ANALYISIS OF MULTIPLE COMPARISON ERRORS USING GROUND REACTION FORCE DATA

Abstract

This technical note examined the effect of uncorrected multiple statistical comparisons that are often inappropriately reported in the applied biomechanics literature. Bilateral ground reaction forces (1000 Hz) were collected for a single subject performing twenty vertical jumps to create multiple comparisons of discrete kinetic variables. Nine of twenty comparisons were statistically significant using multiple t tests, while only seven were truly statistically significant when controlling for inflation of type I error. Most of the differences observed were not of practical significance to performance. The data confirmed the danger of inflation of type I errors in biomechanics and not interpreting the results based on meaningful effects or previous literature.