The application of regression based normative profiling for international male Taekwondo performance

Abstract

The purpose of this study was to compare the predictive accuracy of two regression based normative profiling methodologies (O’Donoghue & Cullinane, 2011; 2- Nonparametric regression) for the assessment of elite male Taekwondo performance. Following ‘inter-operator’ reliability analysis, retrospective performance data (319 performance indicators; 167 matches) was used for forty-eight elite senior male Taekwondo athletes (<58kg - <80kg) during participation across 22 major competitions between 2010 and 2013. The world rankings of all athletes were converted into relative quality ratings (RQ), with an RQ differential calculated for all respective matches. A Chi-Square χ2 test of independence was employed to identify the existence of a significance association between relative quality and match outcome. All data was then subjected to either Pearson’s r (O’Donoghue & Cullinane, 2011) or Kendall’s Tau (τ) (Non-Parametric regression) correlation analysis, where the performance indicators (n=34) deemed to bear the most meaningful relationships with relative quality were then included within the both profiling methodologies. The ‘standard error of estimate’ (SEE) and SEE% values were computed for all performance indicators and subjected to both Mann Whitney-U and binomial testing comparisons, from which the most accurate method was recruited to analyse a selected athlete. A Chi-square test of independence identified the validity of including relative quality within regression based performance profiling (χ2; P < .01). Non-parametric regression was found to exhibit moderately superior mean SEE and SEE% values, in addition to superior SEE and SEE% values for a greater proportion of the performance indicators of interest. It was concluded that non-parametric regression offered an advance upon previous profiling methodologies for the assessment of elite male Taekwondo performance.