Link of the Day - Racial disparities in refereeing

This one has already showed up in the blogosphere (TrueHoop link), and a friend just emailed me the ESPN link:

Report: white refs call more fouls on black players

Two economists concluded that as the number of opposite-race officials increased, the number of fouls called on opposite-race players increased. Most significantly, they found the effect was bigger for white refs officiating black players than black refs officiating white players.

(Note: It's unclear from the article how Yao Ming figures into this analysis, since there are no Asian refs that I know of, and how some Latin American players are categorized.)

Of course, the NBA rapidly denounced this paper and said they had done an even better study that showed no bias. I'm not sure if I believe that. Methodologically, though, the NBA has a point:

The NBA strongly criticized the study, which was based on information from publicly available box scores, which show only the referees' names and contain no information about which official made a call.

That's an important point: more white referees ON AVERAGE may lead to more fouls on black players ON AVERAGE; but that doesn't necessarily imply that the white officials are calling the fouls. It certainly suggests that, though.

But the effects are actually quite small:

The key finding was in regard to foul calls, saying "black players receive around 0.12-0.20 more fouls per 48 minutes played [an increase of 2½-4½ percent] when the number of white referees officiating a game increases from zero to three."

That may be a statistically significant difference, but it's a very small effect size. 0.16 ± 0.04 fouls per 48 minutes breaks down to one extra foul per 8.5 games if you play 36 minutes a game. That's surely enough to piss coaches and players off, and definitely enough to suggest some racial discrimination.

But does it warrant this statement?

The lead researcher, Justin Wolfers, tells Schwarz that an examination of the numbers suggests that "if you spray-painted one of your starters white, you'd win a few more games."

Whoa. If you did that, you'd get 0.16 fewer fouls per 48 minutes assuming an all-white officiating crew compared to an all-black crew. I'm not sure if that translates into winning "a few more games." Sounds like maybe one close game per season could be decided by that.

Let's be clear: any disparity here is bad. There should be no difference based on referee race or player race.

I look forward to hearing more on this; it's an interesting topic and a great use of applied statistics.