Variable Selection in Sequential Hierarchical Regression Imputation

• Activity Number: 474 - New Advances in Modeling Survey Data
• Type: Contributed
• Date/Time: Thursday, August 6, 2020 : 10:00 AM to 2:00 PM
• Sponsor: Survey Research Methods Section
• Author(s): Qiushuang Li* and Recai Yucel
• Companies: University At Albany and STATE UNIVERSITY OF NEW YORK AT ALBANY, School of Public Health
• Keywords: Clustered data; missing data; Markov chain Monte Carlo; multiple imputation; sequential hierarchical imputation; spike-and-slab variable selection

Abstract

We consider the problem of variable selection in the context of sequential (or variable-by-variable) imputation in clustered data. Specifically, we modify the sequential hierarchical regression imputation technique to incorporate variable selection routines using spike-and-slab priors within the Bayesian variable selection routine. Specific choice of these priors allow us to “force” variables of importance (e.g. design variables or variables known to play role in missingness mechanism) into the imputation models. Our ultimate goal is to improve computational speed by removing unnecessary variables. We employ Markov chain Monte Carlo techniques to sample from the implied posterior distributions for model unknowns as well as missing data. We assess the performance of our proposed methodology via simulation study. Our results show that our proposed algorithms lead to satisfactory estimates and in, some instances, outperform some of the existed methods that are available to practitioners.