Thursday, May 03, 2007

Biased towards Bias - What's with those NBA Refs?

The big non-playoff news in the NBA yesterday was the announcement that there may be a racial bias in how NBA officials referee games. Naturally, the minute that the words “race,” “bias,” and “sports” hit the front page of the New York Times, the sports media collectively had an orgasm at the thought of a juicy racial scandal to discuss. In short, the Times’ piece describes an upcoming study by a University of Pennsylvania professor which points to a racial disparity in fouls called during NBA games.

Oftentimes, the root of such potential controversies are sports columnists looking to stir the pot, but this story emerged - and was promptly leapt on - because of a certain authenticity associated with an academic conclusion (as opposed to some blabbering idiot’s opinion). Here’s the problem: while an academic study isn't the equivalent of a yammering columnist blindly pushing a race issue, it still seems like the media is flouting a finding for the sake of scandal-mongering. The main issue is that the media is running with the same headline and synopsis of Wolfers' study that the New York Times presented… but why isn't anyone delving in to the actual data? The immediate reaction has either been to embrace the study or point out superficial criticisms that don't apply when reading the actual study (i.e. assuming that the study is stating the obvious because there are more black players in the league when, in fact, that is accounted for).

I don't claim to be an economist/statistician – in fact, I left academia a year ago to avoid dealing with these kinds of studies in the first place - but I wanted to at least TRY and understand the findings. Why simply take a summary at face value? The first step to interpreting a finding is determining what went into the study in the first place, and it seems like most of the media threw an academic halo around the results without digging deeper.

Admittedly, I can't parse out all of the statistical data and manipulation because some of it is a bit dense and over my head, but there are some fundamental criticisms I took away from the study:

  1. This isn't a criticism of the paper itself, but it's important to remember that this has NOT been published or peer reviewed. Warrants mentioning, because the people most qualified to interpret an academic paper are peers in the field. Personally, I’d like the counterpoints to be available before an orgy of media attention.
  1. The study uses box scores to collect their referee/foul/player data. This means that individual calls are impossible to discern, so unless the officiating team was comprised of one ethnicity, it's hard to tell who's actually making the call. So while a scenario with an all-white or all-black team of refs is possible, it certainly isn’t the most prevalent combination. The box-score data compilation also removes ALL context of fouls, intentional or otherwise.
  1. Most importantly, the heart of the data (which seems to then be extended and extrapolated into more detailed observations with more complicated analysis) does not seem to support the synopsis given by the Times.

Here’s how the finding is summarized by the Times:

"A coming paper […] says that, during the 13 seasons from 1991 through 2004, white referees called fouls at a greater rate against black players than against white players."

But let’s take a look at claim a bit closer, shall we? Here’s a PDF of the study manuscript itself.

What caught my attention was the information presented in Table 3 (Page 36 of the Study). My impression is that most of the study’s farther reaching analyses are based on this initial presentation of foul rates. Simply put, the table shows the difference in rates of fouls called on white and black players by white and black referees. The term foul rate is defined as a player's fouls earned per minute times 48 (48*Fouls/Minutes Played).

Here’s what’s striking about the data in this table. First, there is NO difference in the foul rates for black players between black and white refs (4.33 vs 4.329). There IS a difference in the foul rate called against white players (4.954 called by White Refs, 5.023 by black refs). The second, more detailed part of Table 3 (which measures these changes by the percentage of white or black refs on an officiating team) shows the same finding.

This leads to one of two possible conclusions: either white refs favor white players by calling fouls less frequently, or black refs call more fouls on white players. It is IMPOSSIBLE to determine which one is actually at play.

In the authors' own words (PDF page 30):

"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[…] 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 […] 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."

This is a far cry from what the New York Times article presented, but people ran with that interpretation for the story’s sake. In other words, what we may have on our hands is ridiculous case of rabble rousing, which is especially infuriating because it's veiled by a pretend shield of statistics. At this point several people have come out to criticize the study (PTI and ESPN.com's John Hollinger come to mind), but it almost seems like a fundamental piece of information isn't being given the light of day.

I need to reiterate that I am NOT a statistician, so my ultimate goal would be to see an explanation by someone more qualified than I. Nevertheless, I’m simply trying to point out that the very root of this “scandal” shouldn’t have to do with whether white refs call more fouls on black players than on white players; it’s how they call these fouls compared to black refs. This is the logical apples-to-apples comparison, a common test for one group of referees compared to another. If the study’s most fundamental data shows that there is an ambiguous discrepancy between officials calling fouls (i.e. are white refs under-calling fouls on white players or are black refs over-calling them?), why aren’t we discussing THAT finding instead of immediately pointing a finger of blame?

Now that question I don’t have an answer for.

read more | digg story

2 comments:

Christopher said...

According to this piece by Lester Munson, the study did analyze all-white and all-black crews and the result reaffirmed their findings. Also, Wolfers did present the paper at a U-Chicago economics panel and it has been accepted for publication by a reputable journal. So there has been some measure of peer review applied to the study. I do think you're right that the headlines don't match the content of the paper, but when has journalism ever refused to sensationalize the results of academic inquiry?

ChiTownDan said...

Alright, just a quick thought on those number: (I also have not read/analyzed the study or am not a statistician by any means)

If the 'Rate = 48*Fouls/Minutes', then conversly, 'Fouls = Rate * Minutes / 48'

Can we assume a superstar plays...no more then about 40 minutes for the season? 40 minutes a night for 82 games is 3280 minutes. Plugging that into the formula and using the two different rates for white players, you get either 343.24 fouls per season, or 338.523. Were talking about less then 5 fouls over an entire season? 1.3%?

Seems to me that both rate differences are a little bit irrelevant.