METHODS FOR QUANTIFYING THE VARIABILITY IN DATA

  • David Mullineaux
Keywords: coefficient of variation, confidence intervals, root mean square, trial size

Abstract

Variability in movement affects statistical significance and is important for interpreting data. The aim of this study was to compare methods for quantifying variability, and to use these in assessing the effect of ‘pain’ in the right leg on the running technique of one male English First Division footballer. The player’s sagittal plane movements were filmed while running on a treadmill at 3.58 m.s-1. The variability in 3 strides was quantified using standard deviation, confidence intervals (95%CI) and root mean square difference (RMSD). The kinematics of the left and right legs of the player were different, but did not contain different amounts of variability (e.g. RMSD of both knees at heel strike = 1.2°). To estimate variability the preferred techniques are: 95%CI for n = 1 as the only available; RMSD for small n; normalised techniques only when means are similar. The variability of the player’s movements in other planes and at faster speeds should be explored in future.