feminist and evolutionary perspectives. In D. Buss & N. Malamuth
(Eds.), Sex, power, conflict (pp. 269–295). New York: Oxford Univer-
sity Press.
Malamuth, N. M., Heavey, C. L., & Linz, D. (1993). Predicting men’s
antisocial behavior against women: The interaction model of sexual
aggression. In G. C. N. Hall (Ed.), Sexual aggression: Issues in etiology,
assessment, and treatment (pp. 63–97). Washington, DC: Taylor &
Francis.
Morf, C. C., & Rhodewalt, F. (2001). Unraveling the paradoxes of narcis-
1039
NARCISSISM AND SEXUAL COERCION

sism: A dynamic self-regulatory processing model. Psychological In-
quiry, 12, 177–196.
Muehlenhard, C. L., Danoff-Burg, S., & Powch, I. (1996). Is rape sex or
violence? Conceptual issues and implications. In D. Buss & N. Mala-
muth (Eds.), Sex, power, conflict (pp. 119–137). New York: Oxford
University Press.
Oliver, M. B., & Hyde, J. S. (1993). Gender differences in sexuality: A
meta-analysis. Psychological Bulletin, 114, 29–51.
O’Sullivan, L. F., Byers, E., & Finkelman, L. (1998). A comparison of
male and female college students’ experiences of sexual coercion. Psy-
chology of Women Quarterly, 22, 177–195.
Palmer, C. T. (1988). Twelve reasons why rape is not sexually motivated:
A skeptical examination. Journal of Sex Research, 25, 512–530.
Patrick, J. (1990). Assessment of narcissistic psychopathology in the
clergy. Pastoral-Psychology, 38, 173–180.
Raskin, R., & Terry, H. (1988). A principal-components analysis of the
Narcissistic Personality Inventory and further evidence of its construct
validation. Journal of Personality and Social Psychology, 54, 890–902.
Sankowsky, D. (1995). The charismatic leader as narcissist: Understanding
the abuse of power. Organizational Dynamics, 23, 57–71.
Singleton, J. (Producer & Director). (1995). Higher learning [Motion
picture]. United States: Columbia Tristar.
Wells, G. L., & Windschitl, P. D. (1999). Stimulus sampling and social
psychological experimentation. Personality and Social Psychology Bul-
letin, 25, 1115–1125.
Wicklund, R. A. (1974). Freedom and reactance. Potomac, MD: Erlbaum.
Received May 6, 2002
Revision received October 23, 2002
Accepted October 29, 2002 䡲
1040
BUSHMAN, BONACCI, VAN DIJK, AND BAUMEISTER