Canoe slalom--competition analysis reliability

The aim of this study was to assess intra-observer and inter-observer reliability of data gathered from a lapsed-time time-motion analysis of canoe slalom competition. The data were collected using a definition set developed in conjunction with elite canoe slalom coaches. Competition runs from four national-standard paddlers in a national selection race were analysed in random order three times by three observers. For each run, observers identified various events specific to canoe slalom, including time taken between gates, touched and missed gates, turn times, major and minor avoidance, rolls, paddle in and out of water times, and stroke classification. The error of measurement was determined for each of these variables. For time taken between gates and turn times, the error was < or = 0.21 and < or = 0.39 s for intra-observer and inter-observer analysis, respectively. The error for stroke in and out of water times was < or = 0.08 and < or = 0.13 s for intra-observer and inter-observer analysis, respectively. Analysis of stroke classification identification for intra-observer comparisons revealed that 91% of the time identical stroke identification occurred. Inter-observer analysis revealed identical stroke identification was achieved 81% of the time. These reliability data compare favourably with previous time-motion analysis in other sports using fewer variables.