Estimating Heterogeneous Causal Effects of High-Dimensional Treatments: Application to Conjoint Analysis

• Activity Number: 422 - Recent Development of Machine Learning Methods in Causal Inference
• Type: Invited
• Date/Time: Thursday, August 12, 2021 : 4:00 PM to 5:50 PM
• Sponsor: Section on Statistical Learning and Data Science
• Author(s): Max Goplerud and Kosuke Imai* and Nicole E Pashley
• Companies: University of Pittsburgh and Harvard University and Rutgers University
• Keywords: causal inference; factorial design; mixture model; randomized experiment; regularized regression

Abstract

Exploiting the recent advancement of machine learning algorithms, researchers have developed various ways to infer heterogeneous treatment effects. Most of the existing methods, however, focus on estimating the conditional average causal effects of a single, binary treatment given a potentially large number of individual characteristics. In this paper, we propose a method to estimate the heterogeneous causal effects of high-dimensional treatments using the data from a randomized experiment. Specifically, we use a mixture model to cluster individuals who exhibit similar patterns of treatment effects. By directly modeling cluster membership, the proposed methodology further allows researchers to associate the individual characteristics with the heterogeneity of treatment effects. Our motivating application is conjoint analysis, which is a popular survey experiment in social science and marketing research and is based on a high-dimensional factorial design. We develop a finite mixture model of regularized regressions and apply it to the conjoint data.