A Meta-Meta-Analysis of Behavior Change Interventions: Two Tales of Behavior Change

Abstract

Behavioral science interventions like incentives, nudges, and boosts are increasingly used in public policy, but their effectiveness remains debated. We conducted a meta-metaanalysis on behavior change interventions across health, finance, and sustainability outcomes. 20 Our analysis covers 838 effects from 269 meta-analyses, encompassing 6,327 randomized controlled trials and over 9 million individuals from non-clinical populations of all ages in both developed and developing economies. Our findings tell two stories: First, extracted treatment effects are generally positive but highly variable (M = .173; SD = .195), indicating some interventions impact behavior. However, after adjusting for publication bias, the mean posterior 25 effect pooling domains and interventions is .063 (95% credible interval .044 to .08, BF10 = 139.8) with substantial unexplained heterogeneity (τ̂= .129). Future research requires improved reporting and deeper contextual analysis to address this heterogeneity. Even small effect sizes can yield significant impacts when scaled across populations and sustained over time.