Labor market differentials estimated with researcher-inferred and self-identified sexual orientation

Abstract

The impact of the common practice of inferring sexual orientation via cohabitation status on estimated labor market differentials for sexual minorities is understudied. Using the 2013–2018 National Health Interview Survey, I show that inferring sexual orientation via cohabitation status leads to similar estimated differentials for gay men but inflates outcomes for lesbian women. Estimates for all bisexual individuals are biased upwards, because bisexual individuals are less likely to cohabit and comprise less than ten percent of the same-sex cohabiting sample. Estimates of outcomes for sexual minority members of same-sex households are largely unaffected by the sample contamination resulting from potentially erroneous inclusion of heterosexual individuals. However, cohabitation based researcher inference of sexual orientation masks important heterogeneity in self-identified sexual orientation based labor market differentials. Results highlight the need for inclusion of sexual orientation identity on more large scale surveys.