CMStatistics 2021: Start Registration
View Submission - CMStatistics
B1694
Title: Data-driven heterogeneity detection among subgroup-specific exposure-response functions Authors:  Falco Joannes Bargagli Stoffi - Harvard University (United States) [presenting]
Abstract: Over the past several years, various tree-based methods have been developed to identify subgroups of a population with significantly different conditional average treatment effects compared with the population average effect. Despite their success when applied to settings where the exposure is binary, these methods have not yet been extended to settings with a continuous exposure variable. We develop a flexible method that extends tree-based causal effect heterogeneity identification to settings with continuous exposures, where practitioners can pre-specify a function to summarize exposure-response curves. Our tree-based identification method provides an opportunity to data-drivenly identify important subgroups in a more systematic way.