Bias Is Not Color Blind: Ignoring Gender and Race Leads to Suboptimal Selection Decisions—A Registered Report

Blindfolding—selecting candidates based on objective selection tests while avoiding personal information about their race and gender—is commonly used to mitigate bias in selection. Selection tests, however, often benefit people of a certain race or gender. In such cases, selecting the best candidates requires incorporating, rather than ignoring, the biasing factor. We examined people’s preference for avoiding candidates’ race and gender, even when fully aware that these factors bias the selection test. We put forward a novel prediction suggesting that paradoxically, due to their fear of appearing partial, people would choose not to reveal race and gender information, even when doing so means making suboptimal decisions. Across three experiments (N = 3,621), hiring professionals (and laypeople) were tasked with selecting the best candidate for a position when they could reveal the candidate’s race and gender or avoid it. We further measured how fear for their social image corresponds with their decision, as well as how job applicants perceive such actions. The results supported our predictions, showing that more than 50% did not reveal gender and race information. By contrast, only 30% did not reveal situational biasing information—features of the situation rather than the individual—such as the time of day in which the selection test occurred. Those who did not reveal information expressed higher concerns for their social and self-image than those who decided to reveal. We conclude that decision-makers avoid personal biasing information to maintain a positive image, yet by doing so, they compromise fairness and accuracy alike.