Panel Paper: LGBT Employment Patterns in Public Organizations: Evidence From U.S. Federal Agencies

Saturday, November 9, 2013 : 8:00 AM
Plaza I (Ritz Carlton)

*Names in bold indicate Presenter

David Pitts, American University and Gregory Lewis, Georgia State University
A robust research literature examines workplace outcomes for women and people of color in public organizations, but a lack of data has prevented the development of similar research about lesbian, gay, bisexual, and transgender (LGBT) employees. We know very little about how LGBTs in the public service view their work, whether they are as satisfied with their jobs as heterosexual employees, or whether they are more or less likely to leave their jobs. These questions have important implications for human capital managers tasked to recruit and retain effective individuals/employees, and they are also especially relevant for LGBTs given their history of discrimination in public employment. This manuscript will endeavor to answer some of these empirical questions, focusing specifically on whether there are differences between heterosexual and LGBT employees in turnover intention, job satisfaction, and assessment of work-group relationships.

We will address these questions using data from the 2012 Federal Employee Viewpoint Survey (FEVS). The FEVS is the first federal government survey to include a question about sexual orientation. Although only 2% of respondents identified as LGBT, suggesting some degree of underreporting, a sample of 14,000 LGBTs is still far superior to existing data used for research on LGBTs in public employment. Research in this area typically relies upon Census-generated data in which one must make the assumption that respondents who live with an “unmarried partner” of the same sex are lesbian, gay, or bisexual. Even if this assumption holds, the results reflect only partnered LGBs and the sample is typically not very large. For example, the FEVS sample includes 65 times more LGBT federal employees than the 2010 American Community Survey, which is arguably the next best option.

In addition to providing a larger analysis sample, the FEVS data permit us to control for unobservable organization-level characteristics by including agency and subagency fixed effects, which will in turn strengthen the quality of our estimates. While the primary purpose of the manuscript will be to estimate differences between heterosexual and LGBT employees on the dimensions noted above, we will also examine whether there are interactive effects between sexual orientation and gender, race, and type of organization.