|
Activity Number:
|
104
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, August 4, 2008 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Health Policy Statistics
|
| Abstract - #302403 |
|
Title:
|
Use of Propensities with Unmeasured Confounders
|
|
Author(s):
|
David Nelson*+ and Siamak Noorbaloochi
|
|
Companies:
|
Minneapolis VA Medical Center and VAMC and University of Minnesota
|
|
Address:
|
University of Minnesota, Minneapolis , MN, 55417,
|
|
Keywords:
|
Propensity Theory ; Missing Confounders ; Dimension Reduction
|
|
Abstract:
|
Observational studies assessing causal relationships between an explanatory measure and an outcome can be complicated by hosts of potential confounding measures. Propensity theory effectively addresses confounding when all confounders are measured but the assumptions underlying nonconfounded causal inference using propensities breakdown when there are unmeasured confounders. However, when there are unmeasured confounders, the dimension reduction central to propensities and related summaries can still be utilized in the inferential process. We discuss how propensity theory can be combined with simple covariate free methods for addressing unmeasured confounders to estimate the causal effects.
|
