MEASUREMENT OR HISTORY of sexual behavior and sexual identity (homosexual, bisexual, or heterosexual) is an important variable in the design and targeting of sexually transmitted disease (STD)/HIV prevention projects and in provision of descriptions of populations infected or at risk of STD infection, as well as in clinical case management and partner notification. However, it has been recognized since early in the HIV/AIDS epidemic that the concordance between identity and behavior is imperfect. 1 Furthermore, definitions of what is considered homosexual, bisexual, or heterosexual behavior may vary significantly across cultures 2 or even within cultures, 3 creating discrepancies between reported behavior and sexual identity. Carrier 4 indicates that in Hispanic societies, the male who engages in receptive anal or oral sex may be labeled as homosexual, but not the active partner, on the basis of role rather than gender of partner. It has been suggested that men whose behavior and identity are discordant engage in multiple STD/HIV risk behaviors and greater sexual risk behavior with male partners. Reviewing the data, Doll and Beeker 1 suggest that where racial and ethnic community norms and values narrowly define gender roles and reject homosexuality as “unmanly” (e.g., African American and Latino populations), there will be a higher prevalence of bisexual behavior. There are few data on the situation for women.
Recently, HIV risk-related data have raised the issue of relatively high rates of discordance between reported sexual behavior and sexual identity. Goodenow et al, 5 in a large population-based study of male New England high school students, found that 69% of those with only same-sex partners considered themselves heterosexual, as did 31% of those with partners of both sexes. They suggest that because of young age and the continuing emergence of sexual identity, it is likely that the discordance will be in the direction of those whose behavior is most stigmatized (same-sex partners). More recently, Montgomery et al, 6 in a report on bisexual behavior among HIV-infected men who have sex with men (MSM) in Michigan, noted that among African American men, 34% had also had sex with women, compared with 26% of Hispanic men and 13% of white men. Concordance rates between sexual identity and sexual behavior in this MSM sample were as follows: for African American men who had sex only with men, 82%, and African American men who had sex with both men and women, 61%; for Hispanic men who had sex only with men, 87%, and Hispanic men who had sex with both men and women, 59%; and for white men who had sex only with men, 94%, and white men who had sex with both men and women, 56%. These data suggest that discordance rates between sexual identity and sexual behavior may be significant and vary by race/ethnicity.
We report on a study of street-recruited men and women in four major racial/ethnic groups and the concordance rates between reported sexual behavior and sexual identity.
Data for the present analysis came from a larger community-based anonymous survey designed to determine knowledge, misconceptions, and sources of information in minority populations regarding HIV transmission. The study relied on self-administered questionnaires and respondents were recruited from public parks, mass transit locations, malls, and shopping centers in southwest and downtown areas of Houston, Texas. These neighborhoods have substantial minority populations. Data were collected in January 1997 and June 1998. Inclusion criteria were age >18 years and ability to fill out a questionnaire in the English language. Trained interviewers asked for participation in the study, and all participants were advised that they could refuse to answer any questions and that participation was both voluntary and anonymous. Those who agreed to participate were given the questionnaire to complete and deposit in a sealed box: those who declined to participate were counted as nonresponders. Lack of time was the excuse given by the great majority of nonresponders, followed by lack of facility in English. Return of the questionnaire was taken as evidence of consent. More detail on the study is provided by Essien et al. 7 The study was approved by the relevant university human subjects review board.
The two variables reported in this study were sexual identity and sexual behavior, measured by the questions on the last page of the questionnaire: (1) “What was your frequency of sexual intercourse with partners of the opposite sex during the last 3 months?” and (2) “What was your frequency of sexual intercourse with partners of the same sex during the last 3 months?” For each question, one response was to be circled: never, less than 3 times a month, 1–6 times a week, or once a day. To the sexual identity question (“How do you identify yourself?”), the possible responses were heterosexual, bisexual, and homosexual. Data analysis consisted of cross-tabulating the reported sexual behavior, categorizing it as no sex in the past 3 months (“none”), sex with only same-sex partners (“homosexual”), sex with only opposite-sex partners (“heterosexual”), or sex with both same-sex and opposite-sex partners (“bisexual”). For the purposes of measuring concordance, the individual's reported behavior was compared with the reported sexual identity and considered concordant if the labeled behavioral category and self-reported identity matched. For analysis of primary source of income, respondents were grouped as those with legal employment, welfare, or social security (legal employment); those trading sex for money, sex for drugs, or sex for gifts/favors (sex work); and those who engaged in drug dealing or theft/hustling (illegal employment). Data were analyzed by calculation of percentages and by chi-square test (with Yates correction for discontinuity where appropriate; significance, P < 0.05), with use of SPSS 10.0.
Demographic data (N = 1494) are presented in Table 1. With the exception of Asian males, between one fifth and one quarter of respondents reported they had not had sex in the past 3 months. There were considerable differences between racial/ethnic samples in reported sexual identity, with higher proportions of white and Hispanic males describing themselves as homosexual and high proportions (20–38%) of both males and females describing themselves as bisexual. Concordance between reported sexual behavior in the past 3 months and sexual identity are reported in Table 2. Refusal rates were as follows: for African Americans, 48%; Hispanics, 44%; whites, 42%; and Asians, 43%. The rankings of the four racial/ethnic groups for discordance were identical to those for the proportion of the sample population indicating that they engaged in sex for money, drugs, or gifts (African American, 19.7%; Asian, 10%; Hispanic, 27.5%; and white, 51.4%).
Cross-tabulation between reported behavior and identity concordance rates (concordant versus discordant) and occupational status indicated that when racial/ethnic groups were combined, for those legally employed or on welfare or social security, concordance was 59.9%; for those involved in sex for money, drugs, or gifts, it was 50.4%; and for those involved in drug dealing or theft/hustling, 52.1% (chi-square = 12.6;df = 3;P = 0.006). Computation of concordance rates just for those legally employed or on welfare or social security revealed the following concordance rates for reported behavior/identity (males and females combined): African American, 49.7%; Asian, 75%; Hispanic, 67.7%; and white, 33.3%.
In all racial/ethnic and gender categories but one, the largest discordant group was those who described their sexual identity as “heterosexual” but reported sexual contact with both males and females in the past 3 months. These figures as a percentage of those who were sexually active in the past 3 months are reported in parentheses in Table 2. The exception was Hispanic men, for whom the largest discordant group was those who described themselves as bisexual but had sex only with women in the past 3 months (35% of those who reported themselves as bisexual).
These data must be interpreted with the caveats that they are based on a nonrandom convenience sample, that an English-language questionnaire was used, and that the sample was collected from public places, with a refusal rate approaching 50%. Those not fluent in English would be underrepresented, and those who regularly frequent public places would be strongly overrepresented. This latter point would inflate the proportion of unemployed and probably of those seeking sexual contact or drug-dealing. Since the analyses of concordance rates excluded those with no reported sexual behavior in the past 3 months, it may represent an overestimate of discordance. Discordance may also be overestimated by inclusion of bisexuals who had partners of only one gender in the past 3 months. On the other hand, limiting behavior to the past 3 months may significantly underestimate discordance.
This study raises significant sexual minority sampling issues, as the proportions of reported homosexual and bisexual respondents are an order of magnitude higher than those reported following population-based studies. 8 Because the study was based on street outreach to obtain responses on HIV/AIDS knowledge and questions about sexual identity and behavior were asked toward the end of the questionnaire, we might assume that street-outreach sampling in places of public congregation is likely to recruit a much higher proportion of homosexual/bisexual people and those engaged in sex work and illegal activities and that surveys relating to HIV/AIDS preferentially recruit more sexual minorities. However, our purpose was to determine concordance between self-reported sexual identity and sexual behavior, not prevalence of such reported behaviors.
These data suggest that there is relatively low concordance between reported sexual behavior and sexual identity and that it varies by race/ethnicity. The concordance is, contrary to previous speculation, lowest among the white respondents and highest among Asian respondents. In all cases except the African American sample, concordance is close between males and females in each racial/ethnic group. While the nature of this sample overemphasizes people spending more time in public places, including those with illegal activities or exchanging sex for drugs or money, even if those involved in commercial sex and illegal activities are excluded, the proportions of concordance still range between 66% and 25%. The ranks of the four racial/ethnic groups remain the same, although concordance rates rise markedly for African American and Hispanic populations when just those with legal employment or income are considered. As might be expected, concordance is lowest among those in some form of sex work and those involved in illegal activities as a primary source of income.
These data are consistent with the observation of Humphreys 9 30 years ago in the United States that of MSM, nearly half considered themselves heterosexual. However, the magnitude of the discrepancy has previously not been estimated across racial/ethnic groups with use of the same sampling methodology. Consistent with our findings, data on a large sample of mainly white young men in the northeast United States have demonstrated that more than two thirds of males with only same-sex experience and over a quarter of bisexually experienced males labeled themselves as heterosexual. 5 Further, these discordance rates are consistent with recent data on MSM who are HIV-infected.
Although the current study is based on a biased sample, the fact that the discordance between sexual identity and sexual behavior can be so high—and for both genders—is still surprising, although not inconsistent with recent data. 5,6 These data strongly suggest that the assumption of high concordance between sexual identity and sexual behavior may not hold across race/ethnicity, gender, or social strata. The distinction between doing and being, which is apparent in these data, may be based on stereotypes from studies that are biased in terms of sample culture, recruitment, and stability.
For a variety of reasons, including those based on economic necessity, stigmatization of homosexuality/bisexuality, or differences in cultural interpretation of behavior, sexuality may be more fluid than conceived by those who conceptualize sexual identity as a “master status.” As Chou 10 notes in his analysis of the lack of applicability of Western concepts of sexual identity in China, just because a person likes pork or beef doesn't mean that we label them “porkman” or “beefwoman.” A similar approach to sexual appetite as not conferring identity may be operating in this sample. McIntosh 11 has previously noted that people who do not identify with the classic Western white gay/lesbian role may not necessarily identify their behavior as homosexual: the development of the nomenclature “MSM” has underscored this point.
The implications of this for research measurement and for design and targeting HIV/STD prevention activities are to strongly reinforce previous cautions that reported sexual identity may be discordant with actual behavior. Relying on reports of sexual identity either for provision of HIV/STD prevention activities or for conducting screening in clinical practice would lead to missing between 21.6% and 65.3% of candidates. Furthermore, different cultural interpretations of what behaviors may be associated with sexual identity 3,4 also appear to have a significant influence on the relationship between behavior and self-classification of sexual identity.
Thus, these data have considerable implications for STD/HIV research measurement and STD preventive and clinical practice. 12 Questions on sexual behavior, rather than identity, should be asked, and such questions should not be optional or predicated on preceding questions about sexual identity. It is likely to be highly misleading to use self-reported sexual identity as a guide for the provision of preventive and clinical services across all racial/ethnic groups studied, and it seems that the fluidity of sexual behavior is not always adequately represented by self-labeling of sexual identity.
1. Doll LS, Beeker C. Male bisexual behavior and HIV risk in the United States: Synthesis of research with implications for behavioral interventions. AIDS Educ Prev 1996; 8: 205–225.
2. Tielman RAP, Carballo M, Hendriks AC. Bisexuality and HIV/AIDS. Buffalo: Prometheus, 1991.
3. Ross MW. A taxonomy of global behaviour. In: Tielman R, Carballo M, Hendriks AC, eds. Bisexuality and HIV/AIDS. Buffalo: Prometheus, 1991: 21–26.
4. Carrier J. Some reflections on ethnographic research on Latino and southeast Asian male homosexuality and HIV/AIDS. AIDS Behav 2001; 5: 183–191.
5. Goodenow C, Netherland J, Szalacha L. AIDS-related risk among adolescent males who have sex with males, females, or both: evidence from a statewide survey. Am J Public Health 2002; 92: 203–210.
6. Montgomery JP, Mokotoff ED, Gentry AC, et al. The extent of bisexual behavior in HIV-infected men and implications for transmission to their female sex partners. AIDS Care 2003; 15. In press.
7. Essien EJ, Ross MW, Linares AC, et al. Perception of reliability of HIV/AIDS information sources. J Natl Med Assoc 2000; 92: 269–274.
8. Laumann EO, Gagnon JH, Michael RT, et al. The Social Organization of Sexuality: Sexual Practices in the United States. Chicago: University of Chicago Press, 1994.
9. Humphreys RAL. Tearoom Trade. London: Duckworth, 1970.
10. Chou WS. Tongzhi: Politics of Same-Sex Eroticism in Chinese Societies. New York: Haworth Press, 2000.
11. McIntosh M. The homosexual role. Soc Problems 1968; 16: 182–192.
12. Ross MW, Channon-Little LC, Rosser BRS. Sexual Health Concerns: Interviewing and History Taking for Health Practitioners. Philadelphia: FA Davis, 1999.