Activity Number:
|
505
- Flexible Methods for Causality Research
|
Type:
|
Topic Contributed
|
Date/Time:
|
Wednesday, August 1, 2018 : 10:30 AM to 12:20 PM
|
Sponsor:
|
Section on Statistics in Epidemiology
|
Abstract #329091
|
Presentation
|
Title:
|
A Novel Result on Collaborative Double Robustness
|
Author(s):
|
Ivan Diaz*
|
Companies:
|
Weill Cornell Medicine
|
Keywords:
|
Collaborative double robustness;
TMLE
|
Abstract:
|
The identification and estimation of causal effects from observational data with binary exposures hinges on the assumption that the propensity score is bounded away from zero and one. When this assumption is empirically violated, standard doubly robust estimators of the causal effect suffer from high variance and large finite sample bias. In this paper we discuss a doubly robust estimation method that has smaller or equal variance than standard doubly robust estimators. Our results rely on a novel collaborative double robustness result by which the propensity score dimensionality is reduced through adjustment on an estimate of the outcome regression bias. We discuss some asymptotic properties of the estimator. We present an application as well as simulation based on synthetic data illustrating the variance and bias reduction achieved by our proposed estimator.
|
Authors who are presenting talks have a * after their name.
Back to the full JSM 2018 program
|