A case against default effect sizes in sport and exercise science

Abstract

Recent discussions in the sport and exercise science community have focused on the appropriate use and reporting of effect sizes. Sport and exercise scientists often analyze repeated-measures data, from which mean differences are reported. To aid the interpretation of these data, standardized mean differences (SMD) are commonly reported as a description of effect size. In this manuscript, we hope to alleviate some confusion. First, we provide a philosophical framework for conceptualizing SMDs; that is, by dichotomizing them into two groups: magnitude-based and signal-to-noise SMDs. Second, we describe the statistical properties of SMDs and their implications. Finally, we provide high-level recommendations for how sport and exercise scientists can thoughtfully report raw effect sizes, SMDs, or other effect sizes for their own studies. This conceptual framework provides sport and exercise scientists with the background necessary to make and justify their choice of an SMD.