Activity Number:
|
236
- Causal Modeling Methods in Epidemiology
|
Type:
|
Contributed
|
Date/Time:
|
Monday, July 31, 2017 : 2:00 PM to 3:50 PM
|
Sponsor:
|
Section on Statistics in Epidemiology
|
Abstract #322638
|
View Presentation
|
Title:
|
Methods for Testing of the Relative Bias of an Instrumental Variable Estimator and Confounder Selection
|
Author(s):
|
Byeong Yeob Choi* and Jason Fine and Maurice Alan Brookhart
|
Companies:
|
University of Texas Health Science Center at San Antonio and University of North Carolina at Chapel Hill and University of North Carolina at Chapel Hill
|
Keywords:
|
Bias ratio ;
Confounder selection ;
Instrumental variable ;
Testing of equivalence hypotheses
|
Abstract:
|
An instrumental variable (IV) should satisfy some assumptions to yield an IV estimator less biased than an ordinary least squares (OLS) estimator. Testing of equivalence hypotheses on the bias ratio of an IV estimator and an OLS estimator is proposed. Each measured covariate is used to yield a testing result under an assumption that the measured covariates are related to an unmeasured confounder via linear models. A method to summarize the testing results from each covariates is also proposed. We derive a bias-corrected estimator from the expression of the bias ratio. We show that the standard error of the estimated bias ratio from each covariate has an important role to select the covariates that are the most related to an unmeasured confounder. Simulation studies demonstrate that the proposed methods for testing of the relative bias, confounder selection and bias-corrected estimator work well.
|
Authors who are presenting talks have a * after their name.