“Flawed Measurement of Hiring Discrimination against African Americans.” – Raj Ghoshal
Received July 26, 2019, Accepted for publication December 16, 2019, Published December 20, 2019
Abstract
This article addresses debates about race and hiring discrimination. Many experimental audit studies compare how equal-quality resumes with typically-white versus typically-black names are treated on the job market, often finding a difference in callback rates. However, one recent study (Darolia et al. 2016) argues that many employment audits may overstate discrimination because they use “black names” that are overly stereotypical. Pairing typically-white first names with typically-black last names to signal African American applicants’ race, they send out 9,000 resumes and find no discrimination in callbacks. Recognizing that some employers may not view the names they used, such as “Chloe Jackson,” as belonging to black applicants, they nonetheless argue that employers’ “error rate” in racial attribution would need to be “nearly 60 percent,” which they regard as improbably high, for prior audit findings to hold. Using a national survey of 1,050 Americans, I examine how people perceive the exact name combinations Darolia et al. used. I find that respondents’ overall error rate is indeed “nearly 60 percent” (59.7 percent), that this pattern holds for a subsample of respondents most likely to make hiring decisions, and that the pattern is unique to their approach rather than a problem for audit studies generally. I show that prior findings of persistent discrimination are not just unchallenged, but in fact bolstered by Darolia et al. ’s results and consider implications for future work.
