Sampling bias is an issue for research involving cyclists. The heterogeneity of cyclist populations, on the basis of skill level and riding purpose, can generate incorrect inferences about one specific segment of the population of interest. In addition, a more accurate categorization would be helpful when physiological parameters are not available. This study proposes using self-reported data to categorize amateur cyclist types by varying skill levels and riding purposes, therefore improving sample selection in experimental studies. A total of 986 cyclists completed an online questionnaire between February and October 2016. Two-step cluster analyses were performed to generate distinct groups, and dependent variables of these groups were compared (demographics and characteristics of cycling practice). The cluster analysis relied on 4 descriptors (cycling weekly volume, average cycling speed, riding purpose, and cycling discipline) and yielded five distinct groups: competitive road, recreational road, competitive mountain bike (MTB), recreational MTB and competitive triathlon. Among these groups, averages and distributions for age, height, body mass, body mass index, training volume and intensity, and years of experience varied. This categorization can potentially help researchers recruit specific groups of cyclists based upon self-reported data and therefore better align the sample characteristic with the research aims.
Keywords: Sampling; cycling; mountain bike; road; triathlon.