Designing Effective Interventions

Co-authors: Matthias Sutter and Sebastian Tonke.

Abstract

Most interventions fail to change behavior. We argue that the reason for this failure is that the interventions do not address the underlying problem. We provide a systematic framework to diagnose underlying problems and predict intervention effectiveness ex-ante. For this, we developed a parsimonious and generalizable survey tool (anamnesis).

Our anamnesis classifies the underlying type of problem along three fundamental diagnoses: awareness, intention, and implementation problems. We then test whether our anamnesis results can predict the effectiveness of three typical interventions (reminders, incentives, simplifications) across three different settings in an online experiment with 7,500 subjects. We find very heterogeneous treatment effects across the three different settings. Our diagnosis accurately predicts this heterogeneity. On average, a predicted effectiveness of 10% corresponds to an actual effectiveness of 8.92%. Choosing an intervention based on our diagnosis increases the effect size by around 58% compared to randomly choosing one of the tested interventions. We further demonstrate the applicability of our framework to predict field behavior.