A guide to megastudies

Abstract

How can behavioral insights best be leveraged to solve pressing policy challenges? Because research studies are typically designed to test the validity of a particular idea, surprisingly little is known about the relative efficacy of different approaches to changing behavior in any given policy context. We discuss megastudies as a research approach that can surmount this and other obstacles to developing optimal behaviorally informed policy interventions. We define a megastudy as "a massive field experiment in which many different treatments are tested synchronously in one large sample using a common, objectively measured outcome." We summarize this apples-to-apples approach to research and lay out recommendations, limitations, and promising future directions for scholars who might want to conduct or evaluate megastudies.

Fig. 1.

Left panel: traditional field experiments randomly assign participants to multiple conditions (e.g. Conditions A and B), testing a limited number of related hypotheses. Right panel: megastudies randomly assign participants to a larger set of treatments often clustered by sub-study (different colors indicate different sub-studies), each testing potentially unrelated hypotheses (e.g. Conditions A, B, C, D, etc.).