Mechanisms of physical activity behavior change in an incentive-based intervention: Mediation analysis

Objective: The Physical Activity Loyalty (PAL) scheme was a cluster randomized controlled trial of a six-month complex intervention targeting workplace physical activity. Financial incentives were incorporated in an evidence-based behavior change programme, including self-regulation techniques. This paper examines short-term (<six months) and long-term (≥six months) mediation effects on physical activity. Methods: Participants included 853 adults (n=457 intervention, n=396 control). Physical activity was objectively-assessed using pedometers at baseline, six and 12 months. Hypothesized short-term mediators were assessed at baseline and four weeks (e.g. self-efficacy, intentions, outcome expectations). Hypothesized long-term mediators were assessed at baseline and six months (e.g. habit, intrinsic motivation, planning). Mediation models employed the Structural Equation Modelling product-of-coefficients approach. Results: Intervention participants experienced significant decreases in six-month pedometer steps/day versus controls (b=-336, SE=141, p=0.02), which were partially mitigated by positive indirect effects through six-month integrated regulation (ab=94.7, 95% CI: 18.7, 204.4), intrinsic motivation (ab=59.0, 95% CI: 3.09, 154.5) and habit (ab=198.7, 95% CI: 84.3, 369.9). There were no between-group differences in 12-month pedometer steps/day but positive indirect effects through six-month integrated regulation (ab=128.0, 95% CI: 27.3, 313.2), planning (ab=115.0, 95% CI: 3.71, 285.5) and habit (ab=153.3, 95% CI: 39.3, 333.1). Conclusions: Most examined mediators were non-significant and mediation analyses did not explain decreases in physical activity for interventions versus controls. Decreases were partially mitigated by increased integrated regulation, intrinsic motivation, planning and habit. Results show that, contrary to the hypotheses of self-determination theory, intrinsic motivation is not necessarily adversely impacted if financial incentives are embedded in a complex intervention.