Avoiding Heterosexist Bias in Psychological Research
Reprint
American Psychologist
September 1991, Volume 46, Issue No. 9, 957B963
©1991 by the American Psychological Association, Inc
Gregory M. Herek, University of California at Davis
Douglas C. Kimmel, City College, City University of New York
Hortensia Amaro, Boston University
Gary B. Melton, University of Nebraska-Lincoln
This article is based on the final report of the APA Task Force on Non-Heterosexist Research (Douglas C. Kimmel, Chair), which was convened by the Board for Social and Ethical Responsibility in Psychology (BSERP) in 1985. The authors (who constituted the task force) wish to thank Carol Burroughs, APA Staff Liaison, for her valuable assistance in completing the task force's mission. The authors also thank the many members of BSERP, the Committee on Lesbian and Gay Concerns, and the Board of Scientific Affairs, as well as the many APA members and staff who provided comments on the original task force report and earlier drafts of this article. We dedicate this article to the memory of our distinguished colleague and task force member, Barbara Strudler Wallston.
Correspondence concerning this article should be addressed to Gregory M. Herek, Department of Psychology, University of California, Davis CA 95616-8686.
The authors describe various ways that heterosexist bias can occur in scientific research and suggest ways that social and behavioral scientists can avoid it. Heterosexist bias is defined as conceptualizing human experience in strictly heterosexual terms and consequently ignoring, invalidating, or derogating homosexual behaviors and sexual orientation, and lesbian, gay male, and bisexual relationships and lifestyles. The deleterious scientific, social, and ethical consequences of such biases are discussed. Questions are provided for researchers to use in evaluating how heterosexist bias might affect their own selection of research questions, sampling, operationalization of variables, data collection, protection of participants, and dissemination of results. Suggestions also are offered for reducing heterosexist bias in academic journals, in textbooks, and in colleges and universities.
The social and behavioral sciences have an important role to play in increasing society's knowledge about and understanding of lesbians, gay men, and bisexual people. For example, empirical research by Kinsey and his colleagues (Kinsey, Pomeroy, & Martin, 1948; Kinsey, Pomeroy, Martin, & Gebhard, 1953), Ford and Beach (1951), and Hooker (1957) demonstrated that homosexual behavior is fairly common in the United States and in other cultures and is not inherently associated with psychopathology. These pioneering works continue to be cited today. Unfortunately, the bulk of scientific research has ignored sexual orientation and behavior or has uncritically adopted societal prejudices against gay and bisexual people. For example, Morin (1977) reviewed Psychological Abstracts for the years 1967-1974 and found that most published research on homosexuality treated it within the framework of a sickness model (for detailed discussions of the history of scientific research on homosexuality, see Bayer, 1987; Brub, 1989; Chauncey, 1982; Lewes, 1988).
Ever since 1975, when the American Psychological Association (APA) Council of Representatives adopted its resolution stating that "homosexuality per se implies no impairment in judgment, stability, reliability, or general social and vocational capabilities," the APA has taken a leading role among scientific organizations in "removing the stigma of mental illness that has long been associated with homosexual orientations" (APA, 1975, p. 633). In 1986, for example, the APA submitted a brief amicus curiae to the United States Supreme Court in the case of Bowers v. Hardwick, using data from the social and behavioral sciences to argue for the elimination of state sodomy laws (Melton, 1989). In 1988, the APA submitted a brief amicus curiae to the U.S. Ninth Circuit Court of Appeals in the case of Watkins v. U.S. Army, arguing that empirical data do not support the military's exclusionary policies toward lesbians and gay men (Melton, 1989). In 1986, the APA provided testimony to a congressional committee summarizing scientific research on the sources, prevalence, and consequences of hate crimes against lesbians and gay men (Committee on the Judiciary, 1986; Herek & Berrill, 1990).
Recognizing a general need for gay-affirmative scientific research, the APA Board for Social and Ethical Responsibility in Psychology (BSERP), in 1985, convened a Task Force on Non-Homophobic Research. The task force was charged to "assemble and prepare materials that can be used for educating psychologists about techniques for preventing homophobic bias in research" (Burroughs, 1985).1 In fulfilling its charge, the task force focused on the problem of heterosexist bias in psychological research, which it defined as conceptualizing human experience in strictly heterosexual terms and consequently ignoring, invalidating, or derogating homosexual behaviors and sexual orientation, and lesbian, gay, and bisexual relationships and lifestyles.
In the present article, which is based on the final report of the task force, we describe some of the principal ways in which heterosexist bias can enter each stage of the research process, and offer suggestions for avoiding such bias.2 The discussion is organized as a series of questions that investigators can use to evaluate their own research-projects. The questions are especially relevant to studies conducted with gay male, lesbian, or bisexual populations. Additionally, we suggest that they be used for any research that (a) might incidentally include lesbians, gay men, or bisexual people in a sample of the general population; (b) might affect participants' attitudes, assumptions, or beliefs about sexuality, human relationships, or gay and bisexual people; or (c) might provide results relevant to a general understanding of sexuality, relationships, mental health, minority groups, or related topics.
Because the questions that follow can best be answered by well-informed researchers, we urge all psychologists to learn more about lesbians, gay men, and bisexual people. This self-education should not be restricted to library research and consultations with colleagues, but also should include discussing research ideas and procedures with gay and bisexual women and men, and seeking information from gay community periodicals and other media. Although community consultations do not absolve researchers of primary responsibility for problems inherent in their procedures, they can help researchers to recognize or anticipate such problems and to respond to them sensitively (Melton, Levine, Koocher, Rosenthal, & Thompson, 1988).
Formulating the Research Question
Does the research question ignore or deny the existence of lesbians, gay men, and bisexual people? Failing to recognize that most adolescent and adult samples include some lesbians, gay men, and bisexual participants can weaken a research design. For example, equating interpersonal attraction with heterosexual attraction can cause a researcher to misapply theoretical concepts or to use experimental manipulations that are inappropriate for some participants. By considering male-male and female-female attraction as well as female-male attraction, researchers open themselves to the possibility of a broader understanding of interpersonal attraction and its relation to cultural concepts of gender and sexuality. Similar comments can be made concerning psychological research in many other areas, including stress and coping (which could consider the stress faced by subjects whose sexual orientation is stigmatized), adolescent development (which could address such specific challenges faced by gay adolescents as coming out), and prejudice (which could consider how the social and psychological dynamics of heterosexism compare with racism, anti-Semitism, and other out-group attitudes).
Does the research question devalue or stigmatize gay and bisexual people? Even research questions that appear to be scientifically neutral can be biased by equating an exclusively heterosexual orientation with being "healthy" or "normal." In addition to its potential for reinforcing the stigmatization of lesbians, gay men, and bisexual people, such research can be scientifically weak because it limits inquiry to a single aspect of a multifaceted phenomenon. For example, research on the development of sexual orientation (which treats heterosexuality, bisexuality, and homosexuality as equally requiring explanation) has the potential of improving scientific understanding of human sexuality. In contrast, research is unlikely to yield scientifically important findings if it seeks a cause only for homosexuality and omits heterosexuality from the investigation. Such research also can perpetuate societal prejudices against people who engage in homosexual behavior. Psychological research intended to prevent development of homosexual or bisexual identity, or to change or "cure" a homosexual or bisexual orientation once it has developed, is an even more dramatic example of heterosexist bias. Aside from the general failure of conversion therapies in most cases, the ethics of such attempts are highly questionable (Davison, 1991; Haldeman, 1991).
Does the research question reflect cultural stereotypes of lesbians, gay men, and bisexual people? Lesbians, gay men, and bisexual people do not share any common characteristics apart from their sexual orientation. Nevertheless, numerous stereotypes persist about them (Adam, 1978; Herek, 1991). Erroneously assuming that gay or bisexual people constitute homogeneous groups can lead researchers to overlook the mediating influences of other important variables (age, race, gender, ethnicity, class, etc.). The assumption of homogeneity also can foster unwarranted generalizations about all gay or bisexual men and women from data that were obtained from a single nonrepresentative sample. It also can limit the choice of research topics. For example, few researchers have addressed the experiences of lesbian and gay male people of color, who often encounter both racism and heterosexism (see, e.g., Hidalgo & Hidalgo-Christensen, 1976; Icard, 1986). One probable reason for this deficit in the literature is that researchers have failed to recognize the racial and cultural diversity of the gay community.
Does the research question implicitly assume that observed characteristics are caused by the subjects' sexual orientation? The familiar principle that correlation should not be confused with causation applies to research with lesbians, gay men, and bisexual people. Phenomena should not be assumed to result from sexual orientation simply because they are observed in the gay community. Alcoholism, for example, is a serious problem in some sectors of the gay community. Attributing it to homosexuality per se, however, exemplifies the fundamental attribution bias (Ross, 1977): It explains behavior entirely in terms of personal characteristics while ignoring situational factors (e.g., the sample was recruited from patrons of gay bars).
Sampling
To what degree is the sample representative? When results are to be generalized to the entire population of gay men, lesbians, and bisexuals, researchers should use standard sampling procedures in order to obtain the most representative sample possible (e.g., Sudman, 1976). As with any attempt at probability sampling, researchers should be aware of factors that reduce the sample's representativeness. These include general limitations on any sample (e.g., random-digit dialing techniques exclude persons without telephones from the sampling frame) as well as problems unique to a lesbian, gay male, or bisexual sample (e.g., because of societal stigma, an unknown proportion of eligible respondents will refuse to disclose their sexual orientation to an interviewer).
In the past, researchers often have assumed that lesbians, gay men, and bisexuals could not be sampled through probability methods because of their status as "hidden" minorities in the United States. Recently, however, survey items about sexual behavior and orientation have been successfully administered to probability samples in telephone interviews, face-to-face interviews, and self-administered questionnaires (Miller, Turner, & Moses, 1990). These studies suggest that the problem of nonresponse in surveys of sexual behavior or orientation, although serious, may not be qualitatively different from that encountered with other samples. Researchers, therefore, should seriously consider probability sampling as a strategy for research in this area.
If the sample is not a probability sample, does it include sufficient diversity to permit adequate assessment of relevant variables? When convenience samples must be used, researchers should fully describe their recruitment procedures and sample characteristics, discuss possible sampling biases, and identify particular groups (e.g., ethnic, age, social class) that are likely to be over- or underrepresented. The negative effects of sampling by convenience can be offset to a limited extent by using a variety of recruitment strategies and by targeting diverse sections of the community. Unless bar patrons or political activists are the specific population of interest, for example, restricting recruitment efforts to those individuals will yield a highly selective sample. Instead, researchers should seek participants from throughout the community (e.g., at coffeehouses, bookstores, community centers, parent groups, political organizations, religious organizations, bathhouses, service-oriented programs, social clubs, and student organizations). Respondents also can be successfully recruited through advertisements in community publications and electronic media, and through exit polls at targeted voting sites (e.g., gay-identified neighborhoods).
Is the sample appropriate for the research question? When sampling gay and bisexual people, researchers should ensure that the sample's composition permits valid tests of hypotheses. This requires that researchers be familiar with the participants' communities. For example, a study of changes in gay men's sexual behavior since the outbreak of acquired immunodeficiency syndrome (AIDS) should be sufficiently diverse in terms of age, ethnicity, level of sexual experience, and relationship status (e.g., uncoupled, sexually exclusive relationship, sexually nonexclusive relationship).
Research Design and Procedures
Is sexual orientation the variable of interest? Sexual orientation should be carefully distinguished from such variables as gender identity and gender-role conformity. Specific sexual behaviors should be distinguished from the long-term pattern of attractions, behaviors, and identity known as "sexual orientation." In cross-cultural research, cultural influences on how members of other societies conceptualize sexual behavior should be considered; the categories associated with what is called sexual orientation in Western cultures may not be applicable to all societies (Herdt, 1984; Herek, 1989; Ortner & Whitehead, 1981; Williams, 1986).
Is sexual orientation assessed appropriately? The method of assessing sexual orientation and behavior should fit the sample and the research question. For example, different conclusions are likely to be reached with each of the following operationalizations: (a) self-labeling by respondents as gay male, lesbian, bisexual, or heterosexual; (b) self-reports of sexual behavior during a specified time period (e.g., the previous year; since age 18); and (c) self-reports of sexual desires or fantasies during a specified time period. Additional differences will be observed if these questions are asked among different racial, ethnic, or language groups. For example, Peterson and Marin (1988) noted that Black and Hispanic men may be more likely than White men to have engaged extensively in homosexual behavior while still considering themselves to be heterosexual.
Recruitment procedures should not be confused with assessment of sexual orientation or behavior, because all members of a particular group do not necessarily share the same sexual orientation. Just as samples of the general population should not be assumed to be entirely heterosexual, for example, samples of patrons of gay bars should not be assumed to be entirely gay. They may also include heterosexual and bisexual persons who patronize bars for various reasons (e.g., to socialize with friends, to dance).
Are comparison groups appropriate to the research design? Comparison groups should be carefully selected to avoid confusing other variables with sexual orientation. For example, comparing married women with lesbians confuses relational status with sexual orientation. The two groups are not comparable because (a) not all married women are heterosexual, (b) some lesbians are married to men, and (c) many lesbians are in committed relationships. Depending on the research question, any of the following might be appropriate comparison groups: married heterosexual women with unmarried heterosexual women; married heterosexual women with lesbians in an ongoing relationship; and single lesbians with lesbians in ongoing relationships.
Do questionnaire items or interview protocols assume heterosexuality? Researchers should avoid inappropriate use of standardized measures that equate heterosexuality with sexuality. Such instruments should be modified when possible, or alternative instruments should be devised. For example, personality inventories that assess "need for heterosexuality" as a proxy for "need for sexual intimacy" do not accurately assess the experiences of lesbians, gay men, and bisexual people. Such measures also reinforce heterosexist assumptions among respondents. Generally, questions should not use language that assumes the respondent's heterosexuality. For example, items concerning a respondent's romantic or sexual partner should not reflect unwarranted assumptions about that partner's gender. Nor should questions about marital status be assumed to yield complete information about respondents' committed relationships. Because marital relationships between two persons of the same sex are not legally recognized in the United States, lesbians and gay men who are involved in an ongoing same-sex relationship are legally single. Similarly, attempts to obtain information about an individual's sexual behavior outside of a primary relationship can be limited when only terms such as premarital or extramarital are used. In addition to providing data about only a limited range of behaviors, these terms define all sexuality in relation to a heterosexual marriage, thereby discounting sexual experiences and relationships with members of the same gender.
Do the researchers' personal attitudes and feelings influence participants' responses? Many lesbians, gay men, and bisexual people are justifiably suspicious of all psychological research, which historically has labeled them as sick. Unless researchers are well-informed, candid, and sensitive to such suspicions, they are likely to obtain incomplete or inaccurate responses. Research personnel's lack of knowledge about and comfort with homosexuality and bisexuality can be communicated to participants in subtle ways (e.g., interpersonal distance, nervous speech, unfamiliarity with slang) that can affect research outcomes. Thus, whenever data collection explicitly includes information about sexual orientation or behavior, the members of the research team should educate themselves sufficiently so that they are familiar and comfortable with lesbian, gay male, and bisexual cultures.
Demand characteristics also can affect results when the researcher is openly lesbian, gay male, or bisexual. Participants may feel pressured to provide responses that are not accurate but that are socially desirable within their community. Because of the gay male community's emphasis on safer sex practices to avoid AIDS, for example, male respondents may be reluctant to admit to a gay researcher that they continue to engage in high-risk behaviors.
Do experimental manipulations presume that participants are heterosexual? Data collection procedures can evoke different responses from lesbian, gay male, and bisexual participants than from heterosexual respondents. For example, using "attractive male" confederates in a study of arousal and attraction will evoke different responses from lesbians and heterosexual women. Although the lesbians might well rate the confederate as attractive in an aesthetic sense, they are much less likely than the heterosexual women to translate this attraction into feelings of sexual arousal. Although random assignment of participants to conditions will help to avoid systematic biases in the data in such a situation, such biases might still intrude when small samples are used. Furthermore, using such a procedure conveys an assumption of heterosexuality to participants. Researchers could avoid such problems by advertising the study as one of heterosexual females, by asking lesbian participants to indicate their sexual orientation through an unobtrusive and anonymous procedure, or through simply including a manipulation check that assesses the participants' level of sexual arousal in the presence of the confederate. Generally, unless sexual orientation is an explicit variable in the research, procedures should be neutral concerning respondents' sexual orientations.
Protection of Participants
Is information obtained about sexual orientation and behavior truly confidential, or what are the limits to confidentiality? Because of the stigma attached to homosexuality in American culture, participants who disclose information about their sexual orientation or behavior to a researcher are at heightened risk for negative consequences if that information becomes known to others. At the same time, obtaining such information often is crucial to the success of empirical research that will have beneficial effects for gay communities and for society as a whole. Researchers should not be deterred from addressing issues of sexual orientation and behavior in their research, but should develop procedures for protecting the privacy and welfare of participants. Researchers should carefully consider potential risks to respondents from possible public exposure and subsequent discrimination or harassment, as well as possible psychological stress resulting from requests for disclosure, especially for participants who feel uncertain about their sexual orientation.
Researchers should make a special effort to learn the actual limits to confidentiality of their data, including potential legal requirements for involuntary release of data and legal means for protecting confidentiality. One strategy for protecting participants is to obtain a certificate of confidentiality pursuant to the Public Health Service Act (see Gray & Melton, 1985). When audits are conducted by federal agencies, the researcher should attempt to have them conducted on-site; otherwise the data might become accessible under the Freedom of Information Act (Morris, Sales, & Berman, 1981). If data are subpoenaed, the researcher should attempt to have the subpoena quashed or should otherwise contest it (see Gray & Melton, 1985; Knerr, 1982). Information about the limits of confidentiality and safeguards should be shared with participants.Because of the risks to confidentiality, researchers should avoid keeping identifying information about their data when not absolutely necessary. If data must be identified for later matching in longitudinal or follow-up studies, researchers should use code names or numbers to eliminate the need for keeping names on file.
Does the research procedure reinforce prejudice or stereotypes among heterosexual respondents? If researchers decide to introduce homosexual or bisexual orientation into the procedure (e.g., through a manipulation), they should carefully evaluate the likely effect of such a manipulation on all respondents (heterosexual, bisexual, lesbian, and gay male). For example, if a study exposes participants to gay-related stereotypes, the researcher should develop careful debriefing procedures to ensure that respondents' stereotypes are not increased through their participation. Another example would be a study of person perception that includes a manipulation of a target person's sexual orientation. Again, researchers should use the debriefing period to educate respondents about lesbian, gay male, and bisexual issues, and to counter the effects of any negative stereotypes or prejudice that may have been communicated during the study.
Does the procedure have negative effects on lesbian, gay male, or bisexual participants? Research procedures that explicitly disparage lesbians, gay men, or bisexual people are objectionable on ethical grounds. More subtle assumptions by researchers that all subjects are heterosexual also are problematic both because they negate the experiences of lesbian, gay male, and bisexual participants, and because they obscure the possible effects of a potentially important variable. When participants learn the study's purpose in the debriefing session, they may feel that their experiences do not count-that the researcher considers them to be merely a source of error variance. Such communication can affect respondents' feelings of self-worth as well as their level of respect for psychological research. Furthermore, such a procedure can reinforce heterosexism by encouraging participants to see the world exclusively in heterosexual terms.
Does the recruitment procedure intrude inappropriately on potential participants' privacy? In their quest for participants, researchers run the risk of being overly intrusive in their recruitment procedures, and thereby inadequately safeguarding potential respondents' anonymity or confidentiality. For example, researchers who recruit outside a gay bar or community center may encounter patrons whom they know personally but who have not voluntarily chosen to disclose their sexual orientation. Researchers should consult with colleagues and community members about how best to protect respondents' privacy and confidentiality in those situations. In general, the means of recruitment should be as unobtrusive as possible, with the level of intrusiveness balanced against the possible benefits that will result from the study.
Does the observation procedure intrude inappropriately on participants' privacy? Researchers conducting unobtrusive observation studies should attempt to obtain consent from participants. When such a study violates the participants' expectation of privacy, consent is usually necessary to protect the participants' autonomy and dignity. For example, following respondents from a gay bar to their home would violate their privacy. Simple observation in a public or semipublic setting is more ambiguous, depending on the study and the kinds of observations made.
Does the assessment of sexual orientation create stress for participants? Some participants may experience stress when asked about their sexual orientation and behavior. This is especially likely among those who have not previously labeled themselves (even privately) as lesbian, gay male, or bisexual, and among those who are especially vulnerable to stigma or discrimination (e.g., participants who are in the armed forces, institutionalized, concerned with child custody, or are members of a culture that stigmatizes gay or bisexual people more severely than the dominant culture does). Participants who are concerned about their privacy and about possible discrimination if their sexual orientation becomes known to others need protection as well as sensitive reassurance from the researcher concerning confidentiality.
To permit participants to give their informed consent before agreeing to participate, researchers should inform them about the general nature of the questions being posed. Debriefing procedures also should be designed to detect and alleviate any damaging consequences of the research procedure. In addition to other standard debriefing practices, researchers should make certain that respondents understand the goals of the research and should be prepared to provide referrals to gay-affirmative counselors or information sources when appropriate.
Interpreting and Reporting Results
Is an observed difference assumed to reflect a problem or pathology of lesbian, gay, or bisexual participants? Simply because lesbians, gay men, or bisexual people are observed to differ from heterosexual respondents, it should not be assumed that this reflects a deficit. Value judgments should not be made about different experiences in childhood and adolescence, for example, simply because one pattern is prevalent among heterosexual adults and another is prevalent among lesbian and gay male adults. Similarly, differences between heterosexuals and lesbian, gay male, or bisexual respondents within the normal range of psychological test scores should not in itself be judged as indicative of a deficit in one group (Gonsiorek, 1991).
Does the language reflect heterosexist bias? When reporting results, researchers should carefully avoid heterosexist language, including characterizations that pathologize or ignore the experiences of lesbians, gay men, or bisexual people (see APA Committee on Lesbian and Gay Concerns, 1991, this issue; Morin & Charles, 1983).
Are the limitations of the research findings stated appropriately? As in all areas of psychological research, care should be taken when generalizing from the research sample to larger populations. If the research findings are applicable only to heterosexuals or to specific segments of the gay community (e.g., urban White males), this fact should be clearly stated in the abstract and throughout the research report.
Has the researcher attempted to anticipate distortions or misinterpretations of findings by the lay public and in the popular media? Scientific research on lesbian, gay male, and bisexual populations often is disseminated to the general public through the popular media. For example, studies of developmental or physiological differences between homosexual, bisexual, and heterosexual individuals have been widely reported in popular media as indicating the possible discovery of a presumed "cause" for homosexuality. Such reports usually are followed by immediate discussion of how to prevent or "cure" a homosexual orientation. Although misinterpretations of research results in the mass media are beyond their direct control, researchers should nevertheless try to anticipate and prevent them. This can be accomplished in part by clearly discussing the limitations of results and methodology, taking care that interpretations do not go beyond the data, explaining the qualifications on findings, and interpreting the results for reporters in clear and unambiguous terms. Researchers should be particularly conscientious when their findings are released to the press prior to publication (as in some AIDS research) because their scientific colleagues cannot provide informed criticisms of the research.
Have the results been disseminated to research participants or to the larger gay community? After collecting data from the lesbian, gay male, and bisexual communities, researchers should disseminate the findings in understandable language to the members of those communities. This dissemination can be considered a partial repayment to the communities for participating in the research. It need not violate scientific rules regarding double or prior publication.
Creating an Intellectual Environment Free from Heterosexism
For the advancement of science and society, high-quality research is needed on a wide variety of issues related to sexual orientation. This article and the report (Task Force on Non-Heterosexist Research, 1986) on which it is based signify the commitment of the APA to fostering such research. Further steps clearly are needed as well, especially in the areas of publishing and graduate training.
Journal editors and reviewers should recognize the legitimacy of research involving sexual orientation, should evaluate such research with the same criteria used to evaluate research on other topics, and should not dismiss such research as overly specialized or as frivolous. We applaud the efforts that have been made by many APA journal editors to recruit reviewers from minority populations, including gay and bisexual people, and we encourage the continuation of these efforts.
Textbook authors and editors also have an obligation to avoid statements and explanations that are likely to reinforce heterosexism and foster negative self-images among lesbian, gay male, and bisexual students. For example, discussions of same-gender attractions should not be relegated to chapters on psychopathology and should not be concerned only with consideration of the causes of such orientations. Readers of textbooks should be taught that lesbian, gay male, and bisexual people have unique experiences and concerns that are appropriate topics of psychological study. Explicit mention of lesbian, gay male, and bisexual perspectives should be included for a wide variety of psychological topics, including human development, interpersonal attraction, health, attitudes, and stress and coping. Teachers and professors should actively seek textbooks that present an affirmative view of lesbians, gay men, and bisexual people, and should inform publishers that they are doing so.
Finally, non-heterosexist research will be facilitated by colleges and universities that sensitize their students and faculty to the concerns expressed in this article. In their hiring and promotion procedures, educational institutions should consider research associated with lesbian, gay male, and bisexual topics to be legitimate. Candidates for employment and promotion who conduct such research should be evaluated for the quality of their work, and should not be discounted because of the subject matter. Regardless of their sexual orientation, students should be encouraged to conduct research on lesbian, gay male, and bisexual issues, and to recognize those issues in their research on other topics. Providing such encouragement should not be solely the responsibility of lesbian, gay male, and bisexual psychologists; rather, all psychologists should educate themselves to issues of sexual orientation relevant to their own field of expertise.
Conclusion
The science of psychology requires studying human behavior in all of its diversity. It is most appropriate, therefore, that psychologists address issues relevant to sexual orientation in research with general samples and also conduct research with lesbians, gay men, and bisexual people from diverse cultural backgrounds. Because American society is heterosexist, however, research can reflect cultural ignorance, biases, and prejudices surrounding sexuality and sexual orientation. During the past 15 years, psychologists and other social and behavioral scientists have taken significant steps toward rejecting the negative value assumptions underlying earlier views of sexuality, and have begun to remove the stigma so long associated with homosexual and bisexual orientations. This has led to a new research paradigm that recognizes the legitimacy of lesbian, gay male, and bisexual orientations, behaviors, relationships, and lifestyles.
Because this affirmative approach is such a recent development in the social sciences, many implicit and overt prejudices remain to be overcome. Overcoming these prejudices will lead to better science, as researchers recognize the many ways in which heterosexist bias has influenced formulation of research questions, sampling procedures, methods and measures, and the interpretation of results. Overcoming these biases also will lead to more ethical science, as researchers learn how better to respect the dignity and worth of individuals, to strive for the preservation of fundamental human rights, and to protect the welfare of research participants.
Endnotes
The task force included one representative each from the Board for Social and Ethical Responsibility in Psychology, the Board of Scientific Affairs, the Committee on Lesbian and Gay Concerns, and APA Division 44, as well as one additional expert from the field. It met in December 1985 and completed its report the following year (Task Force on Non-Heterosexist Research, 1986). The task force chose from the outset to refer to itself as the Task Force on Non-Heterosexist Research and to title its report "Avoiding Heterosexist Bias: Guidelines for Ethical and Valid Research." The word homophobia was avoided for several reasons, including its implication that prejudice against gay men and lesbians is a form of psychopathology and its focus on the individual rather than on the larger cultural context. For the present article, heterosexism is defined as an ideological system that denies, denigrates, or stigmatizes homosexual behaviors and gay, lesbian, and bisexual identities, relationships, and communities (see Herek, 1990).
Researchers should display the same sensitivity to heterosexism as they do to other prejudices, such as those based on race, gender, age, and class. Valuable suggestions can be found in the report of the Ad Hoc Committee on Nonsexist Research (Denmark, Russo, Frieze, & Sechzer, 1988), which addresses problems of sexist bias in empirical inquiry. Because sexist bias also can interfere with research on lesbian, gay male, and bisexual populations, that document is a useful supplement to the present article.
References
Adam, B. D. (1978). The survival of domination: Inferiorization and everyday life. New York: Elsevier.
American Psychological Association. (1975). Minutes of the Council of Representatives. American Psychologist, 30, 620-651.
American Psychological Association, Committee on Lesbian and Gay Concerns. (1991). Avoiding heterosexual bias in language. American Psychologist, 46, 973-974.
Bayer, R. (1987). Homosexuality and American psychiatry: The politics of diagnosis (2nd ed.). Princeton, NJ: Princeton University Press.
Bérubé, A. (1989). Coming out under fire: The history of gay men and women in World War II. New York: Free Press.
Burroughs, C. (1985, November 28). Memorandum to members of the Task Force on Non-Homophobic Research from Carol Burroughs, Administrative Associate for Governance Affairs, American Psychological Association. On file at the APA.
Chauncey, G., Jr. (1982). From sexual inversion to homosexuality: Medicine and the changing conceptualization of female deviance. Salmagundi, 58-59, 114-146.
Committee on the Judiciary. (1986, October 9). Anti-gay violence: Hearing before the Subcommittee on Criminal Justice of the Committee on the Judiciary, House of Representatives (Serial No. 132). Washington, DC: U.S. Government Printing Office.
Davison, G. (1991). Constructionism and morality in therapy for homosexuality. In J. C. Gonsiorek & J. D. Weinrich (Eds.), Homosexuality: Research implications for public policy (pp. 137-148). Newbury Park, CA: Sage.
Denmark, F., Russo, N. F., Frieze, I. H., & Sechzer, J. A. (1988). Guidelines for avoiding sexism in psychological research: A report of the Ad Hoc Committee on Nonsexist Research. American Psychologist, 43, 582-585.
Ford, C. S., & Beach, F. A. (1951). Patterns of sexual behavior. New York: Harper.
Gonsiorek, J. C. (1991). The empirical basis for the demise of the illness model of homosexuality. In J. C. Gonsiorek & J. D. Weinrich (Eds.), Homosexuality: Research implications for public policy (pp. 115-136). Newbury Park, CA: Sage.
Gray, J. N., & Melton, G. B. (1985). The law and ethics of psychosocial research on AIDS. Nebraska Law Review, 64, 637-688.
Haldeman, (1991). Sexual orientation conversion therapy for gay men and lesbians: A scientific examination. In J. C. Gonsiorek & J. D. Weinrich (Eds.), Homosexuality: Research implications for public policy (pp. 149-160). Newbury Park, CA: Sage.
Herdt, G. H. (Ed.). (1984). Ritualized homosexuality in Melanesia. Berkeley: University of California.
Herek, G. M. (1989). Sexual orientation. In H. Tierney (Ed.), Women's Studies Encyclopedia Volume 1, (pp. 344-346). New York: Greenwood.
Herek, G. M. (1990). The context of anti-gay violence: Notes on cultural and psychological heterosexism. Journal of Interpersonal Violence, 5, 316-333.
Herek, G. M. (1991). Stigma, prejudice, and violence against lesbians and gay men. In J. Gonsiorek & J. Weinrich (Eds.), Homosexuality: Research implications for public policy (pp. 60-80). Newbury Park, CA: Sage.
Herek, G. M., & Berrill, K. (Eds.). (1990). Violence against lesbians and gay men [Special issue]. Journal of Interpersonal Violence, 5(3).
Hidalgo, H. A., & Hidalgo-Christensen, E. (1976). The Puerto Rican lesbian and the Puerto Rican community. Journal of Homosexuality, 2, 109-120.
Hooker, E. (1957). The adjustment of the male overt homosexual. Journal of Projective Techniques, 21, 18-31.
Icard, L. (1986). Black gay men and conflicting social identities: Sexual orientation versus racial identity. Journal of Social Work and Human Sexuality, 4, 83-93.
Kinsey, A. C., Pomeroy, W. B., & Martin, C. E. (1948). Sexual behavior in the human male. Philadelphia: W. B. Saunders.
Kinsey, A. C., Pomeroy, W. B., Martin, C. E., & Gebhard, P. H. (1953). Sexual behavior in the human female. Philadelphia: W. B. Saunders.
Knerr, C. R., Jr. (1982). What to do before and after a subpoena of data arrives. In J. E. Siever (Ed.), The ethics of social research: Surveys and experiments (pp. 191-206). New York: Springer-Verlag.
Lewes, K. (1988). The psychoanalytic theory of male homosexuality. New York: Simon & Schuster.
Melton, G. B. (1989). Public policy and private prejudice: Psychology and law on gay rights. American Psychologist, 44, 933-940.
Melton, G. B., Levine, R. J., Koocher, G. P., Rosenthal, R., & Thompson, W. C. (1988). Community consultation and socially sensitive research: Lessons from clinical trials of treatments for AIDS. American Psychologist, 43, 573-581.
Miller, H. G., Turner, C. F., & Moses, L. E. (Eds.). (1990). AIDS: The second decade. Washington, DC: National Academy Press.
Morin, S. F. (1977). Heterosexual bias in psychological research on lesbianism and male homosexuality. American Psychologist, 32, 629-637.
Morin, S. F., & Charles, K. A. (1983). Heterosexual bias in psychotherapy. In J. Murray & P. R. Abramson (Eds.), Bias in psychotherapy (pp. 309-338). New York: Praeger.
Morris, R. A., Sales, B. D., & Berman, J. J. (1981). Research and the Freedom of Information Act. American Psychologist, 36, 819-826.
Ortner, S. B., & Whitehead, H. (1981). Sexual meanings: The cultural construction of gender and sexuality. Cambridge, England: Cambridge University Press.
Peterson, J., & Marin, G. (1988). Issues in the prevention of AIDS among Black and Hispanic men. American Psychologist, 43, 871-877.
Ross, L. D. (1977). The intuitive scientist and his shortcomings: Distortions in the attribution process. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 10, pp. 173-220). San Diego, CA: Academic Press.
Sudman, S. (1976). Applied sampling. San Diego, CA: Academic Press.
Task Force on Non-Heterosexist Research. (1986). Avoiding heterosexist bias: Guidelines for ethical and valid research. Washington, DC: American Psychological Association.
Williams, W. L. (1986). The spirit and the flesh: Sexual diversity in American Indian culture. Boston: Beacon Press
