Abstract
Summary
We investigated the relationship of internalized homonegativity/homophobia (IH) to sexual risk behaviours among 216 Ugandan gay and bisexual men, using the 7-item IH scale previously developed on this population. IH was significantly associated with unprotected anal intercourse, and more so with unprotected receptive anal intercourse. Higher IH was also associated with more sex while intoxicated. There was a strong association between anal intercourse of any type and IH, suggesting a complex relationship between anal sex and identification with, or internalization of, homonegativity/homophobia. Specifically, it may be the anal component of sex rather than the sex with another man that is seen as labeling one as homosexual or stigmatizing. Those men who stated that they engaged in sex with other men for love, rather than for the physical feeling or for money, had higher IH scores. These data suggest that there may be an interactive relationship between IH and sexual behaviour, with greater internalization being associated with more stereotypically gay activities, which in turn may lead to more self-identification as gay and thus greater susceptibility to internalization.
Keywords
Introduction
There has been considerable discussion of the psychological variables that may be directly or indirectly associated with unsafe sexual behaviour, and particularly HIV/STI risk behaviour. 1 One such variable that is implicated in risk behaviour in some studies is internalized homophobia/homonegativity (IH). IH is the internalization by gay and bisexual men of negative attitudes and assumptions about homosexual people, 2 and is defined by Meyer and Dean (1998) as ‘the gay person's direction of negative social attitudes toward the self, leading to a devaluation of the self and resultant internal conflicts and poor self-regard’ (p. 161). IH may lead to feelings of guilt, inferiority or inadequacy and lack of self-worth. In turn, these may lead to lack of acceptance or devaluation of the self, possibly including self-destructive behaviours. Cabaj 3 argued that IH is the main dynamic operating in neurotic behaviour in gays, bisexuals and lesbians. This may still be accurate in highly homophobic environments.
The relationship between psychological variables such as neurosis and risk behaviours such as ‘promiscuity’ was described by Kelus 4 from his large randomized controlled study of sexually transmitted infection patients in Poland. He indicated that there were two possible and opposing mechanisms: neurosis (psychological adjustment) leading to high partner numbers and higher risk of contracting infection, and neurosis leading to sexual inhibitions and subsequently lower partner numbers and risk. It is likely that IH may follow this pattern. This model suggests that mental health factors, which may include IH, may be a predictor of HIV risk behaviour through partner numbers and other risk behaviours.
In a comprehensive review, Meyer 5 summarizes the evidence and places IH into the conceptual framework of minority stress, in which stigma, prejudice and discrimination create a hostile and stressful social environment that causes mental health problems. Rosser et al. 6 examined the correlates of IH in a sample of Midwestern US gay and bisexual men and found it was significantly associated with depression, dysthymia, likelihood of being in therapy, poor sexual health, not being ‘out’, discomfort with sexual orientation and underdeveloped peer socialization. Subsequently, Ross et al. 7 reported that in HIV seropositive men who have sex with men (MSM), IH increased HIV transmission risk behaviour through two indirect pathways: sero-discordant unprotected anal intercourse through less frequent disclosure of serostatus to partners, and lower condom self-efficacy through lower sexual comfort. These data suggest that IH may increase risk through lower disclosure and discussion of HIV serostatus, and greater discomfort with discussion of and use of condoms.
While the relationship of IH to minority stress and mental health is relatively uncontested, there is debate about the impact of IH on risky sexual behaviour. Recently in their meta-analysis, Newcomb and Mustanski 8 have argued there is no direct relationship between IH and HIV risk behaviours. Rather, researchers would be well-served to consider abandoning the investigation of IH as a predictor of risky sexual behaviour. Ross et al., 9 however, have suggested that limiting studies reviewed by Newcomb and Mustanski to North American studies (and to direct, linear effects as measured by correlation coefficients) minimized the impact and variance of IH, and that a better test of the question would occur in more homophobic societies where the hostile environment is more pronounced and IH higher.
Uganda is an East African state which has recently sought to introduce the death penalty for homosexual acts, 10 and where ‘Gay Ugandans already describe a world of beatings, blackmail, death threats like “Die Sodomite!” scrawled on their homes, constant harassment and even so-called correctional rape’. 11 It is thus a good environment in which to study IH. This IH is also manifest in difficulties studying homosexual behaviour, with the study from which these data come being prematurely terminated due to public exposure and hostility. 12 This again indicates the level of public hostility to homosexually-related issues, including research. This level of public hostility appears not to have changed since these data were collected. A recent study on MSM in Kampala who had had anal sex in the past three months found a 14% HIV prevalence rate, with predictors of HIV infection being age over 25 (OR = 4.3) and ever being exposed to homophobic abuse (OR = 5.4), suggesting that homophobia is strongly associated with increased risk. The mechanism translating homophobic abuse to risk behaviour is likely to include IH. In order to investigate the relationships of IH to HIV risk behaviour, we carried out an investigation of the relationships between IH and HIV risk behaviours in a sample of 216 Ugandan gay and bisexual men.
Methods
Participants
The data were collected using respondent-driven sampling (RDS) in Kampala, Uganda, the capital and a city of some 1.2 million. RDS is a form of snowball sampling that allows researchers to make estimates about hidden population. 13 Self-identified adult gay and bisexual men (but not all MSM since non-gay or non-bisexually-identified MSM were not included) were recruited using RDS from eight seeds identified through an observation period and interviews with key informants, over a period of two months in 2004, and participants were reimbursed 4000 Uganda shillings (about $US2). Key variables for RDS adjustment (i.e. variables which were required to be non-significant across waves for the sample to have reached equilibrium) were education (completion of secondary school) and having a male sexual partner from outside of Africa. Participants completed an anonymous questionnaire and the data reported here are from the IH scale. Demographic details of the participants appear in Table 1. Complete (unadjusted) data on the IH scale were available on 216 participants. Additional details on recruitment methods, questionnaire and participants in this study are reported by Kajubi et al. 12 and Raymond et al. 14 The study was anonymous and was approved by the relevant US and Ugandan Committees for the Protection of Human Subjects (IRBs), and all participants provided informed consent.
Measures
Internalized homonegativity
The study used the 26-item Reactions to Homosexuality scale. 2 All items were formatted for a 7-point Likert-type response; 0 = strongly disagree, 6 = strongly agree. We analysed an updated short (7-item) version of the scale. 15 The shortened version includes three IH factors (personal comfort with homosexuality, 3 items; social comfort with gay men, 2 items; and public identification as gay, 2 items). The 7-item scale demonstrated invariance in its measurement properties by race/ethnicity and English/Spanish versions in a North American sample. Details and psychometrics of the scale in the Ugandan sample, which was normed on this Ugandan MSM sample and is cross-culturally invariant, are reported by Ross et al. 16 and items appear in the Supplementary Appendix.
Full details of the questionnaire and other measures appear in Kajubi et al. 12 Briefly, questions of homophobic reaction were asked as ‘Have you ever been exposed to any form of violence or abuse?'and if yes followed by ‘Was any of this violence or abuse the result of your being gay/bisexual?’ Listed (response yes/no) were: Physical violence (beaten); Verbal violence (threats, insults); Moral violence (deprivation, humiliation); and Sexual violence (forced to have sexual contact). Sexual behaviour in the past six months was measured on a grid by the interviewer, including each partner, and for anal sex, # times receptive anal, # times insertive anal, with this broken down into # times no condom, and # times no condom, high or drunk. Risk variables were divided into presence or absence of the risk behaviour (insertive or receptive anal intercourse without a condom) in the past six months.
Statistical methods
Unadjusted data were analysed using SPSS version 18. We chose to use the unadjusted data since we did not intend making inferences to the specific population but rather explore the possible link between IH and HIV risk. IH was computed following the scales developed by Ross et al. 16 in this sample, and items were recoded so that a high score reflected high IH. The total IH score was the summed total of the three subscale scores. T-tests (using separate variance estimates where Levene's F was significant at P < 0.05) were computed for risk variables and related variables (unprotected anal insertive and receptive intercourse in the past six months, anal sex while intoxicated, having sex for money, for love, for the physical feeling, number of life partners) and for binary demographic characteristics, and the 3 IH subscales and total score. For multiple independent variables, one-way analysis of variance (all analyses were two-tailed with significance set at the 5% level) was carried out.
Results
Demographic characteristics of the sample
Risk behaviours and internalized homophobia/homonegativity (mean ± SD)
†For all analyses, ‘no anal’ was significantly different from all other categories
*P < 0.01; **P < 0.05
Reasons for sex with men and internalized homophobia/homonegativity (mean ± SD)
For unprotected insertive anal intercourse in the past six months, Personal Discomfort and IH were both significantly predictive. In both cases, higher internalized homonegativity was associated with higher risk. A similar pattern was obtained for having unprotected anal sex while high or drunk, with those indicating intoxicated anal sex scoring higher in discomfort and internalized homonegativity on all the subscales and the total score. In both these analyses, those with the risk behaviour were compared against all other sexual behaviours (Table 2).
The IH subscores and total score for anal sex role were all higher for any anal sex and significantly lower only for those who did not engage in anal sex, suggesting that it is the anal component of the sex (rather than the male partner) which is most associated with internalized homonegativity (Table 2).
Making a comparison of risk behaviours (unprotected anal intercourse) against all other sexual behaviours, given the much lower IH scores for those engaging in non-anal practices, sets up any anal behaviours to have significantly higher scores. To remove this bias, we secondly compared anal unprotected sex only against safe anal sex of the same type (Table 2). Data indicate that despite the smaller n's, for receptive anal sex, IH was still significantly higher for the unsafe sex category (and differences were consistently in the same direction). For insertive sex, however, the differences in IH between safe and unsafe activity failed to reach significance.
Having sex with other men for money was significantly associated only with lower discomfort over public identification as homosexual; having sex with other men for the physical feeling was also associated with lower personal discomfort with homosexuality and social discomfort regarding homosexuality, plus lower total IH score. In contrast, those who indicated that they had sex with men ‘for love’ scored higher in discomfort on all scales than those who indicated that they did not have sex with other men primarily for love (Table 3).
Illustrating the relationships between one potential measure of risk, partner numbers, and the subscales and total scores for IH, it is demonstrated that for three of the four (excepting Personal Discomfort, where the relationship is linear), the relationship is U-shaped.
A total of 61 men (26.8%) reported being subject to some form of violence or abuse as a result of being gay or bisexual: of these, 32.8% indicated it was physical, 83.6% verbal, 42.6% moral (discrimination or humiliation based on being gay or bisexual) and 31.2% sexual: 62.3% indicated that they had been subjected to two or more forms of violence. The only IH subscale predictive of homophobic violence was Personal Discomfort (t = 1.88, df = 226, P = 0.05). Men who engaged in any anal sex were significantly more likely to have experienced violence (29.6%) than those who did not (16.7%: χ2 = 4.3, df = 1, P = 0.04).
Discussion
These Ugandan data are among the first from sub-Saharan Africa on gay and bisexual men and IH. Recently, Vu et al. 17 reported on IH in MSM in South Africa, and found that greater levels of IH were associated with lower education, higher levels of HIV misinformation, bisexual identity and higher levels of HIV-related conspiracy beliefs. The present data are of particular significance in coming from a country where MSM activity is severely proscribed, and where there was serious debate until a few months ago on introducing sentences up to and including the death penalty for some homosexual acts (the existing penalty is up to 14 years imprisonment). Only international outcry has prevented this death penalty legislation from being passed. More recently (2011), the country's leading gay activist was beaten to death after a newspaper published his photo with a banner ‘Hang Them’. Thus, IH can be examined in a hostile environment where there is likely to be high level of expressed homophobia/homonegativity. We have previously established the cultural invariance of the short form of the IH scale in the same sample of Ugandan gay and bisexual men. 16
These data demonstrate that there is a bivariate relationship between IH and its subscales and HIV-associated risk behaviour in this East African context. This is particularly strong for the type of sexual activity, specifically anal intercourse (any role) versus no anal activity. As these data are cross-sectional, it is possible that the anal activity may underlie the IH, since male homosexuality may be culturally defined by anal sex rather than the gender of the partner. It is likely that a complex interaction between IH and anal sex may occur. For example, anal sex may establish or enhance a view of oneself as homosexual, which may in turn make the individual more aware of, or more likely to internalize, anti-homosexual attitudes. These data suggest that the process of development of IH may be multifaceted and that qualitative and developmental methodologies might be best suited to its exploration. This is consistent with the work of Plummer, 18 whose Australian study on the development of anti-homosexual attitudes toward gay men implicated the linkage of perceived gender role deviance with homonegativity, to the point where homosexuality was socially constructed as the ‘antithesis of masculinity’.
Over a quarter (26.8%) of the men in our study reported some form of violence against them. Those engaging in anal intercourse reported almost twice as much violence as those not, and a higher Personal Discomfort score was also significantly associated with a report of violence. The relatively low level of association between violence and IH suggests that IH is associated much less with overt and targeted violence than with general untargeted anti-homosexual and homonegative societal reaction. These data are consistent with the recent work of Hladik et al., 19 also on MSM in Kampala, who found that homophobic abuse was suffered by 39% of their 2009-recruited sample (a nearly 40% increase from our 2004-recruited sample). While sampling differences (they recruited only men who had had anal intercourse in the past 3 months) may account for some of this difference, it is difficult to escape the conclusion that homophobic violence in Kampala has increased over time (given that the median age in both samples was 25).
More than half (52.5%) of the insertive, and 57.6% of the receptive, anal intercourse in the past six months in this sample was reported as unsafe. For unsafe receptive anal intercourse, compared only with safe receptive anal intercourse, IH was a significant predictor, suggesting that those with the greatest discomfort with their sexual orientation are more likely to be unsafe. This is consistent with the data of Ross et al. (2009), which found that IH was indirectly associated with HIV risk behaviour through discomfort with condom use or discussion of HIV status in a US sample of MSM. A similar pattern was found with higher IH predicting anal sex while high or drunk in the present data, suggesting that IH may also be associated with more sex while intoxicated. The lack of significant association in insertive anal intercourse, however, may be associated with that behaviour being seen as at lesser or no risk, or not ‘homosexual’ since it is insertive, and will require more in-depth qualitative research, particularly given the culturally and politically different context.
Data on IH as it relate to reasons for sex illuminates the association of IH with behaviour. There were several significant differences, with men reporting sex for money indicating significantly lower personal discomfort as gay, consistent with their presumed identification as heterosexual and seeing this situation as irrelevant. Those men who reported sex for love consistently reported higher IH scores, presumably because they were more likely to identify themselves as gay. For those men who reported sex with other men for the physical feeling, IH was consistently lower, again presumably for the converse reason that they may identify less as gay and more as having sex with men primarily for physical gratification, and thus not internalize the homophobic hostility or see it as applicable to them.
However, these data may not be generalizable beyond strongly anti-homosexual societies or beyond East Africa. The data may also be limited to those gay and bisexual men prepared to talk with researchers, and may have failed to sample those with the highest levels of internalized homophobia. Thus, they need to be treated with caution until replicated. A major study of IH in MSM in 38 European countries 20 confirmed that IH is associated with HIV risk behaviour, including lower HIV testing rates and lower condom use.
These data not only confirm that there are direct relationships between unsafe sexual behaviour and IH, but that the relationship may be complex in several ways that are not readily apparent with cross-sectional, or with quantitative, data. IH may be more closely related to class of risk behaviour (anal versus no anal sex, partner numbers) than to condom use, and the relationship may be bidirectional. There may, however, be confounders that are not readily apparent in the relationships between IH and sexual behaviour. Religion was not associated with IH and partner numbers and age were, but in the latter variable in a non-linear relationship.
Taken together, these data from gay and bisexual men in a culture where homosexual behaviour is strongly condemned, emphasize the importance of undertaking further research on IH and its development, and its direct and indirect relationships with both social and with sexual risk behaviours in MSM.