You are on page 1of 38

This article was downloaded by: [MPI Fuer Ethnologische Forsc] On: 20 February 2013, At: 09:12 Publisher:

Routledge Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Biodemography and Social Biology


Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/hsbi20

Handsome Wants as Handsome Does: Physical Attractiveness and Gender Differences in Revealed Sexual Preferences
Elizabeth Aura McClintock
a a

Department of Sociology, University of Notre Dame, Notre Dame, Indiana Version of record first published: 24 Oct 2011.

To cite this article: Elizabeth Aura McClintock (2011): Handsome Wants as Handsome Does: Physical Attractiveness and Gender Differences in Revealed Sexual Preferences, Biodemography and Social Biology, 57:2, 221-257 To link to this article: http://dx.doi.org/10.1080/19485565.2011.615172

PLEASE SCROLL DOWN FOR ARTICLE Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material.

Biodemography and Social Biology, 57:221257, 2011 Copyright Society for the Study of Social Biology ISSN: 1948-5565 print / 1948-5573 online DOI: 10.1080/19485565.2011.615172

Handsome Wants as Handsome Does: Physical Attractiveness and Gender Differences in Revealed Sexual Preferences
ELIZABETH AURA MCCLINTOCK
Department of Sociology, University of Notre Dame, Notre Dame, Indiana

Downloaded by [MPI Fuer Ethnologische Forsc] at 09:12 20 February 2013

In this article I evaluate the effect of physical attractiveness on young adults sexual and romantic outcomes to reveal gender differences in acted preferences. Using data from the National Longitudinal Study of Adolescent Health (Add Health), a probability sample of young adults (n = 14,276), I investigate gender differences in desired sexual partner accumulation, relationship status, and timing of sexual intercourse. I nd gender differences in sexual and romantic strategies consistent with those predicted by the double standard of sexuality and evolutionary theory. Specically, compared to men, women pursue more committed relationships, fewer sexual partners, and delayed sexual intercourse.

Introduction
This paper expands the literature on gender differences in pursued sexual and romantic outcomes (such as sexual partner accumulation, relationship status, and timing of sexual intercourse) and adds to the growing body of research that considers both social structural and evolutionary explanations for human behavior (for other examples, see: Howard, Blumstein, and Schwartz 1987; Kenrick et al. 1993; Dietz, Burns, and Buttel 1990; Eagly 1997; DeSteno and Salovey 1996). It contrasts a social structural perspective that posits womens sexuality is constrained by the double standard of sexuality with an evolutionary perspective that posits genetically determined gender differences in sexual strategies. Rather than using a direct measure of sexual and romantic goals, this paper uses reported outcomes to infer goals. The advantage to this approach is that individuals may not know what they most value in partners and relationships and/or may answer questions about preferences and priorities in accordance with gender-stereotyped sexual and
This research uses data from Add Health, a program project designed by J. Richard Udry, Peter S. Bearman, and Kathleen Mullan Harris and funded by Grant P01-HD31921 from the National Institute of Child Health and Human Development, with cooperative funding from 17 other agencies. Special acknowledgment is due Ronald R. Rindfuss and Barbara Entwisle for assistance in the original design. Persons interested in obtaining data les from Add Health should contact Add Health, Carolina Population Center, 123 W. Franklin Street, Chapel Hill, NC 27516-2524 (addhealth@unc.edu). Thanks are due to Paula England and Michael Rosenfeld and to two anonymous Biodemography and Social Biology reviewers for their helpful comments on this article. A previous version was presented at the American Sociological Association Annual Meeting (2010). Address correspondence to Elizabeth Aura McClintock, Department of Sociology, University of Notre Dame, 810 Flanner Hall, Notre Dame, IN 46556. E-mail: emcclint@nd.edu

221

222

E. A. McClintock

Downloaded by [MPI Fuer Ethnologische Forsc] at 09:12 20 February 2013

romantic scripts (Sprecher 1989a; Eastwick and Finkel 2008; Li, Bailey, Kenrick, and Linsenmeier 2002). Indeed, for men there is evidence that the sexual behavior they expect of themselves is not consistent with their actual experiences (Cohen and Shotland 1996). Thus, to the degree that reports of sexual and romantic behavior are reliable, measures of actual behavior may provide less-biased insight into objectives. In addition, actual sexual and romantic outcomes are interesting because they reect compromised rather than ideal choices: Partnership formation is inherently a two-sided negotiation, and individuals ideal preferences may be unattainable, forcing them to compromise and enter sub-optimal unions (or to remain single). This paper argues that more physically attractive individuals are more able to attain desired outcomes. Therefore, when sexual and romantic outcomes differ by physical attractiveness, outcomes for the most physically attractive women and men should be closer to ideal preferences while outcomes for the least physically attractive women and men should represent compromised choices. When the effect of physical attractiveness differs for men and women, this indicates gender differences in sexual and romantic strategies.

Background
Prettiness and the Partner Market Couple formation is often conceptualized as a competitive, two-sided matching process (Choo and Siow 2006; e.g.: Burdett and Coles 1999; Burdett and Coles 2001).1 In this market-based model, individuals implicitly trade their assets for those of a mate, trying to nd the most desirable partner and most rewarding relationship that they can get given their own assets. This market metaphor has primarily been applied to marriage markets and focused on the exchange of income or status for other desired resources such as physical attractiveness (for example, see Burdett and Coles 2001; Elder 1969; Stevens, Owens, and Schaefer 1990; Murstein 1972; Coles and Francesconi 2007; Taylor and Glenn 1976), but it is easily extended to explain partner selection in the young adult premarital dating market. Just as good looks may be exchanged for status and nancial resources, attractiveness may also be traded for control over the degree of commitment and progression of sexual activity. Indeed, prior research suggests that physical attractiveness might impact the distribution of power within relationships. Physically attractive people are perceived and treated more favorably by both strangers and close acquaintances (Langlois et al. 2000; Dion and Berscheid 1972; Mulford, Orbel, Shatto, and Stockard 1998). Physical attractiveness has been shown to matter in marriage markets, and both absolute and relative physical attractiveness continue to matter in established relationships, particularly with respect to mens interactions with their female partners (McNulty, Neff, and Karney 2008; Taylor and Glenn 1976; Byers and Lewis 1988; Coles and Francesconi 2007; Elder 1969; OSullivan 1995;
1 Studies of romantic and sexual relationships often draw on social exchange and equity theory from social psychology (e.g., Sprecher 1998; Udry 1981; Van de Rijt and Macy 2006): This provides an alternative to the market model. Social exchange theory (like the market model) addresses how individuals relative endowments of desired traits inuence partner formation, negotiation within relationships, and relationship dissolution. Which traits are desired in romantic and sexual partners is exogenous to social exchange theory (and to the market model): In the context of this paper, the social structural and evolutionary perspectives address what individuals want in relationships and the market or social exchange model explain how these desires are translated into actual (often compromised) outcomes. Although market and social exchange theories are fundamentally different, in this paper both theories yield the same hypotheses. Therefore, for simplicity, I address only the market model in the main text.

Gender Differences in Revealed Sexual Preferences

223

Margolin and White 1987). Given that physical attractiveness inuences interactions with strangers and spouses, it likely inuences interactions between dating partners. Gender Differences in the Valuation of Physical Attractiveness Compared to women, men are generally assumed to place more importance on their partners physical attractiveness, and men are thought to value appearance very highly (e.g., Elder 1969; Taylor and Glenn 1976). When asked to rank the traits most important in a potential mate, results generally indicate that men value physical attractiveness more highly than do women (Buss and Barnes 1986; Sprecher, Sullivan, and Hateld 1994; Townsend and Levy 1990; Furnham 2009; Hansen 1977; Coombs and Kenkel 1966; Kenrick, Sadalla, Groth, and Trost 1990; Fletcher et al. 2004; Sprecher 1989a; Nevid 1984; Howard et al. 1987). However, this gender difference may be disappearing (Buss et al. 2001), and some recent studies of reported partner preferences among adolescents and young adults have found no gender differences (Regan and Joshi 2003; Regan 1998). In addition, many studies indicate that neither men nor women attach much importance to physical attractiveness. For example, in a study asking individuals to rank the importance of 76 traits, none of the measures of appearance made the top ten (Buss and Barnes 1986), and this low ranking of physical attractiveness is consistent with other similar studies (Howard et al. 1987; Furnham 2009; Nevid 1984; Hansen 1977). In contrast, in experimental studies designed to measure individuals acted preferences (as opposed to stated preferences), physical attractiveness is highly valued by both genders. For example, when undergraduates were given varying budgets to purchase an assortment of desirable characteristics and design a mate, both women and men prioritized physical attractiveness, spending enough to obtain an acceptable level of good looks, and then allocating their remaining funds among secondary characteristics (Li and Kenrick 2006). Similarly, in another experimental study in which undergraduates rated their attraction to stimulus persons, both men and women were more inuenced by physical attractiveness than by high earning potential or high expressivity (Sprecher 1989a). Still, when these same undergraduates were asked to rank the importance of these three traits, women valued mens earning potential and expressiveness over physical attractiveness, implying an inconsistency between their stated preferences and acted preferences. The author suggests that women and men may not be aware of the traits that attract them to a potential partner and instead use culturally provided implicit causal theories to explain their attraction responses. It is also plausible that the undervaluation of physical attractiveness in reported preferences is a result of social desirability bias: Respondents may feel that attaching too much importance to appearance is a sign that one is shallow or supercial. This proposed disjoint between the preferences individuals believe they hold and the preferences revealed by their actions is indicated in a recent study of speed daters. In rating the importance of physical attractiveness, earning potential, and personality as criteria for partners, men attached greater importance (than women) to physical attractiveness, and women attached greater importance (than men) to earning potentialbut no gender differences were evident in the effect of these three criteria on mens and womens interest in pursuing relationships with their speed-dating partners (Eastwick and Finkel 2008). In addition, individuals stated preference (the degree of importance they claimed to attach to physical attractiveness, earning potential, and personality) did not predict their actual interest in their partners. Similarly, a different speed-dating study found that the strongest predictor of interest for both genders was the partners physical attractiveness

Downloaded by [MPI Fuer Ethnologische Forsc] at 09:12 20 February 2013

224

E. A. McClintock

Downloaded by [MPI Fuer Ethnologische Forsc] at 09:12 20 February 2013

(Luo and Zhang 2009) although a third speed-dating study suggests that men value physical attractiveness somewhat more than women (Fisman, Lyengar, Kamenica, and Simonson 2006). Most of the samples used in studies of the valuation of physical attractiveness are unrepresentative and nonrandom. Many are convenience samples of undergraduates (Buss and Barnes 1986; Buss et al. 2001; Coombs and Kenkel 1966; Eastwick and Finkel 2008; Fletcher et al. 2004; Kenrick et al. 1990; Li and Kenrick 2006; Nevid 1984; Regan 1998; Sprecher 1989a; Townsend and Levy 1990). Others are nonrandom samples of high school students (Hansen 1977; Regan and Joshi 2003), graduate and professional school students (Fisman et al. 2006), or other more diverse samples of adults (Buss and Barnes 1986; Furnham 2009; the sampling method is unclear in Howard et al. 1987). Only one study used a large, nationally representative probability sample (Sprecher et al. 1994). Nonetheless, the consistency of ndings over a wide range of studies makes the conclusions reasonably credible: Both women and men value physical attractiveness highly in actual choices but value it less when reporting their preferences. Constrained Choice and Revealed Preference The inconsistency between stated and acted preferences may result from allowing subjects to imagine unrealistically perfect partners. When ranking the importance of various traits, respondents might unconsciously assume a minimally acceptable partner and then decide how to rank preferences to make this already adequate person as desirable as possible. Their underlying priorities become apparent when they are confronted with a partner explicitly possessing the minimal level of a given characteristic (Li and Kenrick 2006; Li et al. 2002). For example, womens valuation of physical attractiveness becomes evident only when they are confronted with the option of selecting an explicitly unattractive mate (Li and Kenrick 2006). These results suggest that individuals may not know what they value in partners and relationships but instead answer in accordance with stereotyped, gender-specic scripts. Alternatively, social desirability bias may be stronger when the consequences of undervaluing physical attractiveness are less tangible. In either case, behavioral indications (actual choices) are more accurate indictors of preferences than reported values and priorities. In fact, although there is a strong relationship between sexual attitudes, behavioral intentions, and behaviors (Gillmore et al. 2002), actual behavior often differs systematically from reported values and preferences. For example, the cohort of women that came of age during the sexual revolution (women born 1937 to 1956) have more liberal attitudes but less liberal behavior than women who came of age in the following era of sexual ambivalence (women born 1957 to 1973; (Haavio-Mannila, Roos, and Kontula 1996). In other words, the ambivalent cohort engages in more sexually liberal behavior, despite approving of this behavior less, suggesting that their stated values may not reect the choices they actually make. Similarly, attitudes toward premarital sexual intercourse do not always align with virginity status, and the direction of disjoint is culturally variant: In Denmark, more adolescents report approval of premarital sexual intercourse than are sexually experienced whereas in the United States, adolescents are more likely to have had sexual intercourse than to approve of this behavior (Christensen and Carpenter 1962). Not only may acted preferences be a better indicator of true preferences than stated preferences, it is acted preferences that most inuence individuals life outcomes. Additionally, partner formation is a two-sided negotiation, so individuals choices will always be constrained, and acted preferences reveal the compromises individuals make

Gender Differences in Revealed Sexual Preferences

225

in response to these constraints. If physical attractiveness is desired in romantic and sexual partners, the physically attractive will have less constrained choices, and differences between outcomes for more and less physically attractive individuals will reect the difference between ideal and compromised outcomes. Are Sexual and Relational Goals Gendered? Women and men may both prefer physically attractive partners, but they may have different preferences for the nature and number of their partnerships. Gendered sexual and romantic goals are predicted by evolutionary theorists and by research demonstrating an enduring, socially constructed double standard of sexuality. Both perspectives predict that compared to women, men will be more willing to engage in uncommitted, short-term sexual encounters with looser criteria for partners. However, whereas the social structural perspective indicates that women will favor committed, longer-term relationships, the evolutionary perspective suggests that women will also pursue uncommitted, short-term sexual encountersbut will be more selective than men in picking casual sexual partners. Evolutionarily Determined Mating Strategies? Theories of sexual selection argue that traits that increase reproductive success will be preferentially selected over time (Buss 1998; Symons 2005). These traits include psychological adaptations that lead individuals to engage in an optimal mating strategy given their own mate value and the demands of their ancestral environment (Symons 2005; Schmitt 2005a). Evolutionary theories argue that because womens minimal investment in children is larger than mens (given that gestation is 9 months) and because women face greater biological limitations on the maximum number of children they are able to produce (at most one child every year, and none after menopause), women and men have adapted to solve different reproductive challenges (those that faced ancestral women and men). Importantly, hypothesized evolutionary adaptations need not align with conscious fertility intentions: The use of contraception does not preclude evolutionarily inuenced sexual motivations, and evolutionary explanations are relevant to sexual behavior among individuals that intend to delay or forgo fertility. Because these evolutionary adaptations are unconscious, one might be inuenced by them when picking a partner without consciously contemplating the benets of procreating with that partner. According to evolutionary theorists, men may adopt a short-term mating strategy in which they attempt to mate with many women without committing resources to support their potential offspring or adopt a long-term mating strategy in which they offer support in raising offspring in exchange for exclusive sexual access (Kruger, Fisher, and Jobling 2003; Schmitt 2005a). Interestingly, some evolutionary psychologists argue that physically attractive men are genetically predisposed to a short-term mating strategy and that these men are particularly desirable to women as short-term partners (Gangestad and Simpson 2000; Sugiyama 2005; Kruger 2006; Kruger et al. 2003). Other authors suggest that all men engage in the same competition for sexual access, with the same level of inclination toward short- and long-term partnering strategies (Daly 1999). Women may also have adapted to pursue a mixture of short- and long-term mating strategies (Daly and Wilson 2005; Hrdy 1981). Because womens minimal investment in offspring is greater than that of men, women are thought to be more selective, picking the men that provide the most benets to the offspring by offering good genes (presumed to be womens motivation for short-term mating and to be identied by physical attractiveness

Downloaded by [MPI Fuer Ethnologische Forsc] at 09:12 20 February 2013

226

E. A. McClintock

Downloaded by [MPI Fuer Ethnologische Forsc] at 09:12 20 February 2013

and bravado) or material support in raising the offspring (the criterion for long-term mates; Gangestad and Simpson 2000; Schmitt 2005a; Sugiyama 2005; Daly 1999; Wiederman 1993; Gangestad and Thornhill 1997). Contrary to the stereotype of women as sexually restrictive, ancestral females may have benetted from cuckolding their primary partners in favor of a male with better genes (Gangestad and Simpson 2000; Hrdy 1981; Gangestad and Thornhill 1997). Females might also have benetted from mating promiscuously to elicit parental support (or at least benign tolerance of the child) from multiple men, each of whom has some probability of being the biologic father and therefore has some interest in the well-being of the resulting child (Hrdy 1981; Schmitt 2005a; Daly and Wilson 2005). For women to benet from a short-term strategy or from cuckolding their primary mate, women must identify those men with desirable genes. Indeed, women favor men with certain genetic characteristics (Gangestad and Simpson 2000; Hrdy 1999; Sugiyama 2005; Wedekind and Furi 1997), and womens increased emphasis on appearance in shortterm partners may result from their attraction to men with desirable genes (Kruger et al. 2003; Kruger 2006). For women to succeed by adopting a long-term mating approach, they must identify reliable and willing providers: This might explain womens greater emphasis on mens earnings power in picking long-term mates (Wiederman 1993), although modern indicators of bread-winning certainly differ substantially from those displayed by ancestral men. For either a short- or long-term approach to be successful for men, they must select a fertile partner, and men are thought to use physical attractiveness as a gauge of fertility and health (Buss 1998; Sugiyama 2005; Wiederman 1993). In fact, there is some evidence that men perceive physically attractive women to be healthier and more likely to have children (Mathes et al. 2005). This explanation, however, assumes a gender differences in the valuation of physical attractiveness for short- and long-term mates that may exist only in stated (versus acted) preferences. Sexual Permissiveness and the Sexual Double Standard Gender differences in sexual strategies might also result from socially structured differences in the costs and benets of sexual and romantic activities. Researchers have proposed a sexual double standard that penalizes women and forgives (or rewards) men for accumulating sexual experience (Sprecher 1989b; Reiss 1967; Crawford and Popp 2003). In a review of articles pertaining to sexual double standards published between 1981 and 2001, Crawford and Popp (2003) conclude that the sexual double standard still exists and likely inuences mens and womens sexual behavior. In one example of the enduring double standard, women are evaluated more negatively than men for the same level of sexual promiscuity (Sprecher, McKinney, and Orbuch 1987; OSullivan 1995). The sexual double standard also affects womens self-evaluations. Regardless of the relational context, women report more feelings of guilt at rst intercourse than men (Sprecher, Barbee, and Schwartz 1995). In adolescence, girls perceive higher relational and personal costs and lower benets to sexual intercourse (Meier 2003), and in high school they are aware that being seen as overly permissive is damaging to their social standing (Eyre, Hoffman, and Millstein 1998). Consistent with a social structural perspective, gendered sexual norms vary over time and place and respond to changing social conditions (Baumeister and Twenge 2002; Crawford and Popp 2003). Womens and mens relative valuation of physical attractiveness and nancial potential have become increasingly similar as womens labor force participation increases, and this trend is stronger in more gender-egalitarian regions of the United States (Buss et al. 2001; Sweeney and Cancian 2004). In cross-national data,

Gender Differences in Revealed Sexual Preferences

227

womens greater economic freedom, female-centered (matrilinear and matrilocal) social structures, and less patriarchal societies are all associated with less restricted female sexuality (Schmitt 2005b; Reiss 1986; Moore and Cassidy 2007). Age, education, feminist ideology, and political orientation are also important, modifying the effect of gender on the valuation of partners physical attractiveness and status (Doosje, Rojahn, and Fischer 1999; Furnham 2009; Bay-Chen and Zucker 2007; South 1991). That gendered sexual norms are inuenced by social context demonstrates that human sexuality is adaptable and at least in part socially constructed, not that evolutionary adaptations are necessarily irrelevant. Still, such substantial variation across social contexts casts some doubt on the modern importance of hypothesized evolutionary adaptations, if not on their very existence.

Downloaded by [MPI Fuer Ethnologische Forsc] at 09:12 20 February 2013

Physical Attractiveness and Gendered Sexual and Relational Goals Both the theory that human mating preferences are inuenced by genetically determined mating strategies and the theory that womens sexuality is constrained by an enduring double standard of sexuality predict differences in the romantic and sexual experiences that women and men pursue. Prior research also indicates that physical attractiveness is a valuable, marketable resource for both genders. Taken together, this implies that physical attractiveness will differently impact sexual and romantic outcomes (such as sexual partner accumulation) for women and men: If physically attractive people are more able to get what they want in romantic and sexual markets and women and men pursue different types of relationships, physically attractive women and men will have different outcomes. Existing studies nd differences in sexual and romantic experience for more and less physically attractive women and men. Presumably because they are preferred as partners, the physically attractive have more romantic experience (Langlois et al. 2000; Berscheid and Dion 1971; Halpern et al. 1999; Wiederman and Hurst 1998). However, though research consistently indicates that physical attractiveness and thinness increase womens romantic experience, ndings are mixed in regard to their sexual experience (Wiederman and Hurst 1998; Kaneshiro et al. 2008). For men, there is little ambiguity: Among undergraduate men, being good looking is associated with accumulating a greater number of sexual partners (Bogaert and Fisher 1995; Gillen, Lefkowitz, and Shearer 2006). In addition, a study of undergraduates found that physically attractive men engage in more risky sexual behavior, relative to other men, whereas physically attractive women engage in less risky sexual behavior, relative to other women et al. 2006). Importantly, in this same study, attitudes toward risky sexual behavior do not depend on physical attractiveness, indicating that physically attractive men and women do not behave differently because they have different values than their less attractive peers. These ndings are suggestive, but existing research has not examined gender differences in the effect of physical attractiveness on negotiated outcomes within relationships, such as the degree of commitment and the timing of sexual activity. Additionally, the majority of studies use unrepresentative samples, usually convenience samples of undergraduate students (e.g., Bogaert and Fisher 1995; Berscheid and Dion 1971; Gillen et al. 2006; Wiederman and Hurst 1998; Halpern et al. 1999). This paper improves upon prior papers by employing a large, nationally representative probability sample of young adults and by examining a wider variety of sexual and romantic outcomes. This paper also advances prior work by attempting to mediate between social structural and evolutionary explanations. As evidenced in the hypotheses section, the two theories generally make similar predictions, so it is difcult to test between them directly. However, when socially constructed standards of physical attractiveness are not aligned

228

E. A. McClintock

Downloaded by [MPI Fuer Ethnologische Forsc] at 09:12 20 February 2013

with characteristics indicating optimal fertility, it may be possible to mediate between them indirectly. In particular, body size is an important dimension of attractiveness, especially for women (Brown and Cash 2001; Sugiyama 2005), and differences in womens body sizes are associated with known differences in fertility. The current culturally ideal female body shape is somewhat thin to very thin (Wiseman et al. 1992; Sugiyama 2005; Fallon and Rozin 1985; Cohn and Adler 1992) despite the reduced fecundity associated with being slender (Grodstein, Goldman, and Cramer 1994; Rich-Edwards, Spiegelman, Garland, Hertzmark, Hunter, Colditz, Willett, Wand, and Manson 2002). In other words, contemporary American society favors a female body type associated with sub-optimal fertility. If outcomes for underweight women resemble outcomes for very physically attractive women, this provides tentative evidence that social forces are most relevant in determining what is valued in partner markets.

Hypotheses
Todays version of the double standard of sexuality does not penalize women for romantic involvement or for having sexual intercourse in committed relationships, so it is unlikely to entirely negate the effect of greater opportunity on physically attractive womens sexual and romantic experience. Physically attractive men will also have more sexual and romantic opportunities. Thus, my rst pair of hypotheses is (H1a) More physically attractive women and men will be more likely to have had a romantic relationship than less attractive women and men. (H1b) More physically attractive women and men will be more likely to have had sexual intercourse than less attractive women and men. These same hypotheses are predicted by the evolutionary perspective, which also holds that men and women value physical attractiveness in partners. Both the evolutionary and social structural theories argue that men are more eager than women to engage in sexual intercourse, particularly casual (non-relational) sexual intercourse, because the costs (whether reproductive or social) are lower for men. If women prefer physically attractive men as partners, these men will have more sexual opportunities. Thus, my second hypothesis is (H2) More physically attractive men will have more sexual partners than less attractive men. The evolutionary perspective gives particularly strong reason to expect physically attractive men to have a greater number of sexual partners because women are thought to select casual partners on the basis of physical attractiveness (the signal for desirable genes), and physically attractive men may be especially inclined to pursue multiple casual partners. The net effect of physical attractiveness on the accumulation of sexual partners is unclear for women. Both the social structural and evolutionary models suggest that men are less discerning regarding the appearance of casual partners; still, as long as men prefer pretty partners, physical attractiveness will increase womens opportunities for casual sexual intercourse. However, it is not clear that physically attractive women would take advantage of these casual sexual opportunities: They might instead form long-term monogamous relationships that would limit their partner accumulation (relative to a short-term mating strategy). Indeed, under both models, physically attractive women will also have

Gender Differences in Revealed Sexual Preferences

229

more opportunities for committed, long-term sexual partnerships. As a result of these conicting forces, neither model makes a clear prediction regarding the effect of physical attractiveness on womens sexual partner accumulation. Another outcome that might be differently inuenced by physical attractiveness for women and men is the timing of sexual intercourse relative to meeting a new partner. Evolutionary theory does not indicate gender differences in the desired timing of sexual intercourse, but the sexual double standard dictates that women delay intercourse until some degree of commitment is reached. According to this perspective, men desire sexual activity sooner in the relationship than women. If physically attractive individuals are more able to get what they want in relationships, then

Downloaded by [MPI Fuer Ethnologische Forsc] at 09:12 20 February 2013

(H3a) More physically attractive men are more likely to have sexual intercourse soon after meeting a new partner, compared to less attractive men. (H3b) More physically attractive women are less likely to have sexual intercourse soon after meeting a new partner, compared to less attractive women. The status of the relationshipwhether it is casual and uncommitted or is exclusive and committedmight also differ for physically attractive women and men. According to the sexual double standard, men are more open to casual sexual encounters than women. In fact, some authors portray young adult men as being hostile to romantic commitment and favoring impersonal, casual sexual encounters (Cross 2008; Kimmel 2008). The evolutionary model does not necessarily indicate that all men will have a greater preference for casual sexual intercourse (than women), but some authors argue that physically attractive men will have an innate preference for casual sexual relationships. In addition, being handsome is considered an indicator of having good genes, and obtaining good genes is womens motivation for short-term mating, so physically attractive men will have more opportunities for casual sexual intercourse. Thus, the evolutionary model suggests that (H4a) More physically attractive men will be more likely to describe their relationships as casual sexual relationships and less likely to describe their relationships as exclusive relationships, compared to less attractive men. If the sexual double standard is thought to be reversed for young adult men so that they avoid romantic commitment and seek casual sexual encounters, the social structural perspective makes the same prediction. Under the sexual double standard, women will avoid casual sexual relationships, and more physically attractive women are likely to have greater ability to negotiate the rules of their sexual involvement. Thus, the social structural model predicts that (H4b) More physically attractive women will be more likely to describe their relationships as exclusive relationships and less likely to describe their relationships as casual sexual relationships, compared to less attractive women. The evolutionary model indicates that more attractive women will have more opportunities for committed relationships because men select long-term partners on the basis of beauty. If women are thought to favor a long-term mating strategy the evolutionary model makes the same prediction as the social structural model. However, some formulations of

230

E. A. McClintock

the evolutionary model imply that women are ambivalent between committed and casual unions or might favor casual unions (Hrdy 1981).

Method
Data I use data from the rst, second, and third waves of the National Longitudinal Study of Adolescent Health (Add Health), a nationally representative longitudinal survey of adolescents in grades 7 to 12 at the time of the initial interview (Bearman, Jones, and Udry 2008; Chantala 2006). From a sample of 134 public, private, and parochial schools, 27,000 adolescents were selected for extensive in-home interviews. Approximately 21,000 of these students completed the interviews. Students used an audio-computer assisted self-interview device (audio-CASI) for several sensitive sections of the interview. Prior research suggests that survey answers regarding sexuality are most honest when respondents are condent that their answers are condential (Alexander and Fisher 2003), so use of the audio-CASI may increase the reliability of these data. The Wave I interviews were conducted during the 19941995 academic year, and the Wave II interviews were conducted about a year later. The Wave III interviews were conducted in 20012002. I use outcomes measured in the third wave of data collection, when respondents are on average about 21.5 years old, and lag measures of most independent variables to the Wave I or Wave II interview. Parental socioeconomic status is measured in Wave I to take advantage of the parent interview, and race is also taken from the Wave I survey (unless missing, in which case I use measures from a subsequent wave). For other independent variables, I use Wave II measures for all respondents except those in twelfth grade at the Wave I interview, for whom I use Wave I measures. There are two reasons for this approach. First, I want a post-pubertal measure of physical attractiveness, so I prefer the Wave II measure of physical attractiveness over the Wave I measure for the younger respondents. Second, this approach maximizes the sample size. Respondents younger than age 15 in the Wave I interview (about 5,500 respondents) were not asked questions about their motivations for sexual activity or abstinence: I use these questions to construct indices of the social costs and personal benets of sexual activity, important control variables. At Wave II, all respondents were asked about motivations for sexual activity, regardless of age. However, respondents in twelfth grade at the Wave I survey (about 3,300 respondents) were not interviewed in Wave II (but were interviewed in Wave III). Using Wave II measures for the younger respondents and the Wave I measures for the Wave I seniors allows me to retain the full sample (using only Wave II would exclude about 3,300 seniors and using only Wave I would exclude about 5,500 respondents younger than age 15). The resulting measures are taken during high school (grades 9 to 12) for nearly all respondents. For consistency, this approach is also used to construct self-esteem. I conduct part of my analysis at the individual level and part at the relationship level, using data from almost a full history of respondents lifetime sexual and romantic relationships. After dropping cases with missing data on the dependent variables, more than 14,000 respondents remain for the individual-level analysis. Detailed information is collected on all sexual relationships and on the two relationships deemed most important, whether or not they involved sexual intercourse. After removing relationships that are missing data on the dependent variables, almost 30,000 relationships remain for analysis. I drop cases missing data on a given dependent variable only when analyzing that variable, so there

Downloaded by [MPI Fuer Ethnologische Forsc] at 09:12 20 February 2013

Gender Differences in Revealed Sexual Preferences

231

are a slightly different number of observations for each dependent variable. I also exclude same-sex relationships. Measures Means and standard deviations of explanatory measures are seen in Table 1. Descriptive statistics on individual-level romantic and sexual outcomes are seen in Table 2 and descriptive statistics on relationship-level romantic and sexual outcomes are seen in Table 3. Individual-Level Sexual and Romantic Outcomes

Downloaded by [MPI Fuer Ethnologische Forsc] at 09:12 20 February 2013

Romantic Experience. This is an indicator variable for whether the respondent has had any heterosexual romantic experience. Respondents that report at least one romantic relationship are coded as romantically experienced. Sexual Experience. This is an indicator variable for whether the respondent has had sexual intercourse. Accumulation of Sexual Partners. The number of partners is a self-reported count of the total number of sexual partners the respondent has ever had at the time of the interview (if non-zero), capped at 50. Individuals who report no sexual partners are excluded. Relationship-Level Sexual and Romantic Outcomes Relationship Status. This variable classies relationships by degree of implied commitment as only having sex, dating once in awhile, dating frequently, or dating exclusively. These categories may be problematic because multiple dimensions of involvement are measured on a single scale: It is not clear that the variable is truly ordinal or even that the categories are mutually exclusive. For example, individuals in a purely sexual relationship might be exclusive and might see each other frequently. I address this concern in the Analysis section of the paper. Length of Acquaintance Prior to Sex. This measures how long the respondent knew her or his partner before having sexual intercourse with the partner. Responses are one day or less, two to seven days, one to two weeks, two to four weeks, one to ve months, six months to a year, and a year or more. This measure is problematic because it does not distinguish couples that were romantically involved for a period of time before having sexual intercourse from couples who knew each other platonically before becoming a romantic and sexual couple. Moreover, my interest is in couples who have sexual intercourse within the rst week after meeting because this violates normative prescriptions regarding the timing of rst sexual intercourse (for example, undergraduates predict that the average man expects sexual intercourse in about 3 weeks and the average woman expects sex in 7 to 10 weeks: Cohen and Shotland 1996). Therefore, I recode this outcome into a binary variable indicating whether the respondent had sexual intercourse within the rst week of acquaintance. Physical Attractiveness. The interviewer rated the respondents physical attractiveness on a ve-point scale from Very Unattractive (1) to Very Attractive (5). Though it would

232

E. A. McClintock

Table 1 Means and standard deviations of explanatory measures. National longitudinal study of adolescent health, Waves I, II, and III, 19942002 Adolescencea Mean Female Age (years) Physical attractiveness Very unattractive Unattractive Average Attractive Very attractive Body mass index (BMI)d Weight category Underweight Normal weight Overweight Obese Sexual motivations Personal benets index Social costs index Self-esteem Socioeconomic status Parental socioeconomic index R is college grad/studente Race White Black Hispanic Other Union history Ever cohabitated Ever married
a

Young adulthoodb Nc 14,276 14,276 14,253 Mean 21.91 .02 .05 .47 .36 .10 26.48 .03 .47 .27 .24 NA NA NA NA .32 14,268 (SD) (2.06) Nc 14,276

(SD) (1.69)

.49 16.37 .01 .04 .46 .35 .13 22.86 .11 .65 .16 .08 0.00 0.00 0.00 0.00 NA

Downloaded by [MPI Fuer Ethnologische Forsc] at 09:12 20 February 2013

(4.73)

13,944 13,944

(6.42)

14,159 14,159

(1.00) (1.00) (1.00) (1.00)

13,465 14,169 14,266 14,188

.68 .14 .11 .07 NA NA

14,276

.47 .21

14,261 14,245

Note: Data are weighted. Age, physical attractiveness, BMI and weight group, sexual motivations, and self-esteem are measured during high school, if possible (at Wave II for most respondents and at Wave I for respondents that were high school seniors in Wave I). Parental socioeconomic status and race are measured at Wave I. b Measures in young adulthood use Wave III data. c The N varies because regression models are estimated using full information maximum likelihood estimation allowing cases missing data to be included. d BMI has a slight positive skew. Median BMI is approximately 22 at Wave II and 25 at Wave III. e About 25% of respondents are currently enrolled in a 4-year degree program. This measure includes 4-year college graduates and individuals who will be 4-year college graduates if they complete their current degree program.

Downloaded by [MPI Fuer Ethnologische Forsc] at 09:12 20 February 2013

Table 2 Bivariate comparisons of individual-level romantic and sexual outcomes by physical attractiveness, weight group, and education. National longitudinal study of adolescent health, Waves I, II, and IIIa , 19942002 No. of sexual partners if sexually experienced Ever sexual experience Men proportion 1st Q Mean 7.27 e 6.01 f 6.97 7.23 9.92 5.61 g 7.61 7.55 5.94 1 2 2 1 2 2 2 2 7.61 7.90 7.25 6.30 9 6 8 9 10 6 9 10 7 8 9 9 8 2 2 2 2 2 3rd Q .86 b .87 .81 c .85 .89 .89 .85 d .89 .85 .84 .88 d .92 .87 .81 .76 c .85 .88 .87 .78 d .87 .87 .80 .88 d .87 .86 .81 Women proportion Men 1st Q 2 1 2 2 2 1 2 2 2 1 2 2 2 Women Mean 5.57 6.03 5.69 5.59 5.14 4.72 g 5.56 6.09 6.33 5.20 g 5.66 5.55 5.90 3rd Q 7 8 7 7 6 6 7 7 8 6 7 7 7 (Continued)

Ever romantic experience Women proportion .98 .96 c .97 .99 .99 .98 d .99 .96 .96 .97 .98 .98 .99

Men proportion

All respondents .96 b Physical attractiveness Unattractive .95 c Average .96 Attractive .97 Very attractive .98 Weight group Underweight .93 Normal weight .97 Overweight .96 Obese .96 Parental socioeconomic status .96 025th percentile .96 2650th percentile .98 5175th percentile .97 76100th percentile

233

Downloaded by [MPI Fuer Ethnologische Forsc] at 09:12 20 February 2013

234 Table 2 (Continued) No. of sexual partners if sexually experienced Ever sexual experience Men proportion 1st Q Mean 7.79 h 5.87 5,637 10 7 2 2 3rd Q .88 d .78 6,711 .91 d .80 7,508 Women proportion Men 1st Q 2 1 Women Mean 6.01 h 4.63 6,500 3rd Q 7 6 Women proportion .98 .98 6,958

Ever romantic experience

Men proportion

Rs college graduation status No .96 Yes .97 N (not missing 6,104 dependent variable)

Note: Data are weighted. a Dependent variables are measured in Wave III when respondents average 21.5 years old. Physical attractiveness and weight group are measured at average age 16.5. Parental socioeconomic status is measured at Wave I and respondents college graduation status at Wave III. b p-value from Pearson Chi2 test of association with gender. c p-value for Goodman and Kruskals gamma (for each gender separately). I use gamma when I hypothesize a monotonic relationship. d p-value from Pearson Chi2 test (for each gender separately). I use Chi2 when I hypothesize a non-monotonic relationship. e p-value from Students t-test of gender difference. f p-value from Spearmans Rho (for each gender separately). I use Rho when I expect a linear relationship (rank correlation). g p-value from ANOVA (for each gender separately). I use ANOVA when I expect a non-monotonic relationship. h p-value from Students t-test (for each gender separately). p < 0.05, p < 0.01, p < 0.001 (two-tailed p-values).

Downloaded by [MPI Fuer Ethnologische Forsc] at 09:12 20 February 2013

Table 3 Bivariate comparisons of relationship-level romantic and sexual outcomes by physical attractiveness, weight group, and education. National longitudinal study of adolescent health, Waves I, II, and IIIa , 19942002 Timing of sexual intercourse Proportion of partnerships described as dating exclusively Men .45 b 54 . .47 d .51 .56 .56 .62 c .55 .46 .45 .42 .44 .46 .50
c

Relationship status Proportion of partnerships described as only having sex Men .31 b .23 .32 .29 .36 .27 c .31 .36 .27 .50 c .54 .54 .58 .34 c .30 .31 .29 Women .19 .27 d .19 .18 .18 .13 c .18 .25 .26 .19 c .21 .19 .17 (Continued)

Proportion of partnerships in which sexual intercourse occurred in the rst week Men .24 b .13 .19 d .15 .13 .10 .51 .45 .46 .44 .47 c .46 .41 .45 .11 c .12 .18 .17 .15 .13 .12 .12 .21 .24 .23 .29 .26 c .23 .27 .24 .26 c .25 .26 .20 Women Women

All respondents Physical attractiveness Unattractive Average Attractive Very attractive Weight group Underweight Normal weight Overweight Obese Parental socioeconomic status 025th percentile 2650th percentile 5175th percentile 76100th percentile

235

Downloaded by [MPI Fuer Ethnologische Forsc] at 09:12 20 February 2013

236 Table 3 (Continued) Timing of sexual intercourse Proportion of partnerships described as dating exclusively Men .52 c .42 13,315 .59 c .51 15,809 Women Relationship status Proportion of partnerships described as only having sex Men .26 c .33 13,315 Women .15 c .21 15,809 Men .18 c .27 12,981 .08 c .16 16,522 Women

Proportion of partnerships in which sexual intercourse occurred in the rst week

Rs college graduation status Yes No N (not missing dependent variable)

a Dependent variables are measured in Wave III when respondents average 21.5 years old. Physical attractiveness and weight group are measured at average age 16.5. Parental socioeconomic status is measured at Wave I and respondents college graduation status at Wave III. b p-value from Chi2 test of association with gender. c p-value from Chi2 test (for each gender separately). I use Chi2 when I hypothesize a non-monotonic relationship. d p-value for Goodman and Kruskals gamma (for each gender separately). I use gamma when I hypothesize a monotonic relationship. p < 0.05, p < 0.01, p < 0.001 (two-tailed p-values).

Gender Differences in Revealed Sexual Preferences

237

Downloaded by [MPI Fuer Ethnologische Forsc] at 09:12 20 February 2013

be better to have the average of several observers ratings, prior studies report consistent evaluations of attractiveness between raters (Langlois et al. 2000; Cunningham et al. 1995; Murstein 1972). In the Add Health data, interviewers ratings are moderately positively correlated across waves, increasing my condence in this measure. They range from a minimum of 0.17 (mens correlation between Wave I and Wave III) to a maximum of 0.35 (womens correlation between Wave I and Wave II), and the average correlation size is 0.23 (average of the six gender-specic correlations). I recode the levels into indicator variables for very attractive, average or attractive (reference), and unattractive (I combine unattractive and very unattractive and call this measure unattractive because of the small number of respondents rated very unattractive). This approach does not use all possible information because it does not incorporate the order of the categories, but the most obvious alternative, treating the variable as metric, implicitly assumes that the distance between the levels is constant and that its effect is linear: If either assumption does not hold, the results may be biased. To test this, I estimated a model including both the original ordinal variable and dichotomous variables for three of the ve levels. If the categories are evenly spaced, the coefcients on these indicator variables should be zero (Scott Long et al. 2006), but I found that the coefcients of the indicator variables are statistically signicant: Treating physical attractiveness as if it were a metric variable may bias coefcient estimates. Therefore, I use the indicator variables described earlier but results (not shown) from models treating physical attractiveness as a metric variable are similar. Weight Group. I use body mass index (BMI), a ratio of height to weight, to approximate body size (BMI equals weight in kilograms divided by the square of height in meters). BMI can be misleading for individuals with unusually low percent body fat, but it provides a reasonably accurate measure of thinness or overweight (Ode, Pivarnik, Reeves, and Knoous 2007; Flegal et al. 2009). Individuals with BMI at or below 18.5 are considered underweight, those with BMI of 25 to 29.9 are overweight, and individuals with BMI of 30 or higher are obese (Department of Health and Human Services 2008). Because I expect the effect of BMI to be non-linear, I use dichotomous variables distinguishing underweight, normal weight (reference), overweight, and obese. To some extent, the effect of BMI is captured by the overall attractiveness rating because, especially for women, body size is an important dimension of attractiveness (Brown and Cash 2001; Sugiyama 2005), but body size is also interesting in its own right because it has been associated with differences in romantic and sexual outcomes (Cawley, Joyner, and Sobal 2006). In particular, I investigate the interactions of the weight group variables with female because prior studies have found differences in sexual outcomes for overweight women (Kaneshiro et al. 2008; Wiederman and Hurst 1998; Halpern et al. 1999). Also, body size ideals vary by sex: The ideal female body shape is somewhat thin to very thin (Wiseman et al. 1992; Sugiyama 2005; Fallon and Rozin 1985; Cohn and Adler 1992) whereas the ideal male body size is muscular and of normal to heavy weight (Cohn and Adler 1992; Sugiyama 2005; Leit, Pope, and Gray 2001; Li and Kenrick 2006; Korn and Lerner 1972). Sexual Motivations. I created two indices from eight measures of the respondents perceived moral, social, interpersonal, and personal costs and benets of having sexual intercourse and from six measures of the respondents perceptions of her parents attitudes (reported at average age about 16.5). One of the indices that emerged from factor analysis captures primarily the social costs of sexual activity (negative evaluations by external actors) and the other captures primarily the personal benets (pleasure, relaxation). Cronbachs alpha is 0.86 for the social costs index and 0.71 for the personal benets index.

238

E. A. McClintock

Higher values indicate greater expected costs or benets. (Please see Appendix 1 a more detailed description of these indices.) Self-Esteem. Self-esteem is an index created from an abbreviated version of the Rosenberg self-esteem scale. Cronbachs alpha is 0.85. (See Appendix 2 for more information on this measure.). Female. All models contain an indicator variable for female (1 = female, 0 = male). Demographic Controls. I control for race and age. Race groups are non-Hispanic Black, non-Hispanic White, Hispanic, and non-Hispanic Other. Race is reported by the respondent and is also evaluated independently by the interviewer (relying on physical indications). I use primarily the respondents self-report of her or his race, but when it is missing, I use the interviewers report. Age is the respondents age in years at the Wave III interview. Socioeconomic Controls. I control for whether the respondent has graduated from college or is currently enrolled in a four-year degree program. I use the respondents expected college graduation status because most respondents are in their early twenties and many are still enrolled full-time in school. I also include an index variable for the respondents parents socioeconomic status, constructed from both parents education, occupational status, and family income (see Appendix 3 for details). When possible, this information is taken from the parental interview (Wave I) but, if unavailable in the parental interview, it is taken from the adolescents report about the parent (Wave I). Union History. I include dichotomous variables indicating whether the respondent has ever been in a cohabiting union and whether the respondent has ever been married (as of the Wave III interview). Analysis The Add Health data are clustered and collected with unequal probability of selection (Chantala and Tabor 1999; Chantala 2006). In accordance with guidelines published by the Carolina Population Center at the University of North Carolina at Chapel Hill (Chantala and Tabor 1999; Chantala 2006), the descriptive statistics and individual-level regression analyses shown in this paper correct for the complex sampling design and unequal probability of selection. The relationship-level regression models likewise correct for design effects and unequal probability of selection and also take into account the clustering of relationships within respondents (this allows for a correlation between relationships reported by the same respondent). I estimate models using the statistical software MPlus, which is able to handle the complex survey design and the additional clustering of relationships within respondents. Coefcients from regression models estimated without weighting data and without adjusting for the complex sampling design or the correlation of relationships within respondents tend to have smaller standard errors, so more results of interest are statistically signicant in those models (not shown): Otherwise, results are similar. The independent variables parental socioeconomic status, the indices representing the personal benets and social costs of sexual activity, and the variables derived from BMI are missing data for a non-trivial number of cases (see Table 1). I address all missing data by using full information maximum likelihood (FIML) estimation (also called direct maximum likelihood). FIML constructs parameter estimates and standard errors using all

Downloaded by [MPI Fuer Ethnologische Forsc] at 09:12 20 February 2013

Gender Differences in Revealed Sexual Preferences

239

available information, including the observed values from cases with data on somebut not allvariables (Buhi, Goodson, and Neilands 2008; Enders 2001). Although FIML produces less-biased estimates than ad hoc approaches such as dropping cases with missing data or using mean substitution, a disadvantage is that it assumes missing completely at random (MCAR) or missing at random (MAR) data (Buhi et al. 2008). Still, alternative methods such as multiple imputation (MI) and Bayesian estimation also assume MAR or MCAR data (Buhi et al. 2008). FIML produces less-biased estimates than listwise deletion, pairwise deletion, mean substitution, and related methods, even if the data are not missing at random (NMAR; Enders 2001; Muthen, Kaplan, and Hollis 1987), and it performs equally well as MI and Bayesian methods (Buhi et al. 2008). To try to establish the direction of potential causal relationships, all independent variables except age, expected college education, and union history are lagged. They are measured at Wave I or Wave II, as discussed earlier. This helps to ensure that predictive variables are measured temporally prior to outcome variables, but when the outcome variable is a measure of lifetime accumulation, this temporal ordering is not ensured. Still, few respondents have had many sexual or romantic partners at Wave I or Wave II, so the majority of events do occur after the independent variables were measured. The number of lifetime sexual partners is a metric variable so I use a linear regression model when analyzing this outcome. To make the distribution closer to Gaussian (reduce positive skew), I use the natural log of the number of sexual partners. Aside from the variables derived from BMI, multicollinearity is low, but errors are mildly heteroskedastic. Results are similar using robust regression models to correct for heteroskedasticity, but these models do not allow the data to be weighted (not shown). Results are also similar treating the number of partners as a count outcome and using Poisson regression (not shown). I show results from a linear regression model using weighted data, despite the heteroskedasticity. I use binary logistic regression to evaluate the effect of physical attractiveness on whether the respondent has had a romantic relationship, whether the respondent has had sexual intercourse, and whether the respondent and partner had sexual intercourse within the rst week of meeting. As mentioned earlier, the variable indicating whether the respondent had sexual intercourse within the rst week of meeting a partner was created from a categorical variable on the grounds that its categories do not classify couples in an ordinal manner in terms of how long they delayed sex after becoming a couple. To verify this decision and check the sensitivity of results to this recoding, I also estimated an ordinal logistic regression model using the original seven levels (not shown). Post-regression tests indicated that the parallel regression assumption does not hold, reinforcing my decision to use the binary measure. The choice of model for the relationship status measure depends on whether the categories are truly ordinal. Analyzing a non-ordered variable as ordinal can generate biased coefcients but treating a variable that is truly ordinal as if it were categorical wastes information. Because there are substantive reasons to suspect that this dependent variable is not ordinal, I present results from a multinomial model as a more conservative test of my hypotheses, but I also estimated an ordinal logistic regression model. As expected, given the more efcient use of information, the results regarding the independent variable of interest are somewhat stronger in the ordinal model (results not shown). All models control for the respondents race and age, parental socioeconomic status, and whether the respondent is a college graduate or enrolled in college. All models also include the personal benets of sexual activity index, the social costs of sexual activity index, and self-esteem. In the regressions of the number of sexual partners, relationship

Downloaded by [MPI Fuer Ethnologische Forsc] at 09:12 20 February 2013

240

E. A. McClintock

status, and whether sexual intercourse occurred in the rst week of acquaintance, I control for past union history (marriage and cohabitation) because this likely affects sexual partner accumulation and the nature of partnerships. I do not include these controls in the regressions of whether the respondent has had a romantic relationship and whether the respondent has had sexual intercourse because union history almost perfectly predicts these outcomes. In results not shown, I found that including them changes the effect sizes somewhat, but ndings remain essentially the same. In all regression models, the independent variables of interest are female, physical attractiveness, weight group, and the interactions of female with physical attractiveness and with weight group.

Results
Downloaded by [MPI Fuer Ethnologische Forsc] at 09:12 20 February 2013
Comparison of Sexual and Romantic Outcomes by Physical Attractiveness and Gender Table 2 compares individual-level romantic and sexual outcomes by lagged measures of physical attractiveness, clinical weight group, and socioeconomic status. For both genders, being more attractive increases the probability of having had sexual intercourse at least once and of having had at least one romantic relationship. Underweight and normal-weight women are more likely to report romantic experience, but for men, weight group is unrelated to the chance of romantic involvement. For neither gender does weight level have a monotonic relationship with sexual experience. Instead, the probability of having had sexual intercourse is generally lowest for women and men in the two extreme categories, underweight and obese. Table 2 also shows that for women, the number of sexual partners decreases monotonically with increasing physical attractiveness, whereas for men the number of sexual partners increases monotonically with increasing physical attractiveness. For women, the number of reported partners also decreases with BMI: Thinner women report fewer partners. Thinness is a dimension of attractiveness for women, so this pattern is consistent with the nding that more attractive women report fewer sexual partners. For men, weight group is non-monotonically related to the number of sexual partners: Normal-weight men report the most partners. The ideal weight for men is considered to be normal to heavy weight, so this is reasonably consistent with the nding that more attractive men report more sexual partners. Table 3 compares relationship-level romantic and sexual outcomes by lagged measures of physical attractiveness, clinical weight group, and socioeconomic status. More attractive and thinner women are less likely to report having sexual intercourse in the rst week after meeting their partner. They are also less likely to classify their relationship(s) as only having sex and more likely to classify their relationship(s) as exclusive relationships. There is not as clear an association between mens attractiveness and their chance of reporting sexual intercourse in the rst week of acquaintance, though the most attractive men are most likely to report sexual intercourse in the rst week. Physically attractive men are most likely to classify their relationship(s) as only having sex whereas physically unattractive men are most likely to classify their relationship(s) as exclusive, the reverse of the pattern shown by women. However, whereas for women physical attractiveness and weight group are monotonically and signicantly related to all three outcomes, for men none of these associations are monotonic, and not all are statistically signicant. The patterns observed in Table 2 and Table 3 might be confounded by socioeconomic status. Sexual and romantic outcomes vary by socioeconomic status (as measured by the parental socioeconomic status index and respondents college graduation status; see Table 2

Gender Differences in Revealed Sexual Preferences

241

Downloaded by [MPI Fuer Ethnologische Forsc] at 09:12 20 February 2013

and Table 3), and in these data, women and men who are more educated and/or who have higher-status parents are rated more attractive and tend to have lower BMI (results not shown). These socioeconomic differences in physical attractiveness ratings and in BMI have been noted in other samples (Elder 1969; Langlois et al. 2000; Wardle, Waller, and Jarvis 2002; Hayes and Ross 1986; Carr and Friedman 2005; Umberson and Hughes 1987; Udry 1977). Thus, for example, that more attractive and thinner women report fewer sexual partners might be explained by their having more education, but controlling for education would likely increase the effect of physical attractiveness on mens sexual partner accumulation. Table 2 and Table 3 may also confound the effects of physical attractiveness and weight group: For both genders, BMI is negatively correlated with attractiveness (results not shown). To parse apart these interrelated effects, I turn to multivariate regression models. Table 4 displays regression coefcients for individual-level outcomes, Table 5 shows regression coefcients for relationship-level outcomes, and Table 6 summarizes and evaluates my hypotheses. Has had a romantic relationship: Compared to their average/attractive peers, very physically attractive men and women are more likely to be romantically experienced, but I had also expected physically unattractive men and women to be less likely to have romantic experience. The effect of physical attractiveness is non-trivial: For both genders, being very attractive (versus average/attractive) increases the odds of romantic experience by a factor of about 2.5 (Model 1). Also, compared to being normal weight, being either underweight or overweight decreases the odds of romantic involvement for both genders (Model 1). As predicted, there is no gender difference in the effect of attractiveness. Has had sex: For both genders, being very physically attractive increases the odds of having had sexual intercourse by a factor of 1.3, compared to being average/attractive, whereas being physically unattractive decreases the odds of having had sexual intercourse by a factor of 0.6 (Model 1). As expected, there is no gender difference in the effect of physical attractiveness. For both genders being either underweight or obese decreases the odds of having had sexual intercourse, compared to being normal weight (Model 1). Number of Sexual Partners: Among those with at least one sexual partner, very physically attractive men report more sexual partners, but the relationship between physical attractiveness and the number of reported sexual partners is approximately nonexistent for women (see Models 2 and 3). All else equal, for men being very physically attractive (versus average/attractive) increases reported partners by 26% (percent change generated from Model 3). Compared to normal-weight men, underweight men are predicted to report 17% fewer partners, and obese men are predicted to report 27% fewer partners (Model 3). The effect of weight group for women differs signicantly from that of men in that overweight women report approximately 10% more partners than normal-weight women whereas obese women report approximately 10% fewer partners (Model 3). Sexual intercourse within the rst week of acquaintance: The chance of having sexual intercourse in the rst week after meeting a particular partner varies by physical attractiveness and gender. In Model 3, the odds that very physically attractive men report having had sexual intercourse in the rst week after meeting the partner are larger (by a factor of 1.2) than those of average/attractive men. In contrast, the odds that very physically attractive women report having had sexual intercourse in the rst week after meeting the partner are smaller (by a factor of .9) than those of average/attractive

Downloaded by [MPI Fuer Ethnologische Forsc] at 09:12 20 February 2013

242
Romantic experience Model 2 2.01 2.00 .67 1.56 1.15 .63 .62 .53 .68 .65 1.34 .77 .57 1.20 .69 1.33 .59 .85 .67 .56 .77 .76 .86 .67 .64 1.45 1.33 .57 .50 1.40 1.41 2.02 1.30 1.07 1.07 .50 1.36 .45 .67 .99 .70 1.30 .72 .95 1.34 .59 1.34 .59 2.18 1.57 1.48 1.52 .03 .11 .14 Model 3 Model 1 Model 2 Model 3 Model 1 Sexual experience No. of sexual partners (log) Model 2 .07 .26 .08 .24 .12 .13 .02 .20 .13 .02 .20 Model 3 .02 .26 .07 .23 .15 .17 .05 .27 .07 .17 .16 .11 .19 .11 .19 .11 .19 .70 .63 .62 .53 1.20 .69

Table 4 Odds ratios from logistic regression predicting romantic experience, odds ratios from logistic regression predicting sexual experience, and coefcients from linear regression predicting number of sexual partners (log). National longitudinal study of adolescent health, Waves I, II, and III, 19942002a

Independent variables

Model 1

2.09

Female Physical attractiveness Very attractive Attractive/average (reference) Unattractive Female very attractive Female unattractive Weight category Underweight Normal weight (reference) Overweight Obese Female underweight Female overweight Female obese Sexual motivations Personal benets index Social costs index

2.46

1.21 .69

Downloaded by [MPI Fuer Ethnologische Forsc] at 09:12 20 February 2013

1.01 1.25 1.58 1.02 .69 NA NA 8,164 8,172 14,219 NA NA NA NA 8,184 1.02 .69 1.02 .69 NA NA 2,758 13,062 2,770 NA NA 1.24 1.57

1.01

1.01

1.17

1.18

1.17

.01 .07 .15

.01 .07 .15

.01 .07 .15

1.25 1.58 NA NA 2,747

Self-esteem Socioeconomic status Parental socioeconomic index Rs college graduation status Union history Ever married Ever cohabitated Constant Bayesian information criterionb N (actual)

.29 .32 1.38 29,990

.29 .32 1.37 29,980 12,137

.29 .32 1.39 29,983

Data are weighted. Models are estimated using FIML and estimates account for sampling design. Models control for age and race. Bayesian information criterion is sample-size adjusted using sample size with complete data on all variables (not FIML sample). p < 0.05, p < 0.01, p < 0.001 (two-tailed p-values).

243

Downloaded by [MPI Fuer Ethnologische Forsc] at 09:12 20 February 2013

244
Status of relationship (vs. only having sex) Dating exclusively Model 1 1.60 .96 1.13 1.27 1.48 .78 .86 1.23 1.42 .99 .78 .92 .79 .71 1.18 1.62 .69 1.53 1.61 1.59 1.51 Model 2 Model 3 Model 1 Model 2 Model 3 1.51 .71 1.17 1.59 .72 Dating frequently Model 3 .56 1.24 .94 .71 1.13 Model 1 1.46 .90 1.09 Model 2 1.40 .73 1.26 1.48 .73 1.20 1.12 1.25 .73 1.14 1.27 .75 .73 .76 .74 1.30 1.29 1.06 .84 .97 1.41 .79 .57 1.12 .83 .74 1.11 .84 .75 .94 .85 .79 1.36 .95 .86 1.16 .81 .97 1.15 .83 .99

Table 5 Odds ratios from logistic regression predicting whether respondent and partner had sexual intercourse within the rst week of acquaintance and odds ratios from multinomial logistic regression predicting relationship status. National study of adolescent health, Waves I, II, & III, 19942002a

Sexual intercourse within rst week of acquaintance

Dating once in awhile Model 3 1.51 .74 1.21 1.42 .82

Independent variables

Model 1

Model 2

.54

.57

1.03

1.24

.99

.93 .69

1.18

1.00

1.01

1.07 .90 1.34 1.16 .80 .54

Female Physical attractiveness Very attractive Attractive/avg. (ref.) Unattractive Female very attractive Female unattractive Weight category Underweight Normal weight (ref.) Overweight Obese Female underweight Female overweight Female obese

1.19 1.41

1.18 1.39

Downloaded by [MPI Fuer Ethnologische Forsc] at 09:12 20 February 2013

1.16 .95 .95 1.04 .69 .99 1.22 .99 1.22 .99 1.21 .97 1.11 .97 1.12 1.13 1.05 1.13 1.05 1.13 1.05 1.09 1.03 1.09 1.03 1.09 1.03 .97 1.12

1.16

1.16

.82

.82

.82

.85

.86

.85

.86 1.10 1.00 .92 1.11

.86 1.11 1.00 .92 1.11

.85 1.11 1.00 .92 1.10

.96 .95

.96 .95

1.04 .69

1.04 .68

Sexual motivations Personal benets index Social costs index Self-esteem Socioeconomic status Parental SES index Rs college grad. status Union history Ever married Ever cohabitated BICb N (actual) .97 1.24 24,951 .67 .83 64,415 .67 .83 64,430 .67 .83 64,445 .74 .86 64,415 .74 .86 64,430 29,124 .74 .86 64,445

.96 1.24 24,944

.96 1.24 24,942 29,503

.89 .84 64,415

.90 .84 64,430

.88 .84 64,445

Data are weighted. Estimates use FIML, account for sampling design, clustering of relationships within respondent, and control for age and race. Bayesian information criterion (BIC) is sample-size adjusted using sample size with complete data on all variables (not FIML sample). p < 0.05, p < 0.01, p < 0.001 (two-tailed p-values).

245

Downloaded by [MPI Fuer Ethnologische Forsc] at 09:12 20 February 2013

246 Table 6 Summary of hypotheses and ndings Findings from bivariate statistics Results of multivariate regression models The chance of having ever had a romantic relationship increases with physical attractiveness for men and women (p < 0.001). The chance of having ever had sexual intercourse increases with physical attractiveness for men and women (p < 0.001). For men, the number of sexual partners increases monotonically with physical attractiveness (p < 0.001). NA For women, the number of sexual partners decreases monotonically with physical attractiveness. ? For men, the pattern is non-monotonic, but the most attractive men are most likely to have sex in the rst week of acquaintance. Being very physically attractive increases the odds of romantic experience by a factor of 2.5 (p < 0.001, Model 1), compared to average/attractive. Being unattractive vs. average/attractive decreases the odds of sexual experience by a factor of 0.6 (p < 0.001, Model 1); being very physically attractive increases them by a factor of 1.3 (p < 0.05, Model 1). For men, being very physically attractive vs. average or attractive increases the predicted number of partners by 26% (p < 0.001, Model 3). NA For women, being very physically attractive vs. average or attractive has essentially no effect on the predicted number of partners (p < 0.01, Model 3). For men, being very physically attractive vs. average or attractive increases the odds of sex within the rst week by a factor of 1.2 (p < 0.05).

Hypothesis: Compared to the less physically attractive, more physically attractive . . .

H1a

. . . women and men will be more likely to have had a romantic relationship.

H1b

. . . women and men will be more likely to have had sexual intercourse.

H2

. . . men will report more sexual partners

H3a

. . . men are more likely to have sex soon after meeting a new partner

Downloaded by [MPI Fuer Ethnologische Forsc] at 09:12 20 February 2013

H3b

. . . women are less likely to have sex soon after meeting a new partner. More physically attractive women are less likely to have sex within the rst week of knowing a partner (p < 0.001).

H4a

. . . men will be more likely to describe their relationships as purely sexual, and less likely to describe them as exclusive.

For women, being very physically attractive vs. average/attractive decreases the odds of sex within the rst week by a factor of 0.9 (p < 0.01, Model 3). For men, being very physically attractive vs. average or attractive decreases the odds of reporting an exclusive (vs. sexual) relationship by a factor of 0.8 (p < 0.05, Model 3). For women, being very physically attractive vs. average or attractive increases the odds of reporting an exclusive (vs. sexual) relationship by a factor of 1.1 (p < 0.01, Model 3).

H4b

. . . women will be less likely to describe their relationships as purely sexual, and more likely to describe them as exclusive.

? For men, the association is non-monotonic but very physically attractive men are most likely to report a purely sexual relationship and least likely to report an exclusive relationship. For women, physical attractiveness is associated with having exclusive relationships (p < 0.001) and not having purely sexual relationships (p < 0.001).

NA = no hypothesis regarding this nding; = hypothesis supported; ? = mixed/unclear ndings; X = hypothesis not supported.

247

248

E. A. McClintock

Downloaded by [MPI Fuer Ethnologische Forsc] at 09:12 20 February 2013

women. However, being physically unattractive does not have an effect for either gender. For both genders, being overweight or obese increases the odds of reporting sexual intercourse in the rst week: This is unexpected especially for obese men: Compared to normal-weight men, obese men should have less ability to attract partners, so they should be less able to have early sexual intercourse. For women, being underweight (versus normal weight) decreases the odds of having sexual intercourse in the rst week by a factor of 0.9 (Model 3). Relationship Exclusivity: For women, being very physically attractive increases (by a factor of 1.1) the odds that a given relationship is described as exclusive rather than as only having sex whereas for men, being very physically attractive decreases (by a factor of 0.8) the odds that a given relationship is described as exclusive (Model 3). For men, being overweight decreases the odds of an exclusive relationship (Model 3), but the effect of weight group is more notable for women. Being underweight increases the odds of an exclusive relationship by a factor of 1.5 whereas being overweight or obese decreases the odds of an exclusive relationship (by factors of 0.7 and 0.6, respectively). Physical attractiveness and weight group are similarly related to the odds of classifying a relationship as dating frequently as opposed to only having sex and of classifying it as dating once in awhile as opposed to only having sex. Thus, although the wording of this variable is not unambiguously ordinal, the independent variables of interest have approximately parallel effects. Limitations and Alternative Explanations In the preceding analysis, I have shown that romantic and sexual outcomes vary by physical attractiveness and gender and that the effect of physical attractiveness often depends on gender. I have attributed this to gender differences in sexual strategies (whether of social or evolutionary origin). However, though it is not feasible to identify and evaluate all possible alternative explanations, I consider some plausible alternatives below. I also discuss some of the limitations of my analysis. One alternative explanation is that desired romantic and sexual outcomes vary by physical attractiveness and gender such that the effect of physical attractiveness on romantic and sexual outcomes is due to differences in sexual strategies. However, in these data, the perceived social costs of sexual activity index and the personal benets of sexual activity index are only weakly related to physical attractiveness (not shown). Moreover, in the regression models, there is no evidence that the effect of physical attractiveness is mediated by these indices or by other plausible mediators such as self-esteem. The lack of mediating variables provides insight into the process that results in different outcomes for more and less physically attractive women and men. Explanations that focus on the supply side assume that individuals pursued goals determine their outcomes. If this were the case, the effect of physical attractiveness would likely be mediated by perceptions of the intrinsic benets and social costs of sexual behavior or would be mediated by self-esteem. In contrast, explanations that focus on the demand side do not imply mediating variables: Under this model, more and less physically attractive women and men may pursue the same outcomes, but those most in demand (the physically attractive) are most able to obtain their ideal outcomes. Findings in this paper support a demand side explanation. Another concern might be that respondents stated behaviors are ltered through preferences and social norms and thus may reect differences in preferences and norms rather than differences in behavior. Women and men might intentionally or unintentionally distort

Gender Differences in Revealed Sexual Preferences

249

their reported behavior to be more consistent with perceptions of gendered social norms and/or with their own preferred outcomes. Though it is not possible to check the validity of self-reports, it is reassuring that simple under- and over-reporting by women and men is not likely to greatly alter my results. This is because my focus is not on between-gender comparisons of sexual and romantic outcomes: It is on within-gender differences between more and less physically attractive respondents and on between-gender comparisons of the effect of physical attractiveness. Although gender likely affects the direction of bias in selfreported behavior, as long as physical attractiveness does not also affect the direction or degree of bias, my within-gender, between-attractiveness level comparisons remain valid. Likewise, between-gender comparisons of the effect of physical attractiveness will also remain valid. An important limitation in this research is the inability to measure the sexual and romantic opportunities that were available to individuals but that the individuals chose to pass up. I assume that lower rates of romantic and sexual involvement reect fewer opportunities, but this might also result from less physically attractive individuals opting out of the partner market. Though it would be ideal to have a measure of opting out, there is ample evidence in prior research that physical attractiveness does increase romantic and sexual opportunities. Moreover, opting out should not alter the associations found in the relationship-level models: Physical attractiveness might inuence the decision to remain single, but the relationship-level data reect negotiated outcomes among individuals who did not opt out. A second limitation is that the measures of relationship status and the timing of sexual intercourse are sub-optimal. As discussed earlier, the categories of relationship status may not be mutually exclusive. Also, the variable measuring the timing of sexual intercourse does not distinguish individuals who were romantically involved for a given time period prior to having sexual intercourse from acquaintances who were not romantically involved for all/any of that time. Delaying intercourse with a romantic partner is substantively different than abstaining from sexual intercourse with a platonic friend or casual acquaintance, but though less-ambiguous measures would be better, there is no reason to expect that any bias generated by using these measures varies systematically by physical attractiveness and gender. Instead, the noise generated by using these variables would tend to result in false-negative results, not in false-positive results. A third limitation is that by focusing on broad patterns of gender difference, this paper obscures potential within-gender variation. Though evolutionarily determined sexual strategies might vary little between individuals, adherence to the culturally dened double standard likely varies substantially. In particular, examples of women who have openly deed traditional double standards and embraced casual sexuality cast doubt on the universality of sexual double standards. Many women, especially those with a more feminist ideology, do not accept the double standard of sexuality (Bay-Chen and Zucker 2007), though the social costs of violating gendered sexual norms may still cause some of these women to limit their sexual expression (Blumberg 2003). It is plausible that women belonging to subcultures supportive of female sexuality and women who are less dependent on social approval (because of nancial independence or simply less psychological need for approval) may be more likely to pursue sexual strategies at odds with the sexual double standard. In fact, being very physically attractive might provide women with the romantic and sexual power to violate the sexual double standard. In future work, it would be interesting to investigate how womens social location and independence might interact with physical attractiveness to determine adherence to gendered sexual norms.

Downloaded by [MPI Fuer Ethnologische Forsc] at 09:12 20 February 2013

250

E. A. McClintock

Discussion
If more-physically attractive women and men are more-desired partners in the young adult sexual and romantic market and are therefore more able to negotiate the rules of their relationships, gender differences in the effect of physical attractiveness on sexual and romantic outcomes are indicative of gender differences in sexual strategies. Because they are inuenced by the double standard of sexuality or because they are inuenced by subconscious, evolutionarily determined mating strategies, results indicate that women have more conservative sexual strategies than men. For example, that more physically attractive men report more partners than less attractive men suggests that men seek a greater number of sexual partners than women: Physically attractive men do better in the resulting competition for sexual access. In contrast, that more physically attractive women do not report more sexual partners than less-attractive women suggests that women do not pursue many sexual partners. I also found differences in relationship status and the timing of sexual intercourse. Very physically attractive women are more likely to form exclusive relationships than to form purely sexual relationships. They are also less likely to have sexual intercourse within the rst week of meeting a partner. Presumably, this difference arises because more physically attractive women use their greater power in the partner market to control outcomes within their relationships. These results are consistent with the social structural model that posits that womens romantic and sexual goals are shaped by the double standard of sexuality. The evolutionary perspective also predicts that more physically attractive women would be more able to nd a long-term mate, but not all evolutionary theorists claim that women favor a long-term mating strategy. Findings regarding relationship status and the timing of sexual intercourse for men were also consistent with the expectations. For men, being very physically attractive increases the chance of reporting purely sexual relationships (versus exclusive relationships) and being physically attractive also increases the chance of having sexual intercourse in the rst week of acquaintance. According to the evolutionary model, when (if) women have casual sexual intercourse, they select men with desirable genes: physically attractive men. Some evolutionary theorists further argue that physically attractive men are genetically predisposed to pursue a short-term mating strategy. Thus, these results are consistent with an evolutionary model in that partnerships characterized as only having sex and partnerships in which sexual intercourse occurs after a brief period of acquaintance are indicative of a short-term mating strategy. These ndings neither support nor contradict the social structural explanation: The double standard of sexuality sanctions women for sexual promiscuity, so it does not directly predict different outcomes for more or less physically attractive men. However, these ndings are consistent with claims by some authors that young adult men are averse to romantic commitment (Cross 2008; Kimmel 2008). In many instances, only very physically attractive individuals differed signicantly from average/attractive individuals (the reference group) whereas unattractive individuals did not differ. In fact, being very physically attractive is related to all of the ve outcomes (for at least one gender), but being physically unattractive is related to only one of the outcomes (it reduces the odds of sexual experience for both genders). It seems that it is primarily the very physically attractive individuals who differ in their ability to control the form and trajectory of their relationships: Very physically attractive individuals are more advantaged than unattractive individuals are disadvantaged. Weight group also inuences romantic and sexual outcomes. In general, outcomes for women and men with closer-to-ideal physiques (underweight to normal weight for women

Downloaded by [MPI Fuer Ethnologische Forsc] at 09:12 20 February 2013

Gender Differences in Revealed Sexual Preferences

251

and normal weight to overweight for men) were similar to those attained by very physically attractive individuals. Outcomes for those of less optimal physiques were more similar to outcomes for physically unattractive women and men. Body physique is a dimension of physical attractiveness so these results increase my condence in the validity of the overall measure of physical attractiveness. It is also interesting that for women the effect of being underweight on within-relationship outcomes resembles the effect of being very physically attractive. This suggests that the factors inuencing romantic and sexual desirability are at least in part socially structured (underweight women are less fertile so the evolutionary perspective predicts that they would be less able to obtain desired outcomes). That physical attractiveness affects relationship status and the timing of sexual intercourse is evidence that individuals make compromises rather than completely opting out of the partner market. The compromises that individuals make to avoid remaining single are interesting because they provide insight into the exibility of sexual and romantic strategies. From an evolutionary perspective, examining ideal and compromised choices reveals alternative mating strategies. From a social structural perspective, ideal and compromised choices indicate how individuals negotiate the competing pressures of gendered sexual standards and constrained sexual and romantic opportunities. It is only by examining acted preferences (versus stated preferences) that these compromises are revealed. Throughout this paper, I have considered two perspectivesevolutionary psychological and social structuralon the source of gender differences in sexual strategies. Although these models are based on very different assumptions, in this study their predictions differ in only three regards. First, the sexual double standard unequivocally indicates that women will favor committed sexual relationships whereas some formulations of the evolutionary perspective suggest that women may pursue both long- and short-term mating strategies. Second, the sexual double standard predicts that women will prefer delaying sexual intercourse whereas the evolutionary model does not make a clear prediction regarding gender differences (or similarity) in the preferred timing of sexual intercourse. Third, the evolutionary model clearly indicates that more physically attractive men will have a greater propensity to form casual sexual relationships whereas the social structural model provides less guidance in predicting mens behavior (the sexual double standard is directed at women). The social structural explanation was supported in both instances when the evolutionary model failed to provide a prediction and the evolutionary explanation was supported in the one instance when the social structural model failed to make a clear prediction. Because the two models never make directly conicting predictions and both are supported, it is difcult to adjudicate between the two perspectives. Examining outcomes for underweight women may provide some leverage in testing between evolutionary and social structural theories (in this paper it favors the social structural perspective), but this is only an indirect test. It is clear that sexual and romantic outcomes are at least partially socially structured: All of the outcomes are strongly related to socioeconomic status and to one or both of the sexual motivations indices, but this does not preclude the inuence of evolutionary adaptations as an additional causal factor. This inability to unambiguously select either perspective as a superior explanation for these ndings is also evident in prior research. It is often difcult to adjudicate between evolutionary and social structural theories because the two approaches make similar predictions and are not inherently incompatible (Kenrick et al. 1993; Howard et al. 1987). Although beyond the scope of this paper, the similarity of predictions from the social structural and evolutionary perspectives is interesting. Perhaps the sexual double standard has persisted in part because the difference in sexual behavior that it enforces is evolutionarily determined and would exist regardless. An alternative explanation is that

Downloaded by [MPI Fuer Ethnologische Forsc] at 09:12 20 February 2013

252

E. A. McClintock

evolutionary theories of human mating behavior have been developed post hoc to explain observed behavior: Insofar as observed behavior is consistent with the sexual double standard, evolutionary theories will tend to make similar predictions. Thus, by exploiting the instrumental value of physical attractiveness in actuating sexual and romantic strategies, this paper demonstrated that women and men pursue different sexual and romantic outcomes, but it was unable to fully discriminate between social structural and evolutionary explanations.

References
Alexander, M. G., and T. D. Fisher. 2003. Truth and consequences: Using the bogus pipeline to examine sex differences in self-reported sexuality. J Sex Res 40(1):2735. Baumeister, R. F., and J. M. Twenge. 2002. Cultural suppression of female sexuality. Rev Genderal Psychol 6(2):166203. Bay-Chen, L. Y., and A. N. Zucker. 2007. Feminism between the sheets: Sexual attitudes among feminists, nonfeminists, and egalitarians. Psychol Women Quart 31:157163. Bearman, P., J. Jones, and J. R. Udry. 2008. The National Longitudinal Study of Adolescent Health: Research Design. Caroline Population Center 2004. Available from http://www.cpc.unc.edu/projects/addhealth/design (accessed July 3 2008) Berscheid, E., and K. Dion. 1971. Physical Attractiveness and dating choice: A test of the matching hypothesis. J Exp Soc Psy 7:173189. Blumberg, E. S. 2003. The lives and voices of highly sexual women. J Sex Res 40(2):146157. Bogaert, A. F., and W. A. Fisher. 1995. Predictors of university mens number of sexual partners. J Sex Res 32(2):119130. Brown, T. A., and T. F. Cash. 2001. Perceptions of physical attractiveness among college students: Selected determinants and methodological matters. J Soc Psy 126(3):305316. Buhi, E. R., P. Goodson, and T. B. Neilands. 2008. Out of Sight, Not Out of Mind: Strategies for Handling Missing Data. Am J Health Behav 32(1):8392. Burdett, K., and M. G. Coles. 1999. Long-Term partnership formation: Marriage and employment. Eco J 109(456):307334. Burdett, K., and M. G. Coles. 2001. Transplants and implants: The economics of self-improvement. Int Eco Rev 42(3):597616. Buss, D. M. 1998. Sexual strategies theory: Historical origins and current status. J Sex Res 35(1):1931. Buss, D. M., and M. Barnes. 1986. Preferences in human mate selection. J Personality Soc Psy 50(3):559570. Buss, D. M., T. K. Shackelford, L. A. Kirkpatrick, and R. J. Larsen. 2001. A half century of mate preferences: The cultural evolution of values. J Marriage Fam 63(2):491503. Byers, E. SANDRA, and K. Lewis. 1988. Dating couples disagreements over the desired level of sexual intimacy. J Sex Res 24:1529. Carr, D., and M. A. Friedman. 2005. Is obesity stigmatizing? Body weight, perceived discrimination, and psychological well-being in the United States. J Health Soc Behav 46(3):244259. Cawley, J., K. Joyner, and J. Sobal. 2006. Size matters: The inuence of adolescents weight and height on dating and sex. Rationality Soc 18(1):6794. Chantala, K. 2006. Guidelines for Analyzing Add Health Data. Carolina Population Center 2006 (accessed July 3, 2008). http://www.cpc.unc.edu/projects/addhealth/les/wt_guidelines.pdf. Chantala, K., and J. Tabor. 1999. National Longitudinal Study of Adolescent Health: Strategies to perform a design-based analysis using the Add Health data. Chapel Hill, NC: Carolina Population Center, University of North Carolina at Chapel Hill. Choo, E., and A. Siow. 2006. Estimating a marriage matching model with spillover effects. Demography 43(3):463490. Christensen, H. T., and G. R. Carpenter. 1962. Value-behavior discrepancies regarding premarital coitus in three Western cultures. Am Sociol Rev 27(1):6674.

Downloaded by [MPI Fuer Ethnologische Forsc] at 09:12 20 February 2013

Gender Differences in Revealed Sexual Preferences

253

Cohen, L. L., and R. L. Shotland. 1996. Timing of rst sexual intercourse in a relationship: Expectations, experiences, and perceptions of others. J Sex Res 33(4):291299. Cohn, L. D., and N. E. Adler. 1992. Female and Male perceptions of ideal body shapes: Distorted views among caucasian college students. Psy Women Quart 16:6979. Coles, M. G., and M. Francesconi. 2007. On the emergence of toyboys: Equilibrium matching with ageing and uncertain careers. Institute for the Study of Labor Working Paper Series Discussion Paper No. 2612. Coombs, R. H., and W. F. Kenkel. 1966. Sex differences in dating aspirations and satisfaction with computer-selected partners. J Marriage Fam 28(1):6266. Crawford, M., and D. Popp. 2003. Sexual double standards: a review and methodological critique of two decades of research. J Sex Res 40(1):1326. Cross, G. 2008. Men to boys: The making of modern immaturity. New York: Columbia University Press. Cunningham, M. R., A. R. Roberts, A. P. Barbee, P. B. Druen, and C.-H. Wu. 1995. Their ideas of beauty are, on the whole, the same as ours: Consistency and variability in the cross-cultural perception of female physical attractiveness. J Personality Soc Psy 68(2):261279. Daly, M. 1999. Darwinism and the roots of machismo. Scientic Am Presents 10(2):814. Daly, M., and M. Wilson. (Eds.). 2005. Human behavior as animal behavior. In Behavior of animals: Mechanisms, function, and evolution, eds J. J. Bolhuis and L. A. Giraldeau. Oxford: Blackwell Publishing. Department of Health and Human Services. 2008. Calculate your body mass index. National Institutes of Health. http://www.nhlbisupport.com/bmi/ (accessed July 3, 2008). DeSteno, D. A., and P. Salovey. 1996. Evolutionary origins of sex differences in jealousy? Questioning the tness of the model. Psychol Sci 7(6):367372. Dietz, T., T. R. Burns, and F. H. Buttel. 1990. Evolutionary theory in sociology: An examination of current thinking. Sociol Forum 5(2):155171. Dion, K. K., and E. Berscheid. 1972. What is beautiful is good. J Personality Soc Psy 24(3):285290. Doosje, B., K. Rojahn, and A. Fischer. 1999. Partner preferences as a function of gender, age, political orientation and level of education. Sex Roles 40(1/2):4560. Eagly, A. H. 1997. Sex differences in social behavior: Comparing social role theory and evolutionary psychology. Am Psychologist 52(12):13801383. Eastwick, P. W., and E. J. Finkel. 2008. Sex differences in mate preferences revisited: do people know what they initially desire in a romantic partner? J Perssonality Soc Psy 94(2):245264. Elder, G. H. 1969. Appearance and education in marriage mobility. Am Sociol Rev 34(4):519533. Enders, C. K. 2001. The Performance of the Full Information Maximum Likelihood Estimator in Multiple Regression Models with Missing Data. Ed Psychol Measurement 61(5):713740. Eyre, S. L., V. Hoffman, and S. G. Millstein. 1998. The gamesmanship of sex: A model based on African American adolescent accounts. Med Anthr Quart 12(4):467489. Fallon, A. E., and P. Rozin. 1985. Sex differences in perceptions of desirable body shape. Jf Abnorm Psy 94(1):102105. Fisman, R., S. S. Iyengar, E. Kamenica, and I. Simonson. 2006. Gender differences in mate selection: Evidence from a speed dating experiment. Quart J Eco 121(2):673697. Flegal, K. M., J. A. Sheperd, A. C. Looker, B. I. Graubard, L. G. Borrud, C. L. Ogden, T. B. Harris, J. E. Everhard, and N. Schenker. 2009. Comparisons of percentage body fat, body mass index, waist circumference, and waist-stature ratio in adults. Am J Clin Nut 89(2):500508. Fletcher, G. J. O., J. M. Tither, C. OLoughlin, M. Friesen, and N. Overall. 2004. Warm and homely or cold and beautiful? Sex differences in trading off traits in mate selection. Personality Soc Psy Bull 30:659672. Furnham, A. 2009. Sex differences in mate selection preferences. Personality Ind Diff 47(4):262267. Gangestad, S. W., and J. A. Simpson. 2000. The evolution of human mating: Trade-offs and strategic pluralism. Behav Brain Sci 23:573644.

Downloaded by [MPI Fuer Ethnologische Forsc] at 09:12 20 February 2013

254

E. A. McClintock

Gangestad, S. W., and R. Thornhill. 1997. The evolutionary psychology of extrapair sex: The role of uctuating assymmetry. Evolution Hum Behav 18:6988. Gillen, M. M., E. S. Lefkowitz, and C. L. Shearer. 2006. Does Body image play a role in risky sexual behavior and attitudes? J Youth Ado 35(2):243255. Gillmore, M. R., M. E. Archibald, D. M. Morrison, A. Wilsdon, E. A. Wells, M. H. Hoppe, D. Nahom, and E. Murowchich. 2002. Teen sexual behavior: Applicability of the theory of reasoned action. J Marriage Fam 64(4):885897. Grodstein, F., M. B. Goldman, and D. W. Cramer. 1994. Body Mass Index and Ovulatory Infertility. Epidemiol 5(2):247250. Haavio-Mannila, E., J. P. Roos, and O. Kontula. 1996. Repression, revolution and ambivalence: The sexual life of three generations. Acta Sociologica 39(4):409430. Halpern, C. T., J. R. Udry, M. E. Campbell, and C. Suchindran. 1999. Effects of body fat on weight concerns, dating, and sexual activity: A longitudinal analysis of black and white adolescent girls. Dev Psy 35(3):721736. Hansen, S. L. 1977. Dating choices of high school students. Fam Coor 26(2):133138. Hayes, D., and C. E. Ross. 1986. Body and mind: The effect of exercise, overweight, and physical health on psychological well-being. J Health Soc Behav 27(4):387400. Howard, J. A., P. Blumstein, and P. Schwartz. 1987. Social or evolutionary theories? Some observations on preferences in human mate selection. J Personality Soc Psy 53(1):194200. Hrdy, S. Blaffer. 1981. The woman that never evolved. Cambridge, MA: Harvard University Press. Hrdy, S. Blaffer. 1999. Mother nature: Maternal instincts and how they shape the human species. New York: Ballantine Books. Kaneshiro, B., J. T. Jensen, N. E. Carlson, S. M. Harvey, M. D. Nichols, and A. B. Edelman. 2008. Body mass index and sexual behavior. Obstetrics Gyn 112:586592. Kenrick, D. T., G. E. Groth, M. R. Trost, and E. K. Sadalla. 1993. Integrating evolutionary and social exchange perspectives on relationships: Effects of gender, self-appraisal, and involvement level on mate selection criteria. J Personality Soc Psy 64(6):951969. Kenrick, D. T., E. K. Sadalla, G. E. Groth, and M. R. Trost. 1990. Evolution, traits, and the stages of human courtship: Qualifying the parental investment model. J Personality 58(1):97116. Kimmel, M. 2008. Guyland: The perilous world where boys become men. New York: Harper Collins Publishers. Korn, S. J., and R. M. Lerner. 1972. The development of body-build stereotypes in males. Child Dev 43(3):908920. Kruger, D. J. 2006. Male facial masculinity inuences attributions of personality and reproductive strategy. Personal Rel 13:451463. Kruger, D. J., Maryanne Fisher, and Ian Jobling. 2003. Proper and dark heroes as dads and cads. Hum Nat 14(3):305317. Langlois, J. H., L. Kalakanis, A. J. Rubenstein, A. Larson, M. Hallam, and M. Smoot. 2000. Maxims or myths of beauty? A meta-analytic and theoretical review. Psychol Bull 126(3):390423. Leit, R. A., H. G. Pope Jr., and J. J. Gray. 2001. Cultural expectations of muscularity in men: The evolution of Playgirl centerfolds. Int J Eating Dis 29:9093. Li, N. P., J. M. Bailey, D. T. Kenrick, and J. A. W. Linsenmeier. 2002. The necessities and luxuries of mate preferences: testing the tradeoffs. J Personality Soc Psy 82(6):947955. Li, N. P., and D. T. Kenrick. 2006. Sex similarities and differences in preferences for short-term mates: What, whether, and why. J Personality Soc Psy 90(3):468489. Luo, S., and G. Zhang. 2009. What leads to romantic attraction: Similarity, reciprocity, security, or beauty? Evidence from a speed-dating study. J Personality 77(4):933963. Margolin, L., L. White. 1987. The continuing role of physical attractiveness in marriage. J Marriage Fam 49(1):2127. Mathes, E. W., C. Arms, A. Bryant, J. Fields, and A. Witowski. 2005. Do men believe that physically attractive women are more healthy and capable of having children? Psychol Rep 96(3):10021008.

Downloaded by [MPI Fuer Ethnologische Forsc] at 09:12 20 February 2013

Gender Differences in Revealed Sexual Preferences

255

McNulty, J. K., L. A. Neff, and B. R. Karney. 2008. Beyond initial attraction: Physical attractiveness in newlywed marriage. J Fam Psy 22(1):135143. Meier, A. M. 2003. Adolescents transition to rst intercourse, religiosity, and attitudes about sex. Soc Forces 81(3):10311052. Moore, F. R., and C. Cassidy. 2007. Female status predicts female mate preferences across nonindustrial societies. Cross-Cultural Res 41(1):6674. Mulford, M., J. Orbell, C. Shatto, and J. Stockard. 1998. Physical attractiveness, opportunity, and success in everyday exchange. Am J Soc 103(6):15651592. Murstein, B. I. 1972. Physical attractiveness and marital choice. J Personality Soc Psy 22(1):812. Muthen, B. O., D. Kaplan, and M. Hollis. 1987. On Structural Equation Modeling with Data that are Not Missing Completely at Random. Psychometrika 52:431462. Nevid, J. S. 1984. Sex differences in factors of romantic attraction. Sex Roles 11(5/6):401411. Ode, J. J., J. M. Pivarnik, M. J. Reeves, and J. L. Knous. 2007. Body mass index as a predictor of percent fat in college athletes and nonathletes. Med Sci Sports Exer 39(3):403409. OSullivan, L. F. 1995. Less is more: The effects of sexual experience on judgments of mens and womens personality characteristics and relationship desirability. Sex Roles 33(3/4):159181. Rich-Edwards, J. W., D. Spiegelman, M. Garland, E. Hertzmark, D. J. Hunter, G. A. Colditz, W. C. Willett, H. Wand, and J. E. Manson. 2002. Physical Activity, Body Mass Index, and Ovulatory Infertility. Epidemiol 13(2):184190. Regan, P. C. 1998. What if you cant get what you want? Willingness to compromise ideal mate selection standards as a function of sex, mate value, and relationship context. Personality and Social Psychology Bulletin 24(12):12941303. Regan, P. C., and A. Joshi. 2003. ideal partner preferences among adolescents. Soc Behav Personality 31(1):1320. Reiss, I. L. 1967. The social context of premarital sexual permissiveness. New York: Holt, Rinehard, and Winston. Reiss, I. L. 1986. Journey into sexuality: An exploratory voyage. New York: Prentice Hall. Schmitt, D. P. 2005a. Fundamentals of human mating strategies. In The handbook of evolutionary psychology, ed. D. M. Buss. Hoboken, NJ: John Wiley & Sons, Inc. Schmitt, D. P. 2005b. Sociosexuality from Argentina to Zimbabwe: A 48-nation study of sex, culture, and strategies of human mating. Behav Brain Sci 28:24731. Scott Long, J., J. Freese, StataCorp LP. 2006. Regression models for categorical dependent variables using Stata. College Station, TX: Stata Press. South, J. S. 1991. Sociodemographic differentials in mate selection preferences. J Marriage Fam 53(4):928940. Sprecher, S. 1989a. The importance to males and females of physical attractiveness, earning potential, and expressiveness in initial attraction. Sex Roles 21(9/10):591607. Sprecher, S. 1989b. Premarital sexual standards for different categories of individuals. J Sex Res 26(2):232248. Sprecher, S. 1998. Social exchange theories and sexuality. J Sex Res 35(1):3243. Sprecher, S., A. Barbee, and P. Schwartz. 1995. Was it good for you, too?: Gender differences in rst sexual intercourse experiences. J Sex Res 32(1):315. Sprecher, S., K. McKinney, and T. L. Orbuch. 1987. Has the sexual double standard disappeared? An experimental test. Soc Psy Quart 50(1):2431. Sprecher, S., Q. Sullivan, and E. Hateld. 1994. Mate Selection preferences: Gender differences examined in a national sample. J Personality Soc Psy 66(6):10741080. Stevens, G., D. Owens, and E. C. Schaefer. 1990. Education and attractiveness in marriage choices. Soc Psy Quart 53(1):6270. Sugiyama, L. S. 2005. Physical attractiveness in adaptationist perspective. In The Handbook of Evolutionary Psychology, ed. D. M. Buss. Hoboken, NJ: John Wiley & Sons, Inc. Sweeney, M. M., and M. Cancian. 2004. The changing importance of white womens economic prospects for assortative mating. J Marriage Fam 66(4):10151028.

Downloaded by [MPI Fuer Ethnologische Forsc] at 09:12 20 February 2013

256

E. A. McClintock

Downloaded by [MPI Fuer Ethnologische Forsc] at 09:12 20 February 2013

Symons, D. 2005. Adaptationism and human mating psychology. In The Handbook of Evolutionary Psychology, ed. D. M. Buss. Hoboken, NJ: John Wiley & Sons, Inc. Taylor, P. A., and N. D. Glenn. 1976. The utility of education and attractiveness for females status attainment through marriage. Am Socl Rev 41(3):484498. Townsend, J. Marshall, and G. D. Levy. 1990. Effects of potential partners physical attractiveness and socioeconomic status on sexuality and partner selection. Arch Sex Behav 19(2):149164. Udry, J. R. 1977. The Importance of being beautiful: A reexamination and racial comparison. Am J Soc 83(1):154160. Udry, J. R. 1981. Marital alternatives and marital disruption. J Marriage Fam 43(4):889897. Umberson, D., and M. Hughes. 1987. The impact of physical attractiveness on achievement and psychological well-being. Soc Psy Quart 50(3):227236. Van de Rijt, A., and M. M. Macy. 2006. Power and dependence in intimate exchange. Social Forces 84(3):14551470. Wardle, J., J. Waller, and M. J. Jarvis. 2002. Sex differences in the association of socioeconomic status with obesity. Am J Public Health 92(8):12991304. Wedekind, C., and S. Furi. 1997. Body odor preferences in men and women: do they aim for specic MHC combinations or simply heterozygosity? Proceed: Biol Sci 264(1387):14711479. Wiederman, M. W. 1993. Evolved gender differences in mate preferences: Evidence from personal advertisements. Ethology Sociobiol(14):331352. Wiederman, M. W., and S. R. Hurst. 1998. Body size, physical attractiveness, and body image among young adult women: Relationships to sexual experience and sexual esteem. J Sex Res 35(3):272 281. Wiseman, C. V., J. J. Gray, J. E. Mosimann, and A. H. Ahrens. 1992. Cultural expectations of thinness in women: An update. Int J Eating Dis 11(1):8589.

Appendix 1. Personal Benets and Social Cost of Sexual Activity


These two indices are derived from a series of questions about the respondents perception of the costs and benets of sexual activity and from a series of questions regarding the respondents perception of her or his parents attitudes. In the rst series of questions, respondents read a series of statements beginning with If you had sexual intercourse, and respond on a 5-level scale from (1) strongly agree to (5) strongly disagree. Statements include your friends would respect you more, your partner would lose respect for you, afterward, you would feel guilty, it would upset [name of mother], it would give you a great deal of physical pleasure, it would relax you, it would make you more attractive to [the other sex], and you would feel less lonely. In the second series of questions, respondents are asked how their parents would feel about the respondent having sexual intercourse at this time in your life, having sexual intercourse with someone who was special to you and whom you know welllike a steady [girlfriend/boyfriend], and using birth control at this time in your life. These questions are asked separately about each parent. Two indices emerged from exploratory factor analysis (two Eigenvalues were above one). The variables measuring perceived parental attitudes, whether the partner would lose respect for the respondent, whether the respondent would feel guilty, whether having sexual intercourse would upset the respondents mother, and whether having sexual intercourse would make the respondent feel less lonely load primarily on the rst index, which I call the social costs of sexual activity index (Cronbachs alpha is .86). The variables measuring the remaining attitudes, whether having sexual intercourse would make the respondent more attractive, whether it would be pleasurable, whether it would be relaxing, and whether

Gender Differences in Revealed Sexual Preferences

257

it would cause the respondents friends to respect her or him more load onto the second index (Cronbachs alpha is .71). I call this the personal benets of sexual activity index. Indices are constructed by averaging scores on the component variables, after reversecoding if necessary.

Appendix 2. Self-esteem
This is an index constructed from an abbreviated version of the Rosenberg self-esteem scales. The respondent is asked to agree or disagree (on a 5-level scale from (1) strongly agree to (5) strongly disagree.) to the following statements: you have a lot of good qualities, you have a lot to be proud of, you like yourself just the way you are, you are doing everything just about right, you feel socially accepted, and you feel loved and wanted. The index is constructed by averaging scores on the component variables. Cronbachs alpha is .85.

Downloaded by [MPI Fuer Ethnologische Forsc] at 09:12 20 February 2013

Appendix 3. Parental Socioeconomic Status Index


This variable is an index made of the respondents mothers and fathers education and occupational status (measured by the seven-level Hollingshead index) and of family income. Education, occupational status, and income are standardized prior to constructing the index and the index itself is also standardized. The index is constructed by averaging scores on the ve component variables. When possible, information is taken from the parental interview in Wave I. Otherwise it is taken from the respondents report about the parent. Still, despite these two sources, a substantial number of cases are missing data, especially on the father. For example, many adolescents do not know about the education and occupation of their nonresident fathers. Data are also missing on occupational status when the parent is not in the labor force. When data are missing (for either reason), the index is constructed from whatever measures are available. (Cronbachs alpha is .76).

You might also like