Abstract
A contest model is constructed to examine the existence of conference bias in college basketball's Ratings Percentage Index (RPI). Though a general RPI bias has been identified in previous literature, this is the first study to address whether the bias is random or systematic in nature. Within the theoretical model, the RPI is shown to be systematically biased against teams in high ability conferences, even when all teams play to expectation and can be transitively compared. Further, the bias can prevent the RPI from producing an ordinal mapping from revealed team ability level to the real number line. Given the longevity of the controversial RPI as the NCAA's primary measure of team ability, these results may indicate that the NCAA is serving a demand for team heterogeneity in selecting for the NCAA Men's Basketball Tournament.
Original language | English (US) |
---|---|
Journal | Economics Bulletin |
Volume | 4 |
Issue number | 34 |
State | Published - Sep 12 2007 |
Externally published | Yes |
ASJC Scopus subject areas
- General Economics, Econometrics and Finance