Men Who Specifically Seek Unprotected Sex Partners via the Internet: Whose Profiles Are the Most Searched for by Other Site Users?

Abstract

Men who have sex with men (MSM) account for the largest number of persons diagnosed with AIDS in the United States, with higher than-average rates of drug use and unprotected sex being cited as the principal reasons underlying their high rates of HIV infection. Recent evidence has suggested that the use of Web sites specifically designed to promote unsafe sexual practices may be particularly common among MSM, therebyfostering their risky behaviots.

In light of these findings, the present study which compares HIV risk behaviors sought by men whose profiles are among the most frequently brought up in searches to those whose profiles ate not aspopular with other site users, is based on a content analysis of 1,316 ads/profiles posted on one of the most popular MSM Web sites that specifically fosters unprotected sex. Ads/profiles were selected randomly based on the American ZIP code of residence. Data were collected between September 2006 and January 2007.

Rates of advertised-for high-risk sexual behaviots were very high both for men whose profiles were among the most popular and those whose profiles were not, particularly for oral sex involving ejaculation into the mouth, anal sex involving ejaculation into the anus, multiple partner sex, and fetching. Analytical comparisons of these groups showed that the most searched for profiles weæ thosepromoting the highest levels ofrisky behaviors and the greatest propensity toward wanting to engage in high-risk activities. Ihe HIV intenention-related implications of these findings are discussed.

INTRODUCTION

To date, approximately 1 million Americans have been diagnosed with AIDS, and estimates suggest that nearly one-quarter million more are living with HIV that has not developed into AIDS (Centers for Disease Control and Prevention [CDC], 2006). Men who have sex with other men (MSM) comprise the largest proportion of these individuals, accounting for 57% of all reported cases of AIDS with a known source of transmission and 53% of all HIV-positive persons who believed that they knew how they became HIV infected (CDC, 2006).¹ Despite the so-called “changing face of AIDS in America”—that is, changes in the proportions of persons of various population subgroups that are more/less affected by HIV now than in previous years—these percentages for MSM have declined very little during the past 10 years (CDC, 1996, 2001).

In light of this, numerous studies have been conducted to identify why, nearly 30 years into the HIV/AIDS epidemic, so many men continue to place themselves at risk for contracting HIV. Many factors have been identified, including the belief that engaging in unprotected sex is an expression of individual choice (Adam, 2005; Carballo-Dieguez & Bauermeister, 2004), the belief that engaging in unprotected sex is an expression of masculinity (Halkitis, Green, & Wilton, 2004; Halkitis & Parsons, 2003; Ridge, 2004), the perception that AIDS antiretroviral drugs have made HIV/AIDS less of a health concern now than in prior years (Halkitis, Parsons, & Wilton, 2003; Sheon & Crosby, 2004), a fear of being rejected sexually by partners who dislike condoms (Sheon & Crosby, 2004), the belief that sex is more pleasurable when condoms are not used (Carballo-Dieguez & Bauermeister, 2004; Dilley et al., 2002; Mansergh et al., 2002), feeling “burned out” by worrying about becoming HIV infected (Dilley et al., 2002; Halkitis et al., 2003), and feeling a greater sense of emotional connectedness to sexual partners with whom one had unprotected rather than protected sex (Mansergh et al., 2002; Theodore, Duran, Antoni, & Fernandez, 2004).

How men who wish to have high-risk sex with other men locate potential sex partners has been the subject of relatively little research, however. For many men, “traditional” avenues of meeting other men (e.g., gay bars, gay/bisexual-oriented social activities, personal ads) remain popular ways of meeting potential sex partners. But these “traditional” avenues often result in thwarted attempts at sexual encounters when, after investing the time in a getting-to-know-you process, the person who actively is seeking sex partners with whom he can have unprotected sex discovers that the person he has met is unwilling to engage in such behaviors. Another common way for men who wish to have unprotected sex with other men to locate potential partners is by frequenting public venues (e.g., parks, rest areas, restrooms) where male-to-male cruising is known to take place. These locales often offer MSM seeking unprotected sex the opportunities they seek to find willing partners, but they have inherent drawbacks as well. Police enforcement efforts often disrupt sex-seeking behaviors in known male-to-male cruising spots (and these men are often prosecuted), and it is not uncommon for non-cruising people to show up in these public venues to utilize them for their intended purposes (e.g., using a restroom facility, taking a walk in the woods).

In recent years, with the proliferation of the Intemet, many men who wish to find other men specifically for engaging in unprotected sex appear to be tuming to MSM-oriented Web sites for this purpose. For example, in a sample of gay men who were recruited into a health promotion study via gay-oriented Internet Web sites, Bolding and colleagues’ (Bolding, Davis, Sherr, Hart, & Elford, 2004) multivariate analysis revealed that the amount of risky sex in which men engaged was a significant predictor of their use of Internet Web sites to locate sex partners. Bolding and colleagues also reported that 47% of the men in their sample said that when they wanted to identify potential sex partners, they preferred using Web sites to frequenting bars or other “offline” venues. In another study (Bull, MacFarlane, Lloyd, & Rietmeijer, 2004), among men actively using the Intemet as a means of locating potential sex partners, 97% reported actually having met someone online for sex, and 86% said that they used Internet MSM sex sites at least once a week to identify possible partners. Halkitis, Parsons, and Wilton (2003) cited Internet Web sites and chat rooms as key sources that are partly responsible for the upsurge of unprotected sexual activities that they have observed among gay and bisexual men in the New York City area. Another study examining the role that Internet usage plays with regard to HIV risk taking found that persons who had a history of meeting sex partners via the Internet reported more frequent involvement in risky sexual behaviors than persons who had not met sex partners online (Mustanski, 2007).

With mounting evidence of the importance of the role that the Internet may play in fostering sexual encounters between men who specifically wish to have unprotected sex with other men, it becomes important to understand male-sexual-content Web site usage in this population. A variety of important questions emerge. First, what specific types of risky behaviors are these men seeking? Second, what beliefs and attitudes do they have vis-á-vis HIV and risk taking—beliefs and attitudes that presumably underlie their risk-taking behaviors? Third, what characteristics are being sought in potential partners? For example, are HIV-negative men frequenting these Web sites seeking to have sex only with other HIV-negative men? Are HIV-negative men who are using these male-sexual-content Web sites seeking overall lower levels of risky behaviors than HIV-positive men are? Also of specific interest and importance is the development of a good understanding of the types of characteristics that lead some profiles to be more searched for than others. For example, are the most popular profiles those posted by HIV-negative men, who might be using Web sites to locate presumably “safe” partners with whom they can engage in behaviors that are usually considered to be unsafe, or is this not the case? Are the most-searched-for profiles those expressing an interest in higher- or lower-than-average rates of risky behaviors? Obtaining the answers to all of these questions is crucial if one wishes to develop an informed prevention and/or intervention effort targeting the risk behaviors of men who use Internet Web sites to locate other men with whom they can engage in risky sex.

METHODS

This research relies upon content analysis as the principal analytical tool. The data were collected between September 2006 and January 2007 using one of the largest MSM-oriented Web sites currently available on the Intemet. The Web site was chosen because it is free to the public, findable by virtually any Internet search utilizing common key words like “bareback,” and because it boasts a large and steadily growing membership. At the time of data collection, the site had more than 145,000 registered users (the large majority of whom reside in the United States) and it was growing at a rate of several hundred persons per month. This Web site allows members to post profiles (including photographs) describing themselves, and there are no length restrictions placed on profiles posted.² In addition, members are instructed to indicate the type(s) of relationships they are seeking (long-term relationships, one-on-one sexual encounters, three-way sexual encounters, and so forth), specific sexual acts that they would like to practice with a willing partner, and a free-for-all field that can be used to provide supplemental information about one’s most-sought-after traits or behaviors. Essentially, the large, stable, and growing membership of this Web site, coupled with members’ ability to describe them selves as fully as they choose, made this particular Web site an ideal candidate for the present content analysis research.

The content analysis was based on a random sample of users’ profiles, randomly selected by ZIP code, which is a searchable feature on the site. Men residing outside of the United States were excluded from this research, so as to keep it an America-focused study. Also excluded from analysis (N = 6) were profiles that had not been filled out completely (i.e., with the user not providing at least one piece of the required information on each profile page on the Web site). [n order to be included in the analyses, a user’s profile had to remain active at the conclusion of the data collection period, to guard against “experimenters” or one-time-only visitors to the site being included in the study. This led to the exclusion of 67 cases (4.8%). In all, 1,316 profiles constitute the sample for this research.

Data Collected

For each profile, the following information was collected: age; race/ethnicity (Caucasian, African American, Latino, Asian, Native American, or biracial/multiracial); self-identification as being a “top,” a versatile top, versatile, a versatile “bottom,” or a bottom³ ; self-reported HIV serostatus (negative, positive, or unknown); desired HIV serostatus in sex partners (must be negative, may be negative, must be positive, may be positive, do not care); self-identified sexual orientation (gay, bisexual, “curious,” heterosexual); willingness to give and receive ejaculatory fluid in the mouth and anus; type(s) of “relationships” sought (one-on-one sexual encounter, long-term relationship, three-person sexual encounter, multiple partner sexual encounter, activities partner); the number and type of self-photographs posted by the user; the user’s ZIP code (which was also used to compute population density as a macro-level analytical variable); and whether or not the user had opted for an expanded, paid membership on the site.

In addition, data collection also entailed coding for a wide variety of specific sexual behaviors, including among others felching (eating ejaculatory fluid that has been inserted into one person’s anus and then feeding it back to that individual by mouth, usually with a kiss), rimming (oral stimulation of the anus), bukkake (ejaculating directly onto another person’s mouth and face), and double penetration (forcing two penises into the same anus simultaneously). Finally, a variety of risk-enhancing practices and attitudes were also coded, including a stated preference for engaging in rough sex, having sexual relations while high (known in the target community as PNP, or “partying and playing”), ovenly stating that they will not use condoms and/or that they will not permit their partners to use condoms, actively trying to become HIV infected (known in the target community as “bug chasing”), actively trying to infect partners with HIV (known as “gift giving”), refusing to withdraw the penis prior to ejaculation and/or refusing to allow a sex partner to withdraw his penis prior to ejaculation, an overt preference for anonymous sex (i.e., sexual encounters in which the name and/or face of the sexual partner(s) is/are unknown), a stated preference for having long-lasting sexual encounters, an expression of seeking sexual encounters that are “uninhibited” or “no holds barred,” and eroticizing ejaculation fluid (known in the target community as being a “cum whore” or a “cum freak” or a “cum lover”).

The principal independent variable used in these analyses is a dichotomous measure indicating whether the user’s profile was one of the most commonly searched profiles on the site on any day during the data collection period. Each day, the site gathers statistics and reports on its home page the 20 members whose profiles were the most searched for on the site the previous day. These “most active profiles” (altemately referred to in this article as “most searched for” and “most popular”) are posted in a prominent place on the Web site, encouraging site users to visit these particular profiles and see for themselves what makes these profiles engaging for other site users.

Research Questions and Analysis

Research Question #1 is “Are there differences in the demographic/ descriptive characteristics of the persons posting profiles that were among the most popular versus those that were not among the most popular?” Analytically, this can be examined with chi-square tests, since the dependent variables in question (e.g., race, HIV serostatus) are categorical in nature. In some instances, odds ratios are used in lieu of chi-square tests to report the findings, where the presentation of the data/findings in odds ratio format is more useful or more informative. Research Question *2 is “Are there differences in the specific risky sexual practices that are being sought by men whose profiles are the most searched for compared to those whose profiles are not?” This can be answered with Student’s t tests whenever the outcome measure is continuous in nature (e.g., overall sexual risk level), and with chi-square tests whenever the dependent variable is dichotomous or categorical (e.g., seeking anonymous sexual encounters, wanting a partner with whom to engage in receptive anal sex). In sorne instances, odds ratios are used in lieu of chi-square tests, where the presentation of the data/findings in odds ratio format is more useful or more informative. Research Question #3 is “Are there differences between the most-searched-for profiles and the profiles that were not alnong the most popular ones on the site in terms of the beliefs and attitudes expressed toward HIV and risk taking?” Most of the pertinent items here (e.g., unwillingness to withdraw the penis or to allow the partner to withdraw prior to ejaculation) were coded in yes/no format during the content analysis process. Therefore, chi-square tests are appropriate for examining differences in responding based on sexual orientation. Again, odds ratios are used instead of chi-square tests in many instances because the former provides more informative data/comparisons than the latter. Research Question #4 is “Are there differences between the types of characteristics that are being sought in potential sexual partners based on the user’s profile status as being or not being among the most-searched-for site profiles?” Again, chi-square tests (or in their alternative, odds ratios) are used for these analyses, as the dependent variables are categorical in nature (e.g., desired partners’ HIV serostatus). Research Question #5 is “What factors contribute uniquely and significantly to the prediction of whose profiles are the most searched for, when the effects of other factors are taken into account?” This question is addressed using multivariate logistic regression, since the dependent variable is dichotomous. The Hosmer-Lemeshow chisquare test is used to assess the multivariate model’s overall goodness of fit, with statistically nonsignificant values of the test indicating a good overall fit of the equation. Results are reported as statistically significant whenever p < .05.

Sample

Men ranged in age from 18 to 63 (mean = 35.8, median = 36, SD = 8.8). The sample approximates the Arnerican population fairly well in terms of its racial composition, with 76.9% of the men being Caucasian, 8.1% African American, 7.8% Latino, 2.7% Asian, 0.2% Native American, and 4.3% biracial/multiracial. One-third of the men (33.9%) self-identified as being “top” or “versatile top” , one-quarter (22.4%) self-identified as being “versatile”; and the remainder (43.7%) self-identified as being a “bottom” or a ‘versatile bottom.” Most (59.1%) said that they were HIV negative, although sizable proportions of the men whose ads were coded said that they knew that they were HIV positive (17.6%) or that they did not know what their HIV serostatus was (23.3%). The sample, like the American population in general, tended to be skewed toward people residing in more densely populated areas. One-fifth of the men (20.0%) lived in an area with fewer than 250 persons per square mile. At the other end of the spectrum, 37.8% of the men resided in an area with more than 5,000 persons per square mile, and half of these men (19.8% of the total sample) lived in a high-density urban area with more than 10,000 persons per square mile.

RESULTS

Research Question #1: Differences in Demographic and Descriptive Characteristics of the People Placing Ads That Are/Not Arnong the Most Searched for

The most popular profiles were those placed by men who were nonCaucasian (x² [1df] = 5.86, p = .02). Results showed a strong tendency for the most-searched-for profiles to have been posted by younger men than by their older counterparts. Men who were under the age of 30 were more than 50% likelier than their counterparts age 30 or older to have one of the most active profiles (OR = 1.68, CI₉₅ = 1.32–2.16, p < .0001). Conversely, men who were age 50 or older were about one-fifth as likely as their younger counterparts to have one of the most popular profiles (OR = 0.21, CI₉₅₌0.11–0.38, p < .0001). Men whose profiles were the most searched for were more likely to describe themselves as versatile, versatile bottom, or bottom than they were to label themselves as a versatile top or a top (x² [1df) = 4.17, p = .05). The most popular profiles were posted by men who considered themselves to be gay, by a factor of about 2:1 compared to their heterosexual, bisexual, or “curious” counterparts (OR = 2.02, CI₉₅ = 1.41–2.90, p < .0001). HIV serostatus was also a key factor differentiating men whose profiles were among the most active and those whose profiles were not. For example, men who were HIV positive were twice as likely as those who were not to have posted one of the most popular profiles (OR = 2.03, CI₉₅ = 1.51–2.72, p < .0001). Likewise, men who said that they did not know what their HIV serostatus was were more than 50% likelier than those who knew their serostatus Eo have a profile that was among the Web site’s most popular (OR = 1.55, CI₉₅ = 1.20–2.01, p = .0009). Conversely, men who said that they were HIV negative were less than half as likely as their HIV-positive and serostatus-unknown counterparts to have posted one of the most active profiles (OR = 0.47, CI₉₅ = 0.38–0.59, p < 0001). Men whose profiles were among the most popular tended to live in more densely populated areas than men whose profiles were not designated as being among the most frequently searched (t = 7.19, p < .0001).

Research Question #2: Differences in the Risk Practices Advertised for in Profiles That Were and Were Not Among the Most Popular on the Web Site

Numerous differences were found with regard to the sexual risk practices advertised for in the profiles that were among the Web site’s most popular and those that were not. For example, the most active profiles were less likely to have been posted by people who were looking for partners who would perform oral sex on them and swallow their semen (x² [1dfl = 4.40, p = .04). Conversely, the men whose profiles were among the Web site’s most popular were more likely to say that they wanted to find persons on whom they could perform oral sex and subsequently swallow their semen (x² [1df] = 4.02, p = .05). Moreover, the most active profiles tended to be those placed by men who were looking for partners who would perform anal sex on them and then ejaculate into their anus (x² [1df] = 15.54, p < .0001). No significant difference was found for the outcome measure of looking for anal sex panners who would allow the profile poster to ejaculate into the anus (x² [1df] = 0.64, p = 43). Men whose profiles were among the most highly searched for were approximately one-third more likely than those whose profiles were not to be looking for partners who were willing to engage in all four of the preceding sexual risk behaviors (x² [ldf] = 7.85, p = .006).

Men who expressed an interest in finding partners with whom they could engage in felching were nearly twice as likely as those who had not expressed such an interest in having a profile that was highly searched for by others (OR = 1.92, CI₉₅ = 1.42–2.58, p < 0001). Posting a profile that specified a desire to find men who would be willing to engage in double penile anal penetration was associated with a sevenfold increase in the likelihood of having that profile be among the Web site’s most popular (OR = 7.21, CI₉₅=1.62–32.06,p=.003).

Research Question #3: Differences in Risk Preferences Mentioned in Profiles That Were and Were Not Arnong the Most Searched for

Men who wanted to find partners who would engage in rough sex with them were twice as likely to have active profiles as men who were not seeking partners with whom they could engage in rough sex (OR = 2.05, CI₉₅ = 1.44–2.90, p < .0001). People whose online profiles specifically stated that they disliked condoms or would refuse to use condoms (or allow a partner to use condoms with them) were more than nine times as likely as those whose profiles did not express a distaste for condoms to have a highly-searched-for profile (OR = 9.48, CI₉₅ = 3.72–24.15, p < .0001). Persons who were actively seeking to become HIV infected (i.e., bug chasers) were five times more likely to have a popular profile than were their counterparts whose profiles did not state that they were trying to become infected with HIV (OR = 5.39, CI₉₅ = 2.05–14.23, p < .0001). Explicitly stating that one was unwilling to withdraw one’s penis prior to ejaculation or to permit one’s partner to do so was associated with a nearly threefold increase in the likelihood of having a highly-searched-for profile (OR = 2.85, CI₉₅ = 1.18–6.88, p = .02). Having a profile that promoted anonymous sexual encounters was associated with a nearly doubled increase in the chance of having a profile that was among the site’s most active (OR = 1.87, CI₉₅ = 1.00–3.51, p = .05). Men whose profiles explicitly made reference to their desired sexual activities as “breeding” or “seeding” were five times more likely to have a highlysearched-for profile when compared to their peers who did not make such references in their profiles (OR = 5.33, CI₉₅ = 2.96–9.62, p < .0001). Wanting to find partners with whom one could engage in long-lasting sexual relations was associated with a doubled chance of having a profile that was among the most frequently searched for (OR = 2.08, CI₉₅ = 1.31–3.32, p = .002). Men with the most active profiles were nearly three times as likely as those whose profiles were not among the most active to say that they wanted to find persons with whom they could engage in “wild” or “uninhibited” sex (OR = 2.90, CI₉₅ = 1.83–4.61, p < .0001). Eroticizing ejaculatory fluids was three times more common in the profiles that were popular than it was in those that were not among the Web site’s most highly searched (OR = 3.01, CI₉₅ = 2.17–4.18, p < .0001). One of the only nonsignificant differences obtained was for the outcome measure “having sex while under the influence of drugs” or “partying and playing (PNP),” which was prevalent in approximately one-half of the men whose ads were among the Web site’s most highly searched (56.0%) and among those whose ads were not as popular (51.4%) (OR = 1.21, CI₉₅ = 0.98–1.49, p = .09).

Research Question #4: Differences in Sought-After Partner Characteristics in Profiles That Were and Were Not Among the Most Popular on the Web Site

The preference that men’s profiles stated for potential sex partners’ HIV serostatus had a significant impact upon whether their profiles were among the most searched for on the Web site (x² [3df] = 87.96, p < .0001). For example, men who said that they did not care about the HIV serostatus of prospective partners were nearly two-and-a-half times more likely to have one of the Web site’s most popular profiles than were their counterparts who did have a preference for the serostatus of their sex partners (OR = 2.40, CI₉₅ = 1.91–3.02, p < .0001), Similarly, men whose profiles stated that they were looking for people who were HIV positive were more than two-and-a-half times more likely than those whose profiles were seeking persons who were not HIV positive to be among the Web site’s most searched for (OR = 2.61,CI₉₅= 1.44–4.75, p < .0001). In contrast, persons who wanted to meet HIVnegative persons were approximately one-third as likely as those who did not specifically stipulate wanting to meet HIV-negative individuals to have profiles that were among the Web site’s most frequently searched (OR = 0.34, CI₉₅ = 0.26–0.43, p < .0001).

The types of relationships sought by the persons placing the online profiles were also found to be key differentiae of whether someone’s profile was among the Web site’s most active. For example, men who were looking for partners for large-group sexual encounters were nearly twice as likely to have a profile that was among the Web site’s most active as their counterparts who were not interested in finding partners for such encounters (OR = 1.89, CI₉₅ = 1.46–2.45, p < .0001). Similarly, men who were looking for partners for three-way sexual encounters were about one-and-a-half times as likely to have a highly-searched-for profile as their counterparts who were not looking for this type of sexual encounter (OR = 1.63, CI₉₅ = 1.25–2.13,p=.0003). Seeking persons with whom one could develop some type of long term relationship was also associated with a greater likelihood of having a popular profile (OR = 1.43, CI₉₅ = 1.15–1.79, p = .002). No significant differences were found for the most popular profiles and those that were not among the most popular profiles in terms of seeking partners for a one-on-one sexual encounter (OR = 1.50, CI₉₅ = 0.98–2.33, p = .06).

Research Question #5: What Factors Contribute Uniquely and Significantly to the Prediction of Whose Profiles Are the Most Searched for?

Table 1 presents the findings of the multivariate analysis undertaken to determine which factors contribute uniquely and significantly to the prediction of whose profiles are the most searched for. In all, 11 variables were retained in the final multivariate equation. First, race/ethnicity was a significant contributor to the multivariate model, with Caucasian men being less likely than their nonwhite counterparts to have profiles that were among the Web site’s most popular (p = .002). Second, two of the measures penained to HIV serostatus. Men who were HIV positive were approximately 50% more likely than those who were not HIV infected to have profiles that were among the site’s most searched for (p = .009). Moreover, not caring whether potential sex partners were HIV positive, HIV negative, or serostatus unknown was associated with a nearly doubled chance of having a popular profile (p < .0001). Third, one particular sexual behavior—namely, receiving oral sex and ejaculating into the partner’s mouth—was also found to be linked with the likelihood of having a highly-searched-for profile on the Web site. In this instance, wanting to receive oral sex from someone who would swallow one’s semen, which is generally construed as a low-risk behavior, was associated with a diminished chance of having a popular profile (p = .03). Fourth, several of the risk preference measures were identified as unique contributors in the multivariate model. For example, wanting to locate persons with whom one could have multiple-partner sex or large-group sex was associated with an increased chance of having a highly-searched-for profile (p = .03). Hoping to find partners for rough sex was associated with a 50% greater likelihood of having a popular profile (p = .04). Overtly stating that one dislikes using condoms led to a five-fold increase in the chance of having a highly-searched-for profile (p = .002). “Bug chasing”—that is, actively trying to find HIV-positive partners who could cause one to become HIV infected—was also associated with nearly a fivefold increase in the odds of having a popular profile (p = .004). Thinking of one’s desired sexual activities as “breeding” or “seeding” (or “being bred” or “being seeded”) was associated with a nearly tripled risk of one’s profile being highly sought after (p = .002). In the multivariate analysis, wanting to identify partners with whom one could have “wild” or “uninhibited” sex was associated with an almost-doubled chance of having a popular profile (p = .02), as was the act of eroticizing ejaculatory fluids (p = .0003). Together, these items accounted for approximately 17.9% of the total variance. The Hosmer-Lemeshow chi-square statistic was nonsignificant and nowhere near attaining statistical significance (x² [8df] = 2.94, p = .94), indicating that the multivariate model presented in Table 1 is an excellent fit.

TABLE 1.

Multivariate Equation to Identify What Factors Are Associated with Whether Someone I-das One of the Web Site’s Most-Searched-For Profiles

Predictor	OR (CI95)	Standardized Coefficient
Race = Caucasian	0.64 (0. 48–0.84)	0.11**
HIV Serostatus = HIV Positive	1.54 (1.12–2.11)	0.09**
HIV Serostatus of Potential Partners = Don’t Care	I .98 (1.54–2.54)	0.18***
Receiving Oral Sex and Ejaculating Internally	0.66 (0.46–0.95)	0.07*
Wanting Large-Group Sexual Encounters	1.39 (1.05–1.84)	0.08*
Wanting Rough Sex	1.50 (1.02–2.19)	0.07*
Overtly Stating a Dislike for Condorns	5.07 (1.91–13.44)	0.16**
Actively Trying to Become HIV Infected	4.64 (1.67–12.93)	0.12**
Conceptualizing One’s Sexual Activities as Breeding	2.88 (1.53–5.43)	0.14**
Wanting “Wild” or “Uninhibited” Sex	1.84 (1.12–3.02)	0.09*
Eroticizing Ejaculatory Fluids	1.91 (1.35–2.71)	0.13***

^*p < .05;

^**p < .01;

^***p < 001.

DISCUSSION

Potential Limitations of This Research

This content analysis research was conducted using one specific Web site and, therefore, there is no way to know whether users of this particular site are similar to or different from those who frequent other sites instead. Other sites were excluded from consideration in this research because of the fees that they charge in order for would-be users to partake of their services⁴ and/or because of the significant limitations they placed on members’ ad/profile content. As a Web site specifically designed to foster contacts between men who wish to locate sexual partners with whom they can have unprotected sex, rather than being a Web site designed to foster male-tomale contacts of a rnore general nature (e.g., dating, friendships, activities partners), the Web site chosen as the focus of this research represents an excellent sampling of men who are using the Internet specifically to locate other Inen with whom they can have unprotected sex.

Another potential limitation of this research is that virtually all of the ads/profiles appearing on the Web site studied are written in English. Although the Web site has a substantial Latino and multiracial membership (11.9% in total), fewer than 0.5% of the ads/ profiles were written in a language other than English. This may prevent non-English speakers from utilizing the Web site, and this may limit the generalizability of the findings somewhat.

Finally, as a content analysis project, this research is unable to assess the extent to which the behaviors advertised for in the ads do or do not Ida represent the behaviors practiced when people meet in person. For example, suppose someone has posted a profile stating that he does not care what his potential sex partner’s HIV serostatus is, and he meets an HIV-positive person who contacted him as a result of his profile. It is impossible to know whether the person would engage in all of the same sexual behaviors with a partner who is known to be HIV positive that he would with a partner who is not HIV positive or a partner with whom there is no discussion of HIV serostatus. In point of fact, the extent to which the risky behaviors advertised for and/or listed as preferences in the ads do or do not reflect actual behaviors practiced when people meet in person is, by necessity, the subject of a different study, following on the heels of the current project. Published studies (cited earlier in this paper) showing that men who use the Internet to locate sexual partners are very likely to meet up with such individuals for sex (i.e., their ads/profiles are, far more often than not, not posted purely for fun, but rather with sexual hook-ups in mind) suggest that there may not be a great disconnect between ad/profile content and behavioral practices. Nevertheless, this needs to be established with additional research.

Conclusions

Despite these potential—and, the present author would contend, minimal—limitations, the present research has much to offer in terns of helping to understand the content of ads posted on Web sites designed to foster unprotected sexual encounters among men seeking to have sex with other men. First and foremost, this research has shown that the profiles that were the most frequently found by other users’ searches of the Web site in question were those promoting the greatest levels of risky sexual behaviors. Compared to those that were less popular, the most popular or active profiles were placed by men who wanted to find partners on whom they could perform oral sex and swallow the semen, partners who would perform anal sex on them and then ejaculate internally, partners with whom they could engage in the high-risk practice of felching, and partners who would be willing to penetrate them anally at the same time that another man was doing the same thing (i.e., “double fucking”). Moreover, the most popular profiles were posted by men who had attitudes and/or behavioral preferences that heightened even more the riskiness of the sexual behaviors they were seeking. For example, men whose ads stipulated an unwillingness to use condoms or to allow a partner to use condoms were more than nine times as likely co be among the most frequently searched as men whose ads did not make such a stipulation. As another example, profiles that stated an unwillingness to withdraw the penis prior to ejaculation or to allow a partner to do so were nearly three times more likely to be among the Web site’s most popular profiles as those that did not explicitly state this. Also associated with an increased likelihood of having a highly-searched-for profile were risk-enhancing preferences such as wanting to find partners for anonymous sexual encounters, seeking “wild” or “uninhibited” sexual encounters, and wanting to find persons with whom they could engage in rough sex, among others. Ali of these behaviors and risk-related preferences are consistent with one very important finding of this research: When MSM use Web sites to locate potential sex partners, they tend to be looking for partners with whom they can engage in very risky behaviors.

From a public health perspective, this finding indicates a great need for HIV prevention and intervention programs to target men who have sex with other men and who use the Internet to locate potential sex partners. Finding ways to convince men such as these, who as a group are probably extremely knowledgeable about the behaviors that place people at risk for contracting and spreading HIV (Kelly & Kalichman, 1998; Korner, Hendry, & Kippax, 2005; Mulry et al., 1997), to use condoms and/or to negotiate for lower-risk sexual behaviors than they currently practice is likely to prove extremely challenging. Perhaps educating the men who use the Internet to locate sex panners about the overall high level of riskiness being sought by the “average” site user would lead some of these individuals to feel more personally vulnerable to contracting HIV and, thus, reduce their willingness to engage in high-risk sexual practices. Previous research has shown that increasing perceived vulnerability to HIV/AIDS can be effective at helping people to reduce their risk for HIV (Belcher, Stemberg, Wolitski, Halkitis, & Hoff, 2005; Longshore, Anglin, & Hsieh, 1997; Rosengard, Anderson, & Stein, 2006). Also potentially worthwhile would be the development and implementation of HIV education and intervention endeavors conducted with the assistance and involvement of the owners/operators of the Websites promoting meetings of men who are looking for other men with whom they can have unprotected sex. Unfortunately, this, too, is likely to prove challenging in many instances, because these Web sites are for-profit ventures and their owners/operators are unlikely to be willing to collaborate with risk-reduction professionals if they (the owners/operators) perceive their collaboration as having the potential of alienating future or current customers (i.e., users of and subscribers to their Web site). Working with these site owners and operators to find ways to diminish the high-risk nature of their Web sites’ content while not compromising their Web sites’ profitability will be the key to making such initiatives work.

As a result of the very risky nature of the behaviors being sought by men whose profiles were among the Web site’s most popular profiles, it is also important to take note of any descriptive characteristics that tend to typify men with the most popular profiles. By noting such traits, we derive a better idea of which subgroups of men are in particular need of the types of education, prevention, and intervention efforts mentioned above—that is, which “types” of men need to receive targeted education, prevention, and intervention efforts. This research revealed several such characteristics, a few of which are discussed below.

First, younger men were more likely to have popular profiles than were older men. Other researchers have found that younger MSM report higher rates of risky practices than their older counterparts (Chen, Weide, & McFarland, 2003; Halkitis et al., 2005; Xia, Osmond, Tholandi, Pollack, Zhou, Ruiz, & Catania, 2006), and the present research is consistent with their findings. Helping younger men to personalize HIV—something that is often much less necessary with older adults, who have lost more friends to AIDS than the current, younger generation of MSM have—may be one crucial part of this strategy. Teaching young men about the reality of HIV/AIDS medications and about what it means to live with HIV—namely, that these medicines do not cure HIV or AIDS, and that even in 2010 living with HIV or AIDS still means living with recurrent illnesses—may prove to be another important component of this strategy, as may increasing young men’s sense of vulnerability to HIV/AIDS.

Second, persons of color were more likely than their Caucasian counterparts to have highly-sought-after profiles. This probably stems from a number of things, including the following: There is a certain amount of “idealized mythology” surrounding penis size and men of color, particularly African-American men. The expectation that many African-American men will have larger-than-average penises makes them attractive partners to many MSM using Web sites to locate potential sex partners. Moreover, men of color, particularly those of African-American and Latino descent, are less likely to have been circumcised compared to their Caucasian counterparts (National Center for Health Statistics, 2007). Many MSM eroticize the state of being uncircumsized and fantasize about having sex with men who are uncircumsized, presumably in part due to the novelty of finding such men as their sex panners. Another factor that is likely to be in operation here pertains to the adherence to cultural stereotypes of being dominated sexually by someone who is perceived to be extremely masculine, as would be the case with Latino and African-American men (Diaz, 1997; Girman, 2004). Research has shown that adherence to such beliefs/stereotypes is associated with a greater involvement in HIV-related risk practices (Carballo-Dieguez et al., 2004; Diaz, 1997; Jarama, Kennamer, Poppen, Hendricks, & Bradford, 2005). Regardless of the specific origin(s) of the greater preference for locating sex partners who are nonwhite, the fact remains that this preference prevailed in this study and it was associated with a greater likelihood of seeking high-risk behaviors. MSM of color, therefore, constitute a group in need of targeted intervention, and this has been noted by other authors as well (Rhodes, Yee, & Hergenrather, 2006).

Third, this study revealed that men who self-identified as being sexually “versatile,” “versatile bottom,” or “bottom” had profiles that were being sought more frequently than men who considered themselves to be “versatile top” or “top” sexually. This, too, emphasizes the higher-risk nature of the intended targets of the profile-searching process, since sexually “versatile” and “bottom”-oriented men are at greater risk of contracting HIV sexually than their “top”-oriented counterparts are by the very nature of the sexual behaviors that they prefer to practice (e.g., receptive anal sex versus insertive anal sex, performing oral sex versus receiving oral sex). As a result, it may be necessary to provide some basic HIV/AIDS education EO these individuals to make sure that they are, indeed, fully aware of the risks that are inherent in their favorite sexual acts. Moreover, educating them about other, safer ways to engage in these activities or in other sexual activities that are less risky but that provide them with many of the same sexual sensations could also prove useful. For example, although performing oral sex on a man who is not wearing a condom is considered a low-to-moderate risk behavior by most, perhaps “bottom”-oriented men could be encouraged not to swallow the semen when their partner is prepared to ejaculate. While this behavior still carries with it a measure of HIV risk, it is a lower level of risk than the same behavior that entails performing the oral sex to the point of completion and intemal ejaculation. The same approach could be taken to receiving anal sex, with the same relative reduction in risk applying. As another example, men who like to receive anal sex from their partners might be talked into having their partners use sex toys (e.g., dildos) on them instead of unprotected penises. The sexual sensations for the “bottom” man would be at least somewhat similar, and this switch-off could help to reduce the person’s level of risk, even if done only some of the time when anal sex would be occurring. The key, the present author believes, is to find ways of eliminating risk whenever possible (through education and increased knowledge, for example) and other ways of reducing risk whenever complete extinction of risk is unrealistic to expect. This is, in the purest sense, an advocacy for the implementation of a harm-reduction approach (Westermeyer, 1994) to dealing with the problem of actively seeking unprotected sex among men who have sex with other men.