Psychology of Wonwn Quarterly, 23 (1999), 50,L517 Printed in the United States of Americd
CONTEMPORARY SEXISM The Relationships Among Hostility, Benevolence, and Neosexism
Barbara Masser and Dominic Abrams Universiiy of Kent
Previous research has examined the relationship between the Modern Sexism k a l e (Swim, Aikin, Hall & Hunter, 1995) and the Ambivalent Sexism Inventory (ASI; Click & Fiske, 1996) or the relationship between the Modern Sexism Scale and the Neosexism Scale (Campbell, Schellenberg, & Senn, 1997; Tougas, Brown, Beaton, & Joly, 1995). The present study examined the relationship between the AS1 and the Neosexism Scale. Across three samples (N= 907), neosexism was found to be associated primarily with the hostile rather than the benevolent component of ambivalent sexism. Hostile sexism, benevolent sexism, and neosexism were all significantly associated with attitudes toward lesbians’ and gay men’s rights, and both hostile sexism and neosexism were significantly associated with attitudes toward women’s rights. Neosexism and hostile sexism were negatively associated with a humanitarian-egalitarian orientation, whereas benevolent sexism was positively associated with a Protestant-ethic orientation.lt is concluded that, although both the Neosexism Scale and AS1 measure contemporary sexism, only the Benevolent Sexism subscale of the AS1 taps the subjectively positive side of contemporary sexism.
Social psychological accounts of prejudce have traditionally focused on blatant forms of discrimination and bias against members of social groups or categories (e.g., Adorno, Frenkel-Brunswik, Levinson, & Sanford, 1950; Tajfel, 1982). However, j u s t as social policies and intergroup relations have changed, so too has the manner in which prejudice i s manifested. Recent accounts of prejudice, particularly racism, This research was supported by an Economic and Social Research Council grant (no: R00429534216) to the first author. Address correspondence and reprint requests to: Barbara Maser, School of Psychology, University of Queensland, St. Lucia, QLD 4072, Australia. E-mail:
[email protected] Published hy Cambridge University Press 0361-6843/99 $9.50
503
504
M A S E R AND ABRAM5
have conceptualized prejudice as an underlying process that is often manifested primarily in subtle or symbolic forms-for example, as in differential helping behavior to menibers of different ethnic groups (Crosby, Bromley, & Saxe, 1980). Contemporary fbrins of‘ racism have been researched for over 20 years (e.g., Caertner & Dovidio, 1986; Katz, Wackenhut, & Hass, 1986; McConahay, 1986; Monteith, 1993; Sears & Kinder, 1971). However, the direct conceptual “crossover” into research on other forms of prejudice, such as sexism, is only recent, despite the inany parallels traditionally drawn between these forms of prejudice (see Hacker, 1951; Myrdal, 1944; Reid, 1988; cf. Ferree & Hass, 1985). One common element in contemporary forms of sexism and racism may be the set ofsocial norms governing the open expression of prejudice. Tougas, Brown, Beaton, and Joly (1995) note that, “as with racism, there have been normative arid legislative changes in many industrialized societies, which make it less acceptable to express sexist ideas openly” (p. 842). Two diffcring operationalizations of contemporary sexism have been proposed. First, two measures have been developed to assess political attitudes related to gender discrimination: the Neosexism Scale (Tougas et al., 1995) and the Modern Sexism Scale (Swim, Aikin, Hall, & Hunter, 1995). Both are based on the theory arid research on symbolic/modern racism (McConahay, 1986; Sears & Kinder, 1985). Second, Glick and Fiske’s (1996) measure of‘ ambivalent sexisin is based broadly on the work of Katz et al. (1986) on ambivalent racism, and focuses on interpersonal relationships. Research into these contemporary forms of sexism has niainly concentrated on the development of scales and inventories to measure individiial differences. NEOSEXISM
Tougas et al. (1995) propose that contemporary sexism, or “neosexisrn,”can be defined as a “nianifestation of a conflict between egalitarian values and residual negative feelings tolvard women” (p. 843). In contrast, tradtional sexism is based simply on negative affect. Tougas et al. (1995) developed a scale to measure neosexisrn using a method and items analogous to those used for measuring symbolic/modern racism in McConahay’s (1986) Modern Racism Scale. Tougas et al. (1995) found with Canadian men that, although traditional sexism and neosexisin were related (because of their joint basis in negative affect), only neosexisrn predicted attitudes toward the symbolic issue of affirmative action.‘ Specifically,Tougds et al. (1995) found that neosexism was directly related to negative reactions to affirmative action programs. In a similar vein, Swim et al. (1995) developed the Modem Sexism Scale based on the premises of symbolicimodem racism. A preliminary assessment showed motlern sexist attitudes to be correlated with (among other things) perceptions of occupational gender segregation. Ambivalent Sexism
In contrast to Tougas et al.’s (1995) view of contemporary sexism as consisting of mainly negative affect, Glick and Fiske (1996) propose that it is characterized by ambivalence-the coexistence of positive and negative affect. Traditional theories
Contemporary Sexism
505
of sexism have assessed only hostility toward women, neglecting the subjectively positive, or “benevolent” side of sexism. Benevolent sexism is defined as “a set of interrelated attitudes toward women that are sexist in terms of viewing women stereotypically and in restricted roles, but that are subjectively positive in feeling tone” (Glick & Fiske, 1996, p. 491). The hypothesized coexistence of both positive aiid negative affect is similar to that proposed in the theory of ambivalent racism (Katz et al., 1986). Katz et al. (1986) suggested that, in North America, both positive and negative affect among Whites toward Blacks could exist because of a tension between egalitarianism and individualism. This results in the Whites viewing Blacks as deserving help (egalitarianism), yet as responsible for their own circumstances (individualism). Glick and Fiske (1996) propose that the positive and negative affect do not arise from opposing value systems (for example, “benevolent” sexism does not stem from egalitarian beliefs) but from evaluative ambivalence. Specifically, Glick arid Fiske (1996) characterize hostile and benevolent beliefs as being ambivalent because, “even if the beliefs about women that generate hostile and benevolent sexism are positively related, they have opposing euuluutive implications, fulfilling the literal meaning of ambivalence” (p. 494). Glick and Fiske (1996) developed a 22-item measure of individual differences in ambivalent sexism-the Ambivalent Sexism Inventory (ASI). They conducted six studies in North America to develop and validate the AS1 using student and nonstudent populations. Factor analyses revealed two main factors (hostile and benevolent), providing the basis for two subscales, both of which had acceptable internal reliability. The Hostile and Benevolent Sexism Scales were found to be strongly positively related in the student samples, and in all studies men scored higher than women. This difference was found to be consistently larger for the Hostile Sexism Scale. The AS1 correlated positively with other traditional measures of sexism such as the Attitudes Toward Women Scale (Spence & Helmreich, 1972), the Modern Sexism Scale (Swim et al., 1995), and the Rape Myth Acceptance Scale (Burt, 1980). For all of these measures, Glick and Fiske (1996) predicted and found that “the relationship between the AS1 and other sexism measures . . . is wholly attributable to the HS (hostile sexism) subscale” (p. 503). Research into contemporary forms of sexism is still at an early stage and is limited to a few published studies (e.g., Beaton, Tougas, & Joly, 1996; Campbell, Schellenberg, I%Senn, 1997; Glick & Fiske, 1996, 1997; Swim & Cohen, 1997; Swim et al., 1995; Tougas et al., 199.5; Wiener, Hurt, Russell, Mannen, & Gasper, 1997) prirriarily conducted in North America. This research has established the relationship of the Modern Sexism Scale to the AS1 (Glick & Fiske, 1996) and of the Modern Sexism Scale to the Neosexism Scale (Campbell et al., 1997). Thc relationship between the Neosexism Scale and the AS1 has not been investigated previously. Campbell et al. (1997) concluded that whereas the Neosexism Scale items are derived from all three tenets of McConahay’s Modem Racism Scale‘ (McConahay, 1986) the Modem Sexism Scale items appear to be derived primarily from the first tenet of‘ McConahay’s scale. Thus, it would be hasty to assume that the Neosexism Scale and the Modern Sexism Scale share the same relationship to the ASI. The first aim of the present study is to extend the research into the interrelationships between different measures of contemporary sexism by examining the relationship of the AS1 (and its subscales) to the Neosexism Scale.
506
MASER A N D ABRAMS
It was hypothesized that, whereas both subscales of the AS1 would be positively related to the Neosexism Scale, the basis for this relationship would primarily be negative affect. Thus, the relationship between benevolence and neosexism should be attributable to the association between hostile and benevolent sexism; hostile but not benevolent sexism should be uniquely associated with neosexism. The second aim, in partial replication of Campbell et al. (1997), is to investigate the relationship of the AS1 and the Neosexism Scale to gender-related attitudes (e.g., women’s rights and lesbian and gay men’s rights). Campbell et al. (1997) found that lower levels of modern sexism and neosexism were associated with more positive attitudes toward feminism, women’s rights, and lesbians and gay men. It was hypothesized that hostility, benevolence, and neosexism would be negatively associated with statements supporting women’s rights and lesbian and gay men’s rig11ts . Swim et al. (1995) found that modern sexism was positively associated with an individualistic value orientation as measured by the Protestant Ethic scale (Katz & Hass, 1988). They also discovered a negative relationship between modern sexism and the Humanitarian-Egalitarian scale (Katz & Hass, 1988). Negative humanitarian-egalitarian and positive Protestant ethic value orientations had previously been found to underlie many contemporary racist attitudes (e.g., in Katz & Hass [1988] formulation of ambivalent racism). In contrast, using the Neosexism Scale, Campbell et al, (1997) found that sexism was associated only with an individualistic value orientation and this relationship was only significant for women. Thus, it appears that the Modem Sexism and Neosexism Scales tap overlapping, but not identical, constructs. A third aim, again in partial replication of Campbell et al. (1997), is to investigate the relationship of the AS1 and the Neosexism Scale to humanitarian-egalitarian and Protestant ethic value orientations. To date, no published study has investigated the relationship of the AS1 or its subscales to humanitarian-egalitarian or Protestant ethic value orientations. It was hypothesized that hostility, benevolence, and neosexism should be positively associated with a Protestant ethic value orientation. click and Fiske (1996) propose that there should be no relationship between benevolent sexism and egalitarian beliefs. However, previous research suggests that hostile sexism and neosexism, but not benevolent sexism, should be negatively related to a humanitarian+galitarian orientation. It is interesting to examine whether the relationship among constructs is relatively stable across different populations. The present study employs two samples of students and a general population sample in order to investigate this issue. Finally, it should be noted that this research represents one of the first attempts to evaluate contemporary sexism measures outside of a North American context. With regard to the Neosexism Scale, this raises some interesting theoretical issues. One of the key concepts underlying the theory of neosexism is the importance of the “symbolic issue” (theorized to be affirmative action). Despite recent attempts by some British organizations to introduce something akin to affirmative action (e.g.,the Labour Party’s “women only” selection lists), Britain does not have affirmative action programs and generally has a lower profile gender-discrimination framework than is common in North America. Because of these cultural and legal differences it is of interest to consider whether the Neosexism Scale has relevance and meaning as a measure of contemporary sexism in a British context.
507
Contemporary Sexism
METHOD
Respondents In order to evaluate the relationship of the AS1 and its subscales (Hostile and Benevolent Sexism) to the Neosexism Scale, samples were collected from three populations.’ Sample 1 (Schools)consisted of 148 male and 147 female 17-year-old students who were studying for university-entryqualifications.Sample 2 (University) consisted of 190 male and 301 female students enrolled in undergraduate university courses who participated in the study at the start of the academic year to gain course credit. The mean age of this sample was 23 years. Sample 3 (Full-Time Employed; FTE) consisted of 60 males and 61 females who were recruited by a number of volunteers in organizations around Britain. The mean age of Sample 3 was 35 years and the majority (80%)had received university entry or universitylevel education. Across the three samples, participants were predominantly Caucasian, were of British origin, and all had English as their first language or were proficient (to university-entry level) in English. A subsample of the University participants (n = 312) completed the women’s rightsflesbian and gay men’s rights measure. A separate subsample of the University participants ( n = 110) completed the humanitarian+egalitarian/protestant ethic individualistic orientation measure. Procedure
Sample 1 (Schools) participants were recruited at University opening day. Each participant was approached by one of a number of Caucasian male or female experimenters and asked if he or she would like to take part in a survey of modern attitudes. The experimenters were selected to be similar in age and status to the participants. Once consent had been obtained, participants were taken to a quiet area and asked to independently and anonymously complete a booklet containing a number of measures, including the AS1 and Neosexism Scale. Once participants had completed the booklet, they were requested to post it in a box provided. After this they were requested to report to an area where full debriefing was given by a number of experimenters and any questions answered. Of those approached, 55 people (16%)declined to take part in the survey. Of those who refused the primary reason was lack of time, and refusal always occurred prior to any of the measures being seen. All participants in Sample 2 (University) volunteered to complete a booklet containing the AS1 and Neosexism Scale (embedded in other measures4) for course credit in a mass “pretest” at the beginning of the academic year. Booklets were distributed by a number of Caucasian male and female experimenters who were similar in age and status to the participants. The experimenters emphasized the anonymity of responses and requested that the participants complete the booklets independently. All those requested to take part in the study agreed. Booklets were returned by participants via an internal mail system. After all the booklets had been returned, a debriefing was given to all the participants. Participants in Sample 3 (FTE) were obtained via “snowball” sampling. Specifically, a number of people employed in a diversity of public- and private-sector organizations in the United Kingdom were approached and asked if they would be willing to assist in the administration of a survey of modern attitudes. These assistants were then sent
508
M A S E R A N D ABRAM.5
packs of questionnaire booklets containing the AS1 and Neosexism Scale (embedded in other measures), debriefing sheets. arid stamped, addressed envelopes to the University. The assistants, who were predominantly Caucasian and were similar in age and status to the subsequent participants, were asked to recivit as many participants as possible for the survey. Each participant was initially given a booklet and a stamped, addressed envelope. Between 2 and 3 weeks later, each participant was given a debriefing sheet by the experimental assistant explaining the purpose of the study. In addition, names, addresses and telephone numbers of the University administrators were provided to enable participants to contact us. A number of participants (approximately 5%)did, primarily to obtain feedback information. Two hundred questionnaire packs were sent out to experimental assistants. A total of 121 were returned completed (60.5%) with an additional 32 (16%) returned by the experimental assistants. Of those who received questionnaire packs, 76.5% completed and returned them. Measures
All measures were presented in English. The order of presentation of the measures was randomized within each sample. The Arnbivulent Sexism Inventory (ASI; Glick & Fiske, 1996): The AS1 is a 22item inventory consisting of two 11-item subscales (Hostile and Benevolent Sexism). The AS1 is comprised mainly of statements concerning relationships between men and women. For this study, participants were requested to indicate their level of agreementidisagreement with the statements using a 7-point ~ c a l eHigher .~ scores indicated more sexist attitudes. The Neosexism Scale (Tougas et al., 1995):An 11-item version6 of the Neosexisin Scale was used in this study. The scale consists of statements about women and work. Participants indicated their level of agreemen~~~isagreerrierit with each statement on a 7-point scale. As with the ASI, higher scores indicated more sexist attitudes. Attitudes toward Women’s Rights and Attitudes toward Leshiun and Gay Men’,s Rights measures (Pratto, Sidanius, Stallworth, & Malle, 1994): These measures consisted ofa number of statements concerned with women’s rights (e.g.,“Equal pay for women”) and Lesbian and Gay Men attitudes (e.g., “Gay or leshian marriage”). Participants were asked to indicate how positive or negative they felt toward the objects, events, or statements, using a 1= verg negative to 7 = very po.sitioe scale. Humanitarian-Egalitarian Scale (Katz & Hass, 1988):The Humanitarian-Ejialitarian scale consists of 10 items designed to measure egalitarian value orientations. Participants in the current study respontled using a 7-point scale, with higher scores indicating a more humanitarian-egalitarian orientation. Protesttint Ethic Scale (Katz & H a s , 1988):The Protestant Ethic scale (Katz & Hass, 1988) consists of 11 items that are designed to measure an individualistic value orientation. Participants in the current study responded on a 7-point scale, with higher scores indicating a inore individualistic orientation. RESULTS Initial analyses for all samples indxated that missing data were distributed randomly and that the assumption of multivariate normality was met.
509
Contemporary Sexism
Internal Reliability of contemporary Sexism Scales
Hostile and Benevolent Sexism Principal components analysis of the AS1 confirmed the two-factor structure of Hostile and Benevolent Sexism' suggested by Glick and Fiske (1996). The internal reliability coefficientsfor each subscale were examined within each sample. Despite the minor differences in factor structure across the three samples, the internal reliabilities of the complete Hostile and Benevolent Sexism subscales of the AS1 were found to be acceptably high (Nunnally, 1978), and in line with those found by Glick and Fiske (1996). Across the three samples, Cronbach's alphas ranged between .7ij and .88 for Hostile Sexism and between .61 and .83 for Benevolent Sexism. Neosexism Principal components analysis ofthe Neosexism Scale confirmed the unidimensional structure found by Tougas et al. (1995) and Campbell et al. (1997). The internal reliability of the scale was also found to be acceptable (Nunnally, 1978) with Cronbach's alphas ranging from .62 to .85. Sexism and Gender
As noted by Swim et al. (1995) and others (e.g., Carnpbell et a]., 1997; DelBoca, Ashmore, & McManus, 1986; Glick & Fiske, 1996) the construct validity of sexisni scales may be reflected by the extent to which they are responded to differently by men and women. Tests of gender differences on all the scales across the three saniples using multivariate analysis of variance (MANOVA) revealed significant main effects for sample and gender for all scales. Across all samples men scored significantly higher than women on all the scales [Hostile Sexism: M,,,,,, = 3.89, SD = 1.01, M,,,,,,,, = 3.16, SD = 0.87; F (1,814)= 94.79, p < ,001; Benevolent Sexism: M,,,,,,,= 3.59, SD = 0.88, M,,,,,,,, = 3.29, SD = 0.88; F (1,814) = 19.25, p < ,001; Neosexism: M,,,,,,,= 3.30, SD = 0.89, M,,,,,,,,,, = 2.70, SD = 0.76; F (1, 814) = 86.84, p < .001]. Participants in the School sample scored higher than participants in the FTE sample [Hostile Sexism: Mrchool= 3.83, SD = 0.99, MFrE= 3.26, SD = 0.98; F (2, 814) = 18.01, p
