When ‘blind’ selection works—and when it doesn't

Efforts to reduce bias in selection often rely on blindfolding—concealing job-irrelevant factors such as race or gender. While this approach can reduce bias, it may also obscure and preserve it when selection tools are themselves biased. This paper introduces a novel framework showing when revealing sensitive information enables more accurate and fair decisions. By leveraging hidden correlations between potentially biasing factors and test scores, decision-makers can correct for bias and improve selection decisions. The framework offers guidance for policymakers and institutions. It also highlights resistance to revealing such information, rooted in the desire to preserve a positive social image. This tension between perceived and actual fairness helps explain reluctance to adjust biased tools.