We apply state-of-the-art machine learning techniques to systematically predict race and objectively measure player skin tone to investigate refereeing decisions in the Women’s National Basketball Association (WNBA). 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. 

Funding / Support: Intramural Funding.

Principal Investigators: Kenneth Colombe, Alex Krumer, Rosa Lavelle-Hill, Tim Pawlowski

Publication:

• Racial bias, colorism, and overcorrection. Available at arXiv