Monday, May 08, 2006

The Subjectivity of NBA Referees


I've watched a lot of NBA games in my life, and I'm always alarmed at the calls made by referees that seem to favor the home team. I think I really started to notice it a few years ago in the Western Conference Finals between the LA Lakers and Sacramento Kings--I think the Kings would have won the series if the officiating had been fairer. Complaints with NBA referees go far further back than this series, and have a rich history. It has often been argued that Jordan "got all the calls" because of his accomplishments. I want to test if the home team does, in fact, get more favorable calls.

What's more, I've always believed that these calls occur during pivitol times of the game. NBA referees are social creatures like anyone else, and I believe that they are influenced by the crowd and the time of the game. I hypothesize that referees in better attended games make more foul calls than usual, and make the calls when the game is close.

Data/Empirical Strategy

My data will come from a 2005-2006 data set made by, a privately-run NBA statistics website. My data will include fouls by quarter, points by quarter, and officials on a game level. I will use STATA code to transform this dataset to a team level, and analyze it as follows.

I want to run a regression that looks at the foul gap between the home and away teams. Specifically, I will regress a "home/away" variable while controlling for day, opponent fixed effects, and team fixed effects. This will show whether the home team does, in fact, receive more calls.

I also want to regress point differential, both by quarter and game, on this foul gap statistic. I think that the lower the point differential, the higher this foul gap will be--in favor of the home team. Conversely, the higher the point differential, the lower the foul gap.

Finally, I want to regress attendance and referees on this foul gap variable, to see the effect of attendance on foul calls and whether certain officials are especially prone to making calls in big games.


There are two problems I forsee in working with my data set. First, I don't know how to use the officials in the regression. Currently, my data has the names of officials in each game. However, there are rotating groups of 3 officials per NBA game that I don't know how to code for in using in a regression.

Secondly, often there are lots of fouls late in the game by teams trying to catch up by stopping the clock and making the other team shoot foul shots. These fouls could distort my results, and I'm not sure how to control for them.

Will, I have actually always wondered this myself but at the same time realized that there are so many things affecting the calls of the officials that it is very difficult to isolate this effect. First off, I think your second worry, that there are a lot of fouls late in the game as teams use a fouling strategy to catch up is a major problem in your second hypothesis. The reason there are more fouls in close games is due to the implementation of a foul strategy that would not occur in a game that was not close. Also, these fouling strategies have a tendency to draw the end score of the games closer. Secondly, another important thing to consider is the advantage of being the home team. One of the key advantages of being the hometeam is obviously the crowd. Basketball is a game of momementum. Teams consistently go on scoring runs that define the course of the game. As a home team, I would hypothesize that these scoring runs are more likely due to occur due to crowd influence. Also, scoring runs are usually correlated with a fast break offense. A fast break offense is also likely to cause more foul calls to go against the away team then a half court offense. Therefore, the correlation in being the away team and a higher number of foul calls could be a crowd effect and not the effect of the officials. Another argument, and this ones actually kind of out there, is based on the fact the away team wears colored uniforms while the home team usually wears white. The effect in officiating could be caused by the fact that officials see fouls commited by brighter colored players. Its probably a combination of all of these factors plus an officiating bias that will show in your result. Isolating the effect of officiating will require controlling for all these factors somehow.
Varun's uniform hypothesis is actually really interesting. I wonder if that would be true for something like the Lakers whose uniforms are not white but yellow, a pretty noticeable and catchy color.

One thing I just wanted to note was whether or not the official track records differentiated personal fouls and technical fouls. Some instances of unsportsman-like behavior or non-contact fouls may have more to do with the player and coach tempers than with the official (though an easily-annoyed official may call technicals more often). This may exaggerate the number of fouls for close games when one particular team gets more riled up than others.
I'm not sure how the issue of fouling strategy at the end of games would affect your regressions unless we can be sure that a home team is more likely to be ahead at the end of games. Because you're looking at close games vs. blow outs and examining the difference in the foul calls for the home team and the away tea, the total number of fouls shouldn't affect your results if we can assume that the probability that a home team or an away team is up near the end of the game is even. I would believe that the home team wins more, but I would also think that it's faily close to even.
Like Varun mentioned, the problem with your second regression (how close the game is and fouls) is essentially simultaneity. Although close games may lead to more foul calls, more foul calls may also cause closer games. Also, I'm not sure if you mentioned this, and it's fairly obvious, but you should probably control for the official as well, right?
I agree with varun on his point that the runs that teams go on have a lot to do with the crowd because of the emotions and adrenaline that it produces however i think it is also possible that the crowd my have an affect on the refs as well. Refs are human too and due too the fact that basketball is a very fast paced game in which there are many calls that have to be made in a slit second and depend on variables that are very hard to differentiate, such as the angle the ref is looking at the foul or which player initiated the contact. So in situations where the ref may not be totally certain of the call they may have a tendency to side with the thousands of people yelling at him.
Post a Comment

Subscribe to Post Comments [Atom]

<< Home

This page is powered by Blogger. Isn't yours?

Subscribe to Posts [Atom]