Bayesian Approaches to Quantifying Uncertainty in Sport and Exercise Measurements

A Guide for Practitioners and Applied Researchers

Abstract

Measurement, analysis and communication of data are key requirements of practitioners and applied researchers in sport and exercise. Many challenges exist, however, in the analysis and communication of data due primarily to noisy measurements and the use of traditional analysis approaches designed for large data acquisitions and long run perspectives. The purpose of this review is to provide an introduction and worked examples to Bayesian statistics that present an alternative to traditional frequentist approaches and explore the means by which they may ameliorate some of the aforementioned limitations. The review focusses on the importance and methods used to include expert knowledge within Bayesian analyses through informative prior distributions. Additionally, the review focusses on the use of subjective probabilities and the subsequent ability to communicate results in an intuitive and informative manner. The review explores both conjugate and sample-based approaches to Bayesian analyses, highlighting their different strengths and the contexts where they may best be used. Worked examples show where substantive prior information can be incorporated, uncertainty can be decreased to obtain more relevant information and thereby underpin appropriate decision making. Given the flexibility afforded by Bayesian approaches and the advantages they can provide, there is a need for more resource development and interdisciplinary work between researchers and practitioners to increase their use in sport and exercise.