Sources of variability in time-motion data; measurement error and within player variability in work-rate

Abstract

The purpose of the current study was to evaluate the intra- and interobserver reliability of work-rate in soccer as well as looking at match to match as well as within match variability in work-rate in soccer. A 60 subject intra-observer reliability study revealed 95% ratio limits of agreement of 1.025×/÷1.155 while a 60 subject inter-observer reliability study revealed 95% ratio limits of agreement of 1.050×/÷1.337 for workrate. The main study used PlayerCam data from 113 FA Premier league matches that were covered by the interactive Sky Sports Extra channel. The PlayerCam facility normally covers 6 different players per match, 3 from each team, for 15 minutes each. This provided a total of 678 fifteen minute observations from 277 individual players. Some players were covered on more than one occasion with one player being covered on 11 occasions. The 61 players who were observed twice, had a mean absolute difference between their first and second observed performances of 2.76±1.94% which was greater than the difference between positional group means. The one player who was observed on 11 occasions had a greater variability in work rate (12.48±2.71%) than the variability between the 115 different midfielders analysed during the current investigation (11.01±2.63%). On one occasion, the PlayerCam facility covered a single subject for the entire duration of the match, showing considerable variation between successive 15 minute periods of the match (12.27%, 9.49%, 8.35%, 9.11%, 7.43%, 6.55%). The current investigation has revealed that the reliability of timemotion analysis is limited. Furthermore, individual match data for subjects may be unrepresentative as the level of within subject variability is comparable with the level of between subject variability.