Preferred aspects of sexually explicit media among men who have sex with men: where do condoms fit in?

Abstract

Sexually explicit media (SEM) is viewed by many men who have sex with men (MSM) and is widely available via the Internet. Though research has investigated the link between SEM and sexual risk behaviour, little has been published about preferences for characteristics of SEM. In an Internet-based cross-sectional study, 1390 adult MSM completed an online survey about their preferences for nine characteristics of SEM and ranked them in order of importance. Respondents preferred free, Internet-based, anonymous SEM portraying behaviours they would do. Cost and looks were the most important characteristics of SEM to participants, while condom use and sexual behaviours themselves were least important. Results suggest that while participants may have preferences for specific behaviours and condom use, these are not the most salient characteristics of SEM to consumers when choosing.

Introduction

Sexually explicit media (SEM) is defined as ‘any kind of material aimed at creating or enhancing sexual feeling or thoughts in the viewer, and containing explicit exposure and/or depictions of the genitals as well as clear and explicit sexual acts (e.g., vaginal intercourse, anal intercourse, oral sex, masturbation, bondage, etc.)’ (Hald & Malamuth, 2008). SEM is viewed by many men who have sex with men (MSM) and is widely accessible (Morrison, Morrison, & Bradley, 2007). The global SEM industry now generates an estimated $100 billion annually, of which $13 billion comes from the United States (Carroll et al., 2008). In the early 2000s, the US market share of SEM portraying sex between men was estimated to be 10–25% of all SEM (Rich, 2001; Thomas, 2000). By 2007, that estimate rose to 33–50% of SEM (Morrison et al., 2007). In the 2000s, however, bareback SEM production became more common, reportedly driven by consumer demand (Holt, 2008).

Currently, there is a growing body of research on SEM focusing specifically on its relation to sexual risk behaviour (Eaton, Cain, Pope, Garcia, & Cherry, 2011; Hald et al., in press; Peter & Valkenberg, 2011; Rosser et al., 2013; Sinković, Štulhofer, & Božić, 2012). In a study investigating the relationship between SEM consumption and HIV risk among MSM, researchers found that preferences for condom use in SEM reflected participants’ preferences for condom use in real life, both with insertive and receptive anal intercourse (Rosser et al., 2013). People who preferred SEM without condoms had higher numbers of unprotected anal intercourse (UAI) partners and people who preferred SEM with condoms had fewer UAI partners (Rosser et al., 2013). Prior research into specific characteristics of SEM has focused largely on portrayed behaviours (e.g., condom use, types of sexual acts) but has not investigated which other characteristics are relevant to consumers, such as cost and medium (Grudzen et al., 2009; Silvera, Stein, Hagerty, & Marmor, 2009).

MSM continue to be at disproportionately high risk for HIV infection (Centers for Disease Control and Prevention, 2010). While new biomedical approaches to HIV prevention are being introduced, condoms remain the best known and least expensive method for HIV prevention, as well as being highly effective (Centers for Disease Control and Prevention, 2013; Hall et al., 2008 ). In spite of this, barebacking remains a highly prevalent behaviour among MSM (Berg, 2009; Blackwell, 2008; Parsons & Bimbi, 2007). Estimates of prevalence vary geographically and range from 10% to 65% of adult US MSM reporting bareback sexual behaviour in cross-sectional, community-based studies (Huebner, Proescholdbell, & Nemeroff, 2006; Mansergh et al., 2002). In Internet-based studies, between 39.2% and 83.9% of MSM report bareback sex (Berg, 2008; Halkitis & Parsons, 2003). Many MSM report learning about sexuality from SEM, which suggests that SEM could potentially influence the condom use behaviours of consumers (Kubicek, Beyer, Weiss, Iverson, & Kipke, 2010; Kubicek, Carpineto, McDavitt, Weiss, & Kipke, 2010; Morrison, 2004; Mustanski, Lyons, & Garcia, 2011). The potential for SEM to increase condom use among MSM and to be a part of sex education merits consideration.

SEM could act as a medium in at least three ways; through sexually explicit public service announcements (PSAs) targeted towards MSM (e.g., on online sex-seeking sites), through sexually explicit educational materials tailored for MSM, or by promoting safer sex within SEM (e.g., through policies or self-governance by the SEM industry). There are several precedents of sex education being delivered via SEM (DCFUK!T, n.d. LaRue, 2010; Swedish Educational Broadcasting Company, 2012), and through the 1990s, the gay SEM industry imposed a self-governance requiring all anal sex to be depicted using condoms.

Though legislation requiring condom use in the production of SEM exists in some jurisdictions (e.g., the city of Los Angeles, California’s ‘Measure B’), these policies focus primarily on protecting the health of performers and not the preferences of SEM consumers (Los Angeles Times, 2012; Safer Sex in the Adult Film Industry Act, 2012). A common argument from people against regulations is that there is a strong consumer demand for SEM without condoms (del Barco, 2013; Los Angeles Times, 2012). There is little published empirical research to support or refute this claim, however. For those in the SEM industry, knowing the preferences of their audience is important. Finally, it is important for researchers studying SEM as it relates to fields such as sexual health and HIV prevention to know which characteristics are most salient.

In this paper, first, we describe the development of a new scale to measure preferences and the relative importance of nine characteristics of SEM. Second, we describe the preferences for characteristics of SEM among a diverse sample of Internet-using MSM noting differences across key demographic characteristics (i.e., age, race and HIV status).

Methods

Participants

Internet-using MSM completed online surveys about their use of SEM and sexual behaviour as part of a reliability study (N = 325) and main study (N = 1390). Many measures used in this survey, including the measures discussed in this manuscript, were developed by our team and were not previously tested for reliability. Thus, we conducted a 7-day test–retest reliability analysis before recruiting a larger sample for our main analysis. Participants in the reliability study were recruited online between January and February 2011, using banner advertisements on gay-oriented websites affiliated with an advertising agency specialising in gay consumers (see Appendix 1). A total of 448,472 impressions were displayed during this period and banners had a click-through-rate (CTR) of 0.31%. Participants in the main study were recruited using the same method between May and August 2011. For that campaign, banner advertisements were displayed on gay-oriented websites for a total of 7,939,758 impressions, with a CTR of 0.16%.

For both studies, banner advertisements directed interested persons to a webpage hosted on a dedicated university server with appropriate encryption to ensure data security. Persons were screened for eligibility, which were being male, having had sex with at least one man over the past 5 years, being 18 years of age or older, and living in the United States (including its territories). Participants in the reliability study were then given the same survey 7 days later to enable the study team to assess test–retest reliability. Participants in the main study answered survey items once. During analysis, we excluded participants who gave suspicious responses that were determined fraudulent (N = 64). To identify fraudulent responses, we used a deduplication protocol that included the following criteria: duplicate IP addresses, inconsistent responses to racial or ethnic identity, zip code and age, confirmed by asking these items multiple times in our survey (Konstan, Rosser, Ross, Stanton, & Edwards, 2006). In addition, we excluded participants with impossible numbers of sexual partners or nonsensical data patterns in responses to sexual behaviour questions. Next, in order to determine the potential impact of missing data, we conducted our analyses both with the full sample using pairwise deletion and including only completers using list-wise deletion and observed no statistically significant differences between estimates. Next, we used two-sample t-tests to measure differences between participants who completed the survey and participants who partially completed based on age (t = −0.84, p = .40), income (t = 0.45, p = .65), number of lifetime male sexual partners (t = −1.45, p = .15) and number of male sexual partners in the past 90 days (t = −1.58, p = .11) and found none of these to be statistically significant at p = .05, though non-completers reported higher mean numbers of both lifetime and 90-day male sexual partners (144.2 and 4.6, respectively) than completers (95.8 and 3.7, respectively). Thus, in order to maximise statistical power, we used a final analytic sample of 1390, with 287 partially complete surveys, handling missing data with pairwise deletion. Demographic characteristics for the final samples obtained are presented in Table 1.

Table 1.

Demographic characteristics of the samples in the reliability and main surveys.

	Reliability survey		Main survey
	n	%	n	%
N	325		1390
Race or ethnicity
White or Caucasian	288	88.34	604	41.46
Black or African American	13	3.99	467	11.46
Latino or Hispanic, any race	32	7.98	441	30.27
Asian or Pacific Islander	13	3.99	108	7.41
American Indian	9	2.76	25	1.72
Other/Multi	4	1.23	112	7.69
Age
18–25	102	31.38	544	37.34
26–35	105	32.31	456	31.30
36–45	57	17.48	217	14.89
Over 45	45	13.85	178	12.22
55+	16	4.92	62	4.26
HIV status
HIV-negative	270	82.82	1119	76.85
HIV-positive	20	6.13	133	20.40
HIV-unsure	36	11.04	204	14.01

Measures

In both studies, participants were asked about their preferences regarding broad characteristics of SEM as well as the importance of those characteristics (see Appendix 2). For the reliability study, participants were asked to choose between two anchors using a five-point semantic differential for each of eight characteristics: cost (free vs. for pay), production (amateur vs. professional), condom (safer sex vs. bareback), medium (online vs. offline), site type (member vs. anonymous), taboo content (things they would do vs. things they would not do), body type of performers (specific looks vs. a range of looks) and genre (vanilla vs. kinky). Participants were then asked to rate the same characteristics in terms of importance using a five-point scale, ranging (‘very important’ to ‘not at all important’).

For the main study, two changes were made to the measures. First, an additional characteristic, types of behaviour (a mix of everything vs. specific acts), was added to the list to yield a total of nine SEM characteristics. Second, the survey was modified to ask relative importance for each characteristic by ranking, rather than absolute importance, using a rating item. This was done because participants in the reliability study rated many characteristics equally important (which provided insufficient variance on which to compare items) and these items had low test–retest reliability (see Tables 2 and 3).

Table 2.

Test–retest reliability of preferences,^†a N = 325.

Characteristic	Weighted Kappa	Std. error	Z	Pr > Z
Cost	0.70	0.063	11.01	<0.0001
Production	0.66	0.0644	10.26	<0.0001
Condoms	0.70	0.0635	11.08	<0.0001
Medium	0.54	0.0645	8.35	<0.0001
Site type	0.63	0.0640	9.83	<0.0001
Taboo	0.55	0.0641	8.61	<0.0001
Actors’ looks	0.61	0.0644	9.47	<0.0001
Genre	0.73	0.0636	11.51	<0.0001

Notes:

^†7-day test–retest reliability.

^aRatings presented in a survey as cost (free vs. for pay), production (amateur vs. professional), condoms (safer sex vs. bareback), medium (online vs. offline), site type (member vs. anonymous), taboo (things you would do vs. things you would not do), Actors’ looks (generic looks vs. specific looks) and genre (vanilla vs. kink).

Table 3.

Test–retest reliability of importance,^† N = 325.

Characteristic	Weighted Kappa	Std. error	Z	Pr > Z
Cost	0.53	0.0645	8.17	<0.0001
Production	0.37	0.0641	5.82	<0.0001
Condoms	0.45	0.0644	6.98	<0.0001
Medium	0.36	0.0645	5.53	<0.0001
Site type	0.44	0.0645	6.87	<0.0001
Taboo	0.22	0.0628	3.55	0.0002
Actors’ looks	0.33	0.0625	5.23	<0.0001
Genre	0.27	0.0628	4.34	<0.0001

Note:

^†7-day test–retest reliability.

Analysis

Test–retest reliability was assessed using weighted Kappa statistics with quadratic weights (Cohen, 1968). To assess preferences, we compared measures of central tendency. For ranked items, mean, mode, median rankings, quartiles and interquartile ranges (IQRs) were used to sort the order of importance. The orders of items using median and mean ranks were very similar. However, because skewness existed in the distributions of several of these characteristics and mean ranks were very similar for many of the items, quartiles of ranks and IQRs were used as our final method to sort these items. First, items were ordered based upon 25th percentile, then in order of 50th percentile where there were ties in the 25th percentile responses and finally, then ordered based upon 75th percentile where ties existed in 50th percentile ranks.

For characteristics posed as semantic differentials, two-tailed t-tests were used to determine whether means reflected a consistent trend in preference across our sample. Means were tested against a null value indicating no preference (i.e., for a semantic differential on a scale from 1 to 5, the null value was 3). In the main survey, t-tests were statistically significant for nearly every characteristic (8 of 9), even when the differences between the mean value and null values were quite small, suggesting high power with little meaning. Where median and mode values equalled the null value while t-tests were statistically significant, these results were categorised as ‘possible artefacts of sample size’. To assess variability in responses across demographics, Pearson chi-square (χ²) tests of homogeneity were calculated to assess differences in responses to these items (both preferences between items and importance of items) across race, age and HIV status, indicated under Tables 4 and 5. Finally, in order to identify the magnitude of these associations while accounting for the large sample size, Cramer’s V was calculated, presented in Tables 6 and 7 (Cohen, 1988).

Table 4.

Preferences in main survey,^a N = 1365.

Characteristic	Preference	M (SD)	t	95% Confidence interval	Mode	Median
CostA	Free	1.35 (0.76)	−79.0**	1.31, 1.39	1	1
ProductionP	Neutral	2.89 (1.04)	−3.90*	2.83, 2.84	3	3
Condom useHP	Marginal	3.24 (1.22)	6.70**	3.16, 3.29	3	3
MediumA	Online	3.75 (1.19)	23.00**	3.68, 3.81	5	4
Site typeA	Anonymous	2.18 (1.14)	−26.89**	2.11, 2.23	1	2
TabooR	Would	2.40 (1.13)	−19.19**	2.35, 2.47	3	2
Actors’ looks P	Specific	3.43 (0.94)	16.76**	3.38, 3.47	3	3
BehaviourRAP	Mix	2.84 (1.16)	−5.07**	2.78, 2.90	3	3
GenreRHP	Neutral	3.01 (1.03)	0.39	2.96, 3.07	3	3

Notes:

^RIndicates p-value from χ² test of homogeneity across race <.05.

^AIndicates p-value from χ² test of homogeneity across age <.05.

^HIndicates p-value from χ² test of homogeneity across HIV status <.05.

^PIndicates possible artefacts of sample size (see Methods section).

^aRating items are presented in a survey as cost (free vs. for pay), production (amateur vs. professional), condom use (safer sex vs. bareback), medium (online vs. offline), site type (member vs. anonymous), taboo (things you would do vs. things you would not do), actors’ looks (generic looks vs. specific looks), behaviour (A mix of everything vs. specific acts) and genre (vanilla vs. kink).

^*Indicates p-value from t-test <.001.

^**Indicates p-value from t-test <.0001.

Table 5.

Importance in main survey, unordered N = 1310.

Characteristic	M (SD)	95% Confidence interval	Median rank	Mode rank
Cost	2.30 (2.32)	2.18, 2.43	1	1
Production	5.1 (2.37)	4.97, 5.23	5	2
Condom useR	5.83 (2.6)	5.69, 5.98	6	9
Medium	5.76 (2.59)	5.63, 5.91	6	9
Site type	5.24 (2.39)	5.11, 5.38	5	3
TabooA	6.18 (2.09)	6.07, 6.30	6	9
LooksA	4.02 (2.13)	3.91, 4.14	4	3
Behaviour	5.05 (2.01)	4.94, 5.16	5	4
Genre	5.86 (2.21)	5.74, 5.98	6	8

Notes:

^RIndicates p-value from χ² test of homogeneity across race <.05.

^AIndicates p-value from χ² test of homogeneity across age <.05.

Table 6.

Chi-squared and Cramer’s V between demographic characteristics and preference items N = 1310.

	Race			Age			HIV status
Characteristic	χ2	V	p-Value	χ2	V	p-Value	χ2	V	p-Value
Cost	30.61	.07	.06	30.02	.07	.02	8.09	.05	.43
Production	27.67	.12	.07	16.70	.06	.41	10.49	.06	.23
Condom use	24.47	.07	.22	21.00	.06	.18	48.23	.13	<.01
Medium	20.54	.06	.43	37.69	.08	<.01	8.06	.05	.43
Site type	24.01	.07	.24	39.13	.08	.08	11.03	.06	.20
Taboo	35.84	.08	.02	20.53	.06	.20	4.61	.04	.80
Looks	21.96	.06	.34	21.04	.06	.18	5.13	.04	.74
Behaviour	42.63	.09	<.01	29.41	.07	.02	734	.05	.50
Genre	34.66	.08	.02	20.18	.06	.21	36.08	.12	<.01

Table 7.

Chi-squared and Cramer’s V between demographic characteristics and rankings N = 1310.

	Race			Age			HIV status
Characteristic	χ2	V	p-Value	χ2	V	p-Value	χ2	V	p-Value
Cost	36.75	.07	.62	28.74	.07	.63	8.09	.05	.43
Production	34.45	.07	.72	33.38	.08	.40	10.49	.06	.23
Condom use	59.44	.10	.03	22.44	.06	.90	48.23	.13	<.01
Medium	34.03	.07	.74	19.87	.07	.95	8.06	.05	.43
Site type	32.46	.07	.80	25.65	.07	.08	11.03	.06	.20
Taboo	41.01	.08	.43	20.53	.06	.20	4.61	.04	.80
Looks	42.09	.08	.38	21.04	.06	.18	5.13	.04	.74
Behaviour	41.94	.08	.39	29.41	.07	.02	7.34	.05	.50
Genre	53.01	.09	.08	20.18	.06	.21	36.08	.12	<.01

Results

Preferred characteristics of SEM

Participants had the strongest preference for free SEM. Participants also preferred SEM that was available online, did not require a membership for viewing, depicted acts they would engage in and featured actors who had a specific looks and portrayed a mixture of behaviours (Table 4). Though t-tests were statistically significant for nearly every characteristic, median and mode values suggested that participants had strong preferences for only a few of these items (see Figure 1). For example, though the mean suggested a preference for SEM without condoms (M = 3.22, t = 6.7, p < .001), the median and mode responses were both equalled 3. Based upon the magnitude of difference in means from null value, the characteristics that respondents of our main survey had the strongest preferences for were cost, medium, taboo and type of site.

Figure 1.

Mean preferences for all items.

Preferences in cost varied across age (χ²(16, N = 1366) = 30.02, p = .018), though this association was weak (V = .07). Younger men (aged 18 and 24) reported the strongest preference for free SEM (M = 1.26, SD = 0.69), while older men (aged 55 and above) had the weakest preference for free SEM (M = 1.42, SD = 0.82); however, no age group preferred ‘for pay’ SEM. Preferences for condom use varied across HIV status, χ²(16, N = 1369) = 48.23, p < .001, and this association was moderately strong (V = .13). HIV-positive participants preferred SEM portraying sex without condoms (M = 3.85, SD = 1.19), while HIV-negative and HIV-unsure participants did not have a strong preference as a group (M = 3.14, SD = 1.21 among HIV-negative MSM, M = 3.27, SD = 1.26 among HIV-unsure). Thirty-five per cent of our main survey’s sample had no preference for or against condom use in SEM and only 25% ranked condom use among their top three most important characteristics of SEM (not displayed in tables).

Preferences for SEM had statistically significant variability across age, χ²(16, N = 1366) = 37.69, p = .002, though this association was weak (V = .07). No age group preferred offline SEM, however, the preference for online SEM was strongest among MSM aged 18–24 (M = 3.85, SD= 1.19), while MSM over 55 had the weakest preference between online and offline SEM (M = 3.46, SD = 1.22). With respect to online SEM, preferences between anonymous and membership sites showed statistically significant variability across age groups, χ²(16, N = 1366) = 39.13, p = .001, which was weak after accounting for sample size (V = .07). No age group preferred membership sites (range 2.07–2.56), however, the preference for anonymous sites was strongest among participants between the ages of 18 and 24 (M = 2.07, SD = 1.11).

Across all races and ethnicities, respondents slightly preferred SEM portraying acts they would engage in to acts they would not do (range of means for Taboo was 2.38–2.5). The differences in preferences were statistically significant χ²(16, N = 1366) = 35.84, p = .016, though it was weak (V = .08). All groups preferred to view SEM portraying acts they would engage in, with white MSM having the strongest preference (M = 2.38, SD = 1.09) and black MSM having the most neutral mean response (M = 2.55, SD = 1.09). Finally, differences in preferences for ‘genre’ of SEM were present between men of different HIV status, χ²(16, N = 1366) = 36.08, p < .001, which was a moderately strong relationship (V = .12). HIV-positive MSM had a mean response indicating a slight preference for ‘kinky’ SEM (M = 3.46, SD = 1.01) while HIV-negative and HIV-unsure participants had mean responses suggesting little to no group-level preferences between ‘kinky’ and ‘vanilla’ SEM (M = 2.95, SD = 1.01 and M = 3.07, SD = 1.07, respectively).

Importance of SEM characteristics

Cost was the most important characteristic to respondents, ranked as most important by 64% of our sample (Table 8). Its median and mode values (both equalled 1) also reflected this finding, along with having the smallest IQR. Aside from cost (M = 2.31, SD = 2.32), mean ranks remained close together (eight items with mean rankings between 4.02 and 6.30; see Figure 2). The importance of cost did not show significant variability across demographics.

Table 8.

Ranked order of items, sorted by quantiles, N = 1310.

Characteristic	25th percentile	50th percentile	75th percentile	Interquartile range
Cost	1	1	2	1
Looks	2	4	5	3
Production	3	5	7	4
Site type	3	5	7	4
Medium	3	6	8	5
Behaviour	4	5	7	3
Genre	4	6	8	4
Condom use	4	6	8	4
Taboo	5	6	8	3

Figure 2.

Mean rankings for all items.

The importance of condom use showed some variability across racial categories (Table 7); on average, condom use had the greatest importance among black MSM (M = 5.21, SD = 2.53) and was least important to other/multiracial MSM (M = 6.48, SD = 2.37). Though condom use was ranked as most important by black MSM, as a group they had no preference between viewing SEM with and without condoms t(152) = 0.53, p = .59 (not shown in tables). Among age groups, men over 55 ranked the category ‘taboo’ with greatest importance (M = 5.92, SD = 2.31) while men between the ages of 18 and 24 had the lowest mean rank (M = 6.23, SD = 2.12). On average, looks of performers was most important to men between the ages of 25 and 34 (M = 3.85, SD = 2.10), and least important to men over 55 (M = 4.71, SD = 2.27).

Discussion

Our study found that cost is the most important characteristic of SEM to this sample of Internet-using MSM. The cost and popularity of SEM have made it an industry with profits on par with Hollywood (Escoffier, 2003). An Internet search of online vendors of SEM in DVD format from major studios found many titles costing between $20 and $60 per DVD; given the popularity of SEM suggested by the revenue it generates, it would follow that consumers want to find SEM at the lowest cost (Channel 1 Releasing, 2012; Duroy, 2013; Titan Media, 2013). Free SEM is more commonly available online, so the high importance of cost offers one explanation of the preference for Internet-based SEM we found among our samples. This also suggests the possibility that consumer behaviour in SEM is similar to other markets (e.g., other media, clothing, books).

Several other features of Internet-based SEM may give it advantages to offline SEM in the eyes of consumers: new SEM can be obtained from a computer without travelling to a store or waiting for delivery, it is more convenient and accessible, there is a greater range of titles and latest releases and finally there are no physical materials (e.g., magazines, DVDs) to store, which could be advantageous to consumers concerned about privacy. In addition, this may be related to other trends in consumption of other media; the popularity of streaming technology for other media as such and movies and music has changed these industries, this may be the same for SEM. There are several plausible reasons why our sample preferred anonymous sites. First, many membership sites are for pay and many anonymous sites are free. Since cost was the most important characteristic of SEM to our sample and free SEM was preferred, this could lead to a preference for anonymous sites as well. Additionally, concerns for confidentiality, due to a desire to maintain anonymity and/or to protect the privacy of personal financial information could explain this preference.

A preference for free Internet-based SEM may present challenges for the wider MSM health movement. Producers of this type of SEM may not be regulated and it may also be self-generated, which could make engagement challenging for health providers. The popularity of free Internet-based SEM also implies a need for change in approach by the MSM health movement in health interventions within the SEM paradigm. Consequently, the MSM health movement needs to be agile in engaging this market.

Though many items varied statistically significantly across demographics, few of these differences seemed substantial; others may have been artefacts of sample size. For example, the difference across racial groups for preferences in the category ‘taboo’ was statistically significant, but the difference between the two most extreme means did not suggest a difference in preference that would lead to different choices in content between groups; it is possible that this may have been due to the highly different numbers of participants within each racial group. Similarly, after accounting for sample size, results suggested that most associations between these items and demographic characteristics were weak. Thus, these findings should be validated in further study, preferably with greater balance of sample sizes between demographic groups.

The general preference in taboo was for acts that one ‘would do’. Despite this preference, taboo was ranked as unimportant. A model of SEM consumption (the sexual risk behaviour model; Wilkerson et al., 2012) can offer a potential explanation. According to this model, sexual intentions change when SEM portraying new behaviours is found arousing by participants and sexual behaviours change when participants both find these new behaviours pleasurable in life and find available sex partners. In this situation, there would be little incentive for someone to continue viewing SEM portraying acts neither arousing nor pleasurable. Additionally, once a person reaches a point where they rarely see SEM portraying novel acts, the question of taboo could become less important, and viewing novel SEM less possible.

Perhaps the most surprising finding was that condom use was ranked so often as unimportant (mode rank = 9). Because such a strong emphasis is placed on condom use in health education strategies targeted towards MSM and there has been substantial cultural discourse about condom-less SEM from both health educators and the SEM industry, we expected condom use to have greater importance. Considering the variety of other characteristics included in this survey, it is possible that condom use is simply less salient than others such as cost, looks of actors and medium. Further exploration of this finding is a necessary next step for our research. In particular, examining previously established links between preferences for portrayal of condom use and condom use behaviours in life with studies of different populations and with designs which can assess temporality can explore the implications of this finding for the SEM industry and HIV prevention (Rosser et al., 2013).

These data came from a cross-sectional survey; having only surveyed these individuals once, inferences about long-term patterns of behaviour cannot be made. Additionally, our use of an Internet-based sample may have led to information bias; the strong preference for online SEM among our respondents may be due to a stronger preference for Internet-based media in general, as a part of being Internet-using MSM. In addition, recruiting a convenience sample from the Internet limits the external validity of our results and their generalisability to other populations, especially non-Internet users and MSM who do not access gay-themed websites. In addition, men who go online specifically to use SEM or engage in riskier sexual behaviours may have been less likely to participate. Though the difference was not statistically significant, non-completers of the main survey had higher 90-day and lifetime male sexual partners.

We did not collect test–retest reliability data for the rankings of items that we used as our final format in the main study. Thus, our results rely on the untested assumption that rankings were a more reliable and better way to measure importance than the rating method we used initially. Alternative methods, such as the ‘Q’ method of sorting, were not used, but may have been (more) appropriate to assess preferences and importance of these characteristics of SEM. However, other studies have found rankings to be superior to ratings at finding distinctions between measures of importance (Alwin & Krosnick, 1985). Our data, then, identify relative importance, but we do not have data from the main survey on the absolute importance of each item. Finally, all results were group-level inferences using measures of central tendency, caution should be taken in translating these group-level response patterns into individual-level preferences.

These findings set up the potential for several future studies. First, researchers can examine these characteristics in other populations (e.g., heterosexuals, women, non-Internet using MSM) and determine if and how preferences and importance differ. Market researchers in the SEM industry could determine the impact of preferences and importance on SEM purchasing and viewing behaviour. Finally, researchers in HIV prevention can examine the relationship between preferences and importance of characteristics of SEM and sexual risk behaviours with more rigorous study designs (e.g., longitudinal studies).

We presented information about the preferences of this sample of Internet-using MSM. SEM is a highly eclectic medium with a variety of content, and though preferences vary greatly between individuals, cost was consistently the most important characteristic to our sample overall, followed by looks of the actors, production and non-membership site. Based on these results, if one could describe the characteristics of SEM that would maximise satisfaction on those characteristics most important, it would be a free video available on the Internet, portraying actors with specific looks on a site that can be accessed without membership. After meeting those criteria, whether or not the sexual acts portrayed by performers are risky is a distal consideration. Condom use was unimportant to our sample overall and a sizable proportion had no preference for or against portrayal of condom use are in SEM. While preferences vary between individuals, these results suggest that risky sex is generally not one of the first characteristics considered when choosing SEM. For health educators and others using SEM for HIV prevention, it is imperative to consider aspects beyond behaviours to reach a target population.

Appendix 1. Reliability study

Site	Clickthrough rate	Impressions
AGB-Style	2.27	17
Another Guy Blog	0.44	2613
Ask Gay Men	25	2
BEARNATION.us – Social Network	2.08	376
BIZ	11.11	9
Back2Stonewall.com	2.67	75
Ben and Dave’s Six Pack	10	14
BigJock	0.89	48
Black Gay Gossip	0.27	724
Break the Illusion	0.46	631
Bryanboy: Le Superstar Fabuleux	0.34	130
Charmants	0.44	229
ChicagoPride	0.64	157
Connexion	1.61	15,653
Cruising Gays	1.32	1257
Dlist	0.6	2671
Dailymotion	0.31	456
Deep Dish	0.74	203
FindFred	0.67	10,306
Gay Authors	1.12	89
GayCities	0.72	3849
Grab Magazine	16.67	6
Hit Dan Back	0.13	427
Homorazzi Media	0.2	3612
Homotron	2.7	37
Homotrophy Gay Blog	0.21	3000
JoeMyGod	2.89	3447
JustGuys	0.37	9224
LA Rag MAg	0.22	1635
Labidos	1.32	227
MANofAUSTIN	1	180
Manjam	0.6	7589
Mark’s List	0.22	273
OHBoyMagazine	9.09	3
OkCupid – Gay	0.06	1678
On Top Magazine	0.13	2258
OneGoodLove – Gay	0.75	12
Out In America Cities Network	0.8	3381
OutLoudBlogs	1.57	366
Outsports	0.68	7564
Pink Kryptonite	1.54	195
Planet Homo	0.88	46
PopWired	0.75	103
Project RunGay – Tom and Lorenzo	0.27	28,732
Qnotes	9.09	22
Queerty	0.45	7339
ROD 2.0	0.85	485
RealJock	0.3	28,232
SportsFags	0.16	1259
Tabloid Heat	0.07	7
Tabloid Prodigy	0.07	8595
TangoWire – Gay	1.89	359
Tap That Guy	0.45	287
The DataLounge	0.17	11,782
The Gay Youth Corner	0.88	5549
The New Civil Rights Movement	0.62	856
The New Gay	0.59	417
Union Cafe	7.06	16
WickedGayBlog	0.19	5473
Windy City Media Group	0.77	1681
World of Wonder – WOW Report	0.26	3458
doorQ	2.5	80
qPDX: The Queer Northwest	2	134
the Celeb Archive	0.44	7764
wayoutwest.tv	0.2	1013
Main study
A Bears Life Magazine	2.04	13
AGB-Style	3.33	101
AKA William	2.18	340
ATLANTAboy	0.76	131
Another Guy Blog	1.17	3037
Antitwink	8.85	56
Antivirus Magazine Greece	2	50
Ask Gay Men	1.32	2
BEARNATION.us – Social Network	5.14	142
BGay.com	5.96	94
Back2Stonewall.com	2.6	271
Ben and Dave’s Six Pack	4.05	74
Best Gay News Magazine	2.28	279
BigJock	3.76	60
Blabbeando	14.29	123
Black Gay Gossip	0.76	856
Body and underwear Model	1.49	3225
BoyGush	40.95	31
Break the Illusion	0.86	3239
BuskFilms	0.76	131
ChicagoPride	3.97	7
Citizen Crain	1.68	119
Click Click Expose (LGBT Media)	19.31	38
Connexion	2.61	86,832
Costa Rica Gay Map	1.45	69
DRAMA DUPREE	0.65	440
DaSeekah	0.42	3577
Daddyhunt	0.53	135,408
David Dust	1.86	325
Deep Dish	0.38	762
Derek and Romaine	9.09	8
Easy Gay Life	20	51
Equalitopia	4	15
Fierth Magazine	2.14	200
FindFred	7.4	40,102
FocusBoy	1.57	184
G.I.R.L. – GayInternetRadioLive.	0.2	5366
GAN – Gay Ad Network	16.67	3
GUIDETOGAY.COM	8.25	82
Gay Authors	4.29	1001
Gay Cruising & Travel	5.98	2636
Gay Indo Forum	1.97	19,081
Gay List Daily	0.45	221
Gay Mexico Map	0.94	6
Gay Party List	2.86	2
Gay Rights Watch	1.49	6
GaySocialites	6.5	223
Gaycast	1.47	238
GaydarGuys	0.52	525
Going Nowhere Queerly	100	2
Golden Girls Forum	0.54	249
Good As You	2.02	964
HIVnet.com	0.13	1
Hit Dan Back	0.65	2016
Homorazzi Media	2.45	2449
Homotography	1.93	12,166
Homotron	1.03	15
Indusgay.com	3.23	41
Instinct Magazine	1.91	1083
InterstateQ.com	1.54	36
Joe.My.God.	3.75	49,714
JustGuys	2.73	39,831
LEATHERPOINT	0.8	22
LGBT News Agency	5.26	6
LGBTQ Nation	7.79	146
Lambda Literary Foundation	2.08	545
Lanzarote Gay Guide	1.45	69
Le Fag	2.6	117
MANofAUSTIN	0.44	452
Manjam	3.32	38,518
Mark’s List	4.12	27
Meet Gay Couples	3.28	76
Meet Gay Professionals.com	21.28	78
MegaMates Men	34.27	10
Michi & Michi	16.67	85
My Fabulous Disease	6.07	72
Nighttours	0.93	7
OHBoyMagazine	20	28
OUTTAKE BLOG™	25	4
OUTview Online	0.11	86
Obama and the Gays	33.33	3
On Top Magazine	2.56	3065
One More Lesbian	0.07	4941
OneGoodLove – Gay	2.28	9
Out In America Cities Network	1.68	9183
OutLoudBlogs	3.75	62
OutTonight	9.58	16
Outsports	5.77	2897
Pams House Blend	1.35	74
Perfect Beat	1.52	324
Petrelis Files	4	103
Pink Kryptonite	10.26	90
Planet Homo	10.84	34
PopWired	0.46	83
Project Q Atlanta	1.33	2946
Project RunGay – Tom and Lorenzo	0.29	105,146
Provincetown Live	22.22	9
QNotes	0.96	24
Queerlife	4.67	177
ROD 2.0	2.46	785
RealJock	23.46	757
Romanian Gay News Blog English	0.24	316
Rosie O’Donnell	0.35	15
Seattle Gay Scene	3.81	160
Sexy Men of Sports	0.25	12
SportsFags	0.85	2850
StiriGay.ro – Romanian Gay News	6.27	2
StudStop.com	19.87	118
THE QIT	0.98	94
Tabloid Prodigy	1.3	4
Tap That Guy	0.58	2079
The Beat San Francisco	7.64	42
The Bilerico Project	2.83	2706
The DataLounge	0.17	120,647
The Drag Queen Posse	6.67	15
The Gay Youth Corner	0.86	20,120
The Georgia Voice	1.56	6
The Gist	0.53	92
The Mad Professah Lectures	0.35	1152
The New Civil Rights Movement	2.12	2843
The New Gay	31.22	520
The Pretty Boys Club	0.85	118
The Queer Village	1.72	374
The Seafront Diaries	3.8	1
This Is FYF	2.88	15
Thought Theater	4	3
Top to Bottom	3.06	81
Unicorn Booty	2.55	5772
Union Cafe	8	7
Up Up and A Gay	13.89	36
VGL	1.49	2722
Velvet Dice Bag	11.55	47
WhatsTheT	0.97	589
WickedGayBlog	0.74	4196
Windy City Media Group	3.27	452
World of Wonder – WOW Report	0.65	1
doorQ	4.02	561
gaelick	0.68	294
gayborhood.tv	36.87	14
glbtq Encyclopaedia	2.72	9
homo-neurotic.com	1.09	366
qPDX: The Queer Northwest	1.69	173
the Celeb Archive	7.69	43
the L word Fan Site	0.36	2602
wayoutwest.tv	7.27	1439

Appendix 2. Items used in survey

The following items ask about different types of pornography that you can access. Please indicate your preferences for the following options. Numbers closer to a description indicate more preference for that description. A number in the center means no preference for either. Here is an example:

	1	2	3	4		5
Red					Blue
Selecting ‘2’ means more preference for Red than Blue.
When searching for pornographic materials, do you prefer?
	1	2	3	4		5
Free					For pay
	1	2	3	4		5
Amateur					Professional
	1	2	3	4		5
Safer sex					Bareback
	1	2	3	4		5
Offline					Online
	1	2	3	4		5
Anonymous site					Membership site
	1	2	3	4		5
Actors doing things you would do					Actors doing things you wouldn’t do
	1	2	3	4		5
Generic looks					Specific looks
	1	2	3	4	5
A mix of everything					Specific acts
	1	2	3	4	5
Vanilla					Kinky
Is there anything else that you prefer when searching for pornographic materials?
	1	Yes
	2	No
	−99	RTA

Please specify what else you prefer to see:_______________

In the last question, you provided information on your preferences when searching for porn. We are also interested in how important those preferences are when you make a decision about which porn to use. Please rate the following characteristics on a scale from one (1) to nine (9), with one (1) being ‘most important’, and nine (9) being ‘least important’.

Cost (free vs. for pay)

Production (amateur vs. professional)

Condom use (safer sex vs. bareback)

Medium (offline vs. online)

Site type (anonymous vs. membership)

Taboo (actors doing things I would do vs. actors doing things I would not do)

Actors’ looks (generic vs. specific)

Behaviour (a mix of everything vs. specific acts)

Genre (vanilla vs. kinky)

[Other]