A reader sends along a study, published in The American Economic Review on racial bias by home plate umpires in calling taken pitches. The authors dig very deep into the data, and find a small but consistent difference in calling pitches that favors the pitcher when he is the same race as the umpire. They also find that the difference tends to disappear in games with a computer monitoring system in place, large crowds, or on terminal pitches. Since PITCHf/x monitors all pitches in all games today, I would suspect the bias has been eliminated.
Two sections I would like you to note as you read the study. Table two shows the raw data. Note that white pitchers get the most called strikes, and black pitchers the least, no matter who is umpiring. Section four uses PITCHf/x data to show that the racial bias effect lies entirely on the edges of the strike zone, those pitches that can go either way.
The one thing that raises a red flag for me is the lack of racial bias where the batter is concerned. One test they did not conduct is what happens when the batter and umpire match race but the pitcher and umpire do not. If the bias shifted toward the batter in those situations, I’d tend to trust the whole study much more.
On the other hand, I’m encouraged that even without monitoring, you only see bias early in the count in low attended games, and even then the bias is extremely small.
I’m curious to see Tom Tango’s take on this, since he’s an expert on linear regressions.
If PITCHf/x is showing bias, then PITCHf/x hasn’t eliminated it, right?
Slideshow Bob » Good point.