The measurement of variability in coordination during locomotion.

Traditionally in sport and exercise biomechanics, variability in movement has been regarded as system noise or error. However, the advent of dynamical systems theory has stimulated a radical reassessment of the concept of variability with it being regarded as functional in some circumstances. With the application of dynamical systems theory to studies in biomechanics, many methodological considerations have emerged. These include the effects of measurement error on observed variability, the suitability of techniques available for quantifying coordination variability and the efficacy of using a treadmill to simulate overground locomotion in terms of coordination variability. The overall purpose of this thesis was to address the issues related to the methodological considerations for studies of coordination variability. In Chapter III, the effects of measurement error on coordination variability were investigated. It was suggested that, of the two major components of measurement error - skin movement and instrument errors - skin movement errors are likely to have a greater effect on measures of coordination variability due to their larger magnitude. If skin movement errors were entirely random, rather than systematic, they would be extremely detrimental to the study of coordination variability, as their magnitude is often greater than the magnitude of variability in human movement. Preliminary analysis of data presented by Holden et al. (1997) revealed that errors due to skin movement were primarily systematic, which suggests that they are less problematic for investigations of variability. In Chapter IV, the suitability of techniques available for quantifying variability in coordination was assessed. The results of an analysis of supporting experimental data suggested that disparate answers to certain research questions might be obtained dependent on the technique used to quantify variability in coordination. It is clear from Chapter IV that comparisons between studies of coordination variability that used different quantification techniques should be made with caution. Researchers should be aware of the benefits and limitations of each technique and the choice of technique should be based on the research question of interest. A comparison of variability in coordination measured overground and on a treadmill was provided in Chapter V. Results indicated that overground running was associated with greater variability in coordination than treadmill running. Therefore, it is possible that performing studies on a motorised treadmill might mask differences in coordination variability between experimental groups. Potentially, the reduced coordination variability during treadmill running can be explained by the artificially constant speed of the treadmill belt externally driving the foot through the stance period. The 'treadmill-on-demand' (Minetti et al., 2003) is an innovative type of feedback controlled treadmill on which the belt speed is not constant so the participant is not constrained to run at a constant speed. Therefore, in Chapter VI, variability in coordination measured during overground, conventional treadmill and treadmill-on-demand running was compared. Treadmill-on-demand running resembled overground running no better, in terms of variability in coordination, than conventional treadmill locomotion and it appears to be no more suited to studies of coordination variability than the conventional treadmill. Although some unanswered questions remain, this thesis has enhanced understanding of important methodological considerations for the study of coordination variability during running.