Page 154 - Contributed Paper Session (CPS) - Volume 7

CPS2050 Marijke Welvaert

The statistical confidence crisis in Sport Sciences:
Is it all “shoddy statistics”?
1,2
Marijke Welvaert
1 Research Institute for Sport and Exercise, University of Canberra, Australia
2 Australian Institute of Sport, Canberra, Australia

Abstract
After decades of gaining popularity within sport science, Hopkins’ magnitude-
based inference has been discredited as a statistical method. While its
rationale, encouraging interpretation of magnitudes of effects rather than
solely relying on p-values, is applaudable, the implementation has
fundamental flaws and did not survive rigorous statistical review. The
questions remains what is the impact of this polarised debate in sport science.
A frequency analysis of recent sport science literature demonstrates that the
majority of studies use traditional statistical inference. The analysis also shows
that statistical practice within sport science could be updated to be more
aligned with more modern statistical methods. A continued focus on
interpreting evidence beyond p-values using validated techniques should
bring this statistical confidence crisis to an end.

Keywords
Frequentist; Magnitude based inference; Effect size; Confidence intervals;
Statistical practice

1. Introduction
Sport science is a multidisciplinary field that studies how the human body
works during exercise and how sport and physical activity promotes health
and performance. Recently, the field attracted the attention of the popular
media following a second statistical review (Sanaini, 2018) of a common
analysis method used in the field, namely “Magnitude-based Inference” (MBI;
Hopkins et al. 2009). In 2015, Welsh and Knight were the first to publish a
statistical review and both reviews agreed that while the rationale behind MBI
should be encouraged (i.e. more focus on the magnitude of effects, rather than
a significance driven interpretation), the implementation has fundamental
flaws.
Despite world-renowned statisticians speaking out against MBI,
proponents of the method persist in its validness and still encourage sport
scientists to use it under a different name (i.e. reference Bayesian inference
with a dispersed uniform prior) to avoid critique during the publication
process (Hopkins & Batterham, 2018). On the other hand, high-ranked
journals in the field (e.g. Medicine and Science in Sports and Exercise (MSSE),
141 | I S I W S C 2 0 1 9