Social structure, social cognition, and physical activity: A test of four models Gaston Godin 1 *, Paschal Sheeran 2 , Mark Conner 3 , Ariane Be ´langer-Gravel 4 , Maria Cecı ´lia B. J. Gallani 5 and Bertrand Nolin 6 1 Canada Research Chair on Behaviour and Health, Faculty of Nursing, Laval University, Quebec, Canada 2 Department of Psychology, The University of Sheffield, UK 3 Institute of Psychological Sciences, University of Leeds, UK 4 Research Group on Behaviour and Health, Faculty of Nursing, Laval University, Quebec, Canada 5 Department of Nursing, Faculty of Medical Sciences, State University of Campinas, Sao Paulo, Brazil 6 Institut National de Sante ´ Publique du Que ´bec, Canada Objective. This study investigated the combined influence of social structural factors (e.g. income) and cognitions in predicting changes in physical activity. Four models were tested: (a) direct effects (social structural factors influence behaviour controlling for cognitions), (b) mediation (cognitions mediate social structural influence), (c) moderation (social structural factors moderate cognition–behaviour relations), and (d) mediated moderation (cognitions mediate the moderating effects of social structural position). Design. Baseline and 3-month follow-up surveys. Methods. A random sample of 1,483 adults completed self-report measures of physical activity at baseline and 3-month follow-up. Measures of age, gender, education, income, material and social deprivation, intention, perceived behavioural control (PBC), and intention stability also were taken. Results. Apart from age, social structural factors exhibited very small or marginal effects on behaviour change, and only education moderated the intention–behaviour relation. In contrast, the magnitude of direct effects of the social cognition variables was comparatively large and intention stability mediated the moderating effect of education. Conclusions. Stable intentions and PBC are the key predictors of changes in physical activity. Consequently, our findings would suggest the value of focusing on cognitions rather than social structural variables when modelling the determinants of physical activity. * Correspondence should be addressed to Professor Gaston Godin, Canada Research Chair on Behaviour and Health, Faculty of nursing, Laval University, Paul-Comtois Building, 4108-A, Quebec, Canada G1K 7P4 (e-mail: gaston.godin@fsi.ulaval.ca). The British Psychological Society 79 British Journal of Health Psychology (2010), 15, 79–95 q 2010 The British Psychological Society www.bpsjournals.co.uk DOI:10.1348/135910709X429901