Field Tested: Assessing the Transferability of Behavioral Interventions

Abstract Applied behavioral science has grown rapidly in the past few decades as organizations and governments seek to incorporate research on human behavior into policy decisions1. At the same time, there is a growing recognition of the need to examine the robustness of empirical findings and their transferability to consequential behaviors2,3,4. Here, focusing on an important policy domain—COVID-19 booster uptake—we investigate the transferability of findings a) from prediction surveys and hypothetical experiments to field settings; and b) from one field context at one point in time to another after evolving circumstances. Across three preregistered randomized controlled trials (RCTs, total N=317,175), we tested text-based interventions that either were expected to work by behavioral scientists and laypeople (RCT1), successfully elevated vaccination intentions in online studies (RCT1 and RCT2), or effectively increased vaccination rates in prior field work (RCT3). Our research highlights the importance of field-testing interventions, as findings based on predictions and intentions fail to transfer to our real-world setting. We also find that despite repeated exposure to COVID-19 messaging in our population, reminders and psychological ownership framing still increased booster uptake. These findings have theoretical and practical implications, as they can contribute to refining both theories of behavior change and experts’ predictions about what interventions are likely to have an impact in the field.