Operational definitions of sexual orientation among adolescents: Consequences for estimates of behavioral health disparities

Identifying disparities among lesbian, gay, and bisexual populations (i.e., sexual minorities) hinges upon measurement of sexual orientation and its subsequent operationalization. Using Massachusetts YRBS data from 2003-2007 (N=10,277) we examined the effect of manipulating operational definitions of sexual behavior and identity on estimates of adolescent risk behaviors. Among respondents, 7.2% reported sexual minority behavior or identity; among sexually active adolescents who claimed a sexual identity, 12.3% had behavioral-identity discordance. Controlling for race, age, survey year, and stratifying by sex, we modeled smoking, suicidal planning, and methamphetamine use as functions of sexual orientation. These behaviors were chosen a priori due to previously documented disparities. Manipulations consisted of three behavioral and three identity operational definitions. The greatest disparities were associated with bisexual behavior and identity, which sometimes fully attenuated effects prior to disaggregation. For instance, compared to males with exclusively heterosexual behaviors, males with any same-sex behaviors had increased odds of smoking (OR: 2.36, 95% CI 1.66-3.35). However, when a distinction between exclusively same-sex (OR: 1.50, 95% CI 0.92-2.46) and bisexual behavior (OR: 4.24, 95% CI 2.55-7.06) was made, smoking was not associated with exclusively same-sex behaviors. Sexual inactivity was associated with decreased odds of all behaviors, being “not sure” of sexual identity was often associated with increased odds of behaviors. Based on estimates produced, behavior and identity are not comparable measures of sexual orientation. Results suggest a need for circumspect measurement and operationalization of sexual orientation in research, particularly among adolescents, lest researchers inaccurately attribute or fail to identify risk.