Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Sexual economics: an econometric study of a university Greek system
(USC Thesis Other)
Sexual economics: an econometric study of a university Greek system
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
Sexual Economics Hernandez USC Economics 2014 1 Sexual Economics: An Econometric Study of a University Greek System Sean Hernandez University of Southern California Department of Economics Sexual Economics Hernandez USC Economics 2014 2 Acknowledgements My sincere thanks go to Professor Ann Majchrzak in the Marshall School of Business, who was the first scholar to lend the assurance that my fraternity observations were a viable research area. I could not have asked for a more supportive committee chair than Professor Yilmaz Kocer. Both his research and professional guidance were instrumental to advancing this project while completing a concentrated course load simultaneous with securing PhD admittance. My most high-powered supporter was Professor Kevin Starr. Through his leadership in the USC Sidney Harman Academy for Polymathic Study, I had a good excuse to advance this research conceptually during my undergraduate years. Many questions in the survey instruments were inspired by Dr. Karin Huebner, the director of programs with the Polymathic Academy. I know she will be proud to read this thesis and discover the high internal consistency of measurements that were our original creation. Professor Jeffrey Nugent was the only other fraternity member on my committee. I thank him for the consistent infusion of wisdom that our interactions produced. My thesis and career have been substantially augmented by contributions from a most clever econometrician, Professor Yu-Wei Hsieh. Although I did not have the foresight to include him in my committee formally, the advice he gave regarding regression specification, identification techniques, literature to review, and statistical coding accelerated this work by months. Outside of USC faculty, the completeness of the surveys employed here are attributable to Shelly McArdle, who has a far more advanced understand of Qualtrics.com than myself. I wouldn’t have been able to measure the key dependent variable – hookups – without her cleverness in the survey design. The USC Fraternity and Sorority Leadership office was a critical contact in this process. I thank its administrators both for directing me through the proper fraternity/sorority institutional channels as well as for providing trustworthy data for fraternity and sorority-level variables. To the National Panhellenic Council, I cannot thank you enough for giving official authorization to this research. The data collection results show very clearly that the authorization was instrumental in assuaging survey respondents’ concerns for anonymity, etc. I thank the Center for the Study of the College Fraternity at Indiana University for their generous financial contribution to the research and I look forward to sharing the results with its scholars in the very near future. I hope that both NPC and CSCF can utilize the results presented in order to strengthen your organizations in years to come. Lastly, it is only right that I thank my fraternity brothers in the Alpha Nu chapter of Theta Xi. They all sat through several sexual-economics related filibusters in meetings while this project was in its infancy. I can only hope that seeing this all in numbers will have made it worth it for them. Sexual Economics Hernandez USC Economics 2014 3 Contents (I) Introduction ................................................................................................................................................................... 4 (II) Literature Review - Transdisciplinary Sexual Economics ................................................................................... 10 (III) Data Collection ......................................................................................................................................................... 17 (IV) Preliminary Results ................................................................................................................................................... 19 (V) Analysis ........................................................................................................................................................................ 29 (VI) Technical Issues ........................................................................................................................................................ 45 (VII) Peer Effects ............................................................................................................................................................. 48 (VIII) Conclusion.............................................................................................................................................................. 59 Appendix A: Factor Analysis of Erotic Capital Items ................................................................................................ 70 Appendix B. Greek Students’ Regrets ........................................................................................................................... 76 Appendix C. Survey Results - Fraternity Treasurers ................................................................................................... 77 Appendix D. ANOVA Results ....................................................................................................................................... 78 Appendix E. Sexual Outcomes Ordered by Fraternity and Sorority Social Status ................................................ 80 Appendix F. Differences between Top/Bottom Greek Women and non-Greek Women .................................. 83 Appendix G: Differences between Top and Bottom Fraternities. ........................................................................... 89 Sexual Economics Hernandez USC Economics 2014 4 “We are the recorders and reporters of facts - not the judges of the behaviors we describe.” Dr. Alfred Kinsey, 1894-1956 “Science by itself has no moral dimension. But it does seek to establish truth. And upon this truth morality can be built.” Dr. William Masters, 1915-2001 “If we all go for the blonde and block each other, not a single one of us is going to get her. So then we go for her friends, but they will all give us the cold shoulder because no one likes to be second choice. But what if none of us goes for the blonde? We won't get in each other's way and we won't insult the other girls. It's the only way to win. It's the only way we all get laid.” Russell Crowe as John Nash in “A Beautiful Mind,” 2001 (I) Introduction Social-constructionist feminism has imbued many with the idea that all of gender studies concerns male subjugation of women and their responses to oppression (Nelson). However, the economics of sexuality insists that forces more complex than mere “exploitation” are at play (Regnerus, 2012; Baumeister and Vohs, 2004). The present research is one step toward elucidating this complexity by investigating an equally pervasive idea; that universities’ Greek societies are fairgrounds for wanton hedonism, debauchery, and sexual exploitation. Econometric analysis using socioeconomic demographics and sexual matching data could confirm or deny this objection. The contribution of economics to sociology has always been rigorous mathematical structure. As such, the study of sexual economics should strive to obtain a mechanistic understanding of the sexual interactions that Greek student institutions create. In the United States, over 800 college campuses are hosts to the special sort of student organization known as a Greek “house” (North American Interfraternity Council). The term “house” has a double meaning. Not only do “Greek” organizations typically own real estate to rent to their members, but the groups’ members collectively identify as “the house” similarly to group identifiers like “the people,” “the nation,” etc. The houses are “Greek” because they are named by 2-4 Greek letter acronyms (for example, “the brothers/sisters of Iota Alpha Theta fraternity/sorority”). Each fraternity and sorority is actually a national foundation, supported by alumni donations and employed staff, which works to extend the influence of that particular fraternity/sorority to new campuses. To differing degrees, the national foundations also govern individual fraternity/sorority chapters at individual universities. In turn, these national foundations are members of two super-organizations, North-American Interfraternity Conference and National Panhellenic Council (NPCWomen.org). The financial and legal support afforded by this national network constitutes a substantial advantage for Greek houses vis-à-vis other university student organizations (Armstrong & Hamilton, pgs. 77-85). The proceeding analysis derives estimates for many of the social, economic, cultural, and especially sexual consequences of the Greek houses’ unique status within university communities. It is chiefly motivated by the distinct attribute of physical capital (real estate) ownership by Greek Sexual Economics Hernandez USC Economics 2014 5 houses, which, I argue, drives the sexual economic activities of Greek students. Armstrong and Hamilton suggested the Greeks’ uniqueness is also uniquely dangerous. As has been shown in countless cinemas and ingrained in the collective American conscious, fraternities’ privately-owned, off-campus real estate frequently serve as havens from underage drinking enforcement (Armstrong & Hamilton, pg. 54). Contemporary attitudes employ the idea that “college is about having the best of one’s life.” This directs many students toward experiences that emphasize socializing and alcohol consumption more so than professional knowledge (Armstrong & Hamilton, pg. 22). In an attempt at remedying the situation, the present analysis intends to explain differing operations of fraternities and sororities in the context of sexual economics theory (Baumesieter and Vohs, 2004). Namely, it attempts to show that the absence of sorority-operated social gatherings gives rise to an oligopolistic model of sexual competition. If they increase sexual-economic welfare, sororities should enable their members to exchange fewer sexual encounters to obtain a higher “price” for each of them. Conversely, fraternities should operate according to oligopsony theory, where the welfare of their members is increased by facilitating sexual encounters at a lower “price” with more sexual encounters in total. Economics is an ideal discipline for addressing concerns from skeptics of the Greek system as well as social-constructivist feminists. By employing theory-driven mathematical constructions, the alleged mechanisms of power can be tested against “as-if” distributions. This is the essence of the econometric technique known as the generalized method of moments (Cameron & Trivedi, 2009, pg. 743). An econometric model of this type proceeds by mathematically defining some assumptions about the population distribution and testing them against empirical data. The most foundational assumption is social exchange theory, which describes that interpersonal relationships occur from expectations that both individuals stand to gain from their interaction. Baumeister and Vohs elaborated social exchange into sexual economics by correctly identifying that nearly all societies have placed considerable value on female sexuality while considering male sexuality effectively worthless. “Sex is a female resource” (Baumeister and Vohs, 2004). Men must provide something valuable to women in order to incentivize any transaction under this value asymmetry. Evolutionary, economic, and social causes underpin the “female resource theory” of sex. For men, the minimum investment in parenthood is essentially zero because they are not biologically tied to the process, whereas for women, sex entails the risk of pregnancy, possible death during childbirth, and less avoidable costs of childrearing overall (Symons, 1979). Townsend & Levy (1990) referred to this as the “parental investment theory.” A second explanation of female resource theory is the “principle of least interest,” which predicts that a sexual or romantic partner who wants the connection less has more power. Accordingly, women have more power in sexual relationships because they demand it less than men. This has been called the “permanent male sex deficit” (PMSD). PMSD is perhaps the most objectionable variant of female resource theory, particularly from certain feminist perspectives. However, it also the most thoroughly tested result in all of sexual economics. As put by Regnerus, “Women do like sex. But men want sex more” (Sociology). Some scholars claimed this fact has been obvious for centuries (Hakim, 2010). Men initiate sexual encounters more often and are more likely to report having five or more sexual partners in one year. Survey results of men reveal more sexually permissive or hedonistic attitudes (Laumann et al, 1994, pg. 509-547; Knox & Cooper et al., 2001; Knox & Sturdivant et al., 2001). Men masturbate more – yielding revealed preference evidence of “latent demand” (Regnerus, 2012). Men are more likely to have sex in the absence of emotional involvement or marital potential (cited in Townsend & Levy, Sexual Economics Hernandez USC Economics 2014 6 1990: Kinsey et al., 1953; Kanin et al., 1970; Wilson, 1981; Houston, 1981; Blumstein and Schwartz, 1983; Carroll et al., 1985; Roche, 1986; Symons and Ellis, 1989). Other research has shown that these sex differences have been consistent over both generations (cited in Townsend & Levy, 1990: Holmes and Hatch, 1938; Kinsey et al., 1953; Hudon & Henze, 1969; Symons, 1979, 1985, 1987; Davis, 1985; Maxur, 1986). For a comprehensive treatment of the permanent male sex deficit approach, readers are directed to (Baumeister et al., 2001). Permanent male sex deficit or not, female resource theory is strongly confirmed by alternative observations. Women never pay for sex. Even male prostitutes almost always service male consumers (Baumesiter and Vohs, 2004). Men uniformly traded economic for sexual resources in a cross-national survey of 37 countries, denying the allegation that female resource theory is merely a cultural artifact (Buss, 1989). The phenomenon of “trophy wives” and absence of symmetrical “trophy husbands” shows that women’s erotic value is greater than men’s (Hakim, 2000, pg. 162). Even studies on married couples showed that wives use sexual access as a bargaining chip with their husbands (Arndt, 2009). “Exploitation” offers one final explanation of female resource theory. In historical- materialism or other totalizing narratives of social development, the growth of economic and political spheres of activity occurred only for men, leaving women at an absolute disadvantage but with comparative advantage in sex production. For some scholars, exploitation is the main driver of sexual economics (Lipman-Blumen, 1984, pg. 89-90). However, the inter-generational and inter- cultural consistency of female resource theory provides good evidence that exploitation is not the only driver of women’s supply incentives. For further work in this area, socio-cognitive theory has been used to explain individuals’ sexual expectations in terms of their parents’ relationship constructs. Arguably, this could result in a dynamic equilibrium that reinforces exploitative sexual norms (Bandura, 1977). Greek systems providetheir own evidence for female resource theory. They offer a natural experiment because fraternity men are institutionally vested with responsibility for social expenditures that are consumed by sorority women for free (at least “free” in the sense of not requiring a monetary exchange, although there is sexual expectation). Anthropologists Armstrong and Hamilton phrased these incentives beautifully, stating that “Fraternities sponsor an erotic marketplace in which students gain status and make connections through ‘hooking up’” (Armstrong & Hamilton, pg. 16). During their investigation, they observed lines of fraternity men’s vehicle arriving to pick up freshman women from their “party dorm.” These transportation services, in addition to the parties, were exchanged (in theory) because the expected value of sexual utility was greater than the expenditure of fuel and vehicle depreciation. This analysis takes Armstrong and Hamilton’s ethnography a step further using quantitative analysis and attempts to show how their discussion of the role of women’s social status may err. Intuitively, one suspects that higher status women are more promiscuous because they have more opportunities or pursuers. However, phenomena like “slut-shaming,” broadly known as “sexual double standard” suggest that, with all other things equal, social status is negatively correlated with hookups for women (Crawford and Popp, 2003). I hypothesize that sororities increase sexual economic welfare for women by serving as a collusive mechanism by which women increase their social status, restrict their supplies of sex, and increase their profits. This only works if high status sororities effectively lock up the supply of the most desirable women. The ethnography did find some evidence of an inverse relationship between status and hooking up. They reported, for example, that “Sororities were on the lookout for women who could damage the sexual reputation Sexual Economics Hernandez USC Economics 2014 7 of the house. An ideal sorority woman was attractive to fraternity men but did not cross the line from sexy to slutty” (Armstrong & Hamilton, pg. 81). The collusive aspects of sorority political economy are well-evidenced in existing policy. Each sorority chapter self-imposes regulations such as curfews or “no boys upstairs” policies which are intended to protect women by preserving the sanctity of their homes. At the foundation of this policy regime is the sororities’ ban on in-house parties, which starkly contrasts the expenditure strategies of fraternities. Female resource theory suggests that there may be devastating unintended consequences. With demand held constant, disallowing parties or male guests at sorority houses artificially limits the supply of sex from sorority women, incentivizing a quasi-black market where men compete all-the-more vigorously to have sex with women in their fraternity houses. Compounding that regulatory oversight, private ownership of fraternity houses endows Greek men with the important right to exclude non-member males from their parties while also ex ante selecting the sizes of their parties and sizes/composition of their memberships. Such a scenario constitutes a very classic public goods problem, where the men bring their individual preferences and attributes and maximize a social optimum (in this case, sexual utility) (Buchanan, 1965). If fraternities employ the proper factors, their parties may afford them incredible “market power” via the ability to engineer a disproportionally small ratio of men to women (Armstrong & Hamilton, pg. 90). Sex ratio distortion motivates a mechanistic and, thereby, economic method for testing the sexual exploitation hypothesis. For example, Regnerus observed that on campuses where the ratio of male to female students was less than one, women gave their male partners lower approval ratings, including lower beliefs in potential romance as well as lower realized values for dates and relationships. The results are economically explicable: with a higher supply of women, relative to men, the price of sex is relatively low, leading to high consumer (male) surplus and low producer (female) surplus. Conversely, with high sex prices, Regnerus predicts more impressive courtship by men, fewer premarital sexual partners, and higher marriage rates (Sociology, 2012).The idea of sexual economics as a feminist project, then, depends on institutional, policy, and welfare analyses of these outcomes. The technology of fraternities counteracts sororities’ supply power in two important ways: (1) the exclusive right to host parties affords both a sex ratio distortion and a “home-turf advantage” over women (2) alcohol intoxication. It was reported that the combination of these two purely institutional attributes left many women in dangerous situations because some of them were not able to make it home without assistance (Armstrong & Hamilton, pg. 88). As previously mentioned, many fraternity men had lined up to drive the women to the fraternity houses. But at the end of the night, the prospect of a hooking up was further incentivized by lack of sober transportation. Through this line of reasoning, Greek systems may offer a refined understanding of the institutional economics of sexuality. I hypothesize that fraternities and sororities build-in incentives which reflect the axioms of sexual economics on an aggregated scale. Fraternities, reflecting the so- called “permanent male sex deficit”, maximize sex and minimize expenditure (which includes relationship commitments). Sororities minimize sex and maximize the value of exchanges in gifts, parties, and relationship commitments (Hakim, 2010; Vohs and Lasaleta 2008). The sexual exploitation theory becomes testable. If fraternity membership grants an individual more market power than membership in a sorority, then fraternities operate as oligopsonists, achieving a price of sex lower than the competitive market equilibrium. This is a high theoretical burden that was first modeled by the first great American economist, John Bates Clarke, Sexual Economics Hernandez USC Economics 2014 8 in his refutation of the broader Marxist theory of exploitation. Buying power means that in various social situations there will be more women than men on average. Men can acquire extra sex because the price ratio favors the transfer of utility/wealth toward them. In a neoclassical microeconomics setting, a monoposonist is usually a firm buying labor. Buying power essentially means that the fraternities face an endogenous utility maximization function, where the amount of sex they consume will have a positive influence on the price. They will essentially shoot themselves in the foot if they consume too much sex because prices will rise. The marginal factor cost, which indicates women’s true value added, will exceed the average factor cost where the wage is set by fraternities. The price of sex (which is a wage in the monopsony model) will be set where the marginal revenue product (men’s demand) is equal to the average factor cost. The difference between the market wage (MRP=AFC) and the point above it on the marginal revenue product curve is the “rate of exploitation,” as defined by Clark (Clark, The Distribution of Wealth, 1899). This is illustrated in Figure 1. Sexual Economics Hernandez USC Economics 2014 9 Figure 1. A monopsony model. For each level of output, the marginal revenue product measures the marginal laborer’s value added. A monoposonist’s endogenous optimization function causes the marginal cost of labor to diverge from the average cost of labor (the supply curve). The difference between the wage at w and the value at the point A is the “rate of exploitation.” Notice that the graph of monopsony outcomes is a vertical reflection of a graph of monopoly outcomes. Sexual Economics Hernandez USC Economics 2014 10 While the style, motivations, and methods of research are primarily economic, the process called for infusions of knowledge from several subfields. Beyond scholars whose work is directly identifiable as “sexual economics,” survey and analytical methods were taken from anthropology, social psychology, and many subfields of economics. In the following section, I detail the contributions of each field to motivate my work and hopefully that of others in the future. In Section 3, the study’s data collection process is detailed. In section 4, preliminary results including ANOVA and t-tests of mean differences convey a general overview of Greek sexual economics. In Section 5, an econometric model is specified to test female resource theory’s ability to explain hookups, hookup partners, and relationships per semester for women. In Section VI, I discuss various econometrics challenges of the research. In Section VII, I discuss the identifiability of peer effects estimates for the sorority data. Finally, I conclude with suggestions to students, administrators, sexual economists, and researchers in other fields. (II) Literature Review - Transdisciplinary Sexual Economics Institutional economics of sexuality has wide theoretical and empirical bases to draw on, including feminist economics, social economics, institutional political economy, social psychology, and peer effects econometrics. In addition, a vast majority of research on sexual behavior has been performed by health professionals, anthropologists, social psychologists, and others. Numerous bridges connect each of the disciplines. As such, my intent is to take readers through each research area beginning with the broadest and ending with the narrowest. I imagine that economists and laypersons alike will be surprised to find out that economics is a growing area of sex research, not all of it relating to intercourse. Feminist economics defines a newly organized subfield (1990’s) that investigates gender topics such as labor markets, (unpaid) household production functions, and the “emotional division of labor” (Bergmann, 1974; Hartmann, 1976; Wagman & Folbre, 1996). An exemplary paper by Hakim (2010) directly bridged feminist and sexual economics with the novel concept of “erotic capital.” Her theory supplied a strong rebuttal to authors who attack the female resource theory from feminist perspectives. Much like the concepts of social and cultural capital, erotic capital describes a category of assets that are valued for their potential in erotic/sexual exchange. Its principal components are (1) beauty, (2) sexual attractiveness, (3) social elements, (4) liveliness, (5) social presentation, (6) sexuality, (7) fertility. Erotic capital categories (1), (5) are given a separate treatment in Appendix A. Just as with social and cultural capital, there are clear challenges in measuring erotic capital. But, Hakim argues there are big conceptual payoffs. For example, because women normally have higher erotic capital values than men, when their incomes rise, they should reduce investments in erotic capital because the marginal utility in terms of men’s economic capital is diminished. A more speculative consequence is that men will increase their erotic investments to become more attractive to high income women. Some feminists would reject female resource theory, alleging that it commodifies female sexuality. Indeed it does, but Hakim reverses this argument, prescribing that women are best served by maximizing the value of their “relative advantage.” Female resource theory emanates from forces external to differential economic status (e.g. the parental investment theory and the permanent male sex deficit). If external forces render erotic capital an inherent advantage for women, then there is a strong sense in which women are more powerful than men. For this reason, erotic capital could be called “erotic power” (Hakim, 2010) Alternatively, Zetterberg (1966) formed “erotic rank” as a Sexual Economics Hernandez USC Economics 2014 11 precursor to the idea. Hakim provides a novel critique of feminism from this perspective by pointing out that the “commodification” argument against female resource theory directly results in a decrease in the value of erotic capital, either by persuading women that attractiveness is not valuable and/or by convincing men that they should not pay for it. The tendency to deny such a thing as erotic capital is, therefore, a patriarchal one. The political economy of the issue is obvious: men will, in equilibrium, endeavor to convince women (or, perhaps, society as a whole) that erotic assets cannot be valued, should not be valued, or are valued at a much lower price than that paid. By doing so, the expected marginal cost of female sex will fall and sex prices will fall, increasing consumer surplus to men and decreasing producer surplus to women. In this way, indicts of erotic capital or female resource theory are not limited critiques of neoclassical sexual economics. They are also necessary components of the “slut-shaming” mentality, sexual repression by conventional religions, negative attitudes towards sex workers, and labeling attractive women who marry wealthy men as “gold-diggers.” Thus, the concept of erotic capital contributes a feminist lenses for understanding female resource theory independently of its evolutionary or demand-side explanations, which may be easy targets for feminist scholars and others. Its utility is broad, including applications to mating markets (Zelizer, 2005), anthropology (Heelas, pg. 234-266), marriage stability studies (Hunter, 2010), and social economics (Webset & Driskell, 1983). I hypothesize that sororities are concentrated sites of erotic power and thus, can lend additional understanding to “women’s changing status in society and the economy” (Hakim, 2010). Feminist economics may be the primary constituent of institutional economics of sexuality but it has a close cousin in a set of research known as “social economics.” Economic activity affects social processes in four chief ways that have been investigated by “social economists”: (1) externalities, (2) discrimination, (3) social interaction, and (4) positional goods. A relevant example: both positive and negative externalities to pornography have been investigated (Cameron, 2005, pg. 171.). Social economic research in discrimination largely parallels the work which has been discussed from feminist economists (Nelson). In Section VII, I draw on a specific econometric technique for measuring social interaction known as “peer effects” to estimate the “social multiplier” of hookups within sororities. Most importantly, my hypotheses about the operations of university Greek systems invoke the social economics concept of social status. Becker et al. (2005) derived a model that contributed two pertinent points about status. First, the price of status positions will increase with economic growth while the price of consumption goods will fall. As growth progresses, the supply of consumption goods increases while no evidence suggests corresponding increases in the number of status positions. The result is that wealthy societies will use more and more consumption goods to compete with increasing vigor to secure the fixed number of high status positions (especially if population increases). Women’s accounts of the competitiveness of sorority rush may corroborate Becker et al.’s theory (Armstrong & Hamilton, pg. 82-86). Second, a market where status is inferred from a “status good” (i.e. “implicit status market) reaches the same equilibrium as an explicit status market where positions are auctioned (Becker et al., 2005). Therefore, we can be confident that students’ fraternity or sorority house status accurately measures each members’ social status, even though group selection is not in the models. Findings from anthropology suggest that there is a lot to be said about the economics of status in Greek systems. Armstrong and Hamilto state that, “Class interests are built into the architecture of the university and…it serves some students better than others. Courses, grades, professors are all against a Weberian backdrop of ‘social closure’” (Paying for the Party, pg. 10). Sexual Economics Hernandez USC Economics 2014 12 They criticized the pre-existing literature, effectively demanding a quantitative analysis of sexual economics outcomes. “Statistical data analyses do not go inside the black box [of the university]” nor the “experiential core of college life” (Armstrong & Hamilton, pg. 4). Class differences and divergent college expectations among the women in their study illustrated three clearly distinct college “pathways”: party, social mobility, and professional. An essential component of the party pathway, Greek life was argued to reinforce economic hegemony via social networking and charismatic “easy majors” (Armstrong & Hamilton, pg. 16). Specifically, the authors made frequent observations of a phenomenon they identified as “status anxiety” (Armstrong & Hamilton, pg. 93; Layte & Whelan, 2013). While the ethnography identified “ceaseless competition” as the cause of status anxiety, social economics results for positional goods illustrate a deeper story. Positional goods are those for which a person’s relative holding enters the utility function. For example, I can write a utility as a function of gross income: U = f(Y), or I can write it as a function of my income rank within a given group, U = h[Ω(Y)], where Ω is a function that coverts dollars incomes to ordinal income rankings (1 n). Frank (1985) identified a negative externality in such positional goods setups. Imagine an individual who gains status and moves from position 2 to position 1. By definition of the positional good, the person who previously occupied position 1 will move to position 2, reducing their utility. Frank (2008) extended this to applied scenarios where position taxes could be used to ameliorate externalities. Some evidence suggests that individuals have preferences over their local rank within smaller communities compared to their national or global rank. If this holds for undergraduates, we could imagine that a preference for local rank could direct a high-status individual into a relatively low-status fraternity or sorority (Frank, 1999). Feminist, social, and sexual economics are each broad in scope. But a model from institutional economics will help identify the few mechanisms under investigation here. An institutional model can include many different actors, groups, organizations, social norms, or almost anything else. Ostrom suggested the general term, “holons,” to refer to any one component of an institutional model (2005). In Figure 2, holons are shown for the Greek sexual economy. It conveys the main theoretical insight that individuals’ relationships are not isolated phenomenon but are dependent on decisions and experiences of all the actors in the market. Sexual Economics Hernandez USC Economics 2014 13 Figure 2: Holons of the Sexual Marketplace Exogenous preferences Sorority members Fraternity members Greek Sex Supply Greek Sex Demand Non-Greek Sex Supply Non-Greek Sex Supply Market Sex Demand Market Sex Supply Market Sex Price and Output Bargaining Sexual Economics Hernandez USC Economics 2014 14 In any institutional setting, rules are a precursor to social norms. The most important norms for this case are sorority controls and fraternity parties. By sorority controls, I refer to those regulations which restrict the supply of sex, such as party bans and “no boys upstairs” policies. These give the means for distinguishing groups. If sorority controls and fraternity parties reflect a neoclassical interpretation of sexual economics, then higher status sorority houses should have more restrictive policies and higher status fraternities should have most successful parties (“successful” in the sexual economics sense). Typically, individuals take norms as exogenous and there is essentially a collective action problem in altering norms (Ostrom, 2005). Social economics literature typically takes social/institutional norms as given and incorporates individuals “deep preferences” within socially contingent “reduced form preferences.” Aristocratic matching models are one critically important example of the distinction, and their logic extends into the results of this analysis. Aristocratic matching refers to a structure where agents with the highest “rank” (rank may be defined in numerous ways within different models) match with the wealthiest principle on the other side of the market. Such a model features deep preferences for wealth as well as the random variable according to which the ranks are constructed. Yet the matching process occurs with respect to the random variable’s rankings, not its real magnitudes. Attention to rank constitutes a reduced form preference for the random variable. Optimization of the reduced form preference can give results quite different from optimization on magnitudes (Cole et al., 1992). Aristocratic matching was given experimental, sexual economic treatment in a study by Townsend & Levy (1990). The experiment tested female resource theory by showing social status – measured by occupation and income prospect – completely dominated physical attractiveness when women selected individuals for both marriage and sexual intercourse. In fact, the result was most robust for sexual intercourse decisions. Even the lowest attractiveness man was selected for marriage and sex as long as his social status was the highest. High status equalized the desirability of low and medium attractiveness men for other sexual economics scenarios such as coffee/conversation and dates. Conversely, high status never compensated for low physical attractiveness in male’s selections. Townsend & Levy (1990) showed that sex differences in selection criteria are most striking for sexual relations. Differences were notable when marriage was involved, but both men and women were quite cautious in selecting hypothetical marital partners. Men and women were most similar in selecting partners for coffee/conversation and dates. This motivated the characterization of sexual behavior employed in this study. I directed both survey materials and literary analysis toward a particular sexual phenomenon known as “hooking up” because it accentuates the sex selection differences that Townsend & Levy found. Many definitions of hooking up have been given in the literature, but they each reflect the common theme of sexual activity devoid of romantic expectation (Fielder & Carey, 2010; Glenn & Marquardy, 2001; Owen et al., 2010; Bogle, 2008). Hooking up reflects the accelerating sexualization of society. People all around the world are having more sex, in many new ways while coming to consider it more and more importantly as a quality of life factor (Saxena et al., 2001). College students, especially, have been the subjects of research into hooking up, although these have been simple cross sections that did not venture into the “black box.” Moreover, Armstrong and Hamilton found troves of qualitative evidence that hooking up is particularly important in university Greek systems. The most comprehensive research was a longitudinal study of college students’ sexual behaviors, attitudes, and risks beginning in 1989. Even in 1989, college students were uniquely likely to engage in “hook up”-style sexual relationships (Herlitz & Ramstedt, 2005). The college hook up rate has been found to be as high as 81%, with others in the neighborhood of 70-80% (Bisson & Kevine; Lambert et al., 2003; Paul et al., 2000; Paul et al., 2002). But U.S. students actually begin their sex lives in high school 47% of the time (CDC, 2006). Sexual Economics Hernandez USC Economics 2014 15 Numbers of sexual partners and romantic relationships increase thereafter, with related increases in pregnancy and sexually transmitted disease (Corbin and Fromme, 2002). There is an abundance of studies that examine social and demographic influences on adolescent and young adult sexual behavior (Miller et al., 1998; Watts & Naggy, 2000; Cooper, 2002). In the creation of a student survey to measure demographics, Greek house membership, Greek social norms, and hookup and relationship behavior, I utilized conceptual and survey items from sexual research that gave specific attention to hooking up. Differences have been found between sexes (Townsend & Levy, 1990; Wetherill et al., 2010; Fielder & Carey, 2010; Paul et al., 2002; Knox et al., 2001), races (Owen et al., 2010), and income groups (Armstrong & Hamilton, pg. 12). Various measurements for alcohol intoxication and pre-college hookup experience were also commonly used predictors (Wetherill et al., 2010; Fielder & Carey, 2010; Knox et al., 2001; Olmstead et al., 2012). Sexual behavior literature motivated the peer effects estimation in Section VII. For instance, Wetherill et al., (2010) predicted a kind-of inverted peer effect hypothesis that “higher levels of parents perceived-awareness-and-caring (PAC) and peer PAC would be associated with fewer sexual behaviors during both high school and college.” If students’ peers were more aware and caring about their sexual behavior, then support and monitoring norms should negatively feedback on hookups in their view. The prediction held for populations that had conservative sexual values. However, when both peer and parental PACs were high and the student had liberal sexual values, they were exposed to the greatest degree of risk. This is an interesting observation to have in mind when considering the interpretation of the “social multiplier” for sorority women’s sexual behaviors. Whereas the social multiplier predicts that peers act as motivators or facilitators of spiraling behavior, Wetheril et al. suggest peer pathways from “empathic validating relationship[s] that include helping with decision-making, empowerment, and developing awareness, and peer monitoring.” They described a focus group of freshman and sophomore students who suggested that low levels of peer monitoring leads to high hookup rates. Equivalently, feelings of being unknown or anonymous are highly correlated with hookup-like behavior. Hookup rates are high at both ends, where peer and parental PAC are both very low and very high. But Wetherill et al. and peer effects estimates seem to be examining different mechanisms. The type of hooking up that occurs at low levels of peer PAC should be mechanistically distinct from that occurring in university Greek systems due to the tightness of peer networks that their houses afford (Figure 3). Section III proceeds from the concepts above by detailing the methods of data collection. The controversy regarding the direction of peer influences on sexual behavior continues in Section VI. Sexual Economics Hernandez USC Economics 2014 16 Figure 3. Conceptually distinct peer effects Hookups per Individual Perceived Peer/Parental Awareness and Caring Anonymity drives hookups Social multipliers drive hookups Sexual Economics Hernandez USC Economics 2014 17 (III) Data Collection Three surveys were conducted. Fielder and Carey (2010) suggested that online surveys are effective at reducing measurement errors in sexual behavior data. Therefore, the first data collection mechanism was a 92-item online survey of three populations: university fraternity members, university sorority members, and female university students who were not sorority members. The subjects were solicited in four ways: (1) paper flyers posted in high traffic areas on campus, (2) three personal solicitations to each of the Greek houses during weekly meeting times, (3) one in person solicitation at a weekly meeting of fraternity and sorority presidents, (4) and three email solicitations sent to the fraternity and sorority presidents as well as Interfraternity Council and Panhellenic Council executive officers. Fraternity and sorority members were incentivized with a house-level cash prize competition mentioned both on flyers and during solicitations. Two amounts of $500 were donated to the fraternity and sorority with the largest numbers of members participating in the survey and another two amounts of $500 were donated to the fraternity and sorority with the highest participation rates. This incentive was designed to be attractive to houses with both small and large populations. The online survey was given in the following 10 blocks: (1) Introduction – Respondents identify their sex and read a matched consent form. Men identified themselves as Greek or non-Greek. Non-Greek men were excluded after this section using the survey logic. (2) Anonymous House Identification – Fraternity and sorority members identified their own houses from a set of edited images of the 22 Interfraternity Council fraternity houses or 9 Panhellenic Council sorority houses. Identifiable characteristics such as the Greek letters were removed from the images. Additional membership categories included “fraternity/sorotiy that does not own a house,” and “Non-Greek female.” (3) Collegiate Demographics – Respondents were asked about various personal attributes like class rank and GPA, race, hair color, and parents’ education level and income range. (4) Sexual Capitalization – Respondents answered questions relating to their social status and sexual preferences. These included randomized measures of extroversion, self-reported attractiveness, the range of attractiveness-acceptability for sexual partners, and retroactive pre-semester hookup and relationship expectations (Gosling et al., 2003). (5) Income, Expenditure, and Economic Behavior – Questions asked for the respondents’ monthly income and estimated expenditure levels for alcohol and 11 product categories describing “erotic capital” (Hakim, 2010). The product categories were (1) clothing, (2) university apparel, (3) shoes, (4) makeup, (5) accessories and “style-items”, (6) haircuts, hair products and tools, hair coloring, (7) nail care and spas, (8) nutritional supplements, (9) gym memberships, sporting goods, (10) cosmetic surgery, (11) teeth whitening, (12) skin care. (6) Fitness – Respondents described the type of exercise they engage in and for what duration. (7) Houses and Row Parties – These questions related to specific aspects of Greek social operations. Respondents answered how many alcoholic drinks they typically consume during Sexual Economics Hernandez USC Economics 2014 18 7 types of student activities (Thursday night parties, football games, registered parties, “invites,” philanthropy events, and before consenting to sex), the percent of their fellow fraternity/sorority members that would successfully hook up on an average night, whether they use birth control or condoms, the fraction of fraternity funds spent on social events, and the average sex ratio in their fraternity’s parties. Female respondents were asked to identify the frequency of 4 types of sexually aggressive experiences (i.e. “creepiness”) at each of the 22 fraternities (Armstrong & Hamilton, pg. 89-90). (8) House Rankings – male respondents sorted the 9 sorority houses into 4 status categories: “Top house,” “Good house,” “Lower-Middle house,” and “Bottom house.” Female respondents sorted the 22 fraternity houses into the same categories. House status was recorded with 2 variables. The first is the category that a respondent sorted the house into and the second was the house’s rank within the category. The two were sorted into a single ordinal measure of house status where status equal to 1 is the “top house.” (9) Hookup/Relationship Inventories – The respondents reported their sexual histories in 3 stages. In Hookup Inventory A, men/women entered integers indicating the number of different partners they hooked up with in each of the sororities/fraternities or with Non- Greeks. In Hookup Inventory B, up to 5 rows appeared representing the 5 most recent hookup partners from each of the sororities/fraternities/Non-Greeks. The respondents entered 2 integers for each of the 5 listed partners, one for the number of non-penetrative hookups with that partner and the second for the number of penetrative hookups with that partner The concepts of non-penetrative and penetrative hookups were defined in the survey based on Olmstead et al. (2012). Hookup Inventory C asked respondents 10 questions pertaining to each partner from the same list in B. Respondents were instructed to consider the most recent hookup with each of their partners (from up to 5 partners per house) and report: the number of alcoholic drinks before the hookup, whether drugs were used before the hookup, whether they enjoyed the hookup, whether they successfully achieved orgasm, whether there was romantic interest before and after the hookup, and whether they were invited to some sort of exclusive Greek event before or after the hookup. Two separate but related relationship inventories were also conducted, similarly to Hookup Inventories A and B. Relationship Inventory A obtained the number of the respondents’ relationship partners per sorority/fraternity/Non-Greek. Relationship Inventory B measured the weekly frequency of non-penetrative and penetrative sexual activity within each of the reported relationships (up to 5 relationship partners per house) as well as whether the respondent was currently involved in the relationship at the response time. (10) Conclusion – at the end of the survey, respondents were asked to consider how many oral sex and intercourse partners they planned to hook up with in the following semester as well as whether their house of membership was the one they wanted. If it wasn’t, they were asked to elaborate prosaically. These qualitative results are shown in Appendix B. The second survey was given in-person to 2 administrators in the university’s Greek governance office. House-level variables were obtained, including total pre-recruitment members, estimated new members recruited, membership price for living in the house, membership price for Sexual Economics Hernandez USC Economics 2014 19 living outside the house, membership price for new members, total new members, and whether the organizations’ real estate is owned by the university. The administrators were briefed on the research topic with their own consent forms, provided the requested data from official documents, and were compensated with $20. Unfortunately, fraternity membership price data was not available to the administrators. The third survey was undertaken in response to the fraternity administrator’s lack of membership price data. Two rounds of mailers, including self-addressed stamped envelopes, were addressed to fraternity treasurers and sent to each of the 22 fraternities two weeks apart. The mail survey requested member counts and prices for the 3 membership types (live-in, live-out, and new) in addition to an itemized budget for the fraternity across some provided categories (Fraternity Sample Budget, tke.org). Treasurers were anonymously compensated with $15 via the mail. Their houses were also awarded a bonus of 10 membership points in the aforementioned prize competition. The membership point bonus was indicated on the treasurers’ informed consent documents. Combined, these created both personal and group incentives to participate in the third survey. Unfortunately, one 3 out of 22 treasurer responses were received. These data were not included in the subsequent models but summary statistics are reported in Appendix C. (IV) Preliminary Results There were three outcomes variables of great interest in the student survey: the number of individuals’ hookup partners, their hookup events, and their relationship partners during the fall semester. The frequency distributions of fraternities and individual female’ outcomes are shown in Figure 4. The participation rate for fraternity members was not large enough for the intended multivariate analyses. The sample included 76/477 men, few of whom completed hookup inventories. Therefore, women’s responses were aggregated at the fraternity level to obtain estimates of mean’s group outcomes. Note that the summary statistics between fraternities and women are not directly comparable due to the aggregation. Because of the low participation rate from fraternity members, Figures 4 and 5 present the same data points aggregated through different dimensions: in the fraternity data, each woman’s hookup/partner/relationship responses are added over each of 24 fraternity column vectors. The Poisson-estimated unconditional probability mass functions for sexual outcomes are given below in Figure 5. Some of the outcome distributions are not Poisson, especially fraternities’ partners. Moreover, each suffers from overdispersion because their variances are much greater than their expectations (Cameron & Trivedi, 2013; Honore). All of the probability mass functions underestimate the probability that the dependent variable is the given, x. The simulations are only lower bounds for the unconditional estimates because of overdispersion. Females’ partners and fraternities’ relationships have the least overdispersion because they have the largest mean-variance Sexual Economics Hernandez USC Economics 2014 20 ratio. With relative confidence, these nonparametric results suggest the probability that a female university student hooks up with 5 partners is at least 7%. The probability that she has 1 relationships is 18%. The probability that a fraternity’s members have a least 2 total relationships is at least 24%. Other distributions suffer from more extreme overdispersion such that their unconditional mass functions are not particularly illustrative. This was not the first study to experience sex-weighted bias in survey sampling. Wetherill also found that women were more likely to complete the peer awareness and caring survey on sexual attitudes. In a panel data study, men who dropped out before T3 reported more hookups than the ones who remained (Olmstead et al., 2012). The absence of individual member fraternity data was not particularly problematic for two reasons. First, sexual economics theory has typically emphasized women’s roles in markets (Nelson; Baumeister & Vohs). The trend in research seems appropriate because sexual health literature suggests women incur more deleterious mental health consequences from hooking up (Fielder and Carey, 2010). Women may also have higher STD risks from systematic misperception of risk (Corbin & Fromme, 2002; Kershaw et al., 2003). Second, relevant fraternity-level measures were constructed using women’s data inputs, as previously mentioned. One-way ANOVA results were computed to aid regression specification. 8 variables were determined to define economically relevant group differences: sex, fraternity number, sorority number, parent’s divorce (1,0), white (1,0), class rank, age, and parents’ income range. Results are combined in the Appendix D table. Sexual Economics Hernandez USC Economics 2014 21 Figure 4. Frequency Distributions of Fraternity and Women’s Outcome Variables Distribution Sample Statistics Mean: 37.35 Variance: 1521.84 Standard deviation: 39.01 Maximum: 153 Proportion of Zeros: 3/26 = 11.54% Mean: 56.96 Variance: 787.96 Standard deviation: 28.07 Maximum: 115 Proportion of Zeros: 0 Mean: 2.73 Variance: 38.04 Standard deviation: 6.17 Maximum: 32 Proportion of Zeros: 8/26 = 30.77% Sexual Economics Hernandez USC Economics 2014 22 Figure 4. Frequency Distributions of Fraternity and Women’s Outcome Variables Distribution Sample Statistics Mean: 5.71 Variance: 123.17 Standard deviation: 11.10 Maximum: 60 Proportion of Zeros: 169/278 = 60.79% Mean: 2.50 Variance: 21.16 Standard deviation: 4.60 Maximum: 31 Proportion of Zeros: 159/278 = 57.19% Mean: 0.23 Variance: 0.32 Standard deviation: 0.56 Maximum: 6 Proportion of Zeros: 220/278 = 79.13% Sexual Economics Hernandez USC Economics 2014 23 Figure 5. Unconditional Poisson Mass Functions Mass Function (Overdispersed) Simulations Probability that: Hookups = 20, Partners = 5, Fraternity Relationships = 2, Women’s Relationship = 1 Fraternities’ Hookups Fraternities’ Relationships Women’s Hookups Women’s Partners Women’s Relationships Sexual Economics Hernandez USC Economics 2014 24 (A) Economics of Men, Economics of Women Consistent with female resource theory and aristocratic matching, the data suggest that men and women have different ranges for acceptable partner attractiveness. The hypothesis that men match with their partners based on attractiveness rather than on social status/wealth is demonstrated because men should have a tighter acceptable attractiveness range. They are searching for the most attractive women to hook up with. In a one-sided t-test, the null hypothesis that acceptable- attractiveness-range is the same between men and women was rejected. The correlation between mother’s sorority membership and sex is positive, showing that women are more likely to have a mother who was in a sorority. It appears that sorority mothers are more influential at encouraging their daughters to join sororities whereas they are relatively ineffective at encouraging their sons to join fraternities. Women responded to the 8 extroversion statements much more agreeably than men. The statements read: I am the life of the party. I don’t talk a lot. I am interested in people. I feel comfortable around people. I sympathize with others’ feelings I seldom feel blue. I start conversations. I am not interested in other people’s problems. This contravenes intuition because men host all the parties. They should be the most extroverted. One-sided t-testing was robust to differences in variance between the sexes as well. The extroversion statements were not employed in regression because they lacked internal consistency. Cronbach’s alpha, a popular statistic for testing combinations of survey scores, was only 0.34 whereas the minimum acceptable score is typically 0.5. Sexual economics theory predicts that men and women should have significantly different hookup and relationship expectations. Men should have higher hookup expectations and women should have higher relationship expectations. One-sided t-tests with variance equality showed that both statements are true. Surprisingly, the sample means showed that women reported more actual hookups than men and the result was extremely robust to the removal of influential responses. Even when all individual hookup values greater than 50 were taken out of the analysis, the t-statistic is significant at the 99% level. Unequal variances were assumed between the two populations. This is quite surprising but almost definitely due to survey selection and completion biases rather than the true characteristics of men. Sexual Economics Hernandez USC Economics 2014 25 It is tempting to conclude that a higher mean of hookups for women signals higher market power for fraternity men. But survey selection bias is more likely because earlier survey questions on historical hookups values (high school and previous semester) did not have different means between men and women. If market power was structurally in favor of fraternities, previous semester hookups should also have been higher for women. Instead, it appears we have obtained a sample of men who didn’t complete the Hookups Inventories. Moreover, the lack of differences for high school hookups shows that the partial responses are most likely from men with low levels of the outcome variables because high school hookups is a significant predictor of college hookups (Wetherill et al., 2010). The correlation between sex and relationships is positive and the one-sided t-test showed that women have, on average, more relationship partners. Both sexual economics theory and sample selection/completion biases from men predict this result. Parents’ education involvement is higher for male students than female students. Perhaps parents monitor their sons’ educations more closely once they have joined a fraternity. Alternatively, male respondents could be more likely to score small levels of parental involvement highly due to an innate male desire for independence or some related latent variable. Hookup status anxiety is different between sexes at the 95% level which was not surprising because it was also 95%-confidently different across sororities and fraternities. Consistent with aristocratic matching, women have higher status anxiety and men have higher income status anxiety. I am 99% confident that there are sex differences in drinks per social event ad private alcohol expenditure. Men drink significantly more and spend more on alcohol. Sexual Economics Hernandez USC Economics 2014 26 Table 1. Significant Sex Differences in Independent Variables t-stat p-value Partner attractiveness range (1-10) -2.25 0.013 Mother’s Sorority Membership (1,0) -1.87 0.062 Extroversion (equal variances) -2.43 0.008 Extroversion (unequal variances) -2.308 0.011 Hookup Expectations 2.50 0.006511 Relationship Expectations -1.98 0.024 Hookups -3.22 0.0017 High School Hookups 0.45 -- Previous Semester Hookups 0.55 -- Relationships -1.33 0.093 Parents Education Involvement 1.97 0.050 Status Anxiety (hookup) -2.33 0.01 Status Anxiety (income) 1.78 0.08 Drinks (average per event) 4.94 9.6*10^ -7 Alcohol Expenditure 4.806 1.27*10^ -6 Sexual Economics Hernandez USC Economics 2014 27 (A) [Not] All Frats are Created Equal Some fraternities have significantly different mean ranges for partner attractiveness. This may indicate that men aggregate into fraternities based on sexual preferences (i.e. “higher standards”) (90% level). There is also very strong evidence of variation in self-reported attractiveness across fraternities (99% level). Some fraternities (or, perhaps merely the distributions of those willing to respond) have more, older members/members with higher class ranks (99% levels). A small variation exists between the private alcohol expenditures by fraternities as well as their fall and spring levels of hookups (90% level). (B) Neither are all Sororities... Some sororities have women who have significantly more siblings (95% level). This is a very neat result because the number of siblings was suggested as a proxy measure of fertility (Armstrong et al., 2012). Sororities might (implicitly or not) direct the most fertile participants into particular houses (presumably the highest status/lowest hookups houses). This reflects the core hypothesis under investigation: women who are most sexually desirable have incentives to join high status sororities in order to receive higher revenues on hookups. Armstrong and Hamilton went to great lengths to argue that erotic status characteristics were the primary interests of sororities when choosing new members (Paying for the Party, pg.80-86). Further evidence is given by Kendall’s tau, a tool used to express correlations over ordinal variables. It showed that the women with the most siblings (i.e. highest fertility) tended to be in the highest status sororities (tau = -0.11, p = 0.06). Sexual selection evidence is found in both high school and spring semester hookup rates. Both are significantly different across sororities at the 99% level. Obviously sororities do not select their members on their number of siblings. They may not even ask. But women may discuss hookup histories of potential recruits, implicitly obtaining an observable proxy for fertility that is useful for member selection. There are measureable differences for hookup status anxiety and social status anxiety between sororities (95% and 90% respectively) but not for income anxiety. In equilibrium, women know that men are selecting them as partners based on their physical and social attributes rather than economic ones. Kendall’s tau proved that higher status houses have most of the anxiety (tau = -0.16, p = 0.006). This provides significant quantitative verification of Armstrong and Hamilton’s stratification hypothesis, where erotic capital items were especially important for the stratification process. The score shows that social status is a significant driver of inter-sorority competition. It was very interesting that body-mass index (BMI) varied significantly across sororities but not fraternities. Weekly fitness time did not very between either sets of groups. Women have lower BMI. One might find this intuitive because men tend to be heavier than women. However, BMI accounts for this to provide the same scale for both men and women. The BMI finding corroborates previous research about eating disorders in sororities (Mize & Valliant, 2012; Adams & Araas, 2006; Allison & Park, 2004; Brown et al, 1990). High status sororities have the lowest BMIs (tau = 0.21, p = 0.000), suggesting this is a major mechanism for competition. Furthermore, this demonstrates that men Sexual Economics Hernandez USC Economics 2014 28 select partners on physical attractiveness (including physique) while women select on status and wealth. Women would not mind dating or hooking up with men who have above average BMIs. Lastly, there was a 90% significant difference in personal income among sororities. Personal income did not vary across any of the other groups except for parents’ income range. (C) Does parents’ divorce make you much different from a sexual economics perspective? Among respondents whose parents are divorced, a 95% difference was found for father’s education. Upon further inspection, the correlation between father’s education and divorce is -0.117, suggesting that higher father’s education is associated with lower divorce probability. Consistently, the correlation between divorce and parents’ income range is negative. Divorced parents have a lower income range. Divorce is associated with higher high school hookups but not with any other type of sexual behavior. (D) Do whites have a sexual-economic advantage? Some fraternities had more white respondents than others (90%) and the same effect existed at the 99% level for sororities. This provides strong evidence to race concerns that have attracted national attention to Greek systems. In the famous case of University of Alabama, a sorority, rather than a fraternity, was the primary object of criticism (The Guardian). It was natural to test whether the race differential traced out the Greek social hierarchy. That hypothesis was confirmed by a Kendall Tau correlation test (z = -2.1, tau = -0.13, p = 0.04). The parents of white students have significantly more education (95%) and were much more likely to have been in a Greek system themselves (99%). These were, perhaps, the most robust findings from ANOVA because they are strongly consistent for both parents. White students had higher gpa’s (99%) even though they had lower parental education involvement (99%). (E) How are older students different from younger students? Older and higher-class-ranked students expected to participate in significantly more committed relationships (99%). They followed through: older members participated in more committed relationships (90%, not robust to class rank). Older students had much higher private alcohol expenditures, showing that college and the Greek system do not eliminate all the transaction costs to alcohol acquisition under age 21. (F) How does parents’ income affect the sexual economy? Unfortunately, it cannot be utilized as a regressor because it was collected as a set of income ranges, rather than monetary values. But this structure does make it ideal for ANOVA. Parents’ income does significantly increase student’s income (95% level), as well as their erotic capital and personal alcohol expenditures (99%). It varied significantly across sororities but not fraternities. Parents’ income also varies with siblings. The correlation is greater than 0, indicating that richer Sexual Economics Hernandez USC Economics 2014 29 families have more kids. Parents’ education and parents’ Greek system membership vary directly with income range, which is not surprising. As previously stated, higher parents’ income decreases the probability of divorce. Quite fascinatingly, if a sorority member has richer parents, she is likely to assess a higher value for the average hookup rate within her sorority (95%). This suggests a relationship between income and peer effects, which are explored in Section VII. Finally, as parents’ income goes up, so does income status anxiety and social status anxiety. That is perhaps counter-intuitive because one might expect that poorer people are more anxious about increasing their incomes. However, the articulation that Armstrong and Hamilton gave with their female party floor ethnography showed that middle class and extremely well-off students were most afraid of not “fitting in” socioeconomically (Paying for the Party, pg. 73-81). The same hypotheses are used for macroeconomic evaluations of income status anxiety (Layte & Whelan, 2013). Parents’ income affected income and social status anxiety but not hookup status anxieties. All of these results are predicted by the social economics theory of status. In high income societies, competition for status will be more vigorous than competition for goods and the prices of status items will rise because the supply remains fixed as growth increases incomes (Becker et al., 2005). (V) Analysis (a) Model Specification My first task is to specify a theoretical model that demonstrates the functions of fraternities and sororities within the overall sexual marketplace. This level of analysis focuses specifically on the market holons Greek Sex Demand and Greek Sex Supply, shown in Figure 6 Recall that according to sexual economics theory, men’s incentives are to maximize the utility of sex subject to expenditure, while women maximize profit by maximizing transfers from men (revenue) subject to their own sexual costs. Women who receive high prices in terms of alcohol are likely to also receive high prices in terms of other goods. Then, the error term in the supply equation is correlated with the dependent variable and an inefficient and inconsistent estimator is obtained. I give attention to this concern while obtaining the model specification. Sexual Economics Hernandez USC Economics 2014 30 Exogenous preferences Sorority members Fraternity members Greek Sex Supply Greek Sex Demand Non-Greek Sex Supply Non-Greek Sex Supply Market Sex Demand Market Sex Supply Market Sex Price and Output Bargaining Figure 6. The Local Bilateral Oligopoly. The relationship between status, alcohol transfers, and hookups is limited to the Greek sexual marketplace. Although non-Greek women also exchange hookups for drinks in the model, their lack of sorority membership leaves them without an explicit social status ranking. They do not have the market protection that sororities are hypothesized to provide. Sexual Economics Hernandez USC Economics 2014 31 This hypothesis can be presented as a supply-demand model: Equation Set 1: Simultaneous equations model (SEM) for Greek Supply and Demand The supply equation indicates that the quantity of hookups given by a woman, i who is in sorority g, is a function of the price she receives, some exogenous attributes, X ig , which relate to herself, and attributes, Y g , which relate to her sorority. In the demand equation, the aggregate price of sex offered by fraternity, f, is a function of the quantity of hookups that the fraternity receives, as well as exogenous attributes, Y f , which relate to the fraternity only. This structure, with individual and group level variables for sororities and only group level variables for fraternities, reflects the result of the data generating process, characterized by significantly more sorority than fraternity responses. Fraternities’ individual-level variables are intractable because there is both status-based selection and a small sample size. Fraternity-level measures for hookups and “prices” were assembled from women’s responses in Hookup Inventory C – where hookup-event attributes were matched to fraternity houses. What do I meant by “price” of sex, exactly? This turns out to be an econometrically important question. Men might trade many valuable commodities to women to inspire sexual interest; not only resources but also “affection investment,” time, and energy Townsend & Levy (1990). To estimate these equations, I begin by treating the model as if alcohol consumption is an appropriate proxy measurement for the resource price of a hookup. This seems an appropriate treatment because an analysis involving “affection,” time, and “energy” proxies could be highly sensitive to the design of those measures. Alcoholic beverages are both a prominent predictor of hookups and a significant component of the variable costs of fraternity parties (Bersamin et al., 2011). Students with the lowest alcohol intoxication frequencies never hook up (Paul et al. 2000). Alcohol increases hookups not only by lowering inhibitions for sexual encounters but also by giving students an “anticipatory excuse” where students blame their behavior on alcohol (Paul, 2006, pg. 151). Therefore, we may have a unique situation where it is appropriate to treat resources as the primary driver of hookup prices, specifically. Fixed costs such as house infrastructure are clearly present. Other costs of parties might include DJs, lighting, security, and decorations. But they must be neglected due to the data quality. A consequence of measuring “price” in terms of alcoholic beverage is effectively a numeraire assumption that the price of alcoholic beverages is equal to 1 and, therefore the equilibrium number of hookups divided by the equilibrium alcohol transfer is equal to the “price of hooking up” (sex). Later, this price could be multiplied by an average price of alcohol to estimate a “realistic” price that Sexual Economics Hernandez USC Economics 2014 32 relaxes the numeraire assumption. But when there are other relevant resources exchanged for hookups, the numeraire assumption is violated and an endogeneity problem arises in the estimation. I begin by estimating the simplest possible model that could confirm the hypothesis. Let the price of hookups be expressed in numeriare terms of alcohol beverages and Y g and Y f include only the fraternities’ and sororities’ statuses. X ig is empty. This model can be written as: Equation Set 2: Status-specification for SEM. The coefficients in the demand equation, Q f , have changed to reflect the fact that Q is now on the left side of both equations. It is preferred because there is a suspected measurement error in number of hookups. Keeping the suspected-error limited to the dependent variable will not upset the model’s consistency (Cameron and Trivedi, 2009). OLS and overdispersed Poisson regression results for both the supply and demand equations are shown in Table 2. The prediction is that in the sororities equation, price (average drinks per party event per woman) will have a positive coefficient and status will have a positive coefficient. Recall that a positive coefficient on status actually indicates a negative relationship between status and hookups because the top house has the minimum value of 1. In Appendix E, the outcome variables are plotted according to each fraternity and sorority member’s status. Sexual Economics Hernandez USC Economics 2014 33 Table 2: OLS and Poisson Results for Estimation of Supply and Demand (Hookups) Variable Fraternity Demand Equation Sorority Supply Equation Ordinary Least Squares Intercept 18.36 (32.96) 4.19*** (1.46) Drinks (price) 0.197 (0.25) 0.44 (0.04)*** Status 0.51 (1.54) -0.51 (0.20) Multiple- R 2 0.032 0.023 Adjusted-R 2 -0.06 0.004 Poisson Regression Intercept 4.18*** (0.07)*** 1.59*** (0.06) Drink (price) 0.22*** (0.06) 0.032*** (0.0009) Status -0.29*** (0.03) -0.032*** (0.009) Residual Deviance/df 0.57 0.48 Quasi-Poisson Results Intercept 4.18*** (0.19) 1.59*** (0.27) Drink (price) 0.22 (0.16) 0.032*** (0.004) Status -0.30*** (0.087) -0.032 (0.042) Dispersion Parameter 7.61 19.11 Sexual Economics Hernandez USC Economics 2014 34 The OLS results produced no significant regressors and R 2 values near zero. Because the dependent variable is mostly Poisson-distributed, conditional homoskedasticity of the error term causes OLS to be inefficient. This is why we see comparatively larger standard errors in the OLS than Poisson models. In the basic Poisson regressions, the coefficient on sorority status reflects the opposite of the hypothesis that higher sorority status leads to fewer hookups. We can run a numerical simulation with the model to verify the direction of the results. Imagine two individual women who receive the same amount of alcohol from fraternities over the semester, let’s say 50 drinks total. One of the women is in the top sorority (1) and the other is in the bottom sorority (9). The Poisson model’s prediction for their hookups is: Equation 3: Simulation of Poisson Regression for Supplies. When the prices in terms of drinks are controlled for, women with higher status are predicted to hook up more. As their status rank decreases from 9 toward 1, a negative coefficient drives their hookups upward. The direction of this relationship of fraternities was correctly predicted by sexual economics theory: lower status fraternities are predicted to hook up less. A numerical example for fraternities follows: Equation 4: Simulation of Poisson Regression for Demands. The quasi-Poisson model corrects for overdispersion by estimating a corrective parameter based on the ratio of residual deviance to degrees of freedom, also known as the “scale factor.” The quasi estimator ascribed all explanatory power to prices rather than sororities’ status. Even though this is the most robust model, this is not an entirely satisfactory result. Ample evidence from the anthropologists, women’s written responses in Appendix B, and ANOVA showed that status plays a significant role in sexual economics outcomes for women. In the quasi-Poisson estimate of the demand equation, the opposite effect was shown. Status took up all explanatory power, leaving prices insignificant. These results are not surprising because the effect of the quasi-Poisson estimator is to correct the standard errors to account for dispersion. For further inquiry, I extend the model based on theoretically important omitted variables and then run regression on the partners and relationships outcomes. Sexual Economics Hernandez USC Economics 2014 35 (b) Heterogeneity and Alternative Prices Omitted-variables endogeneity could explain the reversed-causality of sorority status on hookups. For example, a latent variable such as sexual desirability could be positively correlated with status and with hookups. With an instrument that purified status of its correlation to sexual desirability, the hypothesized protective power of top sororities could emerge in the regression results. Before taking such extreme econometrics measures as Poisson instrumental variables, I implement a proxy variables strategy for measuring the omitted latent variables, sexual desirability. Proxy estimation would be preferred to Poisson-IV for its simplicity. Instrumental variables in a non-linear setting call for distributional assumptions about the status rankings (Cameron & Trivedi, 2009). I must relegate these possibilities to future research for the time. The Poisson regression models with sexual desirability proxies are shown in Table 3. Proxy variables for women’s sexual desirability were her self-reported attractiveness and body-mass index (Figure 7). The proxy measurement of sexual desirability did not reverse the coefficient on sorority status in any models. In the first model of Table 3, self-reported attractiveness is not a significant predictor of hookups (p = 0.12). If the omitted variable does drive hookups and create an endogeneity problem that hides the true direction of sorority status, then self-reported attractiveness is not a good estimator of it. In the second model, both self-reported attractiveness and body-mass index were used a proxies for sexual desirability because several papers in health journals identified significantly greater attractiveness-motivated weight concerns within sororities (Allison & Park, 2004; Hoerr; Schulken, 1997; Adams & Haraas; Brown et al., 1990). The relationship between attractiveness and body mass index was confirmed by a correlation test between self-reported attractiveness and body-mass index (r = -0.16, p<0.0059). BMI even beat self-reported attractiveness as a significant (positive) predictor of hookups (p<0.00). These results are consistent with the sexual economics theory that low attractiveness women should hook up more for a given price. The estimated coefficients directions on the proxies are consistent with measuring sexual desirability but sorority status rank is remains a negative predictor of sorority women’s hookups. In fact, the inclusion of sexual desirability variables drove the status coefficient even further negative. Membership in a high status sorority increases hookups, even for more sexually desirable women. But according to sexual economics theory, the most desirable women should have the market power to hold off on hooking up. If there are many men in queue for their attention, they should select the one with the highest willingness to pay. In equilibrium this bidder offers resource like affection and commitment, which is exclusive with the hooking up. The available evidence, therefore, rejects the hypothesis that sorority status enables women to hook up less. In Appendices F and G, I used a non-parametric method to search for instrumental variables for sexual desirability. By comparing non-Greek women to sorority women, as well as “top” vs. Sexual Economics Hernandez USC Economics 2014 36 “bottom” sororities and fraternities, I find valid and relevant instruments along with new insights about the industrial organization of Greek sexual behaviors. These appendices will be of greate interest to students, faculty, and administrators. The third model in Table 3 shows that the Poisson results are robust to including additional resources besides alcohol as part of the price of sex. It is a distinguishing characteristic of sexual economics that many currencies can be traded at once (Townsend & Levy, 1990; Baumeister and Vohs, 2004). A major challenge for the field is to derive solutions for the econometric problems that arise when certain prices are excluded. The case of university Greek systems is unique in that alcoholic beverage trades were expected to be the most significant price factor for institutional reasons. Taking the specification from Equation Set 2, we can see how unobserved transfers, such as gifts, relationship perceptions would cause the idiosyncratic error, u ig , to correlate with the specified prices and thus lead to inconsistent estimation. For example, the relationships regressor in the third model of Table 3 suggests that women can also be compensated for hookups with a future relationship. Future research that tried accounting for romantic transfers including dates and romantic gifts would significantly advance this line of estimation. Sexual Economics Hernandez USC Economics 2014 37 Table 3. Supply with Proxy Variables Variable Sorority Supply Equation Regressand: # of hookup events Poisson Regression Quasi- Poisson 1 Proxy 2 Proxies 2 Proxies - 2 Prices 2 Proxies - 2 Prices Intercept 2.25*** (0.24) 1.30*** (0.40) 1.14*** (0.40) 0.83 (1.63) Drinks (price) 0.03*** (0.001) 0.03*** (0.001) 0.03*** (0.001) 0.036 (0.40) Relationships (price) 0.21*** (0.000) 0.11 (0.27) Status - 0.05*** (0.01) -0.06*** (0.011) -0.055*** (0.01) -0.054 (0.047) Self-Reported Attractiveness -0.04 (0.03) -0.024 (0.03) -0.005 (0.03) -0.011 (0.12) Body-Mass Index -- 0.041*** (0.003) 0.04** (0.01) 0.07 (0.057) Residual Deviance 2469 2347 14.23 2412 Df 154 148 81 153 Dispersion 18.59 Sexual Economics Hernandez USC Economics 2014 38 Figure 7. Attractiveness and BMI Distribution and Plot Sexual Economics Hernandez USC Economics 2014 39 (c) Partners and Relationships Regressions There is no evidence to suggest that gaining entrance into a higher status sorority allows women to restrict the number of hookups they undertake. But if they reduced women’s number of hookup partners, there would be room for further commentary. Three supply equation specifications are estimated in Table 4. The dispersion factors reveal that the simplest model is the best fit. A scale factor equal to 1 indicates that the variance exactly equals the mean. Because we have a nonlinear model, it is possible to reduce the goodness-of-fit by adding more regressors if they are not powerful enough. Status loses its significance, so there is no evidence that top sorority women have more hookup partners, all other things equal. The same models were calculated over the relationships dependent variable, shown in Table 5.There is no 2-proxy, 2-price model because relationships cannot be exchanged for relationships and hookups are not valuable currencies for women to allocate relationships. We should expect, then, that this relationships model suffers from some level of endogeneity from other resources. Nonetheless, we have the best fit Poisson models because the sample mean was relatively close to the sample variance. As well, we see the scale factors are finally below 1. The scale factor for the second model is the smallest yet achieved although none of the standard errors are small enough to confirm significant of any independents. For this reason, I do not present the quasi-Poisson standard errors. In future research, a survey that was more specifically tailored for relationship responses would lead to more impressive results. Sexual Economics Hernandez USC Economics 2014 40 Table 4. Supply of Hookup Partners Regressand: # of hookup partners Poisson Regression Quasi- Poisson Simple Sexual Desirability Proxies Proxies + 2 Prices 2 Proxies - 2 Prices Intercept 0.92*** (0.097) 0.44 (0.74) 0.57 (0.77) 0.57 (3.63) Drinks (price) 0.006** (0.003) 0.097*** (0.02) 0.12*** (.0.02) 0.12 (0.97) Relationships (price) -1.70*** (0.32) -1.70 (1.50) Status -0.017 (0.015) -0.12 (0.019) -0.01 (0.019) -0.01 (0.09) Self-Reported Attractiveness 0.14*** (0.05) 0.12 (0.05) 0.12 (0.24) Body-Mass Index -0.04 (0.026) -0.04 (0.03) -0.04 (0.13) Residual Deviance 1979.9 1488.2 1431.5 1431.5 Df 235 148 147 147 Dispersion Factors 8.425 10.06 9.74 9.74 Sexual Economics Hernandez USC Economics 2014 41 Table 5. Supply of Relationships Regressand: # of relationships Poisson Regression Simple Sexual Desirability Proxies Intercept -2.17*** (0.48) 0.96 (3.25) Drinks (price) 0.032*** (0.013) 0.11 (0.08) Relationships (price) Status 0.05 (0.064) 0.02 (0.09) Self-Reported Attractiveness -0.33 (0.24) Body-Mass Index -0.03 (0.12) Residual Deviance 96.56 64.68 Df 142 82 Dispersion Factors 0.68 0.79 Sexual Economics Hernandez USC Economics 2014 42 (d) Quasi-Significant Regressors For a final set of models, I used a large number of regressors in a quasi-Poisson model to obtain a few significant findings for each dependent variable (Table 5). Insignificant results are not reported. Nnegative covariates on hookup events and partners are of particular interest for policy reasons. These include fraternity parties per night, private alcohol expenditure, and hookup status anxiety. I suspect that women who attend many parties in one evening more easily deflect sexual advances in individual parties by leaving. But the same women will have more hookup partners. They might be meeting more men over time compared to women who only attend a few parties. The data suggests that women who attend many parties are behaving very competitively. They identify several potential partners but distribute more hookup events to a select few. Arguably, attending more parties per night gives women more power to distribute their sexual resources on their own terms. But, having more partners is also associated with heightened STD risks. Private alcohol expenditure is negatively correlated with hookups and positively correlated with relationships because women who buy alcohol for themselves simultaneously forgo the opportunity to receive a drink from a man, as the theoretical models showed. Notice that partner attractiveness range increases both hookups and relationships. This is a nice result for matching theory: when a man or woman has a lower preference for the opposite sex’s attractiveness, they search to hookup out of a larger pool of individuals. The college spouse variable shows that the more a student intended to find a spouse in college, the less likely they were to find a relationship. The distribution of this regressor showed that nearly all respondents were uninterested in finding a spouse. Therefore those who are interested will not find partners who share their same long-term relationship goals. Regnerus discussed the full implications of asymmetric preferences for marriage (Sociology, 2012). Several other variables should be of note to women because they decrease the probability of having a relationship. Women who have more Facebook friends are less likely to have a relationship. A person in a relationship has less free social time than a person who isn’t (assuming time on social media is correlated with number of Facebook friends). Having more siblings, a mother who was in a sorority, high masturbation frequency, or high self-reported attractiveness also reduce the probability of having a relationships. Some suggestions by Armstrong and Hamilton might be relevant. They believed that the ideology of “college of the best time of your life” is the foundation of the party pathway, and there is likely to be a relationship between how women process the ideology and how their mothers represented the sorority experience to them. Further research would be needed to make any definitive conclusion. Sexual Economics Hernandez USC Economics 2014 43 Table 6. Quasi-Models of Many Regressors Quasi-Poisson Hookup Events Hookup Partners Relationships Intercept White Class Rank Age GPA Siblings -0.77* (0.42) Father Education Mother Education Father’s Fraternity Mother’s Sorority -2.48*** (1.072) Parents’ Divorce Parents’ Involvement Self-Reported Attractiveness -0.79* (0.45) Parents’ Financial Aid Facebook Friends 0.0007*** (0.0002) -0.0017* (0.0009) Student’s Income Hookup Status Anxiety -0.20* (0.12) Social Status Anxiety 0.125* (0.074) Fraternity Parties Per Night -0.45* (0.26) 0.17* (0.100) Birth Control Rate Previous Semester Hookups Masturbation Frequency -0.68* (0.29) Pornography Frequency High School Hookups Hookup Sexual Economics Hernandez USC Economics 2014 44 Expectations Relationship Expectations Private Alcohol Expend. -0.02* (0.008) 0.034** (0.015) College Spouse -1.49*** (0.48) Fitness Minutes Partner Attractiveness Range 0.30* (0.15) 0.7*** (0.25) Avg. Drinks 0.42*** (0.167) Body-Mass Index 0.10* (0.06) Relationships Residual Deviance 493.71 401.9 18.995 Df 32 89 40 Dispersion Factor 15.43 4.52 0.47 Sexual Economics Hernandez USC Economics 2014 45 (VI) Technical Issues Ordinal Regressor The most important regressor in this analysis was the ordinal status variable for fraternity and sorority status. However, I have been incorrect to assume that identically distributed as the other regressors. Status cannot be identically distributed because it is definitely non-normal. Ordinal rankings are not continuous and are bounded at 1 and 9 for sororities or 1 and 24 for fraternities. It has been shown that without additional modification, most linear model specifications will be biased with ordinal regressors (Kukuk). Methods for estimating models with ordinal regressands include cumulative link models (Christensen). Unfortunately, software methods for implementing cumulative-link- like corrections on a regressor were not available. A method for ordinal variables in GLM was given by Johnson (2006) and could be the object of future analyses of Greek systems and social economics. Fixed Effects In Section V, I exploited an observed fixed effect and observed individual regressors and faced the specified error structure (Equation Set 2). Normally, unobserved fixed effects lead to inconsistent estimates. However, Poisson-distributed dependent variables avoid this problem when regressors are strictly exogenous (Blundell, Griffith, and Windmeijer (1997); Lancaster (1997)). This provides a strong basis for the proxy variables approach to sexual desirability and alternative sexual currencies, such as relationships. In order to meet the necessary condition for strict exogeneity, full parameterization implies that all the fixed effects which are correlated with hookups can be summarized by social status. This is a reliable assumption based on Armstrong and Hamilton’s evidence on status along the party pathway (Paying for the Party, pg. 70-80). Dispersion We saw that Poisson regressions produced much more interesting results than their quasi counterparts. The quasi models did not change the interpretation from the coefficients but corrected the size of the standard errors for the dispersion, which is the extent to which the sample distribution departs from the Poisson distribution, where the mean is equal to the variance (Honore). In models where overdispersion is high, the quasi-model’s effect on estimation is much higher standards of proof for regressor significance. The robust quasi- Poisson estimates do produce some sexually-economically relevant information, which was presented in Table 6. Another possibility referenced in the statistics literature is to use the negative binomial distribution for the dependent variable (Cameron & Trivedi, 2013). Sexual Economics Hernandez USC Economics 2014 46 Consistent Specification The generalized linear models (GLM) estimates will be inconsistent if the conditional mean of the distribution is incorrectly specified. In future work on sexual econometrics, it will be vitally important to hypothesize correct distributions for regressors. Residual analysis can be used to discover nonlinear relationships between regressors and dependents (Cameron and Trivedi, 2009). Collinearity When a few regressors are highly correlated with each other, their coefficients can have finite sample bias even though they are asymptotically consistent. If collinearity is bad enough, it is wise to interpret collinear coefficients together. Essentially, their marginal effects are indistinguishable from each other. Marginal analysis can still be conducted by considering quasi-ceteris paribus simulations where both of the collinear regressors increase simultaneously. Table 7 shows the correlation matrix of regressors in the status analyses. Significant collinearities are notes with *. Even though r values seem small by conventional standards, the -0.18 correlation between BMI and self-reported attractiveness passed a correlation test at the 99% level. Collinearities for status were calculating using Kendall’s tau method for ordinal variables which has a slightly different interpretation for tau than for r. The most significant collinearities are with sorority status. Both sexual desirability proxies are very strongly correlated with status. Higher status women report more attractiveness and have lower BMIs. The correlation matrix suggests that these models are not robust for comparing a simulation between a women of attractiveness 10 in a bottom house versus one in a top house. Such a case is almost never observed in the data, which generated the collinearity. When making simulations, it would be more consistent to use reasonable assumptions where the top status women are consistently more attractive. This is perhaps a minor detail. Notice that women’s pre-hookup drinks are not collinear in the model, so we should still have a robust measure of the price of hookups. Sexual Economics Hernandez USC Economics 2014 47 Table 7. Collinearity Matrix of Important Regressors Drinks Status SRA BMI Relationships Drinks 1 Status 0.09 (tau) 1 Self-Reported Attractiveness -0.10 -0.23*** (tau) 1 BMI 0.01 0.30*** (tau) -0.18*** 1 Relationships 0.09 0.03 (tau) -0.02 0.03 1 Sexual Economics Hernandez USC Economics 2014 48 (VII) Peer Effects Peer effects estimators are one of the hottest controversies in contemporary econometrics. A rather simple notion, peer effects hypothesize that there is a “social multiplier” whereby individuals’ peers’ values for a dependent variable have a unique effect on the individual’s own value of the same dependent variable. For example, instead of wondering whether a sorority’s status rank decreases hookups, one could hypothesize that a woman is more likely to hookup if more of her sorority sisters hookup. Similarly, Ali and Dwyer (2011) used Add Health data to get an instrumental variables estimate of the peer effect on adolescents initiating sex. In their study, a 10% increase in the proportion of close friends who initiate sex increased the probability of the individual initiating sex by 5%. A 10% increase in close friends’ number of sexual partners increased the individual’s number of sexual partners by 5%. An additional way of considering it is as a positive feedback within peer groups. Brock and Durlauf (2001) described it as a “classical nonpecuniary externality.” If there is a social multiplier, then it will be an endogenous omitted variable in the above analysis. In this section, I specify a peer effects model to sorority sexual economics after a brief review of the literature. The model is estimated and a unique alternative is considered. Manski (1993) is attributed with first noting the nature of peer effects. He called it “the reflection problem” because the value of the dependent variable is a constituent of the peer-mean. In turn, the peer effect “reflects” the dependent variable, creating a unique type of endogeneity problem. Peer effects have also been hypothesized for exogenous regressors. For example, in Section V, self-reported attractiveness was negatively correlated with a woman’s hookups. Each sorority’s mean self-reported attractiveness could be an exogenous peer effect (also called “contextual effect”) that may increase individual i's number of hookups. Peer effects models have been widely applied while their theoretical soundness has been contested among econometricians (Angrist, 2012). The literature could be categorized by various styles of peer groups. The first type, where households are grouped into peers/neighborhoods, has been used by macroeconomists to explain social ills such as school dropouts and teen fertility (Case and Katz, 91’; Crane, 91’). The second type employs the National Longitudinal Study of Adolescent Health (Add Health) for its peer-to-peer specific construction. Add Health respondents nominate their 5 closest male and 5 closest female friends during an interview. Observations where both friends are Add Health respondents are common and provide for individually-defined peer groups (Ali and Dwyer 2011; Fletcher, 2011). The third type of peer structure is the one presented here. The data is comprised of many individual responses with a natural grouping by Greek organization. The structure can be described as a quasi-panel such as that specified in Equations Set 2. There are G sororities where g is 1-9. Each sorority is composed of i individual respondents. This implies that there are two distinct error terms: a fixed effects error term for the sorority, α g, and an individual error term, u ig. The fourth type of peer effects groups are a set of novel instances of random assignment. Glaser et al. (2003) estimated peer influences on propensity to join Greek houses based on the Sexual Economics Hernandez USC Economics 2014 49 random dorm assignments at Dartmouth. The authors found that high school drinking behavior was a positive predictor of freshman year pledging (about 50% of Dartmouth students go Greek). The identification method used by the original authors cleverly exploited the nature of their data. After establishing very highly granular peer groups based on buildings, floors, and individual roommates, endogenous peer effect were instrumented by the covariates of individuals who are two degrees separated. That is, non-peers of the individual, i, who are peers of individual i's peers, provide some data which is correlated with the peer-meaned dependent variable value (i.e. the peer effect) while avoiding direct correlation with the initial individual’s value. Angrist (2012) used the paper to illustrate his critique of the peer effects method, suggesting it as an extreme example of many-weak instruments. The standard errors of peer effects coefficients increased as the models’ groupings increase in granularity. Angrist presents the peer effects researchers with a double bind: peer effects are either (1) a mechanical result that follows from the definition of statistical expectations or (2) they are relatively vapid measure of group clustering. He, furthermore, warns that adopting viewpoint (2) would motivate many spurious investigations of contextual effects. The gold standard of peer effects proof is a hypothesis test that the coefficient is different from 1 (Angrist, 2012). Can sexual economics weigh in? Most likely, yes. In Section 2, I reviewed a paper by Wetherill et al. (2010) which took survey estimates from college students asking how much they perceived that their peers were aware and caring about their sexual behaviors. They found that more perceived caring was correlated with fewer hookups. In the case of peer effects, a slightly different measurement, peers’ real outcome values, is tested as a predictor of the individuals’ outcomes. Based on the available sexual economics data, I compare the traditional peer effects estimator to an exogenous measurement of the peer effect based on individuals’ subjective expectation. The peer effects model for sorority hookups can take the following form: There are at least three possible identification strategies for the current data. Lee (2007) demonstrated a method to exploit differences in group size for a consistent conditional maximum Sexual Economics Hernandez USC Economics 2014 50 likelihood estimator. Brock and Durlauf (2007) showed that certain nonlinear models might be identified, although this has yet to be shown for the Poisson distribution. The third identification method was Graham and Hahn’s panel data derivation from the within and between estimators (2005). Unfortunately, with only 8 sororities in the data set, there was not enough “between” variation to obtain an identified estimator. As well, the data is not individually matched, so the popular method of instrumenting with second-degree separated peers does not apply. Instead, I obtain results from pooled regressions where all members of the same sorority are assigned the proper peer effects value. The results for coefficients will be inconsistent when the peer effect is measured in the conventional manner as the sorority-mean of the outcome. Therefore, my primary interest is in comparing the goodness of fits between the conventional peer effect and a new subjective/exogenous measurement. In the latter set of models, I expect the estimate to be consistent, albeit with a slightly different interpretation. To reduce the amount of missing data, I employed an alternative drinking independent variable. The data used for hookup prices in Section V was collected toward the end of the survey in Hookup Inventory C. I could match these drink values to fraternities whereas the new drink variable is the individual’s average alcoholic beverage consumption in 7 different types of common student events. The ANOVA in Section 3 suggested including the women’s income because the exogenous measure of the peer effect (referred to as “sister hookup rate”) varies across parents income groups, which are correlated with individual’s income. Baseline results are reported below in Table 8. Note the dispersion factors in Table 8. I treat them as the lower bound for goodness of fit of the peer effects estimator. If it cannot improve the estimation beyond established sexual economics principles, it is rightfully considered useless. To construct an upper bound, I replaced sorority status with dummy variables for each sorority group. Dummies will capture all sorority-level effects, including contextual effects and endogenous peer effects. In the relationships model, specifications with dummies were not identifiable because there are only a few observations. Sexual Economics Hernandez USC Economics 2014 51 Table 8. Baseline Results for Peer Effects Models (Compare to Section V) Poisson Regressions Hookup Events Hookup Partners Relationships Intercept 0.48 (0.52) 2.85*** (0.82) 3.875 (2.94) Drinks 0.08*** (0.012) 0.24*** (0.01) -0.315 (0.23) Relationships -0.12 (0.079) -1.03*** (0.15) Income -0.000034*** (0.000014) -0.000022 (0.0000245) 0.000072 (0.00006) Status -0.44*** (0.013) -0.05*** (0.01) 0.026 (0.089) Self-Reported Attractiveness 0.059 (0.037) -0.09* (0.054) -0.315 (0.23) BMI 0.071** (0.016) -0.051* (0.027) -0.124 (0.106) Residual Deviance 1532.1 607.72 57.295 Df 82 82 83 Dispersion Factor 18.68 7.41 0.69 Sexual Economics Hernandez USC Economics 2014 52 Table 8. Sexual Outcomes with Sorority Dummy Variables Poisson Regressions Hookup Events Hookup Partners Relationships Intercept 0.99 (0.612) 3.50*** (0.9954) Drinks 0.079*** (0.013) 0.258*** (0.0124) Relationships -0.079 (0.08) -0.87*** (0.16) Income -0.00003** (0.000014) -0.000012 (0.000025) Status Self-Reported Attractiveness 0.053 (0.038) -0.113** (0.0565) BMI 0.076*** (0.08) -0.047* (0.027) Sorority 1 -0.61*** (0.21) 0.003 (0.34) Sorority 2 -16.63 (46.93) -16.96 (77.38) Sorority 3 -- -- Sorority 4 -0.65*** (0.20) -0.83** (0.33) Sorority 5 0.326 (0.25) -0.20 (0.44) Sorority 6 -0.78*** (0.202) -1.089*** (0.34) Sorority 7 -0.98*** (0.20) -1.42*** (0.34) Sorority 8 -1.163*** (0.22) -1.15*** (0.37) Sorority 9 -- -- Sorority 10 (No House??) -- -- Sorority 11 (non- Greek???) -- -- Residual Deviance 1461.2 512.04 Df 77 77 Dispersion Factor 18.39 6.65 Sexual Economics Hernandez USC Economics 2014 53 The results are comparable to Tables 2, 3, and 4 in Section V. In addition, women’s income is significant and negatively correlated with hookups, with confirms Hakim’s (2010) predictions for erotic capital theory. Women who acquire more economic capital have fewer incentives to trade erotic capital for economic capital at the margin. Unlike usual goodness of fit measures, the dispersion factor may increase when additional regressors are added (Cameron and Trivedi, 2009). Given that 6 additional regressors were added (several dummies are NA due to in-sample identifiability problems), the peer effects estimators have an unfair advantage in this goodness of fit test because they will consume fewer degrees of freedom which gaurantees smaller dispersion factors. The bias against dummies is corrected in Table 12, where I show the dispersion factors as if dummies consumed only 2 df combined. The dummy’s coefficients are all negative which contrasts the results obtained using the ordinal measure of sororities’ status. Does joining a sorority reduce the individuals’ number of hookups? In Table 9, the significant dummy coefficients are matched to each sorority’s status. Kendall’s tau correlation test verified that the magnitude of the coefficients is not significantly related to the status of the sorority, so the results in Section V are robust. The endogenous peer effects estimates are given in Table 10, and the exogenous ones in Table 11. The exogenous measure was not applied to the relationships model because it specifically incorporates the notion of hooking up. The measurement was obtained in the survey as follows: “If all of the members of your house were to go out to a set of registered parties, what percentage do you estimate would hook up?” To evaluate the relative quality of peer effects estimators, the dispersion factors from Tables 8-11 are collected in Table 12. Sexual Economics Hernandez USC Economics 2014 54 The best possible performance for a peer effect would be to achieve the dispersion factors in row (c). In such a case, it could be concluded that the peer effect summarizes all the attributes of a sorority which make one more or less likely to hook up. The ranges for improvement are calculated as the difference between row (a) and (c). Given this, I can express measure the improvement in dispersion for the two peer effects measures as a percent of the total possible. The last two rows in Table 11 suggest that the endogenous measure of the peer effect contributed to significantly better model fits than the exogenous one, although its coefficient will be inconsistently estimated without more advanced methods. The exogenous peer effect caused the dispersion factor in the hookup events regression to increase from the baseline, indicating it was not worth losing a degree of freedom. The endogenous peer effects estimates provided a significantly more powerful model. Compared to the sorority dummy variables, which include endogenous peer effects and all possible contextual effects, the endogenous peer effects are about 30% as effective for predicting hookups and almost 40% as effective at predicting hookup partners. Such large improvements were expected because they came from an uninstrumented endogenous model. The coefficients on endogenous peer effects are inconsistent and their explanatory power is overestimated due to correlation with the error term. Furthermore, if peer effects are merely a mechanical construction as Angrist suggested, then researchers should not be surprised to obtain big model fit improvements. The exogenous peer effects regressor (sister hookup rate) did not significantly predict hookups. The dispersion factor is larger than in the baseline model, which indicates the regressor was not worth sacrificing an additional degree of freedom. It worsens the fit so the dispersion factor is further from 1. Sister hookup rate did predict hookup partners, however. It was 5.56% as explanatory as the sorority level dummy variables, which is small but still considerable. The coefficient has a small value which bounds its effect between 0 and 1 marginal hookup partner. When a women believes only 10% of her sorority sisters would hookup as proposed in the question,. The marginal effect is exp(-0.009)(0.1) = 0.099, meaning that such a woman would have 0.1 additional partners in expectation. When the estimate changes to 90%, she has 0.89 additional partners in expectation. Sexual Economics Hernandez USC Economics 2014 55 Table 9. Sorority Dummy Variable Coefficients and Status Hookup Events Hookup Partners Status Sorority # 1 -0.61*** (0.21) 7 Sorority # 2 Sorority # 3 Sorority # 4 -0.65*** (0.20) -0.83** (0.33) 3 Sorority # 5 Sorority # 6 -0.78*** (0.202) -1.089*** (0.34) 2 Sorority # 7 -0.98*** (0.20) -1.42*** (0.34) 8 Sorority # 8 -1.163*** (0.22) -1.15*** (0.37) 9 Kendall’s tau (p-value) -0.4 (0.48) -0.33 (0.75) Sexual Economics Hernandez USC Economics 2014 56 Table 10. Endogenous Peer Effect –Sorority Level Mean Poisson Regressions Hookup Events Hookup Partners Relationships Intercept 0.056 (0.53) 1.99** (0.084) 2.959 (2.955) Drinks 0.0757 (0.012) 0.233*** (0.011) -0.116 (0.11) Relationships -0.102 (0.0789) -1.005*** (0.16) Income -0.0000278** (0.000014) -0.000008 (0.000024) 0.00008 (0.00006) Status -0.043*** (0.014) -0.02 (0.02) -0.15 (0.144) Self-Reported Attractiveness 0.052 (0.038) -0.01* (0.056) -0.3 (0.22) BMI 0.078*** (0.016) -0.043 (0.027) -0.116 (0.101) Sorority-Meaned Hookups 0.046*** (0.006) Sorority-Meaned Partners 0.16*** (0.022) Sorority-Meaned Relationships 5.78* (3.44) Residual Deviance 1487.4 561.42 53.751 Df 81 81 82 Dispersion Factor 18.36 6.93 0.66 Sexual Economics Hernandez USC Economics 2014 57 Table 11. Exogenous Peer Effect –Estimated Sister Hookup Rate Poisson Regressions Hookup Events Hookup Partners Intercept 0.51 (0.53) 2.77*** (0.82) Drinks 0.08*** (0.012) 0.225*** (0.011) Relationships -0.12 (0.08) -1.043*** (0.15) Income -0.00003** (0.000014) -0.00003 (0.000025) Status -0.04*** (0.014) -0.053 (0.018) Self-Reported Attractiveness 0.05 (0.04) -0.065 (0.055) BMI 0.068*** (0.017) -0.034 (0.027) Estimated Sister Hookup Rate 0.002 (0.0018) -0.00955*** (0.0026) Residual Deviance 1530.7 594.48 Df 81 81 Dispersion Factor 18.89 7.34 Sexual Economics Hernandez USC Economics 2014 58 Table 12. Dispersion Factors from Peer Effects Estimates Poisson Regressions Hookup Events Hookup Partners Relationships (a) Baseline (Table 7) 18.68 7.41 0.69 (b) Sorority Dummies (Table 8) 18.39 6.65 (c) Adj. Sorority Dummies – df = (77+ 6) 17.60 6.17 (d) Endogenous Peer Effect (Table 9) 18.36 6.93 0.66 (e) Exogenous Peer Effect (Table 10) 18.89 7.34 Maximum Possible Improvement (a) – (c) 1.08 1.24 Dispersion Improvement - Endogenous Effect (a) – (d) 0.32 0.48 Dispersion Improvement - Exogenous Effect (a) – (e) -0.21 0.07 Percent of Possible Improvement - Endogenous Effect 29.63% 38.70% Percent of Possible Improvement - Exogenous Effect -19.44% 5.56% Sexual Economics Hernandez USC Economics 2014 59 (VIII) Conclusion The purpose of this research has been to econometrically describe the interaction between institutional norms and supply side incentives in a competitive sexual marketplace. The hypothesis was that men and women will strive to gain access to high-status fraternities and sororities in order to increase their sexual-economic welfare. The roles of fraternities are to signal prestige and wealth, attract large numbers of women to parties, serve alcohol with price equal to zero, and provide a facility for the exercise of “home-turf advantage” (Paying for the Party, pg. 16). The role of sororities is to acquire as many sexually desirable members as possible, restrict social and sexual uses of their homes, and enforce local norms against promiscuous behavior (Paying for the Party, pg. 81). Data was obtained from women who indicated how highly ranked each fraternity was, how many times they had hooked up with their members, and how many drinks they consumed before each hookup. Similarly, 76 male respondents ranked the “topness” of the 9 Panhellenic Council sororities and this data was matched to the women’s responses about their characteristics and sexual behaviors. The demand-side hypothesis was that higher status and more drinks served would increase the number of hookups per fraternity. It was confirmed. The supply-side hypothesis was that high sorority rank will decrease the number of hookups and hookup partners per woman because a more highly “valued” woman should command a higher price per hookup. Women’s welfare has been the primary interest of sexual and feminist economics. The hypothesis that high-status sororities provide collusive power against fraternities’ sexual technologies was falsified. The evidence suggests that for two women with the same sexual desirability, if they receive the same prices (in terms of alcohol and relationships) from men throughout the course of the semester, the top house woman will hook up more. This result was consistent through each estimation but does not have theoretical ground in sexual economics as of yet. It may be intuitive to think that more desirable women would have more hookup partners. But sexual economics provides a theory for her optimization process given knowledge about her number of suitors (Baumeister and Vohs, 2004, pg. 344-245). It is in women’s interest to hook up with only one partner – the one offering the highest price. In this way, a longer queue of potential partners should only increase prices and, thus, reduce high-status women’s number of hookups, ceteris paribus. So, are the data saying that women are better off in a bottom than in a top sorority? There are a few explanatory options. First, if there is a resource which is paid to top-house women at a higher rate than to bottom-house women, then the result may be explained by an omitted variable problem. For example, the data did not contain information on how much money men spend on dinner dates with sorority and non-sorority women. If men systematically took top-house women to more expensive restaurants, or if top-house women systematically ordered items of higher values on such dates, then including a regressor for dinner-date expenditures could revere the direction of the coefficient. I explained the difficulty of unobserved prices for sexual economists in Section V. Second, the right kind of survey bias would produce this result. Imagine that top-house women really do have fewer hookups and fewer partners in the real world than bottom-house Sexual Economics Hernandez USC Economics 2014 60 women, but bottom house women are more likely to truncate their reports in surveys. If such a bias existed, bottom house women would report fewer hookups given the regressors than top-house women. Survey bias is highly relevant to the study of sexual economics, but there is no reason to believe that such a phenomenon occurred in the present research. The third explanation is to conclude that the oligopsony power of fraternities outweighs the oligopoly power of sororities. Not only the analysis results, but also the structural features of Greek life suggest this is the best explanation. While the fraternities’ advantages are structural and physical, sororities’ advantages rely heavily on social norms with indefinite enforcement mechanisms. First and foremost, fraternities have legal impunity to serve alcohol at zero price in their homes. It is a structural advantage because sorority charters ban parties in their houses. It is a physical advantage because parties attract women to enter the men’s homes and because alcohol both lowers inhibitions to sex and serves as an anticipatory excuse for women to transcend the sororities’ norms (Paul et al., 2006, pg. 151). Meanwhile, sororities’ exert their influence through peer selection, in-house policies (which, I have previously argued, creates a quasi-black-market to attract women into men’s homes for sex), and ostracism (also known as “slut shaming”). In-house policies prohibiting sexual activity are ineffective because sex from women is not a location-specific commodity. And while social norms can have significant implications for governmentality, there is no reason to believe that slut shaming is uniquely more effective at deterring hookups for sorority women, nor that shaming is more effective in top houses. Given these weaknesses, sororities are ineffective at enforcing their cartel positions. They cannot make credible threats against “defectors” who divert from the cartels “quantity target.” But failure to restrict supply does not completely explain why top-house women are exploited more than bottom house women. If sororities fail at their sexual-economic prerogative to cartelize, the outcome should be a return to the competitive equilibrium without sororities. Why are top-house women exploited at higher rates than bottom house women? This “fraternity exploitation” theory calls for an additional layer of explanation because otherwise it would characterize a new equilibrium where women prefer to join the bottom house in order to earn higher sexual profits. Why doesn’t this happen? I perceive two possible explanations: either top sororities provide non-sexual utility at the margin, or they reduce the disutility of hookups. If top sororities provide higher non-sexual utility than bottom sororities, then some women will prefer to sacrifice their sexual utility in order to acquire those benefits. Examples of non-sexual utility gained from sororities could include career networks, low housing prices, or some abstract notion of club-derived utility. Perhaps the most interesting explanation is that sorority “topness” could decrease the disutility of hooking up. Just as Armstrong et al. (2012) showed that women are more sexually satisfied in relationships, there may be some selection mechanisms that cause them to be more sexually satisfied (i.e. less sexually dissatisfied with hookups) if they are in a top house. For example, we can imagine a world where fraternity rush is efficient at sorting men by sexual competency (competent: top; incompetent: bottom). In such a world, top house men still prefer to hook up with Sexual Economics Hernandez USC Economics 2014 61 top house women but now they also give their partners higher sexual satisfaction. The result is that top house women will have more satisfying sex lives (in the physiological sense). This selection condition would cause top house women to transact more total hookups than bottom house women in equilibrium. Several non-physiological extensions of this perspective are possible. If the man’s social status is an argument in the woman’s sexual utility, then top house women will have more satisfactory sex lives because their partners tend to have high statuses (Townsend & Levy, 1990). The higher utility of sex (or, alternatively, lower disutility of hooking up) will incentivize women in top houses to hook up more than bottom house women. Even if status or sexual competency were not direct arguments in women’s sexual utility functions, they might still calculate higher net present expected values for hookups with high-status fraternity men because of their probabilistic value as a boyfriend of husband. If top sororities’ members hook up more because men target them for their top-ness, it follows to ask how sororities can reassert themselves to improve sexual outcomes for their members. If women tolerate men’s targeting do to non-sexual utility from the sorority, then corrective action is imperative. But if top-ness leads to higher sexual utility (that is, better hookups), then sororities should consider doing nothing. There are no easy answers and the hard answers that exist may be beyond the scope of sorority governance. A first category of answers is rather draconian. Sorority leaders could search for more direct mechanisms to enforce restrictions on their members’ sexual behavior. If a social planner had perfect information about sexual behavior, it could issue an efficient “tax” on quota non-compliers to increase the price of sex. Similarly, top sororities’ leaders could identify ways to enhance ostracism/slut shaming within their houses, assuming it was actually an effective hookup deterrent and had no negative externalities for women’s mental health. Second, sororities could provide subsidies or other incentives for women to increase the value of their erotic capitals. For example, if a portion of membership dues were rebated toward makeup, shoes, clothes, or hair enhancements, the price of sex with women of that sorority would rise (Hakim, 2010). Finally, the most important aspect for policy control is the sex ratio. Regnerus showed that the proportion of women to men in a society (universities included) has a direct effect on the prices of sex. In nearly all U.S. colleges, there are more female than male students. This disadvantages women’s sexual welfare because there is more supply than demand: prices will fall. But if the sex ratio in the local community of the Greek system were otherwise sororities would improve the welfare of their members by banning hook ups with the non-Greek male population. In this university under investigation, the university-wide sex ratio was 9288 women / 8712 men whereas it was 2039 Greek women /2429 Greek men. The economics of the sex ratio was also covered by Baumeister and Vohs (2004). I have given these options a purely theoretical treatment and must leave their implementation to the legitimate officials. While I have entertained strict hypotheses, my chief goal has been to show that sexual economics, as defined by Baumesiter and Vohs (2004) in terms of female resource theory, rightfully deserves empirical treatment. The institutional structure of fraternities, sororities, and universities provided a convenient format for data collection and aggregation, but the extent of this field is unlimited by the intensity of humans’ sexually competitive desires. Sexual Economics Hernandez USC Economics 2014 62 Works Cited "About NIC." National Interfraternity Council. N.p., n.d. Web. 11 May 2014. <http://www.nicindy.org/about-nic.html>. Adams, T.B., and T.E. Araas. "Purging and Alcohol-Related Effects in College women." International Journal of Eating Disorders 39 (2006): 240-44. Print. Ali, M.M., and D.S. Dwyer. "Estimating Peer Effects in Sexual Behavior among Adolescents." Journal of Adolescence 34 (2011): 183-90. Print. Allison, K.C., and C.L. Park. "A Prospective Study of Disordered Eating among Sorority and Nonsorority Women." Wiley Periodicals, Inc. (2004): n. pag. Print. Angrist, J. "The Perils of Peer Effects." National Bureau of Economic Research Working Paper (2013): n. pag. Print. Armstrong, E.A., P. England, and A.C.K. Fogarty. "Accounting for Women's Orgasm and Sexual Enjoyment in College Hookups and Relationships." American Sociological Review 77.3 (2012): 435-62. Print. Armstrong, Elizabeth, and Laura Hamilton. Paying for the Party: How College Maintains Inequality. N.p.: n.p., 2013. Print. Arndt, B. The Sex Diaries. London: Hamlyn, 2009. Print. Bandura, A. Social Learning Theory. Oxford, UK: Prentice Hall, 1977. Print. Baumeister, Roy, and Kathleen Vohs. "Sexual Economics: Sex as Female Resource for Social Exchange in Heterosexual Interactions." Personality and Social Psychology Review (2004): n. pag. Print. Baumeister, Roy, Kathleen Vohs, and K.R. Catanese. "Is there a gender difference in stength of sex drive?" Personality and Social Psychology Review 5 (2001): 242-73. Print. Sexual Economics Hernandez USC Economics 2014 63 Becker, Gary, Kevin Murphy, and Ivan Werning. "The Equilibrium Distribution of Income and the Market for Status." Journal of Political Economy 113.2 (2005): 282-310. Print. Bergmann, B. "Occupational Segregation, Wages, and Profits When Employers Discriminate by Race or Sex." Eastern Economic Journal 1 (1974): 103-10. Print. Bersamin, M.M., et al. "Young Adults and Casual Sex: The Relevance of College Drinking Setting." Journal of Sex Research (2011): n. pag. Print. Bisson, M.A., and T.R. Levine. "Negotiating a Friends with Benefits Relationships." Archive of Sexual Behavior: n. pag. Print. Blundell, R., R. Griffith, and F. Windmeijer. "Individual Effects and Dynamic Count Data." University College, London (1997): n. pag. Print. Bogle, K.A. Hooking Up: Sex, Dating, and Relationships on Campus. N.p.: New York University Press, 2008. Print. Brock, W.A., and S.N. Durlauf. "Discrete Choice with Social Interactions." Review of Economic Studies 68 (2001): 235-60. Print. - - -. "Identification of Binary Choice Models with Social Interactions." Journal of Econometrics 140 (2007): 52-75. Print. Brown, J.A., H.J. Cross, and J.M. Nelson. "Sex-Role Identity and Sex-Role Ideology in College Women with Bullemic Behavior." Journal of Eating Disorders 9 (1990): 571-75. Print. Buchana, James. "An Economic Theory of Clubs." Economica 32.125 (1965): 1-14. Print. "Budgeting FAQs." Fraternity Sample Budget. Tau Kappa Epsilon, n.d. Web. 12 May 2014. <http://www.tke.org/member_resources/finance/budgeting_basics>. Buss, D.M. "Sex Differences in Human Mate Preferences." Behavioural and Brain Sciences 12 (1989): 1- 49. Print. Sexual Economics Hernandez USC Economics 2014 64 Cameron, A. Colin, and Pravin K. Trivedi. Microeconometrics: Methods and Applications. N.p.: Cambridge, 2009. Print. Cameron, A.C., and P.K. Trivedi. "Count Panel Data." Oxford Handbook of Panel Data Econometrics (2013): n. pag. Print. Cameron, Samuel. Economics of Pornography. N.p.: Edward Elgar Publishing Limited, 2005. Print. Case, A., and L. Katz. "The Company You Keep: The Effects of Family and Neighborhood on Disadvantaged Families." NBER Working Paper no. 3705 (1991): n. pag. Print. Centers for Disease Control and Prevention. "Morbidity and Mortality Weekly Report: Youth Risk Behavior Surveillance." www.CDC.gov (2006): n. pag. Print. Christensen, R.H.B. "Regression Models for Ordinal Data: Introducing R-package ordinal." Technical University of Denmark: n. pag. Print. Cole, H.I., G.J. Mailath, and A. Postelwaite. "Social Norms, Saving Behavior, and Growth." Journal of Political Economics 100 (1992): 1092-125. Print. Cooper, M.L. "Alcohol Use and Risky Sexual Behavior Among College Students and Youth." Journal of Studies on Alcohol 14 (2002): 101-17. Print. Corbin, W.R., and K. Fromme. "Alcohol Use and Serial Monogamy as risks for Sexually Transmitted Diseases in Young Adults." Health Psychology 21 (2002): 229-36. Print. Crane, J. "the Epidemic theory of Ghettos and Neighborhood Effects on Dropping Out and Teenage Childbearing." American Journal of Sociology 96 (1991): 1226-59. Print. Crawford, Mary, and Danielle Popp. "Sexual Double Standards: A Review and Methodological Critique of Two Decades of Research." Journal of Sex Research 40.1 (2003): 13-26. Print. Fielder, Robyn L. "Predictor and Consequences of Sexual 'Hookups' Among College Students." Archive of Sexual Behavior 39 (2010): 1105-19. Print. Sexual Economics Hernandez USC Economics 2014 65 Fletcher, J.M. "Peer Influences on Adolescent Alcohol Consumption: Evidence using Instrumental Variables/Fixed Effects Approach." Journal of Population Economic 25 (2012): 1256-86. Print. Frank, R.H. Choosing the Right Pond: Human Behavior and the Quest for Status. New York: Oxford, 1985. Print. - - -. Luxury Fever. NY: The Free Press, 1999. Print. - - -. "Should Public Policy Respond to Positional Externalities?" Journal of Public Economics 92 (2008): 1777-86. Print. Glaeser, E.L., B.I. Sacerdote, and J.A. Scheinkman. "The Social Multiplier." Journal of European Economic Association 1.2-3 (2003): 345-53. Print. Glenn, N., and E. Marquardt. "Hooking Up, Hanging Out, and Hoping for Mr. Right: College Women on Dating and Mating Today." Institute for American Values (2001): n. pag. Print. Goldsmith-Pinkham, P. "Social Networking and the Identification of Peer Effects." Journal of Business & Economic Statistics 31.3 (2013): n. pag. Print. Gosling, S.D., P.J. Rentfrow, and W.B. Swan. "A Very Brief Measure of the Big-Five Personality Domains." Journal of Research in Personality 37.6 (2003): 504-28. Print. Graham, B.S., and Jinyong Hahn. "Identification and Estimation of the Linear-in-Means Model of Social Interactions." Economics Letters 88 (2005): 1-6. Print. Hakim, Catherine. Work-Lifestyle Choices in the 21st Century. N.p.: Oxford University Press, 2000. Print. Hartmann, Heidi. "Capitalism, Patriarchy, and Job Segregation by Sex." Signs 1.3 (1976): 137-69. Print. Heelas, P. The Social Construction of Emotions. N.p.: Oxford: Blackwelll, 1986. Print. Herlitz, C., and K. Ramstedt. "Asssessment of Sexual Behavior, Sexual Attitudes, and Sexual Risk in Sweden." Archives of Sexual Behavior 34 (2005): 219-29. Print. Honore, B.E. "Nonlinear Models with Panel Data." Department of Economics, Princeton: n. pag. Print. Sexual Economics Hernandez USC Economics 2014 66 Hunter, A.S. The Rules of the Affair. N.p.: London: Gibson Square, 2010. Print. Johnson, T.R. "Generalized Linear Models with Ordinally-Observed Covariates." British Journal of Mathematical Statistical Psychology 59.2 (2006): 275-300. Print. Kershaw, T.S., et al. "Misperceived Risk Among Female Adolescents: Social and Psychological Factors Associated with Sexual Risk Accuracy." Health Psychology 22 (2003): 523-32. Print. Knox, D., and C. Cooper. "Sexual Values of College Students." College Student Jounral 35 (2001): 24- 27. Print. Knox, D., and L. Sturdivant. "College Student Attitudes Toward Sexual Intimacy." College Student Journal 35 (2001): 241-43. Print. Kukuk, M. "Indirect Estimation of Linear Models with Ordinal Regressors. A Monte Carlo Study and some Empirical Illustrations." University of Tubingen: n. pag. Print. Lambert, T.A., A.S. Kahn, and K.J. Apple. "Pluralistic Ignorance and Hooking Up." Journal of Sex Research 40 (2003): 129-33. Print. Lancaster, T. "Orthogonal Parameters and Panel Data." Brown University (1997): n. pag. Print. Laumann, E.O. The Social Organization of Sexuality: Sexual Practices in the United States. N.p.: University of Chicago Press, 1994. Print. Laumann, E.O., and R.T. Michael. Sex, Love, and Health in America: Private Choices and Public Policies. N.p.: University of Chicago Press, 2001. Print. Layte, Richard, and Christopher Whelan. "Who Feels Inferior? A Test of the Status Anxiety Hypothesis of Social Inequalities in Health." GINI Research (2013): n. pag. Print. Lee, L. "Identification and Estimation of Econometric Models with Group Interactions, Contextual Factors, and Fixed Effects." Journal of Econometrics 140 (2007): 333-74. Print. Lipman-Blumen, J. Gender Roles and Power. New Jersey: Prentice Hall, 1984. Print. Sexual Economics Hernandez USC Economics 2014 67 Manski, C. "Identification of Endogenous Social Effects: The Reflection Problem." Review of Economic Studies 60.1993: 531-42. Print. Miller, K.E., et al. "Athletic Participation and Sexual Behavior in Adolescents." Journal of Health and Social Behavior 39 (1998): 108-23. Print. Mize, B., and M. Valliant. "An Assessment of Panhellenic Sorority Member Meal Consumption Patterns." Oracle: Research Journal of Fraternity/Sorority Advisors 7.2 (2012): n. pag. Print. Mongell, S., and A.E. Roth. "Sorority Rush as a Two-Sided Matching Mechanism." American Economic Review 81.3 (1991): 441. Print. Nelson, Julie A. "Feminist Economics." New Palgrave Dictionary of Economics (2008): n. pag. Print. NPCWomen.org. National Panhellenic Council, n.d. Web. 11 May 2014. <https://www.npcwomen.org/>. Olmstead, Spencer, Kay Pasley, and Frank Fincham. "Hooking Up and Penetrative Hookups: Corrleated that Differentiate Men." Archive of Sexual Behavior (2012): n. pag. Print. Ostrom, Elinor. "Understanding Institutional Diversitty." Princeton University Press (2005): n. pag. Print. Owen, J., et al. "'Hooking Up' Among College Students: Demographic and Psychosocial Correlates." Archive of Sexual Behavior 39 (2010): 653-63. Print. Paul, E.L. Beer Goggles, Catching Feelings, and the Walk of Shame: The Myths and Realities of the Hookup Experience. N.p.: Lawrence Erlbaum Associates, 2006. Print. Paul, E.L., and K.A. Hayes. "The Casualities of 'Casual Sex': A Qualitative Exploration of the Phenomenology of College Students' Hookups." Journal of Social and Personal Relationships 19 (2002): 639-61. Print. Sexual Economics Hernandez USC Economics 2014 68 Paul, E.L., B. McManus, and A. Hayes. "'Hookups': Characteristics and Correlates of College Students' Spontaneous and Anonymous Sexual Experience." Journal of Sex Research 37 (2000): 76-88. Print. Regnerus, Mark. "Mating Market Dynamics, Sex-Ratio Imbalances, and Their Consequences." Sociology 49 (2012): 500-05. Print. Saxena, S., et al. "The WHO Quality of Life Assessment Instrument." Quality of Life Research 10 (2001): 711-21. Print. Symons, D. The Evolution of Human Sexuality. N.p.: Oxford University Press, 1979. Print. Townsend, John Marshall, and Gary D. Levy. "Effects of Potential Partners' Physical Attractiveness and Socioeconomic Status on Sexuality and Partner Selection." Archives of Sexual Behavior 19.2 (1990): n. pag. Print. "University of Alabama orders sororities to end race discrimination." The Guardian. N.p., 17 Sept. 2013. Web. 12 May 2014. <http://www.theguardian.com/world/2013/sep/17/sorority- segregation-ended-university-alabama>. Vohs, Kathleen, and Jannine Lasaleta. "Heterosexual Sexual Behavior Is Governed by Social Exchange and Basic Economic Principles." Minnesota Journal of Science and Technology (2008): n. pag. Print. Wagman, B., and A. Jennings. "Household Services and Economic Growth in the United States." Feminist Economics 2 (1996): 43-66. Print. Watts, G.F., and S. Nagy. "Sociodemographic Factors, Attitudes, and Expectations Toward Adolescent Coitus." American Journal of Health Behavior 24 (2000): 309-17. Print. Webster, M., and J.E. Driskell. "Beauty as Status." American Journal of Sociology 89 (1983): 140-65. Print. Zelizer, V.A. The Purchase of Intimacy. N.p.: Princeton, 2005. Print. Sexual Economics Hernandez USC Economics 2014 69 Zetterberg, H.L. "The Secret Ranking." Journal of Marriage and Family 28 (1966): 134-42. Print. Sexual Economics Hernandez USC Economics 2014 70 Appendix A: Factor Analysis of Erotic Capital Items The theory of erotic capital is an addition to other theories of amorphous personal assets such as cultural and social capital. The purpose of the paper was to provide motivation for researchers to begin constructing measures of erotic capital because of its implications for both marriage and labor markets. For example, other authors have shown that certain industries pay an “attractiveness premium” to employees. A cornerstone of erotic capital theory is that women have relatively more erotic capital than men whereas men have relatively more economic capital. Thus, erotic capital theory also provides a solid theoretical basis for the female resource theory of human social-sexual exchange. In this analysis, I look at survey data from 344 undergraduate students, most of whom were members of a fraternity or sorority during the fall 2013 semester when the survey was deployed. I demonstrate two instances of exploratory factor analysis to illustrate the components of physical/sexual attractiveness. The collection of variables was highly motivated by Hakim's seven constituents of erotic capital: beauty, sexual attractiveness, social elements, liveliness, social presentation, sexuality, and fertility. The differentiation of each of these components is carefully shown in her paper. The data presents measured the following erotic capital attributes: 246 of the observations are from women and 98 are from men. I want to estimate a 2 factor model to show that attractiveness has different components between men and women. There are 17 variables to be included in the analysis. The scree plot suggests that a 2 factor model should be sufficient. Sexual Economics Hernandez USC Economics 2014 71 Sexual Economics Hernandez USC Economics 2014 72 The exploratory model yields some predictable results. In Factor 1, shoes, nails/spa, nutrition, teeth, and skincare, have the highest loadings. A suggested rule of thumb was to consider loadings > 0.7 as relevant. However, I generally relax these standards to the 0.50 level given the finite sample properties and sociological/qualitative nature of the estimation. For Factor 2, the loadings are categorically lower. Loadings with the highest absolute values were clothing and nails/spa but the loadings are negative. I hypothesize that Factor 1 describes erotic capital elements relevant to increasing women's attractiveness and Factor 2 describes erotic capital elements relevant to increasing men's attractiveness. While Factor 1 loadings show that clothing and shoes are some of the most important components of attractiveness for women, Factor 2 suggests that more attractive men tend to spend less on clothing and spas. Men who spend too much on clothing, for example, could be perceived as trying too hard or he may be outcompeted by more casual or athletic styles (which are presumably more inexpensive). Similarly, only a few men in the sample reported any nail/spa expenditure and these results suggest those expenditures were negatively associated with other kinds of attractiveness. The second and third biggest clues for why the 2 factors describe the erotic capital gender divide are the higher loading for nutrition and differential signs on the body-mass index loading. This shows that the components of Erotic Capital (2): Sexual Attractiveness vary between men and women. Women become more attractive with a lower bmi and they use moderate nurtition-based expenditures in order to acheive those goals. But for men, bmi has a slight association with attractiveness. Moreover, I expect that the negative effect for women would be stronger than the positive effect for men because losses are more valuable than gains by the law of diminishing marginal returns. Therefore, a woman who losses 5 bad lbs. will increase her attractiveness by relatively more than a man who gains 5 good lbs. The only problem with this interpretation is that the makeup indicator's loading is higher for the Factor 2 than Factor 1, even though all men reported 0 makeup expenditure. It does not make sense to reverse the interpretation to let women's attractiveness be indicated by Factor 2 because the clothing loading is negative. To clarify, I perform a second 2-MFA excluding the overly-unique indicators discussed below. Several variables had surprisingly high uniquenesses, which indicates that they have a high unique variance and do not have much common variance with the other indicators. The highest uniquenesses were for self-reported attractiveness, partner attractiveness range, Facebook friends, and every variable relating to personal fitness (body-mass index, fitness time, fitness expenditure, abs, and weight lifting and running (1,0)). These uniquenesses and the omissions they created in the loading results suggest a few things. First, self- reported attractiveness is not a reliable indicator of sexual attractiveness. In Table 2, the results show that individuals who spend more to augment their attractiveness do not necessarily rate themselves higher on a scale of 1-10. Similarly, the conventional attractiveness-augmenting resources are not negatively associated with individual's reported range of acceptable partner attractiveness. In figure 2, the density plots for self- reports and partner ranges are shown. Notice that although the distributions appear normal, they both have conceptual ranges of 1-10. The range of self-reports shows that no respondents rated themselves below 5 (as unattractive). We see a distribution of individuals who all consider themselves attractive, which is significantly different from the distribution of the latent variable of sexual attractiveness in the whole population. Sexual Economics Hernandez USC Economics 2014 73 As noted in Table 1, Facebook friends was included as a measure of social elements to erotic capital. The loadings and uniquness demonstrate that Facebook friends is not a covariate of “grace, charm, social skills, the ability to make people like you” (Hakim, 2010). This is not particularly surprising because there are numerous explanations for a person's social network structure. One might earn more Facebook friends if they are attractive on the whole, but, at best, the variance of Facebook friends is too high relative to the covariance of Facebook friends and the beauty/sexual attractivness indicators that compose of the analysis. If there were other Social elements indicators, the uniqueness for this variable could fall. Alternatively, if self- reported attractiveness had higher reliability and covaried with all the indicators, then the reliability for 1 or 2 social elements indicators would fall. Based on the odd makeup result and high uniquenesses, a second round of explortatory factor analysis is undertaken briefly. For the erotic capital indicators: clothing, shoes, makeup, nails, nutrition, teeth, skin, and accessories, a second scree plot is shown in Figure 3 suggesting that a 2 factor model is still appropriate. With fewer indicators a much different interpretation reveals itself. Clothing accounts for nearly all the variation in factor 1, whose other strong components are shoes and nails. Each of these produced loadings greater than 0.7 as suggested above. In factor 2, makeup and nutrition are the most important components. Nutrition and teeth are at the margin of acceptability for factor 2 but not factor 1. Sexual Economics Hernandez USC Economics 2014 74 The new loadings in Table 3 show that makeup is actually more relevant to factor 2 than to 1. Therefore, the previous interpretation of male-female erotic capital differences has to be rejected. I believe that this mode is, instead, showing differences between social presentation (Factor 1) and beauty (Factor 2) forms of erotic capital and their relative contributions to attractiveness. Hakim wrote that social presentation means “style of dress, jewelry, accessories.” Although we don't see much variation in expenditure on accessories, factor 1's components are clothing, shoes, and nails which each have high social visibility. Furthermore, “beauty” was defined as “facial attractiveness.” The loadings for Factor 2 now show that makeup, nutrition, skin care, and teeth are the most relevant components. Each is closely linked to facial attractiveness or beauty. It is only difficult to reconcile the loading for shoes within Factor 2 because it is very near the same values for teeth and skin. However, shoes have nothing to do with facial attractiveness. Shoes perform much better in the social presentation factor (> 0.7). This is expliable by imagining that there are two kinds of women: one kind that specializes in social presentation and one that specializes in facial attractiveness. The model predicts that in such as world there would still be significant variation in shoe expenditure between the subpopulations. Figure 4 displays a factor map of the final MFA model. I can show this final result is robust by using the factor scores in an OLS regression on Facebook Friends. Table 4 shows the result that social presentation (Factor 1) is a significant predictor of the individual's number of Facebook friends (95% confidence). This is consistent with expectation; a person who effectively converys high social status through clothing, shoes, and nice nails is more likely to have more people vying for their attention. They would have more Facebook friends as a result. Consisten with the commentary on uniquenesses, the factors did not predict either self-reported attractiveness nor partner attractiveness ranges. Sexual Economics Hernandez USC Economics 2014 75 Sexual Economics Hernandez USC Economics 2014 76 Appendix B. Sorority Women’s House Membership Regrets Are you in the house you wanted? If not, why? Sorority Responses: 1. I loved [the 9th ranked sorority] and [the 6th ranked sorority], I had family in those sororities 2. A top house- to have more fun 3. [the 4th ranked sorority], because many of my friends are in that house. But I am very happy with my house now. 4. [the 9th ranked sorority] because all my friends were in it, but now I'm happy with where I am 5. A better house, with a better reputation 6. Have you been through formal recruitment? I wanted to be a [the 6th ranked sorority] or [the 4th ranked sorority] because they seemed the nicest. But now I know I wouldn't ever trade my sisterhood. 7. I will not name the house. I wanted to be in that one because more of my friends are in that one and it seems like a better house. 8. [the 9th ranked sorority]Knew more people, nicer house, cheaper dues, better reputation 9. [the 7th ranked sorority]for being a top house 10. [the 7th ranked sorority], they are considered a "better" house and I am a legacy or [the 4th ranked sorority] because I liked them during recruitment 11. [the 3rd ranked sorority], felt more welcoming and sporty. 12. I wanted to be in [the 6th ranked sorority] because everyone seemed really nice and caring for each other. 13. [the 7th ranked sorority]or [the 3rd ranked sorority]- I bonded with them the most during rush week 14. I wanted to be in a higher ranked house. I didn't care which. But now I'm happy where I am, and I think I was always supposed to end up here. 15. [the 6th ranked sorority], [the 1st ranked sorority], [the 5th ranked sorority], [the 7th ranked sorority], [the 3rd ranked sorority], [the 4th ranked sorority] because they have more social events and social events with better social standing houses, also because they have better reputations on campus (appearance and personality), their sisterhood events look more in tune with my likes and activities, their homes are nicer, and they are treated better at parties based on their house reputation. 16. [the 4th ranked sorority]or [the 1st ranked sorority] - I felt a connection with them 17. [the 7th ranked sorority]or [the 6th ranked sorority] 18. [the 8th ranked sorority], [the 2nd ranked sorority], or [the 9th ranked sorority] because they have houses on the row and are more central on the Row 19. A house with a house. I joined when our house had a house and was a legitimate sorority. After losing our house we have lost legitimacy and identity has become very asian and lesbian. Two things I do not identify with and feel ostracized by. 20. didnt care but liked [the 2nd ranked sorority] 21. I dropped out of rush halfway Sexual Economics Hernandez USC Economics 2014 77 Appendix C. Survey Results - Fraternity Treasurers Fraternity Status 18 Fraternity Status 19 Fraternity Status 21 Mean Status Live-In Dues $1,750 $1,200 $1,100 $1,313 Live-Out Dues $1,875 $1,700 $1,400 $1,669 Pledge Dues $1,650 $2,054 $750 $1,614 Live-In Members 17 54 17 35.5 Live-Out Members 30 38 16 29.75 Pledge Members 10 33 2 19.5 Recruitment $8,000 $25,000 $6,000 $16,000 Social $10,000 $45,000 $17,000 $29,250 Philanthropy $2,000 $5,000 $1,600 $2,900 Officers $0 $5,000 $0 $2,500 Scholarship $0 $1,000 $0 $750 Brotherhood Events $2,000 $8,000 $3,000 $5,250 Conferences $1,500 $750 $16,000 $25,813 Other $35,000 $6,400 $20,700 National Issue 0 0 0 0 Sex Ratio (W/M) 1 3 1 2.25 Sexual Economics Hernandez USC Economics 2014 78 Appendix D. ANOVA Results Significance Levels: ***99% level, **95% level, *90% level Sex Frat number Soror number Parents Div White Class Rank Age Parents Income Range sex * * fratnum * *** *** sorornum *** * *** siblings ** ** ** ** motheduc ** *** fatheduc ** ** *** fathfrat *** *** ** * ** mothfrat * ** *** ** parents div * ** partner attract range ** * ** income * ** hookups hs *** ** ** white * *** ** classrank *** age *** * extrovert *** fb selfattract *** * * hookup expect *** relation expect *** *** *** frat parties sister hookup rate ** drink *** private alch *** * *** *** *** fitness *** bmi *** *** ** ** gpa *** *** *** spring hookups * * *** relationships ** * erotick * *** partners hookups * * * Sexual Economics Hernandez USC Economics 2014 79 Parents income range *** * * Parents education involvement ** ** *** * *** Hookup status anxiety ** ** ** Income status anxiety * ** Social status anxiety * *** College spouse Porn frequency *** * ** Masturbation Frequency *** * ** ** Birth Control Rate ** ** Sexual Economics Hernandez USC Economics 2014 80 Appendix E. Sexual Outcomes Ordered by Fraternity and Sorority Social Status Outcome (Status, Outcome) Plot Kendall’s Tau Rank Correlation Test Fraternity- Level Hookups z = 0.20 Tau = 0.03 p = 0.84 Fraternity- Level Partners z = -2.38** Tau = -0.35 p = 0.02 Sexual Economics Hernandez USC Economics 2014 81 Fraternity- Level Relationships z = 0.51 Tau = 0.08 p = 0.61 Sorority-Level Hookups z = 0.18 Tau = 0.010 p = 0.86 Sorority-Level Partners z = 0.39 Tau = 0.020 p = 0.70 Sexual Economics Hernandez USC Economics 2014 82 Sorority-Level Relationships z = 2.37** Tau = 0.14 p = 0.02 (positive z was robust to removing high- relationship respondents in sorority status 8) Sexual Economics Hernandez USC Economics 2014 83 Appendix F. Differences between Top/Bottom Greek Women and non-Greek Women Status is an endogenous variable because if an individual has attributes that make them sexually desirable, they are likely to be in a high status house. In the sorority supply equations, there is a correlation between a fixed effect (status) and individual effects. We have a related problem in fraternity equations because the fraternity-means of individual variables like income/wealth are in the error term and are certainly correlated with fraternity status. The models in Table 2 suggested that without capturing hidden sexual desirability elements, we will get an inconsistent estimate of the status coefficient. But a special type of variable, which is correlated with sororities’ status but not hookups, would be a valid and relevant instrument for the supply equation. Sorority group is conceptually identical to status rank in our data, so the selection process into sorority houses is selecting women into strict status rankings (Becker et al., 2005). If there are differences in the mean of a certain individual-level variables (for example, self-reported attractiveness) between top and bottom sororities, this lends some evidence about which criteria sororities (i.e. status levels) use to select their members. If we find one such variable is not correlated with hookups, then this is a valid and relevant instrument for sexual desirability. Previous research formalized the selection algorithm that sororities collectively employ. It optimizes matches between potential new members and sororities using a 1-3 ranking supplied by each potential new member and a 1 – m ranking submitted by each sorority, where there may be some restriction on m depending on implementation (Mongelll and Roth, 1991). Here, I investigate the assumption that top houses want to increase certain attributes of their members such that their house-mean diverges from the house-mean of the bottom house. The two-sided t-test results are given in Table A. Sexual Economics Hernandez USC Economics 2014 84 Table A. Differences between top and bottom sororities. Variable t-score (p-value) Correlation with Hookups Correlation with Hookups (excluding 0’s) Correlation Change Hookup partners t = 1.80 (0.072) 0.42 0.28 0.14 High School Hookups t = 3.68 (0.0007) 0.15 0.14 0.01 Avg. drinks per social event t = 3.8316 (0.000) 0.11 0.32 0.21 Sum of erotic capital t = 1.717 (0.087) -0.04 -0.09 -0.05 Masturbation frequency t = -2.47 (0.020) -0.01 -0.05 -0.04 Birth Control Frequency t = -3.22 (0.003) 0.05 0.13 0.08 Partner Attractiveness Range t = -1.67 (0.1035) 0.08 0.09 0.01 Rent living inside the sorority (inside – outside dues) t = -15.42 (< 2.2 x 10 -16 ) 0.004 -0.02 -0.024 Dues rate living outside sorority house t = -24.12 (< 2.2 x 10 -16 ) 0.01 -0.03 -0.04 Dues rate living inside sorority house t = -26.61 (< 2.2 x 10 -16 ) 0.01 -0.03 0.04 Avg. of Known LGBTs per sorority t = -18.70 (< 2.2 x 10 -16 ) -0.02 -0.17 0.15 Avg. of Expected Closet LGBTs per sorority t = -25.70 (< 2.2 x 10 -16 ) 0.08 0.025 0.055 Total incumbent members from previous semester t = 24.43 (< 2.2 x 10 -16 ) 0.03 0.08 -0.05 Total new members in the semester t = 36.77 (< 2.2 x 10 -16 ) 0.08 0.13 -0.05 Total sorority members t = 30.76 (< 2.2 x 10 -16 ) 0.05 0.10 -0.05 Status-Invariant Variables Previous semester hookups Relationships Hookups Social Status Anxiety Hookup Status Anxiety Income Status Anxiety White College Spouse Extroversion Intended Next Semester Hookups Parents Income Range Mother’s Sorority Membership Father’s Fraternity Membership Parents’ Education Involvement Parents’ Financial Aid Facebook Friends Fraternity Parties per Night Private Alcohol Expenditure Body Mass Index Siblings Father’s Education Mother’s Education Parents’ Divorce Self-Reported Attractiveness Student’s Income Pre-semester hookup expectations Pre-semester relationship expectations Fitness Minutes per Week Sexual Economics Hernandez USC Economics 2014 85 For variables where t-scores are greater than 0, top houses have higher group-means than the bottom house. This includes the number of hookup partners, high school hookups, drinks, erotic capital expenditure, and the two types of members. On the whole this is consistent with theory. Notice that the number of hookup partners is higher for top house women but the number of hookup events is statistically the same. Because men compete more vigorously for the high status women, there is little incentive to stay with a single partner for a large number of hookup events. There will be plenty more and one will eventually offer more valuable transfers that allow her to exit the hookup market. High status women consume more alcohol which confirms that they are likely to accept more transfers from fraternities than bottom house women. High status women have higher erotic capital, corroborating the commentary on status items from Armstrong and Hamilton (Paying for the Party, pg. 70-80). The bottom sorority was shown to have higher masturbation frequency, higher birth control use, a wider partner attractiveness range, more known and expected/closeted LGBT students, as well as lower rent and membership fees. Notice that top sororities have more members but also higher dues rates and room rental rates than bottom sororities. T-score results suggest that rent, inside and outside members’ dues, and expected house LGBTs could be effective instruments in the supply model. Each has a negative relationship with social status rank but no relationship with hookups. Therefore they are valid and relevant instrumental variables. For empirical implementation of an IV strategy, a mixture model should be used to specify the proper conditional distribution of the status variable. Status is not Poisson-distributed but may call for its own non-linear distribution to obtain the instrumented regressor values in the first-stage of 2SLS estimation. ` These data can also be used to estimate a supply equation for non-Greek women. If we assume status non-Greek women’s status is 0, then status drops out of the regression. Table B. Non-Greek Women’s Supply Regression Dependent variable: hookups Poisson Regression (1) Supply regression from non-Greek subpopulation only (2) Aggregate supply regression with non-Greek women as lowest status Intercept 0.85*** (0.03) 1.62*** (0.06) Drink (price) 0.04*** (0.0008) 0.03*** (0.001) Status -0.042*** (0.008) Table B implies that, if sorority status is not a factor for them (in column 2, I am assuming that sorority status does matter for them and they are ranked below the bottom house), non-Greek women face a lower intercept than Greek women. Greek women have an additional 2.5 hookups (exp(1.59)-exp(0.85)) on average before considering the women’s levels for the regressors. It also suggests that non-Greek women gain additional alcohol per hookup than Greek women. Effectively, the price of their hookups, in terms of alcoholic beverages, is higher. They might, for example, carry an extra aversion to hooking up with fraternity men. The data reflects that they are compensated for their aversion. With the endogeneity problem Sexual Economics Hernandez USC Economics 2014 86 unresolved, the model predicts that the non-Greek woman will hook up the least, followed up the woman in the bottom sorority, with the women in the top sorority hooking up most. If there are two supply curves, where all Greek women have a higher y-intercept for hookups (which is an additional consequence of assuming their status is 0), how are Greek and Non-Greek women differentiated in their individual characteristics. Do non-Greek women actually have higher hookup averages as predicted by the simulation? Do they tend to drink more or less? The results for t-tests of Greek versus non-Greek female means are shown in Table C. These quantitative results can confirm several of the observations made at the “Midwest University” party dorm. Consider women in the Greek system as those on the party pathway and non-Greek women as those on the social mobility or professional pathways (Paying for the Party, pg. 15-20). Women in the Greek system have significantly wealthier parents and receive significantly more financial support toward their education. Yet in ANOVA we say that parents’ incomes were not significantly different between the high and low status sororities. The Greek and non-Greek women themselves do not have different personal discretionary incomes even though survey respondents were prompted to consider their parents’ direct income transfers in their own income calculations. Because the two groups have the same incomes, it is particularly telling to show that women in the Greek system spend more of their personal income on alcohol. Either Greek women have stronger deep preferences for alcohol, which may direct them into the Greek system, or the Greek system embeds a reduced form preference for alcohol that perhaps adds to the women’s sexual desirability. Armstrong and Hamilton gave a lot of attention to the idea of status anxiety, and some conclusions about it can be drawn from these results. The only form of status anxiety that was significantly higher for Greek women was social status anxiety. The three forms of status anxiety were measured in the survey as follows (answers: Strongly Disagree Strongly Agree): Hookup Status Anxiety: Do you feel you have hooked up in order to be accepted by others? Income Status Anxiety: Indicate your level of agreement with the following statement. “Some people look down on me because of my job situation or income.” Social Status Anxiety: Indicate your level of agreement with the following statement. “Some people look down on me because of my social status or appearance.” Sexual Economics Hernandez USC Economics 2014 87 Table C. Mean Attribute Differences between Greek and Non-Greek Women Variable t-score (p-value) Parents’ Income Range t = 1.78* (0.097) Mother’s Sorority Membership t = 8.79*** (4.94x10 -16 ) Father’s Fraternity Membership t = 8.48*** (3.80.x10 -15 ) Parents’ Education Involvement t = -3.67*** (0.002) Parents’ Financial Aid t = 1.73* (0.09) Facebook Friends t = 1.73* (0.10) Social Status Anxiety t = 3.03*** (0.01) Fraternity Parties Per Night t = 1.966* (0.075) Private Alcohol Expenditure t > 1.62* (0.062) College Spouse t = -2.8*** (0.01) Body Mass Index t = -3.43*** (0.004) Relationships t = 2.90*** (0.007) Event Drinking Average Siblings Father’s Education Mother’s education Parents’ divorce Self-Reported Attractiveness Personal Income Hookup Status Anxiety Income Status Anxiety Birth Control Rate Previous Semester Hookups High School Hookups Hookup Expectation Relationship Expectation Fitness Minutes Per Week Partner Attractiveness Range Hookups Partners GPA Sexual Economics Hernandez USC Economics 2014 88 The fact that Greek women do not worry more about using hookups for social acceptance follows because hookup levels are not different between Greek and non-Greek women. If sorority women did have higher hookup status anxiety, this would have given evidence that sororities use status anxiety to maintain their restrictions on the sex market. Notice the minor difference between Income and Social status anxiety measurements. Recall that erotic capital theory predicts that women have higher erotic than economic capital stocks (Hakim, 2010). From this, one can easily understand why higher sexual-economics status might facilitate social rather than income status anxiety. Women’s social status and especially their appearance determine their sexual economic outcomes, not their income or job prospects (Townsend & Levy, 1990; Hakim, 2010; Baumeister and Vohs, 2004). We see the sexual economics corollary is reflected in this result: women with higher social status report higher social status anxiety. However, none of the status anxiety measurements were different between top and bottom sororities (Table A). The mechanisms maintaining and restraining status anxiety within the Greek social hierarchy might inspire some future quantitative research. Sexual Economics Hernandez USC Economics 2014 89 Appendix G: Differences between Top and Bottom Fraternities. Although the predicted coefficients were was obtained for fraternities, it may still be interesting to investigate the differences between top and bottom fraternities. The results of calculations are shown in the table. There were 4 top and 5 bottom fraternities. Due to the lack of individual-level fraternity data, only the institutional variables like rent, dues, and members were tested. Both the Table and Figure suggest that top fraternities incurred sexual aggression reports from women much more frequently than bottom fraternities. The data supports two possible explanations: either top fraternity men are actually more aggressive, competitive or threatening, or women attend their parties more frequency to make the observations. Table 5. Differences between top and bottom fraternities. Variable t-score (p-value) Correlation with Hookups Partners t = 10.35*** (1.60x10 -10 ) 0.26 Total incumbent members from previous semester t = 11.36*** (1.13x10 -10 ) 0.312 Total new members in the semester t = 12.19*** (2.92x10 -11 ) 0.10 USC Owned t = 3.13*** (0.005) 0.52 Survey Participation t = -26.97*** (<2.2x10 -16 ) -0.15 Relationships t =2.06** (0.05) -0.18 Hookups t = 4.12*** (0.000) -- Creepiness: Left fraternity t =4.67*** (0.000) -0.10 Creepiness: Drink Contamination t = 7.23*** (2.35x10 -7 ) -0.19 Creepiness: Witness Threat t = 5.68*** (8.87x10 -6 ) -0.04 Creepiness: Average t =5.62*** (1.02x10 -5 ) -0.12 Status-Invariant Variables Creepiness: Preempt Threat Sexual Economics Hernandez USC Economics 2014 90 Figure. Sexual aggression/creepiness – Relationships with Status and Hookups
Abstract (if available)
Abstract
Sexual economics can be defined as the study of the time, effort, and resources that humans expend to engage in sexual activity. While psychology and popular culture ascribe sexual motivations to broad categories of human action, economics has been slow to provide quantitative models of sexual decision‐making. In this analysis, the principle of social exchange theory is adapted to explain fraternity and sorority members' incentives to engage in casual sexual relationships. The scope of sexual economics is thereby elaborated to feature institutional factors, which, due to supply and demand dynamics, are hypothesized to have opposite effects for men (in fraternities) and women (in sororities). Social exchange predicts that men in high status fraternities should receive casual sex at a lower price whereas women in high status sororities should receive higher prices for casual sex. Original data was collected from 477 individuals across 35 fraternity and sorority houses at a single university. They are analyzed using ANOVA and Poisson regression to achieve a consistent estimator for the ""price of sex"" in terms of relationships and fraternities' alcohol. Finally, the data are used to demonstrate the current controversies of peer effects econometrics. Contrary to the seminal sexual economics theory, it was found that women in the high status sororities receive lower sex prices than others. Conclusions pertaining to sorority governance are offered as concluding remarks.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Pricing strategy in a duopolistic market: aviation industry in Middle East
PDF
Three essays on cooperation, social interactions, and religion
PDF
Essays on health economics
PDF
Comparative study on Asian financial capitals' competitiveness: focused on strengths and weaknesss of the city of Seoul
PDF
What “works” and why it still isn't enough: sexual harassment training measures of effectiveness
PDF
Two essays in econometrics: large N T properties of IV, GMM, MLE and least square model selection/averaging
PDF
Economic policy-making in a changing environment: the politics of success in Peru vis-à-vis China
PDF
The impact of agglomeration policy on CO₂ emissions: an empirical study using China’s manufacturing data
PDF
Explaining well-being: essays on the socio-economic factors accompanying life satisfaction
PDF
The Catholic church in Latin America: an evaluation of the institutional and political impacts of progressive church reforms
PDF
Essays on development and health economics
PDF
The influence of a humorous intervention on parent-child sexual communication
PDF
Three essays on social policy: institutional development, and subjective well-being as a cause and consequence of labor market outcomes
PDF
Three essays on health & aging
PDF
Essays in political economy and mechanism design
PDF
Essays in health economics and provider behavior
PDF
Powerful guts: how power limits the role of disgust in moral judgment
PDF
A social accounting matrix multiplier analysis of a water supply shortage in the Los Angeles County water service area SEZ 386
PDF
Essays on economics of education and private tutoring
PDF
Essays in labor economics: demographic determinants of labor supply
Asset Metadata
Creator
Hernandez, Sean
(author)
Core Title
Sexual economics: an econometric study of a university Greek system
School
College of Letters, Arts and Sciences
Degree
Master of Arts
Degree Program
Economics
Publication Date
07/18/2014
Defense Date
07/18/2014
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
fraternity,Greek,institutional economics,OAI-PMH Harvest,sexual economics,social status,sorority
Format
application/pdf
(imt)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Kocer, Yilmaz (
committee chair
), Huebner, Karin (
committee member
), Majchrzak, Ann (
committee member
), Nugent, Jeffrey B. (
committee member
)
Creator Email
sean.j.hernandez@gmail.com
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c3-445745
Unique identifier
UC11286910
Identifier
etd-HernandezS-2722.pdf (filename),usctheses-c3-445745 (legacy record id)
Legacy Identifier
etd-HernandezS-2722.pdf
Dmrecord
445745
Document Type
Thesis
Format
application/pdf (imt)
Rights
Hernandez, Sean
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Access Conditions
The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA
Tags
fraternity
Greek
institutional economics
sexual economics
social status