Measuring processes of change in behavioural interventions: Insights gained from linking mechanisms of action to associated measures

Testing mechanisms of action (MoAs) hypothesized to drive behaviour change improves intervention efficacy and allows theoretical propositions to be evaluated, enabling evidence accumulation. However, clear correspondence between behavioural MoAs and associated measures is lacking, creating challenges for intervention evaluation. To link well-defined behavioural MoAs to multidimensional measures. Two researchers independently judged whether 44 measures (comprising 131 (sub)scales) in the Science of Behavior Change (SOBC) Measures Repository are suitable for measuring 270 MoAs from the Human Behaviour-Change Project's MoA Ontology (2022). Links were categorized as 'confirmed' (aligned with a prior expert opinion study linking measures to 26 MoAs), 'removed' or 'new'. Judgements were compared, discussed, reconciled iteratively and jointly reviewed for consistency. Six hundred and eighty-six links between SOBC measures and MoAs were identified (397 'confirmed', 289 'new'). Measures were found to tap into multiple MoAs, with 5.24 MoAs linked to each measure on average. These links demonstrated greater granularity than those identified in a previous expert opinion study because MoAs from the MoA Ontology were more specific than MoAs from this previous study. Commonly co-occurring MoAs were identified (e.g., 'self-regulation process' and 'self-regulation capability') and MoAs potentially missing from the ontology were noted. The refined measure-MoA links provide more precise guidance for researchers when designing and/or selecting measures to assess the role of MoAs in theory-based behavioural interventions. Future research should further explore measure-MoA links by, for example, testing the discriminant content validity of ostensibly distinct measures that tap into the same or similar MoAs.