Adaptive Experiments Toward Learning Treatment Effect Heterogeneity

Speaker: Jingshen Wang, UC Berkeley

Title: Adaptive Experiments Toward Learning Treatment Effect Heterogeneity

Abstract:  Understanding treatment effect heterogeneity has become an increasingly popular task in various fields, as it helps design personalized advertisements in e-commerce or targeted treatment in biomedical studies. However, most of the existing work in this research area focused on either analyzing observational data based on strong causal assumptions or conducting post hoc analyses of randomized controlled trial data, and there has been limited effort dedicated to the design of randomized experiments specifically for uncovering treatment effect heterogeneity. In this manuscript, we develop a framework for designing and analyzing response-adaptive experiments toward better learning treatment effect heterogeneity. To be specific, we provide algorithms that revise the data collection mechanism in an adaptive manner according to the accrued evidence during the experiment, which allows for identifying subgroups with the largest treatment effects with enhanced statistical efficiency. The resulting framework not only complements A/B tests in e-commerce but also unifies adaptive enrichment designs and response-adaptive randomization designs in clinical settings. We demonstrate the merit of our design in simulation studies, synthetic e-commerce, and clinical trial data analyses.