SCIENCE WATCH

An Ambivalent Alliance
Hostile and Benevolent Sexism as Complementary Justifications for Gender Inequality
Peter Glick Susan T. Fiske Lawrence University Princeton University

The equation of prejudice with antipathy is challenged by recent research on sexism. Benevolent sexism (a subjectively favorable, chivalrous ideology that offers protection and affection to women who embrace conventional roles) coexists with hostile sexism (antipathy toward women who are viewed as usurping men's power). The Ambivalent Sexism Inventory, first validated in U.S. samples, has been administered to over 15,000 men and women in 19 nations. Hostile and benevolent sexism are complementary, crossculturally prevalent ideologies, both of which predict gender inequality. Women, as compared with men, consistently reject hostile sexism but often endorse benevolent sexism (especially in the most sexist cultures). By rewarding women for conforming to a patriarchal status quo, benevolent sexism inhibits gender equality. More generally, affect toward minority groups is often ambivalent, but subjectively positive stereotypes are not necessarily benign.

If woman had no existence save in thefictionwritten by men, one would imagine her a person . .. very various; heroic and mean; splendid and sordid; infinitely beautiful and hideous in the extreme. Virginia Woolf, A Room of One's Own

This article reviews recent theory and empirical research on hostile and benevolent sexism. Hostile sexism is an adversarial view of gender relations in which women are perceived as seeking to control men, whether through sexuality or feminist ideology. Although benevolent sexism may sound oxymoronic, this term recognizes that some forms of sexism are, for the perpetrator, subjectively benevolent, characterizing women as pure creatures who ought to be protected, supported, and adored and whose love is necessary to make a man complete. This idealization of women simultaneously implies that they are weak and best suited for conventional gender roles; being put on a pedestal is confining, yet the man who places a woman there is likely to interpret this as cherishing, rather than restricting, her (and many women may agree). Despite the greater social acceptability of benevolent sexism, our research suggests that it serves as a crucial complement to hostile sexism that helps to pacify women's resistance to societal gender inequality. In 19 nations, more than 15,000 participants have completed the Ambivalent Sexism Inventory (ASI; Glick & Fiske, 1996; Glick et al., 2000), a 22-item self-report measure of sexist attitudes with separate 11-item Hostile and Benevolent Sexism scales (see the Appendix). Hostile and benevolent sexism are prevalent across cultures, and crosscultural differences in ambivalent sexism are predictable and systematic, with both ideologies relating to national measures of gender inequality. Moreover, underlying the differences between cultures are important consistencies in the structure and consequences of sexist beliefs. What ASI research reveals about the nature of sexism challenges current definitions of prejudice as an unalloyed antipathy
Editor's note. Denise C. Park served as action editor for this article. Author's note. Peter Glick, Department of Psychology, Lawrence University; Susan T. Fiske, Department of Psychology, Princeton University. Correspondence concerning this article should be addressed to Peter Glick, Department of Psychology, Lawrence University, Appleton, WI 54912-0599. Electronic mail may be sent to peter.s.glick@lawrence.edu.

hat Woolf saw as "astonishing extremes" in men's images of women date back to ancient texts. Pomeroy (1975), a social historian, suggested that classical representations of women fit into the polarized categories of goddesses, whores, wives, and slaves. Feminists who analyze contemporary society (e.g., Faludi, 1992) argue that similarly extreme characterizations of women are alive and well in popular culture, such as film depictions that divide women into faithful wives and murderous seductresses. Although what Tavris and Wade (1984) termed the pedestal-gutter syndrome (or the Madonna-whore dichotomy) has long been recognized by psychologists, historians, and feminists, most empirical researchers have identified sexism only with hostility toward women, ignoring the corresponding tendency to place (at least some) women on a pedestal. February 2001 American Psychologist
Copyright 2001 by the American Psychological Association, Inc. 0003-066X/01/S5.00 Vol. 56, No. 2, 109-118 DOI: 10.1037//O003-066X.56.2.1O9

109

business across the globe (United Nations Development Programme. 1998). The standard model of prejudice would suggest, then, that attitudes toward women must be overwhelmingly hostile and contemptuous. Recent research, however, shows that overall attitudes toward women are quite favorable. Eagly and Mladinic (1993) found that both men and women have more favorable attitudes toward women than toward men, attributing an extremely positive set of traits to women. Known as the "women are wonderful" effect, this finding is extremely robust and has been replicated (though more strongly for women than for men) even with implicit (i.e., tionconscious) attitudes (Carpenter, 2000). The preference for women creates a conundrum for prejudice theorists: How can a group be almost universally disadvantaged yet loved? Answers to this riddle come from several quarters. Eagly and Mladinic (1993) pointed out that the favorable, communal traits ascribed to women (e.g., nurturing, helpful, and warm) suit them for domestic roles, whereas men are presumed to possess the traits associated with competence at high-status roles (e.g., independent, ambitious, and competitive). Furthermore, women's stereo typically communal attributes are also the traits of deference that, when enacted in daily interaction, place a person in a subordinate, less powerful position {Ridgeway, 1992). Thus, the favorable traits attributed to women may reinforce women's lower status. Indeed, Jackman (1994), in her persuasive analysis of race, class, and gender relations, argued that subordination and affection, far from being mutually exclusive, often go hand-in-hand. Dominant groups prefer to act warmly toward subordinates, offering them patronizing affection as a reward for "knowing their place" rather than rebelling. Open antagonism is reserved for subordinates who fail to defer or who question existing social inequalities. But can subjectively benevolent attitudes he a form of prejudice? By Allport's (1954) definition of prejudice as an antipathy, the answer is no. Yet, Allport immediately followed his definition by stating that "the net effect of prejudice . . . is to place the object of prejudice at some disadvantage" (p. 9). Allport's afterthought suggests that the crux of prejudice may not be antipathy but social inequality; if so, a patronizing bul subjectively positive orientation toward women that reinforces gender inequality is a form of prejudice.

Peter Glick

and draws attemion to the manner in which subjectively benevolent, paternalistic prejudices (e.g., benevolent sexism) may reinforce inequality between groups.

The Nature of Sexism

Allport (1954), in his foundational book entitled The Nature of Prejudice, defined prejudice as "an antipathy based upon a faulty and inflexible generalization" (p. 9). Although some (e.g., Brown, 1995) have questioned the latter part of (his definition, virtually all psychological theorists have likewise equated prejudice widi antipathy. From antipathy, it is assumed, flow the discriminatory acts that disadvantage targets of prejudice. In addition, because people seek to justify social systems by believing that groups deserve their place in the social hierarchy (Jost & Banaji, 1994; Tajfel, 1981), a group's disadvantaged status reinforces prejudice, presumably creating a vicious positive feedback loop between antipathy and social inequality. Even members of low-status groups may endorse such system-justifying ideologies, despite the fact that these beliefs bolster their group disadvantage (Jost & Banaji, 1994). On the basis of cross-cultural indicators of status and power, women are clearly a disadvantaged group. Although some cultures are more egalitarian than others, patriarchy is widespread (Eagly & Wood, 1999; Harris, 1991; Pratto, 1996), though not necessarily universal (Salzman, 1999). Hunter-gatherer societies (common to an earlier era of human history), in which wealth could not be accumulated, may have been relatively egalitarian, bul the idea that matriarchy was once common has been thoroughly debunked (Harris, 1991). Simply put, men typically rule, dominating the highest status roles in government and
10

Why Benevolent Prejudices Matter

Benevolent sexism is a subtle form of prejudice, ye! the ideology it represents may be far from trivial in promoting gender inequality. Both hostile and benevolent sexism are presumed to be "legitimizing ideologies," beliefs that help to justify and maintain inequality between groups (Sidanius, Pratto, & Bobo, 1994). Ideologies of benevolent paternalism allow members of dominant groups to characterize their privileges as well-deserved, even as a heavy responsibility that they must bear. Consider, for example, the ideology of the "White man's burden" articulated by February 2001 American Psychologist

Susan T. Fiske

Rudyard Kipling (1899). the poel-apologist of British imperialism: Take up the White Man's burden Send forth the besi ye breed Go, bind your sons to exile To serve your captives' need; To wait in heavy harness, On fluttered folk and wild Your sullen, new-caught peoples. Half devil and half child, (p. 290) The widespread belief that Europeans were redeeming the primitive masses was essential to maintaining colonialism {Hochschiid, 1998). The resources (including cheap labor) of occupied territories were viewed as fair payment for European enlightenment. Indigenous peoples who rebelled were perceived as ungrateful children or savages who must be severely disciplined. In the United States, even outright slavery was legitimated through paternalisiic ideologies (Jackman, 1994). Benevolent sexism may serve functions similar to belief in the White man's burden, allowing men to maintain a posiiive self-image as protectors and providers who are willing to sacrifice their own needs to care for the women in their lives. On its own. this ideology may seem unobjectionable, even laudable, but what if (similar to the case of the White man's burden) it is a crucial complement to hostile sexism, helping to justify men's greaier privilege and power? If men's power is popularly viewed as a burden gallantly assumed, as legitimated by their greater responsibility and self-sacrifice, then their privileged role seems justified. Furthermore, women who seek power may consequently be February 2001 American Psychologist

perceived as ungrateful shrews or harpies deserving of harsh treatment, consistent with Kipling's (1899) lament that the White man could expect only to "'reap his old reward/ The blame of those ye better/ The hate of those ye guard" (p. 290). Equally important is the way in which benevolent paternalism may reduce women's resistance to patriarchy (cf. Jackman, 1994). Benevolent sexism is disarming. Noi only is it subjectively favorable in its characterization of women, but it promises that men's power will be used to women's advantage, if only they can secure a high-status male protector. To the extent that women depend on men to be their protectors and providers, they are less likely to protest men's power or to seek (heir own independent status. For instance, Rudman and Heppen (2000) found that college women who implicitly associated male romantic partners with chivalrous images (e.g.. Prince Charming) had less ambitious career goals, presumably because they were counting on a future husband for economic support. In a related study, Moya, Exposito, and Casado (1999) found, in a community sample of Spanish women, that those who did not have paid employment scored higher in benevolent sexism. These researchers also explored the women's reactions to discriminatory scenarios {e.g., losing a promotion to a less qualified man or having their husband forbid them to go out at night). The same acts of discrimination were perceived as less serious when the perpetrators expressed a benevolent, protective justification as opposed to a hostile one. Furthermore, women who scored higher in benevolent sexism were more likely to excuse not only benevolently justified discrimination by nonintimate men (e.g., a boss) but also overtly hostile discrimination by a husband. The latter effect occurred only for women without paid employment, suggesting that women who are highly dependent on male partners are prone to forgive even hostile acts, perhaps reinterpreting them as a sign of the husband's passionate attachment. Thus, women who endorse benevolent sexism are more likely to tolerate, rather than challenge, sexist behavior when the sexist's motivation can be interpreted as being protective.

Hostile and Benevolent Sexism: Universal Prejudices?

We hypothesized that hostile and benevolent sexism are predictable products of structural relations between men and women that are common to human societies: (a) Men are typically accorded more status and power than women (Harris, 1991); (b) men and women are often differentiated in terms of social roles (Eagly & Wood, 1999) and trait ascriptions (Williams & Best. 1982); and (c) male-female relations are conditioned by sexual reproduction, a biological constant that creates dependencies and intimacy between the sexes. These three factorspatriarchy, gender differentiation, and sexual reproductiontogether create
111

both hostile and benevolent attitudes toward the other sex (Glick & Fiske, 1996; Glick et al., 2000).' Patriarchy and gender differentiation create and reinforce hostile sexism because dominant groups seek to justify their privileges through ideologies of their superiority (Jost & Banaji, 1994: Sidanius et al., 1994) and through exaggeration of perceived differences with other groups (Tajfel, 1981). In addition, we suggest that sexual reproduction promotes hostile sexism because men often resent women's perceived ability to use sexual attractiveness to gain power over them. At the same time, men's dependence on women (due to sexual reproduction and role differentiation) fosters benevolent sexism, an ideology that counterbalances sexist hostility with a paternalistic protectiveness toward women as a "weaker" but essential group. Men's recognition of their reliance on women to bear and nurture children, to provide domestic labor, and to fulfill sexual and intimacy needs makes women a valuable resource (cf. Guttentag & Secord, 1983). Thus, even though benevolent sexism presumes women's inferiority, it is subjectively positive (from the perspective of the sexist perceiver), characterizing (at least some) women as wonderful, pure creatures whose love is required to make a man whole. Does sexism encompass separable but related hostile and benevolent components that appear as coherent belief systems in a variety of cultures? Factor analyses of the ASI suggest that it does. The ASI was predicted to have a complex structure with separate hostile and benevolent sexism factors, each of which incorporates attitudes related to the structural factors that affect male-female relations: power (patriarchy), gender differentiation, and heterosexuality. Subfactors emerge empirically only for benevolent sexism (see Glick & Fiske, 1996, in press-a, for speculations as to why this is the case). Confirmatory factor analysis (Joreskog & Sorbom, 1993) for samples from 19 nations (Australia, Belgium, Botswana, Brazil, Chile, Colombia, Cuba, England, Germany, Japan, Italy, the Netherlands, Nigeria, Portugal, Spain, South Africa, South Korea, Turkey, and the United States) ranging in size from 250 to 1,600 men and women has replicated the factor structure illustrated in Figure 1, which outperforms a variety of alternative factor models (see Glick et al., 2000). Hostile and benevolent sexism consistently emerge as separate but positively correlated factors. Furthermore, three benevolent sexism subfactors typically appear: protective paternalism (e.g., women ought to be rescued first in emergencies), complementary gender differentiation (e.g., women are purer than men), and heterosexual intimacy (e.g., every man ought to have a woman whom he adores). Hostile sexism items also address power relations (e.g., women seek to gain power by getting control over men), gender differentiation (e.g., women are easily offended), and sexuality (e.g., many women get a kick out of teasing men by seeming sexually available and then refusing male advances), even though the factor structure of the Hostile Sexism scale has proved to be unidimensional in both the United States and elsewhere (Glick & Fiske, 1996; Glick et al., 2000). 112

Figure 1
Factor Structure of the Ambivalent Sexism Inventory

Note. From "The Ambivalent Sexism Inventory: Differentiating Hostile and Benevolent Sexism," by P. Glick and S. T. Fiske, 1996, Journal of Personality and Social Psychology, 70, p. 498. Copyright 1996 by the American Psychological Association.

The factor analyses indicate that hostile and benevolent sexism are meaningful, coherent ideologies not only in the United States but also in a number of other countries. If the ideologies tapped by the ASI were not recognizably similar in other nations, the same factors would not emerge. Although the set of nations in which the ASI has been administered is not random and, in many cases, participants are mostly undergraduate students, the countries sampled are culturally, geographically, and economically diverse. That the structure of the ASI held up well in such diverse nations provides support for the pervasiveness of hostile and benevolent sexism across cultures. Another cross-cultural consistency emerged: Both factor analyses and correlations of raw scores on the ASI confirmed that, within samples, hostile and benevolent sexism were moderately positively correlated. This correlation, however, often dropped (sometimes to nonsignificance) for high-scoring respondentscorrelations were generally smaller among respondents in the most sexist nations, among those who scored above the median on hostile sexism, and for men (who tend to score higher in sexism)suggesting that for sexist respondents hostile and benevolent sexism are only modestly related or independent. More striking was the correlation between national averages on the two scales across nations. When we used nations as the unit of analysis (so that N = 19), hostile and benevolent sexism means correlated .89 (p < .01), for both men and women. Thus, nations in which hostile sexism was strongly endorsed were those in which benevolent sexism was also embraced, indicating that at the systemic level these ideologies are complementary, mutually supportive justifications of patriarchy and conventional gender relations.

Polarized Images of Women

If hostile and benevolent sexism do in fact predict opposing valences in attitudes toward women, then individuals who
'Ambivalence toward the other sex does not flow in only one direction. Glick and Fiske (1999) explored women's ambivalence toward men (as assessed by the Ambivalence Toward Men Inventory).

February 2001 American Psychologist

endorse both ideologies can be viewed as ambivalent about women. Ambivalence is usually conceptualized as either the simultaneous experience of or an oscillation between conflicting feelings or beliefs, such as loving and hating the same individual at the same time (Blueler, 1910, as cited in Katz, 1981). Such conflicting beliefs ought to be negatively correlated, making it surprising that (among individual respondents) hostile and benevolent sexism were typically either moderately positively correlated or independent. Although we believe that ambivalent sexists do often experience conflicting feelings when dealing with individual women, the underlying ideologies that precipitate these conflicts need not be in opposition. The reason for this is evident when one examines how hostile and benevolent sexism relate to stereotypes about women. A major consequence of ambivalence is polarized responses toward the target of ambivalent feelings (for examples involving race and the stigma of having a physical handicap, see Katz, 1981). In 12 nations in which the ASI has been administered, respondents also indicated their spontaneous stereotypes of women (Glick et al., 2000) by generating up to 10 traits that came to mind as characteristics they associated with women. Respondents then evaluated each trait on a - 3 (extremely negative) to 3 (extremely positive) scale. The valence ratings were averaged for each respondent to yield a score of the positivity-negativity of his or her stereotypes of women. Examples of traits that participants generated included warm, sweet, and sensitive (positively valenced) and sly, touchy, and selfish (negatively valenced). Partial correlations, used to control for the typically positive relationship between hostile and benevolent sexism, consistently showed that hostile sexism predicts unfavorable and benevolent sexism predicts favorable stereotypes or images of women. In the United States, Glick, Diebold, Bailey-Werner, and Zhu (1997) conducted two studies that help to solve the puzzle of how hostile and benevolent sexism can be reconciled in the minds of ambivalent sexists without creating cognitive dissonance. In everyday interaction, women are more commonly stereotyped at the level of subtypes, such as "housewives," "career women," "babes," or "lesbians" (e.g., Six & Eckes, 1991), as opposed to a general, overarching "woman" category. Glick et al. reasoned that this subtyping is what allows hostile and benevolent sexism to be complementary, rather than conflicting, belief systems, even though they predict attitudes of opposing valencesat the ideological level, they target different types of women. Hostile sexism is elicited by women who are viewed as directly challenging or surreptitiously stealing men's power (e.g., feminists, career women, or seductresses), whereas benevolent sexism is directed toward women who reinforce conventional gender relations and serve men as wives, mothers, and romantic objects (e.g., homemakers). In Study 1, Glick et al. (1997) asked participants to generate and then evaluate their own categories of women. Overall, men who scored high on both hostile and benevolent sexism had more polarized ratings of the different types of women they generated; that is, they were more February 2001 American Psychologist

likely to evaluate some of these types extremely positively and others extremely negatively. In Study 2, Glick et al. asked participants to evaluate two specific female types, one nontraditional (career women) and the other traditional (homemakers). Men's hostile sexism scores uniquely predicted negative attitudes toward career women, whereas their benevolent sexism scores predicted positive attitudes toward homemakers. These results suggest that hostile and benevolent sexism can be simultaneously endorsed because they are directed at different female subtypes. The complementarity of these ideologies (and their sexist tone) stems from how women are split into "good" and "bad" types; women who fulfill conventional gender roles that serve men are placed on a pedestal and rewarded with benevolent solicitude, whereas women who reject conventional gender roles or attempt to usurp male power are rejected and punished with hostile sexism. The subtyping explanation solves one problem (how the two forms of sexism can be reconciled in the sexist's mind), but it introduces another: If hostile and benevolent sexism are directed at different targets, is this really a form of ambivalence (i.e., conflicted feelings)? We believe that men who endorse both hostile and benevolent beliefs about women are likely to experience ambivalence toward individual women. At the level of ideology, it may be easy for sexists to classify women into distinct groups that are viewed either favorably or unfavorably, but individual women (e.g., a younger sister who becomes a feminist) may often defy easy categorization. Consider, for instance, the well-known pattern that occurs in domestic abuse with a husband reacting with violence when his authority is challenged but later expressing remorse and affection (the subsequent "honeymoon" period)a pattern that suggests considerable sexist ambivalence. Furthermore, we strongly suspect that another oscillating form of ambivalence is likely when men who score high on both hostile and benevolent sexism find that an initial categorization of a woman does not hold. For example, a sexist man might initially place a woman in whom he is romantically interested on a pedestal but abruptly change his views when she rejects him, reclassifying her from "babe" to "bitch." Demonstrating that the ASI predicts such conflicted or oscillating feelings and behavior toward individual women is an important task for future research.

Gender Inequality
The evidence on evaluations of female subtypes is consistent with the notion that benevolent sexism is used to reward women who embrace conventional gender roles and power relations, whereas hostile sexism punishes women who challenge the status quo. This combination of rewards and punishment may be particularly effective in maintaining gender inequality. Psychologists well know that punishment, by itself, is not the most effective means of shaping behavior. Being subjected to hostility alone would be likely to elicit a hostile counterreaction among women, even among those who do not consider themselves to be feminists (see Glick & Fiske, 1999). Combining punish113

ment for nonconformity with rewards for conformity creates a much more effective system for reinforcing "correct" behavior (cf. Jackman, 1994). Does benevolent as well as hostile sexism serve to justify gender inequality in society? Although it is impossible to conduct an experiment that would demonstrate a causal relationship between sexist ideologies and gender inequality at the societal level, cross-cultural comparisons offer a correlational test of this relationship. The United Nations Development Programme (1998) publishes two indices of cross-national gender inequality. The Gender Empowerment Measure assesses women's (relative to men's) participation in a country's economy (percentage of administrators and managers, professional and technical workers who are women, and women's share of earned income) and political system (percentage of seats in parliament held by women). The Gender Development Index (GDI) is a form of the United Nations' Human Development Index (HDI), which focuses on longevity (life expectancy), knowledge (adult literacy rates and years of schooling), and standard of living (purchasing power). The GDI uses the same measures as the HDI, but a country's score is decreased for gender inequality (e.g., women having a lower literacy rate than men). The greater the gender disparity, the lower the GDI relative to the HDI. Glick et al. (2000) examined the correlation of national averages on hostile and benevolent sexism with the two United Nations indices. In addition to the standard warning that causation cannot be inferred, several cautions must be kept in mind when one is interpreting these correlations. First, the high degree of correlation between hostile and benevolent sexism averages across nations (close to .90) makes it impossible to pull apart their relative contributions to predicting inequality (however, that hostile and benevolent sexism are so strongly correlated is central to our point that these ideologies are complementary justifications of inequality). Second, most samples within each country could not be presumed to be representative, and the number of countries in our set is relatively small for computing correlations (both of these facts, however, would be likely to depress correlations with the United Nations measures, rather than privilege our hypothesis). The correlations of men's and women's average hostile and benevolent sexism scores with the two United Nations gender equality measures across 19 nations are reported in Table 1. Not surprisingly, given that men are the dominant group, men's sexism scores tended to be more strongly related to gender inequality. Men's average scores on hostile sexism significantly predicted greater inequality as assessed by the United Nations measures, and both genders' scores on benevolent sexism tended to do so. Even though the benevolent sexism correlations did not quite reach statistical significance, they were consistently negative and close in magnitude to the hostile sexism correlations. Although causality cannot be inferred and the relative roles of hostile and benevolent sexism cannot be disentangled, these results are consistent with the notion that both forms of sexism serve as justifications for gender inequality. 114

Table 1
Correlations Between Ambivalent Sexism Inventory (ASI) Averages and National Indexes of Gender Equality
ASI : Gender Development Index (GDI) Gender Empowerment Measure (GEM)

Men's averages Hostile Sexism Benevolent Sexism

-.47* -.40f
Women's averages

.53* ,43t
.38|
42T

Hostile Sexism Benevolent Sexism

.03 -.32

Note. All correlations with GDI are partial correlations controlling for overall level of human development (Human Development Index) in each country. Sample sizes were 19 countries for GDI correlations and 18 countries for GEM correlations (because the GEM was not available for Nigeria). From "Beyond Prejudice as Simple Antipathy: Hostile and Benevolent Sexism Across Cultures," by P. Glick et al., 2000, Journal of Personality and Social Psychology, 79, p. 772. Copyright 2000 by the American Psychological Association. Reprinted with permission. tp<-10. *p<.05.

Women's Acceptance of Sexist Ideologies

Jost and Banaji's (1994) system-justification perspective emphasizes how subordinate groups tend to accept systemjustifying ideologies of their own inferiority (at least on status-relevant dimensions) that are propagated by dominant groups. Cross-cultural comparisons of men's and women's sexism averages (see Figures 2 and 3) allowed us to test the system-justification hypothesis by seeing whether women expressed greater endorsement of sexist ideologies in countries where men more strongly expressed these views. Using national means as the unit of analysis, Glick et al. (2000) found that men's averages on both hostile and benevolent sexism strongly predicted women's averages on these scales. Men's hostile sexism means correlated .84 and .92, respectively, with women's mean scores on hostile and benevolent sexism. Men's benevolent sexism mean scores correlated .84 and .97, respectively, with women's hostile and benevolent sexism mean scores. Thus, when men in a nation more strongly endorsed sexist ideologies, women followed suit, providing strong correlational evidence of system justification. These correlations, however, are not the whole story. Initial results in the United States (Glick & Fiske, 1996) had shown that women (relative to men) were more likely to reject hostile than benevolent sexism (see also Kilianski & Rudman, 1998). A central part of our argument is that benevolent sexism is a particularly insidious form of prejudice for two reasons: (a) It does not seem like a prejudice to male perpetrators (because it is not experienced as an antipathy), and (b) women may find its sweet allure diffiFebruary 2001 American Psychologist

Figure 2
Hostile Sexism Across Countries

Note From "Beyond Prejudice os Simple Antipalhy: Hostile and Benevolenl Sexism Across Cultuies," by ? Glick el ol., 2000, Journal o! Personality and Social Psychology. 79, p. 770. Copyright 2000 by the American Psychological Association Reprinled with permission.

cult to resist. Benevolent sexism, after all, has its rewards; chivalrous men are willing to sacrifice their own well-being to provide for and to protect women. In the 19 countries studied (Glick & Fiske, 19%; Glick et al., 2000), without exception, men's average hostile sexism scores are considerably (and significantly) higher than women's (see Figure 2). In contrast, in about half of the countries we have studied, women endorse benevolent sexism about as much as men do (see Figure 3). In the 6 countries in which women scored significantly lower on benevolent sexism than men, this difference was typically significantly less than the sex difference in hostile sexism (with two exceptions, Australia and the Netherlands). Most intriguing is the finding that in the 4 nations with the highest sexism scores (Cuba, Nigeria, South Africa, and Botswana), the gender gap in benevolent sexism was reversed, with women endorsing this form of sexism more strongly than men. Thus, even though women in more sexist, relative to less sexist, countries are more likely to accept both hostile and benevolent sexism, women everywhere (when compared with men) are more resistant to the former than to the latter. In fact, the gender gap between men's and women's hostile sexism scores increases the more sexist the country is. Glick et al. (2000) found that, across nations, the gender gap in hostile sexism scores was positively correlated with men's average hostile sexism (r .61, p < .01) and benevolent sexism (r = .41, p < .10) scores, suggesting that as men's sexism increases, women's acceptance of hostile sexism is increasingly tempered by resistance to men's hostility. In contrast, the gender gap in benevolent sexism scores was negatively correlated with men's hostile {r = -.75, p < .01) and benevolent (r = -.65, p < .01) sexism means; the more sexist the men, the more likely women were to embrace benevolent sexism, even to the point of endorsing benevolenl sexism more strongly than men. February 2001 American Psychologist

Virginia Woolf (1929/1981) argued that true gender equality will happen only when "womanhood has ceased to be a protected occupation" (p. 40), but women's relatively greater acceptance of benevolent sexism suggests that some women may resist this change. Perhaps many women want what benevolent sexism brings (e.g., protection), without the corresponding costs of sexist hostility. There is a fine line, however, between acknowledging women's responsibility in maintaining benevolent sexism and blaming the victim. Another explanation for women's acceptance of benevolent sexism is that it is a form of self-protection in response to men's sexism. Smuts (1996) argued that pairbonding among humans is, in part, an evolved female response to the threat of sexual violence (because a pairbonded male mate offers protection from other men). In a similar manner, endorsing benevolent sexism may be a way in which women cope when many men in a culture tend to be hostile sexists (cf. Jackman, 1994). The irony is that women are forced to seek protection from members of the very group that threatens them, and the greater the threat, the stronger the incentive to accept benevolent sexism's protective ideology. This explains the tendency for women in the most sexist societies to endorse benevolent sexism more strongly than men. Furthermore, the countries in which women (as compared with men) rejected benevolent sexism as strongly as hostile sexism were ones in which men had low hostile sexism scores. As sexist hostility declines, women may feel able to reject benevolent sexism without fear of a hostile backlash.

Implications for Theories of Prejudice: Paternalistic Versus Envious Prejudices

The pervasiveness of benevolent sexism across cultures and its relation to both hostile sexism and women's sub-

Figure 3
Benevolent Sexism Across Countries

Nofe. From "Beyond Prejudice os Simple Amipolhy: Hoslile end Benevolent Sexism Across Cultures." by P. Glick et al.. 2000, journal of Personality and Social Psychology, 79, p. 770. Copyright 2000 by the American Psychological Association. Reprinted with permission.

115

ordination suggest that it is time to rethink the equation of prejudice with antipathy and to acknowledge that there is more than one type of prejudice. Prejudice can manifest itself not only in unalloyed hostility but also in sweet, yet patronizing, guises that may be insidiously effective at maintaining social inequalities. Although benevolent paternalism is most evident in sexist ideology, it is not unique to this form of prejudice. We have already referred to historical examples of paternalistic prejudicesbenevolent justifications of slavery and colonialismthat represent group relations in which (as with gender) the dominant group was dependent on groups they subordinated (for their labor). There are contemporary examples as well. For example, some liberal Whites may have paternalistic attitudes toward African Americans, characterized by pity and an implicit belief that African Americans are incapable of helping themselves (Katz & Hass, 1988). Recent research that we have conducted with our colleagues suggests that paternalistic stereotypes are directed at a number of groups that are perceived to be low in status and capability but not threatening, such as people who are blind, people with handicaps, older individuals, and (of course) housewives (Fiske, 1998; Fiske, Glick, Cuddy, & Xu, 1999; Fiske, Xu, Cuddy, & Glick, 1999; Glick & Fiske, in press-b). These paternalistic prejudices can themselves be considered ambivalent in that they combine subjectively benevolent stereotypes (of the group's warmth) and feelings (sympathy, affection, and pity) with beliefs in the group's incompetence, need to be helped, and unsuitedness for high-status rolesthe ambivalence of liking coupled with disrespect. In contrast, the resentful tone evident in hostile sexism may be similar to prejudice directed toward socioeconomically successful minorities who are perceived as a competitive threat. Just as sexist men resent career women who succeed in traditionally male-dominated areas, minorities who are perceived to be successful (e.g., Jews or Asian Americans) may be envied and viewed as overly ambitious rivals. This can constitute another form of ambivalencea grudging respect coupled with dislike and fearthat we have called "envious prejudice" (Glick & Fiske, in pressfa). Hostile sexists' views of career women, feminists, and "temptresses" seem to fit this category, as evidenced by hostile sexism items that characterize both feminism and female sexuality as usurping men's power and by hostile sexists' tendency to attribute competence to career women but also to say they fear, envy, resent, and feel competitive toward this type of woman (Glick et al., 1997). Paternalistic and envious prejudices may be quite distinct in their causes and their consequences. The one-size-fits-all conception of prejudice, rooted in the idea that prejudice is pure hatred and contempt, has obscured these differences (Young-Breuhl, 1996). One commonality between envious and paternalistic stereotypes, however, is that seemingly favorable traits attributed to a group may only add fuel to the fire of prejudice. Paternalistic stereotypes assign the favorable traits of warmth to low-status groups, but this represents an amiable way of helping to ensure these groups' subordination. For example, the positive value placed on warm, communal traits also lends them a more prescriptive tone that sets up 116

powerful norms for women's behavior. Although progress has been made in combating prescriptions that women ought not to be too competent, prescriptions for women's niceness remain strong (Rudman, 1998; Spence & Buckner, 2000). Praising women's nurturing traits, for some, is part of expressing the belief that women are especially suited to a domestic role (a much more acceptable political position than trumpeting women's lack of suitedness for high-status roles). Furthermore, communal traits are associated with deferent and subordinate behavior; enacting these traits in interactions with men reinforces women's subordinate status (Ridgeway, 1992). Likewise, even though envied groups (e.g., Jews or Asian Americans) may be attributed the normally positive traits of competence (e.g., ambitious or smart), these attributions often become part of the justification for discriminating against them; they are perceived to be too clever and manipulative. The statement "Jews are extraordinarily shrewd" is as likely as not to be diagnostic of an extreme, envious form of anti-Semitism (e.g., the belief in a dangerous international Jewish conspiracy; see Glick, in press). In a similar manner, perceiving certain types of women as overachievers may only add to sexist men's hostility. This perspective on prejudice suggests a closer link between affect and cognition than recent theorists have posited. For example, Eagly and Mladinic (1989) noted a lack of correlation between traditional views about women and affect toward women as a group, but as our research shows, this is not because affect and stereotypes are unrelated. Rather, sexists direct positive affect toward subtypes of women who embrace conventional roles and negative affect toward those who do not. Our approach suggests more generally that group stereotypes are strongly related to the affect experienced toward groups. Rather than predicting undifferentiated negative affect toward out-groups, however, our theoretical perspective predicts more specific and often ambivalent emotional reactions (Cuddy, 2000). The negative emotions directed at members of high-status groups who are viewed as a competitive threat are likely to revolve around feelings of resentment, fear, and envy but also may be accompanied by feelings of grudging admiration. Low-status, cooperative groups, the targets of paternalistic prejudice, may predominantly evoke positive feelings, but these feelings would be confined to patronizing affection and pity rather than respect. In short, it is possible to understand the relationship between affect and stereotypes only by abandoning the notion that prejudice is a generalized antipathy and examining more specific emotional reactions toward groups (cf. Smith, 1993). Our approach suggests that these relationships are systematic and predictable, even though the emotions may be more differentiated and complex than the antipathy model of prejudice implies.

Conclusion
Although sexist antipathy is the most obvious form of prejudice against women, our evidence suggests that sexist benevolence may also play a significant role in justifying gender inequality. Together, these ideologies represent a system of rewards and punishments that provide incentive for women to remain in conventional gender roles. BenevFebruary 2001 American Psychologist

olent sexism, though a kinder and gentler form of prejudice, is pernicious in that it is more likely to be accepted by women, as well as men, especially in cultures in which women experience a high degree of threat from men. Both hostile and benevolent sexism appear to be cross-culturally prevalent, supporting the argument that these ideologies arise from structural aspects of male-female relations that are common across human groups. More generally, we suggest parallels between the two forms of sexism and prejudice against other groups. Hostile sexism is similar to other forms of envious prejudice, directed at groups who are seen as threats to the in-group's status and power, whereas benevolent sexism corresponds to other paternalistic prejudices, directed at groups that are lower in status and viewed as cooperative or nonthreatening. These notions challenge the assumption that prejudice is an antipathy and suggest that emotions toward out-groups, though complex and ambivalent, can be predicted and understood.
REFERENCES Allport, G. W. (1954). The nature of prejudice. Reading, MA: AddisonWesley. Brown, R. (1995). Prejudice: Its social psychology. Oxford, England: Blackwell. Carpenter, S. (2000). Implicit gender attitudes: Group membership, cultural construal, valence-consistency, and stability. Unpublished doctoral dissertation, Yale University. Cuddy, A. C. (2000). [Prejudice and emotion]. Unpublished raw data. Eagly, A. H., & Mladinic, A. (1989). Gender stereotypes and attitudes toward women and men. Personality and Social Psychology Bulletin, 15, 543-558. Eagly, A. H., & Mladinic, A. (1993). Are people prejudiced against women? Some answers from research on attitudes, gender stereotypes and judgments of competence. In W. Stroebe & M. Hewstone (Eds.). European review of social psychology (Vol. 5, pp. 1-35). New York: Wiley. Eagly, A. H., & Wood, W. (1999). The origins of sex differences in human behavior: Evolved dispositions versus social roles. American Psychologist, 54, 408-423. Faludi, S. (1992). Backlash: The undeclared war against American women. New York: Doubleday. Fiske, S. T. (1998). Prejudice, stereotyping, and discrimination. In D. T. Gilbert, S. T. Fiske, & G. Lindzey (Eds.), The handbook of social psychology (4th ed., pp. 357-411). New York: McGraw-Hill. Fiske, S. T., Glick, P., Cuddy, A. C , & Xu, J. (1999, July). Ambivalent content of stereotypes, predicted by social structure: Status and competition predict competence and warmth. Paper presented at the 22nd General Meeting of the European Association of Experimental Social Psychology, Oxford, England. Fiske, S. T., Xu, J., Cuddy, A. C , & Glick, P. (1999). Respect versus liking: Status and interdependence underlie ambivalent stereotypes. Journal of Social Issues, 55, 473-489. Glick, P., (in press). Sacrificial lambs dressed in wolves' clothing: Envious prejudice, ideology, and the scapegoating of the Jews. In L. S. Newman & R. Erber (Eds.), What social psychology can tell us about the Holocaust. New York: Oxford University Press. Glick, P., Diebold, J., Bailey-Werner, B., & Zhu, L. (1997). The two faces of Adam: Ambivalent sexism and polarized attitudes toward women. Personality and Social Psychology Bulletin, 23, 13231334. Glick, P., & Fiske, S. T. (1996). The Ambivalent Sexism Inventory: Differentiating hostile and benevolent sexism. Journal of Personality and Social Psychology, 70, 491-512. Glick, P., & Fiske, S. T. (1999). The Ambivalence Toward Men Inventory: Differentiating hostile and benevolent beliefs about men. Psychology of Women Quarterly, 23, 519-536. Glick, P., & Fiske, S. T. (in press-a). Ambivalent sexism. In M. P. Zanna

(Ed.), Advances in experimental social psychology. San Diego, CA: Academic Press. Glick, P., & Fiske, S. T. (in press-b). Ambivalent stereotypes as legitimizing ideologies: Differentiating paternalistic and resentful prejudice. In J. T. Jost & B. Major (Eds.), The psychology of legitimacy: Emerging perspectives on ideology, justice, and intergroup relations. New York: Cambridge University Press. Glick, P., Fiske, S. T., Mladinic, A., Saiz, J., Abrams, D., Masser, B., Adetoun, B.,Osagie, J., Akande, A., Alao, A., Brunner, A., Willemsen, T. M., Chipeta, K., Dardenne, B., Dijksterhuis, A., Wigboldus, D., Eckes, T., Six-Materna, I., Exp6sito, F., Moya, M., Foddy, M., Kim, H.-J., Lameiras, M., Sotelo, M. J., Mucchi-Faina, A., Romani, M., Sakalli, N., Udegbe, B., Yamamoto, M., Ui, M., Ferreira, M. C , & Lopez, W. L. (2000). Beyond prejudice as simple antipathy: Hostile and benevolent sexism across cultures. Journal of Personality and Social Psychology, 79, 763-775. Guttentag, M., & Secord, P. (1983). Too many womeni Beverly Hills, CA: Sage. Harris, M. (1991). Cultural anthropology (3rd ed.). New York: Harper Collins. Hochschild, A. (1998). King Leopold's ghost: A story of greed, terror, and heroism in colonial Africa. Boston: Houghton Mifflin. Jackman, M. R. (1994). The velvet glove: Paternalism and conflict in gender, class, and race relations. Berkeley: University of California Press. Joreskog, K., & Sorbom, D. (1993). USREL 8: Structural equation modeling with the SIMPUS command language. Hillsdale, NJ: Erlbaum. Jost, J. T., & Banaji, M. R. (1994). The role of stereotyping in systemjustification and the production of false-consciousness. British Journal of Social Psychology, 33, 1-27. Katz, I. (1981). Stigma: A social psychological analysis. Hillsdale, NJ: . Erlbaum. Katz, I., & Hass, R. G. (1988). Racial ambivalence and value conflict: Correlational and priming studies of dual cognitive structures. Journal of Personality and Social Psychology, 55, 893-905. Kilianski, S., & Rudman, L. A. (1998). Wanting it both ways: Do women approve of benevolent sexism? Sex Roles, 39, 333-352. Kipling, R. (1899). The White man's burden: A poem. McClure's Magazine, 12, 290. Moya, M., Exposito, F., & Casado, P. (1999, July). Women's reactions to hostile and benevolent sexist situations. Paper presented at the 22nd General Meeting of the European Association of Experimental Social Psychology, Oxford, England. Pomeroy, S. B. (1975). Goddesses, whores, wives and slaves: Women in classical antiquity. New York: Schocken. Pratto, F. (1996). Sexual politics: The gender gap in the bedroom, the cupboard, and the cabinet. In D. M. Buss & N. M. Malamuth (Eds.), Sex, power, and conflict (pp. 179-230). New York: Oxford University Press. Ridgeway, C. (1992). Gender, interaction, and inequality. New York: Springer-Verlag. Rudman, L. A. (1998). Self-promotion as a risk factor for women: The costs and benefits of counterstereotypical impression management. Journal of Personality and Social Psychology, 74, 629-645. Rudman, L. A., & Heppen, J. (2000). Some day my prince will come: Implicit romantic fantasies and women's avoidance of power. Unpublished manuscript, Rutgers, The State University of New Jersey. Salzman, P. C. (1999). Is inequality universal? Current Anthropology, 40, 31-61. Sidanius, J., Pratto, F., & Bobo, L. (1994). Social dominance orientation and the political psychology of gender: A case of invariance? Journal of Personality and Social Psychology, 67, 998-1011. Six, I., & Eckes, T. (1991). A closer look at the complex structure of gender stereotypes. Sex Roles, 24, 57-71. Smith, E. (1993). Social identity and social emotions: Toward new conceptualizations of prejudice. In D. M. Mackie & D. L. Hamilton (Eds.). Affect, cognition, and stereotyping: Interactive processes in group perception (pp. 297-315). San Diego, CA: Academic Press. Smuts, B. (1996). Male aggression against women: An evolutionary perspective. In D. M. Buss & N. M. Malamuth (Eds.), Sex, power, and conflict (pp. 231-268). New York: Oxford University Press. Spence, J. T., & Buckner, C. E. (2000). Instrumental and expressive traits, trait stereotypes, and sexist attitudes. Psychology of Women Quarterly, 24, 44-62.

February 2001 American Psychologist

117

Tajfel, H. (1981). Social identity and intergroup relations. London: Cambridge University Press. Tavris, C.,& Wade, C. (1984). The longest war (2nd ed.). San Diego, CA: Harcourt Brace Jovanovich. United Nations Development Programme. (1998). Human development report 1998. New York: Oxford University Press.

Williams, J. E., & Best, D. L. (1982). Measuring sex stereotypes Beverly Hills, CA: Sage. Woolf, V. (1981). A room of one's own. New York: Harcourt Brace Jovanovich. (Original work published 1929) Young-Breuhl, E. (1996). The anatomy of prejudices. Cambridge, MA: Harvard University Press.

Appendix The 22-ltem Ambivalent Sexism Inventory

Relationships Between Men and Women
Below is a series of statements concerning men and women and their relationships in contemporary society. Please indicate the degree to which you agree or disagree with each statement using the scale below: 1 2 3 4 5 disagree disagree agree agree agree somewhat slightly slightly somewhat strongly 1. No matter how accomplished he is, a man is not truly complete as a person unless he has the love of a woman. 2. Many women are actually seeking special favors, such as hiring policies that favor them over men, under the guise of asking for "equality." 3. In a disaster, women ought to be rescued before men. 4. Most women interpret innocent remarks or acts as being sexist. 5. Women are too easily offended. 6. People are not truly happy in life without being romantically involved with a member of the other sex. 7. Feminists are seeking for women to have more power than men. 8. Many women have a quality of purity that few men possess. 9. Women should be cherished and protected by men. 10. Most women fail to appreciate fully all that men do for them. 11. Women seek to gain power by getting control over men. 12. Every man ought to have a woman whom he adores. 13. Men are incomplete without women. 14. Women exaggerate problems they have at work. 15. Once a woman gets a man to commit to her, she usually tries to put him on a tight leash. 16. When women lose to men in a fair competition, they typically complain about being discriminated against. 17. A good woman should be set on a pedestal by her man. 18. Many women get a kick out of teasing men by seeming sexually available and then refusing male advances. 19. Women, compared to men, tend to have a superior moral sensibility. 20. Men should be willing to sacrifice their own well-being in order to providefinanciallyfor the women in their lives. 21. Feminists are making unreasonable demands of men. 22. Women, as compared to men, tend to have a more refined sense of culture and good taste.

H H

B(G)

Scoring: Total ASI score = average of all items. Hostile Sexism = average of Items 2, 4, 5, 7, 10, 11, 14, 15, 16, 18, 21. Benevolent Sexism = average of Items 1, 3, 6, 8, 9, 12, 13, 17, 19, 20, 22. Note. Items 3, 6, 7, 13, 18, and 21 are reverse-worded in the original version of the ASI (Glick & Fiske, 1996), though not in the version that appears here because reverse-worded items did not perform well in translation to other languages (other than lower factor loadings for reversed items, similar results have been obtained in the United States and elsewhere when both reversed and nonreversed wordings have been administered; see Glick et al., 2000, footnote 2). B = benevolent sexism; I = heterosexual intimacy; H = hostile sexism; P = protective paternalism; G = gender differentiation. Copyright 1995 by Peter Glick and Susan T. Fiske. Use of this scale for nonacademic purposes (i.e., activities other than nonprofit scientific research and classroom demonstrations) requires permission of one of the authors. 118 February 2001 American Psychologist