Causal Mediation Analysis with the Multiply Robust Method Under Unmeasured Cluster-Level Confounders (Poster 20)

• In Event: Division D Graduate Student Committee In-Progress Research Gala Poster Session

Abstract

In education and behavioral research, existing causal mediation methods for clustered data are limited. We extend a multiply robust method to estimate causal mediation effects in clustered data, accounting for unmeasured cluster-level confounders. We consider individual- and cluster-level effects with continuous and binary mediators and outcomes. To mitigate bias from model misspecification, we integrate machine learning techniques for nuisance model estimation. Through simulations, we evaluate the method’s performance for inference in causal mediation analysis with clustered data where an unmeasured cluster-level confounder is present. We demonstrate its application using data from the National Longitudinal Study of Adolescent to Adult Health.

Authors

• Cameron McCann, University of Texas at Austin

• Xiao Liu, University of Texas at Austin