Online Program Home
  My Program

All Times EDT

Abstract Details

Activity Number: 24 - Recent Advances in Causal Analyses That Tell the Story of Complex Mediation Mechanisms
Type: Topic-Contributed
Date/Time: Sunday, August 8, 2021 : 1:30 PM to 3:20 PM
Sponsor: Biometrics Section
Abstract #317044
Title: Heterogeneous Indirect Effects for Multiple Mediators Using Interventional Effect Models
Author(s): Wen Wei Loh* and Beatrijs Moerkerke and Tom Loeys and Stijn Vansteelandt
Companies: Ghent University and Ghent University and Ghent University and Ghent University
Keywords: conditional effects; direct and indirect effects; effect decomposition; effect modification; natural effect models

Decomposing a treatment effect on an outcome into separate path-specific effects via multiple mediators requires strict assumptions, such as correctly postulating the mediators' causal structure and no unmeasured confounding among the mediators. In contrast, interventional indirect effects for multiple mediators can be identified even when these assumptions are violated. Existing estimation methods require calculating each distinct interventional indirect effect in turn. This can quickly become unwieldy when investigating effects modified by observed baseline covariates. In this article, we introduce simplified estimation methods for such heterogeneous interventional indirect effects using interventional effect models. Interventional effect models are a class of marginal structural models that encode the interventional indirect effects as causal parameters, thus readily permitting effect modification using (statistical) interaction terms. The mediators and outcome can be continuous or noncontinuous. We propose two estimation procedures: one using inverse weighting by the counterfactual mediator density or mass functions, and another using Monte Carlo integration.

Authors who are presenting talks have a * after their name.

Back to the full JSM 2021 program