|
Activity Number:
|
295
|
|
Type:
|
Invited
|
|
Date/Time:
|
Tuesday, August 4, 2009 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
ENAR
|
| Abstract - #303209 |
|
Title:
|
Bayesian Inference About Mediation in Longitudinal Intervention Trials
|
|
Author(s):
|
Joseph W. Hogan*+ and Shira Dunsiger and Bess Marcus
|
|
Companies:
|
Brown University and Brown University and Brown University
|
|
Address:
|
Biostatistics Section, Program in Public Health, Providence, RI, 02912,
|
|
Keywords:
|
mediator ; causal inference ; direct and indirect effects ; sensitivity analysis ; behavioral intervention trial
|
|
Abstract:
|
Behavioral scientists are interested in the effect of 'mediating' variables. Unlike drug treatments, behavioral interventions act through one or more mediating variables rather than directly on the outcome. The mediation problem can be cast in the framework of direct and indirect causal effects, where the effect of treatment that flows through a mediating variable is an indirect effect. In this talk, we use a principal stratification model to parameterize descriptive direct and indirect effects (Pearl, 2001), and show how to identify these up to a sensitivity parameter. The method uses randomization of the intervention and information from baseline covariates to identify the joint distribution of potential mediator values. Two intervention trials designed to increase exercise frequency are used for illustration.
|
- The address information is for the authors that have a + after their name.
- Authors who are presenting talks have a * after their name.
Back to the full JSM 2009 program |
