深yae mm.31w1s1品21 Narcissistic Rage Revisited ZamK0a时o 4o tent wit ing abou Narcissism pation with the n wit (Freud.1932). explosive mix of mistrust,anger,and shame as core ients e rage his rag of narcissism have been inspired by these accounts or have been vulne able.Taken ogether,the studies of na sistic agg 1.p.9 1972). trated by the opening quote absomtion arrogance and entitlement as key ner ality char ested in their self W292014 logy,lowa State f other's tent vith one's own (Am ation.2000:Cair nal behavior.namely.th se o and vulnerability (Cain et al grand University.Ames.IA 50011.E-mailiastae.edu to self-centeredness,defensiveness.insecurity.and resentfulness 78
Narcissistic Rage Revisited Zlatan Krizan and Omesh Johar Iowa State University Narcissists are thought to exhibit “narcissistic rage,” an explosive mix of anger and hostility arising from threats to narcissists’ fractured sense of self. Building on clinical views of narcissism, we present empirical evidence on the nature and sources of narcissistic rage. Findings from 4 studies reveal narcissistic vulnerability (but not grandiosity) as a powerful driver of rage, hostility, and aggressive behavior, fueled by suspiciousness, dejection, and angry rumination. Consistent with theorizing about narcissistic rage, Study 1 showed that vulnerable (but not grandiose) narcissism predicted more anger internalization and externalization, as well as poorer anger control. Study 2 revealed vulnerable narcissism as a stronger indicator of shame and aggressiveness, especially hostility and anger. Study 3 identified distrust of others and angry rumination as key factors accounting for vulnerable narcissists’ reactive and displaced aggression. Study 4 provided behavioral evidence that vulnerable (but not grandiose) narcissism amplifies reactive and displaced aggression in the face of provocation. Taken together, the findings not only establish narcissistic vulnerability as a key source of narcissistic rage but also reveal an important pathway to narcissistic aggression that does not involve competitiveness or exploitativeness. In addition, the results support clinical views of narcissistic aggression and implicate deficient self-esteem as an important driver of aggressive behavior. Keywords: narcissism, aggression, anger I am God ... and zombies will pay for their arrogance, hate, fear, abandonment, and distrust. —Dylan Klebold (2/2/1998), Columbine High School Shooter In his 1932 paper “Libidinal Types,” Sigmund Freud offered a revolutionary description of narcissistic personality, proposing that preoccupation with oneself can lead to narcissistic injury that fuels anger and aggressive behavior (Freud, 1932). Although the idea of “narcissistic rage” as a key aspect of narcissism became widely accepted in clinical theory, the form this aggression takes and its underlying reasons have not been adequately understood. In this article we present empirical evidence on the features and sources of narcissistic rage. Evidence from four studies implicates an explosive mix of mistrust, anger, and shame as core ingredients of narcissistic rage. Furthermore, the data reveal this rage to be especially pervasive and undiscriminating in fueling aggression among the narcissistically vulnerable. Taken together, the studies identify an important driver of narcissistic aggression and implicate self-esteem dysregulation as a key source of aggressive impulses. Finally, they help us understand the extreme cases of rage illustrated by the opening quote. Narcissism Narcissism is a “cognitive–affective preoccupation with the self” (Westen, 1990, p. 226). The concept refers to Narcissus, the mythical character that fell in love with his own reflection. Psychological use of the term originates within psychoanalysis, and most theoretical development and empirical research has focused on narcissism as a personality structure, particularly in the form of a personality disorder (American Psychiatric Association, 2000; Kernberg, 1975; Kohut, 1971; Levy, Ellison, & Reynoso, 2011; J. D. Miller & Campbell, 2008; Millon, 1997; Ronningstam, 2005). These clinical accounts of narcissism are important because they have canonized the narcissistic phenotype (American Psychiatric Association, 1994). Furthermore, most personality measures of narcissism have been inspired by these accounts or have been based on formal diagnostic criteria for the narcissistic personality disorder (e.g., Pincus et al., 2009; Raskin & Hall, 1981). Most important, it is clinical theory that gave birth to the concept of narcissistic rage (Freud, 1921, p. 91; Kohut, 1972). Clinical descriptions of narcissism emphasize vanity, selfabsorption, arrogance, and entitlement as key personality characteristics; narcissistic individuals are overly invested in their selfimage, obsessed about how they are viewed by others, and often dismissive of other’s needs and wants to the extent they interfere with one’s own (American Psychiatric Association, 2000; Cain, Pincus, & Ansell, 2008; Ronningstam, 2005; Westen, 1990). However, there are two distinct themes regarding narcissists’ emotion and interpersonal behavior, namely, those of narcissistic grandiosity and vulnerability (Cain et al., 2008). Narcissistic grandiosity refers to overconfidence, exhibitionism, self-promotion, and exploitativeness. Narcissistic vulnerability, on the other hand, refers to self-centeredness, defensiveness, insecurity, and resentfulness This article was published Online First December 29, 2014. Zlatan Krizan and Omesh Johar, Department of Psychology, Iowa State University. Omesh Johar is now at the Opus College of Business, University of St. Thomas, Minneapolis, MN. We thank Constantine Sedikides and Tom Widiger for helpful comments regarding this article, as well as members of the Self and Social Perception Laboratory for assistance with collecting data. Correspondence concerning this article should be addressed to Zlatan Krizan, W112 Lagomarcino Hall, Department of Psychology, Iowa State University, Ames, IA 50011. E-mail: zkrizan@iastate.edu This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. Journal of Personality and Social Psychology © 2014 American Psychological Association 2015, Vol. 108, No. 5, 784 – 801 0022-3514/15/$12.00 http://dx.doi.org/10.1037/pspp0000013 784
NARCISSISTIC RAGE 785 Critically,empirical research on narcissistic personality traits slights with"disdain,rage,or defiant co ack"(p.659). utline key theor andiosity and vulnerability.For Wink (191) (1)Then.we ce on nan of pe sism dir stic."only grandiosity predi arce of narcissistic rage.As a whole,the studies sent the n only on of narciss e.”Subsequent distinct dim wvide key tests of clinical a ed the high reas vulnerabili What Is Narcissistic Rage? w圆 trom ulsion in the pursuit of all these e the poud be in all sm within If-invo 1006)This i o. 1975,Kh.1972R0.2005. liy disor actions such as depre sion.Moreover,thes and se despite these being key to clinical ent of the disorder.As ng a self-p piral B.Lewi agner, y th at extent existing understanding narcissistic aggression. Anger and Hostility Narcissism and Aggression nd Fre the first self-preo ccupation leads them to word rage itsel ntial reasons for th ink betw As M.Lewis (1992)puts it"a elaborates. Kohut (1972)as 'narcissistic rage. clinical observations the dysfunctionl,as it is dispr irected.Other the individua whom he his to be that ne's se Ker oiee who ard him.(.64 6
exhibited by narcissistic individuals (J. D. Miller & Campbell, 2008; Pincus & Lukowitsky, 2010; Wink, 1991). Critically, empirical research on narcissistic personality traits reveals that people differ on two relatively independent dimensions of narcissistic features, corresponding to the conceptualizations of grandiosity and vulnerability. For example, Wink (1991) linked grandiosity and vulnerability factors extracted from numerous self-report measures of narcissism to spouse reports of personality. Although elevation on both narcissism dimensions predicted being viewed as “arrogant,” “argumentative,” and “opportunistic,” only grandiosity predicted being viewed as a “show-off,” “egotistical,” and “assertive,” whereas only vulnerability predicted being viewed as “complaining,” “bitter,” and “defensive.” Subsequent research has replicated these distinct dimensions in diverse populations and identified their distinct correlates. Specifically, grandiosity is linked to dominance, low emotional distress, and high self-esteem, whereas vulnerability is linked to introversion, high emotional distress, and low self-esteem, although both share a core of entitled and dismissive tendencies (Glover, Miller, Lynam, Crego, & Widiger, 2012; Krizan & Johar, 2012; J. D. Miller et al., 2011; Rathvon & Holmstrom, 1996; Wink, 1991). It should be noted that the Narcissistic Personality Inventory (NPI; Raskin & Hall, 1981), the most commonly used measure of narcissism within personality and social psychology, captures narcissistic grandiosity, not vulnerability (Krizan & Johar, 2012; J. D. Miller & Campbell, 2008; Rathvon & Holmstrom, 1996). This is important because the vast majority of theory and evidence in personality and social psychology is based on the use of this measure, and thus speaks only to grandiose aspects of narcissism. Furthermore, diagnostic criteria for narcissistic personality disorder do not currently include vulnerable symptoms of narcissism, despite these being key to clinical assessment of the disorder. As a result, there is a substantial concern that assessment of narcissism across both personality and clinical domains ignores important aspects of the construct (J. D. Miller, Gentile, Wilson, & Campbell, 2013; Pincus & Lukowitsky, 2010; Ronningstam, 2005). As we elaborate later in the article, this imposes serious limitations on understanding narcissistic aggression. Narcissism and Aggression Sigmund Freud (1932) was the first to suggest that narcissists’ self-preoccupation leads them to aggress against others. The potential reasons for the link between anger, aggression, and narcissism were subsequently suggested by other psychoanalysts, whose clinical observations suggested that narcissistic self-absorption can fuel a vicious cycle of hostility, shame, and reactive aggression (Alexander, 1938; Jacobson, 1964; Saul, 1947). Defined by Heinz Kohut (1972) as “narcissistic rage,” these clinical observations describe a precarious condition in which frustrations of a narcissistically perceived reality and a vulnerable sense of self result in dejection and shame that fuel anger, resentment, and vindictiveness. Moreover, they suggest narcissistic rage to be immature and dysfunctional, as it is disproportionate or misdirected. Other theorists have also suggested that narcissistic individuals exhibit patterns of rage, instigated by rejection that opens childhood wounds or events that contradict one’s sense of specialness (Kernberg, 1975; Millon, 1997). The Diagnostic and Statistical Manual of Mental Disorders (4th ed.; American Psychiatric Association, 1994) itself states that narcissistic individuals react to interpersonal slights with “disdain, rage, or defiant counterattack” (p. 659). Following these views, we first outline key theoretical features of narcissistic rage, building on the pioneering work by Kohut (1972). Then, we review existing empirical evidence on narcissism and aggression with a focus on rage reactions. We argue that the almost exclusive focus on grandiose narcissism yielded little support for theoretical accounts of narcissistic rage. In response, we present four studies that looked to vulnerable narcissism as a key source of narcissistic rage. As a whole, the studies represent the first systematic examination of narcissistic rage within the context of both vulnerable and grandiose narcissism. Furthermore, they provide key tests of clinical accounts of narcissistic aggression so far not adequately addressed in nonclinical populations. What Is Narcissistic Rage? Although it can be traced back to Freud, narcissistic rage was first extensively described by the psychoanalyst Heinz Kohut (1972). According to him, “the need for revenge, for righting a wrong, for undoing a hurt by whatever means, and a deeply anchored, unrelenting compulsion in the pursuit of all these aims . . . are the characteristic features of narcissistic rage in all its forms” (p. 638). He succinctly summarized a key observation of many clinicians, namely, that narcissistic individuals whose self-involved perception of reality are questioned will respond with vitriol and retaliation in addition to shame and depression (Alexander, 1938; Kernberg, 1975; Kohut, 1972; Ronningstam, 2005). In fact, these observations view the lack of a cohesive sense of self as key both to externalizing reactions such as aggression and to internalizing reactions such as depression. Moreover, these other- and selfdestructive reactions are thought to become mutually reinforcing, producing a self-perpetuating “shame–rage” spiral (H. B. Lewis, 1987; Scheff, 1987; Tangney, Wagner, Fletcher, & Gramzow, 1992). As specified next, these analyses imply three key aspects of narcissistic rage. We first outline these features of the narcissistic rage hypothesis and then critically evaluate to what extent existing empirical evidence on narcissism bears on them. Anger and Hostility First and foremost, narcissistic rage is defined by pervasive, intense, and ill-directed anger. As the word rage itself implies, narcissists’ angry responses to even minor provocations should be disproportionate and unfocused. As M. Lewis (1992) puts it, “rage is anger out of control” (p. 153). Numerous psychoanalytic accounts link threats to narcissistic self-views (i.e., “narcissistic injuries”) to intense anger and hostility (e.g., Alexander, 1938; Freud, 1921). As Kohut (1972) elaborates, the fanaticism of the need for revenge and the unending compulsion of having to square an account after an offense are therefore not the attributes of aggressivity that is integrated with the mature purposes . . . the shame-prone individual who is ready to experience setbacks as narcissistic injuries and respond to them with insatiable rage does not recognize his opponent as a center of independent initiative with whom he happens to be at cross purposes . . . the narcissistically injured... cannot rest until he has blotted out a vaguely experienced offender who dared to oppose him, to disagree with him, or to outshine him. (pp. 643– 644) This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. NARCISSISTIC RAGE 785
786 KRIZAN AND IOHAR asthey need to f-control s 1973: the wour sm (Ame ciation.2000 issists are exre cted to react with aggression to and to displace their aggression onto others who response to even minor provocations Does Evidence Support the Narcissistic Rage Hypothesis? Sha ame and Inferioritv add them.R 981 sion when face ster.Parrott.&Byre.02:Tangney &Dearing.2002).S impeachments of one' ften go usly unacknowledged or misidentified.transforn 1993 1988:Rhode walt Mor thers that shamed us may be initially adaptiv nd entitlemer similar ainful fee ather med us (Ko ohut.1972:M.Lewis.1992.pp. 150 iv es to imp shment of on's image o ndio issists" inflated s which ther fuel a neiste 998:Morf g92 see also Scheff.1987).Empirical research sa clo tool out of the se that ha d as 2008 Hill-Rado 106)In shor sistic individuals are expected to experie rage in par 678 ide on grand 19resofn han Reactive and Displaced Aggression 1995 On the oth man studies find tha ithe V.or d Third and final.narcissistic rage should lead to disr and unfocus Atlas Them 2003 ggre ”that is rathe on harming th other indi 1004)Given the intense rage compounded by feelings of shame or s to b "there is utte isregard for reasonable limitation t Mo 1009 &Stak and add fuel he
Accordingly, the narcissistically wounded are thought to distrust others and be suspicious of their intentions, as they need to rationalize why these others are getting in their way or withholding special treatment that the narcissist expects (Bursten, 1973; Millon, 1997; Ronningstam, 2005). In concert with other clinical accounts of narcissism (American Psychiatric Association, 2000; Kernberg, 1975), these observations explicate narcissistic rage as a pervasive and dysfunctional anger coupled with hostile suspicions, presumably resulting in misdirected or disproportionate aggression in response to even minor provocations. Shame and Inferiority Second, narcissistic rage is thought to be aggravated by inferiority and shame (Broucek, 1982; Kohut, 1972). Shame involves feelings of being exposed and devalued for one’s deficiencies, and can be one of the most devastating human emotions (Smith, Webster, Parrott, & Eyre, 2002; Tangney & Dearing, 2002). Scholars have long noted that the pain of shame is so severe that it may often go consciously unacknowledged or misidentified, transforming into vague feelings of depression or anger at sources of shame (Alexander, 1938; H. B. Lewis, 1971). On one hand, anger at others that shamed us may be initially adaptive, as it can help us to forget shame-based pain, to recast the blame for painful feelings on others rather than ourselves, or to eliminate the person that shamed us (Kohut, 1972; M. Lewis, 1992, pp. 150 –151). On the other hand, such responses to shaming, if prolonged, can lead to chronic rage reactions, which further exacerbate existing feelings of shame and guilt, which then further fuel anger, ultimately creating a “shame–rage spiral” (H. B. Lewis, 1971; M. Lewis, 1992; see also Scheff, 1987). Empirical research confirms a close link between shame and anger, and suggests that shame-prone individuals are more likely to experience anger and to engage in destructive behaviors as a result (Bennett, Sullivan, & Lewis, 2005; Harper & Arias, 2004; Tangney et al., 1992; Tangney, Wagner, Hill-Barlow, Marschall, & Gramzow, 1996). In short, narcissistic individuals are expected to experience rage in part because they are especially prone to shame and dejection when flaws in themselves and their narcissistically perceived reality are exposed. Reactive and Displaced Aggression Third and final, narcissistic rage should lead to disproportionate and unfocused acts of aggression in response to provocation. These aggressive behaviors are expected to be “reactive” rather than “proactive,” that is, driven by anger, relatively spontaneous, and focused on harming the other individual (rather than deliberate and calculating; see Anderson & Bushman, 2002; Crick & Dodge, 1994). Given the intense rage compounded by feelings of shame or inferiority, narcissistic individuals are thought to engage in immature, dysfunctional aggressive acts. As Kohut (1972, p. 639) suggests, “there is utter disregard for reasonable limitations and a boundless wish to redress an injury and to obtain revenge.” Furthermore, this rage is expected to fuel displaced aggression as others’ unrelated or minor provocations get in the way and add fuel to the proverbial fire (Vaknin, 2001). Consistent with this reasoning, increased anger and suspicions about others’ behavior generally exacerbate displaced aggression, particularly in the presence of additional minor provocations (i.e., “triggers”; see N. Miller, Pedersen, Earleywine, & Pollock, 2003). In addition, angry rumination may undermine self-control, suggesting an additional level of unpredictability in aggression among the wounded narcissists (Denson, Pedersen, Friese, Hahm, & Roberts, 2011). In short, narcissists are expected to react with aggression to even minor provocations and to displace their aggression onto others who stand in their way or prove to be minor annoyances. Does Evidence Support the Narcissistic Rage Hypothesis? Having summarized the theoretical proposals about narcissistic rage, it is essential to evaluate to what extent the empirical literature addressed or supported them. Research on narcissistic grandiosity using the NPI (Raskin & Hall, 1981) does reveal that narcissistic individuals are more prone to aggression when faced with strong threats to self (e.g., public impeachments of one’s ability, intelligence, or social status; Bushman & Baumeister, 1998; Rhodewalt, Madrian, & Cheney, 1988; Rhodewalt & Morf, 1998; Stucke & Sporer, 2002; Twenge & Campbell, 2003). Research focusing more specifically on narcissistic exploitativeness and entitlement reveals similar results (Campbell, Bonacci, Shelton, Exline, & Bushman, 2004; Konrath, Bushman, & Campbell, 2006; Reidy, Zeichner, Foster, & Martinez, 2008). These aggressive responses to impeachment of one’s image or status are often interpreted as maneuvers aimed at restoring one’s superiority, necessitated by (grandiose) narcissists’ inflated self-esteem and entitlement (Bushman & Baumeister, 1998; Morf & Rhodewalt, 2001). These superiority-imposing tactics are also considered to be only one tool out of the narcissists’ considerable set of selfenhancing strategies (Campbell, Reeder, Sedikides, & Elliot, 2000; Gabriel, Critelli, & Ee, 1994; Krizan & Bushman, 2011; Morf & Rhodewalt, 1993; Vazire, Naumann, Rentfrow, & Gosling, 2008). Although these findings are often taken as support for the narcissistic rage hypothesis (e.g., Rhodewalt & Morf, 1998, p. 678), a careful survey of evidence on grandiose narcissism reveals little to no direct support that it is linked with rage. On one hand, some studies do find a link between narcissistic grandiosity and chronic anger or hostility—signature features of narcissistic rage (e.g., Raskin, Novacek, & Hogan, 1991; Rhodewalt & Morf, 1995). On the other hand, many studies find that grandiosity either does not consistently predict chronic anger nor hostility, or does so only when unique effects of entitlement and exploitativeness are considered (e.g., Atlas & Them, 2008; Besser & Priel, 2010; Bradlee & Emmons, 1992; Ruiz, Smith, & Rhodewalt, 2001; Witte, Callahan, & Perez-Lopez, 2002). Furthermore, studies examining angry and hostile responses to specific hypothetical or real-life situations offer similarly weak support for broad rage reactions among those high in grandiosity, showing their angry responses to be either average or exacerbated only by direct threats to one’s competence or social status (Kernis & Sun, 1994; McCann & Biaggio, 1989; Rhodewalt & Morf, 1998; Smalley & Stake, 1996; Stucke & Sporer, 2002). Taken together, this evidence does not provide convincing support for the notion that chronic and unrestrained anger typifies those with grandiose narcissism. When it comes to shame or dejection, the second signature of narcissistic rage, evidence indicates that grandiose narcissists are actually less likely to experience these reactions than others. First, This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. 786 KRIZAN AND JOHAR
NARCISSISTIC RAGE 781 O'Leary,&Balkin,1989)and beliefs in one's own istic grandiosity does not p wide strong rt for th Sedikide th this sive”p.48.More ntly o idea that nar sistic grandiosity engenders shame:in fact. ve behavior itself,there is surprisingly little eistie rage are ba To be st liose individu edirectly and publi imag itsky.2010:Sedikide 1.2008:Sma ggre Talley aulhu 2010).Even unde Narcissistic Vulnerability as a Source of Na 2002s In order to marshal evidence for the narcissistic rage hypothesis or a critique.see fer Rueda.2009).As the and will thus be stronely linked to all the three core features of and social status and le ctor of e features.namel Rhodewalt.2001).it should not be ising that their of f laced aggressio who has just berated th ression?Only thre examined dis ability and grandiositys predictors of anger vulnerability 2003 Furth given that displaced stron ly with angr hostility et of sion should be J.D.Miller ps020 e) s(Krizan Johar,2012).Second,evidenc suggest tha (Barry et al 00 2009)In addition, nd t ediate the link bet 206:Reidy. high findings suggests the possibility that grandiose individuals" sity(n et al. 009).To our kn vulncrability an with ou han uncontrolled acts of rage osal that it is narc ility (rather than grandios ty)that is the seat of Summary NPI (Raskin Hal Overview of Studies
grandiose narcissism predicts lower depression and anxiety (Sedikides, Rudich, Gregg, Kumashiro, & Rusbult, 2004; Wright, O’Leary, & Balkin, 1989) and beliefs in one’s own superiority rather than inferiority (Campbell, Rudich, & Sedikides, 2002; Krizan & Bushman, 2011). Second, when it comes to shame specifically, grandiosity is either unrelated or negatively related to shame responses (Gramzow & Tangney, 1992; Krizan & Johar, 2012; Pincus et al., 2009; Wright et al., 1989). In short, there is no evidence that narcissistic grandiosity engenders shame; in fact, evidence points to the contrary. Turning to aggressive behavior itself, there is surprisingly little evidence that grandiosity fuels rage-driven aggression. To be sure, grandiose individuals are more likely to aggress against others who have directly and publicly impeached their image of superiority and status (Bushman & Baumeister, 1998; Ferriday, Vartanian, & Mandel, 2011; Reidy et al., 2008; Smalley & Stake, 1996; Twenge & Campbell, 2003). However, there is little evidence that such aggression occurs in the absence of public ego threats (see Bettencourt, Talley, Benjamin, & Valentine, 2006; Ferriday et al., 2011; Jones & Paulhus, 2010). Even under conditions of ego threat, it is not clear that the observed aggression is driven by anger (Bushman & Baumeister, 1998; Stucke & Sporer, 2002). In this vein, note that aggression in studies on narcissism is typically assessed with the Competitive Reaction Time Task (Taylor, 1967; for a critique, see Ferguson & Rueda, 2009). As the name implies, this task assesses competitive noise blasts throughout a course of a multitrial performance competition. Given grandiose individuals’ competitiveness and investment in social status and superiority (Campbell et al., 2000, 2002; Krizan & Bushman, 2011; Morf & Rhodewalt, 2001), it should not be surprising that their engagement in a competitive task with a person who has just berated them motivates competitive noise blasting. Does grandiose narcissism promote displaced aggression? Only three studies examined displaced aggression, and they provide conflicting findings (Bushman & Baumeister, 1998; Martinez, Zeichner, Reidy, & Miller, 2008; Twenge & Campbell, 2003). Furthermore, given that displaced aggression was always assessed with competition-based measures, the exact interpretation of the observed aggression should be tentative. Finally, we should note that narcissistically grandiose (but not vulnerable) individuals frequently report engaging in proactive, instrumental aggression, that is, deliberate acts of aggression that are used to further one’s social status or other goals, often by scheming or manipulating others (Barry et al., 2007; Fossati, Borroni, Eisenberg, & Maffei, 2010). In addition, grandiosity is related to sadism and unprovoked aggression, with enjoyment of aggression found to mediate the link between grandiosity and aggressive behavior (Buckels, Jones, & Paulhus, 2013; Girgis, 2006; Reidy, Foster, & Zeichner, 2010). Taken together, these findings suggests the possibility that grandiose individuals’ aggressive responses to ego threat are deliberate means of asserting superiority and dominance or twisted games to be enjoyed, rather than uncontrolled acts of rage. Summary In short, existing evidence based on the NPI (Raskin & Hall, 1981) reveals weak and inconsistent links between (grandiose) narcissism and chronic anger or hostility, negative links with shame and inferiority, and little evidence for unrestrained or displaced aggression. As a result, we conclude that evidence bearing on narcissistic grandiosity does not provide strong support for the idea of narcissistic rage. In accord with this sentiment, Costa and Widiger (1993) state that “narcissist is not particularly hostile... or even particularly aggressive” (p. 48). More importantly, we contend that many of these studies are not ideal tests of the narcissistic rage hypothesis, as they examine narcissism exclusively in terms of narcissistic grandiosity. Measures of narcissistic grandiosity do not seem to capture individuals on which clinical accounts of narcissism and narcissistic rage are based, given grandiose individuals’ relative social adjustment, lack of depression and anxiety, and lack of treatment seeking (Corruble, Ginestet, & Guelfi, 1996; Peck, 1998; Pincus & Lukowitsky, 2010; Sedikides et al., 2004). This raises additional doubts about whether many existing studies of narcissism and aggression should be taken as tests of the narcissistic rage hypothesis in the first place. Narcissistic Vulnerability as a Source of Narcissistic Rage In order to marshal evidence for the narcissistic rage hypothesis, we looked to narcissistic vulnerability. We hypothesized that vulnerable narcissism will be a powerful predictor of rage reactions and will thus be strongly linked to all the three core features of narcissistic rage outlined earlier. In the paragraph below we briefly discuss existing evidence in support of our proposal that vulnerable narcissism should be a key predictor of these features, namely, anger, shame, and reactive, as well as displaced aggression. We conclude with an overview of four studies we conducted that examined narcissistic rage as a function of narcissistic vulnerability (in addition to grandiosity). First, investigations that directly contrasted narcissistic vulnerability and grandiosity as predictors of anger suggest vulnerability to be a stronger predictor. For example, vulnerability, relative to grandiosity, is associated more strongly with angry hostility (a facet of neuroticism), and with hostile and paranoid personality features (J. D. Miller & Campbell, 2008; J. D. Miller et al., 2011; Okada, 2010). Moreover, vulnerability is a stronger predictor of anger reported in response to hypothetical provocation scenarios (J. D. Miller et al., 2011; Okada, 2010), as well as hostile envy of others (Krizan & Johar, 2012). Second, evidence suggests that vulnerable narcissism predicts stronger shame responses, whereas grandiose narcissism actually predicts less shame (Krizan & Johar, 2012; Pincus et al., 2009); In addition, vulnerability is strongly associated with depressive reactions, a key feature of shameful experiences (Pincus et al., 2009; Tritt, Ryder, Ring, & Pincus, 2010). Third, those high on vulnerability are more likely to report generally engaging in aggressive behavior than those high on grandiosity (Pincus et al., 2009). To our knowledge, no published studies examined the link between vulnerability and displaced aggression. Regardless, these findings are consistent with our proposal that it is narcissistic vulnerability (rather than grandiosity) that is the seat of narcissistic rage. Overview of Studies To marshal evidence for the narcissistic rage hypothesis, four studies examined vulnerable narcissism as a predictor of anger and This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. NARCISSISTIC RAGE 787
788 ed agg SuT ng Expe y Siebrer ance of each on for ra Results and Discussion of ange an iosity factors as rs of chronic ag Gra y,an dy 3 te are prone to intense ted anger but not gran Study 2:Narcissism and Aggressiveness in a Sample of Community Adults Havi we wanted to ing the complex nature of narcissism. his finding an d our analysis to hostility Study 1:Narcissism and Anger e subst reported narci sism and ggression may b Given existing theories of narcissistie aggr focus or t as a k ed bcause of bias ed responding.We expect and towhich it would be indicative ofmore out weexpccted vunerability to havemuchst ou high shame-proneness,and that gran- Method One hundred eieht unders tudents Method online survevs in exch. N( nge for course credit.Th n&H scale (HNS:Hen cities of the United States (Chicago.Los Angeles and At De. sure narcissistie entitlement)to asse dy on penonmliny”Eachesponderwge icipants completed the Anger-Out (my anger" aware of")and Anger Control Scales (ee"I control my temner") nant completed numerous measures includine the followine for 2 6 80(40 nt Scale 0481079 79
hostility, shame and dejection, and reactive and displaced aggression; the key ingredients of narcissistic rage. All studies also examined narcissistic grandiosity, in order to directly evaluate the importance of each narcissism dimension for rage and aggression. Study 1 compared narcissistic vulnerability and grandiosity as predictors of anger externalization, internalization, and control. Study 2 utilized multiple measures of narcissism in a sample of community adults and compared narcissistic vulnerability and grandiosity factors as predictors of chronic aggressiveness, anger, hostility, and shame. Study 3 tested a structural model that specified distrust of others and angry rumination as key factors underlying narcissists’ reactive and displaced aggression. Study 4 sought behavioral laboratory evidence that vulnerable (but not grandiose) narcissism fuels dejection, hostility, anger, and aggression in response to provocation. By employing diverse samples and methodological approaches, the data triangulate on the important role that narcissistic rage plays in aggressive behavior while illuminating the complex nature of narcissism. Study 1: Narcissism and Anger First, we tested vulnerable and grandiose narcissism as predictors of anger experience and expression. We hypothesized that only vulnerable narcissism would be substantially associated with anger externalization, anger internalization, and poor anger control. Given existing theories of narcissistic aggression focus on entitlement as a key reason for narcissists aggressive responses to threat (Bushman & Baumeister, 1998), we also examined entitlement and the extent to which it would be indicative of more maladaptive anger responses. Method One hundred twenty-eight undergraduate students completed online surveys in exchange for course credit. They completed a battery of personality measures, including the NPI (Raskin & Hall, 1981; to measure grandiosity), the Hypersensitive Narcissism Scale (HNS; Hendin & Cheek, 1997; to measure vulnerability), and the Psychological Entitlement Scale (Campbell et al., 2004; to measure narcissistic entitlement). To assess anger expression, participants completed the Anger-Out (e.g., “I express my anger”), Anger-In (e.g., “I’m irritated a great deal more than people are aware of”), and Anger Control Scales (e.g., “I control my temper”) from the State-Trait Anger Expression Inventory (Spielberger, 1999). Results and Discussion Correlations between narcissism dimensions and anger expression are presented in Table 1. Critically, vulnerability was associated with anger externalization, internalization, and poor anger control. Grandiosity and entitlement were also associated with anger externalization, but did not predict other aspects of anger expression. Taken together, these findings imply that only vulnerable narcissists are prone to intense and misdirected anger. Study 2: Narcissism and Aggressiveness in a Sample of Community Adults Having established preliminary support for vulnerable narcissism as the key predictor of uncontrolled anger, we wanted to replicate this finding and expand our analysis to hostility (cynical and negative views of others), shame, and aggressive behavior itself. To aid generalizability, Study 2 involved a sample of community adults of varying ages, and employed multiple measures of narcissism that were used to extract underlying grandiosity and vulnerability factors. Moreover, we assessed socially desirable responding in order to examine to what extent any observed associations between reported narcissism and aggression may be skewed because of biased responding. We expected both dimensions of narcissism to relate to reports of aggressive behavior itself, but we expected vulnerability to have much stronger association with aggressive tendencies overall (i.e., anger, hostility, and aggression). In addition, we expected that only vulnerable narcissism would be associated with higher shame-proneness, and that grandiose narcissism would predict less shame. Method One hundred sixty-one community residents (57% female; aged 18 to 37 years; 62% Caucasian) were recruited online in major cities of the United States (e.g., Chicago, Los Angeles, and Atlanta) via Craigslist (www.craigslist.org) advertisements for a survey study on “personality.” Each respondent was compensated with $10 for their participation. The questionnaires were administered online, and each participant completed numerous measures, including the following (orTable 1 Correlations Between Narcissism and Anger Expression in Study 1 (ns 124 to 128) 1 2 3 4 5 67 Narcissism 1. Narcissistic Grandiosity .80 (40) 2. Narcissistic Vulnerability .04 .81 (10) 3. Psychological Entitlement Scale .42 .31 .87 (9) Anger 4. Anger Externalization .36 .31 .38 .79 (8) 5. Anger Internalization .08 .40 .12 .50 .77 (8) 6. Anger Control .09 .18 .20 .38 .06 .87 (8) Note. Cronbach’s alphas are reported along the diagonal with the number of items in parentheses. p .10. p .05. p .01. This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. 788 KRIZAN AND JOHAR
NARCISSISTIC RAGE 789 997).the NPI (Raskin Hall.1981).the State-Trait Grandiosit The correlations between grandiose and vulnerable factors,ag Ste sm Inventory (Pincus et al trongly asse iated with overall trait ss than wa he Buss-Pem rticinan me pro esscale (Wolf,Cohe In order to examine the uigue relations between the two di Insko.2010). odant and m and aggres ggress eness a ously predicted by both narcissism ng for gender and tor of aggre grandiosity only showing uniqu Results and Discussion Study 3:Ac analysis (with oblimin rotation) on ou The evidence from the first two studies clearly implica nar actors ed two with and ag with the re ccounting for 3 and 23%of the id predict hig rted a two-fac (Horn.1965).The factor th such as inte e-propensity. and grand ggres ion is likely to take.As noted ntitlement is ately expres ed in (a)h idity of NPI and HNS as valid mea e path.To this end.Study 3 tes lty,respectively.Cri ically.we derived na and shame and regression on narcissism factors (smudy 2) 2 5 7 9 10 G V 004 9.Avo 290-0208 二5 11.Impression Management 06二45二2-005二3二20二37620 帐
dered randomly). First, to comprehensively assess narcissistic personality traits, participants completed the HNS (Hendin & Cheek, 1997), the NPI (Raskin & Hall, 1981), the State-Trait Grandiosity Scale (Rosenthal, Hooley, & Steshenko, 2003), the Psychological Entitlement Scale (PES; Campbell et al., 2004), and all the subscales from the Pathological Narcissism Inventory (Pincus et al., 2009). Second, to assess aggressiveness, participants completed the Buss-Perry Aggression Questionnaire, which consists of subscales for physical aggression, verbal aggression, hostility, and anger (Buss & Perry, 1992). Third, to assess shame propensity, participants completed the Shame Proneness scale (Wolf, Cohen, Panter, & Insko, 2010), which assesses avoidant and selfevaluative reactions to hypothetical situations that induce shame. Fourth and final, to assess socially desirable responding, participants completed the Balanced Inventory of Socially Desirable Responding that yields a self-deceptive enhancement and an impression management score (Paulhus, 1998). The reliabilities appear in the diagonal of Table 2. Results and Discussion In order to extract grandiosity and vulnerability factors underlying the various measures of narcissism, we conducted a principal factor analysis (with oblimin rotation) on our 11 measures of narcissism with the aim of extracting two correlated narcissism factors. The analysis revealed two factors with an eigenvalue greater than 1 (accounting for 37% and 23% of the variance after rotation). Similarly, the scree plot, as well as parallel analysis, supported a two-factor solution (Horn, 1965). The factor loadings from the pattern matrix (controlling for factor dependence) are shown in Table 3 and represent the two factors of narcissistic vulnerability and grandiosity (which correlated .31). Examining factor loadings confirms that entitlement is a core feature of narcissism common to both dimensions, and also supports the validity of NPI and HNS as valid measure of narcissistic grandiosity and vulnerability, respectively. Critically, we derived narcissism factor scores, which we then linked to narcissism, aggressiveness, and shame. The correlations between grandiose and vulnerable factors, aggressiveness, shame, and socially desirable responding are presented on the left side of Table 2. First, vulnerability was more strongly associated with overall trait aggressiveness than was grandiosity (.56 vs. .38). Second, only vulnerability was a consistent predictor of all aspects of aggressiveness, including hostility and anger. Finally, whereas vulnerability was positively related to shame responses, grandiosity was negatively related. In order to examine the unique relations between the two dimensions of narcissism and aggressive tendencies, we conducted a series of hierarchical regression analyses in which each feature of aggressiveness was simultaneously predicted by both narcissism factors, controlling for gender and socially desirable responding. The results of these regressions appear on the right side of Table 2. Note that only vulnerability was a unique predictor of aggression, anger, and hostility, with grandiosity only showing unique links with physical aggression. Study 3: Accounting for Narcissists’ Reactive and Displaced Aggression The evidence from the first two studies clearly implicates narcissistic vulnerability as a potent predictor of anger, hostility, shame, and aggressiveness, consistent with the narcissistic rage hypothesis. Although narcissistic grandiosity did predict higher likelihood of reporting aggressive behavior itself, it did not predict other key features of narcissistic rage, such as intense, uncontrolled anger, or shame-propensity. However, the evidence presented thus far does not reveal the particular form that narcissistic aggression is likely to take. As noted earlier, narcissistic rage should be ultimately expressed in (a) hostile, reactive aggression, and (b) displaced aggression toward individuals who find themselves in narcissists’ destructive path. To this end, Study 3 tested a structural equation model linking narcissistic traits to reactive Table 2 Correlations Between Narcissism Factors, Aggression, and Shame, and Regression on Narcissism Factors (Study 2) 1 2 3 4 5 6 7 8 9 10 11 Regression Coefficients G V Narcissism 1. Grandiosity factor (G) — 2. Vulnerability factor (V) 0.04 — Aggressiveness 3. Overall Aggressiveness .23 .56 .94 (29) .05 .53 4. Physical Aggression .31 .38 .83 .85 (9) .18 .31 5. Verbal Aggression .28 .28 .77 .58 .78 (5) .13 .33 6. Anger 0.15 .56 .88 .63 .68 .82 (7) .02 .49 7. Hostility 0.07 .58 .83 .51 .50 .67 .89 (8) .11 .57 Shame Propensity 8. Negative Self Evaluation .32 .29 0.11 0.04 0.05 0.08 .25 .78 (5) .37 .43 9. Avoidance 0.04 .47 .31 .20 0.13 .32 .36 .46 .75 (5) .10 .41 Social Desirability 10. Self-Deceptive Enhancement 0.16 .45 .20 0.13 0.05 .28 .24 0.13 .33 .76 (20) 11. Impression Management .21 .41 .39 .33 .17 .37 .36 0.07 .26 .62 .84 (20) Note. ns 147 to 151. Reliabilities appear in the diagonal with the number of items in parentheses. Standardized regression coefficients in the last two columns were obtained after controlling for gender and social desirability. p .10. p .05. p .01. This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. NARCISSISTIC RAGE 789
790 KRIZAN AND JOHAR Table3 Factor Loadings in Sudy 2 (N=147) Narcissistic Traits Vulnerability sity S l Entit Scale 44 ent self-esteem self-enhancemer 14 el fantasy Entitlement rage isly exam- cally e mistrust in others (a key feat sed with the een vulne able and grandiose r issism,on one ete nery rumination.Finally.we an nmcnwoulaoapretircactiveasgresio ent (as in Studies I and2).After stine this initial model n.Pedersen. Miller.2006). sted (a)that the direct pat angry ruminati n (ie..i ave a direct en As a removed paths from the initial model illustrated as dashed lines) Method Three hundred seventy-four undergraduate students completed ngry rumination and entitlement)and aggressive behavior (re nd the HNS (to me nt for the link be e features zed to a To measure ral model with g for derlie hostile attitudes toward others,and assessing it directly inal model are shown as dashed linesp
and displaced aggression specifically, while simultaneously examining key personality features responsible for this link. Specifically, we modeled mistrust in others (a key feature of hostility), angry rumination, and entitlement as factors accounting for the link between vulnerable and grandiose narcissism, on one hand, and reactive and displaced aggression, on the other. Based on our theorizing and the data from Studies 1 and 2, we expected only vulnerability to predict both reactive and displaced aggression, whereas grandiosity to only predict reactive aggression, if predicting it at all. We further anticipated that mistrust and angry rumination would be key factors accounting for the links of vulnerability to both reactive and displaced aggression, with mistrust itself also predicting increased angry rumination. Finally, we anticipated that entitlement would only predict reactive aggression, and that both vulnerable and grandiose narcissism would predict entitlement (as in Studies 1 and 2). After testing this initial model, modification indexes suggested (a) that the direct path from grandiosity to reactive aggression was unnecessary, and (2) that mistrust only predicted aggression through angry rumination (i.e., it did not have a direct effect on aggression). As a result, both of these paths were dropped from the final model (see Figure 1, with removed paths from the initial model illustrated as dashed lines). Method Three hundred seventy-four undergraduate students completed online surveys in exchange for course credit. They completed a battery of personality measures, including the following sets of measures assessing narcissism (grandiosity and vulnerability), the hypothesized aggression-driving features of narcissism (mistrust, angry rumination, and entitlement), and aggressive behavior (reactive aggression and displaced aggression). First, to assess narcissism, participants completed the NPI (to measure grandiosity) and the HNS (to measure vulnerability), as in Study 1. Second, to measure features of narcissism hypothesized to account for the link between narcissistic traits and aggressive behavior, they completed the following. To measure mistrust in others, they completed the International Personality Item Pool Distrust scale (Goldberg, 1999). Mistrust is a key feature thought to underlie hostile attitudes toward others, and assessing it directly (rather than hostility more broadly) should specifically implicate expectations of others’ behavior as key to narcissistic aggression (Gurtman, 1992). Angry rumination was assessed with the Anger Rumination Scale (Sukhodolsky, Golub, & Cromwell, 2001), which has demonstrated appropriate reliability and validity. Entitlement was measured with the PES, as in Study 1. Finally, to measure reactive aggression, participants completed the Richardson (1998) Conflict Response questionnaire. This measure asks participants to rate their tendency to respond to conflict with either direct (e.g., verbal abuse) or indirect (e.g., property damage) aggression, thus broadly sampling from potential aggressive responses to provocation which do not always involve faceto-face retaliation (see Richardson & Green, 2006, for validity evidence). To measure displaced aggression, participants completed the Displaced Aggression subscale from the Displaced Aggression Questionnaire (Denson, Pedersen, & Miller, 2006), which has been behaviorally validated as a predictor of triggereddisplaced aggression. Table 3 Factor Loadings in Study 2 (N 147) Narcissistic Traits Factor Vulnerability Grandiosity Narcissistic Personality Inventory .18 .83 Hypersensitive Narcissism Scale .64 .05 State-Trait Grandiosity Scale .04 .79 Psychological Entitlement Scale .23 .44 Pathological Narcissism Inventory Contingent self-esteem .90 .07 Exploitativeness .11 .63 Self-sacrificing self-enhancement .49 .21 Self-hiding .61 .11 Grandiose fantasy .54 .34 Devaluing .90 .00 Entitlement rage .86 .08 Note. The coefficients represent unique factor loadings from a principal factors analysis with oblimin rotation. Grandiosity Vulnerability Entlement Mistrust Angry Ruminaon Displaced Aggression Reacve Aggression .18** .45** .42** .32** .24** .39** .12** .40** .45** .33 -.09 ** Figure 1. A structural model with Anger Rumination, Mistrust, and Entitlement as factors accounting for the links between narcissism and aggression in Study 3. Originally hypothesized paths not included in the final model are shown as dashed lines. p .001. This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. 790 KRIZAN AND JOHAR
NARCISSISTIC RAGE 791 Results and Discussion to participantsas atest of people's food preferences and created an has in some e way p voked the partic dictor c wherea the anting for al respo y flavo r.co hoking)is an ctual weapon ents to th .2012).The fit of the mod I was good.y(10) 19.4 0 nar cause harm to another see Ritter&Eslea 2005 n S whethe responses to addition manip late entitlemen whe age Third and ager on).Thi of with the herabiliy-4o-displaced-aggrcsion aggression (ie choosine a hot sauce intended either fo and point to rust and angry rumination as personality features earch on nar sity an predictor of grandio The findings from the first three studies thus consistently reveal 2006:Bus A that out at t who stand in the way.However,the findings oth. ened them an unplea tance.Onl a for the oked the rage hypothesi but not gra of future nggressive behavior u thin a labor aory study presu ehavio while also ass art impressions of the supposed prov hould take the rabl aggres Study 4:Narcissism and Aggression in the laboratory ponsible for the original prov ocation Finally we To marshal direct eviden narcissistic vulnerability fuels those high on vulnerable sism when provoked. administered to another individ ord with theorizing about the role of fragility and shame in ual within a laboratory study.This procedure portrayed the study narcissistic rage Table 4 Correlations Between Narcissism,Anger,Aggression,and Entitlement in Study 3 2 3 M SD 1 _0 90 856
Results and Discussion Correlations between the variables in the model are presented in Table 4. As we anticipated, only vulnerability was a potent predictor of reactive and displaced aggression, whereas grandiosity was not. The structural model that examined mistrust, angry rumination, and entitlement as factors accounting for narcissistic aggression is presented in Figure 1; it was estimated using fullinformation maximum likelihood procedures in Mplus (Muthén & Muthén, 2012). The fit of the model was good, 2 (10) 19.41, p .03, comparative fit index .99, root mean square error of approximation .05, 95% confidence interval (CI) [.01, .08]. Several key results are worth highlighting. First, vulnerable narcissism was a significant predictor of mistrust and angry rumination, which accounted for its links to both reactive and displaced aggression. Second, grandiosity only predicted entitlement, with entitlement playing only a minor role in predicting reactive aggression and no role in predicting displaced aggression. Third and final, we were able to almost fully account for the links between both dimensions of narcissism and aggression, with the sole exception of the vulnerability-to-displaced-aggression pathway. Taken together, these results reveal narcissistic vulnerability as a key facilitator of reactive and displaced aggression, and point to mistrust and angry rumination as personality features responsible for these outcomes. The findings from the first three studies thus consistently reveal narcissistic vulnerability to be a driver of narcissistic rage, an explosive mix of mistrust, anger, and rumination that results in lashing out at those who stand in the way. However, the findings thus far are all based on concurrent self-reports of narcissism, anger, and aggression. To address this limitation and establish direct behavioral evidence for the narcissistic rage hypothesis, Study 4 evaluated narcissistic vulnerability and grandiosity as predictors of future aggressive behavior, unsuspectedly assessed within a laboratory study presumably investigating food tasting. This approach allowed us to objectively examine actual aggressive behavior, while also assessing participants’ emotions and their impressions of the supposed provocateur. Study 4: Narcissism and Aggression in the Laboratory To marshal direct evidence that narcissistic vulnerability fuels actual aggressive behavior, Study 4 measured aggressive behavior as a choice of a noxious stimulus administered to another individual within a laboratory study. This procedure portrayed the study to participants as a test of people’s food preferences and created an opportunity for participants to assign “hot sauce” to a presumed coparticipant who has in some way provoked the participant earlier during the study (as in the “hot-sauce paradigm”; see Lieberman, Solomon, Greenberg, & McGregor, 1999). Given that capsaicin (the chemical responsible for the spicy flavor, coughing, sweating, or choking) is an actual weapon that is used on other people (e.g., in pepper spray), this paradigm has appealing ecological validity (Milne, 1995). Moreover, we made a number of adjustments to the original paradigm in order to strengthen its internal validity (i.e., to ensure that the choice of hot sauce is a freely chosen act intended to cause harm to another; see Ritter & Eslea, 2005). In Study 4 we manipulated provocation and examined whether narcissism (assessed prior to the experiment) augmented aggressive responses to provocation. In addition, we manipulated whether the eventual target of aggression was the initial provocateur, or a third party who was not responsible for the original provocation (yet was rude, i.e., could “trigger” aggression). This enabled a direct assessment of both reactive and triggereddisplaced aggression (i.e., choosing a hot sauce intended either for the initial provocateur or an annoying third party, respectively). Note that the provocation in this paradigm does not involve an explicit “ego threat,” typical of research on narcissistic grandiosity and an established predictor of grandiose narcissists’ laboratory aggression (at least when assessed with the Competitive Reaction Time task; Bettencourt et al., 2006; Bushman & Baumeister, 1998; see also previous discussion). Rather, participants were provoked by believing another person freely assigned them an unpleasantly bitter substance. Only a narcissist that takes everything personally should be especially vengeful toward another person who provoked them in this way. In this vein, we expected narcissistic vulnerability, but not grandiosity, to augment aggressive responses to such provocation. Specifically, only those high on vulnerable narcissism should be especially likely to aggressively retaliate when provoked in this largely nonpersonal way. Furthermore, this retaliation should take place regardless of the target: When provoked, vulnerable narcissists’ anger should lead them to aggress against even those who are a minor annoyance but are not responsible for the original provocation. Finally, we expected those high on vulnerable narcissism, when provoked, to be especially prone to anger, hostility, as well as depression, in accord with theorizing about the role of fragility and shame in narcissistic rage. Table 4 Correlations Between Narcissism, Anger, Aggression, and Entitlement in Study 3 1 2 3 4 5 67 M SD 1. Grandiosity .80 15.69 6.72 2. Vulnerability .09 .84 2.73 0.69 3. Angry Rumination .06 .58 .92 1.98 0.53 4. Entitlement .37 .21 .17 .88 3.36 1.11 5. Mistrust .12 .42 .51 .02 .76 0.30 0.25 6. Reactive Aggression .04 .40 .48 .19 .23 .90 1.98 0.56 7. Displaced Aggression .03 .45 .50 .08 .27 .50 .90 2.05 0.59 Note. ns 374 to 378. Reliabilities appear in the diagonal. p .05. p .01. This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. NARCISSISTIC RAGE 791
792 KRIZAN AND JOHAR Method aboCerview Participants signed up for a study they thought wa supposed they would la have an op participants s with ally fiotiti for nant to tast participants to rate how much they like f from the ste categories (s nd we ing habits.Non ither a mildly bitte sweetened tea ( on)or vile this t vill give them ome information abou and e room for 30 had to tast both sauces before their choic and made the che d that the first step woud inolve the the oice onstitute the was th n the spic d eithe to a supposed I init Every part cipant will have a choice bet o thi f hi r fo you f theml.Th taste they fe s it fits in t Participant nd design Two hundred eight students parti signed to m at a m ginning of the seme used to t thi (o 68).Upor to th ory later in the d food selection made by the other partici ant (who a ut the par -subie d for bitter foods).The experimer reasine -oz containe ried tray back to the any nalyses.Th At this the ch the any key statisticala ister the Foo tions Questi edure and mea Participants came to a other people's food choic spects for them Giver tial that participants If no to finish the bitt st that ent (always of th ny were couraed todoso(all did).The exper the ex d th nlea le's reactions to foods ch en for them b The exn ete it as bonestly and acct as you can.Note tha exp another partici than left to p and then returned he roc basie taste categories)to the other person.The participants were indicated the perceived bitterness of what they tasted ("How would
Method Overview. Participants signed up for a study they thought was about people’s “food preferences,” specifically on how people react to other people’s food choices. In this vein, participants were to exchange information about their tastes with a presumed coparticipant (actually fictitious), who was to assign a particular food item for the participant to taste. This allowed manipulating provocation; the participants were assigned to taste something bitter and were assigned (presumably by their coparticipant) to have either a mildly bitter unsweetened tea (control condition) or a vile, disgusting bitter melon juice (provocation condition). Subsequently, participants thought they were randomly assigned to give hot, spicy food back to a coparticipant and were given a choice to assign them either a mild or a hot version of a spicy sauce. They had to taste both sauces before their choice and made the choice without the experimenter in the room. This choice constituted the main measure of aggression. Furthermore, the participants were to assign the spicy food either to a supposed coparticipant who assigned them the bitter food initially (original-target condition) or to a supposed future participant (displaced-target condition); this constituted the manipulation of the intended target of the participants’ aggression and enabled the assessment of trigger-displaced aggression. Additional “taste reactions” questionnaires administered during the study enabled the assessment of anger, depression, and perceived untrustworthiness of the provocateur, psychological reactions thought to underlie narcissistic rage. Participants and design. Two hundred eight students participated in exchange for course credit. They completed measures of narcissism at a mass-testing session at the beginning of the semester, with NPI–16 used to measure grandiosity ( .60; Ames, Rose, & Anderson, 2006) and the HNS used to measure vulnerability ( .68). Upon arriving to the laboratory later in the semester to participate in a study on “food preferences,” they were randomly assigned to one condition in a 2 (provocation: absent vs. present) 2 (target: original vs. displaced) between-subjects design. Upon the completion of experimental procedures, participants underwent a “funnel” debriefing procedure that probed them for suspicion with increasing levels of specificity. If a participant indicated any suspicion about the purpose of the study (i.e., even a vague sense of doubt about the veracity of the cover story), he or she was removed from the analyses. This resulted in the final sample of 182 participants. Inclusion of the removed participants did not change the outcome of any key statistical analyses. Procedure and measures. Participants came to a study they thought was about people’s “food preferences,” specifically about how people react to other people’s food choices for them. Given that it was essential that participants believed there was another coparticipant in the study, experimenters went to great lengths to suggest that another participant was present (always of the same gender). Following consent, the experimenter introduced the study as examining the “social dimension of eating,” aimed at assessing people’s reactions to foods chosen for them by others. The experimenter then explained that there was another participant in the laboratory who would remain physically separated so as to not bias participants’ taste reactions. The experimenter added that each person would be randomly assigned to give a particular form of a food (from one of the basic taste categories) to the other person. The participants were told that they will first be assigned a bitter food by their (supposed) coparticipant, while they would later have an opportunity to give spicy food to the same coparticipant. Then, the experimenter asked the participant to complete a “Food Preferences Inventory” in order to give some basic information about their tastes to their coparticipant. This questionnaire asked participants to rate how much they like foods from the basic taste categories (sweet, sour, bitter, salty, and spicy), with several examples provided for each, as well as filler questions about their eating habits. None of the participants indicated that they liked bitter food. Once finished, the experimenter took the participants’ inventory, said “Let me go and give this to the other participant—this will give them some information about your tastes. I will be right back,” and left the room for 30 s. Provocation phase. After returning, the experimenter explained that the first step would involve the participant tasting a food chosen for them by the coparticipant and that “bitter” was the assigned flavor. The experimenter added the following: Every participant will have a choice between a milder and a more concentrated version of the same taste category to allow for individual differences in taste preferences. In this case of bitter foods, the person assigning the food to you will have a choice between Black Tea or Bitter Gourd Juice [pointing out the bottles in front of them]. The person assigning the food can select any amount for the other person to taste they feel is appropriate, as long as it fits in the 4oz container. Later on, you will have a chance to assign a food to them. Note that all participants in this study will always sample the food item they are assigning to the other participant themselves first. In addition, the person tasting the food assigned to them will have to ingest the entire amount of that food. Any questions? At this point the experimenter left the room again to pick up the supposed food selection made by the other participant (who would have already presumably received information about the participants’ taste preferences via the Food Preferences Inventory, indicating participants’ dislike for bitter foods). The experimenter waited 30 s and poured either 3 oz of unsweetened tea (the control condition) or bitter gourd juice (the provocation condition) in a 4-oz container with a lid, and carried it on a tray back to the participant. The experimenter explained the following: At this point you will sample the bitter food chosen for you by the other participant. They have chosen the Bitter Gourd Juice [Unsweetened Tea]. Note that you have to ingest the entire amount. Following the tasting, we will administer the Food Reactions Questionnaire in order to assess your reactions to the food. When you ingest it, make sure to do so slowly and notice all the aspects of the taste that you can. Try to focus on the different sensations you are experiencing. If participants were reluctant to finish the bitter gourd juice (many were), they were encouraged to do so (all did). The experimenter concluded, “Thank you for tasting. To get a sense of your reactions, please complete this Food Reactions Questionnaire. Please complete it as honestly and accurately as you can. Note that your responses are completely confidential.” The experimenter than left to presumably check on the coparticipant, waited 60 s, and then returned. The Food Reactions Questionnaire had embedded items to assess key variables. As a manipulation check, the participants indicated the perceived bitterness of what they tasted (“How would This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. 792 KRIZAN AND JOHAR
NARCISSISTIC RAGE 793 you rate your liking of the iuice that you ipant concems. les from the PANAS-X.res Results and Discussion .wh manipu ions,followed by the ma and dep Correlations among pendent variable e (ie st,the expenr ntered [R.con ted [R].dishonest [R].moral.malevolen the bitter gourd juice rated it as mu ch more bitter (M.72.SD n the participant ted the bitter gourd juice rated it as much more unpleasan ond their supp out 28 ld he in the next study n (the displaced-target c ood Prefer the pr ation condition exn ned much more ang M hat the id not like spicy foods at all.Furtherm e in order to d copa ”int d-target col tion,the com ways had note that read."Don't be a jerk and give me Thed' 001.d nd in the 34).F1.173) were ened to all ds.Specifically.you not I any evidenc Buffalo wild wings Mild and Bu Wild win of participants chose the hot e in th that the 9.0%did s dition.F(.173).2.P001.odds ratio (OR)3.58 ner)Befor I will (() tion (F 49).Together with the esults o alysi iv eintent and that participants disp ced their aggression to a cover t with a lid.Their choice of hot over mild o given a Food Taste Preferences Inventory "somewhat"to "very much"wher f the on a scale of I (n aalho)to 9(e pected liking by their coparticipant ("How mu n do you think 4 M SD tremely dislike)to 9 lextremtely like) well as ne p dislike for spicy food(ow much did you rely on their er left rticipant to make their of sa privately.Upon their retum.the study was concluded.the exper g0p<05."p<0
you rate the bitterness of the juice that you sampled?”) on a scale of 1 (not at all bitter) to 9 (extremely bitter), and liking (“How would you rate your liking of the juice that you sampled?”) on a scale of 4 (extremely dislike) to 4 (extremely like). Anger and depression were assessed by aggregating items for the Hostility and Dejection subscales from the PANAS-X, respectively (Watson & Clark, 1994), which was presented, as needed, to assess current psychological state. Finally, we measured impressions of trustworthiness, as participants rated the presumed coparticipants on 12 interpersonal adjectives usinga1(not at all) to 11 (great amount) scale (i.e., likable, arrogant [R], friendly, genuine, trustworthy, self-centered [R], conceited [R], dishonest [R], moral, malevolent [R], considerate, and ethical), presumably to control for any biases. Aggressive-choice phase. The experimenter continued by stating it was then the participants turn to assign spicy food to the coparticipant. Participants were then told that they would assign this food either to their coparticipant (the original-target condition), or because their supposed coparticipant turned out to be allergic to capsaicin and could not taste it, to a participant that would be in the next study session (the displaced-target condition). They were also given the Food Preferences Inventory supposedly e-mailed by this other participant, which always clearly indicated that the person did not like spicy foods at all. Furthermore, in order to insure a “trigger” in the displaced-target condition, the comment section of the inventory from the supposed future coparticipant always had a written note that read, “Don’t be a jerk and give me something I don’t like!” The experimenter continued, Again, you were assigned to allocate spicy foods. Specifically, you will be allocating hot-sauce for them to sample. You will have a choice between Buffalo Wild Wings Mild and Buffalo Wild Wings Hot sauce. Just as you did, the person tasting the food you allocate will be required to ingest the entire amount. Note that the choice of food is completely up to you, and that you can select as little or as much as you want (provided it’s less than the 4 oz container). Before proceeding, you will have to sample both sauces in order to give you a sense of the foods. I will administer a very small amount of both the Mild and the Hot Sauce for you to taste. Both bottles were always full, and the experimenter administered a taste of each sauce to the participant via a straw. The experimenter then instructed the participant to select one of the sauces, pour or spoon the amount they selected in a clean 4 oz container, and cover it with a lid. Their choice of hot over mild sauce constituted the key dependent variable of aggression. The participants were also given a Food Allocation Record, which asked them to rate both of the sauces they tried on heat intensity (“How would you rate the hotness of the sauce that you will allocate?”) on a scale of 1 (not at all hot) to 9 (extremely hot), and expected liking by their coparticipant (“How much do you think will the other person like the sauce that you will allocate?”) on a scale of 1 (extremely dislike) to 9 (extremely like), as well as inquired whether participants looked at the presumed coparticipants’ Food Preferences Inventory (which always indicated a strong dislike for spicy food (“How much did you rely on their Food Preferences Inventory when choosing the food sample for them?”) on a scale of 1 (not at all) to 9 (very much). Finally, the experimenter left the participant to make their choice of sauce privately. Upon their return, the study was concluded, the experimenter conducted the funnel debriefing described earlier, explained the need for deception in detail, and assuaged any participant concerns. Results and Discussion We first present analyses of the manipulation check variables and experimental manipulations, followed by the main results regarding the impact of narcissism on aggression, anger, hostility, and depression. Correlations among dependent variables are shown in Table 5. First, the experimental manipulation of frustration via noxious stimulus was successful—participants who tasted the bitter gourd juice rated it as much more bitter (M 7.72, SD 1.40) than participants who tasted the tea (M 4.19, SD 1.95), F(1, 176) 184.9, p .001, d 2.1. Similarly, participants who tasted the bitter gourd juice rated it as much more unpleasant (M 3.54, SD .82) than participants who tasted the tea (M .24, SD 2.17), F(1, 176) 209.9, p .001, d 2.3 (no other main effects or interactions impacted these variables, all ps .28). Accordingly, a 2 2 ANOVA on anger confirmed that those in the provocation condition experienced much more anger (M 1.79, SD .78) than those in the control condition (M 1.15, SD .41), F(1, 173) 43.0, p .001, d 1.0. Similarly, participants attributed less trustworthiness to the presumed coparticipant in the provocation (M 5.43, SD 1.58) relative to the control condition (M 7.87, SD 1.30), F(1, 175) 113.7, p .001, d 1.7. Finally, participants reported virtually identical, low levels of depression in the provocation (M 1.22, SD .50) and in the control condition (M 1.17, SD .34), F(1, 173) .61, p .44, d .12. These analyses did not yield any evidence for other main effects or interactions, all ps .12. Did participants retaliate aggressively when provoked? Indeed, 26.1% of participants chose the hot version of the sauce in the provocation condition, whereas only 9.0% did so in the control condition, F(1, 173) 8.32, p .001, odds ratio (OR) 3.58. Moreover, the expected target of aggression did not matter much, with similar proportion of participants choosing the hot sauce in the original (20.2%) and the displaced target (14.8%) condition, F(1, 173) .001, p .98, OR 1.44, with no evidence of an interaction (F 1, p .49). Together with the experimental results on anger and trustworthiness reported earlier, this analysis supports the conclusion that choice of hot sauce reflected aggressive intent and that participants displaced their aggression to an “innocent” other when “triggered” by their minor annoyance. Furthermore, 90% of participants reported consulting the others’ Taste Preferences Inventory “somewhat” to “very much” when Table 5 Correlations Between Dependent Variables in Study 4 1234 M SD 1. Sauce Choice — .10 .06 .12 .17 .38 2. Trustworthiness .20 .91 .39 .08 6.66 1.89 3. Hostility .22 .63 .85 .46 1.48 .70 4. Sadness .09 .41 .52 .83 1.20 .43 Note. n 86 to 90 per condition. Control (provocation) condition correlations appear above (below) the diagonal, with reliabilities presented in the diagonal. p .10. p .05. p .01. This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. NARCISSISTIC RAGE 793
