On Minimizing the Risk of Bias in Randomized Controlled Trials in Economics

Estimation of empirical relationships is prone to bias. Economists have carefully identified and addressed sources of bias in structural and quasi-experimental approaches, but the randomized control trial (RCT) has only recently begun to receive such scrutiny. In this paper, we argue that several lessons from medicine, derived from analysis of thousands of RCTs conducted over the past 60 years and establishing a clear link between certain practices and biased effect estimates, can be used to reduce the risk of bias in economics RCTs. We first identify the subset of these lessons applicable to RCTs in economics. We then use them to assess the risk of bias in estimates from economics RCTs published between 2001 and 2011. Most studies fail to report important details on study design necessary to assess risk of bias. Many report practices known to bias effect estimates. We conclude with suggestions on how to remedy these issues