Sexual Orientation, Sexual History, and Inequality in the United States

ABSTRACT

Much of the literature on sexual orientation discrimination reports earnings differentials for gay, lesbian, and bisexual individuals when compared with heterosexuals. The General Social Survey (GSS) has previously been used due to its extensive coverage of demographic variables and sexual behavior in the United States. This study uses updated GSS data to investigate whether the income differentials found in earlier work have persisted and how estimates based on categorizing respondents according to the reported sex of their sex partners compare to estimates based on the respondents’ self-reported sexual orientation. Results for the years 2008–14 indicate that self-identification as an LGB individual and/or same-sex sexual behavior are correlated with a lower income; however, not all the results are statistically significant. In addition, there is a statistically significant negative income differential of 32 percent for men who report having had a same-sex partner at some point, but identify as straight/heterosexual.

KEYWORDS:

• Earnings differentials
• sexual orientation
• discrimination
• wage gap
• inequality
• labor market discrimination

JEL Codes:

• J31
• J71
• B54

ACKNOWLEDGMENTS

I would like to thank the many people who have contributed to this research, including Alexandra Bernasek, Cher Li, Sammy Zahran, M. V. Lee Badgett, Elissa Braunstein, Christopher Carpenter, Dan Black, and Feminist Economics’ associate editors and anonymous reviewers, for their assistance and thoughtful feedback.

Notes

1 This paper, as well as most of the previous literature, is restricted to studies of LGB individuals, leaving out transgender individuals. Authors face the issue that data on transgender individuals is almost nonexistent (Badgett 2006).

2 For a full review of the literature on income differentials based on sexual orientation, see Badgett (2006) and Klawitter (2015).

3 This paper derives from a thesis submitted to the Academic Faculty of Colorado State University in partial fulfillment of the requirements for the degree of Doctor of Philosophy.

4 Regarding the sexual history of those in the selfidLGB category: nearly all of them have had at least one same-sex partner. While some have also had one or more opposite-sex partners, further splitting the selfidLGB group would result in very small sample sizes. Additionally, due to heteronormative culture, it is generally expected that LGB individuals go through a sexuality questioning process where they may experiment with both sexes. This is generally not expected for heterosexual individuals. For more on heteronormativity and the questioning process, see Mindy Stombler et al. (2003).

5 The list of the twenty-one occupational categories and corresponding average incomes is available in Appendix Table A2.

6 I also adjust for inflation using the consumer price index (CPI).

7 A second commonly used method for interval type data is to use an interval regression. Results using interval regressions can be found in Appendix Tables A5 and A6. The interval regression results do not differ vastly from results using basic OLS (see Appendix Tables A3 and A4). In the main text, I use the conditional median method used in Badgett (1995), since this method also allows use of the Heckman correction.

8 Potential experience equals age minus years of education minus 5 years. This proxy must be used, as real experience is not available.

9 Although my sample includes full-time paid workers, “full time” may not be the standard 40 hours a week for each individual. Some work as many as 80 hours per week.

10 I utilize the variable occavg as described above. Using this method, I am able to control for a large number of occupational categories without sacrificing degrees of freedom (or having thirty or more regressors). A list of the twenty-one occupational categories may be found in Appendix Table A2.

11 Percentage difference is given by d=, where d is the percentage difference, and δ is the regression coefficient.

Additional information

Notes on contributors

Christina Curley

Christina Curley is a Lecturer at Ohio University. Her research interests center on social issues, discrimination, and public policy.