A 15-year longitudinal test of the theory of planned behaviour to predict physical activity in a randomized national sample of Canadian adults

Abstract

Background: There have been very limited long-term prospective studies examining social-cognitive models in the PA domain with large, representative population samples. Objective: The primary objective was to examine the Theory of Planned Behaviour's (TPB) ability to explain PA behaviour in a randomly selected national sample of Canadian adults over a 15-year period. A secondary objective was to test the moderating effects of gender on TPB. Design: The study employed a longitudinal design. Methods: A cohort of 1427 randomly-selected adults reported TPB constructs and PA behaviour in 1988, and again in 2003. Structural equation modelling techniques were employed to test TPB. Results: The full model including both 1988 and 2003 TPB measures accounted for 29% and 21% of the variance in intention, and 9% and 22% of the variance in PA in 1988 and 2003, respectively. Gender moderated aspects of the TPB in the full model. The core predictive model (1988 TPB constructs predicting 2003 PA behaviour) explained 13% of the variance in 2003 PA behaviour. Conclusions: The TPB's long-term ability to predict PA behaviour is modest. © 2012.