Batting Averages, Weighted Averages, and Simpson's Paradox
Batting Averages, Weighted Averages, and Simpson's Paradox
Simpson's paradox refers to the reversal of the direction of an association when a categorical variable is ignored. Although Simpson's paradox can and does occur in such important contexts as death penalty rates and gender equity cases, it is illustrated here in the familiar setting of baseball. It is possible for two baseball players to have a season in which player 1 has a higher batting average than player 2 in the first half of the season and also in the second half of the season, yet player 2 has a higher batting average than player 1 for the entire season. Such a possibility exists because a player's overall batting average for the season is a weighted average of his two half-season averages, weighted by the proportion of at-bats (AB) in each half of the season.