|
Activity Number:
|
393
|
|
Type:
|
Invited
|
|
Date/Time:
|
Wednesday, August 1, 2007 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
ENAR
|
| Abstract - #308113 |
|
Title:
|
Accounting for Unobserved Confounders via Data Augmentation
|
|
Author(s):
|
David B. Allison*+
|
|
Companies:
|
The University of Alabama at Birmingham
|
|
Address:
|
Dept. of Biostatistics, Ryals Public Health Bldg, 420C, Birmingham, AL, 35294,
|
|
Keywords:
|
causal effects ; adiposity ; mortality rate ; putative confounding ; weight loss
|
|
Abstract:
|
A recurrently identified challenge in the analysis of data seeking to estimate the causal effects of adiposity on mortality rate is the putative confounding by unobserved variables that both predispose to weight loss and to earlier death. To correct for this, investigators have utilized a variety of ad hoc procedures that typically entail eliminating subsets of subjects in the hopes that the analyses of remaining subjects will yield unbiased estimates. Unfortunately, such ad hoc procedures enjoy little statistical foundation and have not been validated. I will propose an alternative approach in which information on putative but unobserved confounders is taken from other studies and incorporated into analyses of the dataset in hand using techniques adapted from multiple imputation. The approach will be illustrated with both real and simulated data.
|
- 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 2007 program |
