Understanding and Misunderstanding Randomized Controlled Trials

The use of RCTs has a long history in economics, but has recently greatly expanded, particularly in economic development. Like other econometric methods, RCTs are often informative, but like other econometric methods, they have problems and pitfalls that are not always fully understood. Common misunderstandings include balance, standard errors and inference, and blinding. We argue that the concepts of internal and external validity as commonly used are unhelpful. We focus on how to use the results of RCTs,  arguing that simple replication is rarely useful, but rather that the results of RCTs need to be combined with other knowledge and evidence, and used as part of a serious economic analysis. Examples are provided along the way.