“GWM looking for sex—SERIOUS ONLY”: The interplay of sexual ad placement frequency and success on the sexual health of “men seeking men” on Craigslist

Abstract

Over the past few years, researchers have begun to explore the use of the personals section on Craigslist.org as it applies to the sexual behaviors and partner accrual of men who have sex with men (MSM). Yet, the exact ramification of such a free sexual advertisement service on MSM's sexual health and sexual risk-taking behaviors are just being documented. As such, we surveyed 535 men by responding to their sexual ads placed on Craigslist.org to assess their sexual health and sexual health behaviors. Using Self-Efficacy Theory as our foundation, we posited that MSM who placed more ads and reported more success from these ads would also report more negative health behaviors (e.g., lowered condom use) and sexually transmitted diseases. Testing confirmed this hypothesis. There were two important implications. The behavior of Internet cruising, on its own, insufficiently impacted MSM health. It was the interaction of number of ads (self-efficacy) and the success of those ads (outcome-efficacy) that ultimately contributed to negative health behaviors. Second, MSM may have had different intentions for cruising. The anti-theoretical tendency for some men to continually place high numbers of ads while also reporting very low success suggested that the outcome of merely placing the ads might be sufficiently erotic for some men.

INTRODUCTION

A major shift has occurred over the past fifteen years in how men who have sex with men (MSM) find potential sexual partners. In the past, MSM often searched gay bars and bathhouses for casual sexual encounters. Today, such men who look for sexual partners can view advertisements of men on the internet that list demographic information, sexual propensities, and other personal information. These can be created cheaply and easily, in some cases for free. As a result, websites catering to MSM in search of same-sex sexual encounters have proliferated. About 50% of MSM report using such sites to facilitate a sexual encounter (Garofalo, Herrick, Mustanski, & Donenberg, 2007).

Craigslist.org (Craigslist) may be categorized as a website that allows for such internet cruising. This is a site that is community-specific and allows for individuals in close proximity (e.g., chicago.craigslist.org or tampa.craigslist.org) to post and find jobs, merchandise, rentals, sublets, and people. Moreover, this is a free service, as compared with other websites that charge monthly or yearly fees (e.g., manhunt.net or gay.com). In the past few years, researchers have begun to notice and explore the use of Craigslist as it applies to MSM's sexuality and partner accrual (Ward, 2007, 2008), and prostitution (Holt & Blevins, 2007; Holt, Blevins, & Kuhns, 2008). Yet, the exact ramifications of such a free sexual advertisement service on MSM's sexual health and sexual risk-taking behaviors are just being documented.

Admittedly, research has been conducted on the sexual health of MSM who generally use the internet to find sexual partners (e.g. Bolding, Davis, Sherr, Hart, & Elford, 2004; Kim, Kent, McFarland, & Klausner, 2001; Mettey, Crosby, DiClemente, & Holtgrave, 2003; etc). These studies have shown that MSM engaging with sexual partners met through the internet were more likely to contract sexually transmitted diseases (STIs) than those engaging with partners found through other means (Klausner & Wong, 2003; Rietmeijer, Bull, McFarlane, Patnaik, & Douglas, 2003). MSM using the internet for sex were about two times more likely to contract syphilis (Wong, Chaw, Kent, & Klausner, 2005). Research has also shown an increased likelihood for HIV infection among such men (Rietmeijer, Bull, & McFarlane, 2001). Thus, there does seem to be a negative impact on the health of MSM who cruise online.

Sensation seeking has emerged as a potential reason for this association (Horvath, Beadnell, & Bowen, 2006). That is, MSM who are highly sensitive to positive experiences (e.g., anal intercourse) and disproportionately look for these experiences are substantially more likely to take risks, particularly when using the internet. This may be due to the availability and anonymity of individual cruising, which can be carried out at any time from any computer or internet-ready device (e.g., the iPhone^®). Others suggest that sexual compulsivity may play a role (Grov et al., 2007; Parsons, Severino, Grov, Bimbi, & Morgenstern, 2007), whether as an independent or dependent outcome. MSM who are already sexually compulsive may use the internet to facilitate sex with even more speed. This leads to more sexual partners and more frequently enacted unprotected intercourse. Alternatively, the availability and increased access to other potential partners by using the internet may become addictive (Chaney & Chang, 2005), leading many of them to have more sexual partners and more unprotected intercourse.

Yet, there is a theoretical component missing from the research on MSM who use the internet to acquire sex: outcome-efficacy (Bandura, 1977). Research has shown that individuals will decide to enact a behavior according to degrees of self- and outcome-efficacy. Self-efficacy refers to individuals’ judgments regarding their abilities to perform a behavior (Bandura, 1986). Outcome-efficacy refers to whether the behavior will lead to a specific positive or negative outcome (Bandura, 1977). Behavior enactment is contingent on both of these concepts. Merely because an individual shows a high degree of self-efficacy regarding a behavior (e.g., being able to use a condom), does not necessary predict the enactment of the behavior. Outcome efficacy must also be high (e.g., that using the condom produces a positive outcome). Research shows that during instances in which self- and outcome-efficacy contrast, behavioral enactment may be unpredictable. For example, using safer sex, men report knowing how and why to use condoms (i.e., high self-efficacy). However, they also report the outcome of decreased sensation. As a result of their low outcome-efficacy, condom use may not ensue (O'Leary, Goodhart, Jemmott, & Boccher-Lattimore, 1992; Wulfert & Wan, 1993).

A strong degree of self-efficacy is implicit in all of the research previously cited on MSM internet cruising: MSM know how to cruise, and it is an available and employable behavior useful towards acquiring sex. Men's perceptions of outcome-efficacy are virtually unknown. That is, do the men feel that internet cruising produces a quantifiable positive or negative outcome?—such as acquiring a sexual partner. Though individuals may have a high degree of self-efficacy at cruising, they actually may have a low degree of outcome-efficacy. Simply, some MSM may be cruising, but this behavior may not be leading them to actualized sex.

The role of varying outcome-efficacy on the sexual health and sexual health behaviors of MSM is even more uncertain. It would seem that increased self-efficacy regarding internet cruising, as shown through the previously mentioned studies (e.g., Klausner & Wong, 2003; Rietmeijer et al., 2003; etc.), leads to lowered condom use and more STI infections. Because self- and outcome-efficacy have been noted to be related positively (Maddux, Sherer, & Rogers, 1982), usually when the behavior is beneficial, we posited that increased outcome-efficacy will encourage the same tendencies as shown for self-efficacy. That is, as men report more success with advertising for sex online, their condom use and communication about HIV will decrease, and they will report having acquired more STI infections during the past year. This only will be exacerbated by an increased number of advertisements placed.

METHODS

Because Craigslist offers the discrete placement of free ads (one free ad at a time), we selected this website to test our hypotheses. Men should be able to count the number of ads they placed in a given time and assess how many of these ads resulted in actualized sex with a male partner. Also, Craigslist is free of cost and available to all men (as compared against pay-sites such as Manhunt.net).

Procedures

Men placing sexual advertisements in the “men seeking men” personals section of Craigslist were asked to take a brief on-line survey. These men were recruited from all of the Craigslist locations in which English was the predominate language (i.e., all cities in Australia, New Zealand, the United Kingdom, and the United States). Specifically, we responded to each man's listed advertisement with a block message informing him of a research study on the sexual health behaviors of men looking for sex online. MSM interested in taking the survey could click on a link embedded in the email that took them directly to the survey. Participants were not compensated for their time.

Participants

This study focused on MSM, regardless of self-reported sexual orientation. Participants completed the 20-minute survey from February to April 2008. Of the 746 individuals who began the survey, 535 (71.7%) completed the questionnaire in its entirety. Because we did not compensate participants for their time, most men tended to ignore our email, which resulted in a low response rate (5-10% depending on the Craigslist location). We collected a larger sample to compensate for this tendency.

Measures

Ad Placement Frequency and Success

We asked for, “the number of ads the men placed looking to ‘hookup with men’ in the past 90 days.” A hookup or hooking-up refers colloquially to having a sexual experience with another. We followed this question with, “In the past 90 days, what percent of the time were you successful at having a sexual encounter by using these online advertisements?” Success was measured on a continuum from 0% to 100% of the time using 10% increments. Participants who reported 100% success reported that every ad they placed resulted in an actualized sexual experience. Participants who reported 50% success reported that one in two ads they placed resulted in an actualized sexual experience.

Condom Use

Participants were asked, “In the past 12 months, what percent of the time did you use condoms during anal intercourse?” (0 = I never had this sort of sexual partner, 1 = never used a condom, 2 = practically never used a condom, 3 = rarely used a condom, 4 = sometimes used a condom, 5 = usually used a condom, 6 = practically always used a condom, 7 = always used a condom). MSM (n = 129) who reported never having an anal intercourse partner were omitted from the analyses.

HIV Disclosure

To assess the degree to which MSM were communicating their HIV serostatus to partners, we asked for the number of anal sex partners the men had acquired during the past year by using the internet. Each of these questions was followed by, “With how many of these partners did you discuss your HIV-negative, positive, or unknown status?” A ratio was then created in which number of disclosed partners was divided by number of total partners (e.g., a score of 1 signified 100% disclosure, a score of 0 signified 0% disclosure).

Sexually Transmitted Infections (STIs)

Participants could indicate having any of nine STIs over the previous 12 months: chlamydia, gonorrhea, crabs/scabies/pubic lice, genital herpes, genital warts/HPV, hepatitis, syphilis, protozoa/fungal infections, or other. The number of reported STIs was summed for each participant.

Statistical Analysis

First, the data were analyzed using bivariate correlations and analysis of variance to test for potential control variables. Weighted least squares (WLS) multiple linear regression was used for testing for associations between the independent variables and positive or negative sexual health behaviors. We weighted the analyses that featured condom use, HIV disclosure, and number of STIs acquired by number of anal intercourse partners. This was conducted because the variance associated with men who had few actual anal sex partners was not equivalent to the variance associated with men who had many actual anal sex partners—particularly with respect to negative sexual health behaviors.

WLS moderated regression was used to explore amplified effects (e.g., placing more ads and having these ads successful result in a sexual experience). This was conducted by multiplying the given moderator with the independent variable to create the interaction term (Aiken & West, 1991). Variables were entered at different steps of the regression, with controlled variables entered before the independent and moderator variable. The interaction term was entered last. When an interaction term proved to be significant, we deconstructed the term into its different variations in order to interpret its meaning accurately. Continuous variables, such as percent of ads that proved successful, were transformed into low and high variations by subtracting or adding one standard deviation to each participant's response. Thus, low variations represented the strength of the relationship at one standard deviation below the mean and high variations represented the strength of the relationship at one standard deviation above the mean (see Aiken & West, 1991).

RESULTS

Sample

As shown by Table 1, the sample was mostly White and HIV-negative. Most men (88.3%) had at least attended some college. This was a relatively rural sample. Many of the men came from small to medium-sized towns (44.6% from areas populated with under 100,000 people). A sizeable minority reported a sexual orientation other than homosexual (37.6%). On average, the men placed one ad every five to six days and were successful every three to four ads. No notable differences were found regarding the demographic variables, amount of ads placed, and success rate of those ads placed. Success rate was not influenced significantly by amount of ads placed.

Table 1.

Description of the sample

	n	% of N
Location (in people)
< 10K	50	9.3
10K – 50K	100	18.7
51K – 100K	89	16.6
101 – 500K	115	21.5
501 – 1000K	80	15.0
1000K +	101	18.9
Race/Ethnicity
White	462	86.4
Black	20	3.7
Latino	31	5.8
Asian/Pacific Islander	9	1.7
Middle Eastern	2	0.3
Other	11	2.1
Education
Some high school	9	1.7
Finished high school	54	10.0
Some undergraduate	179	33.5
Finished undergraduate	150	28.0
Some graduate	40	7.5
Finished graduate	103	19.3
Sexual Orientation
Heterosexual	20	3.7
Homosexual	334	62.4
Bisexual	149	27.9
Will not label	32	6.0
HIV Status
HIV-negative	468	87.5
HIV- positive	38	7.1
HIV-unknown	29	5.4

	M	SD	Range
Age	40.84	12.26	18-79
Number of Male Anal Sex Partners	7.03	15.65	0-100
Number of Ads Placed	15.69	21.14	1-90
Success Rate for Hookups	29.83	26.79	0-100

Note. N = 535. Number of male anal sex partners was measured retrospectively over the past 12 months and refers to partners met through the Internet. Success rate for Hookups was measured in percent of the time (out of 100%).

Sexual Risk-Taking Behaviors

On our seven-point scale, the men in the sample reported “usually” using condoms during anal intercourse with casual partners (M = 5.30, SD = 2.01). The men discussed with, or disclosed their positive, negative, or unknown HIV status to, four out of five anal sex partners (M = 80.27%, SD = 35.82%). Older men were found to use condoms less (r = -.16, p < .01) than younger men. HIV-positive men were found to both use condoms (r = -.27, p < .001) and disclose their HIV status less (r = -.29, p < .001) than HIV-negative men. Thus, we controlled for these variables across all of the analyses.

The first model contained the control variables, success rate, and number of ads placed as independent variables and condom use as the dependent variable. We added an interaction term to the model to see if the two main independent variables amplified each other (i.e., success rate X number of ads placed). The overall model was statistically significant, F(6, 346) = 22.37, p < .001, R² = .28. Success rate produced a main effect: Men reporting more success with their ads reported using condoms less (t(346) = -2.76, p < .01, β = -.42).

Additionally, the interaction of success rate and number of ads placed was significant (t(346) = -4.87, p < .001, ΔR² = .05, β = -.74). To interpret this result more accurately, we split the interaction by those who were more successful (one SD above the mean, or 56.62% successful), and those who were less successful regarding their internet hookups (one SD below the mean, 3.04% successful). As Figure 1 illustrates, men with increased success at hooking-up used condoms less frequently as the number of ads they placed also increased (t(346) = -1.96, p = .05, β = -.11). Men with decreased success at hooking-up used condoms more frequently as the number of ads they placed increased (t(346) = 4.52, p < .001, β = .38).

Figure 1.

Low and high number of ads predicting condom use by success rate. Note, the x-axis represents generalized low (one SD below the mean) and high number of ads (one SD above the mean), and the y-axis represents increasing condom use. Those who were more successful at hooking-up fell one SD above the mean, and those who were not successful at hooking-up fell one SD below the mean. The horizontal line running across the figure represents the mean for condom use.

The second model contained the control variables, success rate, and number of ads placed as independent variables and HIV disclosure as the dependent variable. We added an interaction term to the model to see if the two main independent variables amplified each other (i.e., success rate X number of ads placed). The overall model was statistically significant, F(6, 358) = 21.78, p < .001, R² = .27. Success rate produced a statistically significant main effect: Men reporting more success with their hookups reported discussing their HIV status more (t(358) = 3.85, p < .001, β = .57).

Additionally, the interaction of success rate and amount of ads placed was significant, (t(358) = 2.32, p < .02, ΔR² = .01, β = .35. To interpret this result more accurately, we split the interaction by those who were more successful (one SD above the mean, or 56.62% successful), and those who were less successful regarding their internet hookups (one SD below the mean, 3.04% successful): Men with increased ad success disclosed no more or less as the number of ads they placed also increased (t(358) = -.02, p = .98, β < .01). Men with decreased success at hooking-up disclosed less as the number of ads they placed increased (t(358) = -2.78, p < .001, β = -.23).

STIs

One in four of the men in our sample reported at least one STI contracted during the previous year (M = .25, SD = .61). We controlled for age and HIV status. We also controlled for condom use because it was correlated negatively with contracting STIs (r = -.12, p = .02). We ran a WLS moderated regression model containing the control variables, success rate, and number of ads placed as independent variables and condom use as the dependent variable. We added an interaction term to the model to see if the two main independent variables amplified each other (i.e., success rate X number of ads placed). The overall model was statistically significant (F(6, 358) = 12.78, p < .001, R² = .18).

Although no main effects were produced, the interaction of success rate and number of ads significantly contributed to the fit of the model, (t(358) = 3.92, p < .001, ΔR² = .04, β = .62). To interpret this result more accurately, we split the interaction by those who were more successful (one SD above the mean, or 56.62% successful), and those who were less successful regarding their internet hookups (one SD below the mean, 3.04% successful). As Figure 2 illustrates, men with increased success at hooking-up reported more STIs as the number of ads they placed also increased (t(358) = 3.12, p < .01, β = .19). Men with decreased success at hooking-up reported fewer STIs as the number of ads they placed increased (t(358) = -2.51, p = .01, β = -.23).

Figure 2.

Low and high number of ads predicting number of STIs by success rate. Note, the x-axis represents generalized low (one SD below the mean) and high number of ads (one SD above the mean), and the y-axis represents increasing condom use. Those who were more successful at hooking-up fell one SD above the mean, and those who were not successful at hooking-up fell one SD below the mean. Because this was a statistical model, the configuration projects that highly successful men with low numbers of ads would have had zero STIs over the previous year.

DISCUSSION

Our research focused on the interplay between online sexual ad placements, the success of these ads towards acquiring a sexual partner, and MSM's health. Considering previous research and theories, we posited that increased success rate at hooking-up from placed ads would be associated negatively with condom use and HIV disclosure. As a corollary, men reporting higher success rates also would report having contracted more STIs. Indeed, our results showed that condom use decreased by success rate; and the number of ads placed amplified this relationship. Our results also showed that while success rate predicted increased HIV disclosure, this relationship was not present in men who placed large numbers of ads. Finally, men placing a large number of ads and reporting successful sexual encounters from those ads admitted to having contracted more STIs.

Another purpose of our research was to show that having high self-efficacy regarding internet cruising, on its own, would not impact MSM's health negatively. Our results showed that such men (i.e., those placing many ads but actualizing few of them) practiced safer sex with their partners and consequently, contracted fewer STIs. In applying the Self-Efficacy Theory (Bandura, 1977) to our results, men with high self- but low outcome-efficacy seemed to be more conscientious regarding sexual health. It was the men with high self-and high outcome-efficacy who were the most at risk regarding internet cruising and negative sexual health outcomes. Previous research found that men who cruise online are more at risk for acquiring HIV and STIs (Klausner & Wong, 2003; Rietmeijer et al., 2001, 2003; Wong et al., 2005). This may not necessarily be true for all men cruising for sex online, as our study illustrates.

There are several implications of our research that extend past the condom use or HIV disclosure by the men. When researching MSM internet cruising, it is important to understand how to measure the behavior. That is, what precisely is the unit for internet cruising as a behavior? Is it the time spent by men online or number of advertisements placed? Is it the number of partners accrued by using the internet? Is it a ratio (high/low) of effort to success (high/low)? Clearly, figuring out the unit will influence the relationship between cruising and sexual health outcomes. For example, we found no independent relationship between numbers of ads placed or the success of the ads and increased STI prevalence. However, when we created a nuanced unit that accounted more holistically for the behavior (i.e., the interaction of the number of ads and success), some men were, indeed, more at risk. This was echoed when exploring condom use and HIV disclosure. Researchers who conduct future studies might account for this more explicitly.

We also found that men tend to cruise with different intensities. Some in our sample placed ads every day looking for potential partners. Others had only placed the one ad to which we responded when we asked them to take our survey. By accounting for the degree to which the men are successful, it appeared that the men might have different purposes or intentions for cruising. The reason men placed ads on Craigslist in the men seeking men section was ostensibly because they were looking for an offline sexual encounter. It seems anti-theoretical (to Self-Efficacy Theory) that men would continue to place a high number of ads (over a 90 day period) while also reporting very low success of the ads. Low outcome-efficacy tends to predict low behavioral enactment in previous research (Bandura, 1977; 1986). Thus, a ratio of high number of ads to low success would suggest that the outcome of merely placing the ads might be sufficient for some men. Emails exchanged with other men might be adequately erotic; this might be a variation of cyber sex that has evolved for some men. Such MSM in our sample seemed to use condoms more with actual partners and reported fewer STIs. Further studies might assess the positive and negative psychological and psychosexual health implications for these MSM.

Limitations and Future Research

Our study was not without its limitations. First, we only collected data from Craigslist users. As mentioned, there are numerous websites that cater to internet cruising (e.g., manhunt.net, gay.com). Our results are not generalizable to all men who place advertisements or profiles online. The response rate was problematic too. Though we addressed this by spending the extra time to collect a large diverse sample of over 500 men, this may have introduced a selection bias to the study. Additionally, we only collected data from Craigslist users who were placing advertisements actively. The sexual health behaviors of Craigslist users who do not place ads but who review the ads of those looking to hookup was unknown. Moreover, in reference to our discussion of unitizing internet cruising, the interaction of the tendency to view advertisements and the tendency to place them was equally unknown. Finally, using number of ads placed and the success rate of the ads might operationalize self- and outcome-efficacy too crudely. Some men reported high numbers of ads and low success rates. We previously conceded that, for such men, an ideal outcome might not be to have actual offline sex. Thus, we may have misinterpreted the exact construction and influence of outcome-efficacy for some in the sample.

Future studies might correct these limitations and further explore the impact of the internet on sexual health. For example, the role of increasingly more portable technology that can access the internet (e.g., the iPhone^®) on the cruising habits of MSM is unknown. Most of the extant research is predicated on men being at home and in front of their computers. Yet, men out at bars, potentially under the influence of alcohol or drugs, can post ads on Craigslist using their phones or cruise any other website. They even might be more motivated to actualize a potential offline sexual encounter because they would be dressed, ready, and already away from home. With new technology and new innovations come new sexual health research questions. As for future research into MSM Craigslist users, studies might explore such users as potential contributors to STI and HIV infections in women. Table 1 shows that 37.6% of the men reported not being homosexual; specifically, 27.9% were bisexual and 3.7% reported being heterosexual. Our results also indicated that the MSM were not always using condoms during same-sex anal intercourse. These men may be engaging in unprotected vaginal sex as well, potentially with their girlfriends or wives. Future research might explore this potential as a means to explain more accurately the influence of the internet on behavioral bisexuality and sexual health.

Despite the discoveries of future research, the internet will continue to be a facilitator of same-sex sexual encounters for many men. The next logical step would be to move HIV and STI interventions online. Our study and many others show that internet cruising that results in a high degree of actualized sexual encounters contributes to negative sexual health and sexual health behaviors. Yet, to date, relatively few attempts have been made towards creating effective internet-based programs for MSM. The most effective programs would be tailored to each particular website, and Craigslist users certainly would benefit from such programs. For example, Craigslist has a stock message that accompanies the individual message sent when a potential sexual partner emails an advertiser looking for sex. A campaign might add sexual health information as a part of the stock message to remind men to wear condoms, discuss HIV and STIs prior to sex, and to get tested regularly. It is clear that the most computer-literate generation of MSM to date is currently at risk for STI infections, including HIV. Therefore, using the same methods that MSM are using to find sexual partners might be the best way to evoke behavioral change.