MLB Umpire Discrimination Research

As a senior at Claremont McKenna College, I authored a senior economics/data science thesis on observed race-based discrimination by Major League Baseball umpires.

You can read the full paper here.

Using pitch tracking data from 2008-2020, I underwent a statistical modeling expercise to predict the probability of an umpire making an error – either incorrectly calling a ball a stike (pitcher favoritism) or calling a strike a ball (batter favoritism). These probabilities were modeled as a function of whether or not the pitcher’s or batter’s race was the same as the umpire’s, and also included a host of other potentially explanatory reasons a mistake might be made: pitch location, game situation, count, pitcher/batter quality and handedness, etc. With over 3 million pitch observations, I was also able to conduct multiple sub-sample and time trend analyses to examine with whom the discrimination lies and how it changed throughout the sample.

The results suggest that umpires were significantly more likely to make calls that favor players of the same race, and that these effects did not diminished between 2008 and 2020. Furthermore, these biases seem mostly held by White umpires, who account for a wide majority of umpires in MLB.

The research won the Chairman’s Award (given by CMC’s econ department), generated significant buzz on social media and in sports media (Yahoo, Baseball Prospectus, etc.), and has been downloaded and read over 4,000 times.

I have been working to update the results to address some potential improvements – including using pitch type and pitch characteristics to isolate the effects for different pitches – and plan to share those updates here (as well as on Twitter) once I have them.