Abstract
Many sexual minority men (SMM) use sex position self-labels, including “top,” “bottom,” “versatile,” as well as combinations like “vers/top” and “vers/bottom” to describe engagement in anal intercourse. Despite the use of these labels, there is limited literature identifying various aspects that may impact SMM’s chosen sex position label. The purpose of this study was to examine factors (i.e., sexual behavior, participants’ anthropomorphic characteristics, and attraction to partners’ attributes) associated with self-reported sex position label use. Participants (n = 4,802) were cisgender SMM from a U.S. nationwide cohort exploring missed opportunities for HIV prevention. The main analyses used iterative logistic regression models to predict self-reported sex position labels based on sexual behavior, participants’ anthropomorphic characteristics, and attraction to partners’ attributes. A higher proportion of participants identified as bottom/versatile (24%) followed by top/versatile (22.5%), versatile (20.3%), top (17.2%), and bottom (15.8%). Sexual attraction, behavior, and anthropomorphic characteristics poor-to-moderately predicted single sex position label categories. However, when considering near-adjacent sex positions (e.g., top/versatile as adjacent to top), model prediction improved substantially, indicating variability in self-labels and factors associated with each label. Sex positionality is complex and requires multidimensional measurement to fully understand this aspect of an individual’s sexuality.
Keywords: Sex position, sex role, attraction, sexual behavior, sexual minority men, SMM
Introduction
Anal sex position self-labels or sexual self-labels (Hart et al., 2003) are well-defined among sexual minority men (SMM) (Dangerfield et al., 2017). Throughout the United States (U.S.), these labels refer to people who identify as “tops” (perform anal insertive role during intercourse), “bottoms” (perform anal receptive role during intercourse), and “versatile” (perform both anal insertive and receptive roles during intercourse). In other regions, such as throughout Europe and Latin America, terms such as “active” and “passive” are used to describe people who identify as “tops” and “bottoms,” respectively. Previous work illustrates the majority of SMM adopt sex position self-labels (Moskowitz & Roloff, 2017) and that these self-labels are highly correlated with enacted sexual behavior (Dangerfieldet al., 2017; Moskowitz et al., 2022; Moskowitz & Hart, 2011; Moskowitz et al., 2008) and vulnerability to acquiring or transmitting human immunodeficiency virus (HIV) and other sexually transmitted infections (STIs) (Dangerfield et al., 2017), thereby underscoring the importance of sexual self-labels for public and population health interventions among SMM. While much of the research describing sexual self-labels among SMM has assessed the association of these positions with vulnerability to HIV or other STIs (Dangerfield, Ober, et al., 2018; Grov et al., 2007; Newcomb et al., 2015; Parsons et al., 2005; Smith & Amaro, 2021; Van de Ven et al., 2002), recent work has sought to describe the influence of various factors on sexual self-label use among SMM.
Multiple personal (e.g., physical attributes, sexual experiences), practical (e.g., physical discomfort), relational (e.g., partner type, relationship power dynamics), and sociocultural (e.g., perceptions of masculinity and gender) factors influence both sex position self-labels and sexual behavior (Dangerfield et al., 2017; Johns et al., 2012; Moskowitz & Roloff, 2017; Pachankis et al., 2013; Swift-Gallant et al., 2018, 2019). When examining personal physical attributes (i.e., body weight, height, muscularity, penis size, hairiness, and masculinity), Moskowitz and Hart (2011) found only penis size and masculinity to be statistically significantly correlated with sex position self-labels. Further, sex position self-labels may also elucidate contexts about sex, gender roles, and power dynamics during sexual encounters that reinforce heteronormative stereotypes (Moskowitz & Roloff, 2017a). Research has suggested that some SMM reinforce stereotypes about gender roles in sexual encounters, such that “tops” are perceived as more masculine and “bottoms” are ascribed as more feminine (Carballo-Diéguez et al., 2004; Johns et al., 2012; Kippax & Smith, 2001; Moskowitz & Hart, 2011). For instance, a qualitative study among male couples discussed “tops” and “bottoms” through a gendered lens, whereby participants described receptive partners as more effeminate (Kippax & Smith, 2001). Similarly, in their qualitative study of young SMM, Johns et al. (2012) described how heteronormative constructs were commonly associated with sex position self-labels among participants. In a cross-sectional study of 294 Latine SMM, participants who perceived their partners to be more masculine (e.g., taller, larger penis, more handsome) were more likely to bottom, whereas Latine SMM who perceived their partners to be more feminine (e.g., shorter, smaller penis, less aggressive) were more likely to top (Carballo-Diéguez et al., 2004).
Yet, despite the influence of multiple factors on sex positionality, research has often ascertained sex position self-labels with a single item, such as “which sexual position do you identify most as?” Respondents are asked to choose from options including top, versatile top, versatile, versatile bottom, bottom, other, not applicable, or prefer not to respond. In a recently published analysis using 530,737 profiles from a SMM dating website, Smith and Amaro (2021) found that 36.5% of participants identified as “tops,” 33.8% as “bottoms,” and 29.7% as “versatile.” Though this measurement method may provide researchers with self-reported anal sex labels (Dangerfield et al., 2017), it does not capture the considerable influence of various individual and partner characteristics that may contribute to one’s sex position self-label.
Given there are multiple factors that could impact one’s chosen sexual role, the purpose of the current study was to explore characteristics related to sex position self-labels in a U.S. national sample of cisgender SMM. First, we describe the sample and assess sociodemographic differences between participants using different sex position self-labels (i.e., top, top/versatile, versatile, bottom/versatile, and bottom). Then, we use self-identified sex position labels, behavior (engagement in insertive vs. receptive anal intercourse), personal anthropomorphic characteristics (height, weight, penis length, and penis girth), and attraction to partners’ physical characteristics (age, height, weight, muscularity, hairiness, butt size, penis size, masculinity) to better describe sex positionality using iterative logistic regression models. The aims were to identify factors associated with sex positionality and highlight the importance of multi-mode measurement, and to situate this within behavioral and health contexts.
Method
Participants
Data used for the current analyses were from the Together 5,000 (T5K) cohort that aimed to identify modifiable factors associated with HIV seroconversion among men, transgender (trans) men, and trans women who have sex with men. Enrollment for this U.S. national, internet-based cohort was conducted between October 2017 and June 2018 using advertisements on geosocial networking phone applications. The cohort and study procedures have been described fully elsewhere (Grov et al., 2019; Nash et al., 2019). Briefly, participants were eligible for inclusion in the study if they were 16–49 years old; reported at least two male sex partners in the past three months; were not currently participating in an HIV vaccine or pre-exposure prophylaxis (PrEP) clinical trial; were not currently on PrEP; lived in the U.S. or U.S. territory; were not known to be HIV-positive; had a gender identity other than cisgender female; and reported at least one additional behavior associated with increased vulnerability for HIV.
Participants that clicked one of the study advertisements were routed to a secured informed consent page and brief enrollment survey (non-incentivized) that asked about demographic characteristics, sexual behavior, and HIV or PrEP status. Those eligible were sent an online survey that included questions on topics such as psychosocial measures, relationship information, and substance use—6,253 participants completed it and received a $15 incentive. Twelve months following enrollment, participants were asked to complete another survey ($25 incentive). Many measures of interest for the present analyses were added to the 12-month follow-up survey; therefore, this study included data from participants for whom we had baseline and 12-month data on all measures of interest (n = 4,850). We further restricted this sample to include only those participants who identified as a cisgender male and did not identify as heterosexual (n = 4,802). T5K was approved by the Institutional Review Board of the institution of the corresponding author.
Measures
Constructs of interest for the present analysis were self-reported sex position label, attraction to characteristics of potential sex partners, self-reported sexual behavior, as well as self-reported anthropomorphic characteristics. Figure 1 helps to illustrate measured variables as they relate to each construct of interest (measured or unmeasured) and overall sex positionality. We chose to extend our analyses to include anthropomorphic characteristics due to the practical and physical considerations people may have when engaging in different sex roles, as reported in the literature (Grov et al., 2010; Moskowitz & Hart, 2011).
Figure 1.

Conceptual model of sex positionality illustrating measured variables’ relationship with measured and latent constructs of interest (i.e., identity, behavior, attraction, and anthropomorphic characteristics), Together 5,000, 2018–2019.
Constructs of sexual self-labels
Sexual self-labels.
Sex position identity was assessed by asking participants, “Which sexual position do you identify most as?” Participants could choose between top, versatile/top, versatile, versatile/bottom, bottom, or not applicable.
Attraction characteristics.
Participants were asked a series of questions about what they found attractive in male sexual partners (see Table 1 for measures). Attraction characteristics were assessed by asking participants, “for each of the following traits, please use the slider to indicate your preferences in male sexual partners.” For traits of weight, muscle definition, height, presence or absence of body hair, butt size, penis size, and masculinity, participants were asked to rate their preferences on a scale of 1 to 5. For example, participants could indicate preferences for shorter (1), average height (3), or taller (5) partners. As another example, they could also indicate a preference that their partner or potential partners have a smaller penis (1), average penis (3), or larger penis (5). For a final example, participants could indicate if they preferred potential partners who were feminine acting (1), masculine but “obviously gay” acting (3), or very masculine or “straight” acting (5).
Table 1.
Measurement of attraction to partner traits, Together 5,000 study (2017–2022).
| Characteristic | Lowest Value | Highest Value | ||||
|---|---|---|---|---|---|---|
|
| ||||||
| Weight | 1 Thinner/ skinnier | 2 | 3 Average weight | 4 | 5 Bigger guys | No preference |
| Muscle | 1 Very little muscle definition | 2 | 3 Average muscle definition | 4 | 5 A lot of muscle definition | No preference |
| Height | 1 Shorter | 2 | 3 Average height | 4 | 5 Taller | No preference |
| Hair | 1 No body hair/ smooth | 2 | 3 Average body hair | 4 | 5 A lot of body hair | No preference |
| Butt size | 1 Smaller ass | 2 | 3 Average age | 4 | 5 Big ass/bubble butt | No preference |
| Penis size | 1 Smaller penis | 2 | 3 Average penis | 4 | 5 Larger penis | No preference |
| Masculinity | 1 Feminine | 2 | 3 Masculine/ “obviously gay” | 4 | 5 Very masculine/ “straight-acting” | No preference |
Scores ranging 1 to 5 were recategorized to above average (score 4–5), average (score 3), and less than average (score 1–2) for the present analysis. For all potential partner characteristics, participants had the option to choose “no preference.” Additionally, responses to these measures were dichotomized by agreement or non-agreement with the modal response within each position label. For example, among “tops,” the mode of attraction to a partner’s body weight was less than average; thus, if a participant identified as a “top” and reported attraction to a partner that was less than average body weight, they would be considered concordant with the modal response. For any participants who identified as a “top” and provided a preference of average or above average, they would be considered discordant to the modal response. By dichotomizing these reported attractions to partner’s characteristics, we were able to examine the level of variation by reported sex position label.
Sexual behaviors.
Anal sexual behavior was assessed by asking participants how many times they had insertive or receptive anal sex (without a condom, as the study from which data were derived was focused on HIV prevention) in the past 3 months. Participants were allowed to enter a number from 0 to 90+ times.
Anthropomorphic characteristics.
Finally, we assessed participants’ own anthropomorphic characteristics—i.e., height, weight, penis length (erect) and penis girth (erect). Weight was self-reported in pounds and height was self-reported in feet and inches. Penis length and girth (erect) were assessed via self-reported size with diagrams provided to aid in correct measuring, in accordance with how this was assessed in a previous study (Grov et al., 2010). Penis length and girth were reported to the closest 0.5 inch but had substantial missing data (12.3% and 52.4%, respectively); thus, we imputed both values when available using common missing data imputation strategies (Tabachnick & Fidell, 2011; Kline, 1998). Missing girth values were replaced with the average girth value among participants who had +/− 0.5 inches of that participant’s reported penis length. Similarly, missing length values were replaced with the average length of participants who had +/− 0.5 inches of that participant’s girth. In our available data (considering those participants with both length and girth measurements), we found a statistically significant, moderate positive correlation between length and girth (Pearson correlation = 0.34, p<0.0001; Spearman correlation = 0.33, p<0.001). We also found that the odds of missing data on penis length and girth were lower among those with greater length and girth values (results not presented). Therefore, we felt it was more accurate to apply the mean girth of those participants with a similar length (+/− 0.5 inch of that participant), to better represent this correlation and reduce overestimating either value based on available data. For example, if a participant was missing penis girth but reported a penis length of 6 inches, the missing girth value was replaced with the average girth among all participants with a reported penis length of 5.5–6.5 inches. Participants who were missing both penis girth and length could not be imputed.
Other measures of interests
In addition to the main construct measures above, demographic factors were assessed to describe the sample and determine their association with the construct variables of interest. These factors were age (continuous), race/ethnicity (Black/African American, White, Latino or all other/multiracial or multiethnic), sexual orientation (gay, queer, or homosexual; or bisexual), highest level of education (less than high school diploma, high school diploma or GED, some college or technical school training, college graduate or more), employment status (full-time, part-time, working or full-time student, unemployed/other), transactional sex in the past 12 months (yes/no), and currently in a main partnership (yes/no) at the 12-month follow-up.
Analyses
Participant characteristics were summarized using descriptive statistics across sex position identity labels (top, top/versatile, versatile, bottom/versatile, or bottom). Differences across categories were identified with Chi-squared (Fisher’s Exact for small counts) and Kruskal-Wallis tests. Characteristics included demographics (age, race, gender, sexual orientation, highest level of education, employment status, transactional sex in the past 12 months), anthropomorphic measures (height, weight, penis size), sexual behavior (insertive sex, receptive sex, relationship status) and attraction metrics (more feminine vs. more masculine, tall vs. short, bigger body size vs. smaller body size, larger penis vs. smaller penis, larger butt vs. smaller butt). To be thorough, we also created descriptive profile summaries for each participant’s responses on all construct variables (less identity)—attraction preferences, behavior, and anthropomorphic characteristics—to assess the frequency with which certain patterns emerged in our data. These data are not presented in detail here but demonstrated few identical and frequently repeated profiles indicating that people varied often within and across sex position labels for attraction preferences, reported sexual behavior, and their own anthropomorphic characteristics (online Supplemental Analysis 1).
Our primary analysis aimed to explore the relationship of attraction preferences, sexual behavior, and anthropomorphic characteristics with self-reported sex position label using a series of logistic regressions. First, we used self-reported sex position to estimate the coefficients of association with all construct variables. We then used these coefficients to predict sex position for each participant. Each sex position was modeled as a binary outcome: that position category compared to all other position categories, resulting in a total of five logistic regressions (e.g., top vs. all other categories, bottom vs. all other categories). The model that provided the highest predicted probability for a specific category was deemed the predicted position. We assessed model performance in terms of predicted probability accuracy (i.e., exact and near) with self-reported label. Accuracy was defined as the proportion of each predicted sex position category that matched a participant’s self-reported sex position. Near-accurate was defined as the proportion of each sex position category that was correctly predicted or was a neighboring category to what the participant reported. For example, a neighboring category for participants indicating that they were “tops” was top/versatile, while a neighboring category for participants who indicated “versatile” was top/versatile and bottom/versatile. We considered adjacent categories in addition to singular sex position label categories based on our understanding of the flexibility of sex position identity and associated behaviors (Moskowitz et al., 2008; Pachankis et al., 2013). That is, in near-accurate models that considered adjacent categories, we assume that certain adjacent sex position self-labels have similar behaviors or may not be meaningfully different. For example, we assume that participants who identify as “versatile/top”, “versatile”, or “versatile/bottom” engage in both receptive and insertive anal intercourse. However, we do expect that within these three versatile categories there will be variability in behavior in that some participants may have, for example, more insertive anal intercourse while others have less.
As a final step to examine model performance, we repeated the analysis and implemented a training (random sampling, by sex position, of 67% of participants) and testing (remaining 33%) approach to ensure more stable accuracy measures that did not contribute to spurious results. All analyses were performed in SAS software (version 9.4) (SAS Institute Inc., 2023) and R (version 4.3.1) (R Development Core Team, 2021) and the significance level was set to 0.05.
Results
Description of the sample
Overall, half the participants in this study were White (53.8%) and most identified as gay, queer, or homosexual (86.1%). The mean age was 31.9 (SD = 8). Large proportions of participants reported having some college or technical school training (44.7%) and working full-time (68.8%). More than half (63%) reported that they were not in a main partnership at baseline. Thirteen percent of participants reported engaging in transactional sex in the past year (see Table 2).
Table 2.
Overall demographic characteristics for analytic sample and stratified by sexual position, Together 5,000, 2018–2019, n = 4802.
| Demographics | Total n = 4802 | Top n = 837 | Top/ Versatile n = 1076 | Versatile n = 977 | Bottom/ Versatile n = 1151 | Bottom n = 761 | P-value* |
|---|---|---|---|---|---|---|---|
|
| |||||||
| Sexual Position Identity N(%) | 837 (17.2) | 1079 (22.5) | 977 (20.3) | 1152 (24) | 761 (15.8) | NA | |
| Race and Ethnicity | |||||||
| White | 2585 (53.8) | 423 (50.5) | 580 (53.9) | 513 (52.5) | 672 (58.3) | 397 (52.2) | <.0001 |
| Black/African American | 478 (10) | 132 (15.8) | 105 (9.7) | 117 (12) | 58 (5.0) | 66 (8.7) | |
| Latino | 1156 (24) | 185 (22.1) | 285 (26.5) | 203 (20.8) | 287 (25) | 196 (25.8) | |
| All Other/Multiracial or multiethnic | 583 (12.1) | 97 (11.6) | 106 (9.9) | 144 (14.7) | 134 (11.6) | 102 (13.4) | |
| Sexual Orientation | |||||||
| Gay, Queer, or Homosexual | 4137 (86.1) | 672 (80.3) | 881 (81.8) | 840 (86) | 1035 (89.9) | 709 (93.2) | <.0001 |
| Bisexual | 630 (13.3) | 157 (18.8) | 185 (17.2) | 125 (12.8) | 111 (9.6) | 52 (6.8) | |
| Other | 35 (0.7) | 8 (0.96) | 10 (0.93) | 12 (1.2) | 5 (0.4) | 0 (0.0) | |
| Highest Level of Education | |||||||
| < High school diploma | 100 (2.1) | 20 (2.4) | 13 (1.2) | 20 (2.1) | 29 (2.5) | 18 (2.4) | 0.0372 |
| High school diploma or GED | 597 (12.4) | 93 (11.1) | 120 (11.1) | 126 (12.9) | 142 (12.3) | 116 (15.2) | |
| Some college or technical school training | 2145 (44.7) | 357 (42.7) | 494 (45.9) | 423 (43.3) | 519 (45.2) | 352 (46.2) | |
| College graduate or more | 1960 (40.8) | 367 (43.9) | 449 (41.8) | 408 (41.8) | 461 (40.0) | 275 (36.1) | |
| Employment Status (current) | |||||||
| Full-time (40 hours per week) | 3302 (68.8) | 628 (75.0) | 741 (68.9) | 646 (66) | 784 (68.1) | 503 (66.1) | 0.0016 |
| Part-time (less than 40 hours per week) | 557 (11.6) | 86 (10.3) | 131 (12.1) | 130 (13.3) | 128 (11.1) | 82 (10.8) | |
| Working or full-time student | 279 (5.8) | 30 (3.6) | 64 (5.9) | 61 (6.2) | 78 (6.8) | 46 (6) | |
| Unemployed/Other | 664 (13.9) | 93 (11.1) | 140 (13.1) | 140 (14.3) | 161 (14.1) | 130 (17.1) | |
| Transactional sex in the past 12 months | |||||||
| No, not in the last year | 4182 (87.1) | 731 (87.3) | 922 (85.7) | 864 (88.4) | 1004 (87.2) | 661 (86.9) | 0.8769 |
| Yes, in the last year | 620 (12.9) | 106 (12.7) | 154 (14.3) | 113 (11.6) | 147 (12.8) | 100 (13.1) | |
| Relationship Status (in a main partnership) | |||||||
| No | 3024 (63.0) | 555 (66.3) | 651 (60.5) | 574 (58.8) | 736 (64) | 508 (66.8) | 0.0007 |
| Yes | 1778 (37.0) | 282 (33.7) | 425 (39.5) | 403 (41.3) | 415 (36.0) | 253 (33.3) | |
| Demographics (mean (SD)) | |||||||
| Age | 31.9 (8.0) | 34.1 (7.9) | 32.2 (7.7) | 30.9 (7.7) | 30.9 (7.5) | 31.6 (8.2) | <.0001 |
| Height (inches) | 70.4 (3.0) | 70.4 (3.0) | 70.4 (3) | 70.3 (3.1) | 70 (3.3) | 70 (3.1) | <.0001 |
| Weight (lbs) | 197.7 (53.3) | 199.7 (49.2) | 198.1 (49.7) | 195.6 (53.7) | 196.5 (55.1) | 201 (61.0) | <.0001 |
| Penis length (inches) | 6.9 (1.2) | 7.1 (1.0) | 7 (1) | 6.9 (1.1) | 6.5 (1) | 6.3 (1.1) | <.0001 |
| Penis girth (inches) | 4.9 (1.2) | 5 (1.1) | 5 (1.0) | 4.8 (1.1) | 4.7 (1.1) | 4.5 (1.2) | <.0001 |
| Penis length (inches) after imputation** | 6.9 (1.2) | 7.1 (1) | 7 (1.0) | 6.9 (1.1) | 6.5 (1) | 6.3 (1.1) | <.0001 |
| Penis girth (inches) after imputation**** | 4.8 (0.9) | 4.9 (0.9) | 4.9 (0.8) | 4.7 (0.9) | 4.7 (0.9) | 4.5 (0.9) | <.0001 |
| Insertive condomless anal sex ***** | 4.3 (7.6) | 8.4 (14.1) | 7.7 (17) | 5.0 (10) | 3.8 (17.3) | 1.1 (5) | <.0001 |
| Receptive condomless sex ****** | 4.6 (10.1) | 0.7 (2.1) | 2.9 (7.4) | 4.9 (10) | 6.9 (13) | 7.8 (14.7) | <.0001 |
P-value is testing for differences across all sex positionalities using Chi-squared tests (Fisher’s for small counts) if categorical or Kruskal-Wallis tests if continuous metrics.
For Penis length (inches) after imputation variable, 90 responses were missing for top, 102 responses were missing for top/versatile, 125 responses were missing for versatile, 125 responses were missing for bottom/versatile, and 99 responses were missing for bottom sex position categories.
For Penis girth (inches) after imputation variable, 90 responses were missing for top, 102 responses were missing for top/versatile, 126 responses were missing for versatile, 126 responses were missing for bottom/versatile, and 101 responses were missing for bottom sex position categories.
For condomless insertive anal intercourse variable, 84 responses were missing for top, 95 responses were missing for top/versatile, 100 responses were missing for versatile, 119 responses were missing for bottom/versatile, and 90 responses were missing for bottom sex position categories.
For condomless receptive anal intercourse variable, 84 responses were missing for top, 95 responses were missing for top/versatile, 100 responses were missing for versatile, 119 responses were missing for bottom/versatile, and 90 responses were missing for bottom sex position categories.
Regarding characteristics and attributes pertinent to the conceptual model (Figure 1), a higher proportion of participants identified as bottom/versatile (24%) followed by top/versatile (22.5%), versatile (20.3%), top (17.2%), and bottom (15.8%). For anthropomorphic characteristics, the average height of participants was 70.4 inches (SD = 3.0), average weight was 197.7 pounds (SD = 53.3), average reported penis length was 6.9 inches (SD = 1.2; imputed values of , SD = 1.2) and average reported girth was 4.9 inches (SD = 1.2; imputed value of , SD = 0.9). The average number of reported insertive condomless anal sex (CAS) encounters was 4.3 (SD = 7.6) and the average number of reported receptive CAS encounters was 4.6 (SD = 10.1) for the past 3 months.
Demographic and anthropomorphic characteristics did vary by sex position category for all factors except transactional sex. Results from these bivariate analyses are presented in Table 2. Of note, participants who identified as “tops” had a higher number of insertive CAS (8.4, SD = 14.1) while those identifying as “bottoms” had a higher number of receptive CAS (7.8, SD = 14.7). Participants who identified as “tops” had, on average, larger penis size (imputed) in both length (7.1 inches, SD = 1) and girth (4.9 inches, SD = 0.9), compared to “non-tops.” Those who identified as “bottoms” had, on average, shorter height (70 inches, SD = 3.1), but heavier weight (201 lbs, SD = 61), compared to “non-bottoms.”
Presence or absence of heteronormative stereotypes in partner attraction
Table 3 presents results from bivariate analyses examining attraction to partner traits stratified by sex position. Overall, a greater proportion of participants were attracted to partners with average weight (n = 1,994, 41.6%), average muscle definition (n = 2,247, 46.8%), higher-than-average heights (n = 1,947, 40.5%), with less than average body hair (n = 2,109, 44.0%), bigger than average butt sizes (n = 1,818, 37.9%), bigger than average penis sizes (n = 2,456, 51.2%), and partners who were more masculine (n = 2,760, 57.5%).
Table 3.
Attraction to partner traits stratified by sex position, Together 5,000, 2018–2019, n = 4802.
| Total n = 4802 | Top n = 837 | Top/ Versatile n = 1076 | Versatile n = 977 | Bottom/ Versatile n = 1151 | Bottom n = 761 | P-value* | |
|---|---|---|---|---|---|---|---|
|
| |||||||
| Attraction to Partner Traits | |||||||
| Weight | |||||||
| Less/lower/smaller than average | 1826 (38.0) | 392 (46.8) | 426 (39.6) | 381 (39.0) | 384 (33.3) | 243 (31.9) | <0.001 |
| Average | 1994 (41.6) | 324 (38.7) | 446 (41.5) | 407 (41.7) | 488 (42.4) | 329 (43.2) | |
| More/higher/bigger than average | 982 (20.5) | 121 (14.5) | 204 (19.0) | 189 (19.3) | 279 (24.3) | 189 (24.8) | |
| Muscle Definition | |||||||
| Less/lower/smaller than average | 1304 (27.2) | 262 (31.3) | 272 (25.3) | 282 (29.0) | 273 (23.7) | 215 (28.3) | 0.0010 |
| Average | 2247 (46.8) | 374 (44.7) | 538 (50.1) | 437 (44.7) | 571 (49.6) | 327 (43.0) | |
| More/higher/bigger than average | 1251 (26.1) | 201 (24.0) | 266 (24.7) | 258 (26.4) | 307 (26.7) | 219 (28.8) | |
| Height | |||||||
| Less/lower/smaller than average | 1529 (31.8) | 350 (41.8) | 391 (36.2) | 331 (33.9) | 286 (24.9) | 171 (22.5) | <0.001 |
| Average | 1326 (27.6) | 270 (32.3) | 324 (30.2) | 276 (28.3) | 287 (24.9) | 169 (22.2) | |
| More/higher/bigger than average | 1947 (40.5) | 217 (25.9) | 361 (33.6) | 370 (37.9) | 578 (50.2) | 421 (55.3) | |
| Body Hair | |||||||
| Less/lower/smaller than average | 2109 (44.0) | 435 (52.0) | 491 (45.6) | 420 (43.0) | 448 (38.9) | 315 (41.4) | <0.001 |
| Average | 1416 (29.5) | 221 (26.4) | 322 (29.9) | 305 (31.2) | 343 (29.8) | 225 (29.6) | |
| More/higher/bigger than average | 1277 (26.6) | 181 (21.6) | 263 (24.4) | 252 (25.8) | 360 (31.3) | 221 (29.0) | |
| Butt Size | |||||||
| Less/lower/smaller than average | 1302 (27.1) | 204 (24.4) | 247 (23) | 254 (26.0) | 312 (27.1) | 285 (37.5) | <.0001 |
| Average | 1682 (35.1) | 254 (30.4) | 386 (35.9) | 335 (34.3) | 418 (36.3) | 289 (38.0) | |
| More/higher/bigger than average | 1818 (37.9) | 379 (45.3) | 443 (41.2) | 388 (39.7) | 421 (36.6) | 187 (24.6) | |
| Penis Size | |||||||
| Less/lower/smaller than average | 903 (19.0) | 291 (34.8) | 219 (20.3) | 164 (17.0) | 137 (12.0) | 92 (12.1) | <0.001 |
| Average | 1443 (30.1) | 277 (33.1) | 372 (34.6) | 301 (30.8) | 305 (26.5) | 188 (24.7) | |
| More/higher/bigger than average | 2456 (51.2) | 269 (32.1) | 485 (45.1) | 512 (52.4) | 709 (61.6) | 481 (63.2) | |
| Masculine v Feminine | |||||||
| Less/lower/smaller than average | 887 (18.5) | 185 (22.1) | 207 (19.2) | 201 (20.6) | 186 (16.2) | 108 (14.2) | <0.0001 |
| Average | 1155 (24.0) | 199 (23.8) | 273 (25.4) | 233 (23.9) | 295 (25.6) | 155 (20.4) | |
| More/higher/bigger than average | 2760 (57.5) | 453 (54.1) | 596 (55.4) | 543 (55.6) | 670 (58.3) | 498 (65.4) | |
P-values are testing if there are associations between attraction traits and sex position using Chi-squared tests (Fisher’s Exact if small counts).
Attraction results illustrated the presence of several heteronormative stereotypes. For instance, a larger proportion of participants who identified as “tops” were attracted to partners with lower-than-average weight (n = 392, 46.8%) and height (n = 350, 41.8%), bigger than average butts (n = 379, 45.3%), and smaller-than-average penises (n = 291, 34.8%), indicating an attraction to physically smaller male sex partners. Further, a larger proportion of participants who identified as “bottoms” were attracted to partners with average weight (n = 329, 43.2%), higher-than-average heights (n = 421, 55.3%), and bigger-than-average penises sizes (n = 481, 63.2%), illustrating attraction toward physically larger male sex partners. Yet, some attraction results were contrary to typical heteronormative stereotypes. A greater proportion of both “tops” (n = 453, 54.1%) and “bottoms” (n = 498 65.4%) were attracted to partners who were more masculine. In addition, a greater proportion of both “tops” (n = 435, 52.0%) and “bottoms” (n = 315, 41.4%) were attracted to partners with less-than-average body hair. Finally, among all self-reported sex positions, the highest proportion of participants reported attraction to average muscle definition.
Predicting Sex Position Identity/Label Use from Attraction, Behavior, and Anthropomorphic Characteristics
Our main analyses focused on whether we could accurately predict sex position identity based on attraction to partner traits, behavior, and anthropomorphic characteristics. Initial modeling results (coefficients) to determine associations between self-reported sex position and the three constructs are reported in Table 4. These coefficients were used in subsequent predictive models. When using all three constructs (attraction, behavior, and anthropomorphic characteristics), our model was able to correctly predict 62% of participants who identified as a “top,” 33% as a “top/versatile,” 11% as a “versatile,” 54% as a “bottom/versatile,” and 44% as a “bottom” (Table 5). When considering adjacent categories (e.g., top/versatile as adjacent to top), our model was able to correctly predict 80% of those who identified as a “top” or “top/versatile;” 70% as a “top/versatile,” “top,” or “versatile;” 77% as a “versatile,” “top/versatile,” or “bottom/versatile;” 78% as a “bottom/versatile,” “versatile,” or “bottom;” and 81% as a “bottom” or “bottom/versatile.” These results suggest that near adjacent categories allow for better prediction of versatile or versatile combination labels.
Table 4.
Logistic model coefficients (β (SE; p-value)) for the full cohort modeling each sex position category as a function of attraction, anthropomorphic, and behavior constructs, Together 5,000, 2018–2019, n = 3850*.
| Construct Covariates | Full Cohort (Top v all other position) | Full Cohort (Top/versatile v all other positions) | Full Cohort (Versatile v all other positions) | Full Cohort (Bottom/versatile v all other positions) | Full Cohort (Bottom v all other positions) |
|---|---|---|---|---|---|
|
| |||||
| Attraction Traits ** | |||||
| Masculinity | |||||
| AA v A (ref) | 0.18 (0.12; 0.1227) | −0.03 (0.10; 0.7769) | −0.02 (0.10; 0.8467) | −0.22 (0.10; 0.0213) | 0.15 (0.13; 0.2476) |
| BA v A | −0.07 (0.15; 0.6617) | −0.06 (0.13; 0.6168) | 0.26 (0.13; 0.0487) | −0.08 (0.13; 0.5298) | −0.23 (0.18; 0.1895) |
| Butt | |||||
| AA v A | 0.34 (0.11; 0.0024) | 0.11 (0.09; 0.1927) | 0.10 (0.10; 0.2894) | −0.09 (0.09; 0.3336) | −0.64 (0.13; <.0001) |
| BA v A | −0.25 (0.14; 0.0727) | −0.17 (0.11; 0.1340) | −0.006 (0.11; 0.9617) | 0.07 (0.11; 0.5208) | 0.45 (0.13; 0.0005) |
| Hair | |||||
| AA v A | 0.06 (0.14; 0.6866) | −0.12 (0.11; 0.2711) | −0.07 (0.11; 0.5483) | 0.13 (0.11; 0.2306) | 0.11 (0.14; 0.4052) |
| BA v A | 0.11 (0.12; 0.3584) | −0.03 (0.10; 0.7638) | −0.13 (0.10; 0.2060) | −0.004 (0.10; 0.9689) | 0.02 (0.13; 0.8919) |
| Height | |||||
| AA v A | −0.49 (0.12; <.0001) | −0.26 (0.10; 0.0087) | −0.15 (0.11; 0.1500) | 0.36 (0.10; 0.0003) | 0.33 (0.13; 0.0082) |
| BA v A | −0.20 (0.13; 0.1205) | 0.08 (0.11; 0.4463) | 0.10 (0.11; 0.3758) | −0.05 (0.12; 0.6636) | −0.27 (0.16; 0.0816) |
| Weight | |||||
| AA v A | −0.32 (0.15; 0.0333) | −0.003 (0.11; 0.9814) | 0.08 (0.12; 0.5186) | 0.07 (0.11; 0.5099) | 0.13 (0.14; 0.3582) |
| BA v A | 0.27 (0.11; 0.0141) | 0.04 (0.09; 0.7005) | 0.09 (0.10; 0.3678) | −0.16 (0.10; 0.0896) | −0.18 (0.12; 0.1335) |
| Muscles | |||||
| AA v A | 0.13 (0.12; 0.2739) | −0.13 (0.10; 0.1906) | 0.09 (0.10; 0.3522) | −0.12 (0.10; 0.2233) | 0.17 (0.12; 0.1661) |
| BA v A | −0.07 (0.13; 0.5589) | −0.16 (0.10; 0.1193) | 0.13 (0.11; 0.2195) | 0.006 (0.11; 0.9539) | 0.31 (0.13; 0.0227) |
| Penis | |||||
| AA v A | −0.57 (0.12; <.0001) | −0.28 (0.09; 0.0021) | −0.05 (0.10; 0.6318) | 0.38 (0.09; <.0001) | 0.50 (0.12; <.0001) |
| BA v A | 0.60 (0.14; <.0001) | −0.11 (0.12; 0.3597) | −0.38 (0.14; 0.0052) | −0.44 (0.15; 0.0025) | −0.04 (0.19; 0.8234) |
| Anthropomorphic Characteristics | |||||
| Height | 0.003 (0.02; 0.8713) | 0.01 (0.01; 0.5543) | 0.01 (0.02; 0.4980) | −0.0001 (0.01; 0.9953) | −0.02 (0.02; 0.2252) |
| Weight | 0.001 (0.001; 0.2159) | 0.0003 (0.0009; 0.7067) | −0.001 (0.0009; 0.1091) | −0.002 (0.0009; 0.0595) | 0.001 (0.001; 0.2419) |
| Penis length | 0.27 (0.05; <.0001) | 0.17 (0.04; <.0001) | 0.16 (0.04; 0.0003) | −0.29 (0.04; <.0001) | −0.32 (0.06; 0.0737) |
| Penis girth | 0.11 (0.06; 0.0684) | 0.14 (0.05; 0.0044) | −0.07 (0.05; 0.1496) | −0.03 (0.05; 0.5188) | −0.12 (0.06; 0.0044) |
| Behavior | |||||
| Insertive condomless anal sex | 0.02 (0.004; <.0001) | 0.02 (0.004; <.0001) | −0.005 (0.004; 0.1998) | −0.02 (0.01; <.0001) | −0.29 (0.03; <.0001) |
| Receptive condomless anal sex | −0.48 (0.04; <.0001) | −0.03 (0.006; <.0001) | 0.003 (0.004; 0.3835) | 0.02 (0.004; <.0001) | 0.05 (0.01; <.0001) |
We had complete data on a total of 3850 participants, 239 had missing sex positions, 63 missing on both sex positions and covariates, 952 missing on covariates – note 3850 + 952 = 4802 (analytic cohort).
AA= above average, A=average (referent level), BA=below average.
Table 5.
Cross-tabulation of reported sexual positionality and predicted sexual positionality with percent of accuracy or near-accuracy among the full analytic cohort*, Together 5,000, 2018–2019, n = 3850.
| Top | Top/ Versatile | Versatile | Bottom/ Versatile | Bottom | Total | |
|---|---|---|---|---|---|---|
|
| ||||||
| Top Predicted | 424 | 279 | 110 | 64 | 29 | 906 |
| Top/Versatile Predicted | 121 | 292 | 227 | 140 | 56 | 836 |
| Versatile Predicted | 20 | 55 | 87 | 69 | 24 | 255 |
| Bottom/Versatile Predicted | 108 | 240 | 280 | 499 | 220 | 1347 |
| Bottom Predicted | 6 | 28 | 66 | 150 | 256 | 506 |
| Total | 679 | 894 | 770 | 922 | 585 | 3850 |
| % Accurate** | 62% | 33% | 11% | 54% | 44% | |
| % Near-accurate*** | 80% | 70% | 77% | 78% | 81% | |
We had complete data on a total of 3850 participants, 239 had missing sex positions, 63 missing on both sex positions and covariates, 952 missing on covariates – note 3850 + 952= 4802 (analytic cohort).
% Accurate is defined those who were correctly reported as their position (i.e., 424 were predicted as “top” among the 679 who reported a top position, equivalent to 62%).
% Near-accurate is defined as those who were correctly reported as their position or the position immediately adjacent to their category (i.e., 424 were predicted as “top” and 121 were predicted as “top/versatile” among the 679 who reported a top position, equivalent to 80%).
Finally, we conducted a training and testing evaluation of our predictive model. Results from the training and testing analyses are presented in online Supplemental Table 1. Overall, mostly small variations in predicted sex positionality were observed resulting in similar accuracies and near-accuracies to the full model.
Discussion
The purpose of these analyses was not to define sexual positions, but to illustrate the complexities of anal sex position self-labels among SMM by applying concepts of identity, behavior, personal anthropomorphic traits, and attraction to partners’ characteristics. Too often, sex position is used to ascertain “risk” for public health interventions without further exploration of the various contexts and nuances that may contribute to sex positionality within sexual situations. Though specific research questions may ask participants to recall the number of sexual partners with whom they engaged in receptive or insertive anal intercourse or explicit number of receptive and insertive sexual acts, sex positionality is often used to generalize the engaging in a preponderance of receptive or insertive sexual intercourse. Specific and generalized research questions equally have limitations in their measurement including lack of accuracy in specific recall or over-simplification of complex human behavior (Catania et al., 1990); here, we expand the conversation for why SMM may or may not engage in receptive or insertive anal intercourse including anthropomorphic characteristics and considerations of attraction. As our current and others’ work has shown, these considerations can be irrespective of (Dangerfield et al., 2017; Johns et al., 2012; Moskowitz & Roloff, 2017; Pachankis et al., 2013; Swift-Gallant et al., 2018, 2019) or with regard to HIV and STI vulnerability (Dangerfield et al., 2017; Dangerfield, Ober, et al., 2018; Grov et al., 2007; Newcomb et al., 2015). We do not suggest that this is the only way in which to measure sex positionality but rather seek to extend the conversation around this aspect of sexuality.
Without a better understanding of factors related to sex positionality, public health interventions may fail to address periods of HIV and STI vulnerability and ignore real-world challenges in applications of interventions. For example, building sexual health literacy around the risk of STI or HIV acquisition by engaging in insertive or receptive anal intercourse for people who may identify as “versatile”, “versatile top”, or “versatile bottom”, and not just those who identify as “bottoms” (i.e., most of the time engage in receptive anal intercourse). Previous research has suggested differing transmission probabilities by sexual acts for HIV (Patel et al., 2014), gonorrhea (Hui et al., 2015), and syphilis (Gray et al., 2010; Schmidt et al., 2019), much of which indicates higher transmission probabilities for insertive action due to rectal environment perturbations for those receiving (Kelley et al., 2017). Though it may seem that sexual behavior is most salient in discussions of disease transmission and vulnerability, it is the variability in sexual behavior and the reasons for that variability we seek to understand. Building on prior research, this study provides a “next step” to better elucidate sex positionality.
Descriptive analyses showed variation in sociodemographic characteristics and concepts of interest (attraction, behavior, anthropomorphic characteristics) across all sex position self-labels. We also found differences between sex positions for self-reported attraction to male partners’ physical traits across varying degrees of muscle definition, height, weight, body hair, butt size, penis size, and masculinity which is consistent with previous work (Swift-Gallant et al., 2019). Some differences—such as attraction to certain partner physical traits—across sex positions may be influenced by sociocultural norms of perceived masculinity and sexual-scripting (Grov et al., 2010, 2015; Moskowitz & Hart, 2011), thereby perpetuating heteronormative stereotypes among SMM when determining sex self-labels. Further, these perceptions of heteronormative stereotypes and masculinity can manifest as internalized homonegativity, leading to complex relationships with one’s sex label (Dangerfield, Smith, et al., 2018; Sánchez & Vilain, 2012) in response to internally held beliefs of masculinity, heteronormative ideals, and homonegativity ascribed to that role (Hart et al., 2003). For example, taller height, larger penises, or more body hair can be perceived as indicators of masculinity that have sometimes, in the case of body hair and penis size, been associated with sex positionality (Moskowitz & Hart, 2011). Further, sociocultural norms and heteronormative stereotypes may impact individuals’ sexual agency and ability to navigate and negotiate protected sex (e.g., condomed sex, PrEP use) (Smith & Amaro, 2021). Our exploration of self-identified sex position labels, and adjacent labels, suggest substantial fluidity and flexibility within the labels themselves as well as factors—like behavior, attraction to partners’ characteristics, and personal anthropomorphic characteristics—associated with each label. That is, there is no “one size fits all” for describing each sex position. This was further emphasized by our failure to identify significant overlap in unique response patterns across all construct variables (online Supplemental Analysis 1), indicating that sex positionality is uniquely influenced by multiple factors for everyone (Pachankis et al., 2013).
Our main analyses aimed to assess whether we could accurately predict someone’s self-reported sex position identity using other concepts such as attraction to male sex partners’ physical traits, sexual behavior, and anthropomorphic characteristics. Results suggest that we are moderately able to predict sex positionality when considering near-accurate categories, but only fairly accurate for certain stand-alone categories (such as “tops” (62%) and “bottom/versatile” (54%)). From these results, we can infer that people who identify as “tops” and “bottoms” may not be meaningfully different on behavior, attraction to partners’ physical traits, or anthropomorphic characteristics from people who identify as top/versatile and bottom/versatile, respectively. These results may also offer evidence of fluidity in sex position identity through discordance in self-identified label and behavior previously found by others (Pachankis et al., 2013). Though participants may simultaneously identify and act as a “top” during most sexual encounters, sexual behavior may vary depending on circumstances. Although this study focused on individual factors related to sex position identity, other studies have highlighted the importance of relationship factors—such as partner’s position, expectations, power dynamics, communication—in determining sex positionality (Pachankis et al., 2013).
Our study also supports previous findings related to anthropomorphic characteristics being associated with sex positionality. Prior research has noted that men identifying as “tops” self-reported having larger penises (Moskowitz & Hart, 2011). Notably, Grov et al. (2010) found that men who had below average sized penises were more likely to identify as “bottoms” while those with above average penises were more likely to identify as “tops.” Our unadjusted bivariate analyses found a similar pattern among our participants who identified as “tops” and “bottoms,” such that “tops” reported larger penis girth and length compared to “bottoms.” These results may also highlight other underlying factors related to anthropomorphic characteristics influencing sex positionality and the fluidity of sex positionality such as sexual self-confidence, exploration of sexual pleasure, or perceptions of masculinity. Pachankis et al. (2013) highlighted how participants’ sexual development and personal sexual satisfaction assisted with the development of their preferred sex position. Additionally, these participants also mentioned changing positionality as they felt more comfortable expressing themselves sexually (Pachankis et al., 2013). Lastly, perceptions of masculinity and penis size may be linked for some participants who perceive larger penises to be associated with being more masculine (Carballo-Diéguez et al., 2004; Grov et al., 2010; Johns et al., 2012; Moskowitz & Hart, 2011).
For most of our participants, larger proportions were attracted to partners who were more masculine than average, though participants who identified as “bottoms” had the largest proportion. As others have noted, this may be due to how perceptions of masculinity inform how people think about their sex positionality thereby ascribing the more masculine role to that of the insertive partner (Grov et al., 2010, 2015; Moskowitz & Hart, 2011). Future research could continue to investigate past, current, and changing perceptions of self and heteronormative ideals to better understand their role in sexual position labels. As we grow our understanding in how people engage in sex, we can also expand our options for how people can engage in a sex in a safe and healthy way.
Strengths and Limitations
This study had several strengths, including the use of a diverse, nationwide sample of SMM and the use of multiple measures to contextualize participants’ anal sex positionality. Despite these strengths, our findings should be interpreted with several limitations in mind. First, we should acknowledge that this sample was not, nor was meant to be, representative of the U.S. population. As with previous literature, the focus of the parent study was that of prevention among a cohort particularly vulnerable to HIV, which may have impacted our findings. Our inclusion criteria focused on groups with the highest burden of HIV seroconversion, including those who were younger and eligible for, but not using HIV PrEP. However, findings from this study may have implications for STI and HIV prevention interventions based on sex position self-labels; thus, this cohort represents people who may benefit most from this research. Further, given the parent study’s focus on HIV prevention, many of the questions assessing behavior were tailored for that purpose. As such, we are unable to truly reflect all instances of insertive and receptive sexual acts (with and without condoms). Despite this measurement limitation, we do believe that our condomless measures are a reasonable estimation of the primary sexual behavior in which participants were engaging. Additionally, our questions related to “attraction” should be further refined to better measure the intended response related to participants’ attraction to differing degrees of gender expression. The measure used for this analysis does not reflect our current understanding nor sensitivities regarding the distinction between gender expression and sexual orientation nor is it culturally appropriate for all people identifying as a sexual and/or gender minority. One final challenge with our measurements is that participants were allowed to progress through the study without answering the penis length and girth measurement questions. This necessitated our imputation of these values for certain participants but the patterns of missingness may have introduced bias into our estimates of penis size. For example, participants who had smaller penises may have been less likely to respond or exaggerate their responses as suggested by prior literature (Grov et al., 2010). Finally, our analyses focused only on individual drivers of sex positionality. Future research could expand this line of inquiry to include longitudinal and dyadic investigations to be more inclusive of relational and sociocultural drivers of positionality, as well as to capture fluidity in sex positionality overtime.
Conclusions
The main goal of this paper was to explore sex position self-labels using a multidimensional approach that could be used to develop better measures of sex positionality rather than relying on single dimensions of positionality alone (e.g., identifying as a “top” or a “bottom”). This, in turn, can be used to build interventions to more accurately address how people engage in sex (e.g., targeting decision-making to practice insertive or receptive anal sex) so that we can help facilitate healthier sex lives. Our results indicate that self-reported sex position labels, attraction to partners’ physical traits, sexual behaviors, and anthropomorphic characteristics all contribute to overall sex positionality. Thus, using single-indicator measures of sex positionality may not elucidate the complexities surrounding sex position identity, impacting our ability to assess changing vulnerability for HIV and STIs.
Supplementary Material
Acknowledgements:
Special thanks to additional members of the T5K study team: Denis Nash, David Pantalone, Sarit A. Golub, Viraj V. Patel, Gregorio Millett, Don Hoover, Sarah Kulkarni, Matthew Stief, Chloe Mirzayi, Alexa D’Angelo, Fatima Zohra, Michelle Dearolf, Javier Lopez-Rios, & Pedro B. Carneiro. Thank you to the program staff at NIH (Gerald Sharp, Sonia Lee, and Michael Stirratt) and our T5K Scientific Advisory Board (Adam Carrico, Michael Camacho, Demetre Daskalakis, Sabina Hirshfeld, Jeremiah Johnson, Claude Mellins, and Milo Santos). Finally, thank you to Isaac Payton for his contributions. While the NIH financially supported this research, the content is the responsibility of the authors and does not necessarily reflect official views of the NIH.
Funding:
Together 5,000 was funded by the National Institutes for Health (UH3 AI 133675, UG3 AI 133675 -PI Grov). Other forms of support include the CUNY Institute for Implementation Science in Population Health, the Einstein, Rockefeller, CUNY Center for AIDS Research (ERC CFAR, P30 AI124414). D.A.W. was supported, in part, by a career development award (K01 AA 029047). J.B. was supported by a training grant awarded to the University of Florida (T32 AA 025877).
Footnotes
Declaration of interest: The authors report there are no competing interests to declare.
Data availability statement:
Data are available upon reasonable requestion from the corresponding author, CG.
References
- Carballo-Diéguez A, Dolezal C, Nieves L, Díaz F, Decena C, & Balan I (2004). Looking for a tall, dark, macho man… sexual-role behaviour variations in Latino gay and bisexual men. Culture, Health & Sexuality, 6(2), 159–171. 10.1080/13691050310001619662 [DOI] [Google Scholar]
- Catania JA, Gibson DR, Chitwood DD, & Coates TJ (1990). Methodological problems in AIDS behavioral research: influences on measurement error and participation bias in studies of sexual behavior. Psychological Bulletin, 108(3), 339. 10.1037/0033-2909.108.3.339 [DOI] [PubMed] [Google Scholar]
- Dangerfield DT, Ober AJ, Smith LR, Shoptaw S, & Bluthenthal RN (2018). Exploring and adapting a conceptual model of sexual positioning practices and sexual risk among HIV-negative black men who have sex with men. The Journal of Sex Research, 55(8), 1022–1032. 10.1080/00224499.2018.1433287 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Dangerfield DT, Smith LR, Anderson JN, Bruce OJ, Farley J, & Bluthenthal R (2018). Sexual positioning practices and sexual risk among black gay and bisexual men: A life course perspective. AIDS and Behavior, 22(6), 1919–1931. 10.1007/s10461-017-1948-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Dangerfield DT, Smith LR, Williams J, Unger J, & Bluthenthal R (2017). Sexual positioning among men who have sex with men: A narrative review. Archives of Sexual Behavior, 46(4), 869–884. 10.1007/s10508-016-0738-y [DOI] [PMC free article] [PubMed] [Google Scholar]
- Gray RT, Hoare A, Prestage GP, Donovan B, Kaldor JM, & Wilson DP (2010). Frequent testing of highly sexually active gay men is required to control syphilis. Sexually Transmitted Diseases, 37(5), 298–305. 10.1097/OLQ.0b013e3181ca3c0a [DOI] [PubMed] [Google Scholar]
- Grov C, DeBusk JA, Bimbi DS, Golub SA, Nanin JE, & Parsons JT (2007). Barebacking, the internet, and harm reduction: An intercept survey with gay and bisexual men in Los Angeles and New York City. AIDS and Behavior, 11(4), 527–536. 10.1007/s10461-007-9234-7 [DOI] [PubMed] [Google Scholar]
- Grov C, Parsons JT, & Bimbi DS (2010). The association between penis size and sexual health among men who have sex with men. Archives of Sexual Behavior, 39(3), 788–797. 10.1007/s10508-008-9439-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Grov C, Saleh LD, Lassiter JM, & Parsons JT (2015). Challenging race-based stereotypes about gay and bisexual men’s sexual behavior and perceived penis size and size satisfaction. Sexuality Research and Social Policy, 12(3), 224–235. 10.1007/s13178-015-0190-0 [DOI] [Google Scholar]
- Grov C, Westmoreland DA, Carneiro PB, Stief M, MacCrate C, Mirzayi C, Pantalone DW, Patel VV, & Nash D (2019). Recruiting vulnerable populations to participate in HIV prevention research: findings from the Together 5000 cohort study. Annals of Epidemiol, 35, 4–11. 10.1016/j.annepidem.2019.05.003 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Hart TA, Wolitski RJ, Purcell DW, Gómez C, Halkitis P, & Seropositive Urban Men’s Study Team. (2003). Sexual behavior among HIV-positive men who have sex with men: What’s in a label? The Journal of Sex Research, 40(2), 179–188. 10.1016/j.annepidem.2019.05.003 [DOI] [PubMed] [Google Scholar]
- Hui B, Fairley CK, Chen M, Grulich A, Hocking J, Prestage G, Walker S, Law M, & Regan D (2015). Oral and anal sex are key to sustaining gonorrhoea at endemic levels in MSM populations: a mathematical model. Sexually Transmitted Infections, 91(5), 365–369. 10.1136/sextrans-2014-051760 [DOI] [PubMed] [Google Scholar]
- Johns MM, Pingel E, Eisenberg A, Santana ML, & Bauermeister J (2012). Butch tops and femme bottoms? Sexual positioning, sexual decision making, and gender roles among young gay men. American Journal of Men’s Health, 6(6), 505–518. 10.1177/1557988312455214 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Kelley CF, Kraft CS, De Man TJ, Duphare C, Lee HW, Yang J, Easley KA, Tharp GK, Mulligan M, Sullivan PS, Bosinger SE, & Amara RR (2017). The rectal mucosa and condomless receptive anal intercourse in HIV-negative MSM: Implications for HIV transmission and prevention. Mucosal Immunology, 10(4), 996–1007. 10.1038/mi.2016.97 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Kippax S, & Smith G (2001). Anal intercourse and power in sex between men. Sexualities, 4(4), 413–434. 10.1177/136346001004004002 [DOI] [Google Scholar]
- Kline RB (1998). Principles and pracitce of structural equation modeling. Guilford Press. [Google Scholar]
- Moskowitz DA, Alvarado Avila A, Kraus A, Birnholtz J, & Macapagal K (2022). Top, bottom, and versatile orientations among adolescent sexual minority men. The Journal of Sex Research, 59(5), 643–651. 10.1080/00224499.2021.1954583 [DOI] [PubMed] [Google Scholar]
- Moskowitz DA, & Hart TA (2011). The influence of physical body traits and masculinity on anal sex roles in gay and bisexual men. Archives of Sexual Behavior, 40(4), 835–841. 10.1007/s10508-011-9754-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Moskowitz DA, Rieger G, & Roloff ME (2008). Tops, bottoms and versatiles. Sexual and Relationship Therapy, 23(3), 191–202. 10.1080/14681990802027259Moskowitz [DOI] [Google Scholar]
- Moskowitz DA, & Roloff ME (2017). Recognition and construction of top, bottom, and versatile orientations in gay/bisexual men. Archives of Sexual Behavior, 46(1), 273–285. 10.1007/s10508-016-0810-7 [DOI] [PubMed] [Google Scholar]
- Nash D, Stief M, MacCrate C, Mirzayi C, Patel VV, Hoover D, Pantalone DW, Golub S, Millett G, D’Angelo AB, Westmoreland DA, & Grov C (2019). A Web-Based Study of HIV Prevention in the Era of Pre-Exposure Prophylaxis Among Vulnerable HIV-Negative Gay and Bisexual Men, Transmen, and Transwomen Who Have Sex With Men: Protocol for an Observational Cohort Study. JMIR Research Protocols, 8(9), e13715. 10.2196/13715 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Newcomb ME, Ryan DT, Garofalo R, & Mustanski B (2015). Race-based sexual stereotypes and their effects on sexual risk behavior in racially diverse young men who have sex with men. Archives of Sexual Behavior, 44(7), 1959–1968. 10.1007/s10508-015-0495-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Pachankis JE, Buttenwieser IG, Bernstein LB, & Bayles DO (2013). A longitudinal, mixed methods study of sexual position identity, behavior, and fantasies among young sexual minority men. Archives of Sexual Behavior, 42(7), 1241–1253. 10.1007/s10508-013-0090-4 [DOI] [PubMed] [Google Scholar]
- Parsons JT, Schrimshaw EW, Wolitski RJ, Halkitis PN, Purcell DW, Hoff CC, & Gómez CA (2005). Sexual harm reduction practices of HIV-seropositive gay and bisexual men: serosorting, strategic positioning, and withdrawal before ejaculation. AIDS, 19 Suppl 1, S13–S25. 10.1097/01.aids.0000167348.15750.9a [DOI] [PubMed] [Google Scholar]
- Patel P, Borkowf CB, Brooks JT, Lasry A, Lansky A, & Mermin J (2014). Estimating per-act HIV transmission risk: a systematic review. AIDS, 28(10), 1509–1519. 10.1097/QAD.0000000000000298 [DOI] [PMC free article] [PubMed] [Google Scholar]
- R Development Core Team. (2021). R: A language and environment for statistical computing. In R Foundation for Statistical Computing. http://www.R-project.org [Google Scholar]
- Sánchez FJ, & Vilain E (2012). “Straight-acting gays”: The relationship between masculine consciousness, anti-effeminacy, and negative gay identity. Archives of Sexual Behavior, 41(1), 111–119. 10.1007/s10508-012-9912-z [DOI] [PubMed] [Google Scholar]
- SAS Institute Inc. (2023). SAS/STAT® 15.3 User’s Guide. SAS Institute Inc. [Google Scholar]
- Schmidt R, Carson PJ, & Jansen RJ (2019). Resurgence of syphilis in the United States: An assessment of contributing factors. Infectious Diseases: Research and Treatment, 12, 10.1177/1178633719883282. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Smith JG, & Amaro G (2021). “No fats, no femmes, and no blacks or asians”: The role of body-type, sex position, and race on condom use online. AIDS and Behavior, 25(7), 2166–2176. 10.1007/s10461-020-03146-z [DOI] [PubMed] [Google Scholar]
- Swift-Gallant A, Coome LA, Monks DA, & VanderLaan DP (2018). Gender nonconformity and birth order in relation to anal sex role among gay men. Archives of Sexual Behavior, 47(4), 1041–1052. 10.1007/s10508-017-0980-y [DOI] [PubMed] [Google Scholar]
- Swift-Gallant A, Coome LA, Skorska MN, Monks DA, & VanderLaan DP (2019). Pubertal onset and anal sex role among gay men. Psychology of Sexual Orientation and Gender Diversity, 6(1), 107–112. 10.1037/sgd0000302 [DOI] [Google Scholar]
- Tabachnick BG and Fidell LS (2001) Using multivariate statistics. 4th Edition, Allyn and Bacon. [Google Scholar]
- Van de Ven P, Kippax S, Crawford J, Rawstorne P, Prestage G, Grulich A, & Murphy D (2002). In a minority of gay men, sexual risk practice indicates strategic positioning for perceived risk reduction rather than unbridled sex. AIDS Care, 14(4), 471–480. 10.1080/09540120208629666 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Data are available upon reasonable requestion from the corresponding author, CG.
