Online Program Home
My Program

Abstract Details

Activity Number: 680 - Imputation and Analysis of Missing Survey Data
Type: Contributed
Date/Time: Thursday, August 2, 2018 : 10:30 AM to 12:20 PM
Sponsor: Survey Research Methods Section
Abstract #327081 Presentation
Title: "Robust-Squared" Imputation Models Using BART
Author(s): Yaoyuan Tan* and Carol A.C. Flannagan and Michael Elliott
Companies: University of Michigan and University of Michigan, Transport Research Institute and University of Michigan
Keywords: Missing Data; Doubly robust estimators; Multiple imputation; Bayesian additive regression trees; Inverse probability weighting; National Automotive Sampling System Crashworthiness Data System
Abstract:

Examples of "doubly robust" (DR) estimators for missing data include augmented inverse probability weighting (AIPWT; Robins et al., 1994) and penalized splines of propensity prediction (PSPP; Zhang and Little, 2009). DR estimators have the property that, if either the response propensity or the mean is modeled correctly, a consistent estimator of the population mean is obtained. However, DR estimators can perform poorly when modest misspecifi cation is present in both models (Kang and Schafer, 2007). Here we consider extensions of the AIPWT and PSPP models that use Bayesian Additive Regression Trees (BART; Chipman et al., 2010) to provide highly robust propensity and mean model estimation. We term these "robust-squared" in the sense that the propensity score, the means, or both can be estimated with minimal model misspecifi cation, and applied to the DR estimator. We consider their behavior via simulations where propensities and/or mean models are misspecifi ed. We also apply our proposed method to impute missing instantaneous velocity (delta-v) values from the 2014 National Automotive Sampling System Crashworthiness Data System dataset.


Authors who are presenting talks have a * after their name.

Back to the full JSM 2018 program