Features of Successful Behavioral Interventions: A Data Driven Analysis of Megastudy Effects

• Analyzed 274 interventions from 15 behavioral science megastudies • Identified 19 structural, modal, and conceptual intervention features • Found outcome specificity uniquely predicts intervention effectiveness • Provides data-driven guidance for designing more effective interventions Megastudies, or the empirical testing of many theoretically based interventions on the same set of outcomes in the same large experiment, have increased in recent years. This method is successful at identifying the most effective theoretical process such as social norms, or efficacy beliefs at influencing socially beneficial outcomes, ranging from vaccination behaviors to climate actions. However, less is known about the influence that underlying structural, modal, or conceptual intervention features have on their effectiveness, which limits both the theoretical interpretation of the effects, as well as the practical construction of theories in their most effective implementation. Here, we empirically examine the impact of 19 intervention features ranging from attentional ease or outcome-specificity, to delivery modality (e.g., text, video, image, open-ended questions), or conceptual construction (e.g., risk, moralization, emotions), on the effectiveness of 274 psychological interventions extracted from 15 megastudies that included N =4.13 million participants. We find that only one feature - the interventions’ specificity to the outcomes measured - significantly and consistently predicts effectiveness, highlighting the importance of tailoring interventions to outcomes of interest. This finding provides empirically guided insights into the interpretation and successful implementation of behavioural interventions in policy-relevant contexts.