I think what this study is trying to pick up is not a conscious level of racial hatred. I think the word racist needs to be replaced by subconscious racial preference. Similar to that of the Implicit Association Test (IAT) study discussed by Malcolm Gladwell in Blink. It’s not stupid to study these issues. The more we can attempt to (ethically) learn about human behavior the better.

just don’t agree with this study on racist referees. I think it’s a waste of time.
And I think the study could give a valuable insight into the ability of prejudices and personal emotion to cloud professional judgment; referees aren’t machines, nor are they all-seeing. Compound that with the fact that technical fouls, which can include greater subjective judgment than regular fouls, were included in the study and we get a better picture of the study’s thrust – when the decision of whether or not to call a foul is not a purely binary “this happened or it didn’t” situation, the study attempts to investigate the question of whether or not the referee’s own mentality could have influenced their call.

No one has said that “no one knows which referee actually blew the whistle”. In fact, the NBA seems to indicate that they have sufficiently granular data that they do, indeed, know who blew the whistle on each and every call of the 150,000 plays in their own study. They’re just not releasing that data outside of their own study.

Creating a database of referee calls from scratch is now very difficult or impossible. It would require: obtaining copies of the 1230 NBA games per year, presumably from the same source that is withholding the database; sitting through the games while tabulating (about 22 16-hour days, if the tabulating is immediate); and guess on the calls made offscreen; and then clean the errors. Not possible, but surely not a legal issue either, as the data is public, if not condensed. Although, the red tape of a corporate legal department can take a long time just to make that evaluation.

Second, and this one’s really short: If the granular data can be re-created via public information (game footage), then the granular data is already public information, and thus your argumentum ad djinn is seemingly moot. Especially in light of Dr. Bradbury’s suggestion:

Supply them with the NBA’s data with the stipulation that the economists not reveal anything more than their findings. This removes the privacy concerns. After all, the NBA was willing to share its data with the private firm it hired to do the study. The data stays secret and we get an independent evaluation of the conflict.

Second, they absolutely shouldn’t turn over the data unless they know what the result is going to be and are comfortable with that being aired in public. What if the independent researchers come back saying referees X, Y, and Z are racists? This is something that the NBA should find out for themselves but not necessarily in this setting – they can conduct the analysis during the offseason and quietly not renew those referees’ contracts.

Finally, the complaint about the NBA not sharing its data seems like so much whining to me. The authors could probably find game film of all those games and log all the foul calls themselves. Sure it would be much easier to use the NBA’s database, but that data could easily be recreated with some effort.

However, I would also like to point out that the paper looks at technical fouls. In my opinion technical fouls would be much more prone to emotion and bias than personal fouls. The coefficient on blackplayer*%white referees is not statistically significant in this case.

Also, why is the dependent variable fouls per 48 minutes instead of fouls per game? This seems to suggest that the dependent variable not give equal weight to those who come in to the game for only a few minutes and commit a foul. This shows up in the data as well:

Mean Fouls per game:

Black players — 2.5
White players — 2.5

Mean fouls per 48 minutes:

Black players — 4.3
White players — 4.97