Note on Research Methodology: Methodological and ethical issues in research on lesbians and gay men

Abstract

The NASW Public Social Policy Statement on Gay Issues condemned sexual orientation discrimination in 1977, four years after the American Psychiatric Association deleted the diagnostic category of homosexuality from its Diagnostic and Statistical Manual of Mental Disorders (Martin, 1997). However, bias against lesbians and gay men continues to affect social work research, literature, and training (Cain, 1996; Mackelprang, Ray, & Hernandez-Peck, 1996; Messinger & Topal, 1997; Morrow, 1996). Not surprisingly, in a recent survey a majority of professional social workers reported receiving little education about lesbians or gay men (Berkman & Zinberg, 1997). Lack of knowledge such as this may contribute to social work practice that is vulnerable to heterosexist bias. One study (Wisniewsky & Toomey, 1987) found negative attitudes toward lesbians and gay men among a third of professional social work participants. A more recent study (Berkman & Zinberg) found such attitudes to be far less common, but it is not clear whether these findings represented a shift in social workers' actual attitudes or an artifact of methodological differences among studies. Research has an important role in educating social workers about lesbians and gay men and their human services needs. However, these issues received little attention in social work research until recently (Mallon, 1997). Studies of lesbians and gay men now constitute an emergent area of research, a trend that is likely to continue in fight of new requirements for inclusion of curriculum on lesbian and gay issues in Council on Social Work Education-accredited social work programs (CSWE, 1994). Because it is an emergent area, some researchers in social work might feel unprepared to enter it. This article aims to help social work researchers prepare for studies involving lesbians and gay men by raising their awareness of methodological and ethical issues in such studies. These issues involve theory and problem formulation, population definition, sampling and generalizability of findings, and preventing harm to study participants. THEORY AND PROBLEM FORMULATION Researchers must always scrutinize for possible bias the theories underpinning their research. Heterosexist bias is especially likely in studies of lesbians and gay men. Theories that assume or promote the superiority or normalcy of heterosexual sexual orientation and the inferiority or marginalization of other sexual orientations contain heterosexist bias (Hunter, Shannon, Knox, & Martin, 1998). For example, some theories assume that normal human development leads to a heterosexual sexual orientation. With the use of these theories, it is logical to conclude that a gay or lesbian sexual orientation in an adult is the outcome of an abnormal developmental process (Bieber et al., 1962; Erikson, 1980; Kronemeyer, 1980; Nicolosi, 1991; Socarides, 1978; Strean, 1996). When underpinning theories assume that gay and lesbian sexual orientations are pathological, research problems are likely to focus on the causes of the orientation and methods for changing it. Morin (1977) illustrated this point by noting that more than two-thirds of all studies of gay men and lesbians before the 1970s focused on sickness, diagnosis, and causation. Heterosexist bias can affect all stages of the research process. For instance, studies may ignore sexual orientation altogether, use identification of sexual orientation to devalue or stigmatize gay men and lesbians, or treat groups of gay men and lesbians as if they were homogeneous (Herek, Kimmel, Amaro, & Melton, 1991). Whereas heterosexist bias in underpinning theories might result in the reinforcement of prejudicial ideas, biases in a study's problem formulation can reinforce popular denial of systemic oppression against lesbians and gay men. In particular, researchers might reach conclusions that inadvertently blame lesbians and gay men for their own oppression as a result of focusing on person-centered variables in the problem formulation. …