Abstract
Using the Online College Social Life Survey, we examine whether the sex ratio of the student body of a college or university affects whether heterosexual students hook up, have relationships, have intercourse, or have attitudes favorable toward casual sex. The gendered dyadic power model predicts that, if men are more interested in having sex than women, as the ratio of women to men goes up, men will increasingly have the upper hand and more sex will occur. Consistent with the prediction, we find that where the ratio of women to men is higher, students of both sexes hook up more and accumulate more sexual partners, but inconsistent with it, students are no more likely to have intercourse in a given hookup where the ratio of women is higher.
A casual observer of the contemporary academic and popular discourse surrounding college students’ sexuality might infer that “hooking up” is creating a sexual crisis on American college campuses. There has been an explosion of recent research on the topic by social scientists (Allison and Risman 2013, 2014; Armstrong, England, and Fogarty 2010; Armstrong, Hamilton, and England 2010; Bogle 2008; England, Shafer, and Fogarty 2008; Hamilton and Armstrong 2009; Owen and Fincham 2011; Paul, McManus, and Hayes 2000; Regnerus 2012; Regnerus and Uecker 2011; Reid, Elliott, and Webber 2011; Uecker and Regnerus 2010) and journalists (Stepp 2007). “Hooking up,” as students use the term, involves anything from “making out” to intercourse, with no necessary implication of an interest in a relationship. Some writing on this new social form suggests that hookups are a problem to be solved and evidence of heightened exploitation of women. Others argue that casual sex shows women’s empowerment to seek sexual pleasure for its own sake. We address one aspect of this debate: whether the scarcity of men on some college campuses enhances their power and leads to more hooking up, fewer relationships, more intercourse, and attitudes more accepting of hooking up and casual sex.
Why would sex ratios affect sex? Guttentag and Secord (1983) famously argued that sex ratios affect men and women’s relative power to be treated better in sex or relationships. Where there is a relative dearth of one sex, the other sex has to compete to find a sexual or relationship partner, and the sex in surplus may do so by offering more of whatever it is that the sex in short supply wants. They called what one sex gains when they are scarce “dyadic power,” affecting how much they can get what they want within relationships. They argued, however, that in most societies, men have much more structural power, and thus this limits the extent to which women can exercise the dyadic power if what they want is economic independence and autonomy. Thus, they predicted that what women’s scarcity would give them is access to relationships in which men would be monogamous, willing to marry, and willing to help support any joint children.
Many scholars have followed Guttentag and Secord in exploring hypotheses regarding effects of sex ratios. Trent and South (2011) found that, in China’s deeply sex-imbalanced society, the abundance of men relative to women has increased women’s number of sexual partners, likelihood of engaging in premarital sex, and likelihood of marrying before age 25. Guttentag and Secord’s ideas would have predicted that the scarcity of women would lead to marriage, on the assumption that women want committed male partners, but less premarital sex. Thus, Trent and South’s (2011) results may be driven by the simpler notion that they refer to as the “demographic opportunity” hypothesis; it implies that a lower ratio of women to men leads women to have more of all kinds of cross-sex relationships, casual and marital, simply because more men are available. One piece of evidence consistent with the dyadic power view is Harknett and McLanahan’s (2004) finding that, among couples who recently shared a nonmarital birth and were still romantically involved at the birth, the lower rates of African American couples marrying each other later were partially explained by the lower availability of men in the African American marriage market. If women of all races want marriage to the partner with whom they have recently had a baby more than men do, then this is what the dyadic power thesis would predict. It cannot be explained by simple demographic opportunity, because the women under study already had a man; they were in a relationship with the father of their child at birth—the research question is whether they went on to marry.
But marriage and a partner to share child rearing with are not generally what women who attend four-year colleges want; few think it is yet the time for either marriage or childbearing. The more salient issue may be how much they want to participate in the casual sexual behavior entailed in hookups. Given that young women are now beginning to outnumber their male counterparts at campuses around the country (DiPrete and Buchmann 2013), one team of researchers suggested that the elevated ratio of women to men is encouraging casual sex (Regnerus 2012; Regnerus and Uecker 2011; Uecker and Regnerus 2010). We build upon and go beyond their research by using a dataset of more than 20,000 students at 21 colleges and universities to test a key claim: whether students on campuses where the ratio of female to male students is higher hook up more, have intercourse more, are less likely to have long-term relationships, and show attitudes more accepting of casual sex than students at campuses where the ratio of female to male students is lower. We also test two views with differing predictions—the demographic opportunity thesis mentioned above, and a gendered culture model that posits that norms and behavior follow the preferences of the more numerous sex.
Past Research and Theory about Campus Sex Ratios
Sex Ratios and Demographic Opportunity
The “sex ratio” is the ratio of the number of members of one sex to the other (we will reverse the convention and put women in the numerator). The simplest hypothesis about sex ratios posits that women’s likelihood of having either sex or a relationship goes up when there are fewer women per man, whereas men’s likelihood of having either sex or a relationship goes up when there are more women per man. Following Uecker and Regnerus (2010) and Trent and South (2011), we call this the “demographic opportunity” thesis. It states that heterosexual men or women can more readily find a partner—for sex or a relationship or both—when more of the other sex is available. When someone has already found a partner—whether for one night or longer—this thesis has no implication for whether the partners will have sex; the external marketplace enters the picture only in affecting availability of partners, but does not affect power within dyadic partnerships.
Sex Ratios and Gendered Dyadic Power
Uecker and Regnerus (2010) used the dyadic power thesis to argue that the higher the ratio of women to men, the more men have the power to get what they want within heterosexual pairings, and, equivalently, the lower the ratio of women to men, the more women have power to get what they want in such dyads. The general idea that conditions in the partner market outside a relationship affect power within a relationship is consistent with several social science perspectives, particularly bargaining models in economics and social psychologists’ power–dependence exchange theory (Emerson 1976; Lundberg and Pollak 1996). In this view, in any dyad, Partner A has more power to get what she or he wants from Partner B if A has more alternatives than B. Because for heterosexuals the sex ratio affects how likely one is to find an alternative partner, the dyadic power thesis implies that the sex ratio affects power within any given dyad. In common sense terms, if you know you can be easily replaced, but you would have a hard time finding another partner, you are more likely to give your partner whatever he or she wants. If repeated behavior changes norms, then we would expect that a higher ratio of women to men will also change beliefs in the direction of whatever is seen as better by the sex in short supply.
To predict what will happen in the heterosexual arena when sex ratios increase one sex’s dyadic power, we need to hypothesize about differences in what men and women want. Guttentag and Secord (1983) implicitly assumed that women want monogamy and commitment from men to raise children. Issues of how children will be raised are not terribly relevant to the life cycle stage in which most of today’s four-year college women find themselves. What is relevant is whether sex is confined to monogamous relationships or occurs in casual hookups. Uecker and Regnerus (2010) and Regnerus and Uecker (2011), while acknowledging that many women indeed like sex, argue that men have more of an interest than women in casual as well as relational sex, and that women more often prefer not to have sex without a relational commitment.
Does the evidence support the assumption by past authors that men want casual sex more than women? Baumeister, Catanese, and Vohs (2001) reviewed substantial cross-cultural evidence for the claim. Bogle’s (2008) qualitative-interview study of largely white students attending residential universities suggests that men do indeed want no-strings-attached sex more than women, who expressed more desire for relationships. England and Bearak (2014), using the data we use here, did not find women any more interested in relationships than men, but found that men report greater interest in hooking up and are less apt to say that they will not have sex without love. Overall, the evidence suggests that, on average, men are more interested than women in casual sex.
Of course, studies that find gender differences in averages always find that men and women have overlapping distributions. For the dyadic power thesis to predict more college sex when women are more numerous, it is not necessary for all men to want or all women to eschew casual sex, although the magnitude of the sex ratio effect may depend on the magnitude of the difference in preferences.
Where does the gender difference in preference for casual sex come from? Some authors rely on evolutionary arguments, suggesting “hardwired” preferences (Baumeister et al. 2001; Regnerus 2012; Regnerus and Uecker 2011). There are also plausible social explanations for why women would have more hesi-tation about casual sex. One is the double standard; women risk more reputational damage for casual sex because they are judged more harshly than men for nonrelational sex (Armstrong, England, and Fogarty 2012; Bogle 2008; Conley 2011; see also Allison and Risman 2013). Another explanation may be that women (perhaps correctly) perceive more danger from prospective casual sex partners than men do; Conley’s (2011) experimental evidence shows that if vignette manipulations equalize the perceived danger to men and women when a stranger invites them to have casual sex, much of the gender difference in response disappears.
We will form hypotheses assuming, as past authors have, that, on average, men are more interested in casual sex than women. We note that if, under suitable controls, schools with a higher ratio of women to men feature more sex and lower odds of relationships for both men and women, this would be evidence both that a surplus of women lowers women’s dyadic power and that women are less desirous of sex and/or more desirous of relationships than men. We call this position the gendered dyadic power theory, because it not only predicts that sex ratios affect power but does so under an assumption about sex differences in preferences.
One published study on college life provides evidence regarding the gendered dyadic power thesis. Uecker and Regnerus (2010) drew upon a survey of 986 women spread across a large number of coeducational colleges and universities. Using a statistical model with numerous controls, they find that women attending schools with a higher ratio of women to men are less likely to have a boyfriend but more likely to have had sex. If women in schools with less men per woman had been less likely to have a boyfriend and less likely to have sex, we could conclude that demographic availability explained the findings—there just were not enough men available, so some women had to forego either relationships or sex. As women have more sex and fewer relationships when there are fewer men available, the authors take this as evidence that, when men are scarce, they are more able to get women to “give them” sex without the commitment of a relationship. One limitation of the data used by Uecker and Regnerus is that it was limited to women, so they were not able to test whether men in schools with fewer men per woman had more sex, as is also predicted by their hypothesis. This limitation is addressed by Uecker’s (2012) unpublished study using the National Longitudinal Study of Freshmen, a sample of students from 28 highly selective colleges and universities. He found, under controls, that a higher ratio of women to men increases the probability that freshmen women have had sex, but does not have a significant effect on freshmen men’s sex. Because the dataset includes only first-year students, it is possible that the older men in the schools with higher ratios of women are the beneficiaries of greater male dyadic power, but the analysis does not allow Uecker to test this. One clear gap in past research is that it does not provide any evidence to support the prediction from the gendered dyadic power thesis that men at schools with a high ratio of women have more sex. One advantage of the dataset we use is that it includes male and female undergraduate students of all grade levels, allowing us to test effects of sex ratios on men’s behavior as well as women’s.
Sex Ratios and Gendered Culture
We offer another possible theoretical perspective that has seen little development in prior literature. It suggests that sex ratios affect college campus culture at least partly through interaction, such that whatever women start out preferring will become a more predominant view on a college campus to the extent that women numerically dominate the institution (and analogously when men are more numerous). The idea is that intergender interaction can change people’s views, and men are more likely to accept a view common to women when, because more of their classmates are women, they have more exposure to women’s view. This cultural change may also lead behavior to be more consistent with the preference of the sex with the greater numbers. The effect of sex ratio on culture might be linear, or it may be nonlinear if it is only present when organizations’ sex ratios move beyond some tipping point that they become gendered in new ways (Acker 2006; Risman and Davis 2013). It may be that, at universities where women have a substantial numerical edge, their lower dyadic power is superseded by their power to shape norms at the campus level.
Hypotheses
Based on the above theories, we test three hypotheses. The second and third hypotheses hold only under the assumption of a gender difference in preference regarding casual sex and relationships. But they require no assumption about the source of any such difference in preference—whether it is evolutionary, a result of differential socialization, or a result of different incentives because of a double standard in how men and women are judged for casual sex.
Demographic Opportunity Hypothesis
In schools with a higher ratio of women to men, women will have fewer sexual partners, hook up fewer times, and be less likely to have had a relationship, whereas men will do more of all these things. This thesis makes no prediction about effects of sex ratios on whether, conditional on being in a hookup, one will have sex.
Gendered Dyadic Power Hypothesis
In schools with a higher ratio of women to men, both men and women will have more sexual partners, hook up more times, be more likely to have sex on hookups, be less likely to have had a relationship, and express attitudes more accepting of casual sex.
Gendered Culture Hypothesis
In schools with a higher ratio of women to men (or, if there is a critical threshold, at least after a certain level of the ratio), both women and men will have fewer sexual partners, hook up fewer times, be less likely to have sex on hookups, be more likely to have had a relationship, and express attitudes less accepting of casual sex.
Data and Measures
Our data come from the Online College Social Life Survey (OCSLS) and are available upon request from the second author, who was the principal investigator. Between 2005 and 2011, more than 23,940 students at 21 four-year colleges and universities took the online survey. The universities include elite private schools (e.g., Stanford and the University of Pennsylvania), flagship state universities (e.g., Ohio State and Indiana University), small liberal arts colleges (e.g., Ithaca College, Whitman College), and middle-tier state schools with more commuter students (e.g., University of Illinois-Chicago). Together, these schools cover all major geographic regions of the continental United States.
Undergraduate college students were re-cruited in classes that offered credit for taking the survey (or offered an alternative assignment taking the same amount of time). 1 While it is not a probability sample, within classes, there was no significant nonresponse; nearly every student opted for the survey. Thus, who opted into the survey within a class is not a source of bias. Rather, any bias in the sample is a result of nonrepresentiveness of the chosen schools or of those who took the (mostly sociology) courses hosting the survey. Yet sociology majors were only approximately 10 percent of the sample. While a probability sample would be preferred, there are no other datasets available with this much detail on sexual behavior in the college hookup context.
The overall sample size is 23,940 cases. Given our interest in heterosexual relations, our analytic sample was limited to students who reported that their sexual orientation was heterosexual. We also deleted the small portion of students over 24 years of age, as they are unlikely to be involved in campus social life. The final analytic sample is 18,974 men and women. Within this analytic sample, 69 percent are female, 66 percent are white, and the mean age is 19.75 years (with a range of 18–24). This is slightly more female and slightly less white than in the population of U.S. college students as a whole (U.S. Census Bureau 2010).
Dependent Variables
Hookup participation
Students were asked a series of questions about hooking up that began with the lead-in: “For this section, use whatever definition of hookup you and your friends generally use.” They were asked, “Since you started college (counting during summers or during a term abroad),” (1) “how many people have you hooked up with whom you didn’t know before that night” and (2) “how many people have you hooked up with in the case where you were not already in a romantic relationship with the person but you did know him or her?” Stranger Hookups, which are arguably more casual, is a count variable originating from the first of these two questions. The mean number of stranger hookups is 1.68 (see Table 1 for means on all variables). Acquaintance Hookups is a count variable originating from the second question. The mean number of acquaintance hookups is 2.42; students are more likely to hook up with someone they know than with a stranger (Table 1). 2
Means and Standard Deviations on All Variables for Men, Women, and All Students.
Note. Standard deviations are in parentheses. For indicator variables, means of the 0 and 1 values are the proportion of cases in the named category.
Total number of sexual partners
It is a measure of the respondent’s total number of intercourse partners in their life, regardless of context (hookups, relationships, etc.). After being asked the age at which they first had “vaginal intercourse” (or whether they had not experienced it), respondents who had had intercourse were then asked, “How many people have you had intercourse with?” Response options in a dropdown box were all integers 1 to 10, and then three grouped categories of 11 to 15, 16 to 20, and 21 and over. To compute means, and assign count scores for our regressions, we assigned 0 to those who had not had intercourse, and for those students in the grouped categories, we assigned them the midpoint of 11 to 15 and 16 to 20, and assigned 21 to all those in the highest category. Using this procedure, students had an average of 3.13 intercourse partners (Table 1).
Sex on last hookup
It is an indicator variable scored 1 if the respondent reports having had vaginal and/or anal intercourse on their most recent hookup, and 0 if they did not. The analysis with this dependent variable is limited to those respondents who have hooked up, and reported on what they did sexually on their last hookup. Thirty-nine percent of these students reported that they had sex on their most recent hookup (Table 1).
Long-term relationship in college
A dichotomous indicator measures whether the respondent has been in a relationship lasting six months or longer since starting college. Almost half of students reported that they had been in such a relationship (49 percent, Table 1). Analyses limited to seniors show that, by senior year, 70 percent have been in a relationship (results not shown).
Belief that love is necessary for sex
Respondents were asked whether they agreed with the statement, “I would not have sex with someone unless I was in love with them.” Originally a 4-point response item, we collapsed categories into agrees at least somewhat = 1 and disagrees at least somewhat = 0. Almost half of students agree at least somewhat (48 percent), but there is quite a large gender difference, with 52 percent of women but only 37 percent of men agreeing (Table 1).
Preference for being single
We construct this from agreement with the item, “I don’t really want to be in an exclusive relationship now because I’d rather be free to date or hook up with multiple people.” We code dichotomously just as we did with the belief that love is necessary for sex (above). Less than a third (29 percent) of students agree, with a gender difference such that 25 percent of women and 38 percent of men agree (Table 1).
Independent Variable of Major Interest
Campus sex ratio
It is a measure of the ratio of female to male students in the student body. Thus, values less than 1 indicate male-majority schools, values greater than 1 indicate female-majority schools, and “1” itself refers to parity. To test for nonlinearity, and to test the threshold version of the cultural tipping point hypothesis, we also include the square of the sex ratio. The number of male and female undergraduates was taken from reports schools submit annually to the U.S. Department of Education (2005–2011). We took these numbers for each of the seven years the survey was live (2005–2011), and created a ratio of women to men for each year, and, for each school, averaged across the seven years.
Independent Control Variables
Parents’ marital status
We control for whether the respondent’s parents are currently married to each other.
Mother’s education
We include the level of respondents’ mothers’ education as a measure of social class background, coded into four categories represented by three indicator variables: (1) mother dropped out of high school or at most earned a high school degree (the reference category), (2) mother attended some college but did not finish, (3) mother earned a bachelor’s degree, and (4) mother earned an advanced degree.
Age
Sexual experience increases with age, so we include detailed indicator variables for every year between 18 (reference) and 24 (those over 24 were excluded from our sample).
Race
We collapsed respondents’ race into four categories represented by three indicator variables: white (the reference category), black, Hispanic, and other race. These are taken from the question, “If you had to pick one racial or ethnic group to describe yourself, which would it be?”
Educational aspirations
We included indicator variables for whether respondents aspired to complete less than a bachelor’s degree (reference), just a bachelor’s degree, a master’s degree, or, as the highest category, a JD, MD, or PhD.
Religion
We include indicators of the religion in which one was brought up, using these categories: (1) no religion (the reference), (2) Catholic, (3) Conservative Protestant (the response option was “fundamentalist or evangelical Protestant”), (4) other Protestant, (5) Mormon, (6) Jewish, (7) Muslim, (8) Hindu, (9) Buddhist, and (10) other religion.
Residence
We account for the respondents’ place of residence by using the categories on campus (the reference category), off campus (but not with parents), and with parents. The majority of students (59 percent) live on campus, with only 10 percent living with parents (Table 1).
Fraternity/sorority membership
We include an indicator variable for whether a student is a member of a fraternity or sorority (1 indicates membership). A total of 14 percent of respondents reported being in a fraternity or sorority (Table 1).
Immigrant status
We enter an indicator variable for whether the respondent was born outside the United States (1) or in the United States (0).
We gathered data on the institutions themselves from the universities’ publicly available Common Data Set (CDS; Common Data Set Initiative 2014), by e-mail request when CDS was incomplete or unavailable, from data on number of male and female undergraduate students submitted annually to the U.S. Department of Education (2005–2011), and from the Princeton Review (2013).
Ln number of undergraduates
We use a natural log transformation of the number of enrolled undergraduates at each institution as a measure of campus size; it may also indicate how much sexual behavior with a fellow student is known to others and how anonymous it can be. It is taken from the CDS discussed above.
Percent commuters
This was taken from the CDS, and is the percent of students not living on campus.
Percentage in fraternity/sorority
We include a school-level measure of the percentage of students who are members of fraternities or sororities, as this may affect the social climate. It is taken from the CDS.
Private school
It is an indicator of whether the school is privately funded and controlled, taken from the CDS. If the school is public, it is coded as a 0. Six of the 21 schools are private, and 19 percent of respondents attend them (Table 1).
Selectivity
It is a measure of how selective each institution is in its admission process. The values are taken from the Princeton Review selectivity scores and range from 60 to 99.
Method
We conducted a series of regression analyses using STATA software. For dependent variables that were a count (number of hookups with strangers, number of hookups with acquaintances, and number of sexual partners), we used negative binomial regression, and we used logistic regression for those with dichotomous outcomes (sex on last hookup, whether has been in a relationship, preference for being single, and belief that love is necessary for sex). We use the same independent variables in all models. For those models where we found that sex ratio and its square are significant predictors, we used the “margins” command in STATA to compute predicted probabilities for the outcome variable that show average marginal effects of sex ratio, while controlling for other covariates. A predicted probability of the outcome was calculated for the observed range of sex ratio in our data.
Results
We focus on the effects of sex ratios, which Table 2 shows to vary from .88 to 1.86, showing substantial variation. Five universities have slight male majorities in their undergraduate student bodies, from .88 to .99. The rest have from slight to strong female majorities. One school, Framingham State University in Framingham, Massachusetts, is an outlier, with a sex ratio of 1.86; the next highest school has 1.37. The other schools with the highest sex ratios are small liberal arts colleges, with ratios from 1.26 to 1.37. As discussed above, to avoid conflating effects of sex ratio with other characteristics of schools with high sex ratios, we have controlled for several institutional characteristics, as well as individual characteristics that may differentially select students into different kinds of schools (Table 1).
Campus Sex Ratios by School.
Note. Sex ratios were computed from data schools provide to the Department of Education annually (http://nces.ed.gov/ipeds/). See text for details.
Tables 3 (women) and 4 (men) show coefficients from logistic and negative binomial regressions predicting the behaviors and attitudes that are our dependent variables from schools’ sex ratio, the square of sex ratio (to test for nonlinearity), and the control variables.
Coefficients from Logistic and Negative Binomial Regressions for Female Students.
p < .05. **p < .01. ***p < .001.
Coefficients from Logistic and Negative Binomial Regressions for Male Students.
p < .05. **p < .01. ***p < .001.
Effects of Sex Ratio on Hooking Up and Having Relationships
We first consider results regarding effects of the sex ratio on how much students hook up with acquaintances or strangers. In models predicting hookups with strangers, as well as those predicting hookups with acquaintances, both sex ratio and its square are significant for both women and men, indicating nonlinear effects (Tables 3 and 4). Based on these two models, Figures 1 and 2 graph men’s and women’s predicted number of hookups with strangers and acquaintances (from the models controlling for other variables). Figure 1 shows that, for men, the higher the ratio of women to men, the more hookups with either acquaintances or strangers they have; although the relationship is significantly nonlinear, it never turns downward in the existing range of sex ratios, but just levels off, with the range between approximately one and six hookups with strangers, a sizable effect of sex ratio. For women, number of hookups with strangers goes up from under one to about three as the ratio of women to men moves from its low to the relatively high sex ratio of the small liberal arts colleges, then turns down slightly to about 2.5 (Figure 1). The turndown of about one half a hookup, which is seen for women only, is driven entirely by one school, Framingham State in Massachusetts, which has the highest female to male ratio at 1.86. 3 The results for hookups with acquaintances (Figure 2) show that both women and men increase their numbers of hookups as the ratio of women to men increases, but, despite significant nonlinearity, the curves just level off, but do not show any nontrivial turndown. Thus, most of the results are consistent with the dyadic power model. The turndown of women’s curve at high sex ratios for hookups with strangers is consistent with a threshold version of the gendered culture hypothesis. Yet, the evidence for the thesis is very weak in that it is driven by one school, is quite small, pertains only to hookups with strangers and not with acquaintances, and is not observed for men, despite the gendered culture argument predicting it for both sexes. Overall, the results suggest that, across most of its range, a higher ratio of women increases hooking up for both men and women, consistent with the dyadic power hypothesis. We now turn to a consideration of whether this elevated level of hooking up crowds out relationships.
Predicted number of hookups with strangers in the observed range of campus sex ratio, for women and men.
Predicted number of hookups with acquaintances in the observed range of campus sex ratio, for women and men.
For men, sex ratio has no significant effect on whether they have been in a relationship of at least six months (Table 4). For women, the effect is nonlinear, with significant squared terms. Figure 3 shows predictions from the model; at the lowest sex ratio of .88, about 60 percent of women have been in a relationship, but when women are in the majority, that ratio is closer to 50 percent. While the relationship is significantly nonlinear, there is no nontrivial inflection back upward of the curve (Figure 3). The results for women are consistent with either the dyadic power hypothesis or, in the range of all but the school with the highest female/male ratio, with demographic availability. But the effects are quite modest in size. The nonsignificant results for men are consistent with none of the hypotheses.
Predicted proportion of women students who have been in at least one relationship of six months or more in the observed range of campus sex ratio.
Effects of Sex Ratios on Having Intercourse
We next examine the effect of sex ratios on how much students are actually having what they typically mean when they say “sex”—intercourse. First, limiting our analysis to the 63 percent of students who have hooked up at least once (Table 1), we examine whether the sex ratio affects whether, on their most recent hookup, they had anal or vaginal intercourse. Just under 40 percent of hookups involve intercourse (Table 1), and Tables 3 and 4 show that neither sex ratio nor its square have statistically significant effects on whether either men or women had intercourse during their last hookup. This conclusion is not sensitive to whether we use a definition of intercourse that includes anal sex as “intercourse”; only 1 percent of heterosexual hookups involve anal penetration, so regression results are almost identical to those shown here if we make having vaginal intercourse the outcome variable (results not shown). 4 Nor is it sensitive to whether we expand the sample to include those who had a hookup with someone of the other sex, but did not report their sexual orientation to be heterosexual; effects of sex ratio and its square are still nonsignificant (results not shown). Finally, we repeated the analysis limiting the sample to those hooking up with someone at their own school, with the rationale that these are the dyads to whom the theory should most apply; the results were similar, with sex ratio having no significant effect.
We next examine the effect of sex ratio on students’ total number of sexual intercourse partners. For both men and women, there is a nonlinear effect, as indicated by significant effects of sex ratio and its square (Tables 3 and 4). As Figure 4 shows, women’s lifetime number of sexual partners is just over 2 at schools where men are in the majority, rising to just over 3, and thereafter flat, with a trivial downturn. Both the flattening and the turndown come in the range of sex ratio above all schools but the one high outlier. Men’s number of partners goes from just over 2.5 at the lowest sex ratios to approximately 6 at the highest, a large effect. Thus, overall, the results regarding total number of sexual partners are consistent with the dyadic power thesis, but the results for having intercourse on hookups is inconsistent with any of our predictions—sex ratio has nothing to do with it.
Predicted number of sexual partners (ever) in the observed range of campus sex ratio, for women and men.
One limitation of our measure of sexual partners for testing the gendered dyadic power hypothesis is that the question asked about students’ lifetime number of intercourse partners, while we are interested in how campus sex ratio affects sex while students are in school. Thus, we did sensitivity tests to make sure that there was not something unmeasured about the schools with higher sex ratios that led them to take in students already more sexually experienced, with no accumulation of more sexual partners while in college; if that explained our findings, it would not be consistent with any of the theories we are testing. We interacted indicator variables for students’ age with sex ratio, and plotted the relationship between sex ratio and number of partners at each age. The graphs show that the gain in partners across the years was indeed greater at higher sex ratios; this reassures us that the effects we observe are not driven by preexisting differences (results not shown). Another concern we had was that the distribution of number of sexual partners is very skewed to the right for men; we wanted to ascertain whether the results were driven largely by men with an extremely large number of partners; to assess this, we excluded those individuals with 21 or more partners, and reran the analysis. Results for women changed little. Results for men retained the same significance and shape, but the range of sex ratio was accompanied by a rise from 2 to about 4.5 partners, rather than from 2 to 6. None of these sensitivity tests negate the conclusion that men and women have more sexual partners as the ratio of women to men rises throughout most all of its range.
Effects of Sex Ratios on Attitudes toward Casual Sex and Relationships
We have two attitudinal measures to gauge norms about casual sex versus relationships. One is whether one prefers to remain single to be free to hook up. Thirty-eight percent of men and 25 percent of women agree with this, a substantial gender difference. Tables 3 and 4 show that this is not affected by the sex ratio or its square for either men or women, so there is no figure for this variable. The second attitudinal measure is whether or not students agree that “I would not have sex with someone unless I was in love with them.” This also shows a substantial gender difference; 37 percent of men and 52 percent of women agree with the statement. Sex ratio has a nonlinear effect on this attitude for both men and women (Tables 3 and 4), which is graphed in Figure 5. As the ratio of women to men increases, the proportion of either women or men saying that they would not have sex without love goes down substantially, by approximately 30 percentage points for women and 40 for men; while the nonlinearity is that the relationship does not continue to descend, neither does it turn up nontrivially in the existing range of sex ratios. In short, results for this attitudinal item best fits the dyadic power thesis, but the other attitudinal variable was not affected at all by sex ratio, inconsistent with all hypotheses.
Predicted proportion of respondents who believe love is necessary for sex in the observed range of campus sex ratio, for women and men.
Summary of Findings Relevant to Hypotheses
How do our findings, taken together, relate to the three hypotheses we presented above?
Demographic opportunity hypothesis
This hypothesis predicts that, at schools with a higher ratio of women to men, women have fewer sexual partners, hook up fewer times, and are less likely to have had a relationship, while men do more of all these things in schools with more women. Under the hypothesis, when sex ratio has a positive relationship for women, it will have a negative relationship for men, and relationships should be linear. But this hardly describes the findings. When significant for both men and women, slopes always go the same way for both. Inconsistent with the thesis, as women’s proportion increases, in most of the range of sex ratio, women had more, not less, hookups and sexual partners. Overall, demographic opportunity does not explain cross-school variation in sexual attitudes or behavior.
Gendered dyadic power hypothesis
This hypothesis predicts that in schools with a higher ratio of women to men, both men and women have more sexual partners, hook up more times, are more likely to have sex on hookups, are less likely to have relationships, and express attitudes more accepting of casual sex. There is a fair amount of support for this thesis: in most of the range of sex ratio, as women’s representation increases, both women and men have more hookups, have more sexual partners, and are more likely to take the view that love is not necessary for sex. Despite this support, other important evidence fails to support the dyadic power hypothesis. A key claim of this view is that men are more able to persuade women to have sex when women have to compete against more other women for their attention, but we do not find that men or women are more likely to have intercourse on hookups as the ratio of women to men rises.5 Moreover, as hooking up and number of sexual partners goes up with an increased proportion of women, preferences do not shift toward being single, nor are relationships less likely for men. Finally, even where the evidence is strongest—predicting number of hookups—the fact that the effect of sex ratio levels flattens to essentially 0 at high levels would not be predicted by the theory.
Gendered culture hypothesis
This hypothesis predicts that in schools with a higher ratio of women to men (or, if there is a critical threshold, at least after a certain level of the ratio), both women and men have fewer sexual partners, hook up fewer times, are less likely to have sex on hookups, are more likely to have relationships, and express attitudes less accepting of casual sex. There is one isolated piece of support for this hypothesis: beyond the threshold of about 1.4 women to every man, as the proportion of women goes up, women hook up less with strangers. However, the inflection of the curve that reverses the slope is driven by only one school with an extreme sex ratio. Moreover, for other outcomes, even when the signs of the squared terms are significant, the graphs show that slopes do not reverse nontrivially, but just level out, and even the leveling out is driven mainly by the one outlier school. Perhaps a threshold version of the gendered culture theory has some explanatory power, but only at high ratios of women to men, and these gendered cultural patterns just neutralize the effects in the opposite direction predicted by the gendered dyadic power thesis. Data from more schools with large female majorities are needed to test this.
Discussion and Conclusion
It has long been argued that scarcity of one sex or the other begets power in dyadic heterosexual relationships—that when men are scarce (as dating partners, boyfriends, or husbands), women suffer. Our application of this hypothesis to the heterosexual college scene suggests both the merits and limitations of the view. Clearly, of the three hypotheses we considered, there was the most support for the gendered dyadic power hypothesis. Throughout most of the range of the sex ratio’s variation, more women encourages more hookups and more sexual partners for both men and women, as well as the belief that love is not necessary for sex. Men do indeed appear to “get more sex” overall where there are more women, as Uecker (2012) found and Regnerus (2012) argued.
However, the findings also show the limitations of what the gendered dyadic power perspective can explain, as well as the limitations of much of the alarmist discourse about college sex, including those of traditionalists concerned about the erosion of chastity and the retreat from marriage, and those of feminists worried that hookup culture means that men have intensified the objectification and exploitation of women. This discourse seems to assume that hookups all involve intercourse, and that the casual sex entailed in hooking up crowds out monogamous relationships. But this is not the picture we get here. It is crucial that clear distinctions are made between “hookups” and intercourse. Those in the “party culture” (Armstrong and Hamilton 2013) may hook up more and have more intercourse, but under 40 percent of hookups involve intercourse, and much intercourse occurs in relationships. Moreover, while we did find evidence that when women are more numerous, it leads to more hookups and, perhaps because of this, the accumulation of more intercourse partners, it does not lead to a higher probability that any one hookup involves intercourse. Moreover, just as our grandmothers warned that men would not marry women who had premarital sex, alarmists and other advocates of the dyadic power thesis often imply that hookups drive out relationships. They miss the fact that most youth engage in both hookups and relationships at some time across their college experience. Indeed, we found that despite the fact that a higher ratio of women to men encourages hookups and sexual activity, it does nothing to men’s or women’s preferences for being “single” versus in a relationship, and it has no effect on whether men actually participate in long-term relationships, nor does it empower men to demand more sexual intercourse in hookups. Changing sexual behaviors may lead to alarmist predictions but our findings do not support the fear that rising proportions of women on college campuses mean men eschew relationships and successfully demand intercourse during hookups.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.