Findings of a Paired-Testing Pilot Study of Discrimination Against Same-Sex Couples and Transgender Individuals in the Rental Housing Market
Housing Discrimination against Same-Sex Couples and Transgender Individuals is a HUD-commissioned pilot study that used paired testing to determine the extent of discrimination in the rental housing market based on sexual orientation and gender identity. Although sexual orientation and gender identity are not explicitly named as federally protected classes in the Fair Housing Act, the act has been used to protect lesbian, gay, bisexual, and transgender individuals against certain discriminatory acts. Regulatory action has also extended some federal rights to these individuals, and many state and local ordinances bar discrimination based on sexual orientation and gender identity.
This first-of-its-kind study had three goals: first, conduct pilot tests to estimate rental discrimination against gay and lesbian couples compared with heterosexual couples, and compared with heterosexual individuals; second, develop and pilot paired-testing protocols to estimate discrimination against transgender individuals; third, compare the utility of remote testing versus in-person testing for same sex couples and transgender individuals.
Study Design and Implementation
The overall sampling design involved 2,000 paired tests, of which 1,800 were same sex and 200 were transgender. Researchers considered state- and local-level protective laws and racial and ethnic diversity in the 10 largest metropolitan statistical areas (MSAs) when choosing sites. Sites required sufficient populations of minority (including African-American, Hispanic, and Asian populations) and white individuals so that researchers could conduct paired testing separately for the two groups. Based on these considerations, researchers targeted 1,800 same-sex tests for the Dallas-Fort Worth and Los Angeles MSAs and 200 paired transgender tests for the Washington, DC MSA.
Of the 1,800 same-sex paired tests, 1,200 were conducted in person and 600 were conducted remotely (by telephone or e-mail). Both the in-person and remote tests were split evenly between paired tests with lesbian and gay male couples. In-person tests were further allocated equally to white and minority testers. Minority tests were paired tests with a mix of African-American, Hispanic, and Asian participants. The 600 remote tests were also split relatively evenly between white and minority testers, although the minority group did not include Asian testers due to lack of availability.
In developing protocols for the same-sex paired tests, researchers decided that no tester would be assigned a single-person profile; instead, the testers were assigned a same-sex relationship that served as a proxy for sexual orientation. Testers used a mix of terms to disclose their sexual orientation to housing providers in as natural a way as possible before discussing available rental units. In half the tests, individuals posed as being married and referenced their spouse as husband or wife, as appropriate. For the other half, individuals posed as unmarried but referenced their significant other as girlfriend, boyfriend, or partner.
Testers for the 200 paired transgender tests were diverse in gender identity and included those who identified as transgender women, transgender men, and gender nonconforming or genderqueer. Focal and control testers included minority (African-American, Hispanic, and Asian testers) as well as white testers. In the transgender study, researchers sought to develop test protocols aimed at conveying transgender status without betraying the test. They also examined whether tests in which transgender status was not disclosed could capture treatment based on gender status. As a result, the 200 paired transgender tests were equally divided according to two testing approaches. In the first, transgender testers verbally disclosed their gender status, although they were not required to express their gender identity in any particular manner. In the second, they did not disclose their gender identity.
Researchers performed regression analysis to determine if treatment of the focal testers varied with the characteristics of the test, testers, rental agents, and MSA. Researchers specifically tested for the following:
- Whether lesbian and gay male testers experienced differential treatment compared with the control testers and whether treatment varied by race and by the term used to describe significant others.
- Whether variations in differential treatment existed between the Dallas-Fort Worth and Los Angeles MSAs.
- Whether the differential treatment of lesbian and gay male testers varied by method of contact (in person versus remote).
- Whether transgender testers experienced differential treatment compared with the control testers and whether treatment varied by race.
- Whether the choice to disclose gender identity affected the differential treatment of transgender testers.
- How the characteristics of testers, rental agents, buildings, and neighborhoods influence patterns of differential treatment.
Researchers found that housing providers discriminated against gay men and transgender individuals on some treatment measures. Compared with heterosexual men, gay men were told about one fewer available rental unit for every 4.2 tests and were quoted average annual rental costs that were $272 higher. Providers told transgender testers about fewer units than they told cisgender testers, although providers quoted the same rent and average net yearly costs for both groups. Researchers found that differences in results by protocol were small and inconsistent across variables. As a result, they did not recommend one approach over the other.
Housing providers treated lesbians and heterosexual women similarly. Providers were equally likely to schedule an appointment, tell about and show the same number of units, and quote the same rent and incentives for both groups of testers. Although the researchers found small differences across treatment measures of availability and inspections that placed lesbians at a consistent disadvantage, these differences were not large or statistically significant.
The researchers found that the treatment of lesbian and gay couples varied little by race or city, results that they characterized as “unexpected.” The researchers suggested, however, that the results might not be accurate due to data shortcomings. The researchers found little difference in outcomes between the in-person and remote testing approaches. Nevertheless, they hesitated to recommend relying on remote testing methods, because in-person testing ensures that testers interact with the staffers who make rental decisions.
Conclusions, Limitations, and Further Research
Compared with earlier studies of housing discrimination based on sexual orientation and gender identity, the extent of discrimination suggested by the current study is more modest. However, as the researchers noted, the results are not directly comparable due to various factors, including study scale, choice of study sites, and methodological differences. Study outcomes may have been skewed by restricting tests to only three metropolitan areas that had ordinances barring discrimination based on sexual orientation and gender identity. Conducting tests in a larger and more diverse set of metropolitan areas might have produced different results. The researchers also noted that because the number of transgender tests was small, the study cannot support an analysis of differences in treatment of transgender women, transgender men, and genderqueer individuals. Moreover, because of the general methodology employed in paired testing for the rental market, this study’s findings are limited to only the initial stages of inquiry about rental units and to individuals who are financially qualified for the requested housing.
Because of the limitations of the present study as well as the number of questions that it raised, the researchers suggested several areas that merited further inquiry and activity. Among them is testing with a larger sample of sites and with sites that vary by region and population size. Another is additional research into best practices for paired-testing studies of housing discrimination based on gender identity. A third is designing study protocols in which testers pose as single homosexual individuals, permitting a direct test for sexual orientation discrimination rather than testing by proxy. A final area is greater knowledge sharing among researchers studying housing discrimination based on gender identity and sexual orientation.