The Anal Sex Stigma Scales: A new measure of sexual stigma among cisgender men who have sex with men

Abstract

We conducted a series of studies to validate a new scale of stigma toward anal sex, culturally tailored to cisgender men who have sex with men (MSM). In Study 1 we conducted in-depth interviews (N = 35) to generate items. In Study 2, we reduced the item pool through an online survey (N = 268), testing scale performance, dimensionality, and convergent and discriminant validity. For Study 3, we recruited another online sample (N = 1605), randomized to exploratory or confirmatory factor analyses to finalize item reduction, then assessed validity among sexually active MSM (n = 1263). Final subscales encompassed self-stigma (6 items, Cronbach’s α = .72), provider stigma (5 items, Cronbach’s α = .79), and omission of information (6 items, Cronbach’s α = .73; full 3-factor scale = .80). We developed a 17-item measure, grounded in the lived experience of cisgender MSM. Future work should examine associations with health-seeking behavior.

Introduction

Anal sex is a prevalent behavior, depicted since antiquity [1–3], and reported across populations and the globe [4–9]. However, anal sex is also under-reported depending on data collection methods and social context [4], likely due to stigma [2,10,11]. Stigma refers to a social process that labels some people, for example those who engage in anal sex, as less valuable than others [12–15]. In the United States, anal sex was illegal in many jurisdictions until 2003, likely inhibiting both research and health policy on the topic [16–18].

Stigma toward anal sexuality may relate to important health outcomes. Sexual education tends to omit anal sexuality, leaving those who engage in anal sex uninformed, anxious, and at risk for embarrassment and physical harm during sex [11,19–21]. Anal sex also functions as a stigma ‘label’ across populations [3,11,20,22–32], and may function to deter disclosure of sexual behavior and thereby impede access to relevant interventions for health and wellbeing, for example HIV prevention tools [11,19,21]. Among gay, bisexual and other men who have sex with men (MSM), anal intercourse, under certain conditions, is the most proximate risk factor for HIV [4,6] and a precursor, if disclosed, to enrollment in promising interventions, like pre-exposure prophylaxis (PrEP) [33]. However, for sexual and gender minority (SGM) populations, stigma toward their sexual identity and behavior is generally a deterrent to health-seeking behavior [34]. Among MSM, stigma toward sexual orientation is associated with acquisition of sexually transmitted infections [35], diminished social support [36], compromised life satisfaction [37], and concealment of sexuality and delayed and intermittent engagement in medical care [38,39] including biomedical interventions like PrEP [23]. Qualitative studies also describe how internalized shame influences sexual behavior among MSM [21,24,40,41] and a recent quantitative study of stigma toward sexual behavior, though not specifically toward anal sex, found significant associations with both HIV risk and poor mental health [42]. This suggests that anal sex stigma, to the extent that it can be measured more specifically, would likely be relevant to MSM health, possibly as one of several fundamental causes of disease [42] that contribute to their HIV-related health disparities [43,44].

During formative work in the United States [11], we developed a conceptual model based on sexual stigma [14], concealable stigmatized identities [45], and fundamental causes of disease [12,46] positing how stigma toward anal sexuality impedes health-seeking behavior among MSM. We validated the model through qualitative inquiry among 10 key informants and 25 MSM [11]. However, a psychometrically valid measure of anal sex stigma would propel research in this area by, in particular, allowing surveillance of trends and guiding intervention development [13]. One limitation of current stigma measures is that they may miss nuances in subpopulations and social context or lack empirical validation [47,48]. Measurement of different manifestations of stigma also matters, in part because experiences of stigma do not always lead an individual to internalize devaluation or to anticipate its reenactment in the future [49]. Indeed, each of these forms of stigma can be associated with distinct health outcomes [e.g., 45,50,51]. Additionally, measures may capture individual-level elements of stigma, like internalized shame, rather than institutional and interpersonal components [12,14,52,53]. While internalized stigma is an important and relevant construct, meta-analyses among MSM suggest that its effects decrease over time [54]. This points to the need to measure social forces further upstream than the process of internalization, because stigma likely manifests not only as internalization but in additional forms that beget subsequent structural inequalities in health [12,44].

Overview: Study 1 (Item Generation), Study 2 (Item Reduction), Study 3 (Scale Validation)

To quantify anal sex stigma, we developed and validated a new measure, informed by a preliminary conceptual model, published elsewhere, that posited how stigma toward anal sex deters health-seeking behavior [11]. In Study 1 (Item Generation), we aimed to construct a set of items that captured culturally specific elements of stigma toward anal sex, tailored to cisgender MSM, based in in-depth interviews with 10 experts in MSM health and 25 MSM. In Study 2 (Item Reduction), we aimed to reduce the number of items by piloting the initial set in a sample of 268 MSM and then examining scale performance, exploratory factor analyses, and convergent and discriminant validity. In Study 3 (Scale Validation), we conducted another survey to finalize our measure, this time with 1605 MSM, repeating scale performance and exploratory factor analyses, adding confirmatory factor analyses, and examining convergent and discriminant validity with the final measure in a subsample of sexually active MSM.

STUDY 1: ITEM GENERATION

Materials and Methods

The Human Subjects Division of the University of Washington and the Institutional Review Board of the New York York State Psychiatric Institute approved procedures.

Participants

We interviewed 10 key informants as experts in MSM health, then 25 cisgender MSM participants from across the United States. Screening, consent and demographic data collection occurred through an online survey platform [55]. Eligible MSM participants needed to report at least one male anal sex partner in the past year; to identify as currently male and assigned male at birth; and to be living in the U.S., fluent in English, and at least 8 years of age.

Procedures

The first author conducted hour-long interviews with key informants in December 2015 either by telephone (n = 5) or face-to-face (n = 5), then with MSM in March and April 2016 either by their preference for telephone (n = 18) or online chat (n = 7). Procedures for in-depth interviews are described in greater detail elsewhere [11]. Discussion guides are included as appendices. After developing an initial pool of items, we conducted 7 cognitive interviews with a subset of MSM in August 2016, all by telephone while participants and the interviewer viewed items simultaneously online. Participants verbalized their emotional and cognitive responses to reading individual items, thereby informing revisions to reduce response errors and to ensure that items more closely reflected their experiential world [56]. All interviews were recorded. Key informant and initial MSM interviews transcribed. MSM could enter a raffle to win one of three $50 gift certificates as compensation for their time.

Measures

We followed established guidelines for item generation in scale development [57]. We phrased items as closely to the original interview language as possible, at times verbatim from transcripts, and diagrammed in MindNode software [58] to visualize clusters of related original language in relation to generated items.

We aimed to distinguish between experienced, internalized, and anticipated forms of stigma posited and studied in published literature [14,45] and supported by our qualitative findings and conceptual model [11]. We also sought to balance brevity with the potential for endorsement across diverse experiences reflected in interviews, in particular racial and ethnic group differences, sexual position preferences, and potential concealment of sexual orientation and behavior. For example, experienced stigma was not limited to enactment directly on the respondent (e.g., “I was shamed for bottoming”) and instead could be witnessed or ‘felt’ [34] by a respondent who was not himself identifiable as a direct target (e.g., “Bottoms are shamed for enjoying anal sex”). Similarly, we chose language to include a breadth of anal sex practices beyond penile-anal intercourse, to capture potential diversity in proclivities.

Analyses

After the completion of cognitive interviews, we assessed readability and grade-level statistics of the items in Microsoft Word and conducted informal review by MSM community members unfamiliar with the study objectives and hypotheses, then further revised our measures.

Results

Table I provides demographic characteristics. MSM who participated in chat interviews were more likely to report a non-Latino White racial identification (Fisher’s Exact Test X² = 5.54, p = .03) and annual income equal to or greater than $30,000 (Fisher’s Exact Test X² = 5.47, p = .03).

Table I.

Demographic characteristics of participants across studies

	n (%)
	Study 1		Study 2	Study 3
	Key Informant Interviews	MSM In-depth Interviews	MSM Survey Completers	MSM Reporting Recent PAI
	n = 10	n = 25	n = 218	n = 1263
Age in years (M, SD)	44.6 (12.6)	34.0 (9.1)	36.4 (11.5)	36.1 (11.0)
Race and ethnicity
Latino (of any race)	3 (30)	6 (24)	63 (29.9)	270 (21.4)
American Indian/Alaska Native	0	0	7 (3.2)	17 (1.3)
Asian or Asian American	0	2 (8)	16 (7.3)	78 (6.2)
Black or African American	2 (20)	11 (44)	46 (21.1)	204 (16.2)
White or Caucasian (non-Latino)	5 (50)	8 (32.0)	78 (35.8)	673 (53.3)
Native Hawaiian or Pacific Islander	0	0	0	4 (0.3)
Bi- or multiracial, Additional	0	0	40 (18.3)	80 (6.3)
Education
Some college or less/2-year degree	1 (10)	12 (48)	63 (28.9)	461 (36.5)
4-year degree	1 (10)	8 (32)	78 (35.8)	415 (32.9)
Graduate degree	8 (80)	5 (20)	77 (35.3)	387 (30.6)
Income
Below $15,000 - $29,999	1 (10)	9 (36)	51 (23.4)	376 (29.8)
$30,000 - $59,999	1 (10)	14 (56)	78 (35.8)	383 (30.3)
$60,000 - $89,999	2 (20)	2 (8)	39 (17.9)	224 (17.7)
$90,000 or more	6 (60)	0	48 (22.0)	270 (21.4)
Housing
Rent	5 (50)	19 (76)	123 (56.4)	717 (56.8)
Own	3 (30)	5 (20)	70 (32.1)	368 (29.1)
Shelter/Dorm/Another’s home	2 (20)	1 (4)	25 (11.5)	178 (14.1)
Relationship status
Single/Casually dating	5 (50)	10 (40)	99 (45.4)	698 (55.3)
Boyfriend or girlfriend	0	12 (48)	36 (16.5)	146 (11.6)
Partner or lover	0	2 (8)	40 (18.3)	207 (16.4)
Legal, civil, committed partnership	5 (50)	1 (4)	43 (19.7)	202 (16.0)
Sexual orientation
Gay	7 (70)	21 (84)	178 (81.7)	1022 (80.9)
Bisexual	1 (10)	1 (4)	9 (4.1)	116 (9.2)
Queer	1 (10)	2 (8)	20 (9.2)	62 (4.9)
No label/Two-Spirit/Additional	0	1 (4)	10 (4.6)	59 (4.7)
Heterosexual	1 (10)	0	1 (0.5)	4 (0.3)
Sexual position preference
’Bottoming’ (receptive)	-	7 (28)	52 (23.9)	391 (31.0)
’Topping’ (insertive)	-	2 (8)	47 (21.6)	282 (22.3)
Versatile’ (both)	-	15 (60)	111 (50.9)	550 (43.5)
No preference/Not sure	-	1 (4)	8 (3.7)	40 (3.2)
HIV status
Never tested/Never received result	-	1 (4)	3 (1.4)	71 (5.6)
HIV seronegative	-	17 (68)	175 (80.3)	999 (79.1)
Current PrEP use	-	4 (22.2)	74 (41.6)	350 (35.0)
HIV seropositive	-	7 (28)	40 (18.3)	183 (14.5)

In Study 2, n = 216 for Income due to missing responses; in Study 3, n = 1253 for Income, Relationship status, and HIV status due to missing responses

Table II includes examples of stigma-related items intended to measure predominantly one of nine underlying though not mutually exclusive stigma categories. Within each category, we developed items to capture how that category might manifest as distinct forms of experienced, internalized and anticipated stigma. Cognitive testing indicated easier comprehension of experienced, internalized and anticipated stigma when each form of stigma was shown on a separate survey screen. Likewise, participants reported improved comprehension when items all aligned with the same prompt (e.g., “How much do you agree?”), rather than prompts with differing time frames (e.g., “In the past” for experienced stigma or “In the future” for anticipated stigma).

Table II.

Study 1 (Item Generation): Stigma categories and examples from the 63-item pool developed in Study 1 based on in-depth interviews with 10 key informants and 25 MSM participants

Theme	Definition	Examples from item pool
		Experienced	Internalized	Anticipated
A. Sexual Position	Valuation or devaluation based on whether a man ‘tops’ or ‘bottoms’ including gendered stereotypes about men who bottom.	People I know have shamed guys who bottom.	Bottoms shouldn’t tell tops how to fuck them.	Tops can sleep around, but people will look down on a bottom for doing that same thing.
B. Intersectionality	The combination of stigma toward anal sex with other forms of devaluation like race-based bias or gender stereotypes.	A lot of men have thought they knew what I wanted sexually, just because they saw my race or ethnicity.	I feel uneasy having anal sex with guys of a certain racial or ethnic background.	When a guy wants to have anal sex with me, I can’t tell if he’s attracted to me or fantasizing about my race or ethnicity.
C. Shame	Internalized aversion to anal sex including discomfort shame and guilt.	People’s attitudes about anal sex have made me feel worse about myself.	I may never let go of the shame I feel about anal sex.	Talking about anal sex with a health worker is almost always going to be awkward.
D. Disclosure Threat	Devaluation based on how much other people know about one’s engagement in anal sex; harm from disclosure to others; urges to conceal.	I’ve seen how hard life can be for guys who are really open about having anal sex.	It feels easier for me to talk about blowjobs than anal sex.	Telling someone how I have anal sex is risky.
E. Harm/Disgust	Devaluation of anal sex as unhealthy harmful to one’s health or potentially dangerous; prompts for aversion and avoidance.	In one way or another, people have told me anal sex is unhealthy.	In my mind, anal sex is always dangerous, no matter how safe you think you are.	Health workers will try to scare me about anal sex.
F. Omission	Devaluation of anal sex related to the absence of information and resources (for example sex education) or research including misinformation and unfounded myths or conclusions about anal sex.	I had to learn about anal sex on my own – nobody taught me the basics	I feel like I don’t know how to have anal sex properly	Most guys don’t understand how to ease into anal sex.
G. Appearance	Devaluation based on perceived appearance.	People have made fun of my butt in a hurtful way.	I’d feel uncomfortable having my asshole looked at directly.	I expect a sex partner to make a negative comment about my ass.
H. Cleanliness	Devaluation based on feces or contact with feces.	My sex partners have looked disgusted at the sight or smell of shit, even when it was an accident.	A ‘mess’ during anal sex is truly embarrassing.	A ‘mess’ during anal sex would ruin the mood for my sex partners.
I. Isolation	The absence of engagement or public discussion about anal sex devaluation of one’s concerns loneliness or isolation of one’s thoughts and feelings about anal sex.	In my experience, people don’t talk very openly about anal sex.	I often feel like nobody else shares my same issues about anal sex.	Even if someone brought it up, most guys would hide their true feelings about anal sex.

Item generation produced a 63-item pool, with roughly equal numbers of experienced (20), internalized (24), and anticipated stigma (19) items, all written on a 4-point Likert-type scale that ranged from 0 (disagree strongly) to 3 (agree strongly). Readability statistics in Microsoft Word indicated ease of reading at a 6^th grade level (Passive Sentences: 0%; Flesch Reading Ease Scale: 77.1; Flesch-Kincaid Reading Level: 5.7).

Discussion

The 63 culturally tailored items we developed spanned several stigma categories related to anal sexuality, with items specifically developed to capture distinctions between forms of experienced, anticipated and internalized stigma. We revised items to better capture participants’ experiential world. The final set included items that could be tested for scale performance and validity within a larger sample and eventually for associations with health behavior.

STUDY 2: ITEM REDUCTION

Materials and Methods

Participants

Recruitment relied on snowball and targeted sampling, including advertisement on social media and men-seeking-men platforms, electronic announcements to affinity groups, and emails to interviewees from Study 1. After recruitment began, we added a quota trigger to preclude enrollment of additional non-Latino White men once they comprised one-third of the initial targeted sample size, to reserve space for racial and ethnic minority men. Eligibility criteria repeated those for MSM in Study 1 but participants additionally needed to be new to the research project. We collected survey responses from January to February 2017.

Procedures

Consent repeated procedures from Study 1. Within each measure in the survey, we randomized items to lessen response bias. Respondents who completed the survey could enter a raffle to win one of three $50 gift certificates to an online retailer.

Measures for Convergent and Discriminant Validity

Internalized heterosexism was measured by the 5-item Revised Internalized Homophobia Scale [59]. Response categories ranged from disagree strongly (0) to agree strongly (4).

Sexual self-consciousness was measured by the 12-item Sexual Self-Consciousness Scale, which comprises two 6-item subscales of sexual self-consciousness and sexual embarrassment [60]. We modified items to specify anal sex and to be more inclusive of same-sex behavior. Response categories also ranged from disagree strongly (0) to agree strongly (4).

Analyses

We analyzed data in SPSS 19 [61] using exploratory factor analyses (EFA) [57]. We hypothesized that experienced, anticipated and internalized stigmas would form distinct factors if all 63 items were analyzed collectively. However, we also held a priori ideas about the nine categories delineated in Study 1 and wanted to explore the possibility that these categories might form factors within subscales for each of the experienced, anticipated and internalized forms of stigma. We therefore conducted EFA separately for each a priori subscale of experienced, internalized, and anticipated stigma. Prior to EFA, we examined basic psychometrics within each subscale (e.g., whether removal of any single item would increase reliability or scale variance, negative correlation between items).

We conducted EFA with promax rotation under the assumption of moderate factor correlation within each form of stigma. Within each subscale for experienced, anticipated and internalized stigma, we extracted factors based on a visual inspection of scree plots and a statistical criterion from parallel tests, examining the point at which eigenvalues exceeded those in the raw data [62].

To maximize power and because we intended to test factors associated with limited engagement in health practices and services by MSM [11], we included respondents who did not complete the full survey, but who did complete each measure under analysis, which resulted in varying analytic samples for each measure. To determine undue influence by respondents who dropped out of participation, we conducted all final EFA solutions a second time among only those who submitted responses to all three scales (n = 230). Results did not substantively change.

We used an algorithm of item performance in the pattern matrix as a decision guide. We eliminated items with low factor loading (≤ .40) or with substantial cross-loading (≥ .30) [63]. The anticipated stigma measure was presented last among the forms of stigma, and included relatively fewer respondents (n = 230 v. n = 245 for experienced and n = 254 for internalized). We therefore used a lower maximum factor loading (.30) with minimal cross-loading (at most .20) for anticipated stigma. We aimed to balance the elimination of badly performing items and factor solutions with the goal of a mostly harmonious, fairly identical solution across all forms of experienced, internalized and anticipated stigma. When our algorithm did not provide a clear path for retention, we considered relevance of an item based on its theoretical importance in our conceptual model [11]. We then repeated EFA and item elimination at most twice, and chose models with relatively low factor correlation (r < .6).

As a preliminary assessment of convergent and discriminant validity, we examined associations with validated measures using Spearman’s rho (r_s). For our subscale of internalized anal sex stigma, we hypothesized convergence both with internalized heterosexism and the overall scale of sexual self-consciousness. While internalized heterosexism may also be associated with experienced and anticipated anal sex stigma, we strove in Study 1 to distinguish between these three forms of stigma as divergent phenomena, and therefore hypothesized lower magnitude associations as a preliminary assessment of discriminant validity.

Results

The survey received a total of 857 views. We checked key variables [64,65] to examine and exclude 8 potentially careless, repeat or fraudulent responses. Sixty-seven respondents did not complete eligibility criteria, 295 did not meet eligibility criteria, and 219 were excluded by our limit on non-Latino White participation. Of the remaining 268 respondents, 218 (81.3%) completed the survey, with no demographic differences between partial and complete respondents (all p > .05). Median completion time was 24 minutes. Within each EFA sample (n = 230 for anticipated, n = 245 for experienced and n = 254 for internalized stigma), no items were missing for more than 1% of the sample.

Sample characteristics are presented in Table I.

Item scale-correlation and scale-variance indicated that removal of any single item would not increase either reliability or scale variance. Most items were significantly and weakly positively correlated (r < .40); none were significantly and negatively correlated.

Within each form of stigma, parallel test results indicated more factors than the scree plots. We explored these higher factor solutions, moving toward models that better fit the scree plots.

As seen in Table III, of the 63 items developed in Study 1 and assessed empirically in Study 2, the within-scale analyses for experienced, anticipated and internalized stigmas resulted in 3-factor solutions for each form of stigma, with support for overall retention of 30 items. This included: (a) 11 items for internalized stigma (M factor loading = .59; variance explained = 38.0%); (b) 11 items for anticipated stigma (M factor loading = .64; variance explained = 38.6%); and (c) 8 items for the externalized stigma (M factor loading = .63; variance explained = 40.7%). Factor correlations were no greater than .40. However, these within-scale analyses included factors characterized by cross-loading and factors indicated by only 1 or 2 items.

Table III.

Study 2 (Item Reduction) and Study 3 (Scale Validation): Loadings from exploratory factor analyses (EFA) of retained anal sex stigma items

	Study 2			Study 3*
	EFA performed within each a priori set of items for Internalized, Experienced, and Anticipated Stigmas			Final EFA performed among 38 items† retained from Study 2
				Provider stigma	Self-stigma	Omission of Information
				% variance explained
				21.1	9.9	5.4
	Internalized Factors (n = 254)
	1	2	3
	% variance explained
A priori internalized stigma items	21.9	10.9	5.3
1. I hate myself for feeling the way I do about anal sex.	.80				.75
2. When I have anal sex, I feel like I’ve done something unhealthy.	.69				.58
3. I often feel like nobody else shares my same issues about anal sex.	.59				.44
4. I may never let go of the shame I feel about anal sex.	.58				.68
5. I feel like I don’t know how to have anal sex properly.	.56				.44
6. A ‘mess’ during anal sex is the bottom’s fault.		.59
7. Farting during anal sex is unacceptable to me.		.51
8. I can tell whether a guy tops or bottoms just by looking at him.		.50
9. Compared to feminine guys, masculine guys are much better at topping.		.47
10. Even with good friends, I’d feel uncomfortable talking about anal sex.			.65
11. Anal sex is a very personal, private topic.			.57
12. In my mind, anal sex is always dangerous, no matter how safe you think you are.					.42
13. Anal sex is always going to involve some amount of pain.
14. I’d feel uncomfortable having my asshole looked at directly.
	Experienced Factors (n = 245)
	1	2	3
	% variance explained
A priori experienced stigma items	19.3	12.5	9.0
15. It’s a hard life for guys who are really out and open about having anal sex.	.65
16. In my experience, people usually don’t like to talk very openly about anal sex.	.54					.55
17. Experience tells me most people think anal sex is disgusting, even if they’ve never said it aloud.	.52					.41
18. People think that a guy who enjoys getting fucked must want to be a woman.	.43
19. People have often assumed that I’m a ‘top’ or a ‘bottom’ just by looking at me.		.91
20. A lot of men have thought they knew what I wanted sexually, just because they saw my race or ethnicity.†		.43
21. Health workers have ignored my concerns about anal health.			.76	.46
22. I’ve been shamed or lectured about anal sex by a health worker.			.61	.64
23. Most guys I’ve had sex with really didn’t know how to have anal sex properly.						.54
24. Most guys I’ve been with think bottoming is supposed to be painful at first.
25. Growing up, I heard things like, “All gay guys love to take it up the ass.”
26. Growing up, the main message I got about anal sex was “It’s dangerous.”
27. I had to learn about anal sex on my own – nobody taught me the basics.
	Anticipated Factors (n = 230)
	1	2	3
	% variance explained
A priori anticipated stigma items	24.1	9.6	4.9
28. Health workers would treat me badly if they knew the ways I have anal sex.	.79			.75
29. Health workers will try to scare me about anal sex.	.70			.78
30. If they knew the ways I have anal sex, most health workers would shame or lecture me to stop.	.68			.66
31. Most health workers probably wouldn’t want to give an anal exam.	.53
32. There’s no point in talking about anal sex with a health worker.	.51
33. Telling someone how I have anal sex is risky.	.46
34. People’s assumptions about whether I’m a ‘top’ or ‘bottom’ are really annoying.		.67	-.22
35. I feel like I have to prove to guys that I’m masculine enough to top them.		.48
36. When a guy wants to have anal sex with me, I can’t tell if he’s attracted to me or fantasizing about my race or ethnicity.†		.45
37. Even if someone brought it up, most guys would hide their true feelings about anal sex.		.44				.52
38. Most guys don’t know how to prepare themselves for bottoming.			.79			.58
39. Most guys don’t understand how to ease into anal sex.						.67
40. Tops can sleep around, but people will look down on a bottom for doing that same thing.

Factor loadings < .2 are suppressed; all factor loadings were significant at p < .001

^*Study 3 EFA (n = 817) performed on 30 items with empirical support from Study 2 as well as 10 additional items (italicized) with relevance to a conceptual model and potential for empirical support in a larger sample

^†After initial EFA, two conceptually distinct items were excluded from the final EFA

Additional items not retained by EFA still had relevance with respect to our conceptual model [11] and the potential for empirical support in a larger sample. We therefore retained an additional 10 items (italicized in Table III).

In convergent and discriminant validity analyses, we included all 40 items. Appendix I includes means, Cronbach’s α, and correlations among respondents who completed the survey, and indicates moderate convergent and discriminant validity, with nearly all associations significant though many at low magnitude. Specifically, internalized stigma was correlated with internalized heterosexism (r_s = .42, p < .001), and more strongly correlated with sexual self-consciousness (r_s = .48, p < .001). Experienced and anticipated stigma were also significantly correlated with internalized heterosexism, at lower magnitude (r_s = .15, p < .05; r_s = .20, p < .001) than internalized anal sex stigma, indicating a degree of discriminant validity.

Discussion

Study 2 provided preliminary evidence for construct validity of our measure of anal sex stigma. Correlations between referent validated instruments were significant and consistent with our a priori hypotheses. Factor analyses helped inform a reduction in the number of overall items, from 63 to 30. For each form of stigma, there was support for a 3-factor solution. However, these solutions included some problematic factors, characterized by cross-loadings, or factors indicated by only 1 or 2 items. Additionally, 10 items we considered to be critical were not supported empirically. Given these findings, in addition to our relatively modest sample size and our a priori notions about the categories that might form factors from within-scale EFA (i.e., within subscales for experienced, anticipated and internalized stigma items), we decided to repeat EFA in a larger sample in Study 3. This time, however, we aimed to explore the factor structure of all 40 retained items together, as a cross-validation test of findings from Study 2, before moving to confirmatory factor analysis.

STUDY 3: SCALE VALIDATION

Materials and Methods

Sample

Recruitment followed the procedures from Study 2, with additional advertisement on men-seeking-men geolocation apps and electronic announcements to websites, blogs, organizations, and clubs across the U.S. Procedures related to eligibility criteria and assessment of careless responses repeated Study 2. From the start of Study 3 recruitment, we included the quota trigger from Study 2, to reserve space for racial and ethnic minority men. The survey occurred over 10 weeks, between July and September 2017.

Procedures

We repeated procedures from Study 2, with the exception that respondents who completed the survey could now enter a raffle to win one of twelve $50 gift certificates.

Measures for Convergent and Discriminant Validity

We assessed convergent and discriminant validity in relation to the broader experiences we heard during in-depth interviews [11]. This included downstream effects on sexual satisfaction and pleasure, discomfort discussing sexuality and anal sex practices with healthcare workers, difficulty accessing emotional and informational social support, and intersectional stigma. Our aim was to ensure that the new measure continued to speak to the experiential world of interview respondents while also seeking brevity within the online survey, which included several HIV-related measures for subsequent testing of our conceptual model [22].

Sexual self-consciousness was assessed with the same 12-item scale used in Study 2 [60].

Social support, based in an 8-item single subscale of the Medical Outcomes Study Social Support Scale (MOS-SSS) [66], was adapted to anal sex to capture the availability of emotional and informational support (e.g., “Someone whose advice about anal sex I really want”) on a 5-point Likert response scale (None of the time to All of the time).

Sexual satisfaction involved a 2-item measure (“How much pleasure did you get from having anal sex?” and “How satisfied have you been with your anal sex life?”) adapted from the Changes in Sexual Functioning Questionnaire [67] and the Sexual Satisfaction Scale [68], with 5-point Likert response options ranging from No pleasure to Great pleasure and None of the time to All of the time.

Perceived discrimination was measured with the abbreviated version of the Everyday Discrimination Scale [69], a validated 5-item scale of perceived frequency of experienced discrimination (e.g., “You are treated with less respect than others”) on a 6-point Likert response scale ranging from Never to Almost everyday.

Comfort discussing sexual orientation and anal sex were measured with two questions about a respondent’s comfort talking with medical providers, one about their sexual orientation and another about their specific anal sex practices, on a 5-point Likert response scale ranging from Not at all comfortable to Extremely comfortable.

Analyses

Validation procedures included EFA with the addition of confirmatory factor analysis (CFA) [57]. Respondents who completed through the stigma item screens were randomly assigned within SPSS to either an EFA (n = 817) or CFA (n = 788) sample. The two samples did not differ significantly on any demographic variables (p < .05).

As noted in the Discussion of Study 2, we repeated the EFA in the larger sample of Study 3, examining together all 40 items from Study 2, to cross-validate findings from Study 2 before validating again with CFA. Given our larger sample size, we used the more stringent EFA algorithm from Study 2 for item performance in the pattern matrix (eliminating items with factor loading ≤ .40 or substantial cross-loading ≥ .30), adding examination of the structure matrix [63]. We used Missing Values Analysis in SPSS to produce an estimation maximization (EM) correlation matrix, to minimize the effects of bias in EFA [70].

For CFA, analyses were performed in MPlus v. 8 [71], and relied on the comparative fit index (CFI), Tucker-Lewis index (TLI), and the root mean square error of approximation (RMSEA), alternative fit indices given our large sample size and the likelihood of a significant X² Goodness of Fit test statistic [72]. We abided by accepted recommendations for model fit (CFI and TLI > .9; and RMSEA < .06, relaxed to < .08 in most circumstances) [63,73]. Missing data were less than 0.1% at the item-level across participants. We used the robust weighted least squares (WLSMV) estimator that produces mean- and variance-adjusted chi-square statistics, given the small range of ordinal Likert response categories (four) in our measure [71,74].

Our examination of dimensionality of all 40 items together lowered expectations for an a priori factor structure and therefore limited our ability to set a priori hypotheses for convergent and discriminant validity. However, we hypothesized that a validated measure for perceived discrimination would be more strongly and positively associated with experienced and anticipated than internalized stigma; sexual self-consciousness would mirror associations from Study 2; and that social support, sexual satisfaction, and comfort discussing anal sex would be negatively associated with each form of stigma, but comfort most strongly correlated with anticipated stigma than with experienced and internalized stigma.

Results

The survey received a total of 4609 views. After removal of 124 potentially careless or fraudulent responses (2.7%), 455 did not complete eligibility criteria, 861 did not meet eligibility criteria, 533 met criteria but did not respond beyond the information statement, and 700 were excluded by our limit to recruitment of non-Latino White participants. Of the remaining 1936 respondents, 1387 (71.6%) completed the survey. EFA and CFA relied on partial and complete respondents, as in Study 2.

Compared to the 1387 survey completers, the 549 partial completers were significantly more likely to report younger age; recruitment from men-seeking-men apps; Black/African-American racial identification; a relationship status other than marriage/civil partnership/commitment ceremony; being in an open relationship; lower income; lower education; less ‘outness’ about attraction to men; living in someone else’s home; residing in a small town; living in the Southern region; never testing for HIV or never receiving a result; and to have skipped questions about sexual orientation and the gender of sexual partners over the past 3 years (p < .05).

Participant characteristics are presented in Table I for the 1263 MSM who completed the survey and who reported penile-anal intercourse in the past 3 months.

Initial EFA of the 40 stigma items suggested a 4-factor solution, with only two items loading on the fourth factor. These items (“A lot of men have thought they knew what I wanted sexually, just because they saw my race or ethnicity” and “When a guy wants to have anal sex with me, I can’t tell if he’s attracted to me or fantasizing about my race or ethnicity”) appeared to measure a distinct factor (r = .51, p < .001), the intersectional stigma category identified in Study 1 (Table II). These items assessed the co-occurrence of racial discrimination and anal sex stigma, thereby making distinctions between the two harder to detect, with low likelihood of generalizing across subpopulations of MSM. We therefore removed these and conducted another EFA on the remaining 38 items, arriving at a 3-factor solution that appeared to measure distinct factors across 17 items (as seen in Table III, which also indicates Study 2 items not retained after Study 3 EFA because their loadings were < .20). These factors aligned more with the content of interviews and the nine stigma categories we delineated in Study 1 than our originally planned forms of experienced, internalized and anticipated stigma. Overall, analyses supported the retention of most of the internalized stigma items from Study 2, now labeled self-stigma, as well as experienced and anticipated stigma items with high loadings from Study 2, now labeled either as provider stigma or, with slightly lower Study 2 loadings, omission of information. This final set of stigma scales included 3 of the 10 items excluded by EFA in Study 2 that we retained for examination in Study 3, specifically items 12, 23 and 39 in Table III.

The CFA of our 17-item scale (Table IV) demonstrated acceptable fit (CFI = .94, TLI = .93, RMSEA = .06, 95% CI: .06–.07) despite a significant chi-square test of difference (X²(3) = 303.08, p < .0001). Six items appeared to measure a latent factor of omission of information and concealment, which was a combination of perceived/experienced omission of knowledge among sex partners (e.g., “Most guys I’ve had sex with really didn’t know how to have anal sex properly”) and anticipation of concealment or silence in general (e.g. “Even if someone brought it up, most guys would hide their true feelings about anal sex”). Six items more clearly formed a latent factor of internalized stigma (e.g., “I hate myself for feeling the way I do about anal sex”) that also contained internalization of omission of information (e.g., “I feel like I don’t know how to have anal sex properly”). The third factor contained 5 items, was specific to maltreatment by health workers, and combined elements of experienced (e.g., “Health workers have ignored my concerns about anal health”) and anticipated stigma (e.g., “If they knew the ways I have anal sex, most health workers would shame or lecture me to stop”).

Table IV.

Study 3 (Scale Validation): Confirmatory factor analysis of 17-item Anal Sex Stigma Scales (n = 788)

		Loadings (Uniqueness)
	Study 2 Items (Table III)	1	2	3
Self-stigma
I hate myself for feeling the way I do about anal sex.	#1		.79 (.37)
When I have anal sex, I feel like I’ve done something unhealthy.	#2		.78 (.39)
I may never let go of the shame I feel about anal sex.	#4		.74 (.45)	`
I often feel like nobody else shares my same issues about anal sex.	#3		.64 (.59)
I feel like I don’t know how to have anal sex properly.	#5		.59 (.65)
In my mind, anal sex is always dangerous, no matter how safe you think you are.	#12		.49 (.76)
Provider stigma
Health workers would treat me badly if they knew the ways I have anal sex.	#28	.81 (.35)
If they knew the ways I have anal sex, most health workers would shame or lecture me to stop.	#30	.79 (.37)
Health workers will try to scare me about anal sex.	#29	.77 (.41)
Health workers have ignored my concerns about anal health.	#21	.68 (.54)
I’ve been shamed or lectured about anal sex by a health worker.	#22	.61 (.63)
Omission of information
Most guys don’t understand how to ease into anal sex.	#39			.65 (.57)
Even if someone brought it up, most guys would hide their true feelings about anal sex.	#37			.65 (.57)
Most guys don’t know how to prepare themselves for bottoming.	#38			.58 (.66)
Experience tells me most people think anal sex is disgusting, even if they’ve never said it aloud.	#17			.56 (.69)
In my experience, people usually don’t like to talk very openly about anal sex.	#16			.55 (.70)
Most guys I’ve had sex with really didn’t know how to have anal sex properly.	#23			.55 (.69)

Correlations between provider and self-stigmas (.24), provider and omission stigmas (.25), and self- and omission stigmas (.29) all p < .001

Given that we arrived at a factor structure that did not match our a priori intentions to develop subscales for experienced, internalized and anticipated forms of stigma, we examined correlations a posteriori between each subscale and validated measures. As seen in Appendix II, perceived discrimination was positively correlated with all subscales (provider stigma r_s = .28; both self-stigma and omission stigma r_s = .24). Sexual self-consciousness was positively correlated with self-stigma (r_s = .48) and omission (r_s = .39), and less so with provider stigma (r_s = .25). Sexual satisfaction was negatively correlated with self-stigma (r_s = −.40), less correlated with omission stigma (r_s = −.22), and minimally correlated with maltreatment (r_s = −.07). Social support followed a similar pattern. Finally, comfort discussing anal sex was negatively correlated with all subscales (self-stigma r_s = −.37; omission stigma r_s = −.29; provider stigma r_s = −.28).

Overall Discussion

We developed and validated a set of scales to measure stigma toward anal sex among cisgender MSM. Final subscales did not conform to our hypothesized distinctions among factors of experienced, internalized and anticipated stigma. The measure did, however, reflect the experiential world of in-depth interview participants in the form of self-stigma, provider stigma, and stigma related to the omission of information.

Future research may test the contributions of specific forms of devaluation on health and point to areas for intervention. Multiple devalued attributes can exist even in the same person [75], and differentiating between anal sex stigma mechanisms and intersectional stigma (e.g., HIV stigma, heterosexism, racism) is important to the extent that they predict different outcomes at different magnitudes and may help identify critical components for intervention across stigma targets and health behaviors [50,51]. In our stigma measure, we intentionally eliminated intersectional items that involved devaluation by race and ethnicity; only two items remained after factor analyses and these measured a conceptually distinct phenomenon that combined anal sex stigma and racism. However, this theme of intersectional devaluation was clearly evident in interviews, and literature suggests interactions between racial and sexual devaluation that negatively influence disclosure of sexual behavior [76–80].

Our findings need to be considered in light of several limitations. We relied on self-report and noted elsewhere [11] that even during anonymous in-depth interviews participants were reticent to discuss certain aspects of their experience for fear of being ridiculed or devalued. A future option might be to measure implicit associations [81] rather than to rely on self-report, which may be biased by a respondent’s need to navigate the potential aversive consequences of disclosing stigma (e.g., experiencing emotional vulnerability, remembering noxious experiences) [82,83]. As an additional limitations, the internal consistency of items was acceptable, for a new measure, but we do not yet know whether measurement remains consistent over time in the same person, which items might be more or less stable and, relatedly, which subscales and constructs are amenable to change over time. We developed items based on interviews with respondents living in the United States and cannot speak to their relevance in other social contexts. However, qualitative work outside the U.S. suggests that stigma toward anal sex is a relevant construct among MSM in both high- and low-stigma contexts [21,84–86] as well as additional populations [87,88], particularly the domains related to the omission of information [21,86,89].

In sum, cisgender MSM endorsed the relevance of anal sex stigma as a social force in their lives. A lack of information and silence about the topic of anal sex, past and anticipated future maltreatment by health workers, and the internalization of shame all captured aspects of men’s experiential world. Future work can now test how these factors might influence behavior, in particular in relation to health disparities among MSM.

Realizamos una serie de estudios para validar una nueva escala de estigma hacia el sexo anal, adaptada culturalmente para hombres cisgéneros quienes tienen sexo con hombres (HSH). En el primer estudio, se hicieron entrevistas a profundidad (N = 35) para generar ítems. En el segundo estudio, reducimos el número de ítems por medio de una encuesta en línea (N = 268), y probamos la función, la dimensionalidad, y la validez convergente y discriminante. En el tercer estudio, reclutamos otra muestra en línea (N = 1605), aleatorizada al análisis del factor exploratorio o confirmatorio para finalizar la reducción de ítems, y después evaluamos la validez entre HSH sexualmente activos (N = 1263). Las últimas sub-escalas incluyeron auto-estigma (6 ítems, Cronbach’s α = .72), stigma del proveedor (5 ítems, Cronbach’s α = .79), y omisión de información (6 ítems, Cronbach’s α = .73; escala entera de 3 factores = .80). Desarrollamos una encuesta con 17 ítems, basada en la experiencia vivida de los HSH cisgénero. En el futuro, las investigaciones deberían examinar las asociaciones con los comportamientos saludables.