Review Copyright 1996 by the American Psychological Association, Inc. 996, VoL 103, No. I, 5-33 0033-295X/96/$3.00 IP•Chological Relation of Threatened Egotism to Violence and Aggression: The Dark Side of High Self-Esteem Roy E Baumeister Laura Smart Case Western Reserve University University of Virginia Joseph M. Boden Case Western Reserve University Conventional wisdom has regarded low self-esteem as an important cause of violence, but the oppo- site view is theoretically viable. An interdisciplinary review of evidence about aggression, crime, and violence contradicted the view that low self-esteem is an important cause. Instead, violence appears to be most commonly a result of threatened egotismwthat is, highly favorable views of self that are disputed by some person or circumstance. Inflated, unstable, or tentative beliefs in the self's superi- ority may be most prone to encountering threats and hence to causing violence. The mediating process may involve directing anger outward as a way of avoiding a downward revision of the self- concept. Only a minority of human violence can be understood as ra- tional, instrumental behavior aimed at securing or protecting ma- terial rewards. The pragmatic futility of most violence has been widely recognized: Wars harm both sides, most crimes yield little financial gain, terrorism and assassination almost never bring about the desired political changes, most rapes fail to bring sexual pleasure, torture rarely elicits accurate or useful information, and most murderers soon regret their actions as pointless and self- defeating (Ford, 1985; Gottfiedson & Hirschi, 1990; Groth, 1979; Keegan, 1993; Sampson & Laub, 1993; .Scm'ry, 1985). What drives people to commit violent and oppressive actions that so of- ten are tangential or even contrary to the rational pursuit of mate- rial self-interest? This article reviews literature relevant to the hy- pothesis that one main source of such violence is threatened ego- tism, particularly when it consists of favorable self-appraisals that may be inflated or ill-founded and that are confronted with an external evaluation that disputes them. The focus on egotism (i.e., favorable self-appraisals) as one cause of violent aggression runs contrary to an entrenched body of wisdom that has long pointed to low self-esteem as the root of violence and other antisocial behavior. We shall examine the arguments for the low self-esteem view and treat it as a rival hypothesis to our emphasis on high self-esteem. Clearly, there Roy E Baumeister and Joseph M. Boden, Department of Psychology, Case Western Reserve University; Laura Smart, Department of Psy- chology, University of Virginia. Joseph M. Boden is now at the Department of Psychology, University of Virginia. We wish to thank Dianne Tice, Brad Bushman, Jim Overholser, and Sandra Russ for comments on a draft of this article. Roy E Baumeister also gratefully acknowledges Research Grants MH 43826 and MH 51482 from the National Institute of Mental Health. Correspondence concerning this article should be addressed to Roy E Baumeister, Department of Psychology, Case Western Reserve Uni- versity, 10900 Euclid Avenue, Cleveland, Ohio 44106-7123. Electronic mail may be sent via Internet to rfb2@po.cwru.edu. are abundant theoretical and practical implications that attend the question of which level of self-esteem is associated with greater violence. The widely publicized popular efforts to bol- ster the self-esteem of various segments of the American popu- lation in recent decades (e.g., see California Task Force, 1990) may he valuable aids for reducing violence if low self-esteem is the culprit--or they may be making the problems worse. Indeed, if high self-esteem is a cause of violence, then the im- plications may go beyond the direct concern with interpersonal harm. Many researchers share the opinion that high self-esteem is desirable and adaptive and can even he used as one indicator of good adjustment (e.g., Heilbrun, 1981; Kahle, Kulka, & Klingel, 1980; Taylor; 1989; Taylor & Brown, 1988; Whifley, 1983), but this one-sidedly favorable view of egotism would have to be quali- fied and revised: Favorable impressions of oneself may not he an unmitigated good from the perspective of society if they lead to violence. In our view, the benefits of favorable self-opinions accrue primarily to the self, and they are if anything a burden and poten- tial problem to everyone else. Hence the widespread norms con- demning conceit and arrogance, as well as the tendency to shift toward modesty when in the company of friends (Tice, Muraven, Buffer, & Stillwell, 1994). E. Anderson (1994) recently even sug- gested that self-esteem among youth gangs and similar groups con- forms to a zero-sum pattern, which means that any increment in stature, respect, or prestige of one person detracts from what is available for everyone else. Although some researchers favor narrow and precise con- cepts of self-esteem, we shall use the term in a broad and inclu- sive sense. By self-esteem we mean simply a favorable global evaluation of oneself. The term self-esteem has acquired highly positive connotations, but it has ample synonyms the connota- tions of which are more mixed, including pride, egotism, arro- gance, honor, conceitedness, narcissism, and sense of superior- ity, which share the fundamental meaning of favorable self-eval- uation. A related set of concepts refers to favorable evaluations of the self by others, including prestige, admiration, public es- BAUMEISTER, SMART, AND BODEN teem, and respect. Favorable evaluations are also implicit in lik- ing and loving, although those terms have additional meanings. Of particular importance for the present review is that our de- liberately broad usage of the term self-esteemis not limited to the direct results of validated trait measures of self-esteem (although we pay close attention to such measures when available). To reduce confusion, we shall favor the term egotism to refer both to favorable appraisals of self and to the motivated preference for such favorable appraisals, regardless of whether they are valid or inflated. Any assumption or belief that one is a superior being, or any broadly favorable assessment of self (especially in comparison with other people ), is relevant. Thus, in brief, the purpose of this article is to understand how self-appraisals are related to interpersonal violence. We hasten to add that we are not proposing a general theory of violence or aggression, and we assume that many aggressive acts may have little or no relation to self-esteem. Moreover, when self-apprais- als are involved, they may be only one of several factors, and so we are not asserting that other causes become irrelevant or secondary. The intent is merely to understand how self-apprais- als affect violence in those cases in which they are involved. There do seem to be many such cases. Traditional View: Low Self-Esteem Causes Violence A long tradition has regarded low self-esteem as a powerful and dangerous cause of violence. This view seems to be so widely and uncritically accepted that it is often casually asserted in the absence of evidence and even in the presence of appar- ently contrary evidence. When reading the literature for this review, we repeatedly found cases in which researchers summa- rized observations that depicted aggressors as egotistical and ar- rogant, but then added the conventional supposition that these individuals must be suffering from low self-esteem. (Hence we shall in some cases cite authors in this section as arguing in favor of low self-esteem but shall then later cite their empirical obser- vations as contradicting it.) One does not have to look far to find examples of the asser- tion that low self-esteem causes violence. E. Anderson (1994) recently cited low self-esteem as a persistent cause of the vio- lence among youth gangs. Similarly, Jankowski ( 1991 ) referred to "self-contempt" of gang members as a cause of violence. Renzetti (1992) said that the jealousy and possessiveness that lead to domestic violence have generally been understood as re- suiting from low self-esteem. Staub (1989) cited as traditional the view that low self-esteem generally causes all manner of vi- olence, although he was careful not to endorse that conclusion himself and in fact supplied some contrary evidence. Gondolf (1985) noted that wife beaters have usually been characterized as having low self-esteem, although he pointed out that the evi- dence for this is largely indirect, namely from clinical case stud- ies of their victims (Walker, 1979). Long (1990) asserted that low self-esteem and feelings of inadequacy are prominent char- acteristics of most terrorists. MacDonald (1975) said that armed robbers "lack self-esteem" (p. 263). Wiehe ( 1991 ) said that a possible motive for sibling violence is "as a way to bolster or increase their low self-esteem" (p. 17). Kirschner (1992) claimed that several murderers (in this case, adoptees who had killed their adoptive fathers) suffered from low self-esteem and viewed themselves as bad. Levin and McDevitt ( 1993 ) casually mentioned low self-esteem as if it were commonly known to be an important cause of hate crimes. In other cases, the weaknesses and fallacies in the low self- esteem view are readily apparent. Thus, in one of the classic works on the psychology of violence, Toch (1969/1993) re- ferred to a "compensatory relationship between low self-esteem and violence" (pp. 133-134 ), and he suggested that people with low self-esteem turn violent as a way of gaining esteem. Yet Toch did not have any direct evidence of low self-esteem; he merely inferred that these men must suspect themselves of weakness because they seemed so concerned with refuting that impression. Alternative interpretations, particularly that they have highly favorable views of self that are threatened by disre- spectful others, are equally possible. Indeed, in the same passage in which Toch referred to the self-doubts and "sense of inade- quacy" of violent men, he also proposed that these same offenders had "exaggerated self-esteem" (p. 136), which is ob- viously the opposite assertion. He also said that such an individ- ual "demands unwarranted respect" (p. 136), which is close to our own argument of an inflated sense of deservingness (e.g., excessively high self-esteem). In short, Toch's explanation is in- ternally inconsistent. In another example, Oates and Forrest (1985) asserted that abusive mothers had low self-esteem. They based this conclu- sion on a purported measure of self-esteem that was actually a single item asking the mother whether she wished her child would grow up like herself; abusive mothers tended to give self- deprecating answers to this question. At the time of data collec- tion the mothers had all recently been referred for child abuse. Under those circumstances, it would seem almost mandatory to show some self-deprecation and to be hesitant about expressing the wish that one's child would follow in one's footsteps. To label that response as low self-esteem seems potentially misleading. Likewise, Schoenfeld (1988) proposed that the high crime rate among American Black people is clue to their low self-esteem. In his analysis, Blacks were reduced by slavery to a state of extremely low self-esteem. When slavery ended, this low self-regard was per- petuated by Jim Crow laws and, more recently, by the modem welfare system, which fosters helplessness and dependency. Thus, in Schoenfeld's view, low self-esteem is responsible for the high crime rate. Unfortunately this analysis suffers from several flaws. First, it does not fit the temporal shifts in crime rate among Blacks, which is now reaching its highest levels as slavery recedes farther and farther into the background. Second, as Crocker and Major's (1989) review showed, self-esteem levels among Blacks are now equal to or higher than the self-esteem levels of Whites. Third, it is far from certain that slaves bad low self-esteem; Patterson (1982) insisted that slaves did not simply internalize the unflattering views society held of them. Our review did not uncover any one definitive or authorita- tive statement of the theory that low self-esteem causes violence, so it is necessary for us to consider several possible versions of that theory. One view (and one that seems implicit in many writings) is that people who lack self-esteem hope to gain it by violent means, such as by aggressively dominating others. In this view, violence would be a technique of self-enhancement, in the sense that it is used as a means of increasing one's esteem. A long tradition has assumed that people with low self-esteem EGOTISM AND VIOLENCE must be strongly oriented toward self-enhancement, because they want to gain more of what they lack. The self-enhancement version of the low self-esteem view is internally plausible, but the accumulation of research findings has now rendered it untenable. The motivation to seek self- enhancement has been shown to be characteristic of people high (rather than low) in self-esteem, and in fact it appears to be weak or absent among people with low self-esteem (Baumeister, Tice, & Hutton, 1989; Tice, 1991, 1993). Indeed, people with low self-esteem appear to be ambivalent about rising in esteem, and they often avoid circumstances that might raise their self- esteem (De La Ronde &Swann, 1993; Swarm, 1987; Swarm, Griffin, Predmore, & Gaines, 1987). A similar contradiction can be found in recent work on the psy- chology of terrorism. Long ( 1990 )summarized what various writ- ers have concluded about the most common personality traits of terrorist individuals as including "low self-esteem and a predilec- tion for risk-taking" (p. 18). Long's explication of the nature of this low self-esteem seemed, ho~, to fit very closely what is known about high self-esteem. In Long's account, these individu- als "tend to place unrealistically high demands on themselves and, when confronted with failure, to raise rather than lower their aspi- rations" (p. 18). High serf-esteem is associated with higher aspira- tions than low self-esteem in general (e.g., Baumeister & Tice, 1985). The laarticular, ironic pattern of responding to failure by raising one's aspirations further was shown by McFarlin and Blas- covich ( 1981 ) to be characteristic of people with high self-esteem; Baumeist~ Heatherton, and Tice (1993) replicated that pattern and showed that it extended to increased risk-taking after failure or other ego threat. It may once have been plausible to think that people with low self-esteem would be prone to take risks and raise their aspirations alter failure, but those patterns have now been linked to high rather than low self-esteem. Thus, Long's purported evidence for low self-esteem among terrorists in fact seems to in- dicate a pattern of high serf-esteem. Another variation of the low self-esteem theory is based on the notion of a subculture of violence. This notion emerged in the late 1960s as one explanation for violence among stigma- tized minority populations. According to this view, members of these minority groups lacked access to the traditional or main- stream sources of self-esteem, so they formed communities in which aggressive behavior was an alternative source. The sub- culture of violence hypothesis has lost ground, however, as re- searchers have been unable to identify any community or sub- culture that places a positive value on violent acts (see Tedeschi & Felson, 1994). Yet another version would propose that all people desire to regard themselves favorably, and people with high self-esteem have satisfied this need and can ignore it, whereas it remains a focal concern of those with low self-esteem. In this view, high self-esteem ought to confer a kind of immunity to ego threats, because the person is so secure in his or her self-appraisal that nothing can diminish it. However, researchers have not found that most people with high self-esteem are so cheerfully in- different to insults, criticism, or disrespect. Indeed, the strong and sometimes irrational reactions of people with high self- esteem to negative feedback have been abundantly documented (Baumeister et al., 1989, 1993; Baumeister & Tice, 1985; Blaine & Crocker, 1993; McFarlin & Blascovich, 1981 ). Some studies suggest that people with high self-esteem are if anything more sensitive to criticism than people with low self-esteem (e.g., Schlenker, Soraci, & McCarthy, 1976; Shrauger &Lund, 1975). Toch ( 1969 / 1993) observed that many violent men seek out or manufacture situations in which their self-worth is chal- lenged, with the result being a violent confrontation. Because Toch espoused the low self-esteem view, we infer that he thought low self-esteem would be a factor that dictated such efforts. Pos- sibly Toch thought that people who lack self-esteem seek out such challenges as a way of gaining esteem. To us, ~ however, it seems implausible that people who hold low opinions of them- selves will seek out situations that will provide tests or other feedback. Low self-esteem would favor an avoidance of such feedback, for several reasons: These people want to protect themselves from bad feedback (Baumeister et al., 1989); they dislike and distrust flattering, enhancing feedback (Swann, 1987); and they are not strongly motivated to gain accurate feedback (Sedikides, 1993). Only the person with a highly fa- vorable opinion of self will be inclined to seek out risky situa- tions to prove his or her merit. Picking fights with dangerous individuals strikes us as a dubious strategy for gaining esteem, and it seems likely to appeal mainly to individuals with irratio- nally high confidence. There is one final and limited variation on the self-esteem view that appears to be more plausible than the others. Some causes of violence may have little to do with self-esteem, and as a result some people at any level of self-esteem may become aggressive. The combination of violent tendencies and low self- esteem might then exert an influence on choice of target. As we said, it would seemingly require high confidence to attack a powerful person, but when the target is seemingly weak and helpless the odds of success may seem quite high. Accordingly, people with low self-esteem may channel their violent tenden- cies into attacks on such weak and helpless targets. Men who attack women and adults who attack children might well have low self-esteem, not because low self-esteem causes violence, but because low self-esteem causes them to seek a victim who is unlikely to retaliate. On an a priori basis, therefore, domestic violence seems like the most promising milieu in which to find evidence of aggression by people who lack self-esteem. In summary, the view that low self-esteem causes violence has been widely asserted but rarely elaborated. Our efforts to reconstruct the theorizing behind the low self-esteem hypothe- sis have resulted in several versions, none of which is broadly satisfactory. Some are internally inconsistent, whereas others seem internally plausible on a priori grounds but run contrary to the accumulated evidence about self-esteem. The most viable view in our version saw low self-esteem not as a cause of vio- lence but as causing a preference for safe, helpless targets, sug- gesting that any violent tendencies that exist among people with low self-esteem will most likely be expressed in situations in which fear of retaliation is minimal. High Self-Esteem and Violence In contrast to the low self-esteem view, we propose that highly favorable self-appraisals are the ones most likely to lead to vio- In fairness to Toch, we have the benefit of several decades of research on self-esteem that was not available to him in 1969. BAUMEISTER, SMART, AND BODEN lence. As noted in the previous section, the traditional theories linking low self-esteem to violence suffer from ambiguities, in- consistencies, and contradictory empirical evidence. The oppo- site view therefore deserves consideration. There are some bases for suggesting that egotism could lead directly to violence. People who regard themselves as superior beings might feel entitled to help themselves to the resources of other, seemingly lesser beings, and indeed they might even ag- gress against these lesser beings without compunction, just as people kill insects or mice without remorse (Myers, 1980). Also, many violent episodes involve a substantial element of risk, and a favorable self-appraisal might furnish the requisite confidence to take such a chance. In plain terms, egotists might be more likely to assume that they will win a fight, and so they would be more willing to start it. Our main argument, however, does not depict self-esteem as an independent and direct cause of violence. Rather, we propose that the major cause of violence is high self-esteem combined with an ego threat. When favorable views about oneself are questioned, contradicted, impugned, mocked, challenged, or otherwise put in jeopardy, people may aggress. In particular, they will aggress against the source of the threat. In this view, then, aggression emerges from a particular discrep- ancy between two views of sell" a favorable self-appraisal and an external appraisal that is much less favorable. That is, people turn aggressive when they receive feedback that contradicts their favor- able views of themselves and implies that they should adopt less favorable views. More to the point, it is mainly the people who refuse to lower their self-appraisals who become violent. One major reason to suggest that violence may result from threatened egotism is that people are extremely reluctant to revise their self-appraisals in a downward direction. This assertion must he understood in the context of the research literature concerning the motivations that surround self-appraisals. This literature has been dominated by two somewhat conflicting hypotheses. One holds that people wish to hold maximally positive views of them- selves and so seek to enhance their self-appraisals whenever possi- ble (e.g., Darley & Goethals, 1980; Greenwaid, 1980; Schlenker, 1980; Taylor & Brown, 1988). The other is that people seek to maintain consistent self-appraisals and therefore seek to avoid changing their self-concepts at all (e.g., De La Ronde &Swann, 1993; Swarm, 1987). Although these two views make contradic- tory predictions in some settings, they agree emphatically that people are reluctant to change toward more unflattering views of themselves. The avoidance of loss of esteem is thus the clearest and presumably strongest pattern of self-concept motivation (Baumeister, 1993). Decreases in self-esteem are aversive for nearly everyone. The relevance of level of trait self-esteem to these two motives (enhancement and consistency) requires elaboration. First, consider the self-enhancement motive. People with favorable 'opinions of themselves have been shown to exceed those with low self-esteem in desire for self-enhancement (Baumeister et al., 1989; Tice, 1991, 1993). The quest for opportunities to prove oneself or to raise one's standing should therefore appeal mainly to people with high self-esteem. For example, a pattern of seeking out situations in which one's worth is challenged or disputed might strike a very confident person as a good chance to refute such threats and show oneself off to be a winner. In contrast, people with low self-esteem will probably tend to avoid such situations. Meanwhile, the orientation toward self-protection (against losing self-esteem) has been shown to be more characteristic of people with low rather than high self-esteem. On the surface, this seems to suggest a contradiction to our portrayal of aggres- sion as resulting from threatened egotism, because low rather than high self-esteem is associated with broad, chronic concern with avoiding loss of esteem. Yet this is misleading. Self-protec- tion characterizes the habitual orientation of people with low self-esteem because they are constantly concerned with avoid- ing situations that could result in a loss of esteem. People with high self-esteem do not show a strong self-protective orientation habitually because they do not anticipate that they will fail or lose esteem. When threats to esteem do arise, however, people with high self-esteem respond in ways that are often drastic and irrational (see Blaine & Crocker, 1993, for review; see also Baumeister et al., 1993; Baumeister & Tice, 1985; McFarlin & Blascovich, 1981 ). Thus, people with high self-esteem do in- deed hate to lose esteem. Most of the time they scarcely think about the possibility that they will lose esteem, and so it is only when a threat emerges that they become extremely defensive. Self-verification theory (Swarm, 1987) is based on the notion that people resist changes to their self-concepts. This motive to maintain consistent self-appraisals means that people who think somewhat poorly of themselves may resist favorable feed- back. People who think well of themselves, on the other hand, may be broadly receptive to favorable feedback because it largely confirms their self-appraisals. However, they will react quite strongly against unfavorable feedback. Thus, the self-enhancement and self-verification motives both predict that the strongest negative reactions to external feedback will arise when people who think well of themselves receive unflattering feedback. In our view, that is precisely the discrepancy most likely to lead to violence--when favorable views of self are met with external, less favorable appraisals. To elaborate this basic theoretical position, we shall proceed as follows. First, we shall examine some likely moderators of the link between egotism and violence. These moderators are based on the assumption that anything that increases either the fre- quency or the subjective impact of discrepancies between favor- able self-appraisals and external ego threats will increase the likelihood of violence. Second, we shall examine the role of affect as mediating between threatened egotism and aggressive behavior. Last, we shall examine the interpersonal context of threatened egotism. Favorable and Inflated Self-Appraisals If threatened egotism causes aggression, then whatever views of self encounter the greatest number of threats should be the ones most commonly associated with violence. On an a priori basis, it would seem that the higher the self-esteem, the greater the range of feedback that would be seen as threatening. Thus, for example, people who believe themselves to be among the top 10% on any dimension may be insulted and threatened when- ever anyone asserts that they are in the 80th, 50th, or 25th per- centile. In contrast, someone with lower self-esteem who re- gards himself or herself as being merely among the top 60% would only be threatened by the feedback that puts him or her EGOTISM AND VIOLENCE at the 25th percentile; indeed, feedback that puts him or her at the 80th percentile, which was threatening and insulting to the person with very high self-esteem, might even be received as praise by someone with much lower self-esteem. In short, the more favorable one's view of oneself, the greater the range of external feedback that will he perceived as unaccept- ably low. To the extent that violence arises from threats to self- esteem (in the sense of receiving external feedback that evaluates one less favorably than one's self-evaluation), violence should he more common among people with high self-esteem. One could dispute the aforementioned reasoning by arguing that feedback is actually very selectively distributed. That is, maybe the person who regards himself or herself as at the 90th percentile is in fact so competent that he or she will almost never receive the 25th percentile feedback. 2 To the extent that feed- back tends to cluster around accurate appraisal of one's true abilities (including social feedback, i.e., being seen accurately by others), there might be no difference in the frequency of threatening feedback received by people at any level of self- appraisal. There would, however, be one major exception to the argu- ment that the general accuracy of feedback would counteract the excess vulnerability of high self-esteem to threat. Favorable views of self that are unwarranted, exaggerated, or ill-founded would be especially prone to disconfirmation by accurate feed- back. Whenever people's self-appraisals are more favorable than their objective qualities would warrant, the result may be a pervasive vulnerability to threatening feedback. And how often does that happen? Evidence suggests that people with favorable self-opinions frequently benefit from distortion, selective per- ception, or exaggeration. The pervasiveness of such inflated views of self, particularly among people with high self-esteem, has been well documented (Taylor & Brown, 1988). Thus, to the extent that feedback tends to be accurate instead of random, its subjective impact will depend on whether the re- cipient's self-appraisals are accurate or inflated. Accurate feed- back will tend to confirm self-appraisals, including favorable ones, if they were realistic to begin with. Accurate feedback will, however, tend to disconfirm self-appraisals that are unrealisti- cally positive. The implication is that unrealistically positive self-appraisals will increase the frequency with which external ego threats are encountered. Inflated views of self should there- fore increase the frequency of violence. Instability, Uncertainty, and Evaluative Dependency In the section on low self-esteem, we mentioned the hypothesis that people with favorable self-appraisals would he indifferent to bad feedback because it would not threaten them. As we said, re- searchers have not generaUy found many people who are immune to criticism, but there may be a kernel of truth in that reasoning. Undoubtedly there are individual differences and situational vari- ations in the degree to which people care about the opinions of others. Such variations would presumably alter the subjective im- pact of bad feedback and ego threats, and as a result they would moderate the degree of aggressive response. Unlike inflated self- appraisals, which increase the frequency with which one encoun- ters ego threats, these variables increase the importance of ego threats and hence magnify the hostile response. One factor that seems likely to moderate the impact of external appraisals is the degree of certainty of the relevant self-appraisal. Someone who is certain of having a particularly good trait may be relatively less affected by contradictory feedback as compared with someone who is less certain. Accordingly, those people with uncertain but positive views of self may also be the ones most prone to elicit defensive responses to ego threats. Many views of identity formation have emphasized that people require the validation of others (e.g., Baumeiste~ 1986; Cooley, 1902/1964; James, 1890/1950; Mead, 1934; Schlenket; 1980, 1986). Probably the most thorough explication of how uncer- tainty of self-appraisal is linked to reliance on external validation was provided in the work of Wicklund and Gollwitzer (1982). In multiple studies, they showed that people vary in the degree to which they are motivated to have others confirm their identity claims. Although nearly everyone requires some social validation, some people become heavily dependent on it, whereas others re- quire much less, and these variations can be partly explained on the basis of having acquired a stock of symbolic affirmations of the self (which boost certainty and what Wicklund and Gollwitzer called "completeness"). Once a person has accumulated abun- dant trappings of success, for example, he or she may not feel any urgent need to acquire more, and a stray pejorative remark can be easily shrugged off. In contrast, a person whose claim to the de- sired identity is tentative or incomplete may feel a frequent need to gain validation by others and may be acutely sensitive to slighting remarks. Thus, people who feel incomplete and who consequently feel a pervasive need for social validation of their favorable self- conceptions are more susceptible to ego threats. Another relevant pattern would be stability of self-appraisal. Some self-appraisals are relatively stable over time, which sug- gests that they are not greatly affected by daily events. In con- trast, other self-appraisals fluctuate more widely from day to day. Kernis (1993) and his colleagues showed that global levels of self-esteem fluctuate more widely in some people than in oth- ers. It seems likely that people who have unstable self-appraisals will tend to become sensitive and defensive, and so bad feedback will produce a quicker and stronger reaction in them than in people with stable self-appraisals. Shifts toward more negative self-appraisals generally bring anxiety, depression, anger, and other forms of negative affect (Heatherton & Polivy, 1991; Hig- gins, 1987). People wish to avoid these unpleasant states, and so people with unstable self-esteem should be strongly moti- vated to ward off any event that might potentially lower their self-esteem. To such an individual, bad feedback or criticism from other people would almost certainly contain the risk of bringing one's self-esteem down, and so one may react strongly to any hint of such ego threats. In contrast, people whose self- esteem remains the same regardless of what happens would have much less reason to fear criticism or other bad feedback. The result would be that people with unstable high self-esteem might well become violent in response to even seemingly minor or trivial threats to self-esteem. Consistent with this reasoning, Kernis, Cornell, Sun, Berry, and Harlow (1993) found that peo- ple with high but unstable self-esteem were most prone to re- spond defensively to unfavorable feedback. This section has reviewed several factors that would likely in- crease the magnitude and subjective impact of ego threats-- 2 If only life were so consistently fair! BAUMEISTER, SMART, AND BODEN instability, certainty, and evaluative dependency on others. The common theme appears to be that favorable self-appraisals that are insecurely held may be most vulnerable to ego threats. Al- though we have said that these self-appraisals will lead to vio- lence by increasing the magnitude rather than the frequency of ego threats, there could be an apparent increase in frequency too: Instances of minor, slight, or minimal bad feedback could elicit strong reactions from such insecure egotists, whereas se- cure egotists would dismiss such events as too trivial to be worth a response. The Mediating Role of Affect Thus far we have proposed that violence tends to follow from a certain pattern of discrepant appraisals (i.e., favorable self- appraisals and unfavorable external appraisals) and that what- ever increases the frequency or subjective impact of such dis- crepancies will increase aggression. Yet it is a long step from inconsistent appraisals to violent action. One crucial interven- ing variable may be affect. Hence it may be helpful to expand our position to say that encountering a discrepancy between public and private self-appraisals will engender aversive arousal states, and these in turn foster aggression. Is negative affect an adequate explanation for aggression? For many years, theorizing was influenced if not dominated by the view that frustration was an essential cause (Dollard, Doob, Miller, Mowrer, & Sears, 1939). Although it was undeniable that many frustrated people become aggressive, contrary find- ings did gradually accumulate. Some aggression did not seem to follow from frustration, and some frustrations did not result in aggression. In a sweeping reformulation of this research, Berkowitz (1989) concluded broadly that aggression results from negative affective states in general. He proposed that frus- tration and anger had been overemphasized. Any negative affect could cause aggression. Although Berkowitz (1989) made a compelling case for ex- panding aggression theory beyond a narrow focus on feelings of anger and frustration, there is not yet sufficient evidence avail- able to conclude that all states of negative affect can cause ag- gression (as he noted). Indeed, Baron (1976) showed that exposing participants to a pitiable injury victim reduced subse- quent aggressiveness, and he concluded that empathic pity is incompatible with aggressive impulses. Likewise, guilt may of- ten inhibit aggressive acts (e.g., Baumeister, Stillwell, & Heath- erton, 1994). Meanwhile, there is little to suggest that sadness leads to aggression. The most appropriate conclusion at present seems to be that some forms of negative affect can produce or increase aggressive tendencies. Meanwhile, it is relatively straightforward to suggest that ego threats can produce negative affect. Decreases in state self-esteem often lead to negative affect (Heatherton & Polivy, 1991 ). Discov- ering that one falls short of ideals or violates one's proper standards of behavior produces various negative affect states (Higgins, 1987). l_eary, Tambor, Terdal, and Downs (1995) recently pro- vided evidence linking self-esteem to interpersonal appeal and sta- tus, and interpersonal rejection or exclusion is a central cause of anxiety (Baumeister & Tice, 1990), so it seems fair to expect that decreases in self-esteem will bring anxiety too. Despite this seemingly straightforward pathway from ego threats to lowered self-esteem to negative affect to aggression, however, careful inspection suggests several potential problems and inconsistencies. Higgins (1987) proposed that perceiving oneself as falling short of ideals should engender low-arousal emotions such as dejection and sadness, and there is little evi- dence that such emotions lead to aggression. Moreover, if drops in self-esteem were responsible for the negative affect that re- suited in aggression, then one would have to make the strong prediction that low self-esteem (if only as a temporary state) was a crucial factor. Last, the view that people revise their self- appraisals readily in response to external threats runs contrary to considerable evidence that people resist such downward revi- sions (Greenwald, 1980; Swarm, 1987; Taylor & Brown, 1988). We propose, instead, that when favorable views of self are confronted with unflattering external feedback, the person faces a choice point. The affective response will depend on which path is chosen. One path is to accept the external appraisal and revise one's self-esteem in a downward direction. Sadness, anx- iety, and dejection might well result from such a course. In con- trast, the other path is to reject the external appraisal and up- hold one's more favorable self-appraisal. The confluence of self- consistency and self-enhancement motives would suggest that this is generally the preferred response. In such a case, the per- son would infer that the external evaluation is mistaken and undeserved, and he or she may well develop anger or other neg- ative affect toward the source of that evaluation. The hypothesis of a choice point was anticipated to some ex- tent by Berkowitz's (1989) observation that many bad experi- ences lead to a choice between fight or flight reactions--that is, between a self-assertive, aggressive response and one of defeated withdrawal. It also suggests how some seemingly contradictory findings and implications can be integrated. Thus, research on shame suggests on the one hand that this global feeling of being a despicable person often leads to a tendency to withdraw or hide from others (Lewis, 1971; Tangney, 1991, 1992). On the other hand, there is evidence that shame-prone people tend to externalize blame and become angry and aggressive toward oth- ers (Tangney, Wagner, Barlow, Marschall, & Gramzow, 1994; Tangney, Wagner, Fletcher, & Gramzow, 1992), and clearly an- gry aggression is a very different response that seems incompat- ible with social withdrawal. It may be, however, that a poten- tially shame-inducing experience causes some people to accept the unflattering evaluation and withdraw, whereas others re- spond by refusing to accept the evaluation and by becoming angry toward the evaluator. A similar choice point is suggested by recent research find- ings about envy. Envy arises when someone else has what the envious person wanted, which can imply that oneself is less wor- thy and less deserving than the other (Salovey, 1991; Salovey & Rodin, 1984). Smith, Parrott, Ozer, and Moniz (1994) found that envy leads to hostility only if the person retains a favorable view of self as deserving the positive outcome, in which case the envied person's advantage is seen as unjust and unfair. In contrast, if the person accepts the implication and feels inferior to the envied person, then hostility does not ensue. Once again, then, the affective response to an ego threat depends on the self- appraisal, and the response that maintains a favorable self- appraisal leads to aggression. By this reasoning, then, aggression can be regarded as a crude technique of affect regulation. Meloy (1988) made this argu- ment in explicit detail for psychopaths, who engage in predatory EGOTISM AND VIOLENCE 1 1 violence to avoid a broad range of unwanted emotions. To avoid certain negative emotional states, such as shame, dejection, sad- ness, and disappointment with oneself, the person refuses to contemplate information that reflects unfavorably about the self. When others attempt to provide such unfavorable feed- back, the person becomes agitated and directs unpleasant emo- tions at them. By focusing on his or her hostility toward the evaluators, the person avoids the dismal cycle of accepting the feedback, revising his or her self-concept, and experiencing the dejected feelings about the self. This affective view dovetails well with the previous analysis of possible moderators. People whose favorable self-conceptions are inflated, uncertain, or unstable may become quite sensitive to unflattering feedback and may react with hostility. Indeed, this analysis has one further implication, which is that the hos- tile response may often seem wildly disproportionate to the ac- tual informational power of the external evaluation or even to any contemplated reduction in self-esteem. Because the angry, hostile response is essentially a means of preventing oneself from having to suffer through a depressing revision of self- appraisal, its function is largely anticipatory. Hence highly sen- sitive individuals may react with considerable hostility to seem- ingly minor ego threats. In other words, once a person becomes familiar with the emotional distress of losing self-esteem, he or she may become watchful for potential or incipient threats and may react strongly to what observers would regard as slight or trivial offenses. Interpersonal Context The last issue to consider is the interpersonal dimension. In most cases, violence is not a random eruption of intrapsychic forces but rather is directed toward a particular target in the context of some meaningful communications. Two interpersonal aspects stand out. First, aggression may be a meaningful and coercive response to the unflattering evalua- tion. We have proposed that aggression results from a discrep- ancy between a favorable self-appraisal and an unfavorable ex- ternal appraisal. The matter is not concluded simply because the recipient decides not to accept the unfavorable evaluation. Even after that decision is made, the person remains confronted by someone who is expressing a negative view (which is now seen as undeserved and unjust). By aggressing against the eval- uator, the person may accomplish several things, including pun- ishing the evaluator for the bad feedback, impugning the other's right to criticize, and discouraging that person (and others) from expressing similar evaluations in the future. Tedeschi and Felson (1994) recently argued that aggression should be recon- ceptualized as coercive behavior. In this context, a violent re- sponse may coerce the other person into withdrawing the bad evaluation. Second, a successful violent attack achieves a symbolic domi- nance over the other person, and so it affirms one's esteem to the extent of being superior to the victim. Violence may there- fore be one form of self-affirmation, which is a common re- sponse to ego threats (Steele, 1988). This response may help explain two otherwise puzzling patterns of aggression. One is the seeming logical irrelevance of violence to most ego threats. For example, someone who beats up someone who has insulted his intelligence does not provide any positive proof of intelli- gence, but self-affirmation theory emphasizes that people who feel their esteem threatened in one sphere often respond by as- serting positive qualities in another sphere (Steele, 1988; see also Baumeister & Jones, 1978; Greenberg & Pyszczynski, 1985). The other puzzling pattern concerns displaced aggres- sion. If aggression is understood as a communicative response to unfavorable feedback, then it would be illogical to aggress against a third person. But such displacement may become comprehensible as a way of asserting superiority over someone else, especially if the evaluator is an unsuitable target for aggression, The link between aggression and superiority may have evolu- tionaryroots. Certain pack animals develop stature hierarchies in which one's position in the hierarchy depends on which others one can defeat in a fight. Human history has certainly contained abundant episodes consistent with that pattern; indeed, in the transition from nomadic, barbarian life to civilization, the most common pattern was for the warriors to become the aristocracy, with the foremost fighters or battlefield leaders becoming the indi- vidual rulers (e.g., McNeill, 1982, 1991 ).3 Thus, in both evolu- tionary and cultural history, high stature has been finked to fighting, It is plausible that some aggressive resIxms~ derive from this deeply rooted impulse to achieve physical dominance over rivals rather than from some calculated response to discrepant self- appraisals. Although the analogy to pack hierarchies is clearly specula- tive, there is one relevant implication that deserves mention. In any small group hierarchy, the amount of prestige available is limited. (Other limited resources, such as material rewards, may also be involved, insofar as these are distributed in propor- tion to stature.) One can only gain at the expense of another. Hence under conditions of scarcity the negotiation of esteem may take on a zero-sum aspect. E. Anderson (1994) proposed that in poor communities in America, self-esteem does indeed conform to zero-sum patterns. Gaining esteem requires taking it away from others. This analysis greatly expands the range of acts that can constitute an ego threat. If the amount of self- esteem is fixed, then positive claims by one person are sufficient to constitute a threat to others. Thus, one does not have to crit- icize a person to threaten his or her self-esteem; merely making favorable claims about oneself is enough. This zero-sum aspect of esteem should mainly apply to small, fixed hierarchies, but some forms of it may be apparent even in a broad society in which the amount of available esteem is less obviously limited. Feather (1994) recently reviewed research on the "tall poppy" phenomenon, namely the seeming pleasure that people may derive from witnessing the downfall of highly successful people. That pleasure could well be linked to such a zero-sum esteem pattern, especially if highly successful people are perceived as unfairly hogging or hoarding esteem that would otherwise be available to many others. Still, it must be noted that Feather (1994) did not find the tall 3 This is a slight oversimplification of McNeilrs argument. The hunt leader tended to rule in hunting societies, and warriors soon emerged as rulers in early civilizations, but in between there may have been an in- terval during which peasant farmers lived in peace under near anarchy or loose social structures dominated by priests. Still, kingship and aris- tocracy were closely linked to leadership in war, which is the relevant point. BAUMEISTER, SMART, AND BODEN poppy effect to be widespread or robust. The zero-sum aspect Discrepancy of esteem-related violence may be limited to highly particular, Favorable between Negative circumscribed patterns, such as those in which there is some viewofself __~ internaland (_~ evaluation explicit sense of competition for a limited amount of stature. -unstable external by other(s) -inflated appraisals -uncertain Summary of High Self-Esteem Theory In summary, we propose that one major cause of violent re- sponse is threatened egotism, that is, a favorable self-appraisal I threatened that encounters an external, unfavorable evaluation. Factors egotism that increase the frequency or impact of such encounters will increase violence. In particular, unrealistically positive or in- flated views of self, and favorable self-appraisals that are uncer- tain, unstable, or heavily dependent on external validation, will be especially vulnerable to encountering such threats. Such threats often elicit anger and other negative affects when the per- son refuses to accept and internalize the unflattering evaluation. reject a~appraisal (If the person accepts the evaluation and revises his or her self- esteem downward, aggression will be less likely.) The anger and maintain lower the aggressive response typically occur in an interpersonal self-appraisal self-appraisal framework: They are most commonly directed at the source of the bad evaluation. Aggression serves to refute and prevent bad q/ evaluations as well as to constitute a means of achieving sym- bolic dominance and superiority over the other person. Figure 1 summarizes this theory. Review of Empirical Findings If reliable data on self-esteem levels of violent and nonviolent citizens were available, it would be relatively easy to resolve the question of who is most violent (although some theoretical questions would remain). Alternatively, if self-esteem had been routinely measured in laboratory studies of aggression, there would be at least one methodologically solid source of evidence. Unfortunately, neither of these is the case. Accordingly, it is nec- essary to look at a broad range of evidence about aggression, violence, oppression, and other forms of evil behavior and to consider carefully how self-esteem might be involved. In partic- ular, claims about the self made during violent incidents, or as- sumptions about the self that make violence possible, deserve close attention. The present review will survey literature on violence, encom- passing both traditional laboratory studies of aggression and prejudice by experimental psychologists and data from outside psychology, most notably criminology. Tedeschi and Felson (1994) noted the irony that aggression psychologists and crim- inologists rarely read each other's literatures despite common interests and despite the obvious value of converging evidence. Widom (1991) cited broad "agreement" that scholars "need to look beyond disciplinary boundaries" (p. 130) for problems such as family violence and child abuse. No single discipline in the social sciences can claim a monopoly on insights into violence. We shall begin by looking at efforts to predict violent, aggres- sive behavior from measures of egotism (including self-esteem and narcissism). Then we shall turn to the complementary strategy of looking at violent criminals, groups, and other ag- gressive individuals to ascertain how favorably they appraise themselves. toward source of toward self threat Figure 1. Schematic representation of the relation of threatened ego- tism to violent behavior. Self- Reported Hostile Tendencies We begin with survey studies that included both measures of self-esteem and measures of self-reported angry, violent, or hostile tendencies. Several studies have sought links between standard measures of self-appraisal and reports of aggressive ac- tions. In particular, Kernis, Grannemann, and Barclay (1989) gave self-esteem measures to their participants on multiple oc- casions, which allowed the researchers to assess both the (mean) level of each participant's self-esteem and the degree of fluctu- ation in self-esteem scores. Kernis et al. used the fluctuation index as a measure of stability of self-esteem. These scores were then used to predict responses on an inventory of anger and hostility. Kernis, Grannemann, et al. (1989) found that the highest levels of self-reported angry and hostile responding were associ- ated with participants who had high but unstable self-esteem scores. Efforts to predict aggressive tendencies from self-esteem scores alone were inconclusive, and in fact people with high but stable scores in self-esteem showed the lowest rates of anger and hostility. In our view, this is profoundly important evidence about the psychology of bullies and other aggressors: Their opin- ions of themselves are very favorable but vulnerable to fluctu- EcKYrlSM AND VIOLENCE ations. Another way of describing this response pattern is that these hostile individuals were mostly quite high in their self- ratings but they did occasionally drop to substantially lower lev- els, which indicates that they were familiar with the distress and other aversive aspects of losing esteem. One must assume that people whose self-esteem occasionally drops will be sensitive and vulnerable to ego threats, in a way that people who show consistent, stable, impervious high self- esteem are not. Kernis, Grannemann, et al?s (1989) results thus seem quite compatible with the view that aggression ensues when people with very favorable views of themselves encounter an ego threat that evokes the possibility of losing esteem, al- though that conclusion requires some inferences beyond their data. Their findings do clearly link self-reported aggressive ten- dencies with unstable high self-esteem. Similar studies have also been conducted with narcissism. The term narcissism is based on the Greek myth about the young man who fell in love with his own reflection, and it is commonly used to refer to self-love; however, psychological (especially clinical) usages of the term have added the implica- tion of artificially inflated egotism. Wink ( 1991 ) analyzed nar- cissism as having several components. All of them were corre- lated with disregard for others, which we have suggested is one factor that contributes to willingness to behave violently. More important, the component that Wink ( 1991 ) defined as grandiosity or exhibitionism was particularly correlated with aggressiveness. That aspect suggests that wishing to show off to others, particularly so as to convince them to hold an unrealis- tically positive view of oneself, has an important link to aggres- sion. Similar findings were reported by Raskin, Novacek, and Hogan (1991). They found positive intercorrelations among grandiosity, dominance, narcissism, and hostility, thus again suggesting that these wildly favorable views of self are involved in aggressive behavior. 4 Critique. These studies may be criticized as subject to vari- ous biases of self-report, but they have the advantage that they can include good, psychometrically sound measures, and so the information about self-appraisals is good. These are in some ways the first substitute for having data that include direct mea- sures of self-esteem and subsequent measures of violent behav- ior. The step from self-reported hostility questionnaires to ac- tual violent action requires several inferences, however. The main risk would be that reluctance to admit hostility would be unequally distributed along the range of self-appraisal re- sponses; presumably, people who wish to present themselves fa- vorably would score high on self-esteem but low on hostility. Such a tendency would work against the obtained findings, how- ever, and so it seems appropriate to accept these findings unless further work contributes contrary evidence. Conclusion. Self-reported hostility does not correlate sim- ply or directly with self-esteem scores, but the most hostile peo- ple seem to be a subset of people having favorable self-apprais- als. Inflated self-appraisals and unstable high self-esteem have been linked to hostility, consistent with two of the hypothesized moderators. Group Differences An indirect strategy is to look at groups that are known to differ in self-esteem or egotism and then compare their rates Of aggressive actions. Obviously, these are correlational patterns and any one of them is inevitably subject to multiple alternative explanations. Only if there is broad agreement from multiple comparisons could one even begin to draw a tentative conclu- sion. Still, in an interdisciplinary literature review it seems de- sirable to examine as many sources of evidence as possible. Gender differences can be considered relevant. Men have higher self-esteem than women (e.g., Harter, 1993; Veroff, Dou- van, & Kulka, 1981 ), although the difference is not large and may be diminishing in the modern world (see also Crocker & Major, 1989). Men are also more aggressive than v~omen, al- though the size of the difference depends on what measure is used. In laboratory studies of aggressive behavior, the difference is about one third of a standard deviation (Eagly, 1985). In vio- lent crime, the difference is much larger. Although the precise figure varies according to crime and nation, men are between 5 and 50 times as likely to be arrested as women (Wilson & Herrnstein, 1985). The crime rate for women has risen in the United States since the 1960s, but most of that is due to prop- erty crimes, and women commit only about 10% of violent crimes (Wilson & Herrnstein, 1985). Victimization surveys, which avoid possible biases in the arresting system, point to similar differences. The only domain of violence in which women have been found to equal men is domestic violence, to which we shall return. In general, then, men are more violent than women, and they also have higher self-esteem. This finding is most consistent with our hypothesis that high self-esteem is a cause of violence. One qualification is that the size of the self-esteem difference seems too small to account for the large difference in violent crime rates. It is also worth noting that the only realm in which women are more violent than men is child abuse, which could fit the view that attacking a safe, weak target may be a strategy among people with low self-esteem. Another group known to have low self-esteem is depressed peo- ple(e.g., Allgood-Merton, Lcwinsohn, & Hops, 1990; Altman & Wittenborn, 1980; Brown, Andrews, Harris, Adler, & Bridge, 1986; Brown & Harris, 1978; Cofer & Wittenborn, 1980; Tennen & Affieck, 1993; Tennen & Herzberg~ 1987). The National Re- search Council's (1993) report on violence noted that many types of mental illness have been linked to violence, but depression had only been found in connection with family violence, and even those findings were subject to multiple ambiguities, including the possibility that depression was the result rather than the cause of violence and the possibility that depression was the result of one's own prior victimizations (given the often reciprocal and genera- tional nature of family violence). More generally, it does not ap- pear that depression is a major cause of violence. 5 That finding is consistent with the high se•stcem hypothesis, although the oper- ative factor in depression could be unrelated to self-esteem (e.g., apathy or lack of energy is the aspect of depression that prevents violence ). 4 They found, oddly, that if one takes the narcissism, grandiosity, and dominance out of hostility, the residual hostility is negatively related to self-esteem. It is not clear what this means; perhaps there is some aspect of violence that is associated with low self-esteem. Or perhaps this is an overcorrection. Another qualification is that depressed people do have elevated rates of suicide. Still, outwardly directed violence is low. BAUMEISTER, SMART, AND BODEN Psychopaths constitute another relevant group on the fringes of normality. Although mental illness as a cause of crime is be- yond the scope of this review, psychopaths are not mentally ill in the usual sense, because they are well in touch with reality and their actions are apparently freely chosen as opposed to be- ing driven by compulsions or irresistible impulses (Hare, 1993). Hare described them as "social predators;' and al- though they are not inherently or even normally criminals, they do commit a disproportionately high rate of violent crimes (in fact, he estimated that they are responsible for 50% of serious crimes). As to their self-views, Hare characterized them as hav- ing a "narcissistic and grossly inflated view of their self-worth and importance [and] a truly astounding egocentricity and sense of entitlement, and [as] see[ing] themselves as the center of the universe, as superior beings" (p. 38). They have grandi- ose conceptions of their abilities and potentialities, which have also been discussed by Meloy (1988). These observations sup- port the link between inflated self-appraisals and aggression. Hare also noted that psychopaths' sense of superiority is accom- panied by a tendency to regard other people as simply objects to be exploited. Thus, psychopaths seem to fit the view of highly favorable opinions of self as a source of violence. Hare (1993) also ob- served them to be "highly reactive to perceived insults or slights" (p. 59). We propose that such hypersensitivity might reflect the use of violence to ward off emotional distress, and Meloy (1988) proposed that link as central to the psychopathic mentality. Although they are not socially sensitive in the sense of having high empathy or concern for others, they are sensitive in the sense of understanding how to manipulate other people, and they are certainly sensitive to any blows to their egotism. Hare's observations are thus consistent with the view that threatened egotism is a main cause of violence, although they also support the view that egotism can cause violence directly because one disregards the other's interests and point of view. Comparing self-esteem across racial or ethnic groups is com- plicated by several factors, such as measurement issues and temporal changes, but the very possibility of temporal shifts presents an appealing chance to look for covariation in self- esteem and violence levels. In the 19th and early 20th centuries, American White men presumably were fairly securely con- vinced of their superior stature. This confidence is generally as- sumed to have eroded in recent decades, and indeed research now indicates that Black people have self-esteem levels equal to or higher than those of White people (see Crocker & Major, 1989, for review). Concerted efforts to boost racial pride and dignity among Black Americans in the 1960s and 1970s may have contributed to this shift. Meanwhile, violence levels also appear to have changed, and these changes directly contradict the view that low self-esteem promotes violence. During the period when White men had the highest self-esteem, they were also apparently the most violent group. Historians believe that rapes of White women by Black men were quite rare, whereas the reverse was relatively common (e.g., Brownmiller, 1975). Likewise, the majority of interracial murders involved White men killing Blacks, a pattern that is still reasonably well documented into the 1920s (e.g., Brearly, 1932; Holfman, 1925; Von Hentig, 1948). These patterns have been reversed in recent decades as Black self-esteem has risen relative to White self-esteem. According to Scully (1990), Black men now rape White women approximately 10 times as often as White men rape Black women. The timing of this reversal appears to coincide with the concerted cultural efforts to boost self-esteem among Blacks: LaFree's (1976) review of multiple studies of interracial rape concluded that researchers found ap- proximately equal numbers of Black-on-White and White-on- Black rape in the 1950s, but since 1960 all studies have found a preponderance of Black-on-White rape (see also Brownmiller, 1975). Similarly, recent murder statistics indicate that the strong majority (80%-90%) of interracial murders now consist of Blacks murdering Whites (Adler, 1994). Clearly, both races have committed far too many horrible crimes, and neither race can find much claim to any moral high ground in these statis- tics, but the shifting patterns on both sides repeatedly link higher or rising esteem with increasing criminal violence to- ward the other. 6 To seek converging evidence regarding cross-temporal shifts in self-appraisals and aggression, we examined research on manic-depressive ( bipolar ) disorders. Although these individu- als are mentally ill and therefore fall outside the main scope of our review, they do provide an appealing chance to examine intraindividual fluctuations in self-esteem. Inflated, grandiose self-esteem occurs in the manic phase and presumably disap- pears or inverts during the depressive phase (American Psychi- atric Association, 1994). Evidence suggests that aggressive ac- tions and general patterns of hostility or irritability coincide mainly with the manic state (Goodwin & Jamison, 1990), which is again consistent with the view that favorable and in- flated self-appraisals are linked to violence. Another variation on the group-differences strategy is to look at groups who are defined by particular states rather than permanent traits. Indeed, although state self-esteem is strongly correlated with trait self-esteem, it does fluctuate around the chronic level (Heatherton & Polivy, 1991; Kernis, 1993). One particularly rele- vant group would be people who consume alcohol. It is well docu- mented that alcohol consumption increases agw~ession (Bushman & Cooper, 1990). This conclusion has been well supported by lab- oratory studies on aggression, although the usual conclusion is that alcohol does not so much create aggression as increase aggressive responding once aggres~on is elicited by other causes. Moreover, it is well established that the majority of violent crimes are com- mitted by people who have consumed alcohol, and indeed this point has been established repeatedly and separately for murder, rape, and a.~sault (National Research Council, 1993; see also Gott- fredson & Hirschi, 1990;Groth, 1979; Norris, 1988). To forestall the drawing of unintended implications from these data, we hasten to add that we are not advocating that any particular group or category should be denied a basis for pride or self-esteem. We do think that a diverse society such as the United States is likely to function best if all groups cultivate an attitude of respect and appreciation toward all others and seek a severely judicious balance between humility and pride regarding their own group's accomplishments. Efforts to impose humility on others are likely to backfire, and indeed some of America's racial problems can probably be traced to past policies of deliberate subjugation that were based on one race's ill-founded assumptions of innate superiority. It is hoped that both races have had enough oppor- tunity to learn that respect ought to be earned as an individual rather than claimed as a member of a racial group. The apparent link between racial egotism and violence toward others may be one of the more unsa- vory demonstrations of the cross-racial universality of human nature. ECK)TISM AND VIOLENCE What, then, is the self-esteem level of people who consume alcohol? Evidence indicates that alcohol raises the favorability of self-appraisals. Intoxicated people rate themselves more fa- vorably than they would otherwise (Banaji & Steele, 1989; Di- amond & Wilsnack, 1978; Hurley, 1990; Konovsky & Wils- nack, 1982; Orford & Keddie, 1985). Apparently, then, alcohol generally helps create a state of high self-esteem. Thus, again, a group that shows elevated egotism also shows unusually high rates of violence. Critique. Data on large groups can furnish quite accurate indexes of rates of violence with high external validity. The drawback is that each group difference is subject to multiple possible explanations. For example, it is difficult to be certain that the favorable self-assessments of intoxicated people are a mediating factor in their violence; it is plausible that there are direct links from alcohol to violence, without self-esteem being involved. Moreover, it is possible that the violence is perpetrated by a minority portion of the group who may be atypical in self- appraisals. For example, men are both more violent and more egotistical than women, but it is possible that most of the vio- lence is perpetrated by men who do not share the egotism com- mon to their gender. Therefore, none of the findings in this sec- tion permits a strong conclusion about the link between self- appraisals and violence. On the other hand, the convergence across multiple comparisons is impressive in its contradiction to the low self-esteem view. To put this another way: Although each of the results covered in this section could be explained with reference to other factors, it would require considerable explaining and an abandonment of parsimony to continue as- serting that low self-esteem causes violence. Conclusion. If low self-esteem did cause violence, one would expect that in general groups with lower self-esteem would be more violent, but the evidence reviewed in this section repeatedly found the opposite. It is difficult to maintain belief in the low self-esteem view after seeing that the more violent groups are generally the ones with higher self-esteem; at best, one would have to assume that the effect is weak enough to be overridden by many other variables. The effort to invoke al- ternative explanations is especially difficult in light of evidence that shifts over time in self-esteem are accompanied by shifts in aggression such that the periods of higher self-esteem are the ones linked to greater violence. Moreover, several findings suggest that inflated or unstable views of self are linked to violence. The grandiose self-apprais- als of psychopaths and manics, and the inflation of self-ap- praisal during alcoholic intoxication, provide support for this view. Murder and Assault We turn now to considering violent crimes by individuals. Studies have examined various samples of offenders, and our goal is to ascertain what direct observations of offenders have suggested about their self-appraisals. A general methodological problem is that offenders are most available for study after ar- rest and imprisonment, but the humiliating process could welt have an effect on self-esteem. Being captured for a crime is a prominent failure experience, and moreover the assertion of humble remorse is often perceived as a prerequisite for parole and early release. As a result, superficial evidence of low self- esteem should be especially easy to find in studies of convicted offenders. Despite any methodological bias toward low self-esteem, however, studies of violent offenders have typically suggested strong tendencies toward egotism and narcissism, and any signs of low self-esteem are at best ambiguous. Thus, the classic study of violent men by Toch (1969/1993) sought to classify them into types. His taxonomy was weakened by the fact that the two most common types could not be reliably distinguished, and the third largest was related to the second, so in a sense what Toch produced was one very large category and an assortment of small exceptions. The large category (the majority) consisted of men for whom threatened egotism was behind the violence. Although, as noted earlier, Toch's remarks were inconsistent as to whether these men secretly had high or low self-esteem, he was clear that these men generally became violent as a means of proving positive self-worth and refuting perceived insults. These individuals often seemed to seek out or manufacture situations in which their image was challenged and they could bolster it by aggressive action. As we have said, this pattern suggests confi- dence and possibly arrogance. Berkowitz (1978) studied a sample of British men impris- oned for assault. The investigation sought evidence for the hy- pothesis of a subculture of violence. More precisely, Berkowitz tried to show that these men were motivated by the desire to look good by showing off through aggressive behavior that would be admired by others. But he was unable to find evidence to support that hypothesis. Instead, most of the fights had begun when one man thought another had insulted or belittled him. "Our impression is that their egos were fragile indeed" (p. 158), said Berkowitz, which could mean low self-esteem or could mean a defensive pattern of high self-esteem. He said that these men seemed excessively prone to regard another's remarks as insulting or belittling which seems consistent with the unstable high self-esteem suggested by Kernis, Grannemann, et al. (1989) and with the hypersensitivity Hare (1993) observed among psychopaths. The impression of egotism is further sup- ported by Berkowitz's finding that most of the men said they had had high confidence that they would win the fight. Thus, in general, Berkowitz was unable to confirm his initial hypothesis that aggression was a means of making a good impression on others or of adhering to subcultural values or securing material rewards: "If anything, pride appears to be far more significant than direct external benefits. Wounded pride certainly seems to enrage them" ( 1978, p. 160). Wounded pride is essentiaUy the same as threatened egotism, as we have proposed. A recent study of homicide by Polk ( 1993 ) confirmed these conclusions. Polk noted that nowadays many homicides occur in connection with other crimes such as robbery, but in the re- maining cases the homicide is often the result of an altercation that begins with challenges and insults. The person who feels he ( or less often she) is losing face in the argument may resort to violence and murder. Several studies have used the Minnesota Multiphasic Person- ality Inventory (MMPI) on various populations of offenders (see Wilson & Herrnstein, 1985, for review). Three of the 10 MMPI scales are relevant to low self-esteem: Depression (2), which includes self-deprecation; Psychasthenia (7), which in- dudes anxiety and indecision; and Social Introversion (0), which includes insecurity and shyness. The weight of the evi- BAUMEISTER, SMART, AND BODEN dence, including some prospective longitudinal studies, as re- viewed by Wilson and Herrnstein, has not consistently shown any relationship between psychasthenia and criminality, but the other two scales are both negatively related to criminality: De- pressed, self-deprecating, insecure, and shy people are under- represented among criminals. These findings are difficult to rec- oncile with the view that links low self-esteem to violent and antisocial tendencies. Wilson and Herrnstein concluded that the "lack of criminal tendencies among those whose highest scores are on scales 2, 5, and 0 are by now commonplace in the empir- ical literature on crime" (p. 188). By this measure, then, the patterns that suggest low self-esteem produce remarkably few criminals. Even within samples of offenders, it appears that indicators of egotism can discriminate violent and troublesome tenden- cies, and it is the favorable views of self that are linked to the worse actions. Gough, Wenk, and Rozynko (1965) adminis- tered the California Psychological Inventory to young men (in their late teens) on parole. The researchers were able to predict future parole violations (recidivism) with some success, and this sort of predictive success had eluded previous researchers. Among the traits that predicted high recidivism were being ego- tistical and outspoken (as well as "touchy," which suggests being easily offended); meanwhile, being modest and unassuming were among the traits associated with men who were least likely to violate parole. These results all seem to fit the view linking favorable views of self to violent tendencies. Similar tendencies are evident even earlier in life. Studies of ag- gressiveness in children are of special interest because aggressive children show substantially higher rates of adult aggression and criminal violence (e.g., Huesmann, Eron, Lefkowitz, & Waldea; 1984). Olweus ( 1994 ) recently summarized his own program of research on bullies, who have been shown in follow-up studies to have four times the level of serious criminality during adult life that nonbullies show. In contrast to victims of bullying (who show multiple indications of low self-esteem), the bullies themselves seemed relatively secure and free from anxiety. "In contrast to a fairly common assumption among psychologists and psychiatrists, we have found no indicators that the aggressive bullies (boys) are anxious and insecure under a tough surface" (p. 100), said Ol- weus, adding that multiple samples and methods had confirmed this conclusion, and concluding that bullies "do not suffer from poor self-esteem" ( p. 100 ). One of the most earnest and empathic efforts to understand the subjective experience of committing crimes was that of Katz (1988). Homicide as well as assault emerged in his study as typi- cally caused by threats to the offender's public image. In Katz's view, the offender privately holds a positive view of self, but the eventual victim impugns that view and implicitly humiliates the offend~ often in front of an audience. The response is unplanned violence resulting in injury or death. Katz insisted that feelings of being humiliated are quickly transformed into rage. Katz argued that many men feel that almost anyone can judge them and im- pugn their esteem, whereas for women self-esteem is most heavily invested in their intimate relationships---with the result that men will attack strangers whereas women mainly just murder their in- timate partners, because only the partners can threaten their self- esteem to a sufficient degree to provoke such a violent response. Furthermore, Katz ( 1988 ) argued that many youthful circles and street subcultures extend substantial respect to the "'bad- ass" sort of person who transcends the pressures to conform to societal norms, rationality, and ideals. This prized identity is cultivated in part by creating the impression of being unpre- dictably prone to chaos and irrational violence. More generally, street violence, whether by individuals or gangs, often revolves around competing claims to hold a special, elite identity. Concern over respect is hardly limited to modern lower-class youths. Upper classes often have had just as much appetite for egotistical gratifications. Wyatt-Brown (1982) said that the slave codes and other racial practices of the antebellum South all had the fundamental theme that Black people should show sincere respect for all Whites. Moreover, he said that in the Northeast, given the open industrial economy and abundant universities, self-worth could be established through scholarly erudition or financial success, but such means were largely un- available to Southern men, who therefore resorted to violence instead. Accordingly, murder rates in the South were many times higher than property crimes and many times higher than the corresponding murder rates in the Northeast. The violence proneness of the American South has been elab- orated by Nisbett (1993). His work emphasized the point that Southern Americans are more inclined than Northerners to en- dorse violence in response to insults. In laboratory and other studies reported by Nisbett, Southern participants were more angry in response to insults than were Northerners, and they advocated more severe and violent solutions to scenarios in- volving conflicts and insults. The Southern "culture of honor" is an important cause of this tendency toward violence. A sim- ilar conclusion was reached by Ellison ( 1991 ), who found that Southerners are more likely than others to condone defensive or retaliatory forms of violence. Violence is therefore instrumen- tal in enhancing one's honor or reputation. Dueling is a traditional and widely disseminated pattern of violence that is similar to the way Nisbett (1993) and others have portrayed the violent honor culture of the South. Dueling provides a ritualized form of aggression that can be regarded simply as a formalized, systematic form of ordinary fighting. According to Kiernan's ( t 989) account of dueling in European history, dueling was intimately tied to highly favorable views of self and to threats to such esteem. Dueling was mainly practiced by the upper classes, who (back when no egalitarian ideologies diminished their sense of being innately superior people) culti- vated their inflated notions of honor, virtue, and entitlement to respect. Minor acts could be construed as insults, prompting the offended person to insist on fighting on the so-called field of honor. Thus, this visible and durable form of violence sprang directly from inflated notions of personal (and familial) superi- ority and from ego threats. Indeed, Wyatt-Brown's (1982) history of the culture of honor in the American antebellum South noted that dueling was com- mon and felt by many to be an obligatory response whenever one was insulted. However, Wyatt-Brown's comments could be construed to fit either the high self-esteem or the low self-esteem theory. In favor of the latter, he said that winners and sometimes even losers of duels gained esteem in the eyes of others, and he speculated that some duelists suffered from an "inner sense of worthlessness" (p. 360) that prompted them to fight as a way of gaining public esteem. Still, that remark was overtly speculative and may be a concession to the conventional wisdom that low self-esteem causes violence. In contrast, Wyatt-Brown's funda- EGOTISM AND VIOLENCE mental analysis of honor began with "the inner conviction of self-worth" (p. t4), to which public validation must be added, and so fighting duels was a means of publicly defending one's claims to a positive identity against external doubts or slights. In Our view, the weight of evidence presented by Wyatt- Brown (1982) supports the view that upper-class Southern men generally held favorable, rather than humble or unfavorable, views of themselves, even ifthere might be some uncertainty or instability attending their egotism. The role of ego threat is clear, in any case, insofar as duels were nearly always initiated in response to derogatory comments by another (about oneself or one's family). Critique. Again, the convergence across many studies is far more conclusive than the individual results themselves. In all studies of violent populations and samples of offenders, it must be acknowledged that they may be atypical of the broader pop- ulation. Hence, although studies have consistently character- ized offenders as egotistical, one cannot assume that all egotists have violent tendencies. Conclusion. Multiple studies of murder and assault have found threatened egotism to be a significant factor. In some spheres, such as dueling, the link is explicit and formal, whereas in most others it emerges as a common factor. The view that low self-esteem leads to violence appears contradicted by studies on offenders, from childhood bullies to convicted murderers. Many of the studies reviewed here included observations as to how seemingly trivial the provocation was. This is consistent with the view that such aggression has an anticipatory nature, designed to head off possible losses of esteem. The pattern of responding violently to slight or incipient threats suggests a hy- persensitivity to bad feedback, and this could well signify antic- ipatory emotional responses and some tentativeness about the favorable self-appraisals that are questioned. We can thus see a consistent pattern across cultural, histori- cal, situational, and class boundaries. Many violent acts by in- dividuals occur in response to derogatory remarks or acts by others, including ones that seem minor or trivial to observers. In most cases, the perpetrators appear to be men who privatel3; believe in their own superior worth but who encounter others who impugn or dispute that belief. Violence may be especially likely when the individual lacks alternative means to prove or establish his superiority. Rape Rape is a complex crime, and there is considerable contro- versy about its definition, causes, and meanings. Some apparent causes, such as displaced revenge for prior mistreatment and belief in rape myths, seemingly have little relevance to self- esteem, but there is some evidence that self-esteem can be in- volved in rape. An eminent book on rape by Groth (1979) reported that in one major pattern rapes were often preceded by various blows to the rapist's self-esteem, causing him to feel that he "'had been wronged, hurt, put down, or treated unjustly" (p. 16), usually by some woman. In the other major pattern Groth identified, either a woman or a man does something to the rapist that "un- dermines his sense of competency and self-esteem" (p. 30), and raping is a means of "restor[ing] his sense of power, control, identity, and worth" (p. 31 ).7 Although Groth did reiterate the standard line of interpreting the rape as reflecting low self- esteem, his argument that the rape "'restores" positive views of self contradicted that analysis, because it implies that positive views of self exist to be restored. It would perhaps have been more precise to say that rape tends to result from a favorable view of self that has recently been impugned by another person or situation. Groth's observations generally seem most consis- tent with the view that high but unstable self-esteem is the cause of sexual violence. Groth insisted further that the appeal of rape is not sexual release but rather enjoyment of the victim's help- lessness and thus of one's own superior power (the rapist "thrives on a feeling of omnipotence;' p. 47). He added that participation in gang rape is often motivated by "an effort to retain stature" (p. 80), and that the leader of a gang rape enjoys both control over the victim and over his cohorts. Over 100 convicted rapists were interviewed by Scully and Marolla (1985) to ascertain their motives and rewards. The en- joyment of power over one's victim was cited by many. A num- ber of respondents made the comment, also found in Groth's (1979) interviews, that one raped a particular woman to dis- abuse her of her sense of superiority. That is, the woman gave the man the impression she thought she was better than he was, and so he raped her as a way of proving her wrong. The implica- tions for self-esteem are quite apparent: Rape is motivated by the man's belief in his own superiority, which has been chal- lenged or disputed by the woman (or occasionally by someone else). The selection of victim on the basis of her own apparent self-esteem is consistent with the zero-sum view of self-esteem, in which one can only gain esteem at the expense of others. In a later work, Scully (1990) reported ample signs of egotism among many of the convicted rapists, especially those who denied their guilt. She said many of these men spontaneously bragged to her about their sexual prowess and about their other attributes and accomplishments, even claiming to be "multitalented super- achievers" (p. 112). It seems fair to regard these as inflated self- appraisals, especially when one considers that all the men were in prison at the time. A large minority even thought their victims would regard them favorably afterward. She too found evidence of selecting a victim on the basis of the victim's perceived high self- esteem, such as the case of the rapist who described his motivation and satisfaction by saying "I felt like I had put her [the victim] in her place" ( Scully, 1990, p. 134). Marital rape is likewise a controversial issue (even to define), and its causes are poorly understood, but again there is some evidence of issues of self-esteem and control. Finkelhor and Yllo (1985) cited a common masculine belief in entitlement as a cause of marital rape. Husbands rape their wives to prove their sexual ownership and rights over their wives, as well as to dem- onstrate superior power and achieve a victory over the wife. The surprisingly high rates of anal intercourse (which is linked to dominance; see Baumeister, 1989a) and forced sex in front of witnesses both suggest that marital rape often is essentially an effort to achieve symbolic proof of the husband's superior sta- tus. This brings up the broader issue of domestic violence, to which we turn in the next section. Critique. The studies reviewed in this section suffer from Groth did identify a third type of rape, based on sadism, but he said that was statistically a very small and rare pattern. It is irrelevant to our hypothesis. BAUMEISTER, SMART, AND BODEN limitations in sampling, measurement, quantification, and ba- sis for comparison. Research into the psychology of rapists re- mains in a preliminary state, partly because researchers have focused mainly on victims, and in many cases strong ideological commitments may have complicated the development of em- pirically based theory. The studies cited here are valuable sources of observations and impressions, but the evidence is not strong enough to justify sweeping generalizations or firm causal conclusions, Conclusion. Preliminary evidence portrays rapists as hav- ing firm beliefs in male superiority and often elaborate beliefs in their own individual superiority, all of which is contrary to the low self-esteem view. Some observations support the view that ego threats figure prominently in the events leading up to rape. In many cases, however, the victim was not the source of the ego threat. Additional observations by several researchers did, however, fit the pattern we noted based on the zero-sum view of esteem, namely that some offenders choose a victim simply because her own apparent self-esteem somehow consti- tutes a threat to the rapist's belief in his superiority, even though she never evaluated him directly. Domestic Violence Between Partners We proposed that domestic violence was the one sphere in which there would he extra reasons to expect that low self- esteem might predict violence, insofar as unconfident people might select safe, relatively helpless targets for their aggressive impulses. As it happens, researchers have devoted more effort to measuring and studying the effects of self-esteem on domestic than on other forms of violence. Despite frequent portrayals of wife beaters as having low self- esteem (e.g., Walker, 1979), the evidence has not provided much support for this view. Stets ( 1991 ) found no link between self-esteem and inflicting violence among men; among women, there were weak correlations between inflicting violence, being the victim of partner violence, and having low self-esteem. Burke, Stets, and Pirog-Good (1988) found "that self-esteem was not related directly to either physical or sexual abuse for men or for women" (p. 283), although there were some "spuri- ous" correlations as a result of shared variance with gender identity measures. This conclusion seems consistent with Ker- nis's (1993) position that measuring self-esteem (as opposed to looking for the pattern of high but unstable, or variable, self- esteem) alone may be of little help in determining the causes of violence. Even studies that have found low self-esteem correlated with inflicting violence suffer from ambiguities that have caused the researchers to question the traditional view that low self-esteem causes violence. Goldstein and Rosenbaum (1985) found sig- nificantly ltr, ver levels of self-esteem among physically abusive husbands than among happily married husbands or among un- happily married but nonviolent husbands. They observed, how- ever, that the correlational findings were inconclusive and that it is "probable" that "abusing one's wife is self-esteem damag- ing" (p. 427); thus, low self-esteem may be the effect rather than the cause. Their sampling method may also have contributed to this, because it consisted of men who had referred themselves for therapy as wife abusers. As Holtzworth-Munroe (1992) has noted, studies of domestic violence typically find severe differ- ences between the minority of abusers who admit to being abus- ers and the majority of them who tend to minimize or deny their violence and who lay blame for violent incidents on external factors, such as the victim's provocations. It does seem that vol- untarily identifying oneself as a wife batterer and reporting for therapy would be incompatible with furnishing a highly favor- able rating of self on a self-esteem scale. One other study that found correlations was done by Russell and Hulson (1992). They used a nonstandard self-esteem mea- sure that they thought would he especially relevant to domestic violence and an unorthodox sample to find several correlations suggesting that low self-esteem among wives was linked to both perpetrating and receiving both psychological and physical abuse and that low self-esteem among husbands was correlated with inflicting psychological abuse (e.g., insulting the partner) but not physical abuse. A multiple regression analysis elimi- nated most of their effects, although they did find that wives low in self-esteem were still more likely to physically attack their husbands. Thus, repeated efforts to link measures of low self-esteem to self-reported physical violence have not yielded much. Possibly clearer evidence comes from studies concerned with under- standing the motives and circumstances that lead to wife heat- ing. Gelles and Strans (1988) summarized a common provoca- tion to domestic violence by saying that people tended to hit their spouses and children "after they felt that their self-worth had been attacked or threatened" (p. 35). They noted that the threat to self-worth may be external, such as at work, or it may originate in the family itself. In the latter case, family members know what others are sensitive and vulnerable about and may say cruel or disparaging things, which elicit physically violent responses. This pattern was found "over and over again" (p. 79) in interviews. Similarly, Gondolf ( 1985 ) characterized wife beaters in his sample as men who strongly endorsed traditional views about family and gender roles, particularly the "male ex- pectation of privilege" (p. 82 ) and an exaggerated sense of re- sponsibility for the family. When family events failed to follow their cxpectations or jcopardizcd their sense of privilege, they turned violent. A historical study of physically abusive husbands around the turn of the century by Peterson ( 1991 ) is typical and relevant. Peterson characterized the typical wife beater in his sample as "not an all-powerful patriarch but rather a husband with but marginal resources" (p. 12) insofar as these husbands tended to lack money, education, and other signs of stature, especially in comparison with their wives. Peterson inferred that the lack of stature would translate into low self-esteem and was conse- quently quite puzzled by the signs that these men believed strongly in male superiority. Indeed, in discussing the findings, Peterson cited what he regarded as an inconsistency in the liter- ature, namely evidence that wife beaters were men who lacked stature and power but who nonetheless held traditional views about male dominance in marriage. To Peterson, these findings seemed to suggest contradictory conclusions about the role of self-esteem. Such findings are only conflicting, however, if one subscribes to the theory that low self-esteem is the cause of family violence. To us, the findings aptly capture the prototypical cause of vio- lent aggression: threatened egotism, or in this case the man's firm belief in his own superiority coupled with the threat (due EGOTISM AND VIOLENCE to some stature superiority enjoyed by the wife) that others may not share that belief. Men who regarded themselves as superior but who saw that their wives had surpassed them on some im- portant dimensions seemed quite likely to feel this insecure, threatened egotism, which may have led them to strike out against their wives. From our perspective, this pattern confirms that a crucial cause of the violence was the men's beleaguered belief in their own superiority. Similar findings have been reported by Gelles and Straus (1988), who noted that "stature inconsistency is an important component of the profile of the battering husband" (p. 88). They said the typical wife beater feels obliged to hold down the traditional male role of superiority and family dominance but feels undermined by having less economic or social resources than his neighbors and often even his wife. Clues and Rosenthal (1990) likewise found that wife abuse was positively correlated with the husband's perception of the wife as having high reward power. Gelles and Straus reported that many wife beaters spoke to them of "needing" to strike their wives to show them who was the boss (e.g., p. 92 ). Once again, this view precisely fits the formula we have proposed: The man regards himself as superior but fears that others do not sufficiently endorse that view. Another methodologically strong and often cited study of spouse abuse was done by Hornung, McCullough, and Sugi- moto (1981 ). They found that, contrary to conventional wis- dom, working wives are attacked by their husbands more than wives who stay at home, presumably because of stature inconsis- tency: The wife who remains at home does not implicitly threaten her husband's superior stature in the family. Thus, again, it is the pattern of beliefs in one's (the husband's) superi- ority, coupled with circumstances that seem to contradict or undermine that superiority, that is most conducive to violence. Indeed, many of Hornung et al.'s ( 1981 ) findings support the view that a threatened sense of male superiority is an important cause of domestic violence by men. Hornung et al. studied only reports by wives and emphasized violence by husbands. Some oftbeir specific findings seem internally inconsistent unless one assumes that educational level sets expectations and occupa- tional level is perceived as the actual achievement outcome. In that view, domestic violence was most common when men held high but frustrated expectations. Highly educated men with rel- atively uneducated wives were violent, which is consistent with the view that seeing oneself as superior is a cause of violence. Yet when the woman's occupational level was higher than the man's, the man tended to become violent. Wives in the top oc- cupational stratum were subjected to high violence; men in the top stratum were relatively nonviolent. Indeed, when the wom- an's job was higher in stature than her husband's, the likelihood of life-threatening violence was six times higher than when the pair's occupations were similar or compatible. Above all, men who had been highly educated but had not attained high-stature occupations were particularly violent, and this was intensified if the wife had achieved high stature. Men who were overachievers, however, in the sense of enjoying occu- pational stature and success above and beyond what their educa- tional level would normally predict, were significantly less vio- lent than control participants. In other words, when men's ex- pectations exceeded their outcomes, they were highly violent, but when their outcomes exceeded their expectations, they were exceptionally nonviolent. This finding shows that not all stature inconsistencies are equally likely to lead to violence. Threat- ened egotism increases the risk of violence, whereas the oppo- site form of inconsistency (success despite humility) reduces it. All of these findings suggest that men beat their wives to maintain the superiority of the husband role that has been threatened or jeopardized. When the man's outcomes fall short of his expectations, he is vulnerable to feeling that his wife will not respect him, and he may be especially prone to reassert his superiority with physical violence. When the wife has reached a level of occupational success that is higher than her husband's, he is again more likely to beat her, presumably again as a way of enforcing his sense of superiority. Related to this is a finding by Goldstein and Rosenbaum (1985), which suggests that abusive husbands are more likely than others to interpret a wife's behavior as threatening or dam- aging to the man's favorable image of self. A man who feels his superior stature is tenuous, possibly because his occupational success has not measured up to his or his wife's expectations, may be extra sensitive to comments or actions by her that might imply a disparaging or disrespectful attitude. To be sure, not all domestic violence is perpetrated by men. Straus (1980) and others have noted that most researchers be- gan with the assumption that spouse abuse is mainly perpe- trated by men, yet often objective data fail to confirm that. Two reasons that have been suggested are that men tend to not report being physically abused by their wives (because they would be ashamed at being physically bested by a woman) and that the superior size and strength of men typically mean that they in- flict greater harm on their wives than wives can inflict on hus- bands. Straus's own data found approximately equal rates of domestic violence by both genders, which surprised him and his colleagues, leading to a series of supplementary analyses aimed at finding the ballyhooed preponderance of male violence. Sev- eral analyses (e.g., analyzing frequency rather than mere inci- dence and restricting analyses to severe violence) failed to yield any difference, but finally one pattern emerged in which men were more violent: In cases in which there was mutual, escalat- ing violence, husbands escalated to higher levels of violence than their wives. This lone finding may be suspect because so many analyses were done before something could be found to fit the research- ers' preconceptions, but if we assume that it is nonetheless cor- rect and valid, it seems to fit very well the notion of threatened egotism as a cause of violence. Mutual violence presumably means that the couple is prone to engage in physical fighting. Assuming, again, that the majority of men are physically supe- rior at fighting and that they would regard being beaten by a woman as a disgrace, it may simply be that when a man finds himself in a physical battle with his wife he escalates to a level of brutality at which he is sure to win the fight. In other words, the inclination to beat one's spouse may be equally distributed across men and women, and men are only more violent toward their spouses when their egotism (i.e., their sense of superiority and immunity to embarrassment) is jeopardized. Renzetti (1992) studied partner abuse in lesbian relation- ships, which provides a valuable complement to the studies that have focused on male perpetrators of domestic violence. Her data, along with several other studies she cited, confirmed the contribution of stature inconsistency to domestic violence. Bat- terers wanted to be the decision makers, but the victims tended BAUMEISTER, SMART, AND BODEN to have more money and other resources (according to victims' reports, which were Renzetti's sole source of data). The two largest direct causes of violence in Renzetti's (1992) data, however, were dependency and jealousy. She noted that equality and independence are particularly strong ideals in les- bian communities because of feminist ideology and other rea- sons; the more one woman felt dependent on the other, the more violent she was likely to be, and her violence seemed to signify power and autonomy that contradicted the implication of her apparent dependency. Meanwhile, Renzetti noted that jealousy is strongly blended with envy in homosexual relationships, thus adding a significant element of ego threat, because the outside person who pursues one's partner would also be an eligible part- ner for oneself, and so apparently the person has chosen one's partner over oneself, suggesting that the interloper has judged oneself as somehow less attractive than one's partner. Thus, the ego threat of partner infidelity is doubled in homosexual as compared with heterosexual triangles. Critique. Consistent with our prediction that the low self- esteem view would fare best in studies of domestic violence, there have been some findings supporting that theory. However, these effects are weak and small and have often failed to repli- cate. Several of the studies contained multiple and fundamental flaws. Problems of sampling are crucial, because the most con- veniently available sample usually consists of people who iden- tify and reproach themselves as violent spouses (such as those who have sought therapy), but these are a seriously atypical mi- nority (Holtzworth-Munroe, 1992). The possibility that low self-esteem was a result of abuse, or especially that low self- esteem causes women to stay in mutually abusive relationships, has been advanced as a likely explanation even when findings have found low self-esteem to correlate with violence. Indeed, one ought to expect considerable self-deprecation among people who have acknowledged beating their wives or partners, given the stigma. The findings that women with low self-esteem ag- gress against their partners are complicated by the reciprocal nature of much marital violence: If a battered woman strikes her husband in self-defense (as many researchers propose is common), it is misleading to depict her act as a result of her low self-esteem. Larger, more careful, and more systematic studies have found no direct link between self-esteem and domestic vi- olence, and presumably for that reason researchers have re- cently shifted emphasis to stature inconsistency. The link be- tween stature inconsistency and domestic violence appears to be reasonably well supported, although not all inconsistencies are equally productive of violence. Conclusion. Much research on domestic violence has been shaped by the traditional belief that low self-esteem is a major cause, but repeated efforts have yielded at best weak and ambig- uous findings consistent with that view. When preliminary findings have linked low self-esteem with violence, the evidence has usually disappeared under the influence of statistical con- trois, rigorous sampling, or prospective methods. It does not appear that low self-esteem causes domestic violence. Mean- while, evidence of egotism among physically abusive husbands is abundant. Studies of these men repeatedly portray them as believing strongly in male superiority, especially in the face of circumstances that might question their own superiority over their female partners. Research has shown strong support for the hypothesis that stature inconsistency is a major cause of domestic violence, es- pecially violence against women. Male perpetrators apparently believe themselves as being entitled to superior regard but find that circumstances fail to confirm these inflated notions of self, and so they attack their partners. Often the attack is a direct effort to reassert the superiority they believed themselves enti- tled to enjoy, such as the men who beat their wives to show them "who's the boss," in their common phrase. Still, it is important to note that not all forms of inconsistency produce violence, and so it is slightly misleading to assert stature inconsistency as a cause. Violence seems to arise when circumstances question favorable assumptions about the self; in contrast, when circum- stances provide equally inconsistent but highly favorable im- plications about the self, violence is low. Thus, only some forms of inconsistency are relevant. The evidence thus points to threatened egotism as the decisive cause of domestic violence between adult partners. Other Domestic Violence: Parents and Children As we suggested, child abuse may be the form of violence most likely to yield evidence of low self, esteem, because people who happen to feel violent while lacking self-esteem would be most likely to choose relatively helpless targets such as children. A series of studies has indeed suggested that child abusers have low self-esteem (S. C. Anderson & Lauderdale, 1982; Evans, 1980; Melnich & Hurley, 1969). More recent work has begun to question that conclusion, however. Shorkey and Armendariz (1985) replicated the lower self-esteem found among child abusers (from a sample in counseling) but concluded that it is not the main causal factor. One of the most important studies, however, was done by Christensen et al. (1994). They suggested that previous studies may have been misled by reliance on samples of incarcerated abusers or abusers in therapy, because being publicly identified as a child abuser may well lower self-esteem (especially on mea- sures that emphasize getting along well with others). They therefore conducted a prospective study and found no differ- ence in self-esteem between the eventual abusers and the com- parison group. They concluded that low self-esteem is not a risk factor for potential physical abuse. One other prospective study, by Altemeier, O'Conner, Vietze, Sandier, and Sherrod ( 1983 ), also provided relevant evidence. They found that abusive mothers differed mainly in that they were more likely to endorse the statement "I'm usually unsuccessful in life" than others. This statement could indicate low self-esteem, but it could also indicate ex- periences that threaten high self-esteem. A more recent study by Dutton and Hart (1992) found that high levels of narcis- sism (particularly narcissistic personality disorders) were as- sociated with violence against family members. Thus, there is evidence to support the view that excessively favorable but threatened views of self lead to violence. Elder abuse is another form ofdomestic violence that is often perpetrated by women, although both genders are well repre- sented among such batterers. Pillemer's ( 1985 ) findings contra- dicted the traditional stereotype of elder abuse as caused by helpless dependency of the victims and by the perpetrator's re- sentment of the victim's neediness. Pillemer found that the vic- tims of elder abuse tended to be more independent than control EGOTISM AND VIOLENCE participants. The abusers, however, tended to be dependent on the victims for money, transportation, or other resources. The abusers apparently were embarrassed by and resented their own dependency. Thus, the refusal to accept an inferior or dependent role appears to be a major cause of elder abuse, which fits the pattern of ~gh but threatened Self-esteem. The most common but ironically least studied form of do- mestic violence is between siblings. Wiehe ( 1991 ) conducted a victim survey but noted that almost nothing is known about perpetrators, except for the cases in which victims become per- petrators. The systematic selection of weaker, vulnerable targets was confirmed in their findings, according to a general pattern in which abusers tended to be older, stronger, and male, and victims tended to be younger, ~veaker, and female. Whether low self-esteem dictated the preference for weaker victims is quite unclear, however. All that was clear was that the most common situation to produce sibling violence was when older siblings were left in charge of younger ones who then defied their au- thority in some way, and so the older siblings used violence to assert their authority, gain compliance, or simply intimidate the younger ones. This finding seems most consistent with the pat- tern we have already seen multiple times of threatened but fa- vorable (in this case situational) views of self. The older, abusive sibling presumably feels entitled to superior stature and author- ity (as conferred by the parents) and feels this superior stature is threatened when the younger sibling fails to obey or comply. Still, this pattern is merely situational, and one cannot draw any clear conclusions about the trait levels of self-esteem among abusers. It is plausible that abusers regard themselves as supe- rior and hence entitled to hit or hurt their weaker, supposedly inferior siblings. It seems less plausible a priori to suggest that older siblings left in charge by their parents feel inferior to their younger siblings (as the low self-esteem view might suggest), but it cannot be ruled out empirically. Critique. Many studies of intergenerational domestic vio- lence are methodologically weak, and very little evidence of any sort exists with regard to sibling violence. Still, some fairly rig- orous work has been done, particularly with regard to parents who abuse children. These studies have found no link between self-esteem and violence. Conclusion. Early studies found occasional support for the traditional hypothesis of low self-esteem as a cause of domestic violence, but that support has eroded in recent years as method- ologically better studies have examined the issue. More recent evi- dence seems to be moving toward a stature inconsistency explana- tion instead of a low self-esteem explanation, which would parallel the evolution of empirically based theory with regard to domestic violence between adult partners. People become violent because they refuse to accept a dependent role or because they feel that their superior role has been challenged or questioned. Some evi- dence has begun pointing to narcissism (i.e., inflated love of self) as a cause of violence against family members. All these findings seem best characterized as indicating that domestic violence arises when privately favorable views of self are impugned by external circumstances or by other people's particular, disrespectful ac- tions, but given the present state of the literature on intergenera- tional abuse that conclusion must be regarded as quite tentative. Violent Youth Gangs and Juvenile Delinquency The classic study of juvenile delinquency by Glueck and Glueck (1950) compared juvenile delinquents against a matched sample of nondelinquent boys. Although the study was an early one and has been criticized on methodological grounds, it benefited from a large sample and extensive work, and accord- ing to the focused review by Wilson and Herrnstein (1985) nearly all of their findings have been replicated by subsequent studies. The Glueck and Glueck study did not measure self- esteem directly (indeed it antedated most modern self-esteem scales), but there were plenty of related variables. The pattern of findings offers little to support the hypothesis that low self- esteem causes delinquency. Delinquent boys were more likely than control boys to be characterized as self-assertive, socially assertive, defiant, and narcissistic, none of which seems com- patible with low self-esteem. Meanwhile, the delinquents were less likely than the comparison group to be marked by the fac- tors that do indicate low self-esteem, including severe insecur- ity, feelings of helplessness, feelings of being unloved, general anxiety (a frequent correlate of low self-esteem), submissive- hess, and fear of failure. Thus, the thoughts and actions of juve- nile delinquents suggested that they held quite favorable opin- ions of themselves. As Sampson and Laub (1993) noted, it is useful to look for convergences between the Glueck and Glueck (1950) study and more recent studies of youthful violence, not only because of the seminal nature of the Gluecks' work, but also because their data were collected several decades ago and on an almost en- tirely White sample, unlike more recent studies. Converging findings thus confer especially high confidence in conclusions that can be supported across time and ethnicity. One of the most thorough research projects on youth gangs was that of Jankowski ( 1991 ), whose work involved I0 years, several cities, and 37 gangs. Although as a sociologist he was disinclined to use self-esteem or personality factors as explana- tory constructs, his study did furnish several important obser- vations. Jankowsld specifically rejected the notion that acting tough is a result of low self-esteem or feelings of inadequacy. In his words, "There have been some studies of gangs that suggest that many gang members have tough exteriors but are insecure on the inside. This is a mistaken observation" (p. 27). He said that for many members, the appeal of the gang is the positive respect it enjoys in the community as well as the respectful treatment from other gang members, which he found to be an important norm in nearly all gangs he studied. He said most gang members "expressed a strong sense of self-competence and a drive to compete with others" (p. 102). When they failed, they always blamed something external rather than personal inade- quacy or error. This last observation is especially relevant be- cause several controlled studies have shown that such behavior is characteristic of high self-esteem and contrary to the typical responses of people with low self-esteem (Fitch, 1970; Ickes & Layden, 1978; Tennen & Herzberger, 1987; see also Kernis, Brockner, & Frankel, 1989). Jankowski's ( 1991 ) characterization of the personal attitudes and worldviews of gang members likewise contains indicators of egotism. He said gang members tended to believe that their parents had capitulated to accept a humble life of poverty and failure, which they refused to do themselves. Mainly, he found, gang members were violent toward people "whom they per- ceived to show a lack of respect or to challenge their honor" (p. 142). Another main cause of gang violence is personal ambi- tion: People behave violently to enhance their stature in the or- BAUMEISTER, SMART, AND BODEN ganization (and harm or discredit, their rivals). The conclusion that the more ambitious people become the more violent ones is difficult to reconcile with the view that violence arises from low self-esteem. Another sociologist, E. Anderson (1994), summarized his observations on Black street gangs by noting that the code of the streets centers around "respect," which gangs regard as an external quality involving being "granted the deference one de- serves" (p. 82 ). Thus, his analysis indicates that gang members believe they deserve to be treated as superior beings--that is, their self-esteem is high--but they are constantly vulnerable to external circumstances that may dispute or fail to recognize their superiority. He said that gang members learn early in life that humility (which is one of the concepts linked to modesty and low self-esteem) is not a virtue, whereas people who win lights also gain the admiration of others. In fact, gang members have often been socialized with lessons that underscore the ne- cessity of correcting someone who shows disrespect. The high level of violence among modern youth gangs is partly due to what E. Anderson (1994) described as a zero-sum aspect, such that prestige and respect are gained by depriving others of them, and so people may look for fights or conflicts as a way to assert their superiority over others. Anderson also said that respect is enhanced by what he calls "nerve" which is es- sentially a matter of acting as if oneself is above the rules that apply to others and as if one disregards the rights of others. In our view, both of these elements of nerve imply a view of one's own superiority and thus should be linked to high self-esteem. In a similar vein, Katz (1988) noted that many youthful circles and street subcultures afford respect mainly to the "badass" sort of person who transcends the pressures to conform to soci- etal norms, rationality, and ideals. This prized identity is culti- vated in part by creating the impression of being unpredictably prone to chaos and irrational violence. The zero-sum hypothesis may help explain the frequent prov- ocations that lead to violence; after all, if showing disrespect often elicits violence, why would people ever show disrespect to youthful gang members? Yet it is clear that many people do, and in particular gang members insult members of other gangs. According to Horowitz and Schwartz (1974), most of what passes for gang violence is actually a matter of conflicts between individuals, and full-scale gang battles are quite rare (see also Jankowski, 1991 ). Still, the group is the operative unit in many cases. Violence is typically precipitated when one person im- pugns the honor or dignity of the other, most commonly by an insult, but also by any violation of etiquette. The code of honor is central to gang life, and gang members regard their own group as superior; the insults are often spoken as a way of asserting the superiority of one's own group, to which the insulted party must respond by defending the esteem of his ( or sometimes her) group. If the zero-sum view is correct, then derogating rival groups would be perceived as an effective way of asserting and boosting the esteem of one's own group. Indeed, one of Katz's (1988) most provocative arguments is that youth gangs, which bring early deaths to so many of their members, actually have a positive investment in sustaining community violence, because that violence offers them a re- spectable justification for existence (as protection against the dangerous urban environment), without which the gang would seem a mere childish association. The pervasiveness of violence thus helps support the gang members' egotism by transforming them from a club of trouble-making boys into a prestigious corps of warriors defending their community. In any case, intergang violence thus comes to revolve around competing claims to be members of a privileged elite, and the occasion for violence is often a merely symbolic aspersion that the rival group's claims are unfounded (such as by making a humorous verbal insult or writing the name of one's gang in the home territory of the other gang). The gang members' preoccu- pation with respect (as in the common neologism dissing) re- flects the ongoing tension between private, exalted views of self (which are shared and supported by the gang) and public per- ception of themselves as potentially falling short. McCall's (1994) recent firsthand account of his own violent youth em- phasizes the concern with maintaining respect by putting down others and violently preventing others from showing disrespect to oneself, and other accounts make similar points (Bing, 1991; Currie, 1991 ). Studies of adult gangs show similar patterns. Members of or- ganized crime tend to regard themselves as superior beings and to command deferential, respectful treatment from others (Anastasia, 1991; Arlacchi, 1992). Likewise, studies of prison gangs have observed that they form along racial and ethnic lines and hold explicit ideologies of their own racial or ethnic superi- ority, which is intensified as they come to operate as an elite group within the prison (Camp & Camp, 1985; Lyman, 1989). Critique. Studies of juvenile delinquents and youth gangs have generally lacked the rigor of experimental studies, but they have used a variety of observational and occasionally quantita- tive measures. Researchers from different disciplines concur in depicting these young men as egotistical in several ways, and they concur emphatically on the apparent preoccupation with respect and self-assertion. They also agree that insults or other disrespectful treatment tends to lead to violence. In view of this convergence, it seems reasonable to accept these conclusions, until or unless contradictory evidence can be marshaled. The evidence about organized crime and prison gangs is largely ex- ploratory and impressionistic and should be regarded as preliminary. Conclusion. Although standardized measures of self-esteem have generally been lacking from studies ofjuvenile delinquents and gang members, there are ample indications of egotism from those studies. Gang members apparently think, talk, and act like people with high self-esteem, and there is little to support the view that they are humble or self-deprecating or even that they are privately full of insecurities and self-doubts. Violent youths seem sincerely to believe that they are better than other people, but they frequently find them- selves in circumstances that threaten or challenge these beliefs, and in those circumstances they tend to attack other people. It also ap- pears that they sometimes manipulate or seek out such challenges to their esteem, in order to enhance their esteem by prevailing in a violent contest. Similar patterns have been observed in adult crimi- nal gangs, but more research is needed. Political Terror (Government Repression, Terrorism, and War) Political organizations perpetrate a great deal of violence. In this section we shall examine terrorism, government repression, assassination, and war. Genocidal activities may also be consid- EGOTISM AND VIOLENCE ered as political violence, but they will be covered in the follow- ing section. An immense amount of suffering has resulted from internal re- pressive campaigns mounted by tyrannical governments. Chirot (1994) provided an authoritative global survey of 20th-century tyrannies. As a political scientist, his primary interest was in polk- ieal structures and developments as causes of tyranny, but he con- duded that threatened collective egotism was often an important factor in leading to tyranny. In case after case he examined, nations or national groups developed a strong ideology that em- phasized their own superiority to other nations and groups. Tyran- nies typically emerged when the ideology of superiority was ac- companied by the perception that they did not receive the respect that was due them. As a fairly extreme example, prior to the Com- munist takeover Russians felt themselves culturally and morally sui~ior to the decadent nations of Western Europe, but they be- lieved that accidental advantages of material wealth and military innovation had given the Westerners the edge and prevented Rus- sia from taking its rightful place of leadership. This made Russians receptive to the emer~ace of a strong government that took the form of the Communist tyranny and whose internal terrors and purges still hold the record for the most killing by any government. Similar conclusions have been suggested by Staub ( 1985, 1989) and Ford (1985). It is also useful to examine the motives and beliefs of the indi- viduals who carry out repressive policies, although such informa- tion is relatively difficult to obtain. In particula~ the psychology of the individuals who administer torture for repressive regimes has received only intermittent study for various obvious reasons, in- eluding the secrecy surrounding the activities and the reluctance of former torturers to participate in research or tell their stories after the regime has been discredited. The methods used to train torturers should have considerable theoretical interest for the present debate. If low self-esteem leads to violence, then one would presumably train torturers by instilling feelings of inferiority and humility in them; in con- trast, if high self-esteem facilitates violence, then the most effective training would instill attitudes of elitist superiority. The evidence appears to favor the latter. Gibson and Haritos- Fatouros (1986; see also Peters, 1985) described procedures used to train torturers during the military regime in Greece in the late 1960s and early 1970s. Relevant aspects of training included telling the trainees that they were special and fortunate to be included in this elite corps, s The sense of superiority was buttressed after the training by conferring many special privi- leges and marks of stature on these men, such as allowing them to wear their hair long (unlike most soldiers) and wear civilian clothes, lending them military cars for personal use, and allow- ing them to eat free meals in good restaurants instead of dining on military fare. The authors noted that this instilling a sense of elitism and privilege was not unique to these groups, because other programs designed to train especially violent individuals (such as the U.S. Green Berets) do the same. In particular, torturers are apparently strongly indoctrinated with the view that the superior culture (as embodied in,their nation with its current regime) is threatened by evil forces, of whom the torture victims are representatives. Thus, the ideol- ogy of inherent but jeopardized superiority is acutely empha- sized among the people who must carry out the violence. Con- quest's ( 1986, 1990) observations about Russian terror and re- pression present a similar picture, although no systematic or quantitative study was involved. Whereas repressive governments have all the forces of laws, police, and state bureaucracy on their side, terrorist groups typ- ically lack all of these. Most commonly, terrorists are campaign, ing for radical political change and thus must live outside the law and in opposition to state institutions. Despite the opposite situation, they seem to share some important characteristics with tyrannical governments. In particular, they seem to culti- vate the same attitude of moral superiority over their victims and enemies (e.g., Reich, 1990). Hee's (1993) detailed account indicated that her training as a terrorist in North Korea empha- sized the pervasive belief in the moral superiority of North Ko- rea in all nonmaterial aspects over everyone else in the world. The backwardness and poverty and other disadvantages of North Korea may have constituted some ego threat, and indeed Hee reported how much dissonance she suffered during her es- pionage and terrorist missions abroad, which showed her how much better off the citizens of other countries were than North Koreans. Still, she and her peers were quite willing to perform violent acts even against unsuspecting, noncombatant citizens because of their belief in their own moral superiority. Post (1990) made a similar argument, which he exemplified with a story in which a new recruit objected to innocent people being killed if the terrorist group carried out its plan to bomb a de- partment store; the group leader patiently explained that any- one shopping in such a store must be a capitalist and hence was not innocent. We already mentioned that Long (1990) said that terrorists had low self-esteem but then provided evidence suggestive of high self-esteem (e.g., terrorists raise aspirations after failure, as do people high in self-esteem). Later in his book, Long partly contradicted his own assertion about low self-esteem by de- scribing terrorist leaders as narcissistic (p. 18 ). If the leaders-- those who are most responsible for the terrorist violencemhold the grandiose and inflated views of self that are the hhllmark of narcissism, it is hard to regard terrorism as deriving from low self-esteem. Two additional forms of political violence, namely assassina- tion and warfare, deserve brief mention. Assassination has al- ways been quite unusual, but its importance makes it worth considering despite its rarity. Ford's (1985) history does not, however, provide much indication of either high or low self- esteem as a prominent characteristic of assassins. This form of violence may well be a product of concerns and causes that do not include self-esteem. Unlike assassination, war has been extremely common; in- deed, Sluka (1992) summarized various estimates that there have been approximately 14,000 wars since 3600 B.c., and the four decades following World War II contained only 26 days of s To be sure, during the training phase there was a period during which the initiates received humiliating treatment, but this appears to be standard for many military training regimens and indeed has been proposed by Aronson and Mills (1959) as an effective aid toward build- ing strong emotional ties to the group. It is perhaps most precise to say that a preparatory phase of humility during training is followed by a more permanent phase of belonging to a superior elite. In any case, what matters is that by the time these men began their duties as torturers they had been led to regard themselves as special, superior individuals. BAUMEISTER, SMART, AND BODEN world peace. 9 Generalization is therefore quite hazardous. Still, recent and salient evidence seems hard to reconcile with the view that low self-esteem (as in lack of national pride) prompts nations to go to war. It is difficult to characterize imperial Ja- pan, Nazi Germany, or Hussein's Iraq, for example, as suffering from low self-esteem; rather, such cases seem to fit the pattern of excessively favorable views of self that produce dreams of glory and anger that the rest of the world fails to pay sufficient respect. Staub ( 1985 ) concluded that cultural attitudes of supe- riority are important causes of warfare and other violence. If we examine war from the perspective of the individuals who carry it out rather than from the perspective of national ideol- ogy, once again there seems ample evidence of egotism. Keegan (1993) has concluded that professional soldiers, from the Ro- mans to the present, were not generally attracted and sustained in military life by financial gain but rather by pride in belonging to a valued group, concern over winning admiration and fellow- ship of colleagues, accumulation of honor, and largely symbolic recognitions of success. Recent efforts to understand the attitudes that make people favorably inclined toward war have been summarized by Fesh- bach (1994). In his research program, two sets of attitudes stood out (see Kosterman & Feshbach, 1989). He called the first of these patriotism, which he explained chiefly in terms of attachment feelings, although some element of pride is in- volved. The second attitude he referred to as nationalism, which he explicitly defined in terms of belief in the superiority of one's nation over others. Both of these attitudes are positively related to militaristic attitudes, but nationalism shows much stronger relationships to prowar and pronuclear attitudes. Nationalism is also positively correlated with individual aggressive tenden- cies. These results indicate that feelings of collective superiority are linked to violent, militaristic inclinations, ranging from per- sonal conflicts to nuclear war. Critique. Most of the work reviewed in this section was done by historians, sociologists, and political scientists. When judged by psychologists' standards of methodological rigor, this work is relatively weak, but when judged on its own terms it fares better. Moreover, the convergence of evidence across different disciplines helps rule out the danger that disciplinary biases or methodological artifacts have shaped the conclusions. Conclusion. Except for assassination, it appears that politi- cal violence is often correlated with ( and preceded by) strongly favorable self-regard and the perception that these views are threatened or disputed by others. In most cases it is the collec- tive self-perception of superiority that is involved. Some signs indicate that individuals who carry out political violence are either indoctrinated with the view of their own superiority or marked by narcissistic traits. Psychologists may question the methodological rigor of these studies, but the conclusion does seem consistent with the general patterns we have already seen in other spheres, and interdisciplinary convergence is itself a persuasive indicator. The only contrary view was Long's (1990) characterization of terrorists as having low self-esteem, but as we noted his elaboration seemed to indicate high self-esteem after all. Prejudice, Oppression, and Genocide It would seem that the argument in favor of high self-esteem would be relatively easy to make in the case of prejudice and OP- pression. Thus, in the United States, there was until recently a long tradition of general discrimination and oppression against Black citizens, and it would be difficult to argue that these things oc- curred because White people believed they were inferior to Blacks. By the same token, the most discussed and sensational pattern of genocide in the modem era was the extensive murder of Jews by Nazi-dominated, traditionally Christian Germany, and it would require some rather severe stretches of the imagination to contend that the Nazis believed themselves inferior to the Jews (whom they denounced as "vermin" ). Indeed, the Nazis styled themselves the "master race," a label that seems hard to reconcile with a theory that they held a low opinion of themselves. Racist prejudice in the United States appears to have had its major origin in the period during which the majority of Black Americans were slaves. In nearly all societies that have practiced slavery, slaves ranked at the bottom of the social hierarchy of self- worth, and the lowest ranking flee people (including freed slaves ) were oRen very concerned to establish their superiority to slaves (Patterson, 1982). Indeed, in America's Civil War, a problem faced by the southern aristocrats was how to enlist the support of the poor White population for a war that offered it little in terms of economic or political benefits, and the main solution was to appeal to these people's sense of self-esteem by pointing out that if the South were to lose the war, the Black slaves would be freed and would become the equals of these poor Whites (McPherson, 1988). Apparently the poor Whites agreed with that argument sufficiently to enroll and fight in the Civil War. The loss of the Civil War constituted a double blow to the immense pride of the southern aristocrats: First, they had been unthinkably defeated by the despised Yankees, and second, the Reconstruction governments sought to make the Blacks fully equal to all other citizens. The infamous Ku Klux Klan was founded and spread in response to these ego threats, stimulated by the perception that "insolent" Blacks now refused to treat Whites as inherently superior beings (e.g., by stepping out of the way on the sidewalk). Initially Klan activities were designed to play humiliating but otherwise harmless pranks on Blacks, as if simply to prove White superiority. Soon, however, the Klan began to become violent against two groups of targets: Black people who seemed upwardly mobile (thereby refuting White supremacy) and White people who helped Blacks or otherwise treated them as equals. Although leadership and initiative came from the upper classes, most Klan violence over the years was perpetrated by lower-class White men who presumably had the most to fear in terms of loss of stature from the notion of Black equality (Wade, 1987 ). Thus, the emergence and history of the Ku Klux Klan seem consistent with the notion that threatened egotism (in this case, firm beliefs in White supremacy that were undermined by political and socioeconomic changes) is a pow- erful cause of prejudice and related violence. Indeed, at the con- gressional hearings on the Klan in 1921, the Imperial Wizard testified that the doctrine of White supremacy was not intended as a matter of"race hatred" but rather of"race pride" (Wade, 1987, p. 164). Although the Klan has been largely discredited and driven underground, the 1980s witnessed a resurgence of racially mo- 9 Even that estimate is high, because it is based on only international wars. If civil wars were counted too, there probably would be no days of peace at all. EGOTISM AND VIOLENCE tivated group activity, most notably among disaffected young White men who formed neo-Nazi groups ("skinheads"). Hamm (1993) noted the irony of such groups emerging during the 1980s, because that era was marked by high graduation rates and low unemployment rates among young White men. Moreover, Hamm found that even within such groups, the members who participated in violent activities had higher ca- reer aspirations and higher levels of education than the nonvio- lent members. The members did, however, frequently express that they resented the advances by and preferential treatment of minorities. To integrate these observations, we would suggest that the high aspirations and sense of entitlement reflected fa- vorable views of self, and these increased the vulnerability to disconfirmation and threat, as symbolized by the perceived gains by non-White minorities. Levin and McDevitt's (1993) recent work on hate crimes paints a similar portrait of young White men as disaffected by their own eroding entitlements and resenting the gains made (presumably at their expense) by minorities. Slavery itself was a major form of oppression, and indeed, it probably exceeded most other forms of prejudice and discrimi- nation. Although there were often economic motives for slavery, Patterson (1982) concluded that slavery in the American South was atypical of most slave systems in several respects, including the relatively high financial rewards it brought to slave owners. In the history of the world, Patterson said, slave owning was neither vital nor particularly helpful for the accumulation of wealth. Instead, the major appeal of slavery was that it increased the honor and prestige of the owners. (Indeed, he said that whenever the practice was allowed, slaves liked to own slaves themselves, because to do so conferred prestige on them--and slaves generally had few means of gaining such prestige.) Thus egotism rather than greed may often have been the major force behind the institution of slavery. Imperialism has been a major form of international oppres- sion for many centuries, although it reached a peak late in the 19th century as most of the main Western industrial nations sought overseas colonies. Although there were clearly economic motives for imperialism, motives of national pride and self- esteem were also relevant. We assume it will not be controversial to point out that the nationalistic attitudes supporting imperi- alism were essentially those of high collective self-esteem and even narcissism. One place to look for evidence would be in the so-called scramble for Africa during the 19th century, which formed a kind of climax to European imperialistic projects (and indeed the subsequent decolonization of Africa signified the end of the imperialistic era). Pakenham ( 1991 ) provided a detailed account of this conquest of Africa. On the one hand, economic greed was undoubtedly a factor; but the economic promise of many colonies was never entirely spelled out, and in retrospect it is clear that most colonies brought net financial losses (often severe) rather than gains. Pakenham concluded that national prestige was often associated with size of empire, and people wanted their nation's empire to expand regardless of financial prospects. Several of the military confrontations over obscure swamps or disease-infested wastelands can hardly be explained on any basis except national pride. In particular, ter- ritorial acquisitions by one of the principal rivals (especially France and England) often produced consternation among the others, who felt that their collective self-esteem was in jeopardy unless they could match or surpass them. Dower (1986) provided considerable historical evidence that the Pacific segment of World War II was seen by both sides as a race war. This racial dimension to the conflict led to more ex- treme derogation of enemies and much higher levels of atrocity (on both sides) than were seen in the European war. (A parallel pattern can be found in the European war, however, insofar as the Germans treated captured British and American soldiers, whom they regarded as fellow Aryans, much better than the Russians and other soldiers, whom they regarded as inferior races.) The role of collective self-esteem in the Pacific war was quite clear: Both sides (i.e., the United States and Japan) re- garded themselves as racially superior but as threatened by the successes and evil intentions of the others. As evidence that the collective superiority went deeper than mere rhetoric, Dower cited multiple examples of strategic errors that were based in underestimating the enemy because of the assumption that the enemy, as an inferior race, would be militarily stupid and in- competent. For example, the United States did not believe the inferior Japanese capable of the strike on Pearl Harbor, and even after the fact it was often assumed that the Germans must have planned the attack for the Japanese! On the other side, the Japanese thought that the decadent Westerners would not be willing to endure the hardship of a protracted war and so would give up easily, perhaps even right after Pearl Harbor. Genocide is undoubtedly the most sinister form or manifes- tation of prejudice. Staub's (1989) psychological analysis of four major genocides repeatedly referred to the aggressor's sense of being superior and being better, which is often aggravated by threatening conditions. In each of the four cases, the genocides were perpetrated by nationalities and regimes with strong be- liefs in their own innate superiority but that had suffered some threat or blow to their sense of superiority. (Moreover, within the society, those responsible for the killing, such as Hitler's SS in Germany and the military in Argentina, constituted a privi- leged, respected elite.) Large-scale mass murder emerged as a means of cleansing the body politic of impure, evil, decadent influences as well as a means of satisfying the wish to blame one's misfortunes on a scapegoat who could then be punished. Indeed, Staub's general conclusions point directly to threatened egotism: "When a sense of superiority combines with an un- derlying (and often unacknowledged) self-doubt, their contri- bution to the potential for genocide and mass killing can be es- pecially high" (p. 19). The combination of high collective self- esteem (and the resulting "sense of entitlement") with the re- cent threats, blows, or losses provides a "belief in unfulfilled greatness" (p. 234), which constitutes an important precondi- tion and motivation for genocide. There have been some controlled studies of the relation of self-esteem to prejudice. The traditional theory was, as usual, that low self-esteem causes prejudice, and so it was assumed that people who lack self-esteem would be the most prejudiced against others. Preliminary work seemed to fit this pattern (e.g., Stephan & Rosenfield, 1978; see Wills, 1981, for review), al- though the evidence was limited, indirect, or ambiguous. Crocker and Schwartz ( 1985 ) showed, however, that this sem- blance of derogating others reflects the general negativity of peo- pie with low self-esteem: Although they may rate out-groups negatively, they rate other people (and themselves) negatively BAUMEISTER, SMART, AND BODEN too, and so it is misleading to say that they are prejudiced against out-groups. A more precise picture of the role of self-esteem in prejudice has emerged from more recent work by Crocker and her col- leagues. Crocker, Thompson, McGraw, and Ingerman (1987) showed a pattern of responding to ego threats by showing more in-group bias (i.e., rating one's own group more favorably in comparison to other groups). Only people with high self-esteem showed this pattern. Likewise, Crocker and Luhtanen (1990) showed that people who scored high in collective self-esteem (i.e., favorable self-evaluation based on one's memberships and affiliations, as opposed to individual self-esteem) showed the same pattern of derogating out-groups in response to threats to their egotism (in this case, being told that the group to which they belonged had performed poorly on a test and was therefore insensitive, immature, and suffered some cognitive and alfective deficits). Although the effects in these studies were not uniform across all measures and conditions, the bulk of their findings suggests that prejudicial responses may be strongest among peo- ple with high self-esteem and particularly when such people are subjected to ego threats. Critique. As already noted, early studies suggested that low self-esteem was linked to prejudice, but more recent and careful studies have reached the opposite conclusion, and the early work appears to have suffered from methodological and interpretive flaws. The studies on prejudice and violence come from multiple fields and point to similar conclusions. Conclusion. Current research has suggested that racial and ethnic prejudice accompany favorable views of self. Meanwhile, abundant evidence from across several disciplines confirms the view that intergroup violence is often linked to prejudiced views that typically depict the in-group as superior to the out-group. Furthermore, some evidence suggests that the most severe vio- lence occurs when a group perceives that its superior position is being eroded or threatened by the rise of a rival group. General Discussion Several main conclusions can be drawn from our survey of relevant empirical evidence. It must be noted that direct, pro- spective studies linking sophisticated measures of self-appraisal to real violence have been quite rare, and so it has been neces- sary to look for converging evidence from diverse sources and multiple methods. The volume and diversity &the evidence are necessary to compensate for the lack of unambiguous, rigorous work focused on the hypotheses. With a topic as full of ethical, practical, and theoretical complexities as violence, this problem may be inevitable. The traditional view that low serf-esteem is a cause of violence and aggression is not tenable in light of the present evidence. Most studies failed to find any support for it, and many provided clear and direct contradictory findings. Aggressors seem to believe that they are superior, capable beings. Signs of low self-esteem, such as self-deprecation, humility, modesty, and self-effacing mannerisms, seem to be rare (underropresented) among violent criminals and other aggressors. The typical, serf-defining statements by both groups and individuals who aggress indicate a belief in their supe- riority, not inferiority. Violent and criminal individuals have been repeatedly characterized as arrogant, confident, narcissistic, ego- tistical, assertive, proud, and the like. By the same token, violent, aggressive, and criminal groups tend to share beliefs in their own superiority, ranging from the "man of honor" designation of Mafia initiates to the "master race" ideology of the Nazis. Also, from individual hate crimes to genocidal projects, violence that is linked to prejudice is ~erally associated with strong views that one's own group is superior and the out-group is inferior, even subhuman. We suggested that domestic violence might be the one sphere in which the low self-esteem view would fare best. That sphere was indeed the only one in which supportive findings (i.e., link- ing low self-esteem to violence) were reported, but even those tended to be weak, exceptional, and contradicted by the findings of more careful and systematic studies. The possibility that peo- ple with low self-esteem may sometimes choose relatively weak and helpless targets as victims remains plausible although even it cannot be asserted as correct at present. The rejection of the low self-esteem view does not mean, how- ever, thathigh self-esteem is a cause of violence. Most bullies, violent criminals, and other aggressors seem to think highly of themselves, but it is not true that most people who think highly of themselves are violent. The most precise conclusion appears to be that violence is perpetrated by a small subset of people with favorable views of themselves. Or, to put it another way, violence is produced by a combination of favorable self- appraisals with situational and other factors. The most important situational factor that interacts with fa- vorable self-appraisals to cause violence is an ego threat. The evidence conformed broadly to the view that violence is often caused by an encounter in which a favorable self-appraisal is confronted with an external, less favorable evaluation. In all spheres we examined, we found that violence emerged from threatened egotism, whether this was labeled as wounded pride, disrespect, verbal abuse, insults, anger manipulations, stature in- consistency, or something else. For huge nationalities, medium and small groups, and lone individuals, the same pattern was found: Violence resulted most commonly from feeling that one's superiority was somehow being undermined, jeopardized, or contradicted by current circumstances. We do not wish to claim that threatened egotism is the sole cause of a~ession, and indeed there is ample room to discuss biochemical or genetic causes, modeling effects, instrumental ag- gression, and other factors. But in terms of the potent link between self-appraisals and violence, the discrepancy between favorable self-views and external threats is the most important cause. The theory that the discrepancy between self-appraisals and external evaluations causes violence led to the further predic- tion that violence would be increased by anything that raised the frequency or impact of such discrepancies. We proposed that inflated or unrealistically positive self-appraisals would tend to lead to violent responses, because to the extent that feed- back clusters around accurate, realistic appraisals, it will tend to contradict such unrealistically favorable opinions of self. There was moderate support for that view, including evidence about tyrants, career criminals, psychopaths, and convicted rapists. Also, some of the most effective direct predictors of vi- olence were narcissism scales, particularly subscales for grandi- osity and exhibitionism. It remains to be determined how these self-enhancing illusions compare with the positive illusions of nonviolent people and how widely disseminated they are. For the present, however, it seems reasonable to accept the view that EGOTISM AND VIOLENCE inflated, overly positive self-appraisals are associated with violence. Another moderator we proposed was that unstable or uncertain beliefs about the serf's good qualities should be especially vulner- able and sensitive to external ego threats. Again, a broad assort- ment of evidence fit this view, but further work is needed. There was direct evidence that unstable, high self-esteem is most closely linked with hostility (Kernis, Grannemann, et al., 1989). The affective component of the theory is relatively straight- forward. It is clear that ego threats elicit negative affect and that negative affect can lead to violence. The evidence is less clear as to whether anger represents a defensive effort to ward off other forms of negative affect that might follow from accepting the bad feedback, but that hypothesis remains plausible. Mean- while, we predicted from the affect theory that severely violent reactions would sometimes follow from seemingly minor or trivial ego threats, and this prediction was confirmed by multiple observations in various investigations. The interpersonal framework offers relevant insights. It would be misleading to suggest that the experience of discrepant self-appraisals causes an aimless eruption of aggressive im- pulses. Rather, aggression is most commonly directed at the source of the unflattering evaluation, and so it makes sense to regard many aggressive acts as communicative responses to un- welcome, disputed appraisals. Also, some sources provided di- rect confirmation of the view that aggression is a means of dom- inating another person and hence symbolizing one's superiority over that person. On the basis of the zero-sum view of self-esteem, we predicted that one person's positive self-assertions could constitute a threat to (and elicit a violent response from) others, and there was some evidence consistent with that view, especially in the selection of targets based on their presumptive feelings of supe- riority. Still, this seems to be an unusual pattern rather than the norm. The view that zero-sum esteem pressures are exceptional and circumscribed phenomena would be consistent with Feath- er's (1994) conclusion about the "tall poppy" effect, namely that it too only occurs under specific, limited, and unusual circumstances. Is It Really High Self-Esteem? A reevaluation of the relationship between self-appraisals and violence is clearly warranted. Indeed, it seems overdue: It is sur- prising that the low self-esteem view has survived so long, and one wonders if there were not something correct about it to al- low it to endure with so little direct support. Is there any way to salvage the view that low self-esteem contributes to violence? And do narcissistic, inflated, arrogant self-appraisals really con- stitute high self-esteem? To be sure, definitions of self-esteem may vary. We have used the term in a broad and inclusive sense to encompass all favor- able self-appraisals, including confidence and self-respect as well as arrogance and narcissism. In contrast, some might pre- fer to define self-esteem in a way that would eliminate all dis- tasteful and problematic forms, and if this were possible then it might be plausible to deny that high self-esteem leads to vio- lence. It is difficult to see, however, what basis other than the mere value judgment itself might be used to differentiate benign from malignant self-esteem. (Obviously, if high self-esteem is defined in a way that stipulates that it can only produce positive, desirable consequences, then it cannot lead to violence or ag- gression, but this is circular.) In our view, the heavily positive connotation that self-esteem has acquired in recent American thought is partly a result of biased and wishful thinking that simply refuses to acknowledge the darker side. If one remains with the simple, literal definition of self-esteem as a favorable appraisal of oneself, than arrogant narcissists and conceited, egotistical bullies do indeed have high self-esteem. A more subtle line of reasoning might propose that the super- ficially favorable self-views of conceited and other violent indi- viduals are actually defensive reactions that are designed to con- ceal unfavorable self-appraisals. Possibly these are defensive versions of high self-esteem, underneath which lies a hidden but truly low self-esteem. Theorists wishing to make this argument might be encouraged by the evidence we have reviewed suggest- ing that not all people with high self-esteem are violent. If only a subset of people with high self-esteem are violent, might this subset consist of people for whom high self-esteem is a false ve- neer to cover up low self-esteem? If so, then one might yet find a way to argue that low self-esteem is a cause of violence. In other words, perhaps some people who regard themselves unfa- vorably become self-assertive and violent as a result, possibly as a way of compensating for this sense of inferiority. Because this theory enjoys the luxury of being able to interpret contrary evi- dence as meaning the opposite of what it literally signifies, it is difficult to disprove. In other words, if favorable self-assertions are taken as signs of low self-esteem, then the hypothesis of low self-esteem is difficult to falsify. Still, there is some relevant evidence. The pattern of respond- ing to bad feedback with defensively positive assertions about the self--which Long (1990) observed among terrorists and Jankowski (1991) mentioned with violent gangs--has been shown in laboratory studies, but it is characteristic of high rather than low self-esteem (Baumeister, 1982; Baumeister et al., 1993; McFadin & Blascovich, 1981 ). A method of distin- guishing high from merely defensive high self-esteem was pub- lished in 1975 (Schneider & Turkat, 1975), but researchers have not identified very many patterns in which the two groups differ, and the lack of such findings seems to indicate that the pattern of positive self-assertion despite privately low self- appraisal is relatively rare. Moreover, as we noted, a number of researchers specifically contradicted the view that the violent individuals they studied were secretly suffering from inferiority complexes or self-loath- ing (e.g., Jankowski, 1991; Olweus, 1994). The basis for these conclusions was not reported, but then again one wonders what sort of basis might be fully satisfactory, given the difficulty of falsifying such a hypothesis. Still, one researcher who made such an assertion (Jankowski, 1991 ) had spent over a decade living among gangs and getting to know hundreds of gang mem- bers, and it seems fair to assume that he would have seen ample evidence of their inner low self-esteem if it existed. There is also a fundamental conceptual problem with the ap- proach of saying that low self-esteem is often concealed beneath a veneer of high self-esteem. Even if one believes that some peo- ple who assert high self-esteem actually have low self-esteem, low self-esteem cannot be regarded as the true cause of violence. There are plenty of people who do clearly have low self-esteem, and as we have shown, they are generally less violent than others. BAUMEISTER, SMART, AND BODEN It is quixotic to assert that egotists are actually self-doubters as a way of salvaging the hypothesis that self-doubters are the violent ones, given the nonviolence of most self-doubters. At best, one would have to concede that individuals with overt low self-esteem are nonviolent and therefore only those with covert low self-esteem are violent. But if one accepts that only the covert version of low self-esteem leads to violence, then seemingly one has already conceded the role of high self-esteem as decisive. In other words, the crucial distinction is between people who admit to having low self-esteem and those whose (putative) low self-esteem is concealed by some veneer of high self-esteem. Insofar as only the latter group are violent, then the decisive factor would be the veneer of high self-esteem. The favorable self-appraisal would thus still be the cause of violence, even if it did coexist with some hidden, unfavorable self- appraisal. We have seen that violence is most common when favorable self-appraisals are threatened, and such episodes might cause the individual to entertain doubts (at least temporarily) about the favorable self-appraisals. We have proposed that violence is a means of evading such doubts and affirming the favorable views of self, but it is plausible that the aggressors did suffer doubts at least momentarily, and some might propose that the doubts were the impetus for the violence. If one can refer to these self-doubts as low self-esteem, then perhaps a very wa- tered-down version of the low self-esteem theory might be up- held after all. Yet that conclusion would be seriously misleading. The oper- ative, indeed decisive beliefs about the self are the highly favor- able ones. Self-doubts only lead to violence in the context of some commitment to highly favorable self-appraisals. The self- doubt point is perhaps best understood in the context of the repeated evidence that inflated or uncertain views about the self were the views most strongly linked to violent action. The com- posite prototype of the aggressor that emerged from our review of the literature is a man whose self-appraisal is unrealistically positive. His exaggerated impression of his superiority is prone to encounter contrary feedback, which may cause him to doubt himself momentarily but to which he soon responds with vio- lence. It would be quite appropriate for him to feel such doubts, because after all the self-view in question is inaccurate. In the end, however, he preserves the unrealistically favorable self- appraisal by attacking the source of the ego threat. To say that he was violent because of low self-esteem is a serious distortion of the episode. Indeed, his momentary doubts seem better de- scribed as the disturbing voice of reality than as low self-esteem. On both empirical and theoretical grounds, therefore, we must reject the view that low seif-esteem causes violence. Ag- gressive, violent, and hostile people consistently express favor- able views of themselves. And even if one could document hid- den low self-esteem beneath the surface of apparently high self- esteem (for which empirical support is scant), it would still be necessary to regard the surface egotism rather than the hidden self-doubts as causally crucial. Why, then, has the low self-esteem theory persisted? One likely answer is that social scientists have failed to distinguish adequately between internal and external appraisals. Violence does ensue on receipt of bad evaluations from other people; it is only the negative self-evaluations that fail to lead to violence. Symbolic interactionism (e.g., Mead, 1934) proposed that self- views are principally derived from the feedback one receives from others, and this style of thought may have encouraged many thinkers to ignore the distinction and assume that people who are criticized by others must consequently have low self- esteem. It has taken decades for the accumulation of evidence to show that self-appraisals are only weakly related to external appraisals and that in many cases people overtly resist revising their self-appraisals in the face of external feedback (Crocker & Major, 1989; Shrauger & Schoeneman, 1979; Swarm & Hill, 1982). For present purposes, the crucial point is that threat- ened egotism is something quite different than low self-esteem. Another possible reason for the persistence of the low self- esteem view is that a broad reaction against blaming the victim (dating back to Ryan's, 1971, critique) may have encouraged writers to phrase the causes of violence in terms of self-evalua- tion rather than in terms of provocative, evaluative acts by the future victim. It may seem that confusing public esteem with self-esteem is a small error or technicality. However, the result of this error has been to promote a view that is precisely the opposite of the cor- rect one. The reason that disrespectful treatment sometimes leads to violence is that the aggrieved individual regards himself or herself quite favorably and hence is unwilling to tolerate be- ing treated in a way that fails to confirm this favorable self- regard. It is thus the favorable views of self that foster violence. Implications for Future Research Several suggestions for further work emerge from this review. Most pressing is the need for direct and careful study of the ways in which egotism leads to aggression and violence. We recom- mend that laboratory researchers turn some attention to the role of views about the selfin producing retaliation. In particu- lar, the mediating roles of emotional states and particular inter- personal contexts deserve further study. As we noted, it is simplistic to assume a direct and unmoder- ated link between self-esteem and aggression, so further experi- mentation may need more than a self-esteem scale and a bogus shock generator. The causal power of ego threats has been well established (although rarely discussed as such) in laboratory work as well as in nonlaboratory research into violence. What is needed, therefore, is systematic exploration of how particular views about the self interact with ego threats to increase aggres- sion. Both situational factors and individual differences may moderate the tendency of ego threats to produce aggression. Re- garding the latter, it may be desirable to give careful thought to how views about the self are to be assessed. As already noted, self-esteem measures alone may be less successful than mea- sures of narcissism (e.g., Raskin et al., 1991; Wink, 1991 ), sta- bility of self-esteem (Kernis, 1993), or defensive self-esteem (Schneider & Turkat, 1975 ). Another potential problem may lie in the fact that most stan- dard self-esteem measures were designed with the assumption that high self-esteem indicates healthy adjustment and good ad- aptation to life, and indeed high self-esteem scores are some- times used as a criterion measure of adjustment (e.g., Heilbrun, 1981; Kahle et al., 1980; Whitley, 1983). In our view, self- esteem should be a relatively value-neutral construct referring to positive evaluation of self, and so an effective and valid scale would identify the arrogant, conceited narcissist just as well as EGOTISM AND VIOLENCE the person who holds an unbiased appreciation of his or her own well-recognized good qualities. Not all scales may be effective in this regard, however. One apt approach may be to focus on variance shared between measures of high self-esteem and nar- cissism: Both concepts imply favorable self-appraisals, but the underlying evaluative bias about the desirability of these self- appraisals pushes in opposite directions, and so the shared vari- ance might be what remains after these opposing biases are removed. A potential complication that may obscure the relationship between egotism and violence is that most people who have gen- erally low self-esteem nonetheless have highly favorable views about themselves in certain limited domains, as Pelham (1993) has recently shown. Moreover, these individuals seem to be es- pecially sensitive and defensive about these positive self-concep- tions. Pelham's work has not examined the role of these self- conceptions in aggressive responses, but it is quite plausible that such responses would parallel the high defensiveness shown by these individuals in other domains of response. Thus, it could happen that even when people with low global self-esteem re- spond aggressively, the cause is still egotism. Further study to refine the concept of ego threat may also be warranted. We have used the term loosely to refer to any evaluation that is noticeably less favorable than the recipient's prior self-appraisal, but this may be too extensive a definition. Moreover, in many circles, ego threats are common whereas vi- olence is rare, io and so one must conclude that violence is only one among many possible responses to ego threats. Further study of what determines whether threatened egotism leads to violent or alternative responses is therefore warranted. The implication that overly favorable self-appraisals may lead to violence is relevant to recent debates over the desirability of such inflated views of self. Taylor and Brown (1988) amassed considerable evidence that positive illusions (i.e., favorable and possibly distorted views of self) are correlated with mental health and good adjustment. Others have suggested that this ev- idence is limited to fairly small distortions and that such illu- sions may chiefly be beneficial when kept within narrow limits (Baumeister, 1989b) or when confined to certain circumstances (Gollwitzer & Kinney, 1989). Colvin and Block (1994) dis- puted the entire argument and suggested that positive illusions are not adaptive. Although a full consideration of all the costs and benefits of positive illusions is beyond the scope of this arti- cle, we do note that positive illusions may be costly in terms of increasing vulnerability to external disconfirmation and hence resulting in violence. We concur with Taylor and Brown to the effect that high self-esteem does have important benefits, but to the extent that inflated self-appraisals lead one into violent encounters (with risks of injury, death, and imprisonment), they cannot be assumed to be an unmitigated good. Although we are not clinicians, it seems necessary to point out that the theoretical understanding of the causes of violence does have implications for interventions as well. If low self- esteem were really the cause of violence, then it would be thera- peutically prudent to make every effort to convince rapists, murderers, wife beaters, professional hit men, tyrants, torturers, and others that they are superior beings. From our reading of the empirical literature, however, these people are often violent precisely because they already believe themselves to be superior beings. It would therefore be more effective to direct therapeutic efforts elsewhere (e.g., at cultivating self-control), and if any modifications to self-appraisals were to be attempted, then per- haps it would be better to try instilling modesty and humility. Conclusion As compared with other cultures and other historical eras, modern America has been unusually fond of the notion that elevating the self-esteem of each individual will be best for soci- ety (e.g., see Huber, 1971 ). America is also, perhaps not coinci- dentally, one of the world's most violent societies, with rates of violent crime that far exceed even those of other modern, industrialized nations. The hope that raising everyone's self- esteem will prove to be a panacea for both individual and soci- etal problems continues unabated today (e.g., California Task Force, 1990), and indeed the allusions in the mass media to the desirability of self-esteem suggest that it may even be gaining in force. In this context, the notion that low self-esteem causes violence may have been widely appealing as one more reason to raise self-esteem. Our review has indicated, however, that it is threatened ego- tism rather than low self-esteem that leads to violence. More- over, certain forms of high self-esteem seem to increase one's proneness to violence. An uncritical endorsement of the cul- tural value of high self-esteem may therefore be counterproduc- tive and even dangerous. In principle it might become possible to inflate everyone's self-esteem, but it will almost certainly be impossible to insulate everyone against ego threats. In fact, as we have suggested, the higher (and especially the more inflated) the self-esteem, the greater the vulnerability to ego threats. Viewed in this light, the societal pursuit of high self-esteem for everyone may literally end up doing considerable harm. ~o For example, among authors of rejected manuscripts! References Adler, J. ( 1994, August 15 ). Murder: A week in the death of America. Newsweek, 124(7), 24-43. Allgood-Merton, B., Lewinsohn, E M., & Hops, H. (1990). Sex differ- ences and adolescent depression. Journal of Abnormal Psychology, 99, 55-63. Altemeier, W. A., O'Conner, S., Vietze, P. M., Sandier, H. M., & Sher- rod, K. B. (1983). Antecedents of child abuse. Journal of Pediatrics, 100, 823-829. Airman, J. H., & Wittenborn, J. R. (1980). Depression-prone person- ality in women. Journal of Abnormal Psychology, 89, 303-308. American Psychiatric Association. (1994). Diagnostic and statistical manual of the mental disorders (4th ed.). Washington, DC: Author. Anastasia, G. ( 1991 ). Blood and honor." Inside the Scarfo mob---The Mafia's most violent family, New York: Morrow. Anderson, E. ( 1994, May). The code of the streets. Atlantic Monthly, 273(5), 81-94. Anderson, S. C., & Lauderdale, M. L. (1982). Characteristics of abusive parents: A look at self-esteem. ChildAbuse and Neglect, 6, 285-293. Arlacchi, P. (1992). Men of dishonor: Inside the Sicilian Mafia (M. Romano, Trans.). New York: Morrow. Aronson, E., & Mills, J. (1959). The effect of severity of initiation on liking for a group. Journal of Abnormal and Social Psychology,, 59, 177-181. Banaji, M. R., & Steele, C. M. (1989). Alcohol and self-evaluation: Is a social cognition approach beneficial? Social Cognition, 7, 137-151. BAUMEISTER, SMART, AND BODEN Baron, R. A. (1976). The reduction of human aggression: A field study on the influence of incompatible responses. Journal of Applied Social Psychology, 6, 95-104. Baumeister, R. E (1982). Self-esteem, self-presentation, and future in- teraction: A dilemma of reputation. Journal of Personality, 50, 29- 45. Baumeister, R. E (1986). Identity: Cultural change and the struggle for self. New York: Oxford University Press. Baumeister, R. E (1989a). Masochism and the self. Hillsdale, NJ: Edbaum. Baumeister, R. E (1989b). The optimal margin of illusion. Journal of Social and Clinical Psychology,, 8, 176-189. Baumeister, R. E (1993). Understanding the inner nature of low self- esteem: Uncertain, fragile, protective, and conflicted. In R. Baumeis- ter (Ed.), Self-esteem: The puzzle of low self-regard ( pp. 201-218 ). New York: Plenum Press. 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, & Jones, E. E. (1978). When self-presentation is con- strained by the target's knowledge: Consistency and compensation. Journal of Personality and Social Psychology, 36, 608-618. Baumeister, R. E, StillweU, A. M., & Hcatherton, T. E (1994). Guilt: An interpersonal approach. Psychological Bulletin, 115, 243-267. Baumeister, R. E, & Tice, D. M. (1985). Self-esteem and responses to success and failure: Subsequent performance and intrinsic motiva- tion. Journal of Personality, 53, 450-467. Baumeister, R. E, & Tice, D. M. (1990). Anxiety and social exclusion. Journal of Social and Clinical Psychology, 9, 165-195. Baumeister, R. E, Tice, D. M., & HuRon, D. G. (1989). Self-presenta- tional motivations and personality differences in self-esteem. Journal of Personality, 5 L 547-579. Berkowitz, L. (1978). Is criminal violence normative behavior? Hostile and instrumental aggression in violent incidents. Journal of Research in Crime and Delinquency, 15, 148-161. Berkowitz, L. (1989). Frustration-aggression hypothesis: Examination and reformulation. Psychological Bulletin, 106, 59-73. Bing, L. ( 1991 ). Do or die. New York: HarperCollins. Blaine, B., & Crocker, J. (1993). Self-esteem and self-serving biases in reactions to positive and negative events: An integrative review. In R. Baumeister (Ed.), Self-esteem: The puzzle of low self-regard(pp. 55- 85). New York: Plenum Press. Brearly, H. C. (1932). Homicide in the United States. Chapel Hill: Uni- versity of North Carolina Press. Brown, G. W., Andrews, B., Harris, T., Adler, Z., & Bridge, L. (1986). Social support, self-esteem, and depression. Psychological Medicine, 16, 813-831. Brown, G. W., & Harris, T. (1978). The social origins of depression: A study of psychiatric disorder in women. London: Tavistock. Brownmiller, S. (1975). Against our will: Men, women, and rape. New York: Simon & Schuster. Burke, P. J., Stets, J. E., & Pir0g-Good, M. A. (1988). Gender identity, self-esteem, and physical and sexual abuse in dating relationships. Social Psychology Quarterly, 51, 272-285. Bushman, B. J., & Cooper, H. M. (1990). Effects of alcohol on human aggression: An integrative research review. Psychological Bulletin, 107, 341-354. 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. Camp, G. M., & Camp, C. G. (1985). Prison gangs: Their extent, na- ture, and impact on prisons. Washington, DC: U.S. Department of Justice. Chirot, D. (1994). Modern tyrants: The power and prevalence of evil in our age. New York: Free Press. Christensen, M. J., Brayden, R. M., Dietrich, M. S., McLaughlin, E J., Sherrod, K. B., & Altemeier, W. A. (1994). The prospective assess- ment of self-concept in neglectful and physically abusive low-income mothers. Child Abuse and Neglect, 18, 225-232. Claes, J. A., & Rosenthal, D. M. (1990). Men who batter women: A study in power. Journal of Family Violence, 5, 215-224. Cofer, D. H., & Wittenborn, J. R. (1980). Personality characteristics of formerly depressed women. Journal of Abnormal Psychology, 89, 309-314. Colvin, C. R., & Block, J. (1994). Do positive illusions foster mental health? An examination of the Taylor and Brown formulation. Psy- chological Bulletin, 116, 3-20. Conquest, R. (1986). The harvesl'of sorrow: Soviet collectivization and the terror-famine. New York: Oxford University Press. Conquest, R. (1990). The great terror: A reassessment. New York: Ox- ford University Press. Cooley, C. H. (1964). Human nature and the social order. New York: Schocken. ( Original work published 1902) Crocker, J., & Luhtanen, R. (1990). Collective self-esteem and ingroup bias. Journal of Personality and Social Psychology, 58, 60-67. Crocker, J., & Major, B. (1989). Social Stigma and self-esteem: The self- protective properties of stigma. PsychologicalReviev~, 96, 608-630. Crocker, J., & Schwartz, I. (1985). Prejudice and ingroup favoritism in a minimal intergroup situation: Effects of self-esteem. Personality and Social Psychology Bulletin, I I, 379-386. Crocker, J., Thompson, L. L., McGraw, K. M., & Ingerman, C. (1987). Downward comparison, prejudice, and evaluations of others: Effects of self-esteem and threat. Journal of Personality and Social Psychol- ogy, 52, 907-916. Currie, E. ( 1991 ). Dope and trouble: Portraits of delinquent youth. New York: Pantheon Books. Dadey, J. M., & Goethals, G. R. ( ! 980). People's analyses of the causes of ability-linked performances. In L. Berkowitz (Ed.), Advances in experimental socialpsychology (Vol. 13, pp. 1-37). New York: Aca- demic Press. De La Ronde, C., &Swann, W. B. (1993). Caught in the crossfire: Positivity and self-verification strivings among people with low self- esteem. In R. Baumeister (Ed.), Self-esteem: The puzzle of low self- regard(pp. 147-165 ). New York: Plenum Press. Diamond, D. L., & Wilsnack, S. C. (1978). Alcohol abuse among lesbi- ans: A descriptive study. Journal of Homosexuality, 4, 205-216. DoUard, J., Doob, L., Miller, N., Mowrer, O., & Sears, R. (1939). Frus- tration and aggression. New Haven, CT: Yale University Press. Dower, J. W. (1986). War without mercy." Race and power in the Pacific war. New York: Pantheon Books. Dutton, D. G., & Hart, S. D. (1992). Risk markers for family violence in a federally incarcerated population. International Journal of Law and Psychiatry, 15, 101-112. Eagly, A. H. (1985). Sex d~brences in social behavior: A social-role interpretation. Hillsdale, NJ: Erlbaum. Ellison, C. ( 1991 ). An eye for an eye? A note on the southern subculture of violence thesis. Social Forces, 69, 1223-1239. Evans, A. L. (1980). Personality characteristics and disciplinary atti- tudes of child-abusing mothers. ChiMAbuseandNeglect, 6, 285-293. Feather, N. T. (1994). Attitudes toward high achievers and reactions to their fall: Theory and research concerning tall poppies. In M. Zanna (Ed.), Advances in experimental social psychology (Vol. 26, pp. 1- 73). San Diego, CA: Academic Press. Feshbach, S. (1994). Nationalism, patriotism, and aggression: A clari- fication of functional differences. In L. R. Huesmann (Ed.), Aggres- sive behavior." Current perspectives (pp. 275-291 ). New York: Plenum Press. Finkelhor, D., & Yllo, K. (1985). License to rape: Sexual abuse of wives.New York: Free Press. Fitch, G. (1970). Effects of self-esteem, perceived performance, and EGOTISM AND VIOLENCE choice on causal attributions. Journal of personafity and social Psy- chology, 16, 311-315. Ford, E L. (1985). Political murder: From tyrannicide to terrorism. Cambridge, MA: Harvard University Press. Gelles, R. J,, & Straus, M. A. (1988). Intimate violence: The causes and consequences of abuse in the American family. New York: Simon & Schuster/Touchstone. Gibson, J. T, & Haritos-Fatouros, M. ( 1986, November). The educa- tion of a torturer. Psychology Today, 20( 11 ), 50-58. Glueck, S, & Glueek, E. T. (1950). Unraveling juvenile delinquency. Cambridge, MA: Harvard University Press. Goldstein, D., & Rosenbaum, A. (1985). An evaluation of the self- esteem of maritally violent men. Family Relations, 34, 425--428. Gollwitzer, P. M., & Kinney, R. E (1989). Effects of deliberative and implemental mind-sets on illusion of control. Journal of Personality and Social Psycholog,A 56, 531-542. Gondolf, E. W. (1985). Men who batter. Holmes Beach, FL: Learning Publications. Goodwin, E K., & Jamison, K. R. (1990). Manic-depressive illness. New York: Oxford University Press. Gottfredson, M. R., & Hirschi, T. (1990). A general theory of crime. Stanford, CA: Stanford University Press. Gongh, H. G., Wenk, E. A, & Rozynko, V. V. (1965). Parole outcome as predicted from the CPI, the MMPI, and a base expectancy table. Journal of Abnormal Psychology, 70, 432--44 I. Greenherg, J., & Pyszezynski, T. (1985). Compensatory self-inflation: A response to the threat to self-regard of public failure. Journal of Personality and Social Psychology, 49, 273-280. Greenwald, A. G. (1980). The totalitarian ego: Fabrication and revi- sion of personal history. American Psychologist, 35, 603-618. Groth, A. N. (1979). Men who rape: The psychology of the offender. New York: Plenum Press. Hamm, M. S. (1993). American skinheads: The criminology and con- trol of hate crime. Westport, CT: Praeger. Hare, R. D. (1993). Without conscience: The disturbing world of the psychopaths among us. New York: Simon & Schuster/Pocket. Harter, S. (1993). Causes and consequences of low self-esteem in chil- dren and adolescents. In R. Baumeister (Ed.), Selfiesteem: Thepuz- zle of low self-regard ( pp. 87-116 ). New York: Plenum Press. Heatherton, T. E, & Polivy, J. ( 1991 ). Development and validation of a scale for measuring state self-esteem. Journal of Personality and So- cialPsycholog~, 60, 895-910. Hee, K. H. (1993). The tears of my soul. New York: Morrow. Heilbrun, A. B. ( 1981 ). Gender differences in the functional linkage between androgyny, social cognition, and competence. Journal of Personality and Social Psychology. 4l, 1106-1118. Higgins, E. T. (1987). Self-diseropaney: A theory relating self and affect. Psychological Revie~, 94, 319-340. Hoffman, E L. (1925). The homicide problem. Newark, NJ: Prudential Press. Holtzworth-Munroe, A. (1992). Attributions and maritally violent men: The role ofcngnitions in marital violence. In J. Harvey, T. Or- buch, & A. Weber (Eds.), Attributions, accounts, and close relation- ships (pp. 165-175 ). New York: Springer-Verlag. Hornung, C. A., MeCullough, B. C., & Sugimoto, T. (1981). stature relationships in marriage: Risk factors in spouse abuse. Journal of Marriage and the Family, 43, 675-692. Horowilz, R., & Schwartz, G. ( 1974 ). Honor, normative ambiguity, and gang violence. American Sociological Review, 39, 238-251. Huher, R. M. (1971). The American idea of success. New York: McGraw-Hill. Huesmann, L R., Eron, L. D., Lefkowitz, M. M., & Walder, L. O. (1984). The stability of aggression over time and generations. Devel- opmenta! Psychology, 20, 1120-1134. Hurley, D. L. ( 1990 ). Incest and the development of alcoholism in adult female survivors. Alcoholism Treatment Quarterly, 7, 41-56. lckes, W., & Layden, M. A. (1978). Attributional styles. In J. Harvey ( Ed.), New directions in attribution research ( Vol. 2, pp. l 19-152). Hiilsdale, N J: Erlbaum. James, W. (1950). The principles of psychology (Vol. 2). New York: Dover Publications. (Original work published 1890) Jankowski, M. S, ( 1991 ). Islands in the street: Gangs and American urban society. Berkeley: University of California Press. Kahle, L. R., Kulka, R. A., & Klingel, D. M. (1980). Low adolescent self-esteem leads to multiple interpersonal problems: A test of social- adaptation theory. Journal of Personality and Social Psychology, 39, 496-502. Katz, J. ( 1988 ). Seductions of crime." Moral and sensual attractions in doing evil. New York: Basic Books. Keegan, J. (1993). A history of warfare. New York: Knopf. 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 tow self-regard (pp. 167-182). New York: Plenum Press. Kernis, M. H., Brockner, J., & Frankel, B. S. (1989). Self-esteem and reactions to failure: The mediating role of overgeneralization. Journal of Personafity and Social Psychology, 57, 707-714. Kernis, M. H., Cornell, D. P., Sun, C. R., Berry, A.,& Harlow, T (1993). There's more to self-esteem than whether it is high or low: The importance of stability of self-esteem. Journal of Personality and Social Psychotogb¢ 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. Kiernan, V. G. (1989). The duelin European history. Oxford, England: Oxford University Press. 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. Konovsky, M., & Wilsnack, S. C. (1982). Social drinking and self- esteem in married couples. Journal of Studies on Alcohol, 43, 319- 333. Kosterman, R., & Feshbach, S. (1989). Toward a measure of patriotic and nationalistic attitudes. Political Psychology, 10, 257-274. LaFree, G. D. (1976). Male power and female victimization: Towards a theory of interracial rape. American Journal of Sociology,, 88, 311- 328. Leary, M. R., Tambor, E. S., Terdal, S. K., &Downs, D. L. (1995). Self-esteem as an interpersonal monitor: The sociometer hypothesis. Journal of Personality and Social Psychology,, 68, 518-530. Levin, J., & McDevitt, J. ( 1993 ). Hate crimes: The rising tide of bigotry and bloodshed. New York: Plenum Press. Lewis, H. B. ( 1971 ). Shame andguilt in neurosis. New York: Interna- tional Universities Press. Lon~ D. E. (1990). The anatomy of terrorism. New York: Free Press. Lyman, M. D. ( 1989 ). Gangland: Drug tra~cking by organized crimi- nals. Springfield, IL: Charles C Thomas. MacDonald, J. M. ( 1975 ). Armed robbery: Offenders and their victims. Springfield, IL: Charles C Thomas. McCall, N. (1994). Makes me wanna holler: A young black man in America. New York: Random House. McFarlin, D. B., & Blascovich, J. (1981). Effects of self-esteem and performance feedback on future affective preferences and cognitive expectations. Journal of Personality and Social Psychology, 40, 521- 531, McNeiU, W. H. (1982). The pursuit of power. Chicago: University of Chicago Press. McNeill, W. H. (1991). The rise of the West. Chicago: University of Chicago Press. McPherson, J. M, (1988). Battle cry of freedom: The Civil War era. New York: Oxford University Press. Mead, G. H. (1934). Mind, self, and society. Chicago: University of Chicago Press. BAUMEISTER, SMART, AND BODEN Melnich, B,, & Hurley, J. R. (1969). Distinctive personality attributes of child-abusing mothers. Journal of Consulting and Clinical Psychol- ogy, 33, 746-749. Meloy, J. R. (1988). The psychopathic mind: Origins, dynamics, and treatment. Northvale, NJ: Aronson. Myers, D. (1980). The inflated self. New York: Seabury Press. National Research Council. (1993). Understanding and preventing vio- lence. Washington, DC: National Academy Press. Nisbett, R. E. (1993). Violence and U.S. regional culture. American Psychologist, 48, 441-449. Norris, J. (1988). Serial killers: The growing menace. New York: Doubleday. Oates, R. K., & Forrest, D. (1985). Self-esteem and early background of abusive mothers. ChiM Abuse and Neglect, 9, 89-93. Olweus, D. (1994). Bullying at school: Long-term outcomes for the vic- tims and an effective school-based intervention program. In R. Hues- mann (Ed.), Aggressive behavior: Current perspectives (pp. 97-130). New York: Plenum Press. Orford, J., & Keddie, A. (1985). Gender differences in the functions and effects of moderate and excessive drinking. British Journal of Clinical Psychology, 24, 265-279. Pakenham, T. ( 1991 ). The scramble for Africa: White man's conquest of the Dark Continent from 1876 to 1912. New York: Avon / Random House. Patterson, O. (1982). Slavery and social death. Cambridge, MA: Har- vard University Press. Pelham, B. W. (1993 ). On the highly positive thoughts of the highly depressed. In R. Baumeister (Ed.), Self-esteem: The puzzle of low self-regard (pp. 183-199 ). New York: Plenum Press. Peters, E. ( 1985 ). Torture.Oxford, England: Blackwell. Peterson, D. ( 1991 ). Physically violent husbands of the 1890s and their resources. Journal of Family Violence, 6, 1-15. Pillemer, K. ( 1985 ). The dangers of dependency: New findings on do- mestic violence against the elderly. Social Problems, 33, 146-158. Polk, K. (1993). Observations on stranger homicide. Journal of Crimi- nal Justice, 21,573-582. Post, J. M. (1990). Terrorist psycho-logic: Terrorist behavior as a prod- uct of psychological forces. In W. Reich (Ed.), Origins of terrorism: Psychologies, ideologies, theologies, and states of mind (pp. 25-40). Cambridge, England: Cambridge University Press. Raskin, R., Novacek, J., & Hogan, R. ( 1991 ). Narcissistic self-esteem management. Journal of Personality and Social Psychology, 60, 911- 918. Reich, W. (Ed.). (1990). Origins of terrorism: Psychologies, ideologies, theologies, and states of mind. Cambridge, England: Cambridge Uni- versity Press. Renzetti, C. M. (1992). Violent betrayal: Partner abuse in lesbian rela- tionships. Newbury Park, CA: Sage. Russell, R. J., & Hulson, B. (1992). Physical and psychological abuse of heterosexual partners. Personality and Individual Differences, 13, 457-473. Ryan, W. ( 1971 ). Blaming the victim. New York: Vintage. Salovey, P. ( 1991 ). Social comparison processes in envy and jealousy. In J. Suls & T. A. Wills (Eds.), Social comparison: Contemporary theory and research (pp. 261-285 ). Hillsdale, N J: Erlbaum. Salovey, P., & Rodin, J, (1984). Some antecedents and consequences of social-comparison jealousy. Journal of Personality and Social Psy- chology, 47, 780-792. Sampson, R. J., & Laub, J. H. ( 1993 ). Crime in the making: Pathways and turning points through life. Cambridge, MA: Harvard University Press. Scarry, E. ( 1985 ). The body in pain: The making and unmaking of the world. New York: Oxford University Press. Schlenker, B. R. (1980). Impression management: The self-concept, so- cial identity, and interpersonal relations. Monterey, CA: Brooks/ Cole. Schlenker, B. R. (1986). Self-identification: Toward an integration of the private and public self. In R. Baumeister (Ed.), Public self and private self(pp, 21-62 ). New York: Springer-Verlag. Schlenker, B. R., Soraci, S., & McCarthy, B. (1976). Self-esteem and group performance as determinants of egocentric perceptions in co- operative groups. Human Relations, 29, 1163-1176. Schneider, D. J., & Turkat, D. ( 1975 ). Self-presentation following suc- cess or failure: Defensive self-esteem models. Journal of Personality, 43, 127-135. Schoenfeld, C. G. (1988). Blacks and violent crime: A psychoanalyti- cally oriented analysis. Journal of Psychiatry and Law, 16, 269-301. Scully, D. (1990). Understanding sexual violence: A study of convicted rapists. New York: HarperCollins. Scully, D., & Marolla, J. (1985). "Riding the bull at Gilley's". Con- victed rapists describe the rewards of rape. Social Problems, 32, 251- 263. Sedikides, C. (1993). Assessment, enhancement, and verification deter- minants of the self-evaluation process. Journal of Personality and So- cial Psychology 65, 317-338. Shorkey, C. T., & Armendariz, J. (1985). Personal worth, self-esteem, anomia, hostility and irrational thinking of abusive mothers: A mul- tivariate approach. Journal of Clinical Psychology, 41, 414-421. Shrauger, J. S., &Lund, A. K. ( 1975 ). Self-evaluation and reactions to evaluations from others. Journal of Personality, 43, 94-108. Shrauger, J. S., & Schoeneman, T. J. (1979). Symbolic interactionist view of self-concept: Through the looking glass darkly. Psychological Bulletin, 86, 549-573. Sluka, J. A. (1992). The anthropology of conflict. In C. Nordstrom & J. Martin (Eds.), The paths to domination, resistance, and terror (pp. 18-36). Berkeley: University of California Press. Smith, R. H., Parrott, W. G., Ozer, D., & Moniz, A. (1994). Subjective injustice and inferiority as predictors of hostile and depressive feel- ings in envy. Personality and Social Psychology Bulletin. 20, 717-723. Staub, E. ( 1985). The psychology of perpetrators and bystanders. Polit- ical Psychology, 6, 61-85. Staub, E. (1989). The roots of evil: The origins of genocide and other group violence. New York and Cambridge, England: Cambridge Uni- versity Press. Steele, C. M. (1988). The psychology of self-affirmation: Sustaining the integrity of the self. In L. Berkowitz (Ed.), Advances in experimental socialpsychology (Vol. 21, pp. 261-302 ). New York: Academic Press. Stephan, W. G., & Rosenfield, D. (1978). Effects of desegregation on racial attitudes. Journal of Personality and Social Psychology, 36, 795-804. Stets, J. E. ( 1991 ). Psychological aggression in dating relationships: The role of interpersonal control. Journal of Family Violence, 6, 97-114. Straus, M. (1980). Victims and aggressors in marital violence. Ameri- can Behavioral Scientist, 23, 681-704. Swarm, W. B. (1987). Identity negotiation: Where two roads meet. Journal of Personality and Social Psychology, 53, 1038-1051. Swann, W. B., Griffin, J. J., Predmore, S. C., & Gaines, B. ( 1987 ). The cognitive-affective crossfire: When self-consistency confronts self- enhancement. Journal of Personality and Social Psychology, 52, 881- 889. Swarm, W. B., & Hill, C. A. (1982). When our identities are mistaken: Reaffirming self-conceptions through social interaction. Journal of Personality and Social Psychology, 43, 59-66. Tangney, J. P. (1991). Moral affect: The good, the bad, and the ugly. Journal of Personality and Social Psychology, 61, 598-607. Tangney, J. P. (1992). Situational determinants of shame and guilt in young adulthood. Personality and Social Psychology Bulletin, 18, 199-206. "Iangney, J. P., Wagner, P. E., Barlow, D. H., Marschall, D. E., & Gram- zow, R. (1994). The relation of shame and guilt to constructive vs. destructive responses to anger across the lifespan. Manuscript submit- ted for publication. EGOTISM AND VIOLENCE 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. E. (1989). Positive illusions: Creative self-deception and the healthy mind. New York: Basic Books. Taylor, S. E., & Brown, J. D. (1988). Illusion and well-being: A social psychological perspective on mental health. Psychological Bulletin, 103, 193-210. Tedeschi, J. T., & Felson, R. B. (1994). Violence, aggression, and coer- cive actions. Washington, DC: American Psychological Association. Tennen, H., & Atfleck, G. (1993). The puzzles of self-esteem: A clinical perspective. In R. Baumeister (Ed.), Self-esteem: The puzzle of low self-regard ( pp. 241-262 ). New York: Plenum Press. Tennen, H., & Herzberger, S. (1987). Depression, self-esteem, and the absence of self-protective attributional biases. Journal of Personality and Social Psychology, 52, 72-80. Tice, D. M. ( 1991 ). Esteem protection or enhancement? Self-handi- capping motives and attributions differ by trait self-esteem. Journal of Personality and Social Psychology, 60, 711-725. Tice, D. M. (1993). The social motivations of people with low self- esteem. In R. Baumeister (Ed.), Self-esteem: The puzzle of low self- regard(pp. 55-85). New York: Plenum Press. Tice, D. M., Muraven, M., Butler, J., & StiUwell, A. M. (1994). Self- presetltations to friends and strangers. Manuscript submitted for publication. Toch, H. (1993). Violent men: An inquiry into the psychology of vio- lence. Washington, DC: American Psychological Association. (Original work published 1969) Veroff, J., Douvan, E., & Kulka, R. A. ( 1981 ). The inner American: A self-portrait from 1957 to 1976. New York: Basic Books. Von Hentig, H. (1948). The criminal and his victim. New Haven, CT: Yale University Press. Wade, W. C. (1987). The fiery cross: The Ku Klux Klan in America. New York: Touchstone/Simon & Schuster. Walker, L. E. (1979). The battered woman. New York: Harper & Row. Whitley, B. E. (1983). Sex role orientation and self-esteem: A critical meta-analytic review. Journal of Personality and Social Psychology, 44, 765-778. Wicklund,R. A., & Gollwitzer, P. M. (1982). Symbolic self-completion. Hillsdale, N J: Erlbaum. Widom, C. S. ( 1991 ). A tail on an untold tale: Response to "biological and genetic contributors to violence--Widom's untold tale?" Psycho- logical Bulletin, 109, 130-132. Wiebe, V. R. ( 1991 ). Perilous rivalry: When siblings become abusive. Lexington, MA: Heath/Lexington Books. Wills, T. A. (1981). Downward comparison principles in social psy- chology. Psychological Bulletin, 90, 245-271. Wilson, J. Q., & Herrnstein, R. J. (1985). Crime and human nature. New York: Simon & Schuster. Wink, P. ( 1991 ). Two faces of narcissism. Journal of Personality and Social Psychology, 61, 590-597. V6yatt-Brown, B. (1982). Southern honor: Ethics and behavior in the old South. New York: Oxford University Press. Received December 9, 1994 Revision received April 25, 1995 Accepted May 30, 1995 •