Research Article Reviews

ThreatenedEgotism.pdf

Home >Law homework help >Criminal homework help >Research Article Reviews

Threatened Egotism, Narcissism, Self-Esteem, and Direct and Displaced Aggression: Does Self-Love or Self-Hate Lead to Violence?

Brad J. Bushman Iowa State University

Roy F. Baumeister Case Western Reserve University

It has been widely asserted that low self-esteem causes violence, but laboratory evidence is lacking, and some contrary observations have characterized aggressors as having favorable self-opinions. In 2 studies, both simple self-esteem and narcissism were measured, and then individual participants were given an opportunity to aggress against someone who had insulted them or praised them or against an innocent third person. Self-esteem proved irrelevant to aggression. The combination of narcissism and insult led to exceptionally high levels of aggression toward the source of the insult. Neither form of self-regard affected displaced aggression, which was low in general. These findings contradict the popular view that low self-esteem causes aggression and point instead toward threat- ened egotism as an important cause.

How do people's thoughts and feelings about themselves in- fluence their propensities to perform acts of aggression against others? Multiple answers to this question can be suggested. FOr decades, clinical psychologists have subscribed to a conven- tional view that low self-esteem underlies aggression. Yet this is difficult to reconcile with common observations that aggressors often think very highly of themselves, as evidenced by national- istic imperialism, "master race" ideologies, aristocratic duel- ing, playground bullies, and street gang rhetoric.

The present research was designed to test the opposing pre- dictions about the link between self-views and hostile aggres- sion. Perhaps surprisingly, the psychology of aggression lacks published laboratory experimental findings on whether self-love or self-hate contributes more to aggressive behavior. One possi- ble reason is that many studies on aggression were conducted before trait differences in self-esteem, narcissism, and similar self-opinions became widely used in research. Alternatively, it may be that researchers have tried but failed to find a direct link. If violent acts are indeed committed by only a small subset of people with favorable self-views, then a simple measure of self-esteem might not show direct correlations with aggression.

S e l f - E s t e e m , T h r e a t , a n d A g g r e s s i o n

In recent decades, American society has come to look on self- esteem as an unmitigated good and as a cure for a broad variety

Brad J. Bushman, DeparUnent of Psychology, Iowa State University; Roy E Baumeister, Department of Psychology, Case Western Reserve University.

We acknowledge the support of National Institutes of Health Grant MH-51482.

We thank Dan Russell for his help with the LISREL analyses and Stacy Burrell, Robin Hunn, Deanna Mackey, Tari Mellinger, Catharine Oien, Yuko Sasaki, Diane Sidari, Cassandra Skuster, Molly Steffen, and Dawn Stevenson for serving as experimenters.

Correspondence concerning this article should he addressed to Brad J. Bushman, Department of Psychology, Iowa State University, Ames, Iowa 50011-3180. Electronic mail may be sent to bushman @iastate.edu.

of personal and social problems (e.g., California Task Force, 1990). Consistent with this view, it has been widely asserted that low self-esteem is a cause of violence (e.g., Kirschner, 1992; Long, 1990; Oates & Forrest, 1985; Schoenfeld, 1988; Wiehe, 1991). According to this theory, certain people are prompted by their inner self-doubts and self-dislike to lash out against other people, possibly as a way of gaining esteem or simply because they have nothing to lose.

A contrary view was proposed by Baumeister, Smart, and Boden (1996). On the basis of an interdisciplinary review of research findings regarding violent, aggressive behavior, they proposed that violence tends to result from very positive views of self that are impugned or threatened by others. In this analy- sis, hostile aggression was an expression of the s e l f ' s rejection of esteem-threatening evaluations received from other people. They noted that the evidence does not suggest a direct link from high self-esteem to violence, and indeed some people with high self-esteem are exceptionally nonaggressive; in general, how- ever, aggressive people form one subset of people with highly favorable, even inflated opinions of themselves.

Stability of self-esteem may form one moderator. Kernis, Grannemann, and Barclay (1989) showed that people with high but unstable (i.e., subject to daily fluctuations) self-esteem re- ported the highest tendencies toward hostility and anger, whereas people with stable high self-esteem reported the lowest. High self-esteem may thus be a heterogeneous category with links to both extremes of behavior (i.e., violent and nonviolent), which could help account for the lack of published findings about self- esteem and aggression (see also Kernis, Cornell, Sun, Berry, & Harlow, 1993). High, stable self-esteem may be indifferent or even impervious to ego threat, because one's self-love remains the same no matter what happens, and so hostility is minimal. In contrast, high but unstable self-esteem would produce height- ened sensitivity to ego threats, because the individual has much to lose and is vulnerable to the miserable feeling of a brief drop in self-esteem, and so his or her sensitivity may lead to maximal hostility (see also Kernis, 1993).

Prior work thus offers competing predictions about the effects

219

220 BUSHMAN AND BAUMEISTER

of self-esteem on aggression. One is that people with low self- esteem would show the highest levels of aggression. Another is that people with high self-esteem who receive an insulting or threatening evaluation would be most aggressive. A third is that the most aggressive responses would be made by a subcategory of people with high self-esteem (and in response to esteem threat). The quest for aggressive subcategories of high self- esteem brought us to the trait of narcissism, to which we now turn.

N a r c i s s i s m a n d T h r e a t e n e d E g o t i s m

Narcissism offers another approach to examining the possible link between egotism and hostile aggression. If threatened ego- tism is indeed the crucial cause of violence, then one may predict that vulnerability to ego threats would be the feature of self- regard most relevant to aggression. In particular, inflated, gran- diose, or unjustified favorable views of self should be most prone to causing aggression, because they will encounter the most threats and be chronically most intolerant of them (Baumeister et al., 1996). These conceptions of excessive self- love are relevant to narcissism, a term coined by Freud in honor of the mythical Greek character Narcissus, who fell in love with his own image reflected in water. Although Kernberg (1975) insisted that "the nature of normal and pathological narcissism can be ascertained only by psychoanalytic exploration" (p. 327), trait scales have been developed and have facilitated the emergence of an empirically based understanding (Emmons, 1987; Raskin & Hall, 1979; Raskin & Terry, 1988).

There are several possible ways to conceptualize the relation- ship between narcissism and self-esteem. One possibility would be that narcissism is simply an exaggerated form of self-esteem, possibly with a more emotional than cognitive character (i.e., the person may have inordinate self-love without firmly holding cognitive beliefs in his or her superior qualities). This view fits the myth of Narcissus as well as the characterization by Kohut (1971) and Kernberg (1975) of narcissism as libidinal invest- ment in the self. A related view would be that narcissism is one subcategory of high self-esteem. In particular, it is plausible that narcissists might have inflated self-esteem, unlike other people whose high self-esteem is well founded.

Another view is that narcissism involves unstable high self- esteem, which has been linked to hostility (Kernis et al., 1989). Consistent with this last view, Rhodewalt, Madrian, and Cheney (1997) found significant correlations between narcissism and instability of self-esteem, although the correlations were not so high as to indicate that the two are the same. Moreover, their data linked narcissism more strongly to instability of self-esteem than to high self-esteem per se.

Correlations between narcissism and self-esteem have varied substantially across studies, making it necessary to consider the possibility that there are some narcissists with low self-esteem. According to analyses by Kohut (1971) and Kernberg (1975), there are at least two ways that a narcissist could score low in self-esteem. One is that the narcissist may be defensive, so he or she develops a veneer of high self-regard that is nonetheless hollow or brittle because it conceals underlying feelings of inse- curity and low self-esteem. The other is that the narcissistic self-love may be an emotional, immature holdover from early

development, so the person may remain emotionally invested in a grandiose self-image despite also having developed a less favorable (and presumably more accurate) self-appraisal. In effect, the person holds two unrelated sets of views about the self, possibly with the aid of unconscious processes or dissocia- tions, so that "haughty grandiosity, shyness, and feelings of inferiority may co-exist in narcissistic personalities without af- fecting each other" (Kernberg, 1975, p. 331 ).

In a sense, then, narcissism may be less a matter of having a firm conviction about one's overall goodness (which is self- esteem in a literal sense) than a matter of being emotionally invested in establishing one's superiority. It may, in other words, be more a matter of motivation and emotion than of cognition per se: Narcissists care passionately about being superior to others, even if they are not yet convinced that they have achieved this superiority. Hence, high or low levels of narcissism could be found together with either high or low self-esteem. This view resembles the one suggested by Raskin, Novacek, and Hogan ( 1991 ) and Morf and Rhodewalt ( 1993 ), who focused on narcis- sism as an attempt to regulate self-esteem. The social behavior of narcissists may be geared toward maximizing self-esteem (e.g., by gaining the approval and admiration of others) as part of the quest to validate their grandiose self-image.

There is ample reason to suggest that narcissism could be associated with increased aggression, especially in response to insults or other negative evaluations. On theoretical and clinical grounds, Kernberg (1975) proposed that narcissism includes patterns of rage that began in response to parental rejection, and rejection by others during adulthood could reactivate that rage. Millon (1981) proposed, contrary to Kernberg's view, that nar- cissism stems from an individual having parents who overvalued him or her as a child and instilled an inflated sense of entitlement and deservingness, which clearly could generate rage whenever events fail to confirm this inflated sense. Such aggressive re- sponses seem parallel to patterns of shame-based rage that have recently been demonstrated (Tangney, 1995; Tangney, Wagner, Fletcher, & Gramzow, 1992). Kernberg (1975) observed that narcissists seem inordinately sensitive to slight insults or criti- cism, and they are prone to react with hostility.

Questionnaire studies have yielded some positive correlations between narcissism and aggressiveness or hostility (Raskin et al., 1991; Wink, 1991). Emmons (1987) linked narcissism to extreme emotional lability and strong reactions, which could well include anger and rage that might increase aggressive ten- dencies. Rhodewalt and Morf ( 1995 ) found a significant correla- tion between narcissism and hostility. In a subsequent work, Rhodewalt and Morf (in press) showed that, when initial success was followed by failure feedback, narcissists became exception- ally angry, in part because they made internal attributions for the success and then presumably believed that these flattering conclusions about themselves were jeopardized by the subse- quent failure. Meanwhile, some factors that normally restrain aggression also seem to be deficient in narcissists, insofar as narcissism is correlated with disinhibiting tendencies (Emmons, 1984) and low empathy (Watson, Grisham, Trotter, & Biderman, 1984). All of these findings suggest that aggression should be high among narcissists, particularly when their anger is pro- voked by criticism or any other esteem threat.

DOES SELF-LOVE OR SELF-HATE LEAD TO VIOLENCE? 221

T h r e a t , D i s p l a c e m e n t , a n d P r e d i c t i o n s

The main prediction for the present r e s e a r c h was that the c o m b i n a t i o n o f h i g h n a r c i s s i s m and e g o threat w o u l d lead to e x c e p t i o n a l l y h i g h levels o f aggression. That is, narcissists w h o r e c e i v e d negative interpersonal f e e d b a c k w o u l d b e strongly in- clined to res pond with aggression t o w a r d the source o f this feedback.

A l t h o u g h this was the main prediction, several additional hypotheses and theoretical issues w e r e investigated. We i n c l u d e d both s e l f - e s t e e m and narcissism and a l l o w e d the t w o variables to c o m p e t e against each other to predict aggression. First, w e sought to e x a m i n e whether s e l f - e s t e e m w o u l d contribute di- rectly to aggressive responding. A s already noted, the traditional v i e w in p s y c h o l o g y is that l o w s e l f - e s t e e m causes aggression, and so this v i e w w o u l d predict higher levels o f aggression a m o n g p e o p l e s coring l o w in self-esteem. This e f f e c t c o u l d w e l l b e independent o f all other factors. T h e o p p o s i t e view, that a g g r e s s i o n will b e highest a m o n g p e o p l e h i g h in self-esteem, is also plausible.

A second issue was whether any effects o f narcissism o r self- e s t e e m w o u l d b e c o n f i n e d to responses to e g o threat. N a r c i s s i s m is c h a r a c t e r i z e d b y feelings o f superiority o v e r other people, and so s i m p l e disregard f o r the rights and feelings o f others c o u l d result in higher aggression, even in the absence o f threat. It is also plausible that narcissists p e r c e i v e social life as a series o f struggles for d o m i n a n c e , and so they m a y attack others re- gardless o f direct threat, simply as a m e a n s o f establishing t h e m s e l v e s in a s u p e r i o r p o s i t i o n b y conquering o r intimidating other individuals. In the present studies, participants found them- selves in the position o f b e i n g evaluated b y another person, w h i c h i m p l i e s a position o f vulnerability and dependency. Later, they also f o u n d t h e m s e l v e s in a direct c o m p e t i t i o n with s o m e o n e ( w h o was either the evaluator o r a different p e r s o n ) . Either o f these c i r c u m s t a n c e s m i g h t cause narcissists to attempt to assert their o w n superiority through aggressive action.

Likewise, i f l o w s e l f - e s t e e m engenders a desire to rise a b o v e others by attacking them, it c o u l d o c c u r in any c o m p e t i t i v e situation. Hence, the traditional v i e w m i g h t predict a main e f f e c t by w h i c h l o w s e l f - e s t e e m leads to high aggression, regardless o f situational factors. Alternatively, the situations o f evaluative d e p e n d e n c y o r c o m p e t i t i o n c o u l d elicit aggression f r o m p e o p l e with l o w self-esteem.

A third issue is d i s p l a c e d aggression. T h e o r i e s about aggres- sion have v a r i e d w i d e l y in the d e g r e e to w h i c h they e m p h a s i z e the interpersonal aspect. To caricature slightly, these theories have ranged f r o m treating aggression as an eruption o f intrapsy- chic forces ( i n w h i c h c a s e the c h o i c e o f target is a l m o s t irrele- v a n t ) to treating it as a f o r m o f interpersonal c o m m u n i c a t i o n (cf. Berkowitz, 1989; Dollard, D o o b , Miller, Mowrer, & Sears, 1939; Tedeschi & Felson, 1994). In Study 2, w e e x a m i n e d ag- g r e s s i o n t o w a r d the source o f the insulting evaluation and ag- gression t o w a r d an i n n o c e n t third person. I f aggression results f r o m e g o threat simply because bad m o o d s o r other inner pro- cesses create aggressive i m p u l s e s ( o r r e m o v e the barriers to instinctive aggressive i m p u l s e s ) , then a g g r e s s i o n should be high regardless o f the target. In contrast, i f a g g r e s s i o n is a m e a n s o f c o m m u n i c a t i n g directly with the evaluator, then there should b e

n o i n c r e a s e in aggression t o w a r d the i n n o c e n t third person, even i f the participant r e c e i v e d a severe b l o w to his o r her self-esteem.

We have used the terms v i o l e n c e and a g g r e s s i o n s o m e w h a t interchangeably in this introduction. Strictly speaking, o u r labo- ratory p r o c e d u r e s m e a s u r e aggression but not violence, insofar as the latter is l i m i t e d to acts that cause serious h a r m to victims. Nonetheless, it is generally assumed that the study o f laboratory a g g r e s s i o n can shed light on the causes o f v i o l e n c e outside the laboratory. In support o f this view, A n d e r s o n and B u s h m a n ( 1 9 9 7 ) have r e v i e w e d e v i d e n c e that laboratory findings general- ize w e l l to nonlaboratory situations.

S t u d y 1

Study 1 was a direct test o f the main hypothesis that threat- ened e g o t i s m w o u l d lead to m a x i m a l aggression. We m e a s u r e d both narcissism and self-esteem, e x p o s e d participants to an eval- uation that constituted either an e g o threat o r an e g o boost, and then m e a s u r e d a g g r e s s i o n toward the person w h o had delivered the evaluation.

M e ~ o d

Trait measures. Self-esteem was measured via the standard scale developed by Rosenberg (1965), a widely used instrument with good psychometric properties. Sample items are " I feel that I have a number of good qualities," " I take a positive attitude toward myself," and " I am able to do things as well as most people." Each item is answered on a 7-point scale, and responses are summed to create a global self- esteem score, with high scores indicating high self-esteem. Narcissism was measured with the Narcissistic Personality Inventory (Raskin & Terry, 1988), which has excellent psychometric properties. The scale contains 40 items that are answered by means of a simple true-false format. Sample items from the scale are " I f I ruled the world it would be a much better place," " I am going to be a great person," and " I am more capable than other people." In the present sample, the alpha coefficients for the self-esteem and narcissism scales were .55 and .78, respectively. The correlation between the two scales was .09 (p > .05 ). Self-esteem scores were higher for men (M = 35.3, SD = 4.6) than for women (M = 34.1, SD = 3.3), t(258) = 2.30, p < .05, d = 0.30. Narcissism scores also were higher for men (M = 19.5, SD = 5.8) than for women (M = 17.4, SD = 5.1), t(258) = 3.14, p < .05, d = 0.35.

Participants. Participants were 266 undergraduate psychology stu- dents (132 men and 134 women) who received extra course credit in exchange for their voluntary participation. Participants were selected randomly from a large pool of students who had completed the self- esteem scale as part of a battery of questionnaires given in mass-testing sessions. The data for 6 participants were discarded as a result of these individuals' suspiciousness or failure to follow instructions. The final sample consisted of 260 participants ( 130 men and 130 women). Also, a separate sample of 10 men and 10 women took part in a manipulation validation study.

Procedure. Experimental participants were tested individually in the laboratory session, but each was led to believe that he or she would be interacting with someone else of the same sex. Participants were told that the researchers were studying how people react to positive and negative feedback. Informed consent was obtained after the participant had been told that the experiment would involve writing essays and then competing on a reaction time task with stressful, noisy stimuli as a possible outcome. After informed consent had been obtained, the partici- pant completed the narcissism scale. The participant was told that the scale was being used to determine whether feedback affects different types of people in different ways.

2 2 2 BUSHMAN AND BAUMEISTER

Each participant was asked to write a one-paragraph essay on abor- tion, either pro-choice or pro-life (whichever the participant preferred). After completion, the participant's essay was taken away to be shown to the other participant (who was, in fact, nonexistent) for evaluation. Meanwhile, the participant was permitted to evaluate the partner's essay, which, by random assignment, was either a pro-choice or a pro-life essay. There was one essay of each type, and every participant saw one or the other. We also controlled for handwriting by having male and female versions of the standard essays. (Which essay the participant saw had no effect on subsequent aggressive behavior, which rules out any explanation that aggression was mediated by perceptions of partner attitude or of similarity between participant and partner.)

A short time later, the experimenter returned the participant's own essay with comments ostensibly made by the other participant. These comments constituted the experimental manipulation of ego threat. By the flip of a coin, half of the participants were assigned to the ego threat condition, and they received bad evaluations consisting of negative ratings on organization, originality, writing style, clarity of expression, persuasiveness of arguments, and overall quality. There was also a hand- written comment stating "This is one of the worst essays I have read!" The other participants received favorable, positive evaluations consisting of high (positive) numerical ratings and the following written comment: " N o suggestions, great essay!"

The next part of the procedure was presented as a competitive reaction time task based on a paradigm developed by Taylor ( 1 9 6 7 ) ) Previous studies have established the construct validity of Taylor's paradigm (e.g., Bemstein, Richardson, & Hammock, 1987; Giancola & Zeichner, 1995). The participant was told that he or she and the partner would have to press a button as fast as possible on each trial and that whoever was slower would receive a blast of noise. Each participant was permitted to set in advance the intensity of the noise that the other person would receive between 60 dB (Level 1) and 105 dB (Level 10) if the other lost. A nonaggressive no-noise setting (Level 0) was also offered. In addition to determining noise intensity, the winner determined the dura- tion of the loser's suffering, because the duration of the noise depended on how long the winner held the button pressed down. In effect, each participant controlled a weapon that could be used to blast the other person if the participant won the competition to react faster.

A Macintosh II computer controlled the events in the reaction time task and recorded the noise levels and noise durations the participant set for the "other person." The white noise consisted of sound files synthesized by a digital waveform editor (FaraUon Soundedit 2.0.5 ) and reproduced through an Audiomedia 2.0 Digidesign 16-bit digit-to-analog converter. The analog output was amplified by an NAD 3225PE inte- grated amplifier and delivered through a pair of Telephonics TDH-39P headphones. A General Radio 156-B sound level meter was used to calibrate the noise levels.

After completion of the reaction time task, the participant was de- briefed and dismissed. A separate sample of participants took part in a validation study to check the ego threat manipulation (see later discussion).

R e s u l ~

Manipulation validation. To v e r i f y the i m p a c t o f t h e e g o t h r e a t m a n i p u l a t i o n , w e c o n d u c t e d a p i l o t study. A s m e n t i o n e d earlier, 10 m e n a n d 10 w o m e n t o o k part. T h e y f o l l o w e d t h e s a m e p r o c e d u r e o f w r i t i n g t h e e s s a y a n d r e c e i v i n g e i t h e r the f a v o r a b l e o r u n f a v o r a b l e e v a l u a t i o n . I n s t e a d o f c o n t i n u i n g o n to t h e a g g r e s s i o n m e a s u r e , however, p a r t i c i p a n t s c o m p l e t e d a q u e s t i o n n a i r e a s s e s s i n g h o w t h e y f e l t o n r e c e i v i n g t h e e v a l u a t i o n a n d h o w t h e y p e r c e i v e d the e v a l u a t i o n .

A l l e f f e c t s w e r e l a r g e a n d significant. T h e b a d e v a l u a t i o n o f t h e p a r t i c i p a n t ' s essay, i n c o m p a r i s o n w i t h t h e g o o d e v a l u a t i o n ,

w a s r a t e d as m o r e t h r e a t e n i n g , t ( 1 8 ) = 2.19, p < .05, d = 0.98; m o r e m a l i c i o u s , t ( 1 8 ) = 4.94, p < .05, d = 2.21; a n d less fair, t ( 1 8 ) = - 5 . 0 8 , p < .05, d = 2.29. A l s o , p a r t i c i p a n t s r e c e i v i n g t h e b a d e v a l u a t i o n ( r e l a t i v e to t h o s e r e c e i v i n g the g o o d evalua- t i o n ) r e p o r t e d t h a t it l o w e r e d t h e i r s e l f - e s t e e m , t ( 1 8 ) = 3.05, p < .05, d = 1.36, a n d m a d e t h e m f e e l angry, t ( 1 8 ) = 2.21, p < .05, d = 0.99. T h e s e r e s u l t s c o n f i r m t h a t the b a d e v a l u a t i o n p r o c e d u r e d i d i n d e e d c o n s t i t u t e a n u p s e t t i n g e g o threat.

Main analysis strategy. N o i s e i n t e n s i t y a n d n o i s e d u r a t i o n w e r e m e a s u r e s o f the s a m e c o n s t r u c t : a g g r e s s i v e b e h a v i o r . T h e s a m e p a t t e r n o f r e s u l t s w a s o b t a i n e d f o r b o t h m e a s u r e s , a n d the t w o m e a s u r e s w e r e s i g n i f i c a n t l y c o r r e l a t e d ( r = . 3 2 ) . 2 A s a m e a n s o f c r e a t i n g a m o r e r e l i a b l e m e a s u r e , the n o i s e i n t e n s i t y a n d n o i s e d u r a t i o n d a t a w e r e s t a n d a r d i z e d a n d s u m m e d to f o r m a t o t a l m e a s u r e o f a g g r e s s i v e b e h a v i o r .

T h e d a t a w e r e a n a l y z e d v i a r e g r e s s i o n a n a l y s i s . I n r e g r e s s i o n a n a l y s i s , r e s e a r c h e r s r e c o m m e n d c e n t e r i n g the p r e d i c t o r vari- a b l e s w h e n t e s t i n g f o r i n t e r a c t i o n e f f e c t s (e.g., A i k e n & West, 1991; J a c c a r d , Turrsi, & W a n , 1 9 9 0 ) . T h i s t r a n s f o r m a t i o n , w h i c h r e d u c e s the c o r r e l a t i o n b e t w e e n the p r o d u c t t e r m a n d t h e c o m p o - n e n t p a r t s o f the t e r m , w a s u s e d i n t h e p r e s e n t a n a l y s e s . T h e r e g r e s s i o n m o d e l i n c l u d e d m a i n e f f e c t s f o r e g o t h r e a t (1 = p r e s e n t , 0 = a b s e n t ) , n a r c i s s i s m ( c o n t i n u o u s ) , s e l f - e s t e e m ( c o n t i n u o u s ) , a n d sex (1 = m a l e , 0 = f e m a l e ) . T h e m o d e l a l s o i n c l u d e d t w o - w a y a n d t h r e e - w a y i n t e r a c t i o n s , w h i c h w e r e c o m p u t e d as m u l t i p l i c a t i v e p r o d u c t s o f t h e m a i n effects. A hier- a r c h i c a l a n a l y s i s o f sets a p p r o a c h w a s u s e d ( C o h e n & C o h e n , 1 9 8 3 ) . T h e m a i n e f f e c t s w e r e e n t e r e d in the first step, the t w o - w a y i n t e r a c t i o n s w e r e e n t e r e d i n t h e s e c o n d step, a n d the t h r e e - w a y i n t e r a c t i o n s w e r e e n t e r e d i n the t h i r d step. T h e f o u r - w a y i n t e r a c t i o n w a s a d d e d to the e r r o r term. T h u s , t h e m a i n e f f e c t s w e r e r e m o v e d f r o m the t w o - w a y i n t e r a c t i o n s , a n d the m a i n

The reaction time task consisted of 25 trials. After the initial (no provocation) trial, the remaining 24 trials were divided into three blocks of 8 trials each. The participant received feedback on the intensity of noise the "opponent" set on each trial. Provocation was manipulated by increasing the intensity and duration of noise blasts the "other per- s o n " set for the participant across trials. In this article, we describe only the results of Trial 1 aggression. Responses on the first trial provided the best measure of unprovoked aggression, because the participant had not yet received noise or feedback from the "other person." After the first trial, aggression converged on reciprocation of what the partner had ostensibly done. This is consistent with many previous findings sug- gesting that reciprocation is a powerful norm in determining aggressive responses during an ongoing aggressive exchange. Only a few other significant effects were found on subsequent trials. In Study 1, men were more aggressive than women, F ( 1 , 245) = 19.93, p < .05, d = 0.57. In Study 2, there was a main effect for aggression target that was qualified by an interaction between threat and aggression target, Fs( 1, 254) = 4.08 and 5.05, respectively, p s < .05. Participants who received a bad evaluation were more aggressive than those who received a good evaluation when the target was the source of the evaluation but not when the target was an innocent third party.

2 The regression analysis for noise intensity revealed main effects for threat, narcissism, and sex, Fs(1, 245) = 7.79, 11.72, and 23.63, respectively, p s < .05. The regression analysis for noise duration re- vealed a main effect for narcissism and a nearly significant interaction between narcissism and ego threat, F ( 1, 245) = 4.08, p < .05, and F ( 1 , 2 4 5 ) = 3.27, p < .10, respectively.

DOES SELF-LOVE OR SELF-HATE LEAD TO VIOLENCE? 223

effects and two-way interactions were removed from the three- way interactions.

Multicollinearity, or correlation among the predictor vari- ables, was tested by means of variance inflation factors (VIFs; e.g., Neter, Wasserman, & Kutner, 1990). A VIF of 1 indicates that the model terms are not linearly related. A maximum VIF value in excess of 10 is often taken as an indication that multicol- linearity may be unduly influencing the least squares estimates. The maximum VIF in the regression analyses for Study 1 was 1.1, indicating that multicollinearity was not a problem.

A g g r e s s i o n . The regression analysis yielded significant main effects for ego threat, narcissism, and sex. Ego threat in the form of insulting evaluation of the essay led to higher aggres- sion than the nonthreatening, favorable evaluation, F ( 1, 245) = 4.41, p < .05, b = 0.39, S E = 0.19, d = 0.25. There was a positive relation between narcissism and aggression, F ( 1 , 2 4 5 ) = 13.92, p < .05, b = 0.06, S E = 0.02, r = .27. Also, men were more aggressive than women, F ( 1 , 245) = 14.54, p < .05, b = 0.71, S E = 0.19, d = 0.56.

More important, there was an interaction between narcissism and ego threat, F ( 1 , 245) = 5.04, p < .05, b = 0.08, S E = 0.03. This interaction, depicted in Figure 1, indicated that (high) narcissists who received the ego threat were exceptionally ag- gressive, even above and beyond what would be expected on the basis of the simple additive combination of the two variables. In Figure 1, different regression lines are plotted for ego threat and praise feedback for the range of narcissism scores obtained in the study (see Aiken & West, 1991). This interaction con- firmed the main hypothesis regarding the effects of narcissism and ego threat on aggression. As can be seen in Figure 1, the relation between narcissism and aggression was stronger when the evaluation was negative than when it was positive, F ( 1 , 245) = 20.36,p < .05, b = 0.11, S E = 0.02, r = .37, and F ( 1 , 245) = 4.59,p < .05, b = 0.05, S E = 0.02, r = .18, respectively. Still, the effect of narcissism on aggression remained significant even when the evaluation was positive.

-1

---- E g o t h r e a t

- - - 0 - - - P r a i s e

- 2

" 1 ._o

O) 0

I i I

0 10 20 30 40

N a r c i s s i s m

Figure 1. Relation between narcissism and aggression for participants who received either a positive or negative evaluation.

When only the main effects were included in the model, the R 2 value was .14. When the two-way interactions were added, the model R 2 value was .16 (i.e., it increased by .02), and when the three-way interactions were added, the value was . 17 (i.e., it increased by an additional .01 ).

The Narcissistic Personality Inventory includes between four (Emmons, 1987) and seven (Raskin & Terry, 1988) subscales. In the present study, the total scale was more strongly related to aggression than any of the subscales.

Self-esteem was allowed to compete with narcissism to pre- dict aggression. But self-esteem was not related to aggression, either alone or in interaction with other variables. Additional analyses were conducted in which narcissism was excluded as a predictor. Even in these analyses, no significant effects involving self-esteem were found.

D i s c u s s i o n

Study 1 found that ego threats in the form of insulting, nega- tive evaluations increased aggressive responding for all types of individuals, as suggested by many prior findings (for reviews, see Baron & Richardson, 1993; Geen, 1990). More important, these aggressive responses were strongest among people who scored high in narcissism. Thus, the highest levels of aggression were found among people who have emotional and motivational investment in extremely favorable, grandiose self-images. These results are consistent with the view that threatened egotism is a particular cause of aggression and violence.

It is also noteworthy that narcissism led to increased aggres- sion independently of ego threat, as indicated by a ma~n effect. Put another way, narcissists were aggressive even toward some- one who had evaluated them favorably. As noted in the introduc- tion, there are several possible reasons for this effect, although we believed that replication was desirable before drawing strong conclusions. Study 2 was conducted in part to ascertain whether narcissists would show increased aggression toward a third per- son who had not evaluated them at all.

Self-esteem yielded no significant effects, either indepen- dently or in interaction with other variables. The lack of effects contradicts the traditional view that low self-esteem causes ag- gression, as well as any suggestion that favorable self-views in general lead to aggression (including in combination with ego threat). The standard view is that null findings are uninterpret- able; in this case, however, the fact that the narcissism scale yielded both a significant main effect and a significant interac- tion makes the failure of the self-esteem measure seem more conclusive.

Still, before self-esteem is dismissed as irrelevant to aggres- sion, several potential drawbacks must be noted. First, the self- esteem measure was given in mass-testing sessions some weeks before the laboratory session, and so its predictive power might have been somewhat weaker than that of the narcissism scale, which was given at the same experimental session as the depen- dent measures. Second, the alpha coefficient (.55) for the Rosen- berg scale was unacceptably low in this sample. Although that scale is often regarded as having good psychometric properties, it emphatically did not have them here, possibly because stu- dents participating in the group testing session failed to respond

2 2 4 BUSHMAN AND BAUMEISTER

to it as carefully, t h o u g h t f u l l y , o r h o n e s t l y as r e s p o n d e n t s i n o t h e r s e t t i n g s m i g h t .

T h i r d , it is p o s s i b l e t h a t t h e R o s e n b e r g scale is i n s e n s i t i v e to r e l e v a n t d i f f e r e n c e s . It is, i n fact, a relatively short, s i m p l e , a n d o b v i o u s m e a s u r e o f s e l f - e s t e e m . P e r h a p s surprisingly, t h e r e w a s n o s i g n i f i c a n t c o r r e l a t i o n b e t w e e n the s e l f - e s t e e m a n d n a r c i s - s i s m s c a l e s ( r = . 0 9 ) , a n d o n e w o u l d e x p e c t t h a t n a r c i s s i s m s h o u l d c o r r e l a t e w i t h s e l f - e s t e e m ( o n the b a s i s o f b o t h a p r i o r i r e a s o n i n g a n d p r e v i o u s f i n d i n g s ; K e r n b e r g , 1975; R h o d e w a l t e t al., 1 9 9 7 ) . T h u s , the l a c k o f c o r r e l a t i o n r a i s e s f u r t h e r s u s p i c i o n a b o u t the efficacy o f the scale as u s e d here.

S t u d y 2.

S t u d y 2

O u r s e c o n d e x p e r i m e n t h a d several p u r p o s e s . First, it s e e m e d d e s i r a b l e to r e p l i c a t e the e f f e c t s o f n a r c i s s i s m a n d e g o t h r e a t f o u n d i n S t u d y 1. S e c o n d , w e w a n t e d to try a d i f f e r e n t m e a s u r e o f s e l f - e s t e e m to d e t e r m i n e w h e t h e r it m i g h t h a v e m o r e s u c c e s s in y i e l d i n g s i g n i f i c a n t e f f e c t s t h a n t h e scale u s e d i n S t u d y 1.

T h i r d , it s e e m e d d e s i r a b l e to i n v e s t i g a t e a g g r e s s i o n t o w a r d s o m e o n e o t h e r t h a n the evaluator. S t u d y 1 r e v e a l e d t h a t n a r c i s - sists w e r e m o r e a g g r e s s i v e t h a n o t h e r s t o w a r d s o m e o n e w h o c r i t i c i z e d t h e m . T h e y w e r e a l s o m o r e a g g r e s s i v e , a l t h o u g h b y a s m a l l e r m a r g i n , t o w a r d s o m e o n e w h o h a d p r a i s e d t h e m . T h u s , n a r c i s s i s m s e e m s to f o s t e r a g g r e s s i v e r e s p o n s e s to e v a l u a t i o n , r e g a r d l e s s o f t h e v a l e n c e o f t h a t e v a l u a t i o n . O n e p o s s i b l e v i e w is t h a t n a r c i s s i s t s are s i m p l y m o r e a g g r e s s i v e t o w a r d e v e r y o n e , i n w h i c h c a s e a g g r e s s i o n t o w a r d a t h i r d p e r s o n w o u l d b e h i g h e r t h a n t h a t o f n o n n a r c i s s i s t s . Alternatively, it m a y b e t h a t s u b m i t - t i n g to e v a l u a t i o n is i n h e r e n t l y t h r e a t e n i n g t o n a r c i s s i s t s , g i v e n t h e i r o v e r r i d i n g s e n s e o f superiority, a n d so t h e y a g g r e s s o n l y t o w a r d p e o p l e w h o e v a l u a t e t h e m .

M e t h o d

Trait measures. In Study 2, participants completed the Narcissistic Personality Inventory (Raskin & Terry, 1988). The alpha coefficient for the scale was .80. Narcissism scores were significantly higher for men ( M = 20.0, SD = 6.5) than for women ( M = 18.5, SD = 5.7), t(278) = 2.00, p < .05, d = 0.24.

To measure self-esteem, we used a version of the Janis and Field (1959) scale. This scale has been widely used to measure self-esteem. Fleming and Courtney (1984) produced a recent version that has been influential. Following Baumeister's usual procedure (e.g., Baumeister, Heatherton, & Tice, 1993), we administered only the first three sub- scales, which measure global self-regard, academic-task esteem, and social self-esteem. This version of the scale contains 26 items that are answered in a 7-point response format. Sample items from the scale are "How often do you dislike yourself? . . . . In general, how confident do you feel about your abilities?" and "How confident do you feel that someday the people you know will look up to you and respect you?" The alpha coefficient for the self-esteem scale was .93. Self-esteem scores were significantly higher for men (M = 123.8, SD = 22.5) than for women (M = 117.8, SD = 25.3), t(278) = 2.12, p < .05, d =

0.25. Unlike the Rosenberg (1965) scale used in Study 1, this self- esteem scale did yield a significant correlation with narcissism ( r = .33, p < .05).

Participants. Participants were 281 undergraduate psychology stu- dents (141 men and 140 women) who received extra course credit in exchange for their voluntary participation. The data for 1 participant were discarded because he expressed suspicion about the procedure. The final sample consisted of 280 participants (140 men and 140 women).

Procedure. The procedure for Study 2 was the same as the procedure for Study 1, with four exceptions. First, participants completed the self- esteem scale as well as the narcissism scale in the laboratory. Second, the ostensible partner's essay was controlled so that it always agreed with the attitudinal position that the participant advocated. That is, if the participant wrote a pro-choice essay, then he or she evaluated a pro- choice essay; if the participant wrote a pro-life essay, then he or she evaluated a pro-life essay. This procedure modification allowed us to completely rule out the possibility that aggression was mediated by perceptions of partner attitude or of similarity between participant and partner. Third, the participant rated how threatening the essay evaluation was. Several other ratings were included as fillers. Ratings were made along a 10-point Likert-type scale ranging from 1 (strongly disagree) to 10 (strongly agree). The ratings were used to test whether threat mediated the relation between narcissism and aggression. Fourth, by the flip of a coin, the partner on the reaction time task was said to be either the same person who evaluated the participant's essay (direct aggression) or a different person (displaced aggression).

R e s u l t s

Analysis strategy. A s i n S t u d y 1, the n o i s e i n t e n s i t y a n d n o i s e d u r a t i o n d a t a w e r e s t a n d a r d i z e d a n d s u m m e d to f o r m a m o r e r e l i a b l e m e a s u r e o f a g g r e s s i v e b e h a v i o r . 3 A l s o as in S t u d y 1, t h e p r e d i c t o r v a r i a b l e s w e r e c e n t e r e d f o r the r e g r e s s i o n a n a l y - sis. T h e r e g r e s s i o n m o d e l i n c l u d e d m a i n e f f e c t s f o r e g o t h r e a t ( 1 = p r e s e n t , 0 = a b s e n t ) , a g g r e s s i o n t a r g e t ( 1 = d i r e c t a g g r e s - sion, 0 = d i s p l a c e d a g g r e s s i o n ) , n a r c i s s i s m ( c o n t i n u o u s ) , self- e s t e e m ( c o n t i n u o u s ) , a n d sex (1 = m a l e , 0 = f e m a l e ) . T h e m o d e l a l s o i n c l u d e d t w o - w a y a n d t h r e e - w a y i n t e r a c t i o n s . T h e i n t e r a c t i o n b e t w e e n t h r e a t a n d a g g r e s s i o n t a r g e t w a s p r e d i c t e d to b e o f a s p r e a d i n g t y p e r a t h e r t h a n a c r o s s o v e r type. Specifi- cally, h i g h e r levels o f a g g r e s s i o n w e r e e x p e c t e d for p a r t i c i p a n t s w h o r e c e i v e d a n e g a t i v e e v a l u a t i o n a n d t h e n a g g r e s s e d d i r e c t l y a g a i n s t t h e p e r s o n w h o p r o v i d e d the e v a l u a t i o n ( c o d e d as 3 ) t h a n f o r p a r t i c i p a n t s in the o t h e r t h r e e g r o u p s ( e a c h c o d e d as - 1 ). T h i s c o d i n g also w a s u s e d f o r a n y t h r e e - w a y i n t e r a c t i o n s t h a t i n c l u d e d t h r e a t a n d a g g r e s s i o n target. A s in S t u d y 1, a h i e r a r c h i c a l a n a l y s i s o f sets a p p r o a c h w a s u s e d ( C o h e n & C o - hen, 1 9 8 3 ) . T h e m a i n e f f e c t s w e r e e n t e r e d in t h e first step, the t w o - w a y i n t e r a c t i o n s w e r e e n t e r e d i n the s e c o n d step, a n d t h e t h r e e - w a y i n t e r a c t i o n s w e r e e n t e r e d in the t h i r d step. H i g h e r o r d e r i n t e r a c t i o n s w e r e a d d e d to t h e e r r o r term. M u l t i c o l l i n e a r i t y w a s t e s t e d b y m e a n s o f V I F s . T h e m a x i m u m V I F in the r e g r e s - s i o n a n a l y s e s f o r S t u d y 2 w a s 4.2, i n d i c a t i n g t h a t m u l t i c o l l i n e a r - ity w a s n o t a p r o b l e m ( N e t e r et al., 1 9 9 0 ) .

3 The regression analyses for noise intensity revealed a Threat x Aggression Target interaction and a Threat × Sex interaction, Fs( 1, 254) = 9.75 and 4.59, respectively, p s < .05. The regression analysis for noise duration revealed a Threat × Aggression Target interaction and a nearly significant Threat x Aggression Target × Narcissism inter- action, F ( 1 , 254) = 5.35, p < .05, and F ( 1 , 254) = 2.66, p < .11, respectively.

DOES SELF-LOVE OR SELF-HATE LEAD TO VIOLENCE? 2 2 5

Manipulation validation. To verify the i m p a c t o f the e g o threat manipulation, w e a n a l y z e d ratings o f h o w threatening participants thought the evaluation was. A l l variables w e r e en- tered in the r e g r e s s i o n m o d e l e x c e p t those i n v o l v i n g a g g r e s s i o n target, w h i c h was m a n i p u l a t e d after participants had rated the evaluations. A s expected, the negative evaluation was j u d g e d to be m o r e threatening than was the positive evaluation, F ( 1 , 2 6 3 ) = 124.79, p < .05, d = 1.46. N o other significant effects were found.

Aggression. The p r e d i c t e d t h r e e - w a y E g o T h r e a t × A g g r e s - sion Target X N a r c i s s i s m interaction was nearly significant, F ( 1 , 2 5 4 ) = 2.89, p < .10, b = 0.03, SE = 0.02. N a r c i s s i s m was positively related to direct aggression when the evaluation was bad, but it was unrelated to direct a g g r e s s i o n when the evaluation was g o o d , F ( 1 , 2 5 4 ) = 9.62, p < .05, b = 0.09, SE = 0.04, r = .25, and F ( 1 , 2 5 4 ) = 0.34, p > .05, b = - 0 . 0 2 , SE = 0.02, r --- - . 1 0 , respectively ( s e e left p o r t i o n o f F i g u r e 2 ) . M e a n w h i l e , narcis s ism was unrelated to displaced aggression, regardless o f whether the evaluation was bad o r good, F ( 1, 2 5 4 ) = 0.99, p > .05, b = 0.02, SE = 0.02, r = .14, and F ( 1 , 2 5 4 ) = 0 . 6 1 , p > .05, b = 0.02, SE = 0.03, r = .10, respectively ( s e e right p o r t i o n o f F i g u r e 2 ) .

Other significant effects less central to the hypotheses w e r e found. There was a main e f f e c t f o r aggression target, F ( 1 , 2 5 4 ) = 5 . 8 9 , p < .05, b = 0.65, SE = 0.03, d = 0.27. This main effect, however, was qualified b y an interaction b e t w e e n evaluation type and aggres s ion target, F ( 1 , 2 5 4 ) = 13.63, p < .05, b = 0.33, S E = 0.09. Participants w h o r e c e i v e d a bad evaluation were m o r e aggressive than those w h o r e c e i v e d a g o o d evaluation when the target was the source o f the evaluation, but n o t when the target was an i n n o c e n t third party, F ( 1 , 2 5 4 ) = 39.72, p < .05, d = 0.93, and F ( 1 , 2 5 4 ) = 1.75, p > .05, d = 0.25, respectively. T h e s e findings support the v i e w o f a g g r e s s i o n as retaliation for a bad evaluation: P e o p l e were most aggressive t o w a r d the person w h o insulted them.

W h e n only the main effects were i n c l u d e d in the model, the R 2 v a l u e was .10. W h e n the t w o - w a y interactions w e r e added,

the m o d e l R 2 v a l u e was .19 (i.e., it increased b y . 0 9 ) , and when the t h r e e - w a y interactions w e r e added, the v a l u e was .24 (i.e., it i n c r e a s e d b y an additional .05).

Given the theoretical questions about what aspect o f narcis- sism leads to aggression, w e r e p e a t e d the main analyses using the various subscales o f the Narcissistic Personality I n v e n t o r y in p l a c e o f the grand total. T h e s e subscale analyses y i e l d e d little useful i n f o r m a t i o n and generally offered simply w e a k e r versions o f the same results that the total s c o r e gave. T h e total score was m o r e strongly related to aggression than any o f its subscales.

A s in Study 1, s e l f - e s t e e m was a l l o w e d to c o m p e t e with n a r c i s s i s m to predict aggression. B u t s e l f - e s t e e m was not related to aggression, either alone o r in interaction with other variables. Even in additional analyses that e x c l u d e d narcissism as a pre- dictor, n o significant effects i n v o l v i n g s e l f - e s t e e m were found.

Threat as mediator. R e c a l l that participants in Study 2 rated the evaluation they had r e c e i v e d in terms o f its level o f threat. W e used this rating to measure perceived threat. A n analysis was c o n d u c t e d to test whether p e r c e i v e d threat m e d i a t e d the relation b e t w e e n narcissism and aggression in the condition in w h i c h participants r e c e i v e d a negative evaluation on their essay and then a g g r e s s e d directly against the same person w h o had evaluated them (i.e., the e g o t h r e a t - d i r e c t aggression condi- t i o n ) . For the other three conditions, it was p r e d i c t e d that narcis- sism, perceived threat, and aggression w o u l d be unrelated to o n e another. Table 1 shows the v a r i a n c e - c o v a r i a n c e matrices.

A s a means o f testing this model, a multiple group analysis was c o n d u c t e d with the L I S R E L 8 c o m p u t e r p r o g r a m ( J 6 r e - skog & S/Srbom, 1993). For participants in the e g o t h r e a t - direct aggression condition, a causal path was specified f r o m narcissism to perceived threat, and another causal path was spec- ified f r o m perceived threat to aggression. For participants in the other three conditions, these t w o causal paths w e r e fixed at zero. T h e hypothesized m o d e l p r o v i d e d a very g o o d fit to the data, X2(10, N = 2 8 0 ) = 7.87, p = .64, g o o d n e s s o f fit index = .98, comparative fit index = 1.00, root m e a n square e r r o r o f a p p r o x i m a t i o n = 0.0. For participants in the e g o t h r e a t - d i r e c t

3 -

- - E g o t h r e a t ~ E g o t h r e a t

- P r a i s e - - - 0 - - - P r a i s e

t -

O ¢::: "~ 0

i ~ . . . . . O . . . . . . • t ~ -1

-2 -2 I I I I ! I 0 10 2 0 3 0 4 0 0 10 2 0 3 0 4 0

N a r c i s s i s m N a r c i s s i s m

Figure 2. Relation between narcissism and direct and displaced aggression for participants who received either a positive or negative evaluation.

2 2 6 BUSHMAN AND BAUMEISTER

Table 1 D a t a f o r the LISRELAnalysis

Measure 1 2 3 M SD

Ego threat-direct aggression condition

1. Narcissism 36.03 6.16 3.12 17.87 6.00 2. Perceived threat of evaluation .33* 9.46 1.51 4.04 3.08 3. Aggression .25* .24* 4.18 0.74 2.04

Ego threat-displaced aggression condition

1. Narcissism 47.13 - 3 . 0 2 1.15 20.08 6.86 2. Perceived threat of evaluation - . 1 5 8.97 0.17 4.99 3.00 3. Aggression .15 .05 1.39 0.16 1.18

Praise-direct aggression condition

1. Narcissism 33.02 0.12 - 0 . 5 6 18.71 5.75 2. Perceived threat of evaluation .02 0.95 0.06 1.26 0.97 3. Aggression - . 1 0 .06 0.99 - 0 . 7 4 0.99

Praise-displaced aggression condition

1. Narcissism 30.37 0.21 0.72 20.34 5.51 2. Perceived threat of evaluation .06 0.39 -0.03 1.20 0.63 3. Aggression .10 - . 0 4 1.72 - 0 . 1 6 1.31

Note. Variances are on the diagonal, and covariances are above the diagonal. Correlations are reported below the diagonal for descriptive purposes, n = 70 in each condition. * p < .05.

ag g ression condition, the causal path f r o m narcissism to per- c e i v e d threat was positive and significant, as was the causal path f r o m perceived threat to aggression ( s e e F i g u r e 3 ) .

To e x a m i n e the e f f e c t o f constraining these t w o causal paths to zero for the other three conditions, w e tested a second m o d e l wherein the t w o paths w e r e specified as free parameters f o r all f o u r conditions. T h e fit o f this latter m o d e l was found to not b e significantly better than the fit o f the hypothesized model, X 2 (4, N = 280) = 5.56, p = .23. T h e s e results indicate that narcissism, perceived threat, and aggression were unrelated to o n e another in the other three conditions.

The results o f these analyses p r o v i d e support f o r the media- tion model. However, it is also possible that narcissism had a direct e f f e c t on aggression among participants in the e g o t h r e a t - direct aggression condition. To evaluate this possibility, w e tested a second modification o f the hypothesized m o d e l wherein the causal path f r o m narcissism to aggression was included as a free parameter. O n c e again, inclusion o f this parameter did not lead to a significant i m p r o v e m e n t in the fit o f the m o d e l to the data, X z ( 1 , N = 280) = 2.55, p = .11. Thus, the direct causal path f r o m narcissism to aggression was nonsignificant

(z = 1.61, p = .11). In summary, these analyses suggest that narcissists aggress directly against individuals w h o give them negative f e e d b a c k because they v i e w the f e e d b a c k as a threat to their ego.

D i s c u s s i o n

Several important findings e m e r g e d f r o m Study 2. W e repli- cated the key findings o f Study 1, in that high narcissism in- creased aggression overall and led to especially high aggr e ssion in direct response to the bad, threatening evaluation. The appar- ent links b e t w e e n narcissism and aggressive b e h a v i o r were thus confirmed.

We also replicated the null results o f Study 1 in terms o f absence o f s e l f - e s t e e m effects. That Study 2 m e a s u r e d self- e s t e e m in the same session as the aggression m e a s u r e and used a different s e l f - e s t e e m scale f r o m Study 1 w o u l d seemingly rule out some o f the possible explanations for the null results o f Study 1 and cast further doubt on the direct relevance o f self- e s t e e m to aggressive b e h a v i o r in this sort o f setting.

N o t all findings o f Study 1 w e r e replicated. In Study 2, narcis-

Narcissism • 3 3" ~ l Perceived

q threat • 24* d

"1 Aggression

Figure 3. Threat as a mediator between narcissism and direct aggression for participants who received a negative evaluation. *p < .05.

DOES SELF-LOVE OR SELF-HATE LEAD TO VIOLENCE? 227

sism did not have a significant effect on aggression in response to the good evaluation. Put another way, in Study 1 narcissists were aggressive even toward the person who praised them, but this effect did not replicate in Study 2.

Study 2 was not confined to replication. We included an aggressive target manipulation that permitted comparison be- tween direct and displaced aggression. Narcissism showed no relation to aggression toward a third person (who had not deliv- ered an evaluation). Combined with the failure to replicate greater aggression by narcissists toward the positive evaluator, this finding suggests that narcissists are fairly selective and spe- cific in their aggression. The seeming implication of Study 1 that narcissism increases aggression toward everyone should probably be discarded.

These results can be interpreted in the following way. Narcis- sism did indeed enhance the tendency to respond aggressively to a bad evaluation, and so narcissists' aggression was high toward the source of this evaluation (but not toward anyone else). When someone praised the narcissist, on the other hand, the narcissist did not show elevated aggression. Narcissism did not alter the level of aggression toward a new person. The most consistent interpretation is that narcissists were exceptionally aggressive toward anyone who attacked or offended them. Oth- erwise, their aggression did not differ from that of other people.

Our mediation analysis confirmed that perceptions of threat did mediate between narcissism and aggression. The more a narcissist perceived the bad evaluation as threatening, the more aggressively he or she behaved. People who scored low in nar- cissism, in contrast, were less likely to perceive the evaluation as threatening, and these lower perceptions of threat led to lower aggression.

G e n e r a l D i s c u s s i o n

The present investigation has confirmed important links be- tween self-appraisal and aggression. Some theorists have pro- posed that people with low self-esteem would be most likely to lash out in response to an ego threat because the unflattering evaluation reminds them of their personal flaws and faults or because their low self-esteem makes them unable to tolerate the prospect of losing any of it. Others might suggest that low self- esteem would cause an increase in aggression regardless of ego threat. Our results contradict such views.

Instead, it appears that people who are emotionally invested in grandiose self-views are the most aggressive, particularly in response to an esteem threat. In both studies, we found that narcissism combilaed with ego threat yielded the highest levels of aggression. The combination of narcissism and ego threat was the primary focus of our investigation. Thus, the most aggressive responding in both studies was found among narcissists who were attacking someone who had given them a bad evaluation. These people were significantly more aggressive than would be predicted simply by adding any broad (main) effects of narcis- sism and ego threat. Moreover, the mediation analyses of Study 2 confirmed that degree of perceived threat determined level of aggression.

Our results also shed light on whether narcissists are more hostile and aggressive in general than other people. Although some of our results supported that view, others did not. People

with high narcissism scores were slightly more aggressive than other people toward someone who had praised them in Study 1, but this effect was not replicated in Study 2. Moreover, Study 2 added measures of displaced aggression, which is particularly relevant to the question of whether narcissists are aggressive in general. There were no significant correlations between narcis- sism and aggression toward a new, third person. Even if the narcissist had received an insulting evaluation from one person, he or she did not become exceptionally aggressive (i.e., any more than nonnarcissists) toward a different person.

Thus, our data suggest that aggression by narcissists is an interpersonally meaningful and specific response to ego threat. Narcissists became exceptionally aggressive toward a person who had given them a negative, insulting evaluation. They were not, generally, indiscriminately aggressive toward all other peo- ple or in all situations, nor did they show elevated tendencies to engage in displaced aggression. (That is, even receiving an insult did not make narcissists aggressive toward anyone except the person who delivered the insult.) Rather, our results suggest that narcissists mainly want to punish or defeat someone who has threatened their highly favorable views of themselves.

The present conclusions would be much broader if they sug- gested that all favorable views of self (i.e., high self-esteem per se) contributed to increased aggression. We did not find any such effect, however. Self-esteem yielded no significant main effects or interactions in either study.

Although nonsignificant findings are inherently difficult to interpret, we believe that our null results regarding self-esteem (combined with the rather surprising lack of other published findings on self-esteem and aggression) at least create serious doubt that self-esteem has any direct relation to aggression. We found significant effects for narcissism even when controlling for self-esteem, which suggests that the role of self-esteem is small at best. The null hypothesis that self-esteem is completely irrelevant to aggression cannot be confidently rejected on the basis of our data.

The failure of self-esteem to predict aggression in the present work cannot be attributed to a weakness in our measures of aggression, because we did find significant effects of other vari- ables (narcissism, ego threat, sex, and aggression target) on those measures. It cannot be attributed to the timing of the self- esteem measure, because the two experiments administered the self-esteem measure at different times. It cannot be attributed to one specific self-esteem scale, because the two experiments used two different (indeed, quite different) scales.

Our results do dovetail with those of Kernis et al. (1989) to suggest that self-esteem, as measured by the Rosenberg scale, does not have any such direct link to aggression and that the most aggressive people are likely to be in one subset of individu- als with highly favorable opinions of themselves. More gener- ally, our results support the growing suspicion that high self- esteem is a heterogeneous category (e.g., Schneider & Turkat, 1975). There are sound theoretical reasons for supposing that it is. High self-esteem is defined simply as having a favorable opinion or evaluation of oneself. This favorable view may be well founded in objective reality and may constitute an accurate appreciation of one's good traits, or it may be a highly dubious sense of personal superiority that is inflated well beyond what the facts would justify. It may be stable and largely impervious

2 2 8 BUSHMAN AND BAUMEISTER

to evaluations by others, o r it m a y demand frequent confirmation and validation b y other p e o p l e and b e p r o n e to fluctuate in response to daily events. This heterogeneity seems to u n d e r m i n e the usefulness o f s e l f - e s t e e m per se for predicting aggression.

M o r e generally, o u r results d o not indicate that aggression flows directly f r o m any f o r m o f self-regard. Neither a chronic pattern o f l o w s e l f - e s t e e m nor a broadly f a v o r a b l e v i e w o f s e l f p r o d u c e d h i g h levels o f aggression in either study. E v e n narcis- sism did not lead to elevated aggression overall; rather, the a g g r e s s i o n o f narcissists o c c u r r e d mainly in direct response to c r i t i c i s m and insult.

We b e g a n this w o r k with the hypothesis that threatened e g o - t i s m w o u l d lead to aggression ( B a u m e i s t e r et al., 1996). In that connection, what does it m e a n to c o n c l u d e that narcissism predicts aggres s ion ( i n r e s p o n s e to t h r e a t ) , whereas s e l f - e s t e e m does not? O n e w a y to answer this is to note that the c o n c e p t o f e g o t i s m has a dual meaning, and high s e l f - e s t e e m and narcis- sism capture the t w o different versions o f it. H i g h s e l f - e s t e e m means thinking w e l l o f oneself, whereas narcissism i n v o l v e s passionately wanting to think w e l l o f oneself. The present find- ings suggest that it is the latter ( e m o t i o n a l and motivational) sense o f e g o t i s m that is decisive for aggression. T h e p r e e m i - n e n c e o f the e m o t i o n a l and motivational pattern was anticipated b y Kernberg ( 1 9 7 5 ) , w h o s e w o r k on narcissism c o n c l u d e d by stating, " t h e r e f o r e , the ultimate nature o f narcissism . . . is dependent upon the d e v e l o p m e n t o f affect d i s p o s i t i o n s " (pp. 3 4 0 - 3 4 1 ). Recent findings by E m m o n s ( 1 9 8 7 ) , M o r f and Rho- dewalt ( 1 9 9 3 ) , and R h o d e w a l t and M o r f ( i n p r e s s ) have like- w i s e e m p h a s i z e d the affective aspect o f narcissism, especially the proneness to feel angry and hostile in r e s p o n s e to criticism.

In plainer terms, it is not so much the p e o p l e w h o regard t h emselves as superior beings w h o are the most dangerous but, rather, those w h o have a strong desire to r e g a r d themselves as superior beings. S o m e p e o p l e m a y be able to brush o f f c r i t i c i s m easily, j u s t as others m a y v i e w it as valid and w e l l deserved, and neither response m a y produce aggression. In contrast, peo- ple w h o are p r e o c c u p i e d with validating a grandiose self-image apparently find c r i t i c i s m highly upsetting and lash o u t against the source o f it. Thinking w e l l o f o n e s e l f is not inherently p r o n e to lead to a g g r e s s i o n - - e v e n in response to c r i t i c i s m and in- s u i t s - - b u t wanting to think w e l l o f o n e s e l f m a y w e l l be.

R e f e r e n c e s

Aiken, L. S., & West, S.G. ( 1991 ). Multiple regression: Testing and interpreting interactions. Newbury Park, CA: Sage.

Anderson, C. A., & Bushman, B. J. (1997). External validity of "triv- ial" experiments: The case of laboratory aggression. General Psychol- ogy Review, 1, 19-41.

Baron, R. A., & Richardson, D. (1993). Human aggression. New York: Plenum.

Baumeister, R. E, Heatherton, T. E, & Tice, D. M. (1993). When ego threats lead to self-regulation failure: Negative consequences of high self-esteem. Journal of Personality and Social Psychology, 64, 141- 156.

Baumeister, R. E, Smart, L., & Boden, J. M. (1996). Relation of threat- ened egotism to violence and aggression: The dark side of high self- esteem. Psychological Review, 103, 5-33.

Berkowitz, L. (1989). Frustration-aggression hypothesis: Examination and reformulation. Psychological Bulletin, 106, 59-73.

Bemstein, S., Richardson, D., & Hammock, G. (1987). Convergent and discriminant validity of the Taylor and Buss measures of physical aggression. Aggressive Behavior, 13, 15-24.

California Task Force to Promote Self-Esteem and Personal and Social Responsibility. (1990). Toward a state of self-esteem. Sacramento: California State Department of Education.

Cohen, J., & Cohen, P. (1983). Applied multiple regression/correlation analysis for the behavioral sciences ( 2nd ed.). Hillsdale, N J: Erlbaum.

Dollard, J., Doob, L., Miller, N., Mowrer, O., & Sears, R. (1939). Frustration and aggression. New Haven, CT: Yale University Press.

Emmons, R.A. (1984). Factor analysis and construct validity of the Narcissistic Personality Inventory. Journal of Personality Assessment, 48, 291-300.

Emmons, R. A. (1987). Narcissism: Theory and measurement. Journal of Personality and Social Psychology, 52, 11 - 17.

Fleming, J. S., & Courtney, B. E. (1984). The dimensionality of self- esteem II: Hierarchical facet model for revised measurement scales. Journal of Personality and Social Psychology, 46, 404-421.

Geen, R.G. (1990). Human aggression. Pacific Grove, CA: Brooks/ Cole.

Giancola, P. R., & Zeichner, A. (1995). Construct validity of a competi- tive reaction-time aggression paradigm. Aggressive Behavior, 21, 199-204.

Jaccard, J., Turrsi, R., & Wan, C. K. (1990). Interaction effects in multi- ple regression. Newbury Park, CA: Sage.

Janis, I. L., & Field, P. B. (1959). Sex differences in factors related to persuasibility. In C. I. Hovland & I. L. Janis (Eds.), Personality and persuasibility (pp. 55-68, 300-302). New Haven, CT: Yale Univer- sity Press.

J6reskog, K. G., & S6rbom, D. ( 1993 ). Windows L1SREL 8.12. Chicago: Scientific Software.

Kernberg, (3, (1975). Borderline conditions and pathological narcis- sism. New York: Jason Aronson.

Kernis, M. H. (1993). The roles of stability and level of self-esteem in psychological functioning. In R. Baumeister (Ed.), Self-esteem: The puzzle of low self-regard (pp. 167 - 182 ). New York: Plenum.

Kernis, M. H., Cornell, D. P., Sun, C. R., Berry, A., & Harlow, T. ( 1993 ). There's more to self-esteem than whether its high or low: The impor- tance of stability of self-esteem. Journal of Personality and Social Psychology, 65, 1190-1204.

Kernis, M. H., Grannemann, B. D., & Barclay, L. C. (1989). Stability and level of self-esteem as predictors of anger arousal and hostility. Journal of Personality and Social Psychology, 56, 1013-1022.

Kirschner, D. (1992). Understanding adoptees who kill: Dissociation, patricide, and the psychodynamics of adoption. International Journal of Offender Therapy and Comparative Criminology, 36, 323-333.

Kohut, H. (1971). The analysis of the self. New York: International Universities Press.

Long, D. E. (1990). The anatomy of terrorism. New York: Free Press. Millon, T. (1981). Disorders of personality. New York: Wiley. Morf, C. C., & Rhodewalt, E (1993). Narcissism and self-evaluation

maintenance: Explorations in object relations. Personality and Social Psychology Bulletin, 19, 668-676.

Neter, J., Wasserman, W., & Kutner, M. H. (1990). Applied linear statis- tical models (3rd ed.). Homewood, IL: Irwin.

Oates, R. K., & Forrest, D. (1985). Self-esteem and early background of abusive mothers. Child Abuse and Neglect, 9, 89-93.

Raskin, R. N., & Hall, C. S. (1979). A narcissistic personality inventory. Psychological Reports, 45, 590.

Raskin, R., Novacek, J., & Hogan, R. (1991). Narcissistic self-esteem management. Journal of Personality and Social Psychology, 60, 911 - 918.

Raskin, R., & Terry, H. (1988). A principal-components analysis of the Narcissistic Personality Inventory and further evidence of its construct

DOES SELF-LOVE OR SELF-HATE LEAD TO VIOLENCE? 229

validation. Journal of Personality and Social Psychology, 54, 8 9 0 - 902.

Rhodewalt, E, Madrian, J. C., & Cheney, S. (1998). Narcissism, self- knowledge, organization, and emotional reactivity: The effect of daily experience on self-esteem and affect. Personality and Social Psychol- ogy Bulletin, 24, 7 5 - 8 7 .

Rhodewalt, E, & Morf, C. C. (1995). Self and interpersonal correlates of the Narcissistic Personality Inventory: A review and new findings. Journal of Research in Personality, 29, 1-23.

Rhodewalt, E, & Morf, C. C. (1998). On self-aggrandizement and anger: A temporal analysis of narcissism and affective reactions to success and failure. Journal o f Personality and Social Psychology, 74, 6 7 2 - 685.

Rosenberg, M. ( 1965 ). Society and the adolescent self-image. Princeton, NJ: Princeton University Press.

Schneider, D. J., & Turkat, D. (1975). Self-presentation following suc- cess and failure: Defensive self-esteem models. Journal of Personal- ity, 43, 127-135.

Schoenfeld, C. G. (1988). Blacks and violent crime: A psychoanalyti- cally oriented analysis. Journal of Psychiatry and Law, 16, 269-301.

Tangney, J. E (1995). Shame and guilt in interpersonal relationships.

In J. Tangney & K. Fischer (Eds.), The self-conscious emotions (pp. 114-139). New York: Guilford Press.

Tangney, J.P., Wagner, P.E., Fletcher, C., & Gramzow, R. (1992). Shamed into anger? The relation of shame and guilt to anger and self- reported aggression. Journal of Personality and Social Psychology, 62, 669-675.

Taylor, S. P. (1967). Aggressive behavior and physiological arousal as a function of provocation and the tendency to inhibit aggression. Journal of Personality, 35, 297-310.

Tedeschi, J. T., & Felson, R. B. (1994). Violence, aggression, and coer- cive actions. Washington, DC: American Psychological Association.

Watson, P. J., Grisham, S. O., Trotter, M. V., & Biderman, M. D. (1984). Narcissism and empathy: Validity evidence for the Narcissistic Person- ality Inventory. Journal of Personality Assessment, 48, 301-305.

Wiehe, V. R. (1991). Perilous rivalry: When siblings become abusive. Lexington, MA: Heath/Lexington Books.

Wink, P. ( 1991 ). Two faces of narcissism. Journal o f Personality and Social Psychology, 61, 590-597.

R e c e i v e d F e b r u a r y 18, 1997 R e v i s i o n r e c e i v e d J u n e 23, 1997

A c c e p t e d O c t o b e r 2, 1997 •