Statistical Methods to Reduce the Effects of Measurement Error in Sport and Exercise

A Guide for Practitioners and Applied Researchers

Abstract

Quantifying uncertainty in measurements is essential to inform, monitor and evaluate interventions in sport and exercise. Many commonly used tests, particularly those that measure maximum performance or fitness exhibit large measurement errors creating uncertainty that complicates interpretations and decision making. Uncertainty in measurements can be especially problematic where expected changes across an intervention are relatively small. The purpose of the present review is to describe statistical approaches to reduce uncertainty in measurements and therein improve interpretation and decision making. These approaches include increased data collection and the use of relatively simple calculations including means and linear regression to reduce uncertainty. The review provides detailed information on the assumptions underlying each approach and the relevant statistical properties. Visuals and worked examples including R code are provided to solidify concepts and better enable practitioners and applied researchers to adopt the approaches.