|
Activity Number:
|
411
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, August 6, 2003 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Statistics in Sports
|
| Abstract - #301230 |
|
Title:
|
Ranking College Football Teams
|
|
Author(s):
|
Jerome P. Keating*+
|
|
Companies:
|
University of Texas, San Antonio
|
|
Address:
|
6900 N. Loop 1604 West, San Antonio, TX, 78249-1130,
|
|
Keywords:
|
shrinkage estimation ; truncated scores ; computer polls
|
|
Abstract:
|
David Harville (2000) established a set of seven conditions for reasonable comparison criteria for college football teams. In this paper, data-driven methods are used to choose a threshold for truncating scores to accommodate Harville's condition of "unobtrusiveness." In the process, a method is developed using linear models that adjusts performance based on the level of competition. The method can be shown to produce traditional nonparametric results when the threshold is reduced to unity. These new rankings can be compared to opinion polls, such as the AP poll or the coaches' poll, to determine whether voters are using a threshold and whether that threshold corresponds to statistical quantiles of the distribution of the score differential. The remarkable result is that polls are consistent with certain thresholds but the value changes by year for reasons that seem unrelated to the underlying distribution. Other variables such as rushing yardage gained or allowed could be used to adjust team yardage values based on the strength of the opposition. "Statistically adjusted" rushing yardage gained and allowed can be given that reflect the level of competition.
|
- The address information is for the authors that have a + after their name.
- Authors who are presenting talks have a * after their name.
Back to the full JSM 2003 program |
