| Summary: | A needs analysis of canoe slalom (coach and athlete led) resulted in a reliable performance analysis system for training and competition. This was achieved using a procedure recommended by O'Donoghue & Longville (2004). Reliability tests concluded that errors for split-times and upstream analyses were unacceptably high if coaches or less trained analysts collected data due to the inconsistent application of operational definitions. Study 2 used percentage times off the K1M and class winners as a measure of performance. Variability was high due to athlete, course and class differences. Race percentages from the 2013-16 Olympic cycle were used to test the probabilities associated with winning, medalling or reaching a final calculated from the 2009-12 Olympic cycle. Signal detection theory determined an appropriate balance between the risk of misses and false alarms (inversely proportional) with results supporting the use of race percentages off the class winner, rather than the K1M, and a 50% level of probability for predicting gaining a medal i.e. a low rate of false alarms (maximum 6%) and a high hit rate (over 70% of medals correctly identified). Study 3 tracked athletes' performances over time using exponentially smoothed ICF race points. Performance funnels were created for winning (previously won a major championship) and winless athletes using the median and 95% confidence intervals for the median. Time series plots for an athlete (from the start of their International career) were synchronised with the performance funnels to allow easy visualisation of performance. Nearly all athletes' time series depicted a period of initial improvement followed by a plateau and then deterioration in performance over a 7½ year period. Athletes were also classified into probable, possible and unlikely to win a future major. This thesis provided coaches and athletes with academically rigorous methodologies to aid their understanding of canoe slalom performance.
|
|---|
