Are Gay Men and Lesbians Discriminated Against When Applying for Jobs? A Four-City, Internet-Based Field Experiment

Abstract

An Internet-based field experiment was conducted to examine potential hiring discrimination based on sexual orientation; specifically, the “first contact” between job applicants and employers was looked at. In response to Internet job postings on CareerBuilder.com®, more than 4,600 resumes were sent to employers in 4 U.S. cities: Philadelphia, Chicago, Dallas, and San Francisco. The resumes varied randomly with regard to gender, implied sexual orientation, and other characteristics. Two hypotheses were tested: first, that employers' response rates vary by the applicants' assumed sexuality; and second, that employers' Response Rates by Sexuality vary by city. Effects of city were controlled for to hold constant any variation in labor market conditions in the 4 cities. Based on employer responses to the applications, it was concluded that there is no evidence that gay men or lesbians are discriminated against in their first encounter with employers, and no significant variation across cities in these encounters was found. Implications of these results for the literature on hiring discrimination based on sexual orientation, the strengths and limitations of the research, and the potential for the Internet-based field experiment design in future studies of discrimination are discussed.

Keywords:

• discrimination
• employment
• field experiment
• hiring
• sexuality

Notes

We thank Krista D'Amelio, Jacqueline Liu, Siobhan O'Malley, and Nolan Sweeney, who provided invaluable assistance in gathering the data. We also thank Nancy Naples and Mary Bernstein for their input.

1. See Badgett, Lau, Sears, and Ho (2007) for a comprehensive review of studies of the sexual orientation earnings gap.

2. As one reviewer pointed out, an ideal experimental design would vary only the sexual orientation of paired applicants, leaving all other qualifications identical. The need to avoid detection and approximate plausible pairings of applicants, however, prevented us from giving pairs such identical qualifications.

3. See the bottom of Figure 1 for specific university affiliations associated with each template in the four cities.

4. To assure authenticity, the city names and street names in applicants' addresses were real. However, the house numbers were either too low or too high to correspond with actual addresses. In addition, we made sure that the four applicants in each target city came from neighborhoods (as designated by their zip codes) with similarly moderate median incomes so employers would not be unduly influenced by either low-status or high-status residence.

5. We ensured that the treatment organization appeared on the first page of the resume if it extended longer than a single page. Of course, it is possible that straight individuals could be members of gay or lesbian organizations, but we assumed that most potential employers would assume that people affiliated with such organizations would be gay or lesbian.

6. We used pay-per-minute TracFones®, which can be set up under any area code, and free Yahoo!® e-mail accounts. Our four TracFones used St. Louis area codes in order to not favor any of the four target cities.

7. In other words, responses to people of the same Gender × Sexuality combination (i.e., female lesbians) from all four target cities were directed to the same phone number. Because the TracFones® we used recorded the area codes of the callers' phone numbers, we were able to use this information, along with other identifying information (e.g., calls asking for “Amy” only originated from Dallas applications), to tell which target city the calls originated from.

8. It was not necessary for us to distinguish among the different applicants' names because these were randomized in the design.

9. Not all jobs that fit these criteria were equally suitable. We avoided jobs that required registration on an external Web site or long pre-application questionnaires due to time constraints and loss of efficiency. Instead, we concentrated on jobs that featured the “quick apply” feature by CareerBuilder.com®. This feature typically directed us to other “quick apply” jobs that fit the same criteria, which saved time and energy in the application process.

10. Weichselbaumer's (2003) study in Belgium was the next largest, involving 1,226 applications to 613 job openings.

11. Riach and Rich (2006b) found that almost 90% of racial discrimination detected in Belgium, the Netherlands, and Spain occurs at this initial stage of the hiring process. It is important to note their study did not involve Internet applications.

12. Kroft and Pope (2008) observed that CareerBuilder.com® charges employers to post job advertisements, whereas postings on Craigslist are generally free. In addition, although CareerBuilder.com is generally regarded as the industry leader, Craigslist job postings have surpassed CareerBuilder.com's in some metropolitan areas (Kroft & Pope, 2008). Especially given Craigslist's recent surge in popularity, it is possible that employers who use Craigslist have different attitudes toward gay and lesbian employees than employers who use CareerBuilder.com.