Role Congruity Theory of Prejudice Toward Female Leaders PDF

Psychological Review 2002, Vol. 109, No. 3, 573–598 Copyright 2002 by the American Psychological Association, Inc. 0033-295X/02/$5.00 DOI: 10.1037//0033-295X.109.3.573 Role Congruity Theory of Prejudice Toward Female Leaders Alice H. Eagly Steven J. Karau Northwestern University Southern Illino...

Psychological Review 2002, Vol. 109, No. 3, 573–598 Copyright 2002 by the American Psychological Association, Inc. 0033-295X/02/$5.00 DOI: 10.1037//0033-295X.109.3.573 Role Congruity Theory of Prejudice Toward Female Leaders Alice H. Eagly Steven J. Karau Northwestern University Southern Illinois University at Carbondale A role congruity theory of prejudice toward female leaders proposes that perceived incongruity between the female gender role and leadership roles leads to 2 forms of prejudice: (a) perceiving women less favorably than men as potential occupants of leadership roles and (b) evaluating behavior that fulfills the prescriptions of a leader role less favorably when it is enacted by a woman. One consequence is that attitudes are less positive toward female than male leaders and potential leaders. Other consequences are that it is more difficult for women to become leaders and to achieve success in leadership roles. Evidence from varied research paradigms substantiates that these consequences occur, especially in situations that heighten perceptions of incongruity between the female gender role and leadership roles. advanced degrees (U.S. Bureau of the Census, 2000). In contrast, statistics pertaining to major leadership roles consistently show inequality: Women constitute 4% of the five highest earning officers in Fortune 500 companies and 0.4% of the CEOs (Catalyst, 2000); 13% of senators, 14% of congressional representatives, and 10% of state governors (Center for the American Woman and Politics, 2001); and 2% of military officers at the level of brigadier general and rear admiral or higher (U.S. Department of Defense, 1998). Similarly small proportions of women in most high positions also characterize other industrialized and industrializing nations (Adler & Izraeli, 1994; Melkas & Anker, 1997). Explanations for this sparse representation of women in elite leadership roles traditionally focused on the idea that a lack of qualified women created a “pipeline problem.” This shortage of women has been ascribed to a variety of causes, including women’s family responsibilities (Greenhaus & Parasuraman, 1999) and inherited tendencies for women to display fewer of the traits and motivations that are necessary to attain and achieve success in high-level positions (e.g., Browne, 1999; S. Goldberg, 1993). However, because the convergence of the sexes on many humancapital variables raises questions about the sufficiency of such explanations, it is vital to evaluate whether prejudice is one of the causes of women’s rarity in major leadership positions. To this end, we propose a role congruity theory of prejudice toward female leaders that (a) extends Eagly’s (1987; Eagly, Wood, & Diekman, 2000) social role theory of sex differences and similarities into new territory; (b) proposes novel, testable predictions about prejudice and its consequences; and (c) yields an effective organizing framework for a very large number of empirical findings from laboratories, field settings, organizations, and public opinion polls. In this article, we first explain the theory in general terms, then elaborate it by explaining the moderating conditions that allow it to make detailed predictions, and finally review empirical literature relevant to the theory. Leadership has been predominantly a male prerogative in corporate, political, military, and other sectors of society. Although women have gained increased access to supervisory and middle management positions, they remain quite rare as elite leaders and top executives. To explain this phenomenon, public and scientific discussion has centered on the idea of a “glass ceiling”—a barrier of prejudice and discrimination that excludes women from higher level leadership positions (Federal Glass Ceiling Commission, 1995; Morrison, White, & Van Velsor, 1987). To further this discussion, we advance a theory of prejudice toward female leaders and test the theory in relation to available empirical research. This integrative theory builds on social psychologists’ tradition of studying prejudice and stereotyping and industrial– organizational psychologists’ tradition of studying perceptions of managerial roles. The popularity of the glass ceiling concept may stem from the rarity of women in major leadership posts, despite the presence of equality or near equality of the sexes on many other indicators. A number of statistics thus suggest equality: In the United States, women make up 46% of all workers (U.S. Bureau of Labor Statistics, 2001b) and 45% of those in executive, administrative, and managerial occupations (U.S. Bureau of Labor Statistics, 2001a); women possess 51% of bachelor’s degrees and 45% of all A draft of this article was written while Alice H. Eagly was a visiting scholar at the Murray Research Center of Radcliffe Institute for Advanced Study and supported by a sabbatical award from the James McKeen Cattell Fund. The writing of this article was also supported by National Science Foundation Grant SBR-9729449 to Alice H. Eagly. We thank Amanda Diekman, Judith Hall, Lauren Gibbs, Eric Hansen, Madeline Heilman, Mary Johannesen-Schmidt, Richard Martell, Tracie Stewart, Janet Swim, Marloes Van Engen, Claartje Vinkenburg, and Wendy Wood for comments on an earlier version of the article. Correspondence concerning this article should be addressed to Alice H. Eagly, Department of Psychology, Northwestern University, 2029 Sheridan Road, Swift Hall 313, Evanston, Illinois 60208, or to Steven J. Karau, Department of Management, Southern Illinois University, Carbondale, Illinois 62901-4627. E-mail: [email protected] or skarau@cba .siu.edu Role Congruity Theory: How Gender Roles and Leader Roles Produce Two Types of Prejudice In general, prejudice can arise from the relations that people perceive between the characteristics of members of a social group 573 574 EAGLY AND KARAU and the requirements of the social roles that group members occupy or aspire to occupy (Eagly, in press). A potential for prejudice exists when social perceivers hold a stereotype about a social group that is incongruent with the attributes that are thought to be required for success in certain classes of social roles. When a stereotyped group member and an incongruent social role become joined in the mind of the perceiver, this inconsistency lowers the evaluation of the group member as an actual or potential occupant of the role. In general, prejudice toward female leaders follows from the incongruity that many people perceive between the characteristics of women and the requirements of leader roles. To introduce this theory, we first explain how gender roles represent consensual beliefs about men and women, then consider the typical degree of congruity between the female gender role and leader roles, and finally present the two forms of prejudice that follow from this analysis. Gender Roles: Expectations About the Actual and Ideal Behavior of Women and Men To represent people’s beliefs about women and men, our theory invokes the construct of gender role. Because social roles are socially shared expectations that apply to persons who occupy a certain social position or are members of a particular social category (Biddle, 1979; Sarbin & Allen, 1968), gender roles are consensual beliefs about the attributes of women and men. As Eagly (1987) noted in her initial presentation of social role theory, “These beliefs are more than beliefs about the attributes of women and men: Many of these expectations are normative in the sense that they describe qualities or behavioral tendencies believed to be desirable for each sex” (p. 13). Thus, important in social role theory is that idea that roles include two kinds of expectations, or norms. To adopt Cialdini and Trost’s (1998) terms, roles thus include descriptive norms, which are consensual expectations about what members of a group actually do, and injunctive norms, which are consensual expectations about what a group of people ought to do or ideally would do. Although the descriptive norms are thus synonymous with psychologists’ usual definitions of stereotypes of group members, injunctive norms add a prescriptive element not traditionally included in the stereotype construct. The term gender role thus refers to the collection of both descriptive and injunctive expectations associated with women and men.1 According to social role theory, perceivers infer that there is correspondence between the types of actions people engage in and their inner dispositions. The descriptive aspect of gender roles thus originates in perceivers’ correspondent inferences from the observed behavior of men and women to their personal qualities— that is, from the activities that men and women commonly perform in their typical social roles to the personal qualities that are apparently required to undertake these activities (Eagly, 1987; Eagly et al., 2000). Gender stereotypes thus follow from observations of people in sex-typical social roles— especially, men’s occupancy of breadwinner and higher status roles and women’s occupancy of homemaker and lower status roles (see Eagly et al., 2000). Both the descriptive and injunctive aspects of gender roles are well documented. Evidence that descriptive norms— or stereotypes—are associated with women and men is abundant: People believe that each sex has typical—and divergent—traits and be- haviors (e.g., Broverman, Vogel, Broverman, Clarkson, & Rosenkrantz, 1972; Diekman & Eagly, 2000; Newport, 2001; J. E. Williams & Best, 1990a). A key proposition of social role theory is that the majority of these beliefs about the sexes pertain to communal and agentic attributes (Bakan, 1966; Eagly, 1987). Communal characteristics, which are ascribed more strongly to women, describe primarily a concern with the welfare of other people—for example, affectionate, helpful, kind, sympathetic, interpersonally sensitive, nurturant, and gentle. In contrast, agentic characteristics, which are ascribed more strongly to men, describe primarily an assertive, controlling, and confident tendency—for example, aggressive, ambitious, dominant, forceful, independent, self-sufficient, self-confident, and prone to act as a leader. Although other types of attributes are also differentially ascribed to women and men (Deaux & Lewis, 1983, 1984; Eckes, 1994), it is these communal and agentic attributes that especially illuminate the issues of prejudice that we discuss in this article. Evidence abounds that gender roles also embrace injunctive norms about male and female behavior. Specifically, the overall approval of communal qualities in women and agentic qualities in men has been demonstrated in research on (a) the beliefs that people hold about ideal women and men (e.g., Spence & Helmreich, 1978; J. E. Williams & Best, 1990b), (b) the beliefs that women and men hold about their ideal selves (W. Wood, Christensen, Hebl, & Rothgerber, 1997), and (c) attitudes and prescriptive beliefs that people hold about the roles and responsibilities of women and men (e.g., Glick & Fiske, 1996; Spence & Helmreich, 1978). Moreover, J. A. Hall and Carter (1999) showed that as behaviors become more sex differentiated in actuality (as assessed by meta-analytic data), people judge them as increasingly appropriate for only one sex. It thus appears that people tend to think that women and men ought to differ, especially in those behaviors that are associated with larger sex differences. Consistent with social role theory (Eagly, 1987; Eagly et al., 2000) and social– cognitive research (see S. T. Fiske, 1998), gender roles have pervasive effects. Not only is sex the personal characteristic that provides the strongest basis of categorizing people, even when compared with race, age, and occupation (A. P. Fiske, Haslam, & Fiske, 1991; Stangor, Lynch, Duan, & Glass, 1992; van Knippenberg, van Twuyver, & Pepels, 1994), but also stereotypes about women and men are easily and automatically activated (Banaji & Hardin, 1996; Banaji, Hardin, & Rothman, 1993; Blair & Banaji, 1996). In addition, encoding processes 1 Consistent with social role theory (Eagly, 1987), other researchers have adopted the distinction between descriptive and injunctive aspects of gender roles and have used varying labels for this distinction, including descriptive stereotypes and prescriptive stereotypes (e.g., Burgess & Borgida, 1999; S. T. Fiske & Stevens, 1993) and gender stereotypes and gender ideology (e.g., Cota, Reid, & Dion, 1991). Typical descriptive measures are positively correlated with typical prescriptive or injunctive measures (see Burgess & Borgida, 1999). Although we too might have adopted the descriptive stereotype versus prescriptive stereotype terminology, we prefer role terminology because (a) most definitions of social role include injunctive (or prescriptive) beliefs, whereas most definitions of stereotype do not; (b) analyses of leadership have generally adopted role terminology by analyzing managerial and leader roles; and (c) role terminology promotes an understanding of organizational behavior because roles are the building blocks of organizations (D. Katz & Kahn, 1978). ROLE CONGRUITY THEORY advantage information that matches gender-stereotypical expectations (von Hippel, Sekaquaptewa, & Vargas, 1995), and spontaneous tacit inferences fill in unspecified details of male and female social behavior to be consistent with these expectations (Dunning & Sherman, 1997). The activation of beliefs about women and men by gender-related cues thus influences people to perceive women as communal but not very agentic and men as agentic but not very communal. Perceivers are likely to correct for the effects of these expectations only under somewhat narrow circumstances that may require both awareness that such expectations have been activated and the intention or motivation to counter their influence (e.g., Blair & Banaji, 1996; Devine, Monteith, Zuwerink, & Elliot, 1991; Kunda & Sinclair, 1999; Plant & Devine, 1998). Congruity of Gender Roles and Leadership Roles Role congruity theory is grounded in social role theory’s treatment of the content of gender roles and their importance in promoting sex differences in behavior (Eagly et al., 2000). However, role congruity theory reaches beyond social role theory to consider the congruity between gender roles and other roles, especially leadership roles, as well as to specify key factors and processes that influence congruity perceptions and their consequences for prejudice and prejudicial behaviors. The potential for prejudice against female leaders that is inherent in the female gender role follows from its dissimilarity to the expectations that people typically have about leaders. Prejudice can arise when perceivers judge women as actual or potential occupants of leader roles because of inconsistency between the predominantly communal qualities that perceivers associate with women and the predominantly agentic qualities they believe are required to succeed as a leader. People thus tend to have dissimilar beliefs about leaders and women and similar beliefs about leaders and men. In Schein’s (1973, 1975) early empirical demonstration of this masculine construal of leadership, male and female managers gave their impressions of either women, men, or successful middle managers. These respondents perceived successful middle managers as considerably more similar to men than women on a large number of mainly agentic characteristics such as competitive, self-confident, objective, aggressive, ambitious, and able to lead. Researchers have replicated Schein’s (1973, 1975) findings, not only in the United States (e.g., Heilman, Block, Martell, & Simon, 1989; Massengill & di Marco, 1979) but also in the United Kingdom, Germany, Japan, China, and Singapore (J. Lee & Hoon, 1993; Schein, 2001). Also, studies in which participants merely rated managers (Arkkelin & Simmons, 1985; Powell & Butterfield, 1979, 1984, 1989) or holders of political offices (Rosenwasser & Dean, 1989) on agentic and communal characteristics have demonstrated similarly masculine construals of leadership. Combining of gender roles and leader roles. Observing an individual as an occupant of a leader role would place expectations based on gender in competition with expectations based on leadership. The idea that gender roles would continue to have some influence is consistent with the general argument that these roles are consequential in organizational settings. For example, Gutek and Morasch (1982) argued that gender roles spill over into the workplace, and Ridgeway (1997) maintained that gender provides 575 an “implicit, background identity” (p. 231) in the workplace. In agreement with these claims, formal tests of models of the integration of expectations from multiple roles supported a weighted averaging model, with weights responsive to the task relevance of the expectations (e.g., Berger, Norman, Balkwell, & Smith, 1992; Hembroff, 1982). To the extent that weighted averaging describes this combinatorial process, the ease with which expectations come to mind—that is, their accessibility—should also affect weighting. Because frequently and recently activated constructs are more accessible in memory (Higgins, 1996) and gender roles are automatically activated by gender-related cues in virtually all situations, the high accessibility of expectations based on gender likely maintains their impact. Therefore, in thinking about female leaders, people would combine their largely divergent expectations about leaders and women, whereas in thinking about male leaders, people would combine highly redundant expectations. Empirical evidence is consistent with the principle that perceivers blend the information associated with a gender role and a leader role. For example, in a 1972 Louis Harris national opinion poll (Mueller, 1986; Sapiro, 1983), respondents reported that female holders of political offices were especially suited for pursuing what could be regarded as the communal goals of public policy (e.g., dealing with children and family problems, assisting the poor, working for peace) and that male officeholders were especially suited for pursuing what could be regarded as the agentic goals of public policy (e.g., directing the military and handling big business, domestic rebellion, the economy, and foreign relations). Further research showed that participants’ inferences about leaders’ sex-typed communal and agentic attributes mediated these perceptions of the abilities of female and male political leaders (Huddy & Terkildsen, 1993b). Similarly, Heilman and her colleagues demonstrated that male managers rated “women managers” as more agentic and less communal than “women in general” but not as close as “men managers” to a group identified as “successful middle managers” (Heilman, Block, & Martell, 1995; Heilman et al., 1989; see also Martell, Parker, Emrich, & Crawford, 1998). Leader behavior as violating injunctive norms of the female gender role. Despite the likely influence of gender roles in organizational settings, clear evidence that a woman conforms to the requirements of the leader role that she occupies should to some extent restrain gender-stereotypical inferences about her. Consistent with this prediction, both male managers (Heilman et al., 1995) and graduate students in business (Dodge, Gilroy, & Fenzel, 1995) perceived that female managers who were described as successful were almost as similar to successful managers in general as successful male managers were. Paradoxically, perceiving a female manager or leader as very similar to her male counterpart may produce disadvantage. This disadvantage can arise from the injunctive norms associated with the female gender role. Because women who are effective leaders tend to violate standards for their gender when they manifest male-stereotypical, agentic attributes and fail to manifest femalestereotypical, communal attributes, they may be unfavorably evaluated for their gender role violation, at least by those who endorse traditional gender roles. This reaction reflects the general tendency 576 EAGLY AND KARAU for deviations from injunctive norms to elicit disapproval (Cialdini & Trost, 1998).2 A woman who fulfills a leader role may thus elicit negative reactions, even while she may also receive some positive evaluation for her fulfillment of this role. Some evidence of this mix of positive and negative evaluations emerged in Heilman et al.’s (1995) finding that, even when the researchers described female managers as successful, participants regarded these women as more hostile (e.g., more devious, quarrelsome, selfish, bitter) and less rational (i.e., less logical, objective, able to separate feelings from ideas) than successful male managers. This research coheres with informal evidence of ambivalence that is inherent in the epithets often applied to powerful women, such as dragon lady and Battle-Ax (Tannen, 1994). More specific examples of such labeling include (a) Dawn Steel, the first woman to head a major movie studio in Hollywood, who was known as “Steel Dawn” and “the Tank” (Weinraub, 1997) and (b) Margaret Thatcher, British prime minister, who was labeled not only as “Iron Lady” but also as “Her Malignancy” and “Attila the Hen” (Genovese, 1993). To the extent that a woman who fulfills a leader role elicits a mixture of positive and negative reactions—that is, an ambivalent reaction—storage of these reactions in memory could have a variety of effects. As attitude researchers have shown, ambivalence can produce less consistency in expressions of an attitude across time and situations and a propensity for reactions to polarize—that is, become very negative or even sometimes very positive— depending on the particulars of the judgment context (e.g., I. Katz, Wackenhut, & Hass, 1986; see reviews by Eagly & Chaiken, 1993, 1998). Contexts that elicit especially unfavorable evaluations of a female (but not a male) leader include receiving a negative evaluation from her or otherwise having one’s ego threatened by her (Atwater, Carey, & Waldman, in press; Kunda & Sinclair, 1999; Sinclair & Kunda, 2000). Two Forms of Prejudice Toward Female Leaders A distinctive feature of our theory is the proposition that prejudice toward female leaders and potential leaders takes two forms: (a) less favorable evaluation of women’s (than men’s) potential for leadership because leadership ability is more stereotypical of men than women and (b) less favorable evaluation of the actual leadership behavior of women than men because such behavior is perceived as less desirable in women than men. As we have explained, the first type of prejudice stems from the descriptive norms of gender roles—that is, the activation of descriptive beliefs about women’s characteristics and the consequent ascription of female-stereotypical qualities to them, which are unlike the qualities expected and desired in leaders. The second type of prejudice stems from the injunctive norms of gender roles—that is, the activation of beliefs about how women ought to behave. If female leaders violate these prescriptive beliefs by fulfilling the agentic requirements of leader roles and failing to exhibit the communal, supportive behaviors preferred in women, they can be negatively evaluated for these violations, even while they may also receive some positive evaluation for their fulfillment of the leader role. Women leaders’ choices are thus constrained by threats from two directions: Conforming to their gender role would produce a failure to meet the requirements of their leader role, and conform- ing to their leader role would produce a failure to meet the requirements of their gender role. These two forms of prejudice would both be manifested in less favorable attitudes toward female than male leaders and potential leaders. In addition, these two forms of prejudice should produce (a) lesser access of women than men to leadership roles and (b) more obstacles for women to overcome in becoming successful in these roles. The lesser access would follow from the tendency to ascribe less leadership ability to women, and the obstacles to success could follow from this aspect of prejudice as well as from the preference that women not behave in ways that are typical of leaders. Key tests of our theory thus examine sex differences in terms of less positive attitudes toward female than male leaders, more difficulty for women than men in achieving leadership roles, and more difficulty for women than men in becoming effective in these roles. Women would not always be targets of prejudice in relation to women in leadership roles, because various conditions would moderate these prejudices. Because the first form of prejudice toward female leaders follows from incongruity between the descriptive content of the female gender role and a leadership role, prejudice would be lessened or absent to the extent this incongruity is weak or absent. In addition, because the second form of prejudice follows from incongruity between a leader’s behavior and the injunctive content of the female gender role, other moderators would affect this form of prejudice. Specifically, the more agentically a leader role is defined or the more completely women fulfill its agentic requirements, the more likely such women are to elicit unfavorable evaluation because their behavior deviates from the injunctive norms of the female gender role. In addition, the role incongruity principle allows for prejudice against male leaders, to the extent that there exist leader roles whose descriptive and injunctive content is predominantly feminine. Because leadership is generically masculine, such leader roles are rare, and ordinarily women but not men are vulnerable to role incongruity prejudice in relation to leadership. Yet, as we show below, this prejudice against female leaders is responsive to various features of the leader’s situation and characteristics of the perceiver. Key tests of our theory thus examine the impact of theory-relevant moderating variables on attitudes toward female and male leaders as well as on women’s and men’s access to and success in leadership roles. Conditions That Moderate Role Incongruity Prejudice First Form of Prejudice To be predictive in a wide range of situations, our role incongruity theory must take into account the variability that exists in how gender roles and leadership roles are defined. Because role definitions are affected by many contextual and dispositional factors, numerous variables should moderate the two forms of prejudice that we have postulated. We first consider variables that affect the first form of prejudice—that is, regarding women as less 2 In addition, negativity could follow from judging each sex’s agentic qualities in relation to a within-sex standard, which would be lower for women than men (Biernat, 1995). A given agentic behavior delivered by a woman could thus be less positively evaluated because its greater subjective extremity makes it seem excessive. ROLE CONGRUITY THEORY qualified than men for leadership. These several moderators can be understood in terms of the single principle that we have noted— namely, that the greater the incongruity between the descriptive norms that define the female gender role and a leader role, the more likely that women are perceived as less qualified for leadership. Masculinity of leader role. Variability in defining leader roles is a major influence on the amount of incongruity between the female gender role and leader roles. Although people ascribe predominantly masculine attributes to the generic leader role (see Bass, 1990; Heilman, 1983, 1995; Kruse & Wintermantel, 1986; Martin, 1992; Nieva & Gutek, 1980, 1981; Ragins & Sundstrom, 1989; Schein, 2001), there is considerable variation in definitions of these roles. Therefore, greater incongruity follows from defining leader roles in particularly masculine terms—that is, in terms of qualities that are more agentic and less communal. To the extent that leader roles are less masculine, they would be more congruent with the female gender role, and therefore the tendency to view women as less qualified than men should weaken or even disappear. Moreover, more masculine definitions of leadership roles would reduce the likelihood that perceivers would spontaneously categorize women as leaders or potential leaders. Yet, such categorization would advantage person information that is consistent with leadership as demonstrated by Phillips and Lord (1982) and Kenney, Schwartz-Kenney, and Blascovich (1996). Men could thus experience a double advantage over women in perceived leadership ability— one increment from activating the male gender role, with its complement of agentic qualities, and a second increment from being categorized as a leader or potential leader and thereby gaining the agentic qualities associated with leadership. It is important to consider the specifics of how leadership roles are defined because research has found that such role definitions vary widely across leadership domains, such as military, educational, health care, business, and political (Lord, Foti, & de Vader, 1984; Lord & Maher, 1993). Another determinant is the functional area of management within organizations, such as production, marketing, and personnel (Gomez-Mejia, McCann, & Page, 1985). Especially relevant to the masculinity versus femininity of leader roles is the level of these roles in organizational hierarchies (Hunt, Boal, & Sorenson, 1990; Lord & Maher, 1993). To examine this variable, researchers have asked managers at differing levels to rate the importance of various abilities or activities for their jobs (e.g., Alexander, 1979; Gomez-Mejia et al., 1985; Paolillo, 1981; Pavett & Lau, 1983). At the lower level, managers favored abilities involved in direct supervision such as leading, monitoring potential problems, and managing conflict. At the middle level, managers reported that their jobs demanded greater human relations skills that involve fostering cooperative effort and motivating and developing subordinates. At the executive level, managers believed that their jobs required a greater range of skills and activities, including monitoring information, serving as a liaison, manifesting entrepreneurial ability, and engaging in long-range planning. Also, a study of male managers’ perceptions of the characteristics of successful executives yielded a set of highly agentic qualities— specifically, the ability to act as a change agent (e.g., inspirational, decisive), managerial courage (e.g., courageous, resilient), results orientation (e.g., action oriented, proactive), and leadership (e.g., leader, strategic thinker; Martell et al., 1998). Corroborating this association between higher level leadership and agentic qualities is 577 a study of the attributes that people ascribe to political officeholders (Huddy & Terkildsen, 1993a; see also Mueller, 1986). Although agentic characteristics were perceived to be more important than communal characteristics for all officeholders, agentic characteristics became even more important for higher level offices. In short, the incongruity between the female gender role and leader roles is likely to be most extreme at the highest levels of leadership. Yet, such incongruity might be somewhat lower for middle managers than for first level, or “line” managers, given the more socially complex elements of middle management activity. In principle, extent of role incongruity could vary in response to differing definitions of gender roles as well as leader roles. However, essentially similar descriptive beliefs about the communal and agentic qualities of the sexes are apparently held by men and women, students and older adults, people who differ in social class and income, and citizens of many nations (e.g., Broverman et al., 1972; Jackman, 1994; Survey Research Consultants International, 1998; J. E. Williams & Best, 1990a). Research thus suggests more constancy in the stereotypes that people hold about women and men than the stereotypes they hold about leaders. Sex of perceivers. Individual differences in social perceivers may also influence the incongruity between leader and gender roles. One such variable is perceivers’ sex, in view of evidence that men often have a more masculine construal of leadership than do women. Evidence of this construal emerges mainly from studies on perceptions of the managerial role similar to those by Schein (1973, 1975). According to Schein’s (2001) review of this research, these newer studies have shown that in the United States, but not in several other nations (the United Kingdom, Germany, Japan, China), women, compared with men, generally have a more androgynous view of managerial roles as requiring communal qualities as well as agentic ones. This difference between men’s and women’s perceptions of managers likely reflects the considerably greater experience of women with female managers. Even beyond the tendency of male and female managers to be clustered in different managerial occupations and functional specialities within organizations (U.S. Bureau of Labor Statistics, 2001a), men are far less likely than women to have a female manager (Reskin & Ross, 1995). Because of their lesser experience with female managers, men would be less likely to imbue the social category with androgynous meaning. As a consequence of their more masculine perception of leadership, male perceivers should show a stronger tendency than female perceivers to view women as less qualified than men for leadership. Another mechanism that may enhance the tendency for men to view women as less qualified for leadership derives from men’s greater social power, which can enhance their tendency to use gender-stereotypical information rather than available individuating information (Goodwin, Operario, & Fiske, 1998). Other moderators. Other likely influences on the perceived incongruity between the female role and leader roles include cultural milieus. Although the association of women with communal characteristics and men with agentic characteristics for the most part has generalized across nations (e.g., J. E. Williams & Best, 1990a; J. E. Williams, Satterwhite, & Best, 1999), cultural and subcultural variation warrants further exploration. For example, leader roles and the female gender role may be more consistent in the African American community than the European American community as Parker and Ogilvie (1996) argued. Change in role 578 EAGLY AND KARAU definitions over time remains a possibility as well, although there is little evidence that the descriptive content of the female and male gender roles has changed the past few decades, at least among American college students (e.g., Lueptow, Garovich, & Lueptow, 1995; Spence & Buckner, 2000). However, to the extent that leader roles become more androgynous (see the Conclusion section), the tendency to perceive women as less qualified than men should lessen. Other interactions would follow from factors that increase the weight given to the female gender role, as opposed to the leader role, when perceivers combine the descriptive content of the two roles. In particular, certain feminine personal characteristics (e.g., pregnancy, feminine dress and grooming) might increase the accessibility of the female gender role and consequently make women seem particularly unqualified for leadership, compared with their male counterparts. Sex ratios should also produce interactions, because people who are in a small minority in a group on the basis of their sex attract more attention and in general are perceived more stereotypically (Kanter, 1977; Taylor & Fiske, 1978). Given that women are increasingly in a minority position as they rise in organizational hierarchies, their female-stereotypical qualities become more salient, and the resulting perception of them in terms of these qualities produces disadvantage (Reskin, McBrier, & Kmec, 1999; Yoder, 1991). Still other interactions may occur because people rely more on stereotypical beliefs when their cognitive resources are limited—for example, when they are under time pressure (e.g., Kruglanski & Freund, 1983) or when they experience information overload (e.g., Macrae, Hewstone, & Griffiths, 1993). The impact of gender roles may thus be enhanced when cognitive resources are reduced. In contrast, other conditions might decrease the weight assigned to gender roles relative to leader roles—for example, perceivers’ own outcomes might be made dependent on accurate assessment of others’ leadership ability (Rudman, 1998; Shackelford, Wood, & Worchel, 1996). Second Form of Prejudice Other moderating conditions affect the second form of prejudice, which takes the form of evaluating the behavior of women who occupy leadership roles less favorably than the equivalent behavior of men—a reaction that follows from the injunctive aspects of the female gender role. Some of these moderators reflect the following general principle: The more agentically a leader role is defined or the more completely women fulfill its agentic requirements, the more likely such women are to elicit unfavorable evaluation because their behavior deviates from the injunctive norms of the female gender role. Deviation from prescriptions about desirable female behavior would be greater to the extent that the leader role that women fulfill has a particularly agentic definition—as military officer roles do, for example. Also affecting a female leader’s perceived deviation from the qualities preferred in women is the agentic style of her own behavior: To the extent that women fulfill their leader roles in a particularly dominant, assertive, directive, or selfpromoting style, they present greater deviation from the injunctive norms of the female gender role and would receive less positive reactions. In addition, adding communal features to leadership behavior, even though these are not required by the leader role, could allow women leaders to fulfill aspects of the female role. They may thereby receive more positive reactions if they include in their repertoire some behaviors that are expressive, friendly, and participative, as long as these are not viewed as inappropriate for their leader role. Finally, another moderator increasing the perceived deviation of female leaders from the injunctive norms of the female gender role and thereby increasing negative evaluation of them is perceivers’ personal endorsement of these norms—that is, their approval of traditional gender arrangements or disapproval of nontraditional arrangements. Researchers have had considerable success in producing measures of individual differences in approval of traditional gender roles (e.g., Glick & Fiske, 1996; Spence & Helmreich, 1978). Approval of the traditional female gender role also varies according to (a) sex of perceiver, with men showing more approval (e.g., Glick & Fiske, 1996); (b) year of data collection, with a decline over time in approval (e.g., Spence & Hahn, 1997; Twenge, 1997); and (c) culture, with citizens of some nations showing more approval (e.g., Best & Williams, 1993; Glick & Fiske, 2001). In summary, the perception of incongruity between leadership roles and the female gender role often results in prejudice toward female leaders. This incongruity arises because social perceivers typically construe leadership roles in agentic terms, whereas they expect and prefer that women exhibit communal characteristics. Yet, the degree of perceived incongruity between a leader role and the female gender role would depend on many factors, including the definition of the leader role, the weight given to the female gender role, and personal approval of traditional definitions of gender roles. Relevance of Other Theories of Gender Prejudice The role congruity theory of prejudice contrasts very sharply with classic theories that view prejudice as arising from holding an unfavorable stereotype and consequently a negative attitude toward a social group (e.g., Allport, 1954; Esses, Haddock, & Zanna, 1993). From that perspective, women’s underrepresentation as leaders would be ascribed to a negative stereotype and attitude toward women in general. However, such a context-free theory could not explain why women are discriminated against in some roles but not in others. Moreover, research on evaluations of women and men as social groups challenges this approach with evidence that women are not regarded as less good than men, even though they are perceived as inferior to men in power and status (see Eagly & Mladinic, 1994; Langford & MacKinnon, 2000; J. E. Williams & Best, 1990a). In fact, in many recent studies, women have been evaluated somewhat more favorably than men, even in investigations using implicit attitudinal measures that assess the strength of association between male and female category labels and evaluative words (Carpenter, 2001). This perceived goodness of women appears to derive mainly from the communal aspects of the female gender role. Therefore, biases toward female leaders cannot result from the mere extension of a generalized negative attitude toward women. Departing from traditional ideas about prejudice toward women, Glick and Fiske (1996, 2001) developed a theory of ambivalent sexism that encompasses both positivity and negativity toward women. These researchers maintained that prejudice toward women encompasses approval of women in traditional roles, la- ROLE CONGRUITY THEORY beled benevolent sexism, and hostility toward women in nontraditional roles, labeled hostile sexism. These two forms of prejudice do not correspond to the two forms that we propose. Instead, they both elaborate the injunctive aspects of role incongruity prejudice by noting that incongruity elicits negative (or hostile) reactions, whereas congruity elicits positive (or benevolent) reactions. Thus, women occupying incongruent— or nontraditional—roles receive relatively negative reactions, whereas women occupying congruent— or traditional—roles receive more positive reactions. Glick and Fiske (1996) considered both types of reactions to be sexist. More related to our role congruity theory of prejudice toward female leaders is Heilman’s (1983) lack-of-fit model of bias in work settings. Heilman (1983) proposed that, to the extent that a workplace role is inconsistent with the attributes ascribed to an individual, she or he would suffer from perceived lack of fit to the workplace role, producing decreased performance expectations, increased expectations of failure, and decreased expectations of success. These effects would lower self-evaluation and evaluation by others. Heilman (1983) further argued that gender stereotypes would affect the perception of individuals’ attributes and produce lack of fit with workplace roles that are perceived to require attributes stereotypical of the other sex. This theory is an important predecessor of our own because we, like Heilman (1983) emphasize the descriptive content of gender roles. Moreover, Heilman (2001) added a recognition of the importance of the prescriptive content of gender stereotypes, in concert with other scholars who have noted that many of the beliefs held about women are descriptively and prescriptively incongruent with many workplace roles (e.g., Nieva & Gutek, 1981). However, our theory transforms these insights into a systematic theory by joining social– cognitive research on stereotyping and prejudice and industrial– organizational research on management and leadership. This conceptual breadth of our theory also allows it to account for a wide range of moderating conditions in terms of common underlying mechanisms. Burgess and Borgida (1999) proposed another related theory in the context of studying sexual harassment and discrimination. They argued that the “descriptive and prescriptive components of gender stereotypes” (p. 666) result in different types of workplace discrimination. Specifically, they held that descriptive stereotypes lead to discrimination that is labeled disparate impact, in which “institutional practices result in hiring and promotion decisions that are biased against a class of people” (p. 666), whereas prescriptive stereotypes lead to discrimination that is labeled disparate treatment, in which “women who violate prescriptive stereotypes of femininity are punished, either through hostile environment harassment or through the devaluation of their performance” (p. 666). Although this approach provides a welcome complement to our own theory in arguing for the potential of descriptive and injunctive gender roles to produce discrimination in workplace settings, the approach is not specifically tailored to provide an analysis of leadership. Examining Research Relevant to the Theory We now present research that has investigated the consequences of prejudice that are predicted by our role congruity theory. This research provides many opportunities to examine the general prejudicial effects that this theory predicts as well as the moderating conditions it proposes. These paradigms are organized according 579 to whether they investigate (a) attitudes toward women and men as leaders, (b) the access of men and women to leadership roles, or (c) the success of women and men in leadership roles. The diversity of paradigms that we consider illustrates the power of the role congruity theory of prejudice to integrate research carried out by social and industrial– organizational psychologists, management specialists, political scientists, sociologists, and economists. If we follow Allport’s (1954, p. 9) rule that “The net effect of prejudice is to place the object of prejudice at some disadvantage not merited by his own conduct,” only disadvantage not merited by group members’ own behavior can be confidently labeled as prejudicial. Therefore, to take Allport’s dictum into account, we show that these consequences of prejudice are intact in experimental research paradigms that equate the objective characteristics of male and female leaders other than their sex. Organizational studies and other field studies that do not equalize the behavior of the sexes can offer advantages of external validity, but such studies are vulnerable to the criticism that actual sex differences might account for apparently prejudicial reactions. For example, holding a less favorable attitude toward female than male leaders could follow from observed behavioral differences between them or from female leaders’ tendency to occupy less powerful positions (Kanter, 1977). Also, in natural settings in which leaders’ behavior is free to vary, any lesser access of women to leadership or lesser success in these roles could reflect the behavioral nonequivalence of the men and women who are compared. Therefore, whenever possible, we examine experiments that equated the attributes of men and women, but we also review field and correlational studies that did not provide this control. Do People Have a Less Favorable Attitude Toward Women Than Toward Men in Leader Roles? The two forms of prejudice that we postulated, stemming from the descriptive and injunctive aspects of the female gender role, should jointly produce less favorable attitudes toward female than male leaders and potential leaders, although this female disadvantage should have decreased over time. The Gallup Poll illustrates the attitudinal approach by asking respondents if they would prefer a male or female boss (Simmons, 2001b). The responses for selected years ranging from 1953 to 2000 appear in Table 1. Although this question once elicited greater preference for male bosses on the part of men than women, it recently yielded greater preference for male bosses on the part of women than men. Yet, women consistently indicated greater preference for female bosses than men did. This seeming paradox is explained by men’s more sharply increasing tendency to volunteer that the sex of bosses makes no difference. Despite these complex trends, a strong preference for male bosses over female bosses was present for both sexes at all time points, and this preference was present in all 22 of the nations that Gallup surveyed in 1995 (Simmons, 2001b). Researchers have developed scales to assess attitudes toward female managers. Most widely used is the Women As Managers Scale, developed by L. H. Peters, Terborg, and Taynor (1974), which contains items such as “It is not acceptable for women to assume leadership roles as often as men” and “Women are not ambitious enough to be successful in the business world.” This measure and a similar instrument developed by Dubno, Costas, Cannon, Wanker, and Emin (1979) have shown less favorable EAGLY AND KARAU 580 Table 1 Responses by Men and Women in Selected Years to Gallup Poll’s Question About Preferring a Man or Woman as Boss Response and sex of respondents Man preferred as boss Men Women Total Woman preferred as boss Men Women Total Either sex or no difference (volunteered) Men Women Total 1953 1975 1982 1995 2000 75 57 66 63 60 62 40 52 46 37 54 46 45 50 48 2 8 5 4 10 7 9 15 12 17 22 20 19 26 22 21 29 25 32 27 29 46 30 38 44 24 33 35 22 28 Note. Data are reported in percentages of respondents giving each of the indicated responses to the question, “If you were taking a new job and had your choice of a boss, would you prefer to work for a man or a woman?” (Simmons, 2001b). Numbers do not add up to 100 because “no opinion” is not shown. that by 1985, this bias against women on the part of male executives had abated considerably but not disappeared. Several items administered in the General Social Survey, a nationally representative U.S. survey, are informative about attitudes toward women and men as political leaders (see Table 2 for items). These data show some overall intolerance of women compared with men in political roles, increasing tolerance for women between 1974 and 1998, and few differences between male and female respondents. The question about voting for a qualified woman for president was also included in Gallup polls between 1958 and 1972 (see Ferree, 1974). Like the later surveys, these earlier surveys yielded gradually increasing willingness to vote for a female presidential candidate of one’s own party, moving from approval by only 54% in 1958 to 69% in 1972. In response to a 1999 Gallup Poll question concerning whether a woman or man would make a better president, everything else being equal, 42% of respondents favored a man, 31% favored a woman, and 22% attitudes toward managerial women among men than women in samples of employees (Terborg, Peters, Ilgen, & Smith, 1977; Tomkiewicz & Adeyemi-Bello, 1995) and students (Dubno, 1985; Dubno et al., 1979) but some favorable change among men between 1975 and 1987 (Russ & McNeilly, 1988). The Harvard Business Review twice published surveys of the attitudes of large samples of male and female executives (Bowman, Worthy, & Greyser, 1965; Sutton & Moore, 1985). Among the items administered in both surveys was the general question, “Overall, would you say your own basic attitude toward women in management is: strongly favorable, mildly favorable, indifferent, mildly unfavorable, or strongly unfavorable?” Choosing the strongly or mildly favorable responses were 35% of the men and 82% of the women in 1965, compared with 73% of the men and 91% of the women in 1985. The general conclusion from these surveys was that in 1965 the majority of male executives, but not female executives, disapproved of women in executive roles and Table 2 Responses by Men and Women in Selected Years to General Social Survey Items Pertaining to Political Leadership Item and sex of respondent Agreement with “Women should take care of running their homes and leave running the country up to men” Men Women Total Agreement with “Most men are better suited emotionally for politics than are most women” Men Women Total Yes response to “If your party nominated a woman for president, would you vote for her if she were qualified for the job?” Men Women Total Note. 1974 1978 1982 1985 1990 1994 1998 36 35 36 31 32 32 30 27 28 25 27 26 18 18 18 15 14 14 15 16 15 48 47 47 41 46 44 39 36 38 38 39 39 28 26 27 22 20 21 21 22 22 81 80 80 83 80 82 86 86 86 85 80 82 90 89 89 92 92 92 91 90 90 Data are reported in percentages of respondents (National Opinion Research Center, 1998). ROLE CONGRUITY THEORY indicated that the sex of the president would not matter; on this question, women were more likely than men to favor a woman— 39% versus 22% (Simmons, 2001a). Opinion polls can suffer from contamination by respondents’ self-presentational concerns, especially to the extent that they believe that indicating a preference for one sex might reveal prejudice. Studies oriented to circumventing this problem have applied implicit measurement techniques to examine attitudes toward female and male leaders. Using the Implicit Association Test (Greenwald, McGhee, & Schwartz, 1998), Carpenter (2001) assessed the strength of student participants’ associations between the terms male leaders and female leaders (and in one experiment the names of actual male and female leaders) and pleasant and unpleasant evaluative words (e.g., laughter, excellent, cancer, terrible). The results of her several experiments suggested that the women had a more favorable implicit attitude toward female than male leaders, whereas men’s attitudes were inconsistent across the experiments (i.e., more positive toward male leaders, toward female leaders, or equal evaluation of male and female leaders). Using a different type of implicit measure involving priming (see Fazio, Jackson, Dunton, & Williams, 1995), Rudman and Kilianski (2000) primed student participants with schematic drawings of male or female authority figures (e.g., doctor, boss, judge) and assessed latency of response to positive or negative adjectives. This method suggested that both male and female students had negative attitudes toward female authority figures and more neutral attitudes toward male authority figures. However, women showed less prejudice than did men on an explicit measure on which respondents indicated their preferences for male versus female authorities. Reconciling the differences in findings from various implicit and explicit measurement paradigms awaits additional research. In general, attitudinal data show some disapproval of female leaders, although on a decreasing basis and sometimes with more disapproval on the part of male respondents. Although it is not surprising that there are some inconsistencies in these findings, in view of the considerable diversity of measures and participant populations, they are supportive of the role congruity theory of prejudice. However, because factors other than prejudice could underlie less favorable attitudes toward female than male leaders (e.g., differing behavior on the part of female and male leaders), we turn to other research paradigms, some of which equate the objective characteristics of men and women. Do Women Have Less Access Than Men to Leadership Roles? One prediction following from our role congruity theory is that achieving leadership is more difficult for women than men, because of the common perception that women have less leadership ability and (often) the preference that women not exhibit this ability and instead engage in communal, supportive behavior. The plausibility of this proposition would be enhanced by evidence that women have less access than men to positions of power in natural settings as suggested by indicators such as wages and promotions. Providing another paradigm are experiments on reactions to hypothetical male and female job candidates whose characteristics have been equated. Also relevant is research on the perception of task competence and leadership ability and on the evaluation of 581 agentic behavior. Studies of the emergence of leaders from initially leaderless groups are also informative. We review relevant evidence from each of these research paradigms. Studies of Actual Wages and Promotion Researchers have examined sex discrimination in wages and promotion to determine whether the returns for skills and job characteristics are greater for one sex than the other, once other determinants of wages are taken into account. Although these methods are generally accepted, they are vulnerable to omitting determinants of wages that may explain sex differentials and to including determinants that may be biased by prior sex discrimination. Despite these ambiguities, there is general agreement that such studies have demonstrated wage discrimination against women (e.g., Bergmann, 1989; Blau & Ferber, 1987; Jacobsen, 1998). Although Stanley and Jarrell’s (1998) meta-analysis of the results of 41 studies estimating wage discrimination showed an unequivocal decrease over time (see also Solberg, 1999), other detailed analyses confirm that some wage discrimination against women remains in the United States (Bayard, Hellerstein, Neumark, & Troske, 1999). The improvement of women’s wages relative to those of men in recent years probably reflects several broader changes in labor market dynamics (e.g., the stagnation of men’s wages due to economic restructuring and the globalization of industry; see Morris & Western, 1999). In addition, potential causes consistent with the role congruity theory of prejudice include the lessening numerical tokenism of women in many management positions and the increasing power of women to make personnel decisions as owners of small businesses and as decision makers in larger organizations (see Reskin et al., 1999). For example, in the savings and loan industry the traditional tendency for women to hold positions at lower but not higher managerial levels decreased between 1975 and 1987 (Cohen, Broschak, & Haveman, 1998). This study’s results suggested the importance of sex ratios and women’s role as decision makers—specifically, promotions of women into a higher job level were more likely to the extent that a higher proportion of women was already at that upper level (up to 50%) as well as at the applicants’ current job level. Jacobs (1992) examined wages in a nationally representative sample of managers. Among these managers, the magnitude of the discriminatory wage gap favoring men exceeded the wage gap in the labor force as a whole, although this managerial gap decreased over time. Averaging across the sectors of the economy, female managers were given less authority and earned less than male managers, even when managerial level and tenure in organizations were controlled (Reskin & Ross, 1995). Other studies showed that female managers, compared with their male counterparts at the same level, had less access to the high-level responsibilities and the complex challenges that are likely precursors to promotion (Lyness & Thompson, 1997; Ohlott, Ruderman, & McCauley, 1994). In addition, Olson and Frieze’s (1987) narrative review of research on managers’ incomes, including 12 studies of recipients of the masters of business administration (MBA) degree, found little evidence for a sex difference in starting salaries but evidence of a later wage gap favoring men that increased the longer that individuals were followed. This trend, which was typically explained only in pa