Abstract
In recent years, beauty ideals and practices have been explained almost exclusively using evolutionary psychological frameworks, to the exclusion of more proximate factors such as psychosocial and individual psychological variables. To overcome this limitation, we examined the associations among sexist beliefs, objectification of others, media exposure, and three distinct beauty ideals or practices. Across three studies, a total of 1,158 participants in a British community sample completed a series of scales that measured their attitudes toward women, hostility toward women, benevolent sexism, hostile sexism, their tendency to objectify others, media exposure, and endorsement of the thin ideal and (for women) body dissatisfaction (Study 1); height preferences in an other-sex partner (Study 2); and endorsement of cosmetic use (Study 3). Across the three studies, results supported the idea that sexist beliefs predicted beauty ideals and practices, although the strength of these associations varied according to the ideal or practice in question. These results support feminist critiques that beauty ideals and practices in Western societies are linked with sexist attitudes. Furthermore, our results suggest that programmes aimed to reduce or eliminate sexist attitudes, or that promote more gender egalitarian attitudes, may result in healthier beauty ideals and practices.
Beauty ideals have played an extremely important role in most human cultures, and so it is not surprising that a substantial body of work has sought to understand the factors that shape such ideals in contemporary cultures (for reviews, see Moradi & Huang, 2008; Swami, 2007; Swami & Furnham, 2007, 2008; Swami & Salem, in press). One particular aspect of this literature seeks to understand the function and purpose of beauty ideals, as exemplified in feminist critiques of Western beauty standards. A common theme shared by feminist authors is the idea that corporeal experiences are shaped by patriarchal structures in society, gendered identities, and power relations between women and men (e.g., Bartky, 1990; Bordo, 1993; Calogero, Boroughs, & Thompson, 2007; Jeffreys, 2005; Smolak & Murnen, 2007). Seen from this perspective, beauty ideals and practices are believed to originate, in part at least, from oppressive beliefs and attitudes directed at women.
Specifically, a number of feminist authors have argued that beauty ideals and practices arose due to pressures that women faced in male-dominated societies (e.g., Bartky, 1990; Bordo, 1993). Dworkin (1974), and more recently Wolf (1990), rejected the “masculine aesthetics” that shift awareness away from women's real competencies to superficial aspects associated with beauty and appearance. Beauty standards, in this view, reduce women to the status of sex objects, causing them to feel that their bodies are inadequate and forcing them to engage in practices that leave them feeling inauthentic if they do not live up to the idealized image of femininity. As such, women in patriarchal societies do not have any real control over their bodies, and the body remains an important site for oppression (Calogero et al., 2007; Smolak & Murnen, 2007).
In this view, beauty ideals and practices are identified as a form of oppression against women as well as a means for masculine society to denigrate and impede women's liberation and gender equality (Baker-Sperry & Grauerholz, 2003; Bordo, 1993; Wolf, 1990). For example, Dworkin (1974) has argued that beauty practices, aside from being time wasting, expensive, and painful, also harm women's bodies and psyches:
Standards of beauty describe in precise terms the relationship that an individual will have to her own body. They prescribe her mobility, spontaneity, posture, gait, the uses to which she can put her body. They define precisely the dimensions of her physical freedom (p. 112; see also Tiggemann & Rothblum, 1997).
For Dworkin (1974), then, beauty standards are “the major substance of male-female role differentiation, the most immediate physical and psychological reality of being a woman” (p. 112). Male-dominated culture uses beauty ideals and practices to ensure that the domination of one sex over the other continues (Jeffreys, 2005).
A handful of studies has attempted to apply these perspectives to practical research. For example, one recent study postulated that “beauty ideals are oppressive” (BIO; Forbes, Collinsworth, Jobe, Braun, & Wise, 2007), which refers to both the processes and consequences of that oppression and, consequently, focuses attention on the negative aspects of beauty ideals and practices. The BIO hypothesis is an important contribution to the extant literature for a number of reasons. First, as Forbes et al. (2007) note, it encourages researchers to shift the focus of research from individual psychologies to the social meaning and function of beauty ideals and practices (p. 266). In addition, the BIO perspective allows researchers to approach various beauty practices from the same starting point, given that most beauty practices are thought to serve the same functions and motivations (Forbes et al., 2007).
In our study, we sought to build on and extend the work of Forbes et al. (2007) by using their delineation of the BIO hypothesis. Specifically, we examined the association of three specific beauty ideals and practices (namely, the thin ideal and body dissatisfaction, height preferences in a potential partner, and cosmetic use) with various sexist beliefs among heterosexual women and men. Following Forbes et al. (2007), sexism was operationalized through the measurement of sexist attitudes toward women. In addition to sexism, however, we extended previous work by including measures of the tendency to objectify others and media exposure. Taken together, the focus on specific beauty ideals and practices, along with the inclusion of additional measures, allows for a comprehensive test of the BIO hypothesis. In the following sections, we introduce the key variables we included in our study.
Sexism
Sexism generally refers to the belief that one sex is inferior to the other and to the corollary assignment of roles and privileges based on gendered identities (Abrams, Viki, Masser, & Bohner, 2003). In patriarchal societies, the roles and privileges accorded to women are inferior to those assigned to men, and as such, sexism plays a central role in the continuing oppression of women. Moreover, and as predicted by the BIO hypothesis, the existence of patriarchal structures and attitudes should result in significant relationships between sexist attitudes and the endorsement of beauty ideal and practices (Forbes et al., 2007).
Although there are a number of scales that can be used to measure sexist attitudes among women and men, the present study followed Forbes et al. (2007) in operationalising sexism using three distinct scales: the Attitudes Toward Women Scale (AWS; Spence, Helmreich, & Stapp, 1978), the Hostility Toward Women Scale (HTWS; Lonsway & Fitzgerald, 1995), and the Ambivalent Sexism Inventory (ASI; Glick & Fiske, 1996, 2001). The AWS is one of the most widely used scales for the measurement of blatantly sexist attitudes toward women (particularly in terms of women's rights and gendered roles) and, despite its age, continues to be a useful measure of traditionally sexist attitudes (McHugh & Frieze, 1997; Spence, 1998). The HTWS, on the other hand, does not describe appropriate gendered roles or behaviors for women, and as such, provides a more direct measure of hostility toward women (Lonsway & Fitzgerald, 1995).
Finally, the ASI provides a more sophisticated measure of sexism because it recognizes two distinct components of sexism. First, hostile sexism (ASI-H) refers to a tendency to explicitly denigrate women through the justification of patriarchy and gendered assignment of roles as well as to impose restrictions on the roles attributed to women. Second, benevolent sexism (ASI-B) refers to a tendency to idealize women's traditional roles while simultaneously limiting them to subservient positions in society (Glick & Fiske, 1996, 1997). Although this ideology accords women limited protection and privilege in terms of legislation (e.g., protecting women from physically demanding employment) or social rules (e.g., privileging them by opening doors for women), it is also predicated on perceptions of women as subordinate to men (Glick & Fiske, 2001; Glick, Diebold, Bailey-Werner, & Zhu, 1997; Masser & Abrams, 1999). The measurement of benevolent sexism is important because research suggests this ideology may be more subtle than hostile sexism, because women are less likely to recognize it and, in some circumstances, may participate in its perpetuation (Kilianski & Rudman, 1998).
Other Objectification
In addition to the previous measures of sexism, we assessed the tendency for individuals to objectify others, that is, to view the body as a sexualized object separate from nonphysical characteristics (e.g., Brausch, Muehlenkamp, & Swanson, 2005; Moradi, Dirks, & Matteson, 2005; Strelan & Hargreaves, 2005). In Bartky's (1990) view, a person is objectified when they are reduced to “the status of mere instruments” available for visual inspection, evaluation, and the use of others (p. 26). Indeed, there is a great deal of evidence suggesting that women are objectified in contemporary Western cultures, ranging from the prevalence of idealized images of women's bodies to sexual comments about appearance, sexual harassment, and sexual violence (e.g., Frederickson & Roberts, 1997).
Numerous studies and commentaries have documented negative psychological consequences for women as a result of objectification and objectified body consciousness (e.g., Frederickson & Roberts, 1997; McKinley & Hyde, 1996; Roberts & Gettman, 2004). Moreover, objectification can be viewed as one form of gendered oppression: Whereas both concepts reflect the way in which women are perceived and treated in contemporary societies (Glick et al., 1997), objectification in particular reduces women to the status of objects that exist for the pleasure of others (Bartky, 1990). Conceptually, then, the tendency to perceive women as being inferior to men may be associated with the tendency to treat women as objects, devoid of any real nonphysical characteristics (McKee, 2005). As such, other-objectification, operating in tandem with sexism, may result in a stronger tendency to endorse beauty practices and ideals, although it should be noted that studies assessing this possibility are limited.
Traditionally, objectification was viewed primarily in terms of the male gaze directed at women (McKee, 2005), but recent research has highlighted the fact that both men and women objectify others (Strelan & Hargreaves, 2005). Within the literature on objectification, the majority of work among women has focused on self- rather than other-objectification (e.g., Prichard & Tiggemann, 2005; Quinn, Kallan, Twenge, & Frederickson, 2006; Roberts & Gettman, 2004), but it is worth noting that women are also objectified by other women and that self- and other-objectification are linked (e.g., Strelan & Hargreaves, 2005). In terms of our study, it is other- rather than self-objectification that factors into women's oppression more generally (Frederickson & Roberts, 1997) and, as such, we use the term “other-objectification” to refer to the general trait of objectifying others—rather than to a single act of objectification or the act of self-objectification (following Gurung & Chrouser, 2007, pp. 91–92).
Media Exposure
There is a large body of research highlighting the role played by media exposure in the endorsement of beauty ideals, particularly the thin ideal (for a review, see Calogero et al., 2007). Within this perspective, it is argued that some media communicate and reinforce beauty ideals, which are then internalized and used by individuals when judging others (Levine & Harrison, 2004). Moreover, there is a growing body of work implicating the media in shaping ideas about sexuality (e.g., Cowan, 2002; Ward, 2003), the objectification of women (for a review, see Ward, 2003), and the social denigration of women (e.g., Shields & Heinecken, 2002). As such, media exposure (which we operationalized as lifetime exposure to Western television, movies, magazines, and music) can be expected to be associated with beauty ideals and practices, either directly or through its influence on sexism and other-objectification.
The Present Studies
To summarize, the aim of the present series of studies was to examine the associations among measures of sexism, other-objectification, media exposure, and three specific beauty ideals and practices. In Study 1, we explored the associations of these measures with endorsement of the thin ideal for women and (among women) levels of body dissatisfaction; in Study 2, we investigated the association of the aforementioned measures with height preferences in a potential partner; and in Study 3, we examined the association of these measures with the belief that women should use cosmetic products. Specific hypotheses are outlined in the following sections.
STUDY 1: THE THIN IDEAL AND BODY DISSATISFACTION
Only a handful of studies has directly examined the associations between sexism with endorsement of the thin ideal and with body dissatisfaction among women. In an early study, Franzoi and Chang (2000) hypothesized that the thin ideal itself, and practices associated with achieving it, was an example of benevolent sexism in operation (e.g., being thin was expected to increase women's influence over men in certain areas, such as receiving praise and admiration, but also weaken it in others, especially given the association between the thin ideal and body image disorders). Even so, Franzoi (2001) reported that women who scored higher on benevolent sexism were not more likely than women who scored lower to express positive attitudes toward aspects of body esteem, including “weight concern” or “physical condition” (although we should note that the focus of his study was on body esteem specifically and not the thin ideal or body dissatisfaction).
In a more recent study, Forbes et al. (2007) examined the associations among several measures of sexism and several measures of the thin ideal—that is, endorsement of the importance of thinness via a statement-based subscale and two sets of “weight discrepancy” scores derived from the Figure Rating Scale (Stunkard, Sørenson, & Schulsinger, 1983). Forbes et al. (2007) reported that their Importance of Thinness subscale was significantly and positively correlated with HTWS, AWS, and ASI-H (r = .19–.47), but not with ASI-B. On the other hand, there was only one significant correlation between their weight discrepancy scores derived from the Figure Rating Scale and measures of sexism (specifically with AWS, r = .15).
The lack of significant results reported by Forbes et al. (2007) in terms of weight discrepancy, however, may have been caused by problems associated with the uncertain validity of the way in which discrepancy was measured (see Forbes et al., 2007, p. 272) as well as by limitations associated with the use of line-drawn stimuli. (For a discussion, see Swami, Salem, Furnham, & Tovée, 2008a.) Other relevant work has suggested that women who hold feminist attitudes are more likely than those who do not to perceive a wider range of female body sizes as being attractive (Swami, Salem, Furnham, & Tovée, 2008b) and to have more positive body image (e.g., Myers & Crowther, 2007; Rubin, Nemeroff, & Russo, 2004).
In Study 1, we examined the associations among sexism, other-objectification, media exposure, and measures of the thin ideal and body dissatisfaction. The latter measurements were made using the newly validated Photographic Figure Rating Scale (PFRS; Swami et al., 2008a), which allowed us to overcome some of the measurement limitations associated with previous work, as noted previously. Based on the PFRS, we examined ratings of the figure that participants perceived as being maximally attractive, the range of figures that participants perceived as being physically attractive, and (for female participants only) body dissatisfaction scores (calculated by subtracting ideal ratings from current ratings). Based on the available literature, we expected significant correlations among all three dependent variables and measures of blatant sexism, other-objectification, and media exposure, but possibly not benevolent sexism given previous work (Forbes et al., 2007; Franzoi, 2001).
Method
Participants
The participants were 182 women and 169 men recruited from the community in Greater London. The majority of participants were of European Caucasian descent (72%), with others being of Asian (17%), African Caribbean (2%), or other descent (9%). Fully 44% of participants were Christians, whereas others self-reported as being atheists (29%), Muslims (8%), unsure of their religious affiliation (6%), or of some other religion (13%). In terms of education, 16% had been educated to a General Certificate of Education (GCE) Secondary Level (generally taken by students 15–16 years of age), 39% had GCE Advanced Levels qualification (generally taken by students 16–18 years of age), 28% completed an undergraduate degree, 10% had a postgraduate degree, and 6% had some other academic qualification. Finally, 44% of participants were single, whereas the remainder were married (26%), in a relationship (25%), or of some other status (5%). Mean age and self-reported body mass index (BMI) for women and men in this sample are reported in Table 1.
Descriptive Statistics as Well as Between-Sex Comparisons and Interscale Correlations for Study 1
PFBS
The PFRS (Swami et al., 2008a), an advance over the Contour Drawing Figure Rating Scale (Thompson & Gray, 1995), comprises 10 gray scale photographic figures of real women in front view and dressed in standard leotards. The images represent two women from each of the established BMI categories: emaciated (<15 kg/m2), underweight (15– 18.5 kg/m2), normal weight (18.5–24.9 kg/m2), overweight (25.0–29.9 kg/m2), and obese (>30 kg/m2). All participants were asked to rate (a) the figure they considered most physically attractive, (b) the largest figure they considered physically attractive, and (c) the thinnest figure they considered physically attractive. The latter two items were used to calculate an attractiveness range by subtracting the thinnest figure from the largest figure (following Fisak, Tantleff-Dunn, & Peterson, 2007; Swami, Buchanan, Furnham, & Tovée, 2008c). In addition, women were asked to rate the figure that (a) most closely approximated their current body and (b) they would most like to possess. These two items were used to calculate a body dissatisfaction score by subtracting ideal ratings from current ratings. Responses to all five items were made on a 10-point scale—with 1 representing the woman with the lowest BMI and 10 representing the woman with the highest BMI. Previous work has shown that the PFRS has high construct validity and test-retest reliability (Swami et al., 2008a).
Other Objectification
A modified version of the Self-Objectification Scale (Frederickson, Roberts, Noll, Quinn, & Twenge, 1998) was used to measure other-objectification. The original Self-Objectification Scale asked participants to rank how important 10 body attributes were to their physical self-concept. Following previous work (Gurung & Chrouser, 2007; Strelan & Hargreaves, 2005), we asked participants to rate these attributes according to how important they are when looking at others. Five of the attributes are competence-based (strength, physical coordination, energy level, health, and physical fitness), whereas five others are appearance-based (weight, sex appeal, physical attractiveness, firm/sculpted muscles, and measurements). Participants ranked each attribute on a scale ranging from 0 (least impact) to 9 (greatest impact). A single total score was computed by subtracting the sum of the competence-based items from the sum of the appearance-based items. Scores ranged from −25 to +25, with higher scores indicating greater emphasis on appearance and, by extension, objectification of others (Frederickson et al., 1998; cf. Gurung & Chrouser, 2007). Although it is not possible to compute an internal reliability coefficient (because the scale consists of ranked items), Hill and Fischer (2008) noted that the correlation between the sum of the appearance attributes and the competence attributes provides a useful measure of item interrelation. Specifically, they pointed out that, although the appearance-based and competence-based items were not intended to be used as independent measures, the two sets of items should be negatively correlated (that is, individuals who prioritize appearance should conversely denigrate competence). In their study, Hill and Fischer (2008) reported that the correlation between the sum of rankings for the appearance-based items and the competence-based items was −81. Likewise, in the present study, the correlation between the sum of appearance- and competence-based items was −89.
HTWS
The HTWS (Lonsway & Fitzgerald, 1995) is a 10-item scale that does not contain any items that describe appropriate roles, behaviors, or gendered identities for women and, as such, is said to provide a relatively pure measure of hostility toward women (sample item: “I think that most women would lie just to get ahead”). Items were rated on a 7-point Likert-type scale ranging from 1 (strongly disagree) to 7 (strongly agree), and certain items were reverse coded before an overall mean score was computed (higher scores indicated greater hostility toward women). Cronbach's alpha with our sample was .80.
AWS
The AWS (Spence et al., 1978) is a 25-item measure of blatantly sexist attitudes toward women (sample item: “Swearing and obscenity are more repulsive in the speech of a woman than of a man”). Items were rated on a 4-point scale ranging from 1 (agree strongly) to 4 (disagree strongly), and certain items were reverse coded before an overall mean AWS score was computed. In its original form, higher scores indicated a more profeminist, egalitarian attitude. In the present study, the order of anchors was reversed prior to analyses (with higher scores indicating more sexist attitudes) so as to maintain directional consistency with other measures (α = .81).
ASI
The ASI (Glick & Fiske, 1996, 2001) is a 22-item scale that measures two types of sexism: hostile (ASI-H sample item: “Most women fail to appreciate fully all that men do for them”) and benevolent (ASI-B sample item: “A good woman should be set on a pedestal by her man”). All items were rated on a 6-point Likert-type scale ranging from 1 (disagree strongly) to 6 (agree strongly), and certain items were reverse coded prior to analyses so that higher mean scores indicated higher sexism. Cronbach's alphas were .90 for ASI-H and .87 for ASI-B.
Media Exposure
All participants rated their lifetime exposure to Western media across four sources—namely television, movies, magazines, and music, coded 1 (every day) to 7 (1ess than once a month). The four items were subsequently subjected to a factor analysis. The significance of Bartlett's test of sphericity, χ2 = 34.68, df = 6, p < .001, and the size of the Kaiser-Meyer-Olkin measure of sampling adequacy, KMO = .80, revealed that these items had adequate common variance for factor analysis (Tabachnick & Fidell, 2007). A factor analysis was, therefore, conducted using Varimax (orthogonal) rotation, and the number of factors to be extracted was determined both by factor eigenvalues above 1.0 and inspection of the Scree plot (Cattell, 1966). Based on these criteria, a single factor was extracted, which in total explained 38.5% of the variance. All items loaded onto this factor with loadings greaterthan .45 (eigenvalue = 1.54). We, therefore, calculated a composite Western media exposure score by taking the mean of all four items. In order to maintain consistency with other scales, the anchor of this measure was reversed so that higher scores indicated greater media exposure (Cronbach's α = .84).
Demographics
All participants were asked to provide their age, sex, ethnicity, religion, highest educational qualification, marital status, weight, and height. The latter two items were used to calculate participants' BMI, and previous work has shown that this type of self-reported data is reliable when the anonymity of respondents is ensured (Davis, 1990).
Procedure
All participants were recruited on a convenience basis, which in practice meant a multitude of researchers approaching potential participants at several public locations (e.g., train stations, parks, and shopping centres). The only constraint on participation was that potential participants must be adult and heterosexual. The nature of the experiment was explained to participants, and once they provided informed consent, they were given a six-page, paper-and-pencil questionnaire to complete along with brief instructions. The order of presentation of each scale within the questionnaire was randomized for each participant. All participants took part on a voluntary basis and were not remunerated for their participation. The questionnaire took approximately 10 minutes to complete. They were verbally debriefed after they had completed and returned the questionnaire to the experimenter.
Results and Discussion
Descriptive Statistics
Descriptive statistics for each scale, along with participant age and BMI, are reported in Table 1. There were no significant gender differences in the photographic figure rated as the most attractive: Both women and men rated an underweight female figure as the most attractive (consistent with previous work using the PFRS, e.g., Swami et al., 2008b, 2008c). On the other hand, there were significant gender differences on HTWS, AWS, ASI-H, and ASI-B, with men endorsing more sexist attitudes on each of these scales. Effect sizes were largest on ASI-H (d = 0.61) and smallest on HTWS (d = 0.21). No significant gender differences emerged on any of the other measures, or in mean age and BMI.
Interscale Correlations and Multiple Regressions
Interscale correlations are reported in Table 1. For women, ratings of the most attractive figure were significantly and negatively correlated with objectification, AWS, and ASI-H, and positively correlated with age and BMI. In addition, the attractiveness range was significantly and negatively correlated with objectification, HTWS, AWS, and ASI-B, and positively correlated with body dissatisfaction. Finally, body dissatisfaction was significantly and negatively correlated with objectification and ASI-B, and positively correlated with age and BMI. Among men, ratings of the most attractive figure were significantly and negatively correlated with objectification and all four measures of sexism as well as positively correlated with BMI. In addition, the attractiveness range was significantly and negatively correlated with all four measures of sexism. In general, the pattern of correlations between the sexism measures and the ideal figure and the attractiveness range, respectively, were similar for women and men.
We then conducted a series of three multiple linear regressions with the figure rated as the most attractive, the attractiveness range, and (for women only) body dissatisfaction, respectively, as the dependent variables. Eight items were entered simultaneously as predictor variables, namely other-objectification, HTWS, AWS, ASI-B, ASI-H, media exposure, age, and participant BMI. Given the sufficiently large sample size (Tabachnick & Fidell, 2007), we conducted separate regressions for women and men.
For women, the first regression with the figure rated as most attractive returned a significant result, F(8, 181) = 5.08, p < .001, adj. R 2 = .17. Of the variables entered into the regression, more sexist attitudes on the AWS (β = −27, t = −3.65, p < .001), a higher BMI (β = .24, t = 3.36, p = .001), and a greater tendency to objectify others (β = −22, t = −3.11, p = .007) significantly predicted ratings of a thinner body as attractive. When the attractiveness range was entered as the dependent variable, the regression for women was significant, F(8, 181) = 2.97, p < .001, adj. R 2 = .08, with a greater tendency to objectify others being associated with a narrower attractiveness range (β = −32, t = −4.03, p < .001). Finally, for body dissatisfaction, the regression was likewise significant, F(8, 181) = 7.78, p < .001, adj. R 2 = .24. Of the variables entered into the regression, a higher BMI (β = .39, t = 5.84, p < .001) and greater benevolent sexism (β = −14, t = −1.93, p = .045) were associated with greater body dissatisfaction.
For men, the regression with ratings of the most attractive figure was significant, F(8, 168) = 5.16, p < .001, adj. R 2 = .18, with more sexist attitudes on the AWS (β = −36, t = −4.20, p < .001) and greater benevolent sexism (β = −18, t = −2.32, p = .027) predicting a preference for a thinner figure. The regression for the attractiveness range with male participants was also significant, F(8, 168) = 4.09, p < .001, adj. R 2 = .14, with more sexist attitudes on the AWS (β = −29, t = −3.68, p < .001) predicting a narrower attractiveness range.
Conclusions
Overall, the results of Study 1 showed that there were a number of significant predictors of body weight ideals and, for women, body dissatisfaction. For both women and men, it was notable that AWS, a measure of blatant sexist beliefs, significantly predicted ratings of thinner female figures as being maximally attractive and (among men) a narrower range of figures as being attractive. These results are generally consistent with findings by Forbes et al. (2007), who showed that sexist attitudes were associated with a stronger belief that thinness was important for women (where the latter was measured using a statement-based scale). More generally, these findings support the feminist argument that individual differences in sexism are related to endorsement of the thin ideal and that oppressive patriarchal beliefs shape beauty ideals among women and men (e.g., Bartky, 1990; Bordo, 1993; Jeffreys, 2005; Smolak & Murnen, 2007).
In contrast to previous work (Forbes et al., 2007; Franzoi, 2001), our results also showed that benevolent sexism was associated with a preference for a more slender figure among men and with greater body dissatisfaction among women. In general, this set of results supports the argument that one avenue in which benevolent sexism is expressed is in idealized notions of feminine beauty, which ultimately serves to maintain gender inequalities (Franzoi, 2001). Of course, given the correlational design of the present study, it is difficult to ascertain distal motives in these associations. Nevertheless, it might be speculated that benevolently sexist men view their attractiveness judgements as being positive (i.e., benevolent), whereas women high in benevolent sexism develop self-beliefs that they must live up to idealized notions of bodily attractiveness in order to achieve positive gains (e.g., admiration; cf. Franzoi, 2001). To the extent that women are unable to live up to those ideals, they may experience body dissatisfaction and, in extreme cases, symptoms of eating disorders.
Our results also showed that, among women, a greater tendency to objectify others significantly predicted ratings of a thinner figure as being maximally attractive and a narrower range of figures as being physically attractive. A possible reason why other-objectification significantly predicted the ratings of women—but not of men—may be that, given women are the primary targets of objectification (via the male gaze), they may internalize the belief that women must be thin in order to be valued. Moreover, the tendency among women to objectify others is associated with self-objectification: One study showed that women who placed greater importance on their own weight also prioritized this dimension when evaluating other women (Beebe, Holmbeck, Schober, Lane & Rosa, 1996).
Another noteworthy (and expected) association was that of higher female BMI predicting a preference for a heavier figure and greater body dissatisfaction. Previous work has shown that BMI is the primary component of women's attractiveness (e.g., Swami & Tovée, 2005a, 2006a, 2006b; Swami, Miller, Furnham, Penke, & Tovée, 2008d; Swami, Neto, Tovée, & Furnham, 2007; Tovée, Swami, Furnham, & Mangalparsad, 2006) and that ratings of BMI-related physical attractiveness are dependent on the observer's own BMI (see Tovée, Emery, & Cohen-Tovée, 2000). Moreover, the association between women's BMI and body dissatisfaction is consistent with previous work (e.g., Paxton, Eisenberg, & Neumark-Sztainer, 2006). In contrast, we found no association between the thin ideal and media exposure. The most likely explanation for this nonfinding is that our measure of media exposure was not sensitive enough to tap this construct, particularly its influence in shaping beauty ideals. Future work could overcome this limitation by assessing acceptance of media messages about beauty ideals, rather than media exposure per se.
Overall, these results corroborate previous work (e.g., Forbes et al., 2007; Franzoi, 2001) in showing that measures of sexism and other-objectification are related to individuals' differences in the endorsement of the thin ideal and, among women, levels of body dissatisfaction. In Study 2, we extended these findings by examining the association of the same predictor variables with height preferences in a potential partner.
STUDY 2: HEIGHT PREFERENCES
Height preferences in a potential partner have been extensively studied in the psychological literature, with most studies documenting a “male-taller norm,” such that both women and men prefer to be in relationships where the man is taller than the woman (e.g., Jackson & Ervin, 1992; Pawłowski & Jasieńska, 2005; Pawłowski & Koziel, 2002). Of course, height preferences are predicated on the relative height of women and men, which Pawłowski (2003) operationalized as a sexual dimorphism ratio (SDR) in stature (measured as men's height divided by women's height). Recent work has shown that individuals adjust their SDR preferences in relation to their own height (e.g., Fink, Neave, Brewer, & Pawłowski, 2007), although there remains a male-taller norm in most preferences.
Almost without exception, the male-taller norm has been explained in terms of an evolutionary adaptation, where height and stature in men is assumed to signal preferred traits (e.g., the ability to secure resources or intimidate potential rivals; Salska et al., 2008) or heritable qualities (taller men have greater reproductive success than men of average height; Nettle, 2002a; Pawłowski, Dunbar, & Lipowicz, 2000). Given that there is no reproductive advantage for women in being taller than average (Nettle, 2002b), it is argued that height is less important to men when judging the physical attractiveness of women. As argued by Swami et al. (2008e), however, the evolutionary psychological explanation of height preferences fails to appreciate that such preferences often reflect basic gender stereotypes that prescribe normative SDR-related differentials in relationships (e.g., Martel & Billier, 1987).
Specifically, height preferences may reflect the cultural transmission of gender-appropriate behavior, where tall men are socially constructed as being masculine and fulfilling gender norms and tall women as violating those norms (Helgeson, 1994). Thus, men and women should enter relationships where the man is taller in order to avoid contravening gendered norms and any associated stigma. If this is the case, then studies should expect to find significant associations between height preferences and the endorsement of gendered norms of identity. However, two previous studies have found that SDR preferences are only weakly correlated with endorsement of traditional male roles (as measured by the Male Role Attitudes Scale and the Male Role Norms Scale, respectively; Salska et al., 2008; Swami et al., 2008e).
These weak correlations were explained as a function of the limited utility of the aforementioned scales to tap beauty ideals and practices (Swami et al., 2008e). In Study 2, we sought to extend the available research on gendered norms and SDR preferences by examining the latter's association with measures of sexism, other-objectification, and media exposure. Based on the above argument that SDR preferences reflect, in part at least, endorsement of gendered norms of behavior, we expected significant associations with our three measures of sexism. On the other hand, our reading of the literature did not highlight any reasons as to why other-objectification and media exposure should be related to height preferences in a potential partner (although it may be related to endorsement of the male-taller norm in general).
Method
Participants and Procedure
Participants in Study 2 were 165 men and 218 women recruited from a community sample in Greater London (age and BMI data are reported in Table 2). Most participants were of European Caucasian descent (70%), whereas others were of Asian descent (22%) or of some other ancestry (9%). In terms of religion, 41% were Christians, 29% were atheists, 7% were Muslims, 7% were unsure, and 16% were of some other religion. Of this sample, 14% had been educated to a GCE Secondary Level, 41% reported having Advanced Levels, 31% had an undergraduate degree, 9% had a postgraduate degree, and 5% had some other academic qualification. Finally, 49% of participants were single, whereas 26% were married, 21% were in a relationship, and 5% were of some other status. The procedure of this study was identical to that used in Study 1, including the measures of demographics.
Descriptive Statistics as Well as Between-Sex Comparisons and Correlations for Study 2
Ideal Height and Sexual Dimorphism Ratio
Participants indicated the ideal height of someone with whom they would like to be in a romantic heterosexual relationship. Following previous work (Salska et al., 2008; Swami et al., 2008e), the ideal SDR for men was calculated as participants' own height divided by their ideal partner's height, and for women as their ideal partner's height divided by participants' own height. In both cases, therefore, higher SDRs indicated a desire for a relationship in which the male partner is taller than the female partner. All responses were coded in metres.
Other-Objectification, Sexism, and Media Exposure
We used the same scales as in Study 1 to measure other-objectification, sexism (HTWS, AWS, ASI-H, ASI-B), and media exposure. Cronbach's alpha coefficients for these scales were: HTWS = .79, AWS = .85, ASI-H = .89, ASI-B = .86, and media exposure = .87. The correlation between the sum of appearance- and competence-based other-objectification items was −87.
Results and Discussion
Descriptive Statistics
Descriptive statistics for each measure used in Study 2, along with participant age and BMI, are reported in Table 2. There were only two significant gender differences with this sample. First, men had significantly higher scores than women on ASI-H (d = 0.52) and second, men were significantly older than women (d = 0.45). There were no gender differences on any of the other measures, or in mean BMI.
Interscale Correlations and Multiple Regressions
All interscale correlations are reported in Table 2. Women's SDR was significantly and positively correlated with AWS, ASI-H, ASI-B, and BMI, and negatively correlated with age, whereas men's SDR was significantly and positively correlated with HTWS, AWS, ASI-H, and ASI-B.
We then conducted multiple linear regressions with women's and men's SDR, respectively, as the dependent variables and all other measures as predictor variables entered simultaneously. The regression for women's SDR was significant, F(8, 217) = 3.13, p = .004, Adj. R 2 = .08, with greater sexism on the ASI-B (β = .16, t = 2.37, p = .019), AWS (β = .17, t = 2.00, p = .047), and ASI-H (β = .16, t = 1.94, p = .049) predicting a preference for the male-taller norm in relationships. The regression for men's SDR was also significant, F(8, 164) = 4.00, p < .001, Adj. R 2 = .16, with greater sexism on ASI-B (β = .27, t = 3.00, p = .005) and AWS (β = .16, t = 2.00, p = .046) significantly predicting a preference for the male-taller norm.
Conclusions
Consistent with our hypotheses, the results of Study 2 showed that sexist beliefs were significant predictors of SDR among women and men. Importantly, our results showed that, of all our measures of sexism, benevolent sexism was the strongest predictor of the male-taller norm for both women and men. Based on this pattern across measures, we argue that benevolent sexism motivates a preference for the male-taller norm in relationships because of the promise of social or gendered reward for individuals in such a relationship, at least in Western societies. That is, the association between benevolent sexism and SDR may be explained in part by the characterisation of women as being weaker than men and in need of men's protection. Individuals who conform to the male-taller norm in relationships may receive social reward for the conformity (or conversely, individuals who do not conform may be socially punished). As a consequence, the status of women may be undermined insofar as they are continually characterized as the weaker sex and to the extent that they are treated unequally in relation to men (cf. Glick et al., 1997).
However, our results also showed that attitudes toward women and (among women) hostile sexism predicted the male-taller norm. That is, scores on the AWS (and, among women, the ASI-H) partly explained the preference for a taller man in relationships, once the variance of ASI-B had been accounted for. As in Study 1, these results are consistent with the feminist argument that overtly sexist beliefs shape beauty ideals among women and men (e.g., Bartky, 1990; Bordo, 1993; Jeffreys, 2005; Smolak & Murnen, 2007). More generally, our results highlight the importance of considering the social context, and particularly gendered norms of behavior, when considering height preferences. This point is important given that almost all recent work on SDR preferences has utilized a biological framework that does not explicitly consider the social construction of height preferences (e.g., Fink et al., 2007; Pawłowski, 2003; Pawłowski & Jasieńska, 2005; Pawłowski & Koziel, 2002).
STUDY 3: ENDORSEMENT OF COSMETIC USE
There now exists a small body of research examining the use and effects of cosmetics, both across cultures and time (e.g., Marwick, 1988; Richetin, Croizet, & Huguet, 2004). Recent work has tended to address three separate but related issues: first, the association between the use of cosmetics and improvements in judgements of attractiveness (e.g., Cox & Glick, 1986; Kligman & Graham, 1989; Nash, Fieldman, Hussey, Lévěsque, & Pineau, 2006); second, the way cosmetics affect perceptions of women as well as self-perceptions (e.g., Nash et al., 2006; Richetin et al., 2004); and third, individual difference predictors of cosmetic use (e.g., Robertson, Fieldman, & Hussey, 2008).
To our knowledge, however, only two previous studies have examined the association between sexist beliefs and cosmetic use. Franzoi (2001) reported that women who endorsed benevolent sexist beliefs were more likely to use cosmetic products when going out on a romantic date, although they did not have a stronger desire to use such products generally or actually use more products on the day of testing. More recently, Forbes, Jung, and Haas (2006) replicated these results among college and adult samples. As argued by Franzoi (2001), the use of cosmetic products allows women to conform to beauty ideals and thus increases their chances of receiving gender-related rewards. Even so, the use of cosmetics may also reduce women's social power: Studies have shown that women who wear makeup in occupational settings are judged to be less capable than those who do not wear makeup (Cox & Glick, 1986; Kyle & Mahler, 1996). Overall, this double-bind (increased perceptions of attractiveness versus decreased social power) was argued to be a defining feature of benevolent sexism (Franzoi, 2001).
However, one limitation is that these previous studies have used measures of cosmetic use that have uncertain validity. To overcome this limitation, we measured endorsement of cosmetic use with a recent scale that has been assessed for validity and reliability (Robertson et al., 2008). In addition, it seems conceptually plausible that other-objectification and media exposure should be strongly related to cosmetic use, but previous work has not examined this possibility. In the first instance, it might be argued that the perceived need to conform to feminine beauty ideals by artificial means is a central component of objectification. That is, individuals who tend to objectify others may more strongly endorse attempts to increase their own physical attractiveness through artificial methods. Second, much of the information about the methods and necessity of cosmetic use may be garnered through the media, particularly through advertising but also indirectly through the association between cosmetics and the perceived success of individuals who use such products (see Cohen Ragas & Kozlowski, 1998). Although this part of our study was more speculative, we nevertheless expected significant associations among other-objectification, media exposure and endorsement of cosmetic use.
Method
Participants and Procedure
Participants were 424 individuals from the community, comprising 266 men and 158 women (age and BMI data are reported in Table 3). The majority were of European Caucasian descent (67%), whereas others were of Asian (20%) or other descent (13%). In terms of religion, 39% self-reported being Christian, 31% as atheists, 9% as Muslims, 5% as being unsure, and 17% as being of some other religion. In total, 12% had been educated to a GCE Secondary Level, 42% had Advanced Levels, 33% had an undergraduate degree, 9% had a postgraduate degree, and 4% had some other academic qualification. The majority of these participants were single (52%), whereas others were in a relationship (21%), married (21%), or of some other status (5%). The procedure of this study was identical to that used in Study 1.
Descriptive Statistics as Well as Between-Sex Comparisons and Correlations for Study 3
Endorsement of Cosmetic Use
Robertson et al. (2008) developed a 9-item scale that measures customary cosmetic use and the discomfort individuals feel when not using cosmetics (sample item: “I apply facial cosmetics every day”). In our study, women completed the original measure, whereas men responded to an adapted measure that focused on women's use of cosmetics (sample item: “Women should apply facial cosmetics every day”). All items were rated on a 5-point Likert-type scale ranging from 1 (most agree) to 5 (most disagree). Several items were reverse coded before a mean score was computed. In its original form, higher scores indicated lower cosmetic use, but in our study, we reversed the anchors so that higher scores indicated greater endorsement of cosmetic use. Previous work has shown that this scale has good internal consistency and construct validity (Robertson et al., 2008). In our study, Cronbach's alpha was .87.
Other-Objectification, Sexism, and Media Exposure
Other-objectification, sexism (HTWS, AWS, ASI-H, ASI-B), and media exposure were measured using the same scales as introduced in Study 1. Cronbach's alpha coefficients for these scales were: HTWS = .80, AWS = .83, ASI-H = .92, ASI-B = .91, media exposure = .88. The correlation between the sum of appearance- and competence-based other-objectification items was −87.
Results and Discussion
Descriptive Statistics
Descriptive statistics for each measure used in Study 3, along with participant age and BMI, are reported in Table 3. Women were more likely to endorse cosmetic use for themselves than men were for women (d = 0.86). In contrast, men had higher scores on HTWS, AWS, ASI-H, and ASI-B (d ranged from 0.20 to 0.48). There were no gender differences on any of the other measures, or in mean age and BMI.
Interscale Correlations and Multiple Regressions
Cosmetic use among women was significantly and positively correlated with other-objectification, AWS, HTWS, and media exposure as well as negatively correlated with age (see Table 3). Among men, the belief that women should use cosmetics was significantly and positively correlated with objectification, HTWS, ASI-H, and media exposure as well as negatively correlated with BMI.
We next conducted multiple linear regressions with cosmetic use as the dependent variable and all other measures as predictor variables entered simultaneously. The regression for women's cosmetic use was significant, F(8, 157) = 7.80, p < .001, Adj. R 2 = .27, such that a heightened tendency to objectify others (β = .29, t = 4.20, p < .001), younger age (β = −26, t = −3.10, p = .002), stronger sexist attitudes (AWS: β = .24, t = 2.92, p = .007; ASI-H: β = .22, t = 2.70, p = .010), and more media exposure (β = .18, t = 2.24, p = .028) predicted greater cosmetic use. For men, the regression was also significant, F(8, 265) = 3.87, p < .001, Adj. R 2 = .09. Of the variables entered into the model, higher media exposure (β = .24, t = 3.72, p < .001), a greater tendency to objectify others (β = .16, t = 2.48, p = .016), and elevated sexism on the AWS (β = .14, t = 2.06, p = .040) predicted a stronger belief that women should use cosmetic products.
Conclusions
The results of Study 3 revealed that there were a number of significant predictors of cosmetic use among women and of men's perceived desire for women to use cosmetics. Overall, and as predicted, other-objectification appeared to be an important predictor of endorsed cosmetic use. This result lends credence to the view that cosmetic use is associated with the sexualisation of women, possibly because it focuses attention on physical attributes to the detriment of nonphysical characteristics such as ability or competency. In other words, cosmetic use may focus attention away from women's abilities and reinforce notions of women as decorative objects that remain in passive and subordinate roles.
Our results also revealed that media exposure predicted women's cosmetic use and men's perceived need for women to use cosmetics. It seems likely, as argued previously, that information about the (perceived) necessity of cosmetic products is constructed by what Bartky (1990) has termed the “fashion-beauty complex” (corporate interests involved in the fashion and beauty industries) promulgated through the media. By continually displaying images of purported ideals of women's bodies, the fashion-beauty complex ensures that women feel their bodies to be deficient, requiring “heroic measures” to rectify those deficiencies (Bartky, 1990, p. 39). Part of those heroic measures, of course, is the use of cosmetic products sold by the fashion-beauty complex.
The results of Study 3 also highlight the relationships between sexist beliefs, particularly hostile sexism, and endorsement of cosmetic use. It seems that individuals who hold blatantly sexist attitudes or hostile attitudes toward women feel a greater need to use cosmetics or have women use cosmetics. Interestingly, and in contrast to the results of previous work (Forbes et al., 2006; Franzoi, 2001), we did not find any significant correlations between benevolent sexism and endorsement of cosmetic use. It should be noted, however, that we used a different (and improved) measure of cosmetic use, which means that our results may not be directly comparable to past work.
One specific limitation of Study 3 is that we asked women to answer questions about cosmetic use for themselves, whereas men answered questions about what women should do. In hindsight, it may have been more appropriate to ask both women and men to answer questions about what women should do in relation to cosmetic use. Given this discrepancy, our interpretations of gender differences in relation to cosmetic use may be problematic, because women may be similar to men, for example, in not believing that women should use cosmetics, but still comply themselves.
SUMMARY AND CONCLUDING DISCUSSION
Overall, the results of our three studies provide strong support for the BIO hypothesis (Forbes et al., 2007). First, we showed that the endorsement of various beauty ideals and practices was associated with individual differences in sexist beliefs and the tendency to objectify others. Specifically, we found that preferring the thin ideal was associated with stronger sexism (particularly on the AWS) and (among women) heightened objectification of others (Study 1). Similarly, a male-taller norm in romantic relationships was predicted by benevolent sexism and sexist attitudes as measured on the AWS for both women and men (Study 2). Finally, endorsement of cosmetic use for women was associated with greater objectification of others and higher sexism scores as measured on the AWS (Study 3).
In itself, these results are noteworthy because they add to the body of literature attempting to contextualize beauty ideals within corporeal, gendered, and social worlds as well as to extend the feminist critique of Western beauty standards (e.g., Bartky, 1990; Bordo, 1993; Jeffreys, 2005; Smolak & Murnen, 2007). In addition, our results support the generalisability of the BIO hypothesis to a range of beauty ideals and practices. As noted by Forbes et al. (2007), “individual differences in sexism and hostility toward women should be related to all beauty standards and practices and not be restricted to isolated concepts such as the widely studied thin body ideal” (p. 271).
Our results may have important implications for negative body image and health-related behaviors among Western women and men. In particular, it might be argued on the basis of our results that programmes aimed at reducing or eliminating sexist attitudes (or, by contrast, promoting more egalitarian attitudes) would be associated with healthier beauty ideals and practices. That is, individuals who hold more gender-equal attitudes may be more likely to challenge beauty ideals, and their beliefs may act as a buffer against the negative effects of beauty ideals. Indeed, the available evidence does suggest that profeminist attitudes are associated with more inclusive beauty ideals (e.g., Murnen & Smolak, 2009; Swami et al., 2008b).
Of course, our study is not without its limitations. First, although we recruited community samples for all three studies, the opportunistic sampling method means that we cannot be certain that our participants were representative of the wider community. In particular, it appears to be the case that certain demographics were overrepresented in our samples, such as individuals with higher educational qualifications. This oversampling necessarily limits the generalisability of our results, although we do not have much reason to think that our convenience samples would respond differently from a random sample drawn from the same population. Moreover, we would stress that our sample overcomes the reliance on student populations that is prevalent in previous work and, given the ubiquity of beauty ideals and practices (Jeffreys, 2005), it seems likely that our results capture meaningful and reliable associations that can be replicated in most Western settings.
In a similar vein, it is important to note that our samples included a wide range of ethnic groups, and it is possible that there may have been intergroup differences both in the endorsement of Western beauty ideals as well as its predictors (e.g., Swami, Airs, Chouhan, Padilla Leon, & Towell, 2009). Second, it is important to note that our study was correlational by design. Although the direction of effects may seem clear (given that sexist beliefs are more distal than beauty ideals), our results should be interpreted with the caution due any correlational study. Future work could overcome this limitation by manipulating the endorsement of sexist beliefs and examining its impact on beauty ideals and practices.
Were our study to be repeated, there are a number of other ways in which it could be improved. First, it may be useful to include a wider array of measures that tap different components of sexist attitudes, or conversely, egalitarian or profeminist attitudes (e.g., see Murnen & Smolak, 2009). The inclusion of additional variables, such as social conformity and better measures of media exposure (or, more pertinently, acceptance of media messages of beauty ideals) may also allow future researchers to obtain a more comprehensive picture of the factors shaping beauty ideals and practices. Finally, a range of other beauty ideals and practices (e.g., acceptance of cosmetic surgery; see Slevec & Tiggemann, 2010) could be meaningfully examined under the umbrella of the BIO hypothesis.
In conclusion, our results support the BIO hypothesis in showing that measures of sexism are related to the endorsement of various beauty ideals and practices. In addition, we showed that other-objectification (a trait related to sexism) was associated with the same endorsement (albeit not to the same extent as sexism) and that media exposure predicted beliefs about cosmetic use—but not the thin ideal or the sexual dimorphism ratio. These results have important implications for the extant literature on attractiveness ideals and practices. Although our results do not deny a role for distal evolved psychologies, they do emphasize the need for researchers to take into account an array of proximate factors that shape beauty ideals. That is, it is important to acknowledge the shaping of beauty ideals and practices by the social processes that colour the lived experiences of women and men in contemporary societies.
