On Causal Mediation Effect for Nonlinear Outcomes

• Activity Number: 356
• Type: Contributed
• Date/Time: Tuesday, August 11, 2015 : 10:30 AM to 12:20 PM
• Sponsor: Mental Health Statistics Section
• Author(s): Pan Wu*
• Companies: CHRISTIANACARE
• Keywords: Causal Inference ; Medication Effect ; Non-linear Outcomes ; Robust Inference ; Mental Health

Abstract

Causal mediation effects have been well developed under potential outcomes framework. However, most literature focused only on the continuous outcomes for its mathematical convenience. VanderWeele and Vansteelandt (2009, 2010) introduced an approach to estimate causal medication effect for binary and count outcomes through the measurements of causal odds ratio and causal relative risk respectively. Muthen (2011) extended the causal mediation effect by estimating the indirect effect for non-linear outcomes via probit link function, but the inference is very complicated. In the talk, we present a new class of models to easily estimate causal medication effect for non-linear outcomes, which can provide robust inference and good asymptotic properties under regularity condition.