A General Approach to Causal Mediation Analysis

• Activity Number: 568
• Type: Contributed
• Date/Time: Thursday, August 6, 2009 : 8:30 AM to 10:20 AM
• Sponsor: Biometrics Section
• Author(s): Kosuke Imai*+ and Luke Keele and Dustin Tingley
• Companies: Princeton University and The Ohio State University and Princeton University
• Keywords: causal inference ; sensitivity analysis ; identification ; randomized experiments ; observational studies ; causal mechanisms

Abstract

In an influential paper, Baron and Kenny (1986) proposed a statistical method to conduct a causal mediation analysis. We generalize this method in order to accommodate linear and nonlinear relationships, parametric and nonparametric models, continuous and discrete mediators, various types of outcome variables, and single or multiple mediators. We provide the formal justifications for the proposed generalization by placing causal mediation analysis within the formal statistical justifications for the proposed generalization of the Baron-Kenny procedure by placing causal mediation analysis within the potential outcomes framework. Finally, we develop a set of sensitivity analyses that allow applied researchers to formally examine the robustness of their empirical conclusions. Such sensitivity analysis is important since causal mediation analysis typically rests on a strong assumption.