Online Program Home
My Program

Abstract Details

Activity Number: 497
Type: Contributed
Date/Time: Wednesday, August 3, 2016 : 8:30 AM to 10:20 AM
Sponsor: Section on Statistics in Epidemiology
Abstract #321015 View Presentation
Title: Mediation Analysis with Multilevel Additive Models
Author(s): Qingzhao Yu* and Bin Li and Richard Scribner
Companies: Louisiana State University Health Sciences Center and Louisiana State University and Louisiana State University Health Sciences Center
Keywords: Causal Effect ; Indirect Effect ; Mediation Analysis ; Multilevel Additive Models

Mediation refers to the effect transmitted by mediators that intervene in the relationship between an exposure and a response variable. Mediation analysis has been broadly studied in many fields. However, it remains a challenge for researchers to differentiate indirect effect from multiple mediators, especially when the involving variables are of hierarchical levels. Yu et al. (2014) proposed general definitions of mediation effects that were consistent for all different types of response, exposure, or mediation variables. With these definitions, multiple mediators can be considered simultaneously, and the indirect effects carried by individual mediators can be separated from the total effect. We extend the definitions to the multilevel data sets, where multilevel additive models are adapted to model the variable relationships, based on which mediation effects can be calculated at different levels. Moreover, transformations on variables are allowed for potential nonlinear relationships. An R package, mlma, was created to carry out the proposed analysis. Simulations show that the proposed method can effectively differentiate and estimate mediation effects from different levels.

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

Back to the full JSM 2016 program

Copyright © American Statistical Association