Racial Bias In NBA Refereeing?

I’ve blogged before about my friend Justin Wolfers’ research on point shaving in college basketball and the death penalty.

Now Justin is back stirring up more controversy with a paper that claims that there is racial bias on the part of NBA referees, written up in the New York Times by Allen Schwarz. The claim of the paper is that Black officials call more fouls on White players and vice-versa.

I’ve read through the paper and could not find any mistakes. It appears to be carefully done. It mirrors findings John Donohue and I have in a different context. We found that when police departments hire more White officers, arrests of minorities rise, but not arrests of Whites. The opposite is true when more minority officers are hired.

One other thing worth noting. I have never seen a reporter (other than Dubner, of course) take more care to try to get the facts and interpretation right than Allen Schwarz did on this piece. He was in contact with me a number of times and even arranged to have the NBA’s counter-analysis evaluated by independent experts. Hats off to him for his reporting.


Chris Mealy

Levitt is such a softie I took his praise with a grain of salt. But it's completely deserved. I was startled by the paper's depth and care. I've never seen a paper test so many hypotheses, and not obvious ones either.

This paragraph seems to be the big payoff:

There are also two ways in which these own-race biases may emerge: they may reflect referees favoring players of their own race, or alternatively disfavoring those of the opposite race. The arbitrary assignment of referees to games means that we can test whether our estimates reflect an influence of referee race on black players, or on white players. Table 3 is instructive, showing that the rate at which fouls are earned by black players is largely invariant to the racial composition of the refereeing crew. By contrast the rate at which fouls are earned by white players responds quite strongly to referee race. Further regression-based tests yield a similar pattern (see in particular the coefficient on %white referees in Table 4), suggesting that the impact of the biases we document is on white players, who are either favored by white referees, or disfavored by black referees.

Read more...

cisenburg

Let us not forget that causation is independent of correlation.
Is it possible that white players commit more fouls when the referees are black? It seems unlikely, but this blog consistently discusses the unlikely.

chappy8

I guess I wonder a couple of things:

1/ This study doesn't actually have data on WHICH referee actually called the foul, just the racial compostion of the crew, so we have no idea how this is distributed.

2/ I couldn't figure out if 'end-of-game' minutes or 'blow-outs' are really controlled for. Anyone that watches a lot of basketball knows that there is a flurry of fouls at the end of games. Not sure if there might be substitution patterns that could affect this. Of course a game that is a late blow-out might witness the opposite affect. Just wondering how these results might change if end of game minutes (in a 'blow out' versus 'close' game) might change things.

RutgersPainTrain

I haven't read the paper but the NY Times article seemed to imply that the researchers counted the composition of the 3 refs instead of counting which of the refs (white or black) were blowing the whistle each time. Is it possible that when there is only 1 black ref, he feels he needs to be tougher in the eyes of the two white refs and therefore calls more fouls on black players?

waterbird

I was one of the readers who asked you guys to comment on this. I got 2 email responses and 2 blog entries as a reward! The question about end-of-game "stalling" fouls is an interesting one, since they are typically intentional and thus would seem to throw off the count. I'll check out the paper. It's nice to know, at least, that it was conducted thoroughly. It's too bad what it says about us as humans, though.

karl-smith

Is it possible that white players commit more fouls when the referees are black?

So yeah, interaction effects are important. Suppose that white players feel more pressure to abide by the rules when they are being judged by white refs.

I haven't read the paper yet and there may be enough controls to ensure that any interaction effects are racially based but thats a little different than racial bias.

cica

I have a question. What basis is there to label this sort of bias "taste-based" rather than information-based, aside from the fact Becker once classified racial bias as generally falling into the that category? In Levitt's study of The Weakest Link, he was able to distinguish between the discrimination faced by the old (taste-based) and by Latinos (information-based) because of how the winning strategy changes midway into the game--otherwise, both kinds of discrimination would have been indistinguishable from the data, and most people would have believed that the anti-Latino discrimination was also taste-based (out of deference to Becker). But it seems that the racial bias in NBA refereeing could be explained in either theories...

peter123

This is one of those examples where yes, they found a statistical significance, but not a practical one: As John Hollinger pointed out, Lebron James played 3,190 minutes this season, and racial bias would account for a whopping...11 extra fouls.

Fun idea for a study, but not an exciting result.

mtbube

What if the bias is due less to race but more with appreciation of a certain style of play? Maybe referees simply don't see the violations because of the way that they play themselves or think the game should be played. Were I a referee, I would tend to see certain fouls as a necessary component of the game (and therefore not fouls) based on my belief on how the game should be played. Ultimately, does this bias affect the outcome of the game? If the racial composition of each team is the same, then the bias by the referees, whether in favor of whites or blacks, would be null on the outcome of the game.

wintercow20

Isn't Joe Price a co-author? Would be nice to give him credit too.

Police

Clydes a knick hater. Compare his Enthusiasm and his replays to that of any other Knick announcer.
I remember during a game that maubury and Telfair was in that Clyde commented that Maubury was being "schooled" by his cousin. Very harsh words for someone he endorses now. Clyde's a knick basher. a BOOOO generator.

Let's go Knicks

Keith S

I saw the same phenomenon watching CNN anchors Wolf Blitzer and Lou Dobbs on CNN over supper.

In a sense, they "referee" election campaigns.

I could definitely see them pushing for the current underdog, McCain.

Michael Jordan Shoes

This is one of those examples where yes, they found a statistical significance, but not a practical one: As John Hollinger pointed out, Lebron James played 3,190 minutes this season, and racial bias would account for a whopping…11 extra fouls.

Chris Mealy

Levitt is such a softie I took his praise with a grain of salt. But it's completely deserved. I was startled by the paper's depth and care. I've never seen a paper test so many hypotheses, and not obvious ones either.

This paragraph seems to be the big payoff:

There are also two ways in which these own-race biases may emerge: they may reflect referees favoring players of their own race, or alternatively disfavoring those of the opposite race. The arbitrary assignment of referees to games means that we can test whether our estimates reflect an influence of referee race on black players, or on white players. Table 3 is instructive, showing that the rate at which fouls are earned by black players is largely invariant to the racial composition of the refereeing crew. By contrast the rate at which fouls are earned by white players responds quite strongly to referee race. Further regression-based tests yield a similar pattern (see in particular the coefficient on %white referees in Table 4), suggesting that the impact of the biases we document is on white players, who are either favored by white referees, or disfavored by black referees.

Read more...

cisenburg

Let us not forget that causation is independent of correlation.
Is it possible that white players commit more fouls when the referees are black? It seems unlikely, but this blog consistently discusses the unlikely.

chappy8

I guess I wonder a couple of things:

1/ This study doesn't actually have data on WHICH referee actually called the foul, just the racial compostion of the crew, so we have no idea how this is distributed.

2/ I couldn't figure out if 'end-of-game' minutes or 'blow-outs' are really controlled for. Anyone that watches a lot of basketball knows that there is a flurry of fouls at the end of games. Not sure if there might be substitution patterns that could affect this. Of course a game that is a late blow-out might witness the opposite affect. Just wondering how these results might change if end of game minutes (in a 'blow out' versus 'close' game) might change things.

RutgersPainTrain

I haven't read the paper but the NY Times article seemed to imply that the researchers counted the composition of the 3 refs instead of counting which of the refs (white or black) were blowing the whistle each time. Is it possible that when there is only 1 black ref, he feels he needs to be tougher in the eyes of the two white refs and therefore calls more fouls on black players?

waterbird

I was one of the readers who asked you guys to comment on this. I got 2 email responses and 2 blog entries as a reward! The question about end-of-game "stalling" fouls is an interesting one, since they are typically intentional and thus would seem to throw off the count. I'll check out the paper. It's nice to know, at least, that it was conducted thoroughly. It's too bad what it says about us as humans, though.

karl-smith

Is it possible that white players commit more fouls when the referees are black?

So yeah, interaction effects are important. Suppose that white players feel more pressure to abide by the rules when they are being judged by white refs.

I haven't read the paper yet and there may be enough controls to ensure that any interaction effects are racially based but thats a little different than racial bias.

cica

I have a question. What basis is there to label this sort of bias "taste-based" rather than information-based, aside from the fact Becker once classified racial bias as generally falling into the that category? In Levitt's study of The Weakest Link, he was able to distinguish between the discrimination faced by the old (taste-based) and by Latinos (information-based) because of how the winning strategy changes midway into the game--otherwise, both kinds of discrimination would have been indistinguishable from the data, and most people would have believed that the anti-Latino discrimination was also taste-based (out of deference to Becker). But it seems that the racial bias in NBA refereeing could be explained in either theories...