Governments are increasingly using randomized controlled trials (RCTs) to evaluate policy interventions.1 RCTs are often understood to provide the highest quality evidence regarding the causal efficacy of an intervention.2 By randomly assigning participants to intervention and control groups, for example, investigators can minimize selection bias—i.e., systematic differences between those subject and not subject to the intervention that are potentially correlated with the outcome of interest. With observational studies, investigators can never be certain whether outcomes are the result of the intervention, or of some distinctive characteristic(s) for which they have not controlled.