Structural equation model of physical activity in Turkish schoolchildren: an application of the integrated behavioural model

OBJECTIVES: Childhood obesity is increasingly prevalent in the developing world including Turkey. This study examined constructs of the integrated behavioural model associated with physical activity in a sample of schoolchildren in Ankara, Turkey using structural equation modelling. DESIGN: Cross-sectional survey by probability sampling. SETTING: Fifteen schools of different socioeconomic strata in Ankara, Turkey with grade 4 students. PARTICIPANTS: 2066 (969 girls and 1097 boys) grade 4 schoolchildren and their parents selected using a probability-based sampling frame. PRIMARY OUTCOME MEASURES: Three primary outcomes were used: moderate-to-vigorous physical activity, team sport participation, sedentary behaviour. RESULTS: Data were collected from 2066 fourth-grade children from schools of three socioeconomic strata. Missing data were imputed using multiple imputation. To examine the integrated behavioural model, a structural equation model containing latent constructs for physical activity outcome expectancies, self-efficacy, home environment and social norms were fitted with the three outcomes above. Adequate model fit was achieved in the structural equation model (χ(2)=1821.97, df=872, p<0.001, Comparative Fit Index=0.91, Tucker Lewis Index=0.91, root mean square error of approximation=0.02, standardised root mean square residual=0.04). All scale items were significantly associated with their respective latent constructs (all p<0.001). Several significant pathways between latent constructs and outcomes of interest were observed (p<0.05). Self-efficacy was positively associated with moderate-to-vigorous physical activity (p<0.001) and team sport participation (p<0.001) and negatively associated with sedentary behaviour (p<0.001). Negative outcome expectancies were negatively associated with moderate-to-vigorous physical activity (p<0.01) and sedentary behaviour (p<0.01) while positive outcome expectancies were positively associated with team sport behaviour (p<0.001) and negatively associated with sedentary behaviour (p<0.05). Home support was positively associated with moderate-to-vigorous physical activity (p<0.01) and team sport participation (p<0.05). Finally, physical activity social norms were negatively associated with sedentary behaviour only (p<0.05). CONCLUSIONS: This study supported the extension to Turkish children of the integrated behavioural model in relation to physical activity behaviours. Results illustrate multiple targets for interventions to increase physical activity.