Physical exercise participation: A continuous or categorical phenomenon?

Objective Measures of exercise participation are usually one-dimensional continuous variables (e.g., participation frequency). However, there is evidence that projecting exercise participation onto only one dimension cannot adequately reflect the complex multi-dimensional nature of this behaviour. The present study served to identify distinct patterns of exercise participation, to introduce a normative system that allows individual classification of these participation patterns, and to test whether the prediction of exercise participation through psychological variables benefits when one chooses a categorical (multi-dimensional) instead of a continuous operationalisation of the behaviour. Method Exercise participation of NÂ =Â 174 customers of a fitness centre was recorded electronically for 32 weeks. Subjects completed a questionnaire including the psychological variables perceived behavioural control, outcome expectations, strength and self-concordance of goal intention which are known to be relevant predictors of exercise participation. Results Four different participation patterns were identified by cluster analysis: maintenance, fluctuation, late dropout and early dropout. Based on these findings a normative classification system (NOCLEP) was developed to allow for a sample-independent assignment of individuals to these four participation patterns (categorical measure of exercise participation). For some psychological variables the prediction of exercise behaviour improved markedly when this categorical measure instead of a continuous measure was used. This improvement only occurred when the psychological predictors exhibited a non-linear relation to the continuous exercise measure. Conclusion Analyses with categorical criterion measures may allow a deeper understanding of the role of specific psychological predictors in exercise participation. Furthermore, NOCLEP might be used as a diagnostic tool in the practice of exercise psychology