xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam Ka
controlfield tag 001 001966674
007 cr mnu|||uuuuu
008 081023s2007 flu sbm 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0002244
The influence of masculinity ideology on high-risk sexual behavior among men who have sex with men
h [electronic resource] /
by Christopher Wheldon.
[Tampa, Fla.] :
b University of South Florida,
ABSTRACT: Epidemiologic evidence shows that the incidence of HIV among men who have sex with men (MSM) is on the rise. high-risk sexual behaviors are the primary mode of transmission and are responsible for increased rates of infection. a growing body of evidence suggests that endorsement of normative beliefs and values regarding manhood and masculinity, or masculinity ideology, may help to explain certain high-risk sexual behaviors. this study investigated the associations between a measure of masculinity ideology and sexual risk taking among a sample of MSM. in addition, indirect pathways in which masculinity ideology may lead to sexual risk taking were explored, including the use of the internet to meet sex partners and the use of drugs during sexual encounters. a cross-sectional, web-based survey was used to collect data from participants recruited via the internet (n=907) Correlation and multiple regression analysis supported the hypotheses that masculinity ideology was positively associated with the number of sexual partners and frequency of unprotected anal intercourse in the previous six months. Higher endorsement of masculinity ideology was also positively associated with greater intentions for unprotected anal sex. This study adds to the growing body of literature on the subject of gender ideologies and sexual behavior and offers additional avenues for public health research.
Thesis (M.S.P.H.)--University of South Florida, 2007.
Includes bibliographical references.
Text (Electronic thesis) in PDF format.
System requirements: World Wide Web browser and PDF reader.
Mode of access: World Wide Web.
Title from PDF of title page.
Document formatted into pages; contains 82 pages.
Adviser: Jeannine Coreil, Ph.D.
x Public Health
t USF Electronic Theses and Dissertations.
The Influence of Masculinity Ideology on High Risk Sexual Behavior among Men who Have Sex with Men by Christopher Wheldon A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science in Public Health De partment of Community and Family Health College of Public Health University of South Florida Major Professor: Jeannine Coreil, Ph.D. Julie Baldwin, Ph.D. Joseph A. Vandello, Ph.D. Date of Approval: October 31, 2007 Keywords: Masculinity, Sexual Behavior, HIV/AIDS, MSM, Barebacking Copyright 2007, Christopher Wheldon
Dedication For JP and all the men who graciously volunteered their time.
Acknowledgments I owe a debt of gratitude to Dr. Jeannine Coreil for her guida nce, time and patience in serving as my major professor. Her enthusiasm and motivation for this research was greatly appreciated. I would also like to thank my committee, Drs. Julie Baldwin and Joe Vandello, for their advice and support.
i Table of Con tents List of Tables ................................ ................................ ................................ .................. ii List of Figures ................................ ................................ ................................ ................ iii Abstract ................................ ................................ ................................ .......................... iv Chapter One: Introduction ................................ ................................ ............................... 1 Chapter Two: Background ................................ ................................ ............................... 6 Barebacking ................................ ................................ ................................ ..... 6 Highly Active Anti Retroviral Therapies (HAART) ................................ ............. 8 Substance Use ................................ ................................ ................................ ...... 9 The Internet ................................ ................................ ................................ ....... 14 Theoretical Framework ................................ ................................ ...................... 16 The Social C onstruction of Gender ................................ ......................... 18 Theories of Masculinities ................................ ................................ ....... 20 Masculinity and Health ................................ ................................ ...................... 23 Chapter Three: Method ................................ ................................ ................................ .. 27 Study Design and Sample ................................ ................................ .................. 27 Objectives ................................ ................................ ................................ .......... 27 Internet based Research ................................ ................................ ..................... 29 Participants and Procedures ................................ ................................ ............... 33 Measurement ................................ ................................ ................................ ..... 36 Data Analysis ................................ ................................ ................................ .... 42 Chapter F our: Results ................................ ................................ ................................ ... 45 Descriptive Analysis ................................ ................................ .......................... 45 Correlation Analysis ................................ ................................ .......................... 49 Regression Analysis ................................ ................................ ........................... 51 Chapter Five: Discussion ................................ ................................ ............................... 60 References ................................ ................................ ................................ ..................... 69 Footnotes ................................ ................................ ................................ ....................... 82
ii List of Tables Table 1. Summary of Variables ................................ ................................ ..................... 42 Table 2. Demographic Characteristics of the Sample ................................ ..................... 46 Table 3. Descriptive Statistics for High risk Sexual Behaviors ................................ ...... 47 Table 4. Correlation Matrix for Masculinity Ideology and Outcome Measures ............... 51 Table 5. Regression analysis for the prediction of number of anal sex partners .............. 52 Table 6. Regression anal ysis for the prediction of frequency of unprotected sex ............ 55 Table 7. Regression analysis for the prediction of barebacking intentions ...................... 58 Table 8. Regression analysis for full model with drug and Internet variables ................. 59
iii List of Figures Figure 1. Concep tual Model ................................ ................................ .......................... 29
iv The Influence of Masculinity Ideology on High Risk Sexual Behavior among Men who Have Sex with Men Christopher Wheldon ABSTRACT Epidemiologic evidence s hows that the incidence of HIV among men who have sex with men (M SM ) is on the rise. High risk sexual behaviors are the primary mode of transmission and are responsible for increased rates of infection. A growing body of evidence suggests that normative beliefs and values regarding masculinity, or masculinity ideology, may help to explain the variability in certain high risk sexual behaviors. This study investigated the associations bet ween a measure of masculinity ideology and sexual risk taking among a sample of MSM. In addition, behavioral mechanisms in which masculinity ideology may le ad to sexual risk taking were explored including the use of the Internet to meet sex partners and the use of drugs during sexual encounters. A cross sectional, web based survey was used to collect data from participants recruited via the Internet (n=907) Correlation and multiple regression analyses supported the hypotheses that masculinity ideology is positively associated with number of sexual partners and frequency of unprotected anal intercourse in the previous six months. Higher endorsement of mascul inity ideology was also positively associated with greater intentions for unprotected anal sex. This study adds to the growing body of literature on the subject of gender ideologies and sexual behavior and offers additional avenues for public health resea rch.
1 CHAPTER 1 INTRODUCTION Men who have sex with men (MSM) 1 continue to be a vulnerable population at risk for contracting HIV/AIDS. Despite the considerable efforts directed toward the primary prevent ion of HIV/AIDS there is growing evidence to sug gest increasing infection rates among subpopulations of MSM, causing some public health experts to speculate a possible resurgence of the epidemic (Wolitski et al., 2001). In general, epidemiologic surveillance of MSM is fragmented and affords limited gen eralizability; however, improved data collection and reporting from selected states provide strong evidence confirming increased rates of infection. Data from 29 states showed a 17% increase in new infections among MSM during the three year time span betwe en 1999 and 2002 (CDC, 2003). The number of HIV/AIDS diagnoses increased 11% among MSM between 2001 and 2005 (CDC, 2005 ). In 2 005, MSM accounted for two thirds of all new HIV i nfections among male adults and adolescents in the 33 states with long term con fidential name based reporting (CDC, 2007 ). In this same year, homo sexual contact was the source of infection for the overwhelming majority of new HIV/AIDS diagnoses among men (67%), followed by heterosexual contact (15%) and injection drug use (13%) (CDC, 2007). Increasing HIV infection rates point s to changing patter n s of sexual behavior among some MSM, suggesting a trend toward increased risk taking. Behavioral research
2 confirms the relatively high prevalence of risky sexual practices, suc h as having an al sex with a high number of partners and not u sing condoms during intercourse; however, s ampling and measurement issues preclude definitive generalization regarding the prevalence of certain practices, like unprotected anal intercourse (UAI). Social preju dice and marginalization make probability based sampling difficult with hidden populations like MSM. Furthermore, discrepancies in the measurements of UAI make comparisons across studies problematic. For example, some studies do not differentiate between U AI that occurs within the context of a monogamous or committed relationship versus that with anonymous or casual partners. Likewise, t he use of a generic measurement for anal sex is problematic because it ignores th e significantly greater risk for the rece ptive partner and it prohibits more detailed exploration of insertive and receptive anal sex as distinct behaviors with unique antecedents. Given the se limitations, some estimate the prevalence of unpro te cted anal sexual practices to range between 14% and 45% ( Guenther Grey et al., 2005; Webster et al., 2003; Kalichman, Nachimson, Cherry, & Williams, 1998; Mansergh et al., 2002; Strathdee et al., 2000 ) One populatio n based study of MSM reported a 45% prevalence rate of UAI w ithin the previous 12 months amo ng 18 to 29 year olds, 31% occurring with non primary partners ( Webster et al., 2003 ) Some studies have demonstrated marked differences in prevalence of UAI among demographic groups. Bradford et al. studied demographic and behavioral characteristics assoc iated with actual rates of HIV seroincidence and found that older men (mean age = 38.6) with a large amount of sexual partners were the most likely to report UAI and, subsequently, had the highest rate of seroconversion (2006).
3 Parallel increases in sexu ally transmitted infections ( STIs ) with similar infection patterns confirm the prevalence of these behaviors. For example, the incidence of gonorrhea and syphilis suggest s significant increases in the high risk sexual practices that lead to HIV transmissi on. In 2003, the Gonococcal Isolate Surve illance Project reported a 15.6% increase in positive test results for gonorrhea since 1988 among MSM (CDC, 2003). Similarly, i ncreases in syphilis rates have also been found among MSM in several urban areas ( Stolte & Coutinho, 2002 ) Collectively this body of research yields convincing evidence for a renewed focus of public health research and action directed toward the primary prevention of HIV/AIDS among MSM. The focus of such activities should be on understandin g and responding to the social, cultural, and psychological determinants of high risk sexual behaviors. In novative, theoretically driven research is needed in order to understand and formulate adequate public health responses. Reformulating the problem of HIV/AIDS among MSM within a social ecological framework allows for the identification of new variables that may yield promising finding s for future resear ch and program planning. I ncorporating theoretical developments from the social and behavioral science s can help to bridge the gap between seemingly disconnected risk factors. The result of such efforts can illuminate the complex relationships that exist between individuals and their social environment and how together, through interacting systems, those relationships produce varying patterns of risk behaviors throughout a population. Of particular interest to this research project are the ways in which broad social categories such as gender interact at the macro and micro levels to create dynamic
4 trends in sexual risk taking. Utilizing theories of gender and sexuality provides the theoretical basis on which gender is understood as a social category and not solely a proxy for biological sex. This approach deviates considerably from a categorical treatment of gender common to behavioral epidemiologic research. It also fits within a paradigm shift occurring in the emerging field of men's health that calls for a "gender relations" approach that is "informed by a social approach of the kind that presently under pins the new public health," emphasizing the role played by social determinants of health and illness (Schofield, Connell, Walker, Wood, & Butland, 2000). T his research s ought to understand gender as a complex, multidimensional, sociocultural construct tha t structures sexual interactions and is therefore an important component of the social ecology of HIV/AIDS among MSM. The broad objectives of the present work include d the application of social constructionist and feminist theories of gender, sexuality and power to a contemporary public health issue HIV/AIDS risk behaviors among MSM. This approach deviates from similar attempts to integrate theories of gender and power with in public health research in that the focus was not on the deleterious effects on women but on a marginalized group of men In this sense, this work offers a valuable contribution to the growing body of literature in men's health studies. The specific aims of this research were to examine associations between cultural norms and belief s surrounding men and masculinity (masculinity ideology) and sexual risk taking among MSM. Masculinity is understood as a set of interconnected cultural beliefs and attitudes regarding the nature and appropriate behavior for the male gender.
5 Masculinity id eology is a measure of the degree to which an individual endorses or internalizes these cultural standards of male behavior. In addition, some possible pathways in which masculinity ideology may lead to sexual risk taking w ere considered, including the use of the Internet to meet sex partners and the use of drugs in conjunction with sexual encounters. These factors are important because there is evidence that they individually contribute to sexual risk taking among MSM. They can also be understood as gender ed health behaviors, structured by cultural n orms and expectations for men. Specifying behavioral mechanisms that mediate the associations between masculinity and sexual risk taking enables this research to elaborate on these relationships and theoretical underpinnings. Sexualized drug use, or drug use immediately before or during sex, and meeting sex partners online were empirically explored as mediating factors that both encourage and enable the social construction of masculinity. In sum, the two main o bjectives for this research endeavor were to describe the sexual behaviors of a sample of MSM recruited through the Internet and to test a conceptual model based on cultural norms of masculinity as an explanatory factor in risk taking behaviors related to HIV transmission.
6 CHAPTER 2 BACKGROUND Several reasons for the increase in sexual risk taking among MSM have been proposed. Among the more recent and prolific explanations found in the literature include changes in social norms reflected in the "bareba cking" phenomena the development and availability of highly active antiretroviral therapy (HAART), substance use, and the emergence of the Internet as a means to solicit sexual partners Research exploring the ability of these factors to explain the preva lence of sexual risk taking among MSM has garnered some empirical support; however, there are important theoretical and methodological limitations that should be considered. Furthermore, this line of research limits the analysis of sexual behavior to the i ndividual level, failing to connect individual beliefs, attitudes and behaviors to overarching social institutions and cultural norms. In this chapter, the theoretical framework for such an approach is addressed, but first a brief overview of the current l iterature is warranted. Barebacking Typical public health campaigns seeking to prevent HIV infection among MSM focus primarily on increasing knowledge self efficacy, and access to condoms during sexual encounters. Largely informed by individual focused psychological models of behavior change, these campaigns are based on the assumptions that high risk sexual behavior results from uninformed, irrational or impaired decision making strategies. Recent research, however, suggests that many MSM are making ra tional, complex decisions regarding their sexual health that include risk taking behaviors (Suarez and
7 Miller, 2001). Rather than resulting from episodic lapses in otherwise consistent condom use, some MSM are intentionally forgoing the use of condoms duri ng anal sex. Barebacking is the popular term used to describe unprotected anal intercourse (UAI). The initial use of this term within the context of sex between men has been traced back to an article written by Stephen Gendin en titled, "Riding Bareback" (H alkitis, Wilton, & Drescher, 2005). The article published in 1997 in POZ Magazine (a magazine and website targeted to HIV positive men), describes the author's personal desires and the desires of other HIV positive men to abandon the use of condoms. It a ppears initially that barebacking was used to describe condom less anal sex between HIV positive men only; however, the term has since evolved to include a variety of unprotected sexual acts between men, and less commonly between men and women. It has also been established that a large majority of gay and bisexually identified men are familiar with the term (Mansergh et al., 2002 & Halkitis, Parsons, Wilton, 2003); however, the meanings they ascribe to it are varied. Halkitis, Wilton, & Galatowitsch found s ignificant differences in the ways in which HIV + and HIV men defined barebacking (2005). HIV me n were more likely to define barebacking as unprotected anal intercourse (UAI) that typically occurs between HIV negative men, whereas HIV positive men underst ood barebacking to include HIV negative and positive men (Halkitis, Wilton, & Galatowitsch, 2005). In addition, some men understand the term to connote the intentionality of UAI and even the positions of the partners (either insertive or receptive). Esti mates of individuals reporting barebacking, when the term barebacking was undefined by the survey instrument range as high as 45% of the sample (Halkitis,
8 Parsons, Wilton, 2003); however, in studies where barebacking was overtly defined as intentional UAI with a non primary partner, the prevalence was 10% (Mansergh et al, 2002). Stressing the intentionality of unprotected sex in the definition of barebacking is of more importance to public health research because it represents more than condom less sex. I n this context, barebacking becomes a widely recognized rejection of institutionalized safe r sex messages. The reification of barebacking as sociocultural practice is exemplified by recent studies that show barebacking has evolved into an identity among su bgroups of MSM (Grov et. Al, 2007; Parsons, Bimbi, 2006). In their study of single or nonmonagamous MSM, 12% of men identified themselves as a barebacker. Self identified b arebackers in this study were more likely to be HIV positive, more likely to use d rugs during sexual encounters, and reported more unprotected anal sex compared to non identified Barebackers. It is clear from the research described here that barebacking is a current and ev olving sociocultural phenomena that warrants separate considerati on as a unique contributor to the observed increases in sexual risk taking. Highly Active Anti retroviral Therapy (HAART) The advent and availability of HAART may affect sexual risk taking behavior in two distinct ways: by changing perceived risk or susc eptibility of infection due to decreased viral loads ; and by decreasing perceptions regarding the severity of HIV infection (Huebner and Gerend, 2001). HAAR T ha s been shown to be an effective treatment leading to decreased viral loads in many HIV positive men. This leads some to believe that UAI with an HIV positive partner who is undergoing treatment is less of a
9 risk. Still others believe that HIV is no longer the serious disease it once was, in effect likening it to a chronic condition manageable with m edication. Perceptions pertaining to susceptibility of infection as well as the severity of the disease have been found to be associated with sexual risk taking (Kalichman, Nachimson, Cherry, & Williams, 1998; van der Snoek, de Wit, Mulder, & van der Meijd en, 2005) ; however these studies employed a cross sectional design leaving the temporal direction of the relationships ambiguous. Findings from a recent prospective cohort study contests the inferred causative direction proposed by previous studies. Simil ar to these studies, the authors reported a positive correlation between treatment optimism and UAI with non primary partners but the temporal relationship of this association appeared to be contrary to expected findings (Huebner, Rebchook, & Kegeles, 200 4) Treatment optimism resulting from the availability of HAART did not predict UAI but rather those men who were already engaging in high risk sexual behavior s were more likely to reference the availability of HAART as a reason for their risk taking. In other words, the availability and effectiveness of HIV/AIDS treatments became more salient to those men engaging in sexual practices that put them at risk for infection. Focusing on the decreased transmission risk and disease severity thus becomes a ration alization used to decrease cognitive dissonance associated with high risk behavior in the context of HIV/AIDS. Substance Use The use of alcohol and illicit drugs is another important factor to consider when exploring possible explanations for high risk s exual behavior among MSM. Historically, people who seek out same gender sexual relationships have been forced underground due
10 to the social prejudice surrounding the expression of homosexuality As a consequence, social organization for gays and lesbians h as more or less centered on bars and dance clubs where participants go to freely express their sexuality and create social relationships unfettered by heteronormative pressures. The subcultures formed out of these social processes exist outside of larger social mores and institutions that serve to constrain sexual behavior for many heterosexual adults. As a result, the norms governing many networks of MSM are nested in sexual institutions that provide the context in which HIV risk behaviors are cultivated Green explains: Blocked from the institutionalized rites of passage that define and organize (heterosexual) adulthood, some gay men find that commercial sexual institutions have a powerful and enduring presence in their lives well past the age when th eir heterosexual counterparts move out of nightlife and into sexual trajectories tied to marriage and family. Indeed, for this latter group of men, commercial sexual institutions [bars, dance clubs, sex parties, bathhouses, etc] come to replace marri age and family as a primary vehicle for anchoring social and sexual life. (2003) This is the context in which high risk sexual behavior among many MSM must be understood. It is through interaction with dominant cultural norms regarding sexuality that such marginalized, alternative social institu tions are created. These institutions have direct effects on the formation of social and sexual interactions among MSM that include substance use and HIV risk. This is particularly true for young men and adolescent s whose primary soci alization into gay community centers around commercial sexual
11 institutions ; however, research highlights the prevalence of alcohol and drug use among MSM of all ages Data from a probability sample of urban dwelling MSM found a high pre valence of alcohol and illegal drug use (Stall, Paul, Greenwood, et al., 2001). While the rates of alcohol use were roughly comparable to those of males in the general population, the prevalence of illegal drug use is 10 times greater in urban dwelling MSM as compared to all males 12 years of age or older (Stall, Paul, Greenwood, et al., 2001;University of California San Francisco, 2006). Other studies have found an association be tween increase s in the use of "club drugs 2 ," such as Ecstasy, Ketamine, GH B, a nd Methamphetamine, and high risk sexual behaviors (CDC Fact Sheet, 2006). McKirnan et al. found that those who frequently combined drugs with sex were more likely to report higher rates of UAI with non primary partners as well as higher rates of Hepatitis B infection (2001). Methamphetamine (also known as meth, crystal, ice, Tina, crank, speed), in particular, has received considerable media attention for its popularity among MSM and its use in the gay circuit club scenes as well for its use during sexual activity. Research indicates that MSM who use methamphetamines are 2 to 3 times more likely to engage in high risk sexual behavior (University of California San Francisco, 2006; Hirshfield, Remien, Walavalkar and Chiasson, 2004; Mansergh, Shouse, Marks, e t al., 2006). In one study, methamphetamine users were more than twice as likely to en gage in receptive but not insertive anal intercourse (Mansergh, Shouse, Marks, et al., 2006). The differential rates of UAI for receptive and insertive partners reflect i mportant physiological e ffects of
12 methamphetamines in particular the drug increase s sex drive but decrease s the ability to get and maintain erection s The physiological suppression of erections explains the higher proportion of men using methamphetamines during receptive sex as well as the frequency with which methamphetamines are combined with erection enhancing drugs such as sildenafil (Viagra). In fact, the same study found a significant association between the use of sildenafil and unprotected insertiv e anal sex (Mansergh, Shouse, Marks, et al., 2006). These studies highlight the relevance of drug use as a contextual factor important to social ecological models of HIV/AIDS risk; however, these studies fall short of providing a causative explanation for high risk sexual behaviors. In other words, using methamphetamines and other drugs appears to coincide with certain sexual practices, like UAI, but may not be a fundamental cause of them. The causal relationship between substance use and high risk sexual b ehavior is equivocal The temporal relationship is confounded by a lack of experimental and longitudinal studies. Furthermore, research regarding individual expectations for the effects of drugs and alcohol add complexity to these relationships. Motivated by the need to escape "self awareness of personal vulnerability to HIV," individuals may turn to drugs or alcohol to "shift self awareness from the abstract or long term implications of [their] behavior to immediate, here and now' sensations or actions" ( McKirnan, Vanable, Ostrow & Hope, 2001). Individual expectations for this kind of temporary escape have been found to be important considerations explicating the relationship between recreational substance use and high risk sexual behaviors. One study foun d that expectancies of sexual "escape" directly modified the association between substance use
13 during sex and UAI (McKirnan, Vanable, Ostrow & Hope, 2001). The use of substances in this study represented an active strategy to decrease cognitive dissonance resulting from risky sexual behavior. The authors noted that neither "habitual use nor exposure to substances induced risks." Rather, risks were pronounced among men who reported strong expectancies for "cognitive escape" through the use of substances (McK irnan, 2001) Therefore, drug use and high risk sexual behavior appear to have common antecedents but may not constitute the simple linear relationship typically proposed. A qualitative study that explored the contextual antecedents of methamphetamine us e among gay an d bisexual men in Manhattan illuminates the statistical associations reported in the literature. This study found that apart from individual level factors such as self esteem and desire to form social connections, methamphetamine use repre se nted an attempt to "negotiate sexual sociality and increase sexual pleasure" (Green and Halkitis, 2006). These findings are mirrored by another qualitative study that found empirical support for two dependent but distinct motivations for "club drug" use am ong gay and bisexual men: Drugs for sexual performance and drugs for community (Green, 2003). Drugs for sexual performance represent strategies to "negotiate tensions that arise from sexualized interactional patterns within urban gay communities" (Gree n, 2003). Men reported drug use in this context in order to take advantage of the psychosocial and physiological effects of the drugs, such as an increase sexual desire, longevity and self confidence as well as to provide an anesthetic agent for sexual e ncounters that "push the body to new limits" (Green, 2003). Men also reported using drugs in order to "experience membership in large social venues among participants who
14 might otherwise have little in common, or hold little knowledge of each other (Green 2003). In this case, drug use represent ed an active strategy to create temporary communities where users experience decreased social inhibitions, decreased self consciousness, and increased sense of intimacy and solidarity (Green, 2003). The Internet Recent research has focused on the Internet as a medium facilitating risk behavior for HIV/AIDS transmission. One frequently referenced outbreak investigation implicated a particular cha t room as the common factor in four out of six cases of syphilis amon g MSM in San Francisco (Klausner, Wolf, Fischer Ponce et al, 2000). Several studies have declared the importance of the Internet in providing a social space for men to meet other men who are seeking sexual encounters (Bull & McFarlane, 2000; Kim, Kent, McF arland, & Klausner, 2001; McFarlane, Bull, & Rietmeijer, 2000) In this sense, the Internet plays a role similar to bathhouses and other venues utilized by men to arrange anonymous sexual encounters ; however, the Internet is unique in that it provides a "v irtual" meeting place for men to efficiently seek sexual encounters ou tside the confines of physical venues like bathhouses, which are to some extent identifiable and controlled environments. Sexual encounters first initiated through Internet communication may occur in private residences or other agreed upon public venues. The Internet also expands the ease and availability of meeting casual sex partners for men in rural areas or places without established gay communities. This enables some men to participa te in certain sexual practices they may have otherwise not had the opportunity because of issues of anonymity or lack of information regarding active sexual venues.
15 Furthermore, the Internet as a social environment structures sexual interactions in ways th at may affect the spread of HIV. For instance, many Internet sites used by men to meet sexual partners enable them to communicate facts about themselves through web based profiles, resulting in a type of catalog of potential sexual partners that can be sea rched for a variety of characteristics, including HIV status. The availability of such information may be considered a positive aspect of the Internet for it allows for a quick and efficient way to screen out serodiscordant partners; however, evidence sugg ests th at many of the men who may describe themselves as HIV negative may not be aware of their actual HIV/ADIS status due to lack of sufficient testing or may intentionally misrepresent themselves Furthermore, the availability of this information removes the need for sexual partners to discuss HIV status and may provide a false sense of safety. Another way in which the Internet may facilitate HIV transmission is by bridging previously disconnected social networks of individuals most likely to engage in high risk sexual activities. Boily, Hogben and Bastos have argued that non volitional changes in sexual risk taking occurred after the initial AIDS epidemic due to population level changes in the "pool of high risk taking individuals" that resulted from high mortality rates (2005). Today those individuals would have the opportunity to remain sexually active due to trea tment advances (such as HAART) increasing the pool of "high risk taking" individuals. The Internet could exacerbate this process by providi ng a medium in which high risk individuals are able to locate one another By bridging these types of networks, the Internet can significantly impact the spread of HIV. These examples highlight the ways in which the Internet may facilitate high risk sexual behavior and
16 consequently the spread of HIV, but once again, they do not suggest that the Internet is a cause of this behavior. Missing from this explanation and the other explanatory factors previously described is a focus on sexuality as a socially embe dded practice. THEORETICAL FRAMEWORK Fundamental to the success of HIV prevention is a comprehensive understanding of human sexual behavior. The ability of public health analysis to explain, and subsequently respond to, patterns of sexual behavior label ed as high risk depends on the capacity of behavioral epidemiologic research to account for a wide range of intrapersonal, contextual and societal factors. The social psychological models of health behavior commonly employed in research on HIV/AIDS fail to acknowledge the social nature of the epidemic. Typically focused on individuals' perceptions of the disease and their confidence and ability to use prophylactics, these models fail to address the HIV/AIDS epidemic as an "epidemic of desire" (Dowsett, 2 003) propagated by human sexuality. An alternative approach recognizes sexual behaviors as meaningful and pleasurable "socially embedded" practices. The subjective meanings and corporeal pleasures resulting from these sexual practices rely on context, whic h in turn exemplifies culture, and is ultimately structured by history and discourse (Dowsett, 2003). This line of reason relates individual acts to dynamic social institutions that structure the contexts in which sexuality is defined and performed. Gende r represents one such social institution and is the focus of this study. Of particular interest are the ways in which broad social categories, such as gender, interact at the individual level to create predictable patterns of high risk sexual behaviors.
17 Until recently the study of gender and health has been nearly synonymous with the study of women's health ( Schofield, Connell, Walker, Wood, & Butland, 2000 ) Additionally, much of public health research operationalizes gender as a single variable that rep resents two dichotomous levels ("ma le" or "female"); thus, gender as a social process, is typically ignored. Instead, gender serves as a proxy measure for biological sex. Some have argued for a new classification of gender and in turn a paradigm for unders tanding women's and men's health (Schofield et al., 2000) The theoretical framework proposed to guide this analysis draws from several areas of gender and feminist theories. West and Zimmerman's reconceptualization of the sex/gender dualism as three ana lytically independent categories is used to explore the relationship between the biological and social dimensions of gender and how these distinctions help to explain the construction of gender in everyday interactions (1987) This process of constructing a gendered self is embedded in preexisting power structures that determine the resources available to an individual for the performance of gender. In other words, gender is multidimensional and is hierarchically ordered by institutional power structures. Research exploring these types of relationships has generally focused on the impact of gender inequality on women's HIV risk. Wingood and DiClemente utilized a theory of gender and power in order to outline a framework in which to understand women's unique exposures, social and behavioral risk factors and biological properties that interact to create vulnerability for HIV infectio n (2000). Their analysis offers detailed consideration of the multiple levels in which gender inequalities impact women's HIV
18 ris k starting with the most distal, societal level and narrowing down to proximate variables such as personal and biological risk factors. For instance, social norms and expectations structure interpersonal relationships and other social institutions like chu rches and families. In turn, these social institutions affect individual exposures to HIV by proscribing or prescribing patterns of sexual behavior. This in depth analysis of gender as a social structure opens multiple avenues for public health research an d strengthens social ecological models of HIV/AIDS. Similar analysis that incorporates gender as a structuring variable across multiple levels is needed when considering men's HIV/AIDS risk. Fortunately, the lack of scholarship in this area is changing wit h the growth of the men's health movement, which has stimulated research on the effects of masculinity on health (Sabo, 1995). International efforts have recently focused attention on the importance of understanding the ways in which gender relations and g ender ideologies affect men's sexual behavior in relation to the HIV/AIDS epidemic (UNAIDS, 2000). This study is envisioned as a small piece of this larger body of research. The Social Construction of Gender West and Zimmerman argued that the widely recog nized conceptualization of sex and gender, where sex is the biological and gender is the social category, is not sufficient for an adequate understanding of the relationships between the biological and social categories and the situational nature of "doing gender" (1987). They proposed, instead, that sex, sex categories and gender be treated as distinctly independent. In this sense, sex describes the biological criteria used to separate males from females. The criteria used for this purpose, as well as the motivation to make the distinction at all, are social
19 pronouncements that manipulate biological, physiological and/or genetic information in order to validate existing cultural beliefs about gender (Fausto Sterling, 2000) Because these biological criteria are not readily available for evaluation in everyday social interactions, sex categories stand as proxies for biological sex. It is the sex category that individuals express through these interactions. This is accomplished by conforming to the socially pr escribed behaviors and displays appropriate for one's sex category. Gender, then, becomes the activities utilized to manage the perception of one's sex category and also the framework used to interpret the sex category of others. The resources used to ma nage an individual's gender come from socially constructed differences between men and women or boys and girls. Behaviors, attitudes and beliefs are categorized as male, female, or neutral. Individuals can pull from this knowledge acquired through socializ ation and manipulate it in order to represent themselves according to their specific goals in specific situations. The motivation to accurately represent one 's self accordingly is very strong and defying society's gender norms often results in serious con sequences. Because of the ubiquity of gender in social interactions, individuals are consistently held accountable for the accurate representation and maintenance of their sex category. As a result, "doing" gender "appropriately" is an important aspect of an individual's sense of him or her self as a competent social actor. The "appropriateness" of the gendered self is determined by institutional power structures that order social classifications, such as race/ethnicity, class, and sexuality into hierarchie s of privilege and marginalization. The result is a plurality of femininity and masculinity.
20 Theories of Masculinities Social psychological research has generally taken two distinct theoretical approaches to the male gender or masculinity. The "trait" app roach, based initially in psychoanalytic theory and then later in theories of personality, focuses on the "extent to which men actually have characteristics culturally defined as masculine (Pleck, Sonenstein and Ku, 1993). A single construct is often used to place individuals on a continuum ranging from more to less masculine based on characteristics such as being active, instrumental, and tool oriented (Smiler, 2004). The "normative" approach, informed largely by a social constructionist paradigm, focus es on the extent to which one believes that men should have the characteristics culturally defined as masculine. Characteristics of appropriate male behavior revolve around themes of emotional stoicism, status orientation, dominance, aggression and independe nce. This distinction between the trait and normative approaches while on the surface seems slight, completely transforms the ways in which masculinity is conceptualized and empirically measured. The former approach positions masculinity within the indivi dual as a stable personality trait whereas the later views masculinity as culturally defined and acted upon by individuals through performative aspects of behavior. In other words, masculine men are viewed as such not because they possess some stable cons tellation of "male" or "manly" traits but because they represent physically and conform behaviorally to cultural definitions and expectations of how men should look, behave, think, and feel. By conceptualizing masculinity in this way, theoretical analysi s is better able to explain the historical evolution of standards of masculinity as well as the social variations that exist
21 among different groups. This also enables the conceptualization of multiple masculinities and femininities embedded in social relat ions of "alliance, dominance, and subordination" (Connell, 2005), thus allowing for more nuanced analysis of the construction of gender over multiple dimensions of social positions, including race/ethnicity, class and sexuality. In other words, by using a plurality of masculinities and femininities, gender can be understood at multiple levels of meaning (Knaak 2004 ) For instance, "doing" masculinity for a white, heterosexual, married, middle class man would most likely result in a very different masculini ty than the one constructed by a black, bisexual, working class man. The differences arise from the dominant position of one over the other: In this case, the dominance of white over black, heterosexual over bisexual, and middle class over working class. T he power to dominate is provided by the institutionalization of patriarchy that determines "legitimacy" of one form over another. Connell refers to this as hegemonic masculinity. He explains: Hegemonic masculinity can be defined as the configuration of gen der practice which embodies the currently accepted answer to the problem of the legitimacy of patriarchy, which guarantees (or is taken to guarantee) the dominant position of men and the subordination of women. (2005) Hegemonic or normative masculinity n ot only guarantees the dominance of men over women but also the dominance of men over other men. The fundamental notion lies in the legitimization of certain groups of men who occupy specific social positions. Therefore, working class me n lack access to so cial and political power needed to construct a hegemonic masculinity; however, this is not true across all contexts.
22 Working class men are excluded in various contexts but still participate in the construction and legitimization of hegemonic masculinity (C onnell, 2005) Drug use, crime, and violence are all intimately related to cultural standards of masculinity such as the need for dominance, aggressiveness and acceptance of risk taking and are readily available to working class men as resources for the pe rformance of male hegemony. This is equally true for gay, bisexual and other MSM since homosexual desire currently assumes a marginalized position in contemporary society. But this is not to say that these men do not actively participate in the constructi on and legitimization of dominant or hegemonic standards of masculinity. Instead, they may in fact over conform to aspect s of hegemonic masculinity in order to engage the gender system and the marginalized status of same sex desire. For instance, expressin g anti feminine attitudes towards effeminate behaving men (Taywaditep, 2001) or endorsing emotionally detached and goal driven sexuality are both ways in which gay, bisexual and other MSM may actively participate in the construction of hegemonic masculini ty. In order to statistically explore the patterns and effects of hegemonic masculinity among larger groups of men the concept of masculinity ideology was created (Pleck, Sonenstein, and Ku, 1993). Masculinity ideology is defined as the "beliefs about t he importance of men adhering to culturally defined standards for male behavior" and is operationalized by measures of "attitudes toward the male gender role" (Pleck, Sonenstein, and Ku, 1993). Congruent with a social constructionist account of masculinity instruments designed to quantify masculinity ideology measure an individual's endorsements or internalizations of the culturally "appropriate" behavior for
23 men. In other words, masculinity ideology is a measure of an individual's endorsement of hegemonic norms of masculinity. Individual endorsements of hegemonic norms of masculinity become important to public health research when those norms have deleterious effects on men's health behaviors. Masculinity and Health Health beliefs and behaviors are a res ource for both men and women in the construction of masculinity and femininity. Some have described health related behaviors and cognitions as a form of "currency in the transactions that are continually enacted in the demonstration of gender" (Courtney, 2 003) In this sense, health beliefs and behaviors are resources utilized by individuals for the social construction of gender. Researchers have demonstrated that some men who adopt traditional beliefs about masculinity may engage in behaviors that have gre ater health consequences than other males with less traditional beliefs about masculinity (Courtenay, 2003) For instance, men who hold more tradition al beliefs about masculinity may be more likely to smoke, use alcohol and drugs, and engage in more risky behavior (Courtenay, 2003) Courtenay theorizes: denial of risk and other unhealthy behaviors are used by men in the negotiation of social status and to enact idealized forms of masculinity that enable them to assume positions of social power relative to women or less powerful, marginalized men, such as gay and lower class men. (2000) Pleck, Sonenstein and Ku provide d empirical support for a social constructionist view of masculinity and its relative association with measures of sexual behavior (1993 ). Using data from the National Survey of Adolescent males, they found that among males ages 15
24 to 19, higher score s on their masculinity scale were positively correlated with number of sexual partners, negative attitudes toward condom use, and less overal l condom use. Shearer partially replicated these findings on a sample of male and female college students using the full version of Thompson and Pleck's Male Role Norms Scale (1986). She found that masculinity ideology was a significant predictor of three sexual behavioral outcomes: sex without a condom, engaging in casual sex, and endorsement of risky condom related beliefs (Shearer, 2005). In particular, she found that the "antifemininity" subscale of the MRNS was positively related to all three outcomes. Expanding this line of research, Noar and Morokoff conducted a similar study on a sample of male college students. They found that higher endorsement of masculinity ideology as measured by the MRNS was related to more negative attitudes related to condom use but was not directly related to condom use (Noar and Morokoff, 2002). Instead, they found support for an indirect relationship between masculinity ideology and condom use that operates through condom related attitudes. These findings are also support ed by similar findings from qualitative research. A study conducted by the San Francisco AIDS Foundation focused on the social contextual and situational circumstances surrounding UAI among self identified gay and bisexual men. The researchers concluded fr om open ended interviews that many men who engaged in UAI did so in order to meet non sexual needs relating to a broad array of factors including the need to affirm masculinity (Executive Summary of the SF AIDS Foundation's Qualitative Interview Study, 199 7) The men reported that at times these non sexual needs overpowered their concern for possible infections that might occur as a
25 result of their behavior. Another qualitative study that focused specifically on barebacking found that sexual acts and encoun ters took on multiple meanings for their participants and that these meanings were organized around concepts of masculinity (Ridge, 2004). The research summarized here underscores the importance of gender and masculinity in a comprehensive understanding of sexual risk taking; however, these relationships require further study. These findings need to be replicated among more diverse samples of men that include MSM. Furthermore, there is no consensus regarding a gold standard measure of masculinity. Each of the studies previously described utilized the MRNS; however, the evolution of masculinity measurements is extensive and reflects important developments in the theories of gender and sexuality (Smiler, 2004). This body of literature can be strengthened if consistent associations described previously can be substantiated with other measures of masculinity. Exploring the relationships of masculinity and sexual behaviors among gay, bisexual and other MSM is a necessary step toward a comprehensive understandin g of gender as a social determinant of men's health and illness. Therefore, this research project seeks to explore the relationships among masculinity and high risk sexual behavior. Others have proposed that masculinity is constructed among MSM through t he use of culturally available sexual scripts that emphasize sexual conquest, emotional detachment, and the "pursuit of sexual gratification for its own sake, and by the association of danger and excitement" in sexual encounters (Levine & Kimmel, 1998) If MSM do utilize such scripts to define and
26 construct masculinity through sexual encounters, then it would follow that engaging in "safe" responsible sex is counter productive. If [s] ex is about danger, excitement, risk" then safe sex is about "comfort, se curity and softness" (Levine & Kimmel, 1998) By this logic, norms of masculinity may encourage sexual risk taking among MSM who judge these norms to be indicative of the way "men are" or "should be."
27 CHAPTER 3 METHOD Study Design This study utilized a cross sectional self report survey design Items used to measure the constructs under investigation were taken from previous research when available or developed specifically for this study. Participant s were recruited and the data were collected using the Internet. Study Objectives Th e goals of the current study were to: (1) describe the sexual behaviors of a sample of MSM recruited from the Internet focusing on high risk behaviors that could lead to HIV infection/transmission; (2) determine the preva lence of recreational drug use (in particular "Club Drugs") focusing specifically on use before and/or during sex; (3) ascertain the importance of the Internet in meeting sex partners; (4) explore the association between concepts of masculinity and high r isk sexual behavior and intentions; (5) explore possible mediating effects of sexualized drug use and using the Internet to meet sex partners. The proposed relationships are summarized in Figure 1. The following hypotheses are offered:
28 Hypothesis 1: Masc ulinity Ideology will be independently associated with the number of sexual partners over the p revious six months both as the insertive (Top) and receptive (Bottom) partner. a. Masculinity Ideology is positively associated with the total number of sexual partners where the respondent was the insertive partner (Top) in the past six months. b. Masculinity Ideology is positively associated with the total number of sexual partners where the respondent was the receptive partner (Bottom) in the past six months. Hypothesis 2: Masculinity Ideology is positively associated with the frequency of unprotected anal intercourse during the previous six months both as the insertive (Top) and receptive (Bottom) partner. a. Masculinity Ideology is positively associated with the frequency of insertive unprotected anal intercourse (IUAI) in the previous six months. b. Masculinity Ideology is positively associated with the frequency of receptive unprotected anal intercourse (RUAI) in the previous six months. Hypothesis 3: Masc ulinity Ideology is positively associated with the intention to seek out bareback sexual encounters. Hypothesis 4: Substance use before or during sex and using the Internet to meet sex partners will partially mediate the association between masculinity ide ology and a dichotomous measure of high risk behavior (Unprotected Anal Sex in the previous months).
29 Figure1. Conceptual model of Masculinity Ideology and High Risk Sexual Behavior. M a s c u l i n i t y I d e o l o g y H i g h R i s k S e x u a l B e h a v i o r / B a r e b a c k I n t e n t i o n s + U s e o f D r u g s D u r i n g S e x u a l E n c o u n t e r s U s e o f I n t e r n e t t o M e e t S e x P a r t n e r s + + H y p o t h e s i s 4 H y p o t h e s i s 1 2 & 3 Internet based Research An increasing num ber of researchers are turning to the Internet as a means to collect both quantitative and qualitative data (Mann, 2000), including those who have used the Internet to study sexual risk taking behavior among MSM (Bull & McFarlane, 2000; Bull, McFarlane, Ll oyd, & Rietmeijer, 2004; Elford, Bolding, Davis, Sherr, & Hart, 2004; Rhodes, DiClemente, Cecil, Hergenrather, & Yee, 2002) There are several advantages to using the Internet to recruit participants as well as a data collection tool. For instance, it pro vides the ability to reach geographically, socially isolated, or "hidden" populations that may not have been available otherwise. This offers a great advantage when conducting research with MSM populations One
30 major limitation of much of the sexual health research focusing on MSM is the use of convenience samples recruited from established gay venues or street intercept surveys (Mansergh et al, 2002; Halkitis, Parsons, Wilton, 2003; Halkitis, Wilton, Wolitski, et al., 2005, Parsons & Bimbi, 2007 ). Much of this research takes place in large urban cities within well established gay communities and may not be generalizable to MSM who reside outside these urban centers The Internet provides a means to recruit MSM from a variety of Internet "venues" that attrac t a diverse sample of men. This i ncludes gay and non gay identified men who are not acculturated into an established gay community as well as MSM from rural communities or cities without a gay "scene." A second advantage of conducting Internet based resear ch is the flexibility of time and resources. By recruiting participants and administering surveys online researchers can eliminate some of the material and non material costs of research, which includes time needed to recruit and administer surveys as wel l as the monetary associated costs (e.g. paper, postage, research assistants, data entry, etc.). In the past few years there has been an increase in the availably of prepackaged software specifically designed to assist researchers in condu cting web based s tudies. The cost and functionality of these packages vary considera bly; therefore, the use of such packages depends highly on the actual needs of each research project (Wright, 2005). Furthermore by recruiting and collecting data online participants ar e able to respond with a greater degree of actual and perceived anonymity. This is a major advantage when collecting sensitive information like sexual histories and is especially relevant when collecting this information from a socially stigmatized populat ion like
31 MSM (Rhodes, Bowie & Hergenrather, 2003; Riggle, Rostosky & Reedy, 2005). Therefore, Internet based research may help to decrease social desirability bias which is a major concern with face to face interviews (Rhodes, Bowie & Hergenrather, 2003). Conversely, increased anonymity means decrease d control ov er who responds to the survey as well as the conditions surrounding its completion. Other concerns associated with Internet based research include differential rates of Internet access among soc ioeconomic groups as well as methodological issues related to the ability to generate a probability sample. Recent scholarship in these areas indicate that, as the cost of personal computers has decreased along with the ubiquitous presence of computers in public domains, the percent of individuals with Internet access continues to increase and at the same time the demographic profile of Internet users is beginning to normalize (Rhodes, Bowie & Hergenrather, 2002; Yun & Trumbo, 2000). There are, however, iss ues regarding the representativeness of samples derived online; therefore, the results of Internet based studies should be interpreted with these limitations in mind. Particular to this study and many others, the use of convenience samples limits the gener alizability of the results. And while random sampling is possible in certain cases, too often random sampling is not feasible and is especially difficult in this type of Internet based research; however, this is hardly an issue unique to web based research Ethical Issues in Internet Research In addition to these methodological issues, there are a number of ethical challenges to consider when conducting web based research. Similar to the methodological concerns addressed, there are some unique ethical pr oblems raised by
32 Internet research. A workshop convened by the American Association for the Advancement of Science (AAAS) underscored several ethical considerations specific to Internet research. Among these is the anonymity of online participants, the co mplexities of obtaining consent, the issues of privacy and the difficulty in transferring traditional standards of public and private domains to Internet research (Frankel and Siang, 1999). Some of these issues, such as the disagreement over th e public/pr ivate distinction, are important consideration for research conducted with online communities or communications. For the purpose of the present study, the Internet was used as a means of recruitment and survey delivery. Online conversations or other commun ications w ere not recorded. To ensure the protection of the participants of this study as highlighted by the CITI Course in The Protection of Human Research Subjects the following procedures were followed: Consent from participants was obtained prior to t he completion of the survey; Participants w ere asked to read a written statement containing the basic elements of consent as well as the requirements for participation (e.g. must be 18 years old or older) ; and then offer their consent by clicking on a lin k to begin the survey. The risk involved to participants was considered minimal; however, the survey d id ask sensitive questions that may be perceived as intrusive or inappropriate by some. Full disclosure of the personal nature of the survey w as included in consent form. Finally, no identifying information w as collected from participants to ensure their confidentiality. Approval from the University of South Florida's Institutional Review Board w as given prior to data collection.
33 Participants and Procedu res A non random convenience sample was used for this analysis. All participants were recruited via the Internet and completed a web based survey instrument. The target population included single, sexually active MSM between the ages of 18 and 65 who were able to read and comprehend English. Participation was not limited to those who met inclusion criteria in order to maximize the sample size by not discouraging potential respondents with the use of access passwords and other such restrictions. Recruitment The first step in the data collection phase was to construct a website to promote the study and provide information about the study to visitors. Potential participants were referred to this website by all subsequent recruitment strategies. The primary rec ruiting strategy utilized electronic postings and ads placed on Internet websites frequented by MSM such as weblogs, social and sexual networking sites, and listservs. The administrators of these sites were contacted and asked to post an ad or description of the study as well as a link to the study's website. Some visitors of these sites in turn posted descriptions of the study on their own sites. An open ended item at the end of the survey inquiring how participants were referred suggested that the majorit y of respondents were initially referred to the survey website from these electronic postings on weblogs and message boards. This was confirmed by the statistics available from a web counter installed on the site that tracked the number of visitors as well as the originating websites from which the visitors were directed. According to this information, a majority of
34 visitors were referred from weblogs and Internet sites that had a description and link to the survey, all of which catered to a variety of MSM A second recruiting approach included the use of personal contacts and acquaintances who initiated a snowball sample by emailing a link to the survey to their contacts, who in turn emailed a link to the survey to their own contacts. All potential parti cipants were directed to the study's website This strategy accounted for the rest of the sample; however, a small percent of respondents indicated that they came across the survey while "surfing the Internet." Procedure After being directed to the study' s website, participants were presented with information about the study, including an electronic consent form. Visitors were asked to read the information and offer their consent by clicking a button that directed them to the actual survey instrument. The survey instrument contained seven sections presented in chronological order beginning with demographic and background information followed by sections dealing with sexual venues, recent sexual encounters, barebacking, substance use, masculinity ideology, a nd health. No identifying information was collected from the participants. They were asked for the month and year of their b irth and the first three numbers of their ZIP codes. This information was used to calculate participants' age s and to check for dupl icate responses After completing the survey, respondents were directed to a separate page thanking them for their participation. This page also included contact information for the researcher as well as links to relevant HIV/AIDS educational sites.
35 Anal ytic Sample A total of 4,126 respondents initiated the survey and completed the demographic prescreen questions. There was a surprisingly large international response to the survey (n=1456) that was unanticipated in the proposal process. It was decided th at these participants would not be included in the analytical sample for two reasons. Firstly, upon preliminary screening of the data, it appeared from open ended responses that there were some language barriers with some of the international respondents. Furthermore, the masculinity scale used in the current study was developed with a sample drawn from a U.S. population of men, and therefore psychometric data was not available for international samples. In an effort to maintain the validity of the survey i nstrument, it was decided to limit this analysis to U.S. residents only. The analytical sample was refined further by eliminating those that did not meet inclusion criteria. This included those individuals who reported being female, in a current relations hip, having no sexual contact with males during the previous year, and those outside the age requirement. In addition, five duplicate responses were found and subsequently removed. Attrition was high as 340 respondents logged off the survey before comple ting the masculinity scale. These cases were deleted because of a substantial amount of missing data on key variables. For those remaining participants with less than 20% missing items on the masculinity scale, mean item imputation was used to replace miss ing data (Roth, Switzer & Switzer, 1999). The resulting analytical sample consisted of 907 cases.
36 MEASURES Demographic and Descriptive Information The survey instrument contained questions about participants' ages, years of formal education, race/ethnic ity, country of residence, and the first three numbers of U.S. ZIP code (this information was be used along with year of birth in order to search for replicated data ). Participants were also asked about their HIV s tatus any recent STIs, and frequency of H IV testing Sexual identity was an important descriptive characteristic; particularly in this study beca use the recruiting strategy did not focus exclusively on self identified gay men. Assessing the degre e to which the sample identified as gay, bisexual or heterosexual wa s important in determining the degree to which the research findings could be generalized to non gay identified MSM. Because discordance between self reported sexual orientation and actual behavior has found to be associated with demogra phic characteristics (i.e. race/ethnicity and education level) as well as lower condom use and HIV testing (Pathela, Hajat, Schillinger, Blank, et al, 2006), sexual orientation was assessed with a question of sexual identity (e.g. gay, bisexual, heterosexu al) and a question about the sex of previous sexual pa rtners (men only, females only, both males and females). Sexual Behavior Sexual behavior was measured by asking several questions about sexual activity in the previous six months. Respondents were ask ed to report the total number of partners with whom they had anal sex, differentiating between insertive and receptive partners.
37 Responses were open ended to minimize response bias. Similar measurements of self reported sexual behavior have been evaluated and found to be reliable (Catania, Gibson, Chitwood, & Coates, 1990; Kauth, St. Lawrence & Kelly, 1991). Furthermore, using aggregate data as a proxy for the more rigorous detailed partner by partner analysis of HIV risk has also be found to be reliable (P inkerton, Benotsch & Mikytuck, 2007). In addition to the total number of partners, the frequency of insertive and receptive anal sex without condoms was measured on a 6 point scale ranging from "not at all" to "50 or more times." Barebacking Intentions I ntentions for barebacking was assessed by five items that asked participants to agree or disagree on a 5 point scale to statements such as "I purposely seek out bareback sex (no condoms) as a top," "I don't seek out bareback sex, but if it happens that's okay only if I'm the bottom" and "I consider myself a Barebacker." These items were used in previous research (Grov et al., 2007; Parsons & Bimbi, 2007); however, they were not combined as a scale in these studies. The prevalence of harm reduction strateg ies, such as strategic positioning, highlights the importance of analyzing insertive and receptive anal sex as separate behaviors (Grov et al., 2007; Parsons et al., 2005; Kippax & Race, 2003). Rather than analyzing intentions for bareback sex separately f or receptive and insertive sex, this study combines the five items into a single index of barebacking intentions. This was done in an effort to reduce the number of outcome variables. The items demonstrated good internal consistency (alpha=0.87) lending em pirical support to this decision. A higher score on this scale indicates greater intentions for unprotected anal sex.
38 Internet Use Two items measured the frequency of using the Internet for seeking and meeting sexual partners within the previous six months Participants were asked how many times in the last six months they used the Internet "to find a hookup (casual sex partner)" and about how often they met and had sex with a guy they met online. Responses were recorded on a 6 point scale ranging from "not at all" to "50 or more times." This analysis was most interested in qualitative differences between those who have had sex with a partner they met online in the previous six months and those who have not. Therefore, a dichotomous measure of using the Inte rnet to meet sex partners was created for regression analysis. Substance Use Participants reported yes or no to any use of the following substances: alcohol, Viagra, Methamphetamine, 3 4 methylenedioxy methamphetamine ("Ecstasy"), Ketamine, inhalant nit rates ("Poppers"), Gamma Hydroxybutyrate ("GHB"). Only substance use in the last six months that occurred immediately before or during sexual activity was measured. A global measure of drug use w as not used due to questions regarding their utility in evalu ating associations with sexual behavior. Measures of substance use during and or before sex have been shown to be more useful in establishing associations with sexual risk taking (Leigh & Stall, 1993); similar measures have been used in other research on M SM (Kalichman, Heckman, & Kelly, 1996; McKirnan, Vanable, Ostrow & Hope, 2001). A dichotomous measure of drug use was created for regression analysis. Men who reported any use of methamphetamine, ecstasy, Ketamine,
39 inhalant nitrates, or GHB were categorize d as recent drug users in sexual context. Masculinity Ideology It has been argued that there are two distinctly independent categories of masculinity measurements; one measures gender orientation and the other measures gender ideology (Thompson, Pleck, & Ferra 1992; Thompson & Pleck, 1995). The instruments grouped into these categories were developed under notably different theoretical frameworks. The current research conceptualize s sexual behavior as a means by which men participate in gender relations. Therefore, it is necessary to utilize masculinity instruments that were developed using a social constructionist view of gender. Such an instrument would seek to tap into an individual's endorsement of normative beliefs about what men should be and what me n are like. The norms th at encompass these dominant cultural definitions of masculinity and manhood can be understood as hegemonic representations of masculinity. There are several instruments available that proclaim to measure masculinity ideology (for a full review see Smiler, 2004); therefore, for the purpose of this study, five criteria were used to evaluate the appropriateness of any on e instrument : 1) Congruent with Social Constructionist Theoretical Framework; 2) Measure Multiple Dimensions of Mascul inity; 3) Items are appropriate for use with MSM; 4) F ound to have sufficient psychometric properties; 5) Short in Length. After a thorough review of the available scales, it was decided that a revised version of the "Masculinity, Attitudes, Stress, and Conformity Questionna ire" (MASC) would be used in this study. The MASC was developed in order to measure masculinity
40 ideology as well as an individual's own conformity to and distress resulting from male role expectations (Nabavi, 2004). Measuring these t hree distinct constructs within one instrument is an innovative approach that has not been previously explored in the masculinity literature; however, for the purposes of this research and in accordance with the criteria previo usly described, only the item s measuring attitudes towards masculinity were used. The MASC is reported to measure six dimensions of masculinity ideology including constructs relating to emotional restrictiveness, independence, achievement, dominance, aggressiveness, and sexuality. Th e MASC attitudes scale consists of 36 items and was found to have excellent internal consistency as measured by Cronbach's alpha ( = .95) (Nabavi, 2004). Convergent validity was enumerated by examining the magnitude and direction of the bivariate associat ions between the MASC and the following scales: Male Role Norms Inventory (MRNI), Gender Role Conflict Scale (GRCS), Stereotypes about Male Sexuality Scale (SAMSS), and the Attitudes Toward Women Scale (AWS) (Nabavi, 2004). The MRNI is a widely use d measur e of masculinity ideology (Levant et al.,1992). A positive correlation between the MRNI and MASC attitudes scale is an indicator of the scale's convergent validity. Nabavi reported a strong positive association ( r = .70) between the two scales (2004). Sim ilarly, the MASC attitudes scale was positively correlated ( r = .54) with the SAMSS, an instrument designed to measure the degree to which an individual endorses normative cultural ideals of mal e sexuality. These findings lend empirical support for the con vergent validity of the MASC attitudes scale. Two items from the MAS C were not included because they appeared inappropriate for
41 use with a sample of gay, bisexual and othe r MSM ("A man should be the head of the family if he has a wife or partner" and "a m an should take the lead when it comes to sex"). In addition to the MASC, items from the "Meanings of M asculinity Scale" (MMS) were included This instrument was developed to measure gay men's conceptions of masculinity based on three separate dimensions: Masculinity as Physical Appearance, as Social Behavior, and as Sex (Halkitis, Green, and Wilton, 2004). Five items taken from the "Masculinity as Physical Appearance" subscale w ere used in this study. The purpose of these items is to address an often unex plored dimension of masculinity physical appearance. Three items were not included from the original subscale in order to reduce redundant items and to eliminate esoteric items that are inappropriate for use with non gay identified MSM (i.e."The guys in To m of Finland portraits represent the masculine idea"). Internal cons istency of the "Physical Appearance" subscale was determined to be good with !=0.81 (Halkitis, Green, and Wilton, 2004). A final set of five items was included in order to further explore aspe cts of masculinity as it relates to cultural norms surrounding sexuality. Three items were adapted from the Stereotypes About Male Sexuality Scale (SAMSS), specifically from the "Spontaneous Sex" subscale that seeks to measure the cultural norm that sexual behavior for men is natural and spontaneous (Snell, Belk, and Hawkins, 1986). Two additional items were created that relate to this norm ("Most men don't like to think about the consequences of sex" and "For most men, good sex is about taking risks"). In sum, a total of 44 items were included in the survey instrumen t to measure multiple
42 dimensions of masculinity (See Table 1 for a summary of these variables). DATA ANALYSIS Data w as downloaded from the online survey directly into a spreadsheet format. The SAS statistical software version 9.1 for Microsoft Window s w as used to conduct all statistical analysis (SAS Institute, Cary, NC). Variable Type of Measurement Values Masculinity Ideology Interval 1=Strongly Disagree 6=Strongly Agree Mediators Drug Use before or during sexual activity Categorical 0=No 1=Yes Internet Use to meet sex partner(s) Categorical 0=No 1=Yes Control Variables Age Ratio Age in Years Race/Ethnicity Categorical 0=White/Caucasian 1=Not White/Caucasian Education Categorical 0=Not College Graduate 1=College Graduate Sexual Identity Categorical 0=Gay/Other 1=Bisexual HIV Status Categorical 0=HIV Negative/Never Tested 1=HIV Positive Table 1. Summary of Independent Variables
43 Preliminary Scale Development To date, no known published research has empirically confirmed the multidimensionality of the MASC subscales. Furthermore, a composite of three separ ate scales were used to measure masculinity ideology in the present study. For these reasons, principal axis factor analysis using the varimax rotation method was conducted to confirm the multidimensionality of the scale. As an exploratory step, the number of factors to retain was determined using parallel analysis, which is argued to be a superior method for determining factor retention compared to more commonly used criteria (Velicer, Eaton, Fava, 2000). Parallel analysis involves a comparison of the aver age eigenvalues generated from random correlation matrices with the eigenvalues from the real data correlation matrix. A factor is retained if the eigenvalue from the real data is greater than the eigenvalue from the randomly generated data (Hayton, Allen & Scarpello, 2004). This analysis failed to confirm the multidimensionality of the masculinity measures used in this study. All but two items significantly loaded (minimal factor loading of .32) on one factor. A composite variable of the 42 items measuring masculinity ideology, excluding the two items that did not significantly load on the factor, was created based on these findings. Reliability analysis using Cronbach's Alpha demonstrated strong internal consistency of this composite measure ( =0.95). An alyses Frequencies of the demographic data and substance use items were calculated for descriptive analyses. Univariate analysis w as conducted on all of the interval and continuous items in order to generate measures of central te ndency, distribution,
44 skew ness, and to identity outliers. Bivariate analys es (zero order correlations) were performed in order to provide a cursory examination of the hypothesized relationships. Ordinary le ast squares multiple regression analyse s were conducted to ascertain the p ercentage of unique variance explained by the independent variable (masculinity ideology) over and above that accounted for by the controls (Age, Race/Ethnicity, Education, Sexual Identity, HIV Status). Hierarchical multiple regression models were construc ted in two steps. Two models were generated for each of the five dependent variables. Age, race/ethnicity, education, sexual identity and HIV serostatus were entered in step one as a block of control variables. The composite masculinity variable was added in step two. This procedure allowed for the analysis of the change in the percentage of variance explained by the addition of the masculinity variable, independent of that explained by the control variables. A similar approach using logistical regression m odels for a dichotomous outcome (any unprotected anal sex in last six months) was used to test the mediation hypothesis. Mediating effects were evaluated based on the following criteria: (1) The independent variable (i.e. Masculinity Ideology) is associat ed with the mediator variables (i.e. Substance Use and Internet Use); (2) The mediator variables are associated with the dependent variable (i.e. UAI); (3) The independent variable should be associated with the dependent variable; (4) The association betwe en the independent and dependent variable is statistically decreased (evidenced by the semi partial coefficient) with the inclusion of the mediating variable (Baron and Kenny, 1986).
45 CHAPTER 4 RESULTS DESTRIPTIVE STATISTICS Sample Characteristics (Contro l Variables) Characteristics of the study sample are described in Table 2. The age of the sample ranged from 18 to 65 years with a mean age of 37.9 (SD=12.3). Respondents were primarily Caucasian or white (83.1%), college educated (60.2%), gay identified ( 80.5%), and HIV negative (73.8%). Approximately 5% reported a recent STI diagnosis. Sexual Activity and Barebacking Intentions (Dependent variables) The majority of the sample reported having anal intercourse at least once within the previous six months (7 6.9%), forty nine percent did so at least once without a condom. On average, respondents reported 3.56 (SD=9.48) and 3.52 (SD=11.55) partners with whom they engaged in insertive and/or receptive anal sex, respectively. Nearly 64% of the sample reported no insertive unprotected anal intercourse (IUAI) and 68% no receptive unprotected anal intercourse (RUAI) within the previous six months. Measured on a 5 point scale, the mean for barebacking intentions was 1.76 (SD=.90), indicating disagreement with the stat ements; however, the scores ranged between 1 (Strong Disagree) and 5 (Strongly Agree). Descriptive statistics for the five outcome variables can be found in Table 3.
46 Characteristic No. (%) Age (Mean = 37.9) 18-20 72 (7.9) 21-30 242 (26.7) 31-40 190 (21.0) 41-50 245 (27.0) 51-60 130 (14.3) 60-65 28 (3.1) Race/Ethnicity White/Caucasian 754 (83.1) Hispanic/Latino 69 (7.6) Asian/Pacific Islander 32 (3.5) Black/African American 23 (2.5) Mixed Race/Other 29 (3.2) Education High School or less 56 (6.2) Some College or Tech School 304 (33.6) Bachelor's Degree 302 (33.3) Graduate/Professional School 244 (26.9) Sexual Orientation Gay 730 (80.5) Bisexual 143 (15.8) Straight/Heterosexual 16 (1.8) Other Specified 14 (1.5) HIV Status Negative 669 (73.8) Positive 77 (8.5) Never Tested 161 (17.8) Table 2. Demographic Characteristics of Study Sample Note Numbers might not sum to total because of missing data.
47 All five of the dependent variables were positively skewed to varying degrees pulli ng the tails of the distributions to the right. Number of partners in which the respondents reported insertive intercourse contained the most extreme outlier, more than double that of the next highest value (100 partners). There was no indication that thes e extreme observations were erroneous. Furthermore, highly skewed distributions with similar measures of sexual behavior among samples of MSM are not uncommon in the literature (CDC, 2004; Benotsch, Kalichman, Cage, 2002; Tabet et al., 1998), suggesting Mean Standard Deviation Min Max Skewness coefficient Skewness coefficient after transformation No. of Sexual Partners Anal insertive 3.56 9.48 0 123 6.61 1.15 Anal receptive 3.52 11.55 0 250 12.95 1.22 Frequency of Unprotected Anal Sex a Anal insertive 1.62 1.03 1 6 1.91 1.1 Anal receptive 1.58 1.03 1 6 2.07 1.25 Intentions for Barebacking b 1.76 0.9 1 6 1.07 Table 3. Descriptive Statistics for Dependent Variables Note. Dashes indicate the variable was not transformed and the original values were used in subsequent analyses. a 1= Zero (not at all), 2= one to two times, 3= three to nine times, 4= ten to nineteen times, 5= twenty to fortynine times, 6= fifty or more times. b 1=Strongly Disagree, 2=Disagree, 3=Neither agree nor disagree/Neutral, 4=Agree, 5= Strongly Agree.
48 th at the distribution for this population is non normal. Non parametric analyses are not affected by non normal distributions and provide an alterative analytic approach; however, these tests are not immune from the influence of outliers (Zimmerman 1995; Zim merman, 1994). Considering these factors, logarithmic transformations of four of the five dependent variables were calculated (the degree of skew of the barebacking intentions variable was acceptable and did not warrant transformation). Logarithmic transfo rmations are mathematical operations that pull in the tails of a distribution, reducing skewness and improving normality, while retaining the rank order of the values. The degree of skewness of the transformed variables was markedly reduced and smaller tha n the suggested maximum value of 2.0 (Tabachnick & Fidell, 2007). Standardized scores (z scores) were calculated before and after transformation to determine if the highest values on the skewed variables were disconnected from the rest of the scores, indi cating extreme outliers. As expected, large standard scores (i.e. 21.3 and 12.60) disconnected from the distribution of standardized scores were found in the untransformed variables; however, after transformation, the standard scores of the highest values decreased and no long appear disconnected from the rest of their distributions (Tabachnick & Fidell, 2007). Proceeding with parametric analysis was determined to be acceptable considering these improvements in the distributions of the variables, combined w ith a large sample size that safeguards against violations of normality. Masculinity Ideology (Independent Variable) The scores on the measure of masculinity ideology (MI) ranged from 1 to 6 with high scores indicating stronger endorsement of hegemonic MI. On average, the sample
49 appeared to moderately or slightly disagree with the items measuring MI (M = 2.75, SD=.77). Drug and Internet use ( Mediating Variables) Ninety six respondents skipped the section on drug use. Of those who completed these items (n= 811), approximately 26% reported using drugs (Poppers, Cocaine, Ecstasy, Methamphetamine, GHB and/or Ketamine) before or during sex on at least one occasion within the previous six months. The most commonly reported drug used before or during sex was inha lant nitrates ("Poppers") with nearly 24% of respondents using it at least once. Cocaine was the second most common (5.5%) followed by Ecstasy (5.4), Crystal Meth (5.4%), GHB (3.9), and Ketamine (1.4). A dichotomous variable was created for use with regres sion analysis. The variable categorized drug use within a sexual context in the previous six months as yes or no. The majority of respondents (78%) reported ever having had sexual relations with someone they met using the Internet. Within the previous s ix months, 65% of men reported meeting and having sex with men they met using the Internet. Fifty two percent of men specified having anal sex with a partner they met using the Internet. CORRELATIONAL ANALYSIS Bivariate associations were explored in ord er to obtain a cursory understanding of the size and direction of the primary relationships. Specifically, masculinity ideology (MI) was correlated with each of the five outcome variables, which included the number of insertive and receptive sexual partner s, frequency of insertive and receptive unprotected anal sex (IUAI & RUAI), and barebacking intentions. MI was positively
50 associated with the number of insertive and receptive partners within the previous six months ( r = .17, p<.0001 and r = .20, p<.0001, respectively). On average, those who endorsed more hegemonic masculine norms had a greater number of sexual partners both insertive and receptive roles. Positive associations were also found between MI and the frequency of unprotected anal intercourse. IUA I was positively associated with MI ( r = .16, p<.0001) indicating that those who endorsed hegemonic norms had a greater number of unprotected sexual encounters when they were the insertive partner. Similarly, those who endorsed hegemonic norms also reporte d greater frequency of RUAI ( r = .14, p<.0001). Like the behavioral outcomes, barebacking intentions were also positively associated with masculinity ideology ( r = .23, p<.0001). It is not surprising that the five outcomes were also all significantly assoc iated, indicating that those engaging in one risk behavior are more likely to be engaging in other behaviors. Pearson product moment correlational coefficients for these associations can be found in Table 4. Results from these preliminary analyses lend su pport for hypothesis 1, 2 and 3; however, further analysis was needed to see if these associations hold when controlling for age, race/ethnicity, education, sexual identity, and HIV serostatus. The potential intervening influence of the drug and Internet u se variables were also explored using logistical multiple regression.
51 REGRESSION ANALYSIS Hypothesis 1 It was hypothesized that MI would be independently associated with the number of sexual partners after controlling for relevant individual character istics. Number of sexual partners was analyzed separately for insertive and receptive anal sex as previous research indicates that these represent two distinct sexual behaviors with distinct predictors. Five control variables were entered as a single block and included age, race, education, sexual identity, and HIV serostatus. The second step involved the addition of the MI variable in order to ascertain its unique contribution to the amount of variance explained by the models. Results from these analyses a ppear in Table 5. Variable 1 2 3 4 5 6 1. MI --2. No. Insertive partners .17** --3. Freq. of IUAI .16** .64** --4. No. Receptive partners .20** .34** .22** --5. Freq. of RUAI .14** .23** .37** .65** --6. Barebacking Intentions .23** .28** .46** .34** .46** --Table 4. Correlation matrix for masculinity ideology (MI) and dependent variables. Note MI = Masculinity Ideology; IUAI = Insertive unprotected anal intercourse; RUAI=Receptive unprotected anal intercourse. p < .01. **p <.0001
52 Explanatory Variables R 2 F df Insertive Anal Sex (Log) Step1 0.03 5.73*** 5,894 Age -.06 Non-white -.02 College Graduate -.10** Bisexual Identity .03 HIV Positive .16*** Step2 0.06 9.37*** 6, 893 Age -.03 Non-white -.02 College Graduate -.08* Bisexual Identity .01 HIV Positive .15*** Masculinity .17*** Receptive Anal Sex (Log) Step1 0.07 14.41*** 5, 893 Age -.15*** Non-white .02 College Graduate -.03 Bisexual Identity -.03 HIV Positive .24*** Step2 0.11 18.62*** 6, 893 Age -.12** Non-white .02 College Graduate -.02 Bisexual Identity -.06 HIV Positive .24*** Masculinity .20*** *p<.05 **p<.001 ***p<.0001 Table 5. Regression analysis for the prediction of number of anal sex partners within the previous six months.
53 Number of Insertive Partners The first step containing the control variables showed that graduating from college and HIV positive serostatus were independent correlates of the number of insertive partners. The independent contribution of education was considerably less than HIV serostatus evidenced by the smaller standardized regression coefficient (see Table 5) and lower levels of significance and are most likely artifacts of the large sample size. In the second step, MI was independen tly associated with number of insertive partners (" = .17, p<.0001) after adjusting for the effects of the control variables (F (5, 894) = 9.37, p<.0001). The total model explained 6% of the variance in number of insertive partners with 3% of that accounted fo r by MI. A comparison of the standardized regression coefficient, or Betas, showed that MI was the most important variable in terms of its individual contribution to the total model. Men who endorsed more normative attitudes about masculinity were more lik ely to also have a greater number of anal sex partners in which they assumed the insertive role. Number of Receptive Partners The first step containing the control variables showed that age and HIV positive serostatus were independent correlates of the nu mber of receptive partners after adjusting for the other variables in the model (p<.0001). MI was added in the second step. This model was statistically significant (F (6, 893) = 18.62, p<.0001) accounting for 11% of the total variance. MI was independent ly associated with the number of receptive partners (" = .21, p<.0001) and accounted for 4% of the explained variance. These results indicate that men with more normative attitudes towards masculinity were more likely to
54 have had a greater number of anal s ex partners in which they assumed the receptive role. Hypothesis 2 It was hypothesized that MI would be positively associated with the frequency of unprotected anal intercourse within the previous six months. Insertive (IUAI) and receptive (RUAI) unprote cted anal intercourse were analyzed separately. The control variables were entered in the first step followed by the addition of the MI variable. Results from these analyses appear in Table 6. IUAI The control variables that were statistically significan t predictors of IUAI included age and HIV positive serostatus. Age was marginally significant (p<.05) and is most likely an artifact of the large sample size. Age did not remain statistically significant in the second step, confirming this assumption (See Table 6 for Betas and significance levels). The addition of MI to this model increased the R 2 value to .05 and was statistically significant, F (6, 894) = 8.05, p<.0001. A comparison of the Betas indicate that MI was the strongest predictor (" = .16, p<.0001) in the model and independently contributed 2% of the explained variance. The data indicate that men who endorsed more traditional attitudes about masculinity were more likely to frequently engage in insertive anal sex without a condom.
55 Explanatory Variables R 2 F df IUAI (Log) Step1 0.03 5.13*** 5, 895 Age -.07* Non-white -.05 College Graduate .01 Bisexual Identity .02 HIV Positive .16*** Step2 0.05 8.05*** 6, 894 Age -.05 Non-white -.05 College Graduate .01 Bisexual Identity -.01 HIV Positive .15*** Masculinity .16*** RUAI (Log) Step1 0.07 14.41*** 5, 893 Age -.12** Non-white -.02 College Graduate .02 Bisexual Identity .05 HIV Positive .26*** Step2 0.09 14.90*** 6, 892 Age -.10 Non-white -.02 College Graduate .03 Bisexual Identity .03 HIV Positive .26*** Masculinity .13*** *p<.05 **p<.001 ***p<.0001 Table 6. Regression analysis for the prediction of the frequency of unprotected anal intercourse within the previous six months.
56 RUAI Similar to IUAI, age and HIV positive serostatus were among the two significant control variables related to RUAI. HIV serostatus was the strongest contributor (" = .26, p<.0001) to the prediction of RUAI. The addition of MI to this model increased the R 2 value to .09 and was statistically significant F (6, 892) = 14.90, p<.0001. A comparison of the Betas indicate that MI was the second strongest predictor (" = .13, p<.0001) in the model and independently contributed 2% of the explained v ariance. Men who endorsed more traditional attitudes about masculinity were more likely to frequently engage in receptive unprotected anal sex. Hypothesis 3 In contrast to the behavioral outcomes previously tested, the b arebacking intentions outcome purp orted to measure the degree to which men i ntentionally seek out unprotected anal intercourse. It was hypothesized that MI would be positively associated with greater intentions to bareback as ind icated by higher scores on the barebacking measure Of the co ntrol variables, e ducation, bisexual identity, and HIV positive serostatus were among the significant predictors associated with barebacking intentions. Unlike the behavioral outcomes, those who graduated college were more likely to seek out barebacking se xual encounters compared to men who did not complete college (" = .13, p<.0001). Also, unlike the behavioral outcomes, bisexual identity was statistically associated with barebacking intentions (" = .06, p<.05); however, the small size of the Beta and the higher alpha level warrant caution in interpreting this findi ng due to the large sample size and number of hypothesis tests. By far, the greatest contributor to the
57 model was HIV positive serostatus (" = .30, p<.0001) indicating the HIV positive men were more likely to seek out unprotected sexual encounters. The add ition of MI to the model increased the R 2 value to .17 and was statistically significant, F (6, 895) = 30.1, p<.0001). MI explained 5% of the total explained variance in the model. Men who endorsed more traditional attitudes toward masculinity were more li kely report intentions for barebacking. See Table 7 for a complete listing of Betas and significance levels. Hypothesis 4 In order to explore causal pathways that may help to explain the relationship between MI and high risk sexual behavior, two behaviora l variables were tested as potential mediators. Based on theoretical analysis and previous research, it was hypothesized that the use of the Internet to meet sex partners and the use of drugs within sexual contexts may be proximate causes that mediate the associations between MI and risky sexua l behaviors. A series of logistic regression models were constructed in order to test the mediation hypothes e s. These hypothese s were tested using the procedures described by Baron and Kenny (1986) and summarized in t he data analysis section in Chapter 3. For this analysis, risky sexual behavior is operationalized as insertive or receptive anal intercourse without the use of a condom. In the first step, meeting sexual partners via the Internet was regressed on the set of control variables and the masculinity variable. This step was repeated with using the drug use variable as the outcome. MI was not associated with using the Internet to meet sexual partners (OR = 1.08, 95% CI: 0.90, 1.3) or the use of drugs before or du ring sex (OR = 1.19, 95% CI: 0.96, 1.5); therefore th e first criteria for mediation that the independent and mediator variables are
58 correlated was not satisfied. N o further mediation analyses were conducted. There was no support for hypothesis 4 The ad dition of the drug and Internet use variables, however, increased the amount of variance explained in each outcome measure. The percentage increase ranged from two to twelve percent, with the largest increase in explained variance in the number of receptiv e partners and the smallest for barebacking intentions. These results highlight the importance of these two variables in explaining the variation in sexual risk behaviors, above and beyond the explanatory power of masculinity ideology and the block of cont rol variables. See Table 8 for the results of these analyses. Explanatory Variables R 2 F df Step1 0.12 23.38*** 5, 896 Age -.01 Non-white -.06 College Graduate .12*** Bisexual Identity .10* HIV Positive .31*** Step2 0.17 30.01*** 6, 895 Age .03 Non-white -.06 College Graduate .13*** Bisexual Identity .06* HIV Positive .30*** Masculinity .23*** *p<.05 **p<.001 ***p<.0001 Table 7. Regression analysis for the prediction of barebacking intentions.
59 Explanatory Variables R 2 F df Step1 0.12 23.38*** 5, 896 Age -.01 Non-white -.06 College Graduate .12*** Bisexual Identity .10* HIV Positive .31*** Step2 0.17 30.01*** 6, 895 Age .03 Non-white -.06 College Graduate .13*** Bisexual Identity .06* HIV Positive .30*** Masculinity .23*** *p<.05 **p<.001 ***p<.0001 Table 7. Regression analysis for the prediction of barebacking intentions. Explanatory Variables R 2 2 ) F df No. Insertive Partners (Log) .14 ('.08) 16.71*** 8, 795 Control Variables + MASC Drug Use .17*** Internet Use .21*** No. Receptive Partners (Log) .23 ('.12) 29.38*** 8, 795 Control Variables + MASC Drug Use .27*** Internet Use .19*** IUAI (Log) .08 ('.03) 8.71*** 8, 795 Control Variables + MASC Drug Use .15*** Internet Use .06 RUAI (Log) .14 ('.05) 17.18*** 8, 795 Control Variables + MASC Drug Use .15*** Internet Use .16*** Barebacking Intentions (Log) .19 ('.02) 23.73*** 8, 795 Control Variables + MASC Drug Use .13*** Internet Use .00 Table 8. Regression analysis with drug and internet use variables. Note Values enclosed in parentheses represent the change in R 2 after internet and drug use variables were entered into the model. MASC=Masculinity Ideology Measure; IUAI = Insertive unprotected anal intercourse; RUAI=Receptive unprotected anal intercourse. *p<.05 **p<.001 ***p<.0001
60 CHAPTER 5 D ISCUSSION A growing body of research points to increasing incidence of HIV and other sexually transmitted infections among MSM. High risk sexual behaviors are responsible for t he vast majority of these new infections, charging health researchers with the task of better understanding the determinants of these behaviors. Fundamental to this understanding is the acknowledgment that sexual behaviors are not the result of simple proc esses along a linear chain of causes, but rather complex, socially embedded practices involving individual, intrapersonal and institutional forces. This research sought to explore the impact of gender as a social institution, which structures sexual pract ices by communicating norms and expectations for how men "should" behave, think and feel. In line with a social constructionist framework, it was theorized that men who endorsed hegemonic or normative m asculinity to a greater degree would be more likely t o exhibit risk behavior patterns consistent with those norms. In this study, sexual risk taking was defined as the number of anal sex partners (both insertive and receptive), the frequency of unprotected anal intercourse (both insertive and receptive), and the intentions to seek out bareback sexual encounters. It was hypothesized that masculinity ideology a measurement of the degree to which an individual endorses normative aspects of masculinity would be positively associated with sexual behavior. Addition ally, it was hypothesized that masculinity ideology would independently explain a percentage of the variance in these behaviors, ab ove that accounted for by a set of control variables commonly found to be associated with sexual behavior. This study provide s empirical
61 support for these hypotheses. Masculinity ideology (MI) was positively associated with the number of anal sex partners, the frequency of unprotected anal sex, and the intentions to seek out and have bareback or unprotected sex. As hypothesized, on average men who endorsed normative masculinity to a greater degree were also more likely to have more sexual partners, have unprotected sex more frequently, and were more likely to intentionally seek out unprotected sex. The degree of variance explaine d for each outcome variable was small, ranging from 2% to 5%; however, these results suggest that MI does play a role in the social and behavioral determinants of high risk sexual behaviors among MSM. For insertive sex, MI was the strongest predictor for n umber of partners and frequency of unprotected sex surpassing the explanatory power of HIV serostatus. The strongest predictor of receptive sexual behaviors was HIV positive serostatus, followed by MI. Insertive anal sex is widely acknowledged to be more c onducive to transmitting HIV. In an effort to reduce the risk of transmission, HIV positive men will position themselves as the receptive partner commonly referred to as strategic positioning. The relatively strong association between HIV serostatus and re ceptive sexual behaviors lends further evidence for the prevalence of this harm reduction strategy. Similar to the observed associations among the sexual behavioral variables, MI was positively associated with barebacking intentions. In other words, men w ho endorsed normative standards of masculinity were also more likely on average to report greater intentions for unprotected anal intercourse (barebacking). Recent research has suggested that barebacking has evolved into a social identity ("Barebackers") f ormed in opposition
62 to characteristics and behaviors widely associated with the gay community (Grov et al., 2007; Parsons and Bimbi, 2006). The observed association between MI and barebacking intentions is given greater meaning when understood within this context of transgressive identity construction. Conceptualizing barebacking as an identity formed in opposition to the ubiquitous representations of gay men as effeminate is consistent with a constructionist theory of gender and masculinity. Men who violat e social proscriptions against homosexual behavior lack access to "legitimate" interpersonal and societal practices most notably, heterosexual relations that provide meaning and context for men to construct masculinity. This type of marginalization, Connel l theorizes, can result in the construction of a "protest masculinity," exaggerating hegemonic norms such as sexual prowess and risk taking (2005). This reasoning underscores the theoretical thinking behind this study and is supported by the empirical obse rvations summarized above. Furthermore, these findings are consistent with the limited number of studies that have reported comparable relationships between masculinity and sexual behaviors (Shearer, 2005; Jarama, Kennamer, Poppen, Hendricks, & Bradford, 2005; Halkitis, Green, and Wilton, 2004; Pleck & O'Donnell, 2001, Pleck, Sonenstein & Ku, 1993), as well as those that have found indirect relationships with factors such as attitudes towards condoms, condom use, and partner communication mediating the rel ationship (Shearer, 2005; Noar & Morokoff, 2002). Collectively, this body of research provides sufficient argument for the serious consideration of gender ideologies, specifically those relating to masculinity and manhood, in the evaluation of sexual risk behaviors. At the very least,
63 these works can be cited as empirical testimony for future research seeking to elucidate these relationships. This type of research should seek to discern cognitive, emotional, and/or behavioral factors that mediate the relati onship between gender ideologies and high risk sexual behaviors. Attitudes toward condom use, communications regarding safer sex, as well as issues surrounding internalized homophobia are germane psychosocial factors that should be investigated further. Fu ture research should also explore interaction effects of masculinity across social groups including race/ethnicity, social class, sexual identity and HIV status. While these research endeavors would expand upon current knowledge and strengthen explanatory models, quantitative analyses are ill equipped to fully capture the complexity of gender as a multidimensional, dynamic construct. In depth qualitative research designs may be most suitable at uncovering the process by which masculinity is defined and cons tructed within the context of sexual practices. Data from these types of studies could prove invaluable to translating this information into effective public health responses. The results of the current study should be considered hypothesis generating in the sense that they do not yield immediate points for intervention. The primary implication for public health research and practice is to move beyond a categorical treatment of gender. Sex as in the distinction between biological males and females is freq uently described as a distal, unchangeable risk factor. However, gender as a socially constructed factor is dynamic and potentially amenable to change. The current findings suggest that cultural standards of masculinity may work against conventional safe s ex messages and sexual health education. Translating the findings of this and similar studies
64 into practical public health responses could be accomplished by directly challenging endorsement of harmful gender ideologies or targeting social and cultural ant ecedents that continue to legitimize such ideologies. One example of a "gender transgressive" intervention includes a classroom based prevention program for adolescents designed around traditional educational materials, but also addresses critical thinkin g skills pertaining to gender ideologies and sexual scripts (Laub, Somera, Gowen & Diaz, 1999). This type of approach could be easily tailored for gay youth within the context of gay/straight alliances or pride organizations found in many high schools and colleges. Another gender transgressive program titled "Stepping Stones" is an adult workshop developed in Africa that focuses on gender, sexual health, HIV/AIDS, gender violence, communication and relationship skills. The program deals directly with implic it gender ideologies and interpersonal dynamics as a part of a comprehensive workshop. These two programs serve as promising examples of the ways in which issues of femininity and masculinity can be incorporated within comprehensive educational programs. Theoretical and empirical information presented in this study also points to the exclusion of sexual minorities from social institutions (e.g. marriage) as possible targets for policy level interventions. Opening access to such institutions may decrease se xual risk taking by providing sanctioned means of constructing masculinity within contexts that constrain certain risky sexual practices. The recent legal sanctioning of marriage and civil unions for same sex couples in certain states provides a unique opp ortunity for researchers to study the impact of such changes to the sexual behavior of subpopulations of MSM.
65 Findings from this study should be evaluated against the limitations in the design and methodology. A cross sectional survey was used to collect the data precluding any causative claims. The models were tested as if MI was the cause of high risk sexual behaviors, but it is equally possible that MI is a result of certain sexual practices, or perhaps they are both outcomes and causes of each other. I t is impossible to determine these temporal relationships given the study design. Courtenay conducted a longitudinal study in which he reported on the temporal relationships between a measure of MI and an aggregate measure of health risk behaviors (1998). He found that beliefs about manhood emerged as the strongest predictor of risk taking behavioral style after a three year follow up. Interestingly, he also found that beliefs about manhood changed over these three years and were in turn influenced by behav ioral changes in risk behavior, suggesting an interactive relationship. Other limitations of this study involve the measurement of the variables. Sexual behaviors were assessed over the previous six months and may be subject to recall bias. In addition, this analysis was restricted to single men at the time of participation. Relationship status over the previous six months in which sexual behaviors were assessed was not taken into consideration; therefore, some occurrences of unprotected sexual encounters may have occurred within the context of a monogamous relationship. Similarly, the HIV status of partners was not considered, prohibiting the analysis of harm reduction techniques, such as serosorting. Concerns regarding the measure of masculinity ideology are also worth mentioning. Despite the high degree of internal consistency for the masculinity scale, the
66 construct validity for the subscales was not confirmed. This held true even when the additional items were removed, and the analysis was conducted on ly on the original MASC attitude items. Furthermore, there has been disagreement over the number of dimensions scales of this nature are actually measuring (Luyt, 2005). Parallel analysis was used in this study to determine the number of latent factors. Th is method is believed to decrease bias in the estimates and produce more valid findings compared to those produced by other more commonly used methods. Perhaps this technique is what differentiates this study from others who have empirically confirmed mult iple dimensions of masculinity. Due to the ambiguity in the literature as well as a lack of empirical support, analyses were not conducted using the subscales individually. Future research should address the psychometric properties of the MASC scale as wel l as other, more widely used scales purporting to measure multiple dimensions of masculinity ideology. In addition, the four behavioral outcome variables were highly skewed. It is possible that the distribution of sexual behaviors for this population of m en is non normal and highly variable. Rather than truncate the data or use non parametric tests, which are still affected by extreme outliers, the distributions were corrected using logarithmic transformations. These transformations resulted in vastly impr oved distributions, though still positively skewed. Furthermore, eliminating the most extreme outliers (those with z scores in excess of 3.29) only produced trivial changes in the regression estimates. Finally, the sample was highly educated, mostly gay i dentified and white/Caucasian preventing generalizations to a diverse population of MSM. The majority of the sample was also recruited from weblogs and online networking sites and
67 therefore may be more representative of established Internet users. Despite these limitations, this study makes a unique contribution to the literature on the association of gender norms and sexual behaviors. This is the first study to explore these associations with a sample of gay, bisexual, and other MSM adding further credibil ity to previous findings, which were limited by samples consisting mostly of heterosexual adolescents and college students. It also calls attention to the importance of the Internet in the sexual lives of many MSM. This is important for future research as well as an area for program development. Conclusion This research utilized theoretical and empirical work from a variety of social science disciplines as well as critical studies in cluding feminism It was theoretically driven and informed by these works in all phases of the research process: beginning with the initial evaluation o f a current public health issue, followed by the genesis of the study hypothes es, design, and methodology and finally to the interpretation of findings. Theoretically based res earch is a vital component of public health that is often overlooked, evidenced by the volumes of research some have referred to a "risk factorology" (McKinlay & Marceau, 2000). Understanding human sexual behavior the ultimate driving force behind the HIV /AIDS epidemic requires pubic health researchers to move beyond conventional predictors of risk. We know that men in the United States, particularly MSM, are disproportionately at risk for HIV infection. In this sense, male gender is a risk factor for HIV infection. Ultimately, this observation is meaningless unless it is understood within a theoretical context that explains what gender actually is
68 and what it is about the male gender or more specifically, masculinity that is fundamental to the observed dis parities in HIV infections. Common sense tells us that MSM are more likely to become infected with HIV/AIDS because of their risky sexual practices (i.e. anal sex), multiple sex partners, and the failure to use condoms. But what is it about male gender tha t predisposes men to these types of behaviors and why do some men engage in them to a greater degree than others? These questions were fundamental to this investigation. The answers provided here informed by the empirical data point in part to the social a nd cultural norms of manhood and masculinity, and the degree to which individuals endorse these norms.
69 REFERENCES Baron, R. M., & Kenny, D. A. (1986). The moderator mediator variable distinction in social psychological research: conceptual, strateg ic, and statistical considerations. J Pers Soc Psychol, 51 (6), 1173 1182. Bartholow, B. N., Goli, V., Ackers, M., McLellan, E., Gurwith, M., Durham, M., et al. (2006). Demographic and behavioral contextual risk groups among men who have sex with men partic ipating in a phase 3 HIV vaccine efficacy trial: implications for HIV prevention and behavioral/biomedical intervention trials. J Acquir Immune Defic Syndr, 43 (5), 594 602. Benotsch, E. G., Kalichman, S., & Cage, M. (2002). Men who have met sex partners vi a the Internet: prevalence, predictors, and implications for HIV prevention. Arch Sex Behav, 31 (2), 177 183. Boily, M. C., Godin, G., Hogben, M., Sherr, L., & Bastos, F. I. (2005). The impact of the transmission dynamics of the HIV/AIDS epidemic on sexual behaviour: a new hypothesis to explain recent increases in risk taking behaviour among men who have sex with men. Med Hypotheses, 65 (2), 215 226. Bolding, G., Davis, M., Hart, G., Sherr, L., & Elford, J. (2005). Gay men who look for sex on the Internet: is there more HIV/STI risk with online partners? Aids, 19 (9), 961 968. Bull, S. S., Lloyd, L., Rietmeijer, C., & McFarlane, M. (2004). Recruitment and retention of an online sample for an HIV prevention intervention targeting men who have
70 sex with men: the S mart Sex Quest Project. AIDS Care, 16 (8), 931 943. Bull, S. S., & McFarlane, M. (2000). Soliciting sex on the Internet: what are the risks for sexually transmitted diseases and HIV? Sex Transm Dis, 27 (9), 545 550. Bull, S. S., McFarlane, M., & King, D. (20 01). Barriers to STD/HIV prevention on the Internet. Health Educ Res, 16 (6), 661 670. Bull, S. S., McFarlane, M., Lloyd, L., & Rietmeijer, C. (2004). The process of seeking sex partners online and implications for STD/HIV prevention. AIDS Care, 16 (8), 1012 1020. Bull, S. S., McFarlane, M., & Rietmeijer, C. (2001). HIV and sexually transmitted infection risk behaviors among men seeking sex with men on line. Am J Public Health, 91 (6), 988 989. Catania, J. A., Gibson, D. R., Chitwood, D. D., & Coates, T. J. (1 990). Methodological problems in AIDS behavioral research: influences on measurement error and participation bias in studies of sexual behavior. Psychol Bull, 108 (3), 339 362. CDC. HIV/AIDS Surveillance Report, 2005. [Electronic Version], Vol. 17. Retrieve d July 12, 2006 from http://www.cdc.gov/hiv/topics/surveillance/resources/reports CDC. (2003). Gonococcal Isolate Surveillance Project Annual Report [Electronic Version]. Retrie ved July 14, 2006 from http://www.cdc.gov/nchstp/dstd/Stats_and_Trends.htm CDC. (2005). HIV/AIDS among Men who have Sex with Men [fact sheet] [Electronic Version]. Retrieved July 22, 200 6 from http://www.cdc.gov/hiv CDC. (2007). HIV/AIDS among Men who have Sex with Men [fact sheet]. [Electronic
71 Version]. Retrieved October 19, 2007 from http://www.cdc.gov/hiv/topics/msm/resources/factsheets/msm.htm Connell, R. W. (2005). Masculinities (2 ed.). Berkeley: University of California Press. Courtenay, W. H. (1998). Better to die than cry? A longitudinal and constructionist stu dy of masculinity and the health risk behavior of young American men (Doctoral dissertation, University of California at Berkeley). Dissertation Abstracts International, 59 (08A), 232 pp. (Publication number AAT 9902042). Courtenay, W. H. (2000). Constructi ons of masculinity and their influence on men's well being: a theory of gender and health. Soc Sci Med, 50 (10), 1385 1401. Courtenay, W. H. (2000). Engendering Health: A Social Constructionist Examination of Men's Health Beliefs and Behaviors. Psychology o f Men & Masculinity, 1 (1), 4 15. Courtenay, W. H. (2003). Key Determinants of the Health and Well Being of Men and Boys. International Journal of Men's Health, 2 (1), 1 30. Dowsett, G. W. (2003). Some considerations on sexuality and gender in the context of AIDS. Reprod Health Matters, 11 (22), 21 29. Elford, J., Bolding, G., Davis, M., Sherr, L., & Hart, G. (2004). The Internet and HIV study: design and methods. BMC Public Health, 4 39. Executive Summary of the SF AIDS Foundation's Qualitative Interview Stu dy [Electronic (1997). Version]. Retrieved May 1, 2006 from http://www.sfaf.org/prevention/execsumm.html Fausto Sterling, A. (2000). Sexing the body: gender polo tics and the construction of
72 se xuality (1 ed.). New York, NY: Basic Books. Frankel, M. S., & Siang, S. (1999). Ethical and legal aspects of human subjects research on the Internet [Electronic Version]. Washington, DC: American Association for the Advancement of Science. Retrieved Decemb er 1, 2006 from http://www.aaas.org/spp/dspp/sfrl/projects/intres/main.htm Gendin, S. (1997, May 5, 1997). Riding Bareback. POZ: Health, Hope, and HIV, 64 65. Green, A. I. (2003). "chem friendly": the institutional basis of "club drug" use in a sample of urban gay men. Deviant Behavior, 24 (5), 427 447. Green, A. I., & Halkitis, P. N. (2006). Crystal methamphetamine and sexual sociality in an urban gay subculture: an elective affini ty. Cult Health Sex, 8 (4), 317 333. Grov, C., DeBusk, J. A., Bimbi, D. S., Golub, S. A., Nanin, J. E., & Parsons, J. T. (2007). Barebacking, the Internet, and harm reduction: an intercept survey with gay and bisexual men in Los Angeles and New York City. A IDS Behav, 11 (4), 527 536. Guenther Grey, C. A., Varnell, S., Weiser, J. I., Mathy, R. M., O'Donnell, L., Stueve, A., et al. (2005). Trends in sexual risk taking among urban young men who have sex with men, 1999 2002. J Natl Med Assoc, 97 (7 Suppl), 38S 43S Halkitis, P., Wilton, L., & Galatowitsch, P. (2005). Gay and bisexual men's understanding of barebacking. Journal of Gay & Lesbian Psychotherapy, 9 (3/4), 35 48. Halkitis, P. N., Green, K. A., & Wilton, L. (2004). Masculinity, Body Image, and Sexual Behav ior in HIV Seropositive Gay Men: A Two Phase Formative Behavioral Investigation Using the Internet. International Journal of Men's Health, 3 (1), 27
73 42. Halkitis, P. N., & Parsons, J. T. (2003). Intentional unsafe sex (barebacking) among HIV positive gay me n who seek sexual partners on the internet. AIDS Care, 15 (3), 367 378. Halkitis, P. N., Parsons, J. T., & Wilton, L. (2003). Barebacking among gay and bisexual men in New York City: explanations for the emergence of intentional unsafe behavior. Arch Sex Be hav, 32 (4), 351 357. Halkitis, P. N., Wilton, L., Wolitski, R. J., Parsons, J. T., Hoff, C. C., & Bimbi, D. S. (2005). Barebacking identity among HIV positive gay and bisexual men: demographic, psychological, and behavioral correlates. Aids, 19 Suppl 1 S2 7 35. Hayton, J. C., Allen, D. G., & Scarpello, V. (2004). Factor retention decisions in exploratory factor analysis: A tutorial on parallel analysis. Organizational Research Methods, 7 (2), 191 205. Hirshfield, S., Remien, R. H., Walavalkar, I., & Chiasson M. A. (2004). Crystal methamphetamine use predicts incident STD infection among men who have sex with men recruited online: a nested case control study. J Med Internet Res, 6 (4), e41. Huebner, D. M., & Gerend, M. A. (2001). The relation between beliefs a bout drug treatments for HIV and sexual risk behavior in gay and bisexual men. Ann Behav Med, 23 (4), 304 312. Huebner, D. M., Rebchook, G. M., & Kegeles, S. M. (2004). A longitudinal study of the association between treatment optimism and sexual risk behav ior in young adult
74 gay and bisexual men. J Acquir Immune Defic Syndr, 37 (4), 1514 1519. Jarama, S. L., Kennamer, J. D., Poppen, P. J., Hendricks, M., & Bradford, J. (2005). Psychosocial, behavioral, and cultural predictors of sexual risk for HIV infection among Latino men who have sex with men. AIDS Behav, 9 (4), 513 523. Kalichman, S. C., Heckman, T., & Kelly, J. A. (1996). Sensation seeking as an explanation for the association between substance use and HIV related risky sexual behavior. Arch Sex Behav, 25 (2), 141 154. Kalichman, S. C., Nachimson, D., Cherry, C., & Williams, E. (1998). AIDS treatment advances and behavioral prevention setbacks: preliminary assessment of reduced perceived threat of HIV AIDS. Health Psychol, 17 (6), 546 550. Kauth, M. R., St L awrence, J. S., & Kelly, J. A. (1991). Reliability of retrospective assessments of sexual HIV risk behavior: a comparison of biweekly, three month, and twelve month self reports. AIDS Educ Prev, 3 (3), 207 214. Kim, A. A., Kent, C., McFarland, W., & Klausne r, J. D. (2001). Cruising on the Internet highway. J Acquir Immune Defic Syndr, 28 (1), 89 93. Kippax, S., & Race, K. (2003). Sustaining safe practice: twenty years on. Soc Sci Med, 57 (1), 1 12. Klausner, J. D., Wolf, W., Fischer Ponce, L., Zolt, I., & Katz M. H. (2000). Tracing a syphilis outbreak through cyberspace. Jama, 284 (4), 447 449. Knaak, S. (2004). On the Reconceptualizing of Gender: Implications for Research Design. Sociological Inquiry, 74 (3), 302 317. Laub, C., Somera, D. M., Gowen, L. K., & Di az, R. M. (1999). Targeting "risky" gender
75 ideologies: constructing a community driven, theory based HIV prevention intervention for youth. Health Educ Behav, 26 (2), 185 199. Leigh, B. C., & Stall, R. (1993). Substance use and risky sexual behavior for exp osure to HIV. Issues in methodology, interpretation, and prevention. Am Psychol, 48 (10), 1035 1045. Levant, R. F., Hirsch, L. S., Celentano, E., Cozza, T. M., & et al. (1992). The male role: An investigation of contemporary norms. Journal of Mental Health Counseling 14 (3), 325 337. Levine, M. P., & Kimmel, M. S. (1998). Gay Macho: the life and death of the homosexual clone New York: New York University Press. Luyt, R. (2005). The Male Attitude Norms Inventory II: A Measure of Masculinity Ideology in South Africa. Men & Masculinities, 8 (2), 208 229. Mann, C., & Stewart, F. (2000). Internet Communication and Qualitative Research: a handbook for researching online London: Sage Publications. Mansergh, G., Marks, G., Colfax, G. N., Guzman, R., Rader, M., & Buch binder, S. (2002). "Barebacking" in a diverse sample of men who have sex with men. Aids, 16 (4), 653 659. Mansergh, G., Shouse, R. L., Marks, G., Guzman, R., Rader, M., Buchbinder, S., et al. (2006). Methamphetamine and sildenafil (Viagra) use are linked to unprotected receptive and insertive anal sex, respectively, in a sample of men who have sex with men. Sex Transm Infect, 82 (2), 131 134. McFarlane, M., Bull, S. S., & Rietmeijer, C. A. (2000). The Internet as a newly emerging
76 risk environment for sexually transmitted diseases. Jama, 284 (4), 443 446. McKinlay, J. B., & Marceau, L. D. (2000). To boldly go. Am J Public Health, 90 (1), 25 33. McKirnan, D. J., Vanable, P. A., Ostrow, D. G., & Hope, B. (2001). Expectancies of sexual "escape" and sexual risk among drug and alcohol involved gay and bisexual men. J Subst Abuse, 13 (1 2), 137 154. Nabavi, R. (2004). The "masculinity attitudes, stress, and conformity questionnaire (MASC)": A new measure for studying psychology of men Dissertation Abstracts Internationa l: Section B: The Sciences and Engineering. Noar, S. M., & Morokoff, P. J. (2002). The Relationship Between Masculinity Ideology, Condom Attitudes, and Condom Use Stage of Change: A Structural Equation Modeling Approach. International Journal of Men's Heal th, 1 (1), 43 58. O'Neil, J. M., Good, G. E., & Holmes, S. (1995). Fifteen years of theory and research on men's gender role conflict: New paradigms for empirical research. In A new psychology of men (pp. 164 206). New York, NY: Basic Books. Parsons, J. T., & Bimbi, D. S. (2007). Intentional unprotected anal intercourse among sex who have sex with men: barebacking from behavior to identity. AIDS Behav, 11 (2), 277 287. Parsons, J. T., Kutnick, A. H., Halkitis, P. N., Punzalan, J. C., & Carbonari, J. P. (200 5). Sexual risk behaviors and substance use among alcohol abusing HIV positive men who have sex with men. J Psychoactive Drugs, 37 (1), 27 36. Pathela, P., Hajat, A., Schillinger, J., Blank, S., Sell, R., & Mostashari, F. (2006).
77 Discordance between sexual behavior and self reported sexual identity: a population based survey of New York City men. Ann Intern Med, 145 (6), 416 425. Pinkerton, S. D., Benotsch, E. G., & Mikytuck, J. (2007). When Do Simpler Sexual Behavior Data Collection Techniques Suffice? Evalu ation Review, 31 (4), 401 412. Pleck, J., Sonenstein, F., & Ku, L. (1993). Masculinity Ideology: Its Impact on Adolescent Males' Heterosexual Relationships. Journal of Social Issues, 49 (3), 11 29. Pleck, J. H., & O'Donnell, L. N. (2001). Gender attitudes an d health risk behaviors in urban African American and Latino early adolescents. Matern Child Health J, 5 (4), 265 272. Rhodes, S. D., Bowie, D. A., & Hergenrather, K. C. (2003). Collecting behavioral data using the World Wide Web: Considerations for researc hers. Journal of Epidemiology and Community Health, 57 (1), 68 73. Rhodes, S. D., DiClemente, R. J., Cecil, H., Hergenrather, K. C., & Yee, L. J. (2002). Risk among men who have sex with men in the United States: a comparison of an Internet sample and a con ventional outreach sample. AIDS Educ Prev, 14 (1), 41 50. Ridge, D. T. (2004). 'It was an Incredible Thrill': The Social Meanings and Dynamics of Younger Gay Men's Experiences of Barebacking in Melbourne. Sexualities, 7 (3), 259 279.
78 Riggle, E. D., Rostosky, S. S., & Reedy, C. S. (2005). Online surveys for BGLT research: issues and techniques. J Homosex, 49 (2), 1 21. Roth, P. L., Switzer, F. S., & Switzer, D. (1999). Missing data in multi item scales: A Monte Carlo Analysis of missing data techniques. Organiz ational Research Methods, 2 211 232. Sabo, D. F., & Gordon, D. F. (1995). Men's health and illness: Gender, power, and the body Thousand Oaks, CA: Sage Publications, Inc. Schofield, T., Connell, R. W., Walker, L., Wood, J. F., & Butland, D. L. (2000). Un derstanding men's health and illness: a gender relations approach to policy, research, and practice. J Am Coll Health, 48 (6), 247 256. Shearer, C. L., Hosterman, S. J., Gillen, M. M., & Lefkowitz, E. S. (2005). Are Traditional Gender Role Attitudes Associa ted With Risky Sexual Behavior and Condom related Beliefs? Sex Roles, 52 (5 6), 311 324. Smiler, A. P. (2004). Thirty Years After the Discovery of Gender: Psychological Concepts and Measures of Masculinity1. Sex Roles, 50 (1 2), 15 26. Snell, W., Belk, S., & Hawkins, R. (1986). The Stereotypes About Male Sexuality Scale (SAMSS): Components, correlates, antecedents, consequences and counselor bias. Social and Behavioral Science Documents, 16 (1), 9. (Ms. No. 2746). Stall, R., Paul, J. P., Greenwood, G., Pollack L. M., Bein, E., Crosby, G. M., et al. (2001). Alcohol use, drug use and alcohol related problems among men who have sex with men: the Urban Men's Health Study. Addiction, 96 (11), 1589 1601. Stolte, I. G., & Coutinho, R. A. (2002). Risk behaviour and sex ually transmitted diseases
79 are on the rise in gay men, but what is happening with HIV? Curr Opin Infect Dis, 15 (1), 37 41. Strathdee, S. A., Martindale, S. L., Cornelisse, P. G., Miller, M. L., Craib, K. J., Schechter, M. T., et al. (2000). HIV infection a nd risk behaviours among young gay and bisexual men in Vancouver. Cmaj, 162 (1), 21 25. Studies, C. f. A. P. (2006). University of California San Francisco. How does methamphetamine use affect HIV prevention? [Electronic Version]. Retrieved October 14, 2006 from http://www.caps.ucsf.edu Suarez, T., & Miller, J. (2001). Negotiating risks in context: a perspective on unprotected anal intercourse and barebacking among men who have sex with men -where do we go from here ? Arch Sex Behav, 30 (3), 287 300. Tabachnick, B. G., & Fidell, L. S. (2007). Using Multivariate Statistics (5 ed.). Boston: Allyn and Bacon. Tabet, S. R., Krone, M. R., Paradise, M. A., Corey, L., Stamm, W. E., & Celum, C. L. (1998). Incidence of HIV and s exually transmitted diseases (STD) in a cohort of HIV negative men who have sex with men (MSM). Aids, 12 (15), 2041 2048. Taywaditep, K. J. (2001). Marginalization among the marginalized: gay men's anti effeminacy attitudes. J Homosex, 42 (1), 1 28. Thompson E. H., Jr., & Pleck, J. H. (1995). Masculinity ideologies: A review of research instrumentation on men and masculinities. In A new psychology of men (pp. 129 163). New York, NY: Basic Books. Thompson, E. H., & Pleck, J. H. (1986). The structure of male r ole norms. American
80 Behavioral Scientist, 29 (5), 531 543. Thompson, E. H., Pleck, J. H., & Ferrera, D. L. (1992). Men and masculinities: Scales for masculinity ideology and masculinity related constructs. Sex Roles, 27 (11 12), 573 607. van der Snoek, E., d e Wit, J., Gotz, H., Mulder, P., van, T. H. A., Verkooyen, R., et al. (2004). Demographics, sexual behaviour and STD/HIV prevalence in two groups of men who have sex with men, in Rotterdam, The Netherlands. Acta Derm Venereol, 84 (2), 145 150. van der Snoek E. M., de Wit, J. B., Mulder, P. G., & van der Meijden, W. I. (2005). Incidence of sexually transmitted diseases and HIV infection related to perceived HIV/AIDS threat since highly active antiretroviral therapy availability in men who have sex with men. Sex Transm Dis, 32 (3), 170 175. Velicer, W. F., Eaton, C. A., & Fava, J. L. (2000). Construct explication through factor or component analysis: A review and evaluation of alternative procedures for determining the number of factors or components. In D. N. Jackson, R. G. Goffin & E. Helmes (Eds.), Problems and solutions in human assessment: Honoring Douglas N. Jackson at Seventy (pp. 41 71). Boston: Kluwer Academic Publishers. Webster, R. D., Darrow, W. W., Paul, J. P., Roark, R. A., Woods, W. J., & Stempel, R. R. (2003). HIV infection and associated risks among young men who have sex with men in a Florida resort community. J Acquir Immune Defic Syndr, 33 (2), 223 231. West, C., & Zimmerman, D. H. (1998). Doing gender. In Feminist foundations: Toward
81 transform ing sociology (pp. 167 190). Thousand Oaks, CA: Sage Publications, Inc. Wingood, G. M., Scd, & DiClemente, R. J. (2000). Application of the theory of gender and power to examine HIV related exposures, risk factors, and effective interventions for women. He alth Educ Behav, 27 (5), 539 565. Wolitski, R. J., Valdiserri, R. O., Denning, P. H., & Levine, W. C. (2001). Are we headed for a resurgence of the HIV epidemic among men who have sex with men? Am J Public Health, 91 (6), 883 888. Wright, K. (2005). Research ing Internet based populations: Advantages and disadvantages of online survey research, online questionnaire authoring software packages, and web survey services. [Electronic Version], Journal of Computer Mediated Communication, article 11 from http://jcmc.indiana.edu/vol10/issue3/wright.html Yun, G. W., & Trumbo, C. W. (2000). Comparative response to a survey executed by post, e mail, & web form. Journal of Computer Mediated Communication, 6 (1). Zimmerman, D. W. (1994). A note on the influence of outliers on parametric and nonparametric tests. Journal of General Psychology, 121 391 401. Zimmerman, D. W. (1995). Increasing the power of nonparametric tests by detecting and downweighting outl iers. Journal of Experimental Education, 64 71 78.
82 Footnotes 1 This study focused on men who have sex with men (MSM) in order to be inclusive of all men who share the risks associated with certain sexual practices but who may not identify as gay or bi sexual. 2 The term "club drugs" is a dynamic category that includes recreational drugs which are associated with dance clubs, parties and raves.