Limited access Past event
Applied Microeconomics Group Seminar Combining Experimental and Observational Studies in Meta-Analysis: A Debiasing Approach
Speaker: Dr Rachael Meager, LSE
Abstract: We propose a method for aggregating evidence from observational studies, which may be subject to internal bias in estimating causal effects, and randomized controlled trials (RCTs), which may be subject to site selection bias. We use an instrument for research design choice (observational or experimental) to nonparametrically identify average internal and site selection bias and remove both. Since we observe study results with error and often have small samples of studies, we develop a parametric hierarchical Bayesian approach to estimation. Our specific implementation uses the entry of an RCT-facilitating organisation such as the Jameel Poverty Action Lab (JPAL) or Innovations for Poverty Action (IPA) as a differences in differences instrumental variable. Applying this strategy to the conditional cash transfer program (CCT) and microcredit literature shows substantial internal selection bias in observational studies of CCTs but little such bias for microcredit. Neither application shows evidence of RCT site selection bias.