Estimating Heterogeneous Treatment Effects by Combining Experimental with Observational Data

• Activity Number: 64
• Type: Topic Contributed
• Date/Time: Sunday, August 9, 2015 : 4:00 PM to 5:50 PM
• Sponsor: Biometrics Section
• Author(s): Jasjeet Sekhon*
• Companies: UC Berkeley
• Keywords: Causal Inference ; Heterogeneity ; Experimental Design ; Observational Studies

Abstract

Randomized controlled trials (RCTs) can provide unbiased estimates of average treatment effects. However, even in large experiments, RCTs are usually underpowered to provide reliable estimates for subgroups. With the growth of large observational dataset an opportunity arises to combine RCTs with information from massive non-randomized studies (NRSs). We provide identification assumptions, and new methods for both experimental design and effect estimation to make this possible. Special attention is paid to methods for validating and testing the underlying assumptions.