Can an angry woman get ahead?

P SY CH OL OG I C AL S CIE N CE Research Article Can an Angry Woman Get Ahead? Status Conferral, Gender, and Expression of Emotion in the Workplace Victoria L. Brescoll1 and Eric Luis Uhlmann2 1 Yale University and 2Northwestern University ABSTRACT—Three studies examined the relationships among Though politicians have always used such tactics to defame anger, gender, and status conferral. As in prior research, their opponents, this particular case raises interesting questions men who expressed anger in a professional context were about whether expressing anger impedes a woman’s chances at conferred higher status than men who expressed sad- winning a political race or gaining status in other professional ness. However, both male and female evaluators conferred arenas. Generally, emotion theorists suggest that displays of lower status on angry female professionals than on angry certain emotions, such as anger, can communicate that an male professionals. This was the case regardless of the individual is competent and is entitled to high social status actual occupational rank of the target, such that both a (Shields, 2002, 2005; Tiedens, 2001). In a study consistent with female trainee and a female CEO were given lower status if this hypothesis, Tiedens (2001) found that men who expressed they expressed anger than if they did not. Whereas wom- anger in professional settings were more likely to be hired than en’s emotional reactions were attributed to internal char- men who expressed sadness and were also given more status, acteristics (e.g., ‘‘she is an angry person,’’ ‘‘she is out of power, and independence in their jobs. control’’), men’s emotional reactions were attributed to As Senator Clinton’s experience suggests, however, profes- external circumstances. Providing an external attribution sional women who express anger may experience a decrease, for the target person’s anger eliminated the gender bias. rather than an increase, in their status. Women are expected to Theoretical implications and practical applications are be kinder and more modest than men, and they evoke negative discussed. responses from other people if they fail to conform to this pre- scriptive stereotype (Heilman, 2001; Rudman, 1998). Female professionals who express anger violate this feminine norm and During an appearance on a national news program, the chairman therefore may not experience the boost in status enjoyed by of the Republican National Committee asserted that Senator angry men (a possibility acknowledged by Tiedens, 2001). Hillary Clinton was too angry to be elected president (Nagourney, This idea converges with Ekman and Friesen’s (1969) concept 2006). This comment caught the media’s attention because it of emotional display rules, which are ‘‘overlearned habits about appeared to be motivated by her gender: who can show what emotion . . . males should not cry; females, except in a maternal role, should not show anger’’ (Ekman, 1984, They are casting Hillary Clinton as an Angry Woman, a she- p. 320). Such an effect is likewise predicted by Rudman and monster melding images of Medea, the Furies, harpies . . . . This Fairchild’s (2004) integrative model of stereotype-based back- gambit handcuffs Hillary: If she doesn’t speak out strongly against lash. In their model, counterstereotypical actions are expec- President Bush, she’s timid and girlie. If she does, she’s a witch tancy violations that provoke negative reactions from social and a shrew. (Dowd, 2006, p. A21) perceivers. However, this backlash occurs only if a justification for derogating the counterstereotypical individual is available. In typical workplace situations in which a woman has expressed Address correspondence to Victoria Brescoll, Department of Psy- chology, Yale University, 309 Edwards St., New Haven, CT 06520, anger, there is probably enough ambiguity about the reason for e-mail: victoria.brescoll@yale.edu. her anger that some basis for derogation can be found. As a 268 Copyright r 2008 Association for Psychological Science Volume 19—Number 3  Victoria L. Brescoll and Eric Luis Uhlmann result, negative responses to women who express anger could $1.00. Most were Caucasian (85% on average), well educated, occur readily in professional contexts. and politically moderate (M 5 3.91 on a 7-point liberalism- Moreover, the expectation that a woman will not express anger conservatism scale). publicly should affect attributions for the cause of her anger. Kelley’s (1967, 1973) attribution model stipulates that when a STUDY 1 person’s behavior is characterized by low consensus (i.e., is different from that of peers), social perceivers are likely to at- Method tribute the behavior to internal characteristics (e.g., perceive Participants and Procedure anger as stemming from disposition, as opposed to features of the Thirty-nine males and 30 females (mean age 5 38.75 years) situation). According to this model, because anger and pride are were randomly assigned to view a videotaped job interview in the only emotions that people believe men express more than which either a male or a female professional described feeling women (Plant, Hyde, Keltner, & Devine, 2000), a woman’s anger either anger or sadness. The study employed a 2 (target’s gender: should be seen as internally caused (e.g., ‘‘she is an angry per- male vs. female)  2 (emotion: anger vs. sadness) between- son,’’ ‘‘she is out of control’’) rather than externally instigated subjects design. (e.g., ‘‘the situation was frustrating’’). Thus, people should view a We obtained the videotapes used by Tiedens (2001) and man’s anger as a response to objective, external circumstances, created two additional videotapes using a female target pre- but a woman’s anger as a product of her personality. As a result, a tested to be equivalent to Tiedens’s male target in attractiveness, professional woman’s anger may imply that she is not competent age, and ethnicity. In all three studies reported here, targets at dealing with workplace situations, and may therefore lead wore professional attire and were ostensibly being interviewed perceivers to accord her less status. for a job while sitting at a table; the interviewer was out of view of the camera. Targets described an incident in which they and a OVERVIEW OF THE PRESENT RESEARCH colleague lost an important account, and when asked by the interviewer how it made them feel, responded that the incident We report here three studies that tested these hypotheses. Study made them feel either angry or sad. 1 examined whether participants conferred lower status on an angry woman than on an angry man and whether attributions for emotional reactions underlay the relation between expression of Dependent Measures anger and conferral of status. Study 2 examined whether any Participants completed dependent measures in the order listed low-status individual, and not just women, would be given low in this section. status after expressing anger and whether the internal attribu- tion that angry female targets were ‘‘out of control’’ mediated the Status Conferral. Following Tiedens (2001), we created a compos- relation between expression of anger and conferral of status. ite measure of status conferral. This measure included four items Finally, Study 3 examined whether an angry woman who offered assessing how much status, power, and independence the candi- an external attribution for her anger would be accorded status as date deserved in his or her future job (1 5 none, 11 5 a great high as that of an angry man. Generally, people should view a deal) and whether the participant would hire the target person man’s anger as a response to objective, external circumstances, (1 5 never, 11 5 definitely; a 5 .91). but a woman’s anger as a product of her personality. As a result, a woman who expresses anger in the workplace will be accorded Salary. Participants reported the yearly salary they would pay lower status than a man who does the same, unless she em- the target. phasizes the external circumstances that caused her anger. All analyses reported in this article included gender of the Competence. Participants rated the target on the trait dimen- participant as an independent variable. No interactions involving sions of competent-incompetent and knowledgeable-ignorant, this variable were significant, and therefore we collapsed across using 11-point trait semantic differential scales (a 5 .79). gender for all analyses. For all the studies reported in this article, we recruited adult Attributions. Two questions assessed internal attributions (i.e., participants in order to obtain a relatively representative, non- ‘‘she/he became angry because of her/his personality,’’ and ‘‘she/ college-student sample with workplace experience (Sears, 1986). he is an angry person’’), and two questions assessed external We recruited these adult participants by placing flyers adver- attributions (‘‘she/he became angry because of the situation with tising the study at locations throughout a public park in Con- her/his colleague,’’ and ‘‘her/his colleague’s behavior caused necticut. Participants contacted us if they were interested in her/his anger’’). These four questions were answered using 11- participating, completed the experiment individually in isolated point scales (1 5 completely disagree, 11 5 completely agree). laboratory rooms on campus or at a private off-campus location, We reverse-scored the internal-attribution items and summed and were compensated with their choice of a lottery ticket or all four responses to create a composite measure (a 5 .72). Volume 19—Number 3 269  Anger, Gender, and Status Conferral TABLE 1 Results From Study 1: Mean Ratings of Male and Female Targets Expressing Anger or Sadness Anger Sadness Dependent variable Male Female Male Female Status conferral 6.47 (2.25) 3.75 (1.77) 4.05 (1.61) 5.02 (1.80) Yearly salary ($) 37,807 (13,825) 23,464 (10,496) 30,033 (9,255) 28,970 (9,884) Competence 7.55 (1.08) 5.44 (2.79) 5.79 (1.08) 6.17 (1.79) External attributions 7.72 (1.99) 5.80 (2.52) 6.57 (2.08) 6.94 (1.55) Note. Standard deviations are given in parentheses. Results and Discussion as significantly more competent than the sad male, t(32) 5 3.91, Table 1 presents mean scores for all four targets for all of the prep 5 .996. As expected, participants also viewed the angry male dependent measures. We hypothesized that there would be an as significantly more competent than the angry female, t(32) 5 interaction between emotion and the target’s gender. Specifi- 2.99, prep 5 .966. However, the sad female was not seen as signif- cally, we expected to replicate Tiedens’s (2001) finding that an icantly more competent than the angry female. angry man receives higher status, a higher salary, and higher competence ratings than a sad man. We also expected that partic- Attributions ipants would give the angry woman lower status and lower salary Attributions also showed a significant interaction between the than the angry man, would perceive her as less competent than target’s gender and emotion expression, F(1, 63) 5 5.20, prep5 the angry man, and would be more likely to attribute her anger .915. As expected, participants attributed the woman’s anger than his to internal, dispositional causes. more to internal factors and less to external factors than the male’s anger, t(31) 5 2.44, prep 5 .927. Moreover, attributions Status Conferral partially mediated the effects of expressing anger on the status A 2 (target’s gender: male vs. female)  2 (emotion: anger vs. accorded female professionals (Baron & Kenny, 1986). When we sadness) analysis of variance (ANOVA) conducted on the status- controlled for attributions, the coefficient (b for anger was sig- conferral scores revealed a significant interaction, F(1, 64) 5 nificantly reduced, from .59 to .35, Sobel z 5 2.20, p < .05. 16.38, prep 5 .996. As in Tiedens’s (2001) study, participants However, the effects of anger remained significant, which sug- conferred higher status on the angry male target than on the sad gests that attributions partially accounted for the observed male target, t(32) 5 3.59, prep 5 .986. Participants also con- effects. ferred significantly higher status on the angry male than on the angry female, t(31) 5 3.85, prep 5 .986.1 Furthermore, partic- STUDY 2 ipants conferred significantly higher status on the sad female than on the angry female, t(32) 5 2.07, prep 5 .882. The results of Study 1 suggest that expressing anger is an effective means of attaining higher status for professional men, Salary but not for professional women. To replicate Tiedens (2001), we The salary measure also showed a significant interaction be- used sadness as a comparison emotion in Study 1. However, tween gender of the target and emotion expression, F(1, 55) 5 sadness has its own unique connotations and therefore may not 5.46, prep 5 .921. Participants were willing to pay the angry male be a neutral control. Thus, in Study 2, we compared the effects of more than the sad male, t(26) 5 1.77, prep 5 .840, and signifi- anger and of expressing no emotion. cantly more than the angry female, t(25) 5 3.05, prep 5 .966. In Study 2, we also varied the targets’ occupational rank. Participants were willing to pay the angry female marginally less Perhaps angry women are given lower status than angry men than the sad female, t(29) 5 1.51, prep 5 .778. simply because women, on average, have lower status than men initially. People may find it presumptuous for any low-status Competence person, male or female, to display a high-status emotion such Perceptions of competence likewise showed a significant as anger. But if an angry woman receives lower status than an interaction between the target’s gender and emotion expression, angry man because of her gender, expressing anger should re- F(1, 65) 5 7.56, prep 5 .956. Participants viewed the angry male duce her status regardless of whether she is a powerful executive or a lowly trainee. 1 The findings of this study were replicated in a study in which participants In Study 2, we also attempted to clarify why angry women are read a transcript of the videos from Study 1 and rated the targets’ status. The angry female target received lower status than the angry male target and sad accorded low status by employing a more targeted measure female target, whereas the angry male received higher status than the sad male. of internal attributions—belief that angry women are out-of- 270 Volume 19—Number 3  Victoria L. Brescoll and Eric Luis Uhlmann control individuals. To the extent that anger is attributed to the asked how the situation made them feel and, in general, were individual’s personality rather than external circumstances, instructed to not express emotion. In pretesting, participants expressing that anger is likely to be perceived as a self-regu- accurately labeled the angry actors’ emotion and rated the differ- lation failure. Therefore, we hypothesized that participants ent actors’ anger as similarly intense and sincere, Fs < 1; all would view an angry woman as being out of control, and that this participants indicated that the no-emotion actors were not ex- internal attribution would help explain why angry women are pressing emotion. accorded low status. Results and Discussion Method Table 2 presents the mean scores for the eight targets for all of Participants and Procedure the dependent measures. Participants (70 males, 110 females; mean age 5 42.46 years) were randomly assigned to view one of eight videos. The study Status Conferral utilized a 2 (target’s gender: male vs. female)  2 (emotion: anger A 2 (target’s gender: male vs. female)  2 (emotion: anger vs. no vs. no emotion)  2 (occupation: high vs. low rank) between- emotion)  2 (occupation: high vs. low rank) ANOVA conducted subjects design. on status conferral revealed a main effect of the target’s occu- pational rank, F(1, 174) 5 9.25, prep 5 .974, and an interaction Dependent Measures between the target’s gender and emotion expression, F(1, 174) 5 As in Study 1, participants completed measures of perceived 6.69, prep 5 .95. For female targets, only a main effect of emotion competence, status conferral, and salary allocation. An addi- expression emerged, F(1, 85) 5 5.12, prep 5 .915; women were tional item assessed the specific attribution that the target was, accorded lower status when they expressed anger than when in general, an ‘‘in control’’ or ‘‘out of control’’ person (1 5 out of they were unemotional. For the male targets, only a main effect control, 11 5 in control). of target’s occupational rank emerged, F(1, 85) 5 7.28, prep 5 .956. Stimulus Materials Professional actors, different from those in Study 1 and matched Salary for age (middle to late 30s), ethnicity (Caucasian), and attrac- Results for salary paralleled those for status conferral. A three- tiveness, played the part of the interviewee in the videos. (In way ANOVA conducted on salary allocation revealed a main pretesting, 19 participants rated the actors as similarly attrac- effect of the target’s occupational rank, F(1, 155) 5 7.07, prep 5 tive and believed they were the same age, Fs < 1.) The script .953, and a significant two-way interaction between the target’s was identical to that used in Study 1 except that at the beginning gender and emotion expression, F(1, 155) 5 3.03, prep 5 .840. of the interview, targets described their occupation (either low- Participants’ suggested salary for the female targets was not rank assistant trainee or high-rank CEO) and, in the no-emotion based on their occupational rank, but rather was based on condition, the targets were not asked how they felt as a result of whether the targets expressed anger or remained unemotional. the botched work situation. Participants were willing to pay the unemotional female targets In the anger condition, the actors appeared moderately angry more than the angry female targets, even when the targets were when responding to the interviewer’s question, ‘‘How did that high-rank CEOs, F(1, 80) 5 4.13, prep 5 .883. In contrast, make you feel?’’ In the no-emotion condition, the actors were not participants were willing to pay the male CEOs more than the TABLE 2 Results From Study 2: Mean Ratings of Male and Female High- and Low-Rank Targets Expressing Anger or No Emotion Anger No emotion Dependent High rank Low rank High rank Low rank variable Male Female Male Female Male Female Male Female Status conferral 6.19 (2.23) 4.69 (2.03) 5.31 (2.02) 3.92 (1.60) 5.90 (2.27) 5.65 (2.31) 4.34 (2.14) 4.98 (2.38) Yearly salary ($) 66,434 42,526 30,781 24,590 82,368 54,404 41,404 56,318 (53,823) (17,765) (10,458) (10,069) (92,456) (25,731) (55,340) (90,441) Competence 7.36 (2.31) 5.39 (2.25) 6.70 (2.12) 6.26 (2.04) 6.66 (2.23) 7.85 (2.25) 6.64 (1.77) 6.54 (2.18) Out of control 4.12 (2.70) 6.41 (3.16) 4.77 (2.78) 6.61 (2.17) 3.73 (2.62) 4.24 (2.55) 4.59 (2.72) 4.35 (2.72) Note. Standard deviations are given in parentheses. Volume 19—Number 3 271  Anger, Gender, and Status Conferral male trainees, regardless of their expression of emotion, F(1, 75) 5 that preventing negative attributions for violations of gender- 7.61, prep 5 .959. based norms reduces penalties for counterstereotypical behav- ior. Therefore, in Study 3, we tested the hypothesis that if an Competence angry female professional provides an objective, external reason The competence scores showed a two-way interaction between for being angry, she should evoke less negative reactions. gender of the target and emotion expression, F(1, 172) 5 7.52, Experimentally manipulating this proposed mechanism (i.e., prep 5 .959, as well as a three-way interaction, F(1, 172) 5 4.82, attributions)—rather than treating it as a continuous variable— prep 5 .910. The two-way interaction mirrored the pattern found would further establish its validity as a mediator of the bias for status conferral and salary, and the three-way interaction against angry women (Spencer, Zanna, & Fong, 2005). indicated a particularly negative response to the high-rank an- The potential effectiveness of such an intervention is also gry female target. A planned contrast indicated that participants supported by Rudman and Fairchild’s (2004) model of backlash rated the angry female CEO as significantly less compe- effects. They proposed that violations of prescriptive gender tent than all the other targets, t(172) 5 3.01, prep 5 .953. stereotypes are most likely to provoke negative reactions when a potential rationalization for derogating the stereotype violator is Mediational Analyses available. To the extent that an angry female professional can For the internal-attribution ratings, a three-way ANOVA re- provide an objective, external reason for being angry, she should vealed the theoretically expected two-way interaction between evoke less negative reactions. gender of the target and emotion expression, F(1, 172) 5 5.78, prep 5 .938. A planned contrast indicated that participants viewed Method the angry female targets as significantly more out of control than Participants and Procedure the angry male targets and the unemotional male and unemo- Participants (51 males and 82 females; mean age 5 38.70 years) tional female targets, t(174) 5 4.80, prep 5 .986. watched one of six videos. The study utilized a 2 (target’s gender: A series of regression analyses tested the prediction that male vs. female)  3 (emotion: unexplained anger vs. explained perceptions of angry women as out of control would explain their anger vs. no emotion) between-subjects design. There was no failure to attain high status. For female targets, expression of external-attribution condition for the no-emotion condition be- anger was significantly related to the internal attribution of being cause it was not plausible to offer an external attribution for not out of control, r(89) 5 .39, p < .01, and status conferral, r(89) 5 expressing emotion. .23, p < .05. Furthermore, internal attribution was related to status conferral, r(89) 5 .46, p < .001. When we used internal Dependent Measures attribution and emotion expression as independent variables Participants completed measures of status conferral, salary al- to predict status conferral, only the coefficient for internal- location, and competence. attribution ratings remained significant, b(89) 5 .44, p < .001. The coefficient for emotion expression fell from .23 to .06 and was Stimulus Materials no longer significant, Sobel z 5 2.87, p < .01. Thus, the internal The videotapes from Study 2 were used, but with two modifi- attribution that an angry woman was out of control fully mediated cations. The information about the target’s occupational status the relationship between her expression of anger and the status was removed, and in the explained-anger condition, the target she was accorded (Baron & Kenny, 1986). made an external attribution for his or her anger—that a co- worker lied to the target by telling him or her that he had di- rections to the client’s office. This lie caused the target and STUDY 3 coworker to lose the account. Studies 1 and 2 found that angry women are accorded lower status than angry men, and Study 2 indicated that this is true Results and Discussion regardless of the woman’s actual status (i.e., whether she is a Table 3 presents the mean ratings for the targets for all of the CEO or an assistant trainee). Moreover, attributions appear to dependent measures. play a role in this phenomenon: People may confer low status on an angry woman because they see her behavior as arising from Status Conferral something deep and inherent—that is, from her being an angry For status conferral, a 2 (target’s gender: male vs. female)  3 and out-of-control person. If inherent, internal attributions un- (emotion: anger without external attribution vs. anger with ex- derlie this phenomenon, then an intervention designed to direct ternal attribution vs. no emotion) ANOVA revealed a significant attributions away from internal factors and toward external interaction between the target’s gender and emotion expression, factors might be effective at mitigating the bias. Lending support F(2, 34) 5 9.72, prep 5 .999. Examining each gender separately, to this hypothesis, Heilman and Okimoto (2007) demonstrated we found that the angry male without an external attribution 272 Volume 19—Number 3  Victoria L. Brescoll and Eric Luis Uhlmann received significantly higher status than the unemotional male, GENERAL DISCUSSION t(44) 5 2.55, prep 5 .95, and the angry male with an external attribution, t(45) 5 2.11, prep 5 .892. Results for the female The present studies document new phenomena at the nexus of targets supported our prediction; the angry female target who gender, emotional display rules, perceptions of status, and at- provided an external attribution for her anger received signifi- tributions for behavior. For men, expressing anger may height- cantly higher status than the angry female target who did not en status: Men who expressed anger in a professional context provide an external attribution, t(44) 5 3.53, prep 5 .986, but were generally conferred higher status than men who expressed did not receive higher status than the unemotional female tar- sadness. get, t(45) 5 0.22, prep 5 .251. Notably, the status conferred on For women, however, expressing anger had the opposite ef- the angry female target with an external attribution was not fect: Professional women who expressed anger were consistently significantly different from the status conferred on the angry accorded lower status and lower wages, and were seen as less male targets with or without a reason for their anger. This finding fur- competent, than angry men and unemotional women. And unlike ther suggests that our intervention attenuated the backlash against men’s occupational rank, women’s occupational rank (i.e., CEO the angry female. vs. trainee) did not influence status conferral, salary allocation, or judgments of competence. The derogated status of angry Salary women appeared to be due to the degree to which their behavior The results for salary paralleled those for status conferral. A 2 was seen as internally motivated—in particular, to the percep- (target’s gender: male vs. female)  3 (emotion: anger without tion that they were out of control. But when an angry woman external attribution vs. anger with external attribution vs. no offered an external attribution for her anger, she did not suffer emotion) ANOVA revealed a significant interaction between the the same loss in perceived status and competence. target’s gender and emotion expression, F(2, 112) 5 6.90, prep 5 These findings are consistent with the hypothesis that anger is .986. Looking at each gender separately, we found that the angry a status emotion (Shields, 2002; Tiedens, 2001). Moreover, they male target without an external attribution received a signifi- converge with prior theories regarding the nature of emotional cantly higher salary than the unemotional male target, t(38) 5 display rules (Ekman, 1984), as well as with the predictions 1.75, prep 5 .84, and the angry male target with an external of Rudman and Fairchild’s (2004) model of stereotype-based back- attribution, t(35) 5 1.79, prep 5 .842. For female targets, the lash. Finally, the finding that the nature of behavioral attributions angry female who provided an external attribution for her anger partly underlies status conferral for female targets is consistent received a higher salary than the angry female who did not pro- with Kelley’s (1967, 1973) attribution model, which predicts vide a reason for her anger, t(40) 5 3.24, prep 5 .979, but did not that behaviors perceived as different from those of one’s peers receive a higher salary than the unemotional female target, t(36) (e.g., for women, expressing anger) are often given personality- 5 0.44, prep 5 .383. The salary allocated to the angry female based explanations. Participants’ attributions of a woman’s an- target who provided an external attribution was not significantly ger to internal causes (e.g., ‘‘she is an angry person,’’ ‘‘she is different from the salary allocated to the angry male targets with or out of control’’) helped to explain the low status they conferred without a reason for their anger. on her. The present findings complement, but can be distinguished Competence from, those of previous work examining discrimination against For competence, the interaction between target’s gender and women who violate prescriptions for self-promotion. Women who emotion expression did not reach statistical significance, F(2, 134) promote their abilities are perceived as less likeable and less 5 2.24, prep 5 .807. Thus, although having an external attribution hireable than women who do not promote their abilities—but for her anger gave the angry female target a boost in status, it they are still seen as competent (Rudman, 1998). At the same apparently did not influence perceptions of her competence. time, women as a group are seen as warm, but relatively in- TABLE 3 Results From Study 3: Mean Ratings of Male and Female Targets Expressing Anger With or Without an External Attribution or Expressing No Emotion Anger without external attribution Anger with external attribution No emotion Dependent variable Male Female Male Female Male Female Status conferral 5.42 (1.63) 3.40 (1.44) 4.14 (2.46) 5.02 (1.66) 4.19 (1.67) 4.92 (1.65) Yearly salary ($) 46,024 (40,483) 21,130 (13,130) 27,171 (16,708) 34,368 (13,234) 29,100 (15,255) 32,421 (14,037) Competence 6.35 (2.03) 5.48 (2.17) 5.71 (2.62) 5.78 (2.03) 5.83 (1.68) 6.83 (2.14) Note. Standard deviations are given in parentheses. Volume 19—Number 3 273  Anger, Gender, and Status Conferral competent (Eagly & Mladinic, 1993; Fiske, Cuddy, Glick, & Xu, National Science Foundation Graduate Research Fellowships 2002). The explanation for why self-promoting women are seen while this research was undertaken. as competent, but angry women are seen as relatively incom- petent, may be that women who self-promote are explicitly as- REFERENCES serting their competence, whereas angry women are not—they are simply emoting. Albright, M. (2003). Madam secretary: A memoir. New York: Miramax. Although women’s anger appears to be sanctioned in a pro- Baron, R.M., & Kenny, D.A. (1986). The moderator-mediator variable fessional context, it may not be sanctioned in a family context distinction in social psychological research: Conceptual, strate- (Kring, 2000; Tavris, 1982). Also, it is not clear whether this bias gic and research considerations. Journal of Personality and So- cial Psychology, 51, 1173–1182. against angry women applies in professional contexts other than Dowd, M. (2006, February 8). Who’s hormonal? Hillary or Dick? The job interviews and early impression formation. It is possible that New York Times, p. A21. coworkers acquire more individuated information about each Eagly, A.H., & Mladinic, A. (1993). Are people prejudiced against other over time, so that this bias becomes less relevant. Addi- women? Some answers from research on attitudes, gender ste- tionally, the targets in the present studies were White American reotypes and judgements of competence. In W. Stroebe & M. Hewstone (Eds.), European review of social psychology (Vol. 5, pp. men and women, and it remains an empirical question whether 1–35). New York: Wiley. our findings hold for individuals of other ethnic groups, such as Ekman, P. (1984). Expression and the nature of emotion. In K. Scherer African American and Hispanic American men and women. & P. Ekman (Eds.), Approaches to emotion (pp. 319–343). Hills- Further, the importance of maintaining harmony in collectivistic dale, NJ: Erlbaum. cultures (Markus & Kitayama, 1991; Triandis, 1995) suggests Ekman, P., & Friesen, W.V. (1969). The repertoire of nonverbal behavior: that in much of the world, public expressions of anger by both Categories, origins, usage, and coding. Semiotica, 1, 49–98. Fiske, S.T., Cuddy, A.J., Glick, P., & Xu, J. (2002). A model of (often women and men may evoke negative reactions. Future research mixed) stereotype content: Competence and warmth respectively should examine these possibilities. follow from perceived status and competition. Journal of Per- Turning to real-world situations, such as the remarks about sonality and Social Psychology, 82, 878–902. Senator Clinton, one might ask, what can women do to mitigate Heilman, M.E. (2001). Description and prescription: How gender potential backlashes in response to anger? Professional women stereotypes prevent women’s ascent up the organizational ladder. face a dilemma: On the one hand, anger may serve as a powerful Journal of Social Issues, 57, 657–674. Heilman, M.E., & Okimoto, T. (2007). Why are women penalized for professional tool—for instance, to compel other people to fulfill success at male-typed tasks? The implied communality deficit. their responsibilities, or to castigate them for incompetence Journal of Applied Psychology, 92, 81–92. (Shields, 2002). On the other hand, to achieve and maintain Kelley, H.H. (1967). Attribution in social psychology. In D. Levine high social status, professional women may also have to behave (Ed.), Nebraska Symposium on Motivation: Vol. 15 (pp. 192–238). ‘‘unemotionally’’ so that they are seen as rational (Albright, Lincoln: University of Nebraska Press. 2003). Thus, it is important to identify strategies that profes- Kelley, H.H. (1973). The processes of causal attribution. American Psychologist, 28, 107–128. sional women can use to express anger without incurring a social Kring, A.M. (2000). Gender and anger. In A.H. Fischer (Ed.), Gender penalty. and emotion: Social psychological perspectives (pp. 211–231). The present studies take a step in this direction, demonstrating New York: Cambridge University Press. that external, situational explanations for anger ameliorate Markus, H., & Kitayama, S. (1991). Culture and the self: Implications negative responses to angry women. Specifically, a woman can for cognition, emotion, and motivation. Psychological Review, 98, express anger without incurring a drop in status to the extent that 224–253. Nagourney, A. (2006, February 5). Calling Senator Clinton ‘angry,’ her behavior evokes an external attribution for her anger. These G.O.P. chairman attacks. The New York Times, p. A16. results suggest a strategy that the social target herself can use, in Neuberg, S.L., & Fiske, S.T. (1987). Motivational influences on im- contrast to interventions that focus on social perceivers (e.g., pression formation: Outcome dependency, accuracy-driven at- accuracy motivation; Neuberg & Fiske, 1987) or institutional tention, and individuating processes. Journal of Personality and factors (e.g., outcome dependency; Rudman, 1998). Thus, by Social Psychology, 53, 431–444. Plant, E.A., Hyde, J.S., Keltner, D., & Devine, P.G. (2000). The gender offering external, situational explanations for anger, professional stereotyping of emotions. Psychology of Women Quarterly, 24, women may be able to express anger, while simultaneously ful- 81–92. filling one of the most basic social motivations: gaining status and Rudman, L.A. (1998). Self-promotion as a risk factor for women: The power. costs and benefits of counterstereotypical impression manage- ment. Journal of Personality and Social Psychology, 74, 629–645. Rudman, L.A., & Fairchild, K. (2004). Reactions to counterstereotypic Acknowledgments—We thank Larissa Tiedens for providing behavior: The role of backlash in cultural stereotype maintenance. Journal of Personality and Social Psychology, 87, 157–176. her videotapes; Geoff Cohen, George Newman, and Betsy Levy Sears, D.O. (1986). College sophomores in the laboratory: Influences of Paluck for feedback; and the American Psychological Associa- a narrow data base on social psychology’s view of human nature. tion (Division 35) for funding. Both authors were supported by Journal of Personality and Social Psychology, 51, 515–530. 274 Volume 19—Number 3  Victoria L. Brescoll and Eric Luis Uhlmann Shields, S.A. (2002). Speaking from the heart: Gender and the social Tiedens, L.Z. (2001). Anger and advancement versus sadness and meaning of emotion. London: Cambridge University Press. subjugation: The effect of negative emotion expressions on social Shields, S.A. (2005). The politics of emotion in everyday life: ‘‘Ap- status conferral. Journal of Personality and Social Psychology, propriate’’ emotion and claims on identity. Review of General 80, 86–94. Psychology, 9, 3–15. Triandis, H.C. (1995). Individualism and collectivism. Boulder, CO: Spencer, S.J., Zanna, M.P., & Fong, G.T. (2005). Establishing a causal Westview Press. chain: Why experiments are often more effective than mediation- al analyses in examining psychological processes. Journal of Personality and Social Psychology, 89, 845–851. Tavris, C. (1982). Anger: The misunderstood emotion. New York: Simon & Schuster. (RECEIVED 3/16/07; REVISION ACCEPTED 8/31/07) Volume 19—Number 3 275