Ratio of Mediator Probability Weighting for Estimating Natural Direct and Indirect Effects

• Activity Number: 392
• Type: Topic Contributed
• Date/Time: Tuesday, August 3, 2010 : 2:00 PM to 3:50 PM
• Sponsor: Biometrics Section
• Author(s): Guanglei Hong*+
• Companies: The University of Chicago
• Keywords: Causal inference ; Counterfactual outcomes ; Identification ; Mediation ; Non-parametric method ; Treatment-mediator interaction

Abstract

Decomposing the total effect into natural direct and indirect effects is central to revealing causal mechanisms. The standard method of path analysis does not apply when the direct effect depends on the mediator value. Alternative methods typically involve specifying an outcome model as a function of the treatment, the mediator, and the observed covariates and require a linear association between the outcome and the mediator. I propose a non-parametric approach that estimates the marginal mean of each counterfactual outcome through weighting. The weighted distribution of the mediator under the experimental condition approximates their counterfactual mediator distribution under the control condition. The method applies regardless of the distribution of the outcome or the functional relationship between the outcome and the mediator, and is suitable for handling many covariates.