Today, gender stereotypes run so deep in popular culture that women and men are portrayed almost as members of different species. These stereotypes in turn shape gender roles that dictate sexual behavior, from preliminary dating rituals to the bedroom (Canary, Emmers-Sommer, & Faulkner, 1997). Wood and Eagly’s (2002) biosocial model suggests that these gender differences in attitudes and sexual behavior result largely from conformity to socially defined gender roles: Women know that they are expected to be expressive and supportive; men know that they are expected to be withdrawn and preoccupied with sex. Wood and Eagly’s (2002) model suggests that these differences are neither stable nor innate, but change depending on social and cultural context. Thus women and men demonstrate stereotypical differences to the degree that they identify with socially defined gender roles.

Sexually restrictive gender roles are socially damaging because they polarize behavioral expectations for women and men and create self-handicapping and minimally overlapping cognitive frameworks. In effect, gender roles restrict the sexual behavior of women and expand that of men. Stereotypically—albeit questionably in practice—men initiate sexuality, and women guard the gates of sexuality (Allgeier & Royster, 1991; McCormick & Jesser, 1983). A deeper understanding of the processes that give rise to socially defined, sexually restrictive gender roles is an intriguing and increasingly important area of psychological research.

Research has shown that people endorse the stereotype that men are more interested in sex than women are (e.g., Broverman, Vogel, Broverman, Clarkson, & Rosenkrantz, 1972; Deaux & Major, 1987), and, to some extent, the literature supports this stereotype. For example, compared to women, men are more sexually permissive, endorse premarital and casual sex at higher rates, and masturbate more often (Oliver & Hyde, 1993). S. S. Hendrick and C. Hendrick (1995) also found that men are more likely than women to seek out or initiate sexual activity. It is not surprising then that men report a greater interest in sex, greater frequency of sex, and greater enjoyment of sex than women do (Murstein & Tuerkheimer, 1998).

These gender differences in attitudes and sexual behaviors warrant a brief review of research on sexual motivation in women and men. Klusmann (2002) found that men tend to report a greater desire for sex, whereas women report a greater desire for tenderness. Other research suggests that men pursue sex for physical release, whereas women pursue sex to show and receive affection (Canary et al., 1997; Sprague & Quadagno, 1989). Also, Browning, Hatfield, Kessler, and Levine (2000) found that women are more likely to initiate sex for love and pleasure, and men are more likely to initiate sex for recognition by their partners (i.e., to be seen as a “good lover”). In other words, women prefer emotional intimacy to sexual intimacy, and the reverse is true for men.

Evolutionary psychology offers its own explanation of these observed differences between women and men. Buss and Schmitt (1993) proposed that because of differences in biology, women and men adopt different sexual strategies. According to this view, a man benefits from sex with as many fecund partners as possible, whereas a woman benefits from a reliable, long-term partner who provides resources to her and their children. In this way, evolutionary psychologists conceptualize women as less interested in physical sex and more interested in securing resources through a long-term relationship with a mate.

Though much theory and empirical research suggest differences between women and men in attitudes and sexual behaviors, opposing evidence also points to similarities. Implications of the Wood–Eagly model of situation-specific gender differences appear throughout the literature; gender differences appear to be contextually dependent, as opposed to universal. For example, as mentioned, S. S. Hendrick and C. Hendrick (1995) found gender differences in their study of sex, but most of the variables they measured (49 of 60) supported similarities between genders. And, although Oliver and Hyde (1993) also found some gender-based differences in attitudes toward sex, many of these differences were overshadowed by similarities in ratings of sexual satisfaction, incidence of oral sex, and attitudes toward masturbation. Browning, Kessler, Hatfield, and Choo (1999) found that the types of sexual behavior did not differ by gender. However, significant correlations emerged between the types of sexual behavior and such contextual factors as power, status, and personality type. In two of four samples of Heesacker et al. (2007, manuscript in preparation), men reported a significantly greater desire for sex in relationships than women did. However, gender differences in desire for sex were inconsistent across the four samples, and the two significant differences were small effects according to Cohen’s (1988) standards.

These gender similarities may indicate a convergence, over time, between the attitudes toward sex and sexual behaviors of women and men. Oliver and Hyde (1993) found that gender differences in attitudes toward sex and sexual behavior had decreased in magnitude from the 1960s to the 1980s. This work set the stage for Baumeister’s (2000) review of changes in women’s sexuality since the 1960s, which supported the idea that gender-based differences in attitudes and sexual behaviors are diminishing. Similarly, Dekker and Schmidt (2002) found that women as well as men are beginning to masturbate at an earlier age than was the case in the 1960s or the 1980s. This change in sexual behavior implies a corresponding change in attitudes toward sex over time.

As this summary indicates, the published research paints a complex and sometimes contradictory picture of evolving attitudes toward sex and sexual behaviors. In the present study—rather than starting with beliefs about sex as articulated by researchers—we took an inductive approach, by examining beliefs about sex as articulated by the participants. Most published studies regarding gender differences in sexuality are constructed to produce data to support an a priori hypothesis. Such an approach runs the risk of finding support for relatively small and perhaps unstable effects while ignoring larger and more enduring effects. In contrast, the current project was designed to maximize ecological validity and to allow conclusions to emerge naturally from the narratives of the participants, rather than to impose upon them a possibly ill-fitting narrative. The inductive approach holds promise as a way to gain a fresh understanding of differences and similarities in women’s and men’s beliefs about sex. The current study addressed two central questions: (a) Do women and men differ in their self-reported beliefs about sex, and (b) why or why not?

Method

The study proceeded in two phases with participants from a single, large undergraduate psychology class on separate days. Student-generated items from Phase 1 were used to construct a questionnaire for Phase 2.

Phase 1

Participants

Participants were 380 students (219 women, 161 men) enrolled in a lower-level undergraduate psychology of personal adjustment course at a large university in the southeastern United States. Age and ethnicity demographics were collected for Phase 2 participants, but not for Phase 1 participants; for clarification, these demographics appear under “Phase 2.” There was significant overlap among participants in Phase 1 and 2 because participants for each phase were drawn from the same class. Because the data were collected anonymously, the precise degree of overlap is unknown, but it is reasonable to expect that demographics are similar for the two samples. Participants received two extra credit points in the course for completing both phases of the study; one extra credit point was awarded after each phase. Other in-class surveys were available to participants who did not want to partake in the current study.

Measures and procedures

After participants completed an informed consent form, they responded to two open-ended questions regarding their attitudes toward and beliefs about sexuality and sexual behavior. To avoid dual role complications, the instructor did not present the survey. The researcher read the two questions aloud during class, and participants received approximately 20 min to respond in writing. The first statement was “On a sheet of paper please list all of your thoughts that influence whether you will or won’t have sex. For example: ‘His body looks sexy to me,’ ‘I always feel sexier when I am drinking,’ and ‘My roommate is gone for the weekend.’” The second statement was “Now on the same sheet of paper draw a line, and below that line, list the opinions of other people that affect you regarding whether you will or won’t have sex. And next to each opinion tell whose opinion it is. For example: ‘People should not have sex before marriage: father,’ ‘Forget about AIDS, it is a gay disease: girlfriend,’ and ‘If you want to sleep with someone, go to the bars and drink: best friend.’”

We utilized what Aiken (1996) called the inductive method of test construction, wherein participants provide items that are then factor analyzed. Our participants provided 160 non-redundant items. All items reflected participants’ verbatim beliefs about and attitudes toward sex and sexuality. All of these items, with some minor editing for clarity, underwent factor-analysis and were then used in Phase 2.

Phase 2

Participants

As noted, participants in Phase 2 were drawn from the same undergraduate class as those in Phase 1. Participants were 345 undergraduate students (208 women, 131 men, 6 gender unreported, mean age = 18.6 years). The ethnic self-identification of participants was as follows: European American/non-Hispanic = 260, Hispanic American = 37, African American = 12, Asian American = 16, Native American (American Indian) = 1, and Other = 5; 14 respondents did not answer this question. After they had completed Phase 2, participants received the second point of extra credit.

Measures and procedures

Participants completed an informed consent form, and, again, to avoid dual role complications, the instructor did not present the survey. The 160 statements about sexual behavior and attitudes from Phase 1 appeared in paper format, and participants in Phase 2 rated each of them from 0 = “strongly disagree” to 9 = “strongly agree.” Three additional questions examined intentions in risky sexual behavior and assessed participants’: (a) intent to have sex, (b) intent to use alcohol as a way to increase the chances of having sex, and (c) intent to use alcohol to increase sexual pleasure. Participants were given an entire class period (up to 50 min) to complete the survey. As with the beliefs data from Phase 1, the data from these items were examined to generate deductive gender-based theories rather than to test any a priori hypotheses.

Results

Creation of four factors

The first analysis of the belief items was a common factor analysis (CFA) and a scree test (Cattell, 1966) to identify underlying factors; four factors were retained. An oblique rotation was then performed, and we kept the items that achieved a simple structure (Thurstone, 1935; see also Cattell, 1952, 1966). Simple structure was defined in the present study as the highest eigenvalue of each item that achieved an absolute value of at least .4 and all other eigenvalues that achieved an absolute value of no more than .3. The rationale for simple structure is that items that load on one and only one factor best reflect the underlying factor structure. The results of this procedure appear in Table 1.

Table 1 Factor loadings of attitude items retained from an oblique four-factor rotation.
Full size table

The attitude factor labeled “Sex as Personally and Physically Pleasurable” reflects the extent to which participants see sex as beneficial in that it feels good physically and bestows positive intrapersonal benefit (e.g., improved self-confidence) or interpersonal benefit (e.g., meeting the expectations of a peer group). Factor 1 also includes items that reflect beliefs that alcohol is instrumental in having or improving sex. Examples of items from Factor 1 are “I should get drunk to enhance my sexual experience,” “I should have sex with as many people as possible,” and “Everyone is having sex, so I should be also.” Factor 1 consists of 13 items (See Table 1).

The second attitude factor reflects the potential of sexual relationships to enhance the self and others. It includes items such as “I feel attractive after sex,” “Sex makes me feel secure in my relationship,” and “I trust my partner more after sex.” This factor was labeled “Sex as Beneficial to the Self-in-relationship.” Fourteen items are included in Factor 2 (See Table 1).

The third attitude factor addresses the practical issues of sexual relationships. Items that loaded on this factor reflect concerns with issues such as protection against STDs and pregnancy. Items that typify this factor are “Sex might ruin my relationship,” “S(he) will leave the relationship if I give him/her sex,” and “I will get a STD by having sex.” This factor was labeled “Sex as Personally Costly.” Factor 3 contains 10 items (See Table 1).

The last attitude factor reflects moral, ethical, value-driven, and practical concerns regarding sex. More specifically, this factor includes items that prohibit sexual behavior. Example items include “I should not have sex because I’m too young,” “I should be with my partner a long time before I have sex,” and “I should not have sex because it is a sin.” This final attitude factor was labeled “Sex as violation of Social Injunctions.” Eight items constitute Factor 4 (See Table 1).

Coefficient α’s for the four retained attitude factors were: Sex as Personally and Physically Pleasurable, α = .85; Sex as Beneficial to the Self-in-relationship, α = .88; Sex as Personally Costly, α = .87; and Sex as a Violation of Social Injunctions, α = .77. The intercorrelations varied widely in magnitude (see Table 2). The largest overlap was .33 (p < .01), between the Sex as Personally and Physically Pleasurable factor and the Sex as Personally Costly factor, and the smallest overlap was .09 (p > .10), between the Sex as Personally and Physically Pleasurable Factor and the Sex as Beneficial to the Self-in-relationship factor.

Table 2 Correlation matrix of factors.
Full size table

Gender differences

We analyzed differences between women and men using four Bonferroni corrected t-tests, with gender (women vs. men) serving as the independent variable and with each of the four attitude factors serving as the dependent variables. Results of these analyses reveal that women and men significantly differed on three of the four attitude factors (see Table 3 for means and standard deviations). Men scored significantly higher than women on the Sex as Personally and Physically Pleasurable factor, t(338) = −11.05, p < .0001. The effect size for this factor was 1.20, which is a large effect according to Cohen (1988). Men scored significantly higher than women on the Sex as Personally Costly factor, t(338) = −8.49, p < .0001. The effect size here was .92, which is a large effect. Women scored significantly higher than men on the Sex as a Violation of Social Injunctions factor, t(338) = 2.97, p < .004, and the effect size was .32, which is a small effect. Women and men did not differ significantly on the Sex as Beneficial to the Self-in-relationship factor.

Table 3 Summary of t-tests for gender differences on factors (n = 338).
Full size table

Three t-tests indicated that men intended to engage in the behaviors assessed by the intention measures to a greater degree than women did. For “I intend to have sex with another person sometime within the next 30 days,” the mean for women was 4.20 (SD = 3.90) and for men was 5.87 (SD = 3.41), t(337) = −4.04, p < .0001. However, the effect size was small, d = .44. For “I intend to drink alcohol as a way to increase the chances that I will have sex with another person sometime within the next 30 days,” the mean for women was 0.85 (SD = 1.94) and for men was 2.69 (SD =3 .24), t(338) = −6.54, p < .0001. In this case, the effect size was medium, d = .71. For “I intend to drink alcohol as a way to increase my sexual pleasure with another person within the next 30 days,” the mean for women was 1.09 (SD = 1.99) and for men was 2.39 (SD = 2.84), t(337) = −4.99, p < .0001. The effect size was .54, a medium-sized effect.

As a result of the significant gender differences on three of the four attitude and belief factors, we conducted follow-up regression analyses to assess possible gender differences in the prediction of sexual intentions. The interactive effects of gender combined with each of the attitude and belief factors served as predictor variables in these regressions. The three intention measures each served as the criterion variables in the regressions. Results from these analyses yielded no significant gender interaction effects.

Discussion

The purpose of the current study was to examine stereotypes and sexually-restrictive gender roles, as evident in the actual expressed beliefs of participants, not to validate an a priori hypothesis about those beliefs. Existing literature provided mixed evidence about the consistency of gender differences in sexuality, often showing more similarities than differences. Our data show that women and men significantly differed on three of four sets of sexual beliefs, sometimes with large effect sizes. The findings raise the question of whether these differences are innate or are the result of restrictive gender roles to which women and men often conform.

Factor 1: sex as personally and physically pleasurable

Men endorsed at a much higher rate than women did the Sex as Personally and Physically Pleasurable factor. The very large effect size of 1.20 suggests that, compared to the other factors, the gender difference here was the largest. Many of the items in this factor concern risky sexual behaviors, such as multiple partners and use of alcohol before sex. Our finding that men are more inclined than women are to engage in risky sexual behaviors concurs with that of earlier research (e.g., Sprague & Quadagno, 1989).

Six items in this factor deal with sexual frequency and—more centrally to our discussion—public opinion about individuals because of sexual frequency. Items such as “I should have sex a lot so people will be impressed” or “Everyone is having sex, so I should be also” strongly resonate of the social norms that surround sex. The close ties between these norms and gender roles may explain the observed difference between women’s and men’s item endorsements in the current study. Gender roles restrict women in the sense that, if a woman has frequent sex, others (both men and women) may deem her promiscuous. Equally—though not as frequently talked about—gender roles restrict men to a narrow range of acceptable sexual behavior in the sense that others deem him immature and un-masculine if he does not have frequent sex (e.g., the popular 2005 film comedy, “Forty Year Old Virgin”). Among members of the demographic group represented by the sample population, a man romantically devoted to a single partner may be derided by others as “whipped” or “wrapped around” the partner’s “little finger.” The underlying sexually-restrictive gender roles, which probably account for most of the gender difference on Factor 1, appear to dictate how women and men view even their own sexuality. A clearer understanding of this point may help women and men to break free of their restrictive gender roles.

Because six of the 13 statements that load on this factor concern alcohol use, it is important to consider gender difference in drinking patterns between college women and men. Wechsler, Dowdall, Maenner, Gledhill-Hoyt, and Lee (1998) sampled 14,520 college students from across the United States, and they found that 38.9% of women binge drink, whereas 48.4% of men do (binging was defined as four drinks in a row for women and five for men, at least once within the past 2 weeks). They also found that 23.8% of women reported having been drunk three or more times in the past month, whereas 33.7% of men did. These gender differences in self-reported drinking styles are fairly large. If these self reports are accurate, and if the Wechsler et al. sample of collegians reasonably represents the current sample, gender differences in drinking patterns among participants could account for at least some of the gender differences found in Factor 1. Also, fear of date rape after alcohol use might be a strong deterrent for women (but not for men) to drink in sexual settings. To get a clearer understanding of the source of the Factor 1 gender difference, future researchers could statistically control for the amount of alcohol that participants consume. However, because the other seven of 13 items in this factor make no mention of alcohol, it is unlikely that gender-based drinking patterns fully explain the observed differences.

Factor 2: sex as beneficial to the self-in-relationship

We also found that women and men endorse similar beliefs about sex in the context of intimate relationships. Both genders thought that sex was beneficial in creating self-directed positive feelings and in establishing and maintaining healthy relationships. Although men may want more risky sexual behavior outside of a relationship than women do, when sex is examined within the context of an intimate relationship, women’s and men’s responses were much more alike than different. Women and men both expressed that sex was important in bringing partners closer together. Perhaps, then, the sexually restrictive gender roles examined in the current study apply primarily to casual sex. Both genders appear to experience much more sexual freedom from society’s gender roles once they enter a relationship.

A possible explanation for this gender similarity is that people in romantic relationships may give more primacy to their own feelings and their partners’ than they do to social expectations about sexual behavior. Ongoing research in a related realm (Tiegs & Heesacker, 2005) supports this interpretation with the finding that people tend to endorse gender-stereotyped beliefs about what women and men desire sexually from romantic relationships. Yet when the same participants told what they, themselves, desired from relationships, their answers did not match the gender stereotypes they had previously endorsed. Participants’ gender-stereotyped beliefs about what women and men typically want sexually in a relationship were simply inaccurate.

Factor 3: sex as personally costly

The items in Factor 3 concern the concrete physical and emotional consequences of sex much more than they do sexually-restrictive gender roles. In that sense, these items relate to the items in Factor 1, which showed that, compared to women, men believed more strongly than women that risky sexual behaviors are desirable. The data from Factor 3 extended the findings in Factor 1 by showing that men also believed more than women did that negative outcomes result from such risky sexual behaviors. This effect size of .92, like that of Factor 1, was also large, which indicates a large difference between women’s and men’s responses. Perhaps because men engage more often in risky sexual behaviors than women do, men reported sex as more personally costly. The more partners and the more sex one has, the more likely one is to encounter consequences, such as unwanted pregnancies and sexually transmitted diseases. Men in our sample appear to walk a fine line between wanting the risky sex that society says they should have and paying the price for having had it.

Factor 4: Sex as a violation of social injunctions

Women believed more than men did that being sexually active has a greater number of negative social ramifications and violates the social expectations of others, though the observed difference produced only a small effect size of .32. This finding is consistent with gender role and social constructionist theories (e.g., Wood & Eagly, 2002), which hold that women’s and men’s different socialization processes create their different beliefs about sex. Women endorsed at higher rates than men did such items as waiting longer to have sex and not engaging in premarital sex, which is consistent with the notion of women as sexual gatekeepers. Perhaps women are more interested than men are in waiting for the right person and the right moment to have sex. However, it is important to note that these differences might be the result of women and men conforming to their respective traditional gender roles.

Gender differences on attitude measures

Gender differences appeared on all three intention measures. Men, compared to women, intended to have, not only more sex, but more risky sex by using alcohol to increase the chances of having sex and to increase sexual pleasure. Thus, the gender differences in beliefs about sex translated directly into concrete intentions for women and men to engage in risky sexual behaviors. To examine this link specifically with respect to gender, three multiple regressions were performed in which the interaction term of gender and belief-about-sex factors served as a predictor of sexual intention. Because these regressions yielded no significant results, the magnitude of the association between beliefs and intentions did not differ as a function of the gender of the person holding those beliefs. In other words, the discrepancies between women’s beliefs and intentions, and between men’s beliefs and intentions, were similar. This finding suggests that being a woman or a man does influence sexual intentions, but not above and beyond the specific gender-based sexual stereotypes that give rise to the respective gender roles.

Social and contextual implications of the data

Our data help elucidate a complex picture of how women and men approach sex and provide support for a social influence perspective (e.g., Wood & Eagly, 2002) on beliefs and sexual behaviors as reported by our participants. Though the data from the current study appear to suggest that men are more focused on the physical aspects of sex and that women are more focused on its emotional aspects, gender roles alone may be responsible for this difference. Sexually restrictive gender roles too often become self-fulfilling prophesies because women know that they are expected to be less sexual than men, and men know that they are expected to be more sexual than women (Shields, 1995). Perhaps one of our study’s most interesting findings is that participants’ responses differed when their focus was behavior within relationships vs. behavior outside of relationships. Accordingly, the data suggest that gender roles of women and men in relationships differ from their gender roles outside of relationships, a finding consistent with previous research (Vogel, Tucker, Wester, & Heesacker, 1999; Vogel, Wester, Heesacker, & Madon, 2003). This finding also supports the notion of gender role plasticity—rather than behavioral invariance as a function of gender, which suggests that women and men, therefore, have the behavioral flexibility to conform in varied degrees to sexually restrictive gender roles.

Limitations and future research

As with much of the available research, the current study is limited in its employment of a convenience sample. Future researchers would benefit by including a greater diversity of people from different ages, religious affiliations, ethnicities, and sexual orientations. Examination of the degree to which any of these characteristics predicts stereotypes about sexuality would strengthen the field’s understanding of gender-role conformity. In addition, our participants completed the surveys in large groups. With such sensitive material, participant assessment in smaller groups could ensure more honest responses and eliminate excessive variance in responses due to perceived or actual social group pressures (e.g., comments made by classmates, fear of others seeing their responses). On the other hand, this same large group setting was used for university examinations, so chairs were far enough apart that participants could not readily see the responses of neighbors. Another limitation of the current research is that no information was collected regarding the relationship status or sexual orientation of the participants. Because of the difference in result patterns among Factor 2 and the other three factors, outcomes could likely vary depending on whether a participant is currently in a marriage or committed relationship vs. not dating anyone, for example. Also, potentially-different norms for gay or lesbian couples than for heterosexual couples could have impacted the results.

Future researchers might examine the accuracy of women’s and men’s stereotypes toward sexuality. Do women’s and men’s stereotypes reflect the actual degree to which they conform to their respective gender roles? Do men actually experience sex as more physically and personally beneficial? Also, do men experience more of the negative personal costs of having risky sex? Do women feel the negative impacts of breaking social injunctions more than men do?

One strength of the present study is its method, which is an inductive approach to exploring the differences and similarities of beliefs between women and men about sexual behavior and gender roles. By asking participants what they believe about sex, we were able to compile a list of individually relevant and ecologically valid items. This inductive approach not only grows from the beliefs of real people, it also offers insight into similarities and differences between women and men, based on their own experience.