Comments on: Implicit Bias in the NBA

By: A Guide to Evaluating Models « The Wages of Wins Journal

A Guide to Evaluating Models « The Wages of Wins Journal — Mon, 19 Nov 2007 05:30:51 +0000

[...] illustrate this point, consider the Price-Wolfers paper. This study found evidence of racial bias in how NBA referees called personal fouls. What was [...]

By: The 2007 Western Economic Association Meetings for Sports Economists « The Wages of Wins Journal

Sun, 08 Jul 2007 23:28:15 +0000

[...] most prolific writer in North America on sports economics), Joseph Price (co-author of the famous Price-Wolfers study which made headlines a few weeks ago), Dennis Coates (President of NAASE), Victor Matheson (a [...]

By: JChan

JChan — Thu, 03 May 2007 04:15:41 +0000

I’m surprised that some people want to just dismiss this report out of hand. From what I have read today and heard on the radio, the conclusions made in the report are almost just common sense.

The only thing I highly disagree with is the phrase “racial discrimination”. In my eyes, this phrase implies that the white referees are purposefully calling more fouls on the black players. And I don’t think that is the case.

This seems to me to be a case of people making split-second decisions and being ever so slightly influenced by their subconscious. I don’t think I am a racist, but I would be a fool if I didn’t admit that I see white and black people differently.

All of us are slightly biased just by being human. Everyone feels a bit more comfortable with people who are similar to themselves. Whether it is religion, race, economic status or whatever, we identify with those who are like us.

Really what this study shows is that referees are human. I have never heard of anyone saying a bad call was made because of racism, and I sincerely hope that no one uses this to begin accusing our current referees of racism on tough calls. Heaven knows their jobs are hard enough as it is.

By: Owen

Owen — Thu, 03 May 2007 00:51:38 +0000

Well, the study is getting killed by the player-commentators. Barkley went off on it at the start of the Spurs game, though he sounded unsophisticated, suggesting the study didnt control for the amount of time black players spent on the court. From what I can tell, black players, both past and present are unanimous in their rejection of the conclusions of the study. I dont know quite why they would reject it so vehemently, since I don think any of the players would suggest racism doesnt exist. Perhaps, they cherish the idea of the NBA as a meritocracy. Hardly surprising if so, since it has to be the most discrimination-free zone in the entire world for people of African descent.

The other thing i have noticed is the criticism that “all they did was look at the box score,” and “they didn’t even watch the games.” Like it or not, this may cast a shadow over box score approaches like the WOW. Or perhaps, it just reflects a popular distaste for the notion that box scores can capture what is going on out there.

By: MT

MT — Wed, 02 May 2007 21:14:46 +0000

I was puzzled by the extrapolation of the statistical data (which posits incremental fouls of 0.1 to 0.2 per player per 48 minutes being called based on race) to the proposition (which does not seem to be supported directly by the data) that the bias is large enough to cause a “predominantly black team” (to quote the authors) to lose a game refereed by an all-white crew. I had trouble getting from the first point to the second especially in today’s NBA. I guess you would have to assume the opposing team had several more white players, to get enough bias in its favor to generate an incremental number of fouls called, because nobody gets 0.1 foul shots in a game in the real world. And then you have to assume the incremental foul results in a made foul shot or converted turnover, and then there are assumptions about the talent of the teams being otherwise equal, playing time of the racially different members of the team, and so on. It seems like there are so many real world variables that can be held equal only in a hypothetical sense to generate the circumstances in which the extrapolation would be true. But if two teams playing each other had similar racial composition, which I imagine is pretty common in the NBA, the bias identified would not seem to affect the outcome at all.

If anyone wants an anecdote consistent with the authors’ point, watch the play where Chris Bosh picks up his fourth foul in last night’s Game 5 against the Nets, when the replay clearly showed it was Bargnani’s foul. But note the Raptors won anyway, which is consistent with my issue with the extrapolation. There was still a foul to be called.

By: anon

anon — Wed, 02 May 2007 21:14:30 +0000

I actually think the study is interesting, but here’s my problem. Let’s say we could somehow figure out ex post which foul calls were good or bad (video obviously helps to some extent, but assume there was a more objective method) and we aggregate all this data and did regression analysis and figured out how many games a year bad calls cost the average team. At what point would this figure matter? Obviously, anything over 5 would be cause for concern. But under five, while statistically significant, just might not be sufficient to cause concern. The reason is that we know that refereeing is an imperfect science and there are bound to be errors. The critical issue then is whether the errors are so large that the indicate something is seriously amiss. And if not, if they are statistically significant but don’t suggest anything sinister, can we really improve over baseline.

I think this study is interesting in the larger sociological debate over how race colors our private and professional lives. That said, I’m not sure there’s anyway the NBA could really improve over the baseline. Ultimately, empirical studies are only as good as the positive policy they encourage (otherwise, what’s the point. You might as well be a theoretician). And in this case, I can’t see any way the NBA can somehow internalize this information to improve inherent bias above the baseline.