Saturday, March 6, 2010
Bias in officiating
By Kevin Arnovitz
In addition to the lively panels, the MIT Sloan Sports Analytics Conference features the presentation of research papers by up-and-coming thinkers in the field. Brian Robb of Celtics Hub attended one such presentation today, "Whistle Swallowing: Officiating & the Omission Bias." An omission bias can best be defined as a referee's willingness to make an incorrect call rather than make an incorrect non-call.
Robb explains the nut of the findings by Tobias Moskowitz and Jon Wertheim, who wrote the paper:
So where does the NBA fit into this type of bias? The research showed a couple much maligned problems in the league are as big of an issue as many fans of the league would have presumed.
The first is star treatment.
The study compared how likely officials were to call loose ball fouls on stars compared to non-star NBA players they were contesting in loose ball foul situations. The results were found over a three-year study in which 1.5 million plays were examined in 3,500 plus games. “Star” criteria was based on players' MVP votes. The results:
- 42 percent of loose balls fouls called on stars in “regular” situation compared to 57 percent of the time on non-stars in plays.
- The numbers show a much more dramatic shift, favoring the star players when they are in “foul” trouble with only 28 percent of foul calls being called on them, a huge drop from the earlier 42 percent.
- When the roles are reversed however, and the non-star is in foul trouble, the numbers normalize again with 48 percent of the fouls called on the non-star compared to 51 percent for the star.
The other study involving the NBA involved a look at subjective calls (offensive fouls, traveling, double dribble, etc.) being made compared to non-subjective calls (kick ball, 24-second violation, etc.) over the course of the game. The tendency to want to let the players decide the game in close as well as late game situations showed itself once again in the form of omission bias, with the rate of calls falling dramatically from the first half to the second half. Another even sharper drop in subjective calls was apparent in overtime games with the subjective or “judgment” calls. The non-subjective call rates remained very level over those time spans.