In this meta-analysis, we sought to examine the 'active ingredients' (or behaviour change techniques; BCTs) used within theory-based physical activity interventions compared to interventions with no stated theory. We retrieved 171 peer-reviewed studies (224 total interventions) that used a controlled experimental design from 68 previous reviews of physical activity interventions. Data from each intervention were coded with regard to their use of theory and inclusion of 16 BCT clusters within the physical activity intervention. There were no significant differences in the overall effect sizes between theory-based (k = 148, d = 0.48) and no-stated-theory (k = 77, d = 0.37) interventions. Theory-based interventions incorporated a greater number of BCT clusters on average (6.1) compared to no-stated-theory interventions (4.5). Significant effects were shown for interventions that incorporated at least three BCT clusters (d = 0.48) but not for those that used one or two (d = 0.20). Several BCT clusters were more likely to be present in theory-based interventions than no-stated-theory interventions. Significant effects on physical activity were also shown for theory-based interventions that incorporated any of the 16 BCT clusters coded, but only for 9 out of 11 no-stated-theory interventions in this regard (for which effect sizes could be calculated). Taken together, these findings suggest that although the overall effects on physical activity do not differ significantly between theory-based and no-stated-theory interventions, these interventions often differ in their composition of BCTs. Moreover, for interventions utilising certain BCT clusters (namely, 'self-belief' and 'association'), theory may be necessary to derive significant effects.
Keywords: Behaviour change; exercise; health; intervention; physical activity; theory.