Another Look at Clutch Hitting, With a Note On the Bias of the Sign Test for Differences in Proportions

Abstract

“Clutch hitting” in baseball is when a batter performs well in high-stress (“clutch”) situations. We compared the clutch and non-clutch batting averages, for several definitions of clutch, for every player with a minimum number of clutch batting averages in any of the seasons 2000-2007. There does appear to be a league-wide difference between clutch and non-clutch averages, even accounting for other situational variables. However, we found no evidence that certain players are better at clutch than non-clutch hitting year after year. One seemingly natural way to examine this latter hypothesis is to look at sign (clutch – non-clutch) for each player for each year, since this adjusts automatically for differing numbers of at-bats, and to tabulate the distributions of signs across players. However, that sign’s expected value is not zero, and in fact the expected value displays some surprising properties across sample sizes and “true” batting averages.

Bio

Samuel E. Buttrey received a bachelor's degree in statistics from Princeton University and Masters and Doctoral degrees from the University of California, Berkeley, the latter in 1996. In 1996 he joined the faculty of the Department of Operations Research at the Naval Postgraduate School in Monterey, California, where he is an Associate Professor. His interests include data analysis, classification, statistical computing, and data mining.

Unless otherwise stated, the content of this page is licensed under Creative Commons Attribution-ShareAlike 3.0 License