Attempts to make hiring processes ‘blind’ to characteristics like gender or race are often intended to promote fairness. But new research published in Journal of Experimental Psychology: General suggests that hiding such information can reduce fairness instead of improving it. The study shows that people frequently avoid learning candidates' gender or race, despite knowing that ignoring this information will lead to more biased and less accurate hiring decisions. The paper reveals that fear of appearing biased often prevents decision-makers from using crucial bias-correcting information.

A blindfold that backfires

Organisations increasingly rely on ‘blindfolding’ - removing personal information from CVs or applications—to prevent discrimination and promote diversity. But the authors demonstrate a paradox: when evaluation tools themselves are biased (for example, when interview scores systematically favour men or White candidates), correcting the bias requires knowing the candidate’s gender or race, not concealing them.

Across 3 preregistered experiments (with 3,621 subjects), professional hiring managers decided whether to reveal personal information before selecting a candidate for a position. Even when explicitly told that gender or race unfairly influenced interview scores and informed on how to use gender and race to correct the bias, more than half of the managers chose not to reveal the information. They avoided it despite knowing that doing so meant making a less fair and less accurate decision. By contrast, when the same bias stemmed from a situational factor, such as the time of day of an interview, most participants revealed the information and corrected the bias.

Why avoid fairness?

So why would someone knowingly choose a worse candidate? The answer is concerns about their image. The key driver was not confusion or lack of understanding, but concern about how others would perceive them. Participants feared that revealing gender or race information, even for the purpose of addressing bias, would make them appear discriminatory. Behind this hesitation was not a moral objection, but a reputational one. ‘What surprised me the most was that decision-makers traded perceived fairness with real fairness and, while doing so, compromised accuracy,’ says Rabinovitch.

Importantly, job seekers themselves shared this perception: they perceived a hiring manager who reveals gender information as more biased than one who reveals situational information. This means that people’s fears about how they will be perceived are not irrational and align with social norms.

From insight to impact: a cautious path forward

Blindfolding can increase equality in some contexts, but can reinforce bias in others. This work challenges common assumptions about objectivity, diversity, and the design of fair evaluation systems. Rabinovitch adds: ‘Rather than assuming that less information equals less bias, we suggest a more nuanced truth: fairness sometimes requires seeing what we are inclined to ignore.’

Behavioural Ethics at UvA Economics and Business

Rabinovitch is a postdoctoral researcher at Shalvi's lab for Behavioural Ethics at the UvA Economics and Business. He is currently studying bias, exclusion, and discrimination, aiming at understanding the mechanisms leading people to engage in corrective behaviour.

Publication details

Research article: Rabinovitch, H. et al. Bias is not color blind: Ignoring gender and race leads to suboptimal selection decisions—A registered report, (2025), in Journal of Experimental Psychology: General