Racial bias, colorism, and overcorrection

2025-08-14

AI Summary

This study investigates whether heightened media coverage of racial bias and colorism triggers “overcorrection” in officiating decisions within the WNBA. Method: Employing a quasi-natural experiment, we combine machine learning–based prediction of player race and skin tone, quasi-random referee assignment, and high-dimensional fixed-effects regression to identify causal effects. Contribution/Results: Prior to media attention, referees exhibited no statistically significant racial or colorist bias. Following coverage, short-term systemic declines in foul rates across cross-race/skin-tone dyads emerged—indicating conscious overcorrection driven by heightened awareness. This effect decayed gradually over time and converged to zero bias. To our knowledge, this is the first empirical documentation in a real-world sports setting of time-varying overcorrection following DEI interventions, revealing unintended short-term consequences of awareness-raising initiatives. The findings underscore the importance of monitoring transient behavioral responses in organizational governance and provide novel evidence and methodological guidance for evaluating the phased impacts of diversity, equity, and inclusion policies.

Technology Category

Computer Vision: Bias, Fairness & PrivacyNatural Language Processing: Ethics — Bias, Fairness, Transparency & PrivacyReasoning under Uncertainty: Causality

Application Category

Social Networks and Social Media: Fairness and bias in social network and social media analysisUser Modeling, Personalization and Recommendation: Fairness-aware retrieval and rankingEconomics, Online Markets and Human Computation: Fairness, privacy, and diversity in economic environments

Abstract

This paper examines whether increased awareness can affect racial bias and colorism. We exploit a natural experiment from the widespread publicity of Price and Wolfers (2010), which intensified scrutiny of racial bias in men's basketball officiating. We investigate refereeing decisions in the Women's National Basketball Association (WNBA), an organization with a long-standing commitment to diversity, equity, and inclusion (DEI). We apply machine learning techniques to predict player race and to measure skin tone. Our empirical strategy exploits the quasi-random assignment of referees to games, combined with high-dimensional fixed effects, to estimate the relationship between referee-player racial and skin tone compositions and foul-calling behavior. We find no racial bias before the intense media coverage. However, we find evidence of overcorrection, whereby a player receives fewer fouls when facing more referees from the opposite race and skin tone. This overcorrection wears off over time, returning to zero-bias levels. We highlight the need to consider baseline levels of bias before applying any prescription with direct relevance to policymakers and organizations given the recent discourse on DEI.

Problem

• Examines awareness impact on racial bias in sports officiating
• Measures skin tone and race effects on foul calls
• Identifies overcorrection in referee decisions post-media scrutiny

Innovation

• Machine learning predicts player race and skin tone
• Quasi-random referee assignment with fixed effects
• Measures overcorrection in racial bias post-media