Variable selection for longitudinal survey data

TL;DR

This paper introduces a new variable selection method for longitudinal survey data using a survey-weighted quadratic inference approach, providing theoretical guarantees and demonstrating finite sample performance through simulations.

Contribution

It presents a novel penalized survey-weighted quadratic inference estimator with proven properties under the joint model-design framework.

Findings

Estimator exists under certain conditions

Estimator is weakly consistent and asymptotically normal

Simulation study shows good finite sample performance

Abstract

In this article we propose a new variable selection method for analyzing data collected from longitudinal sample surveys. The procedure is based on the survey-weighted quadratic inference function, which was recently introduced as an alternative to the survey-weighted generalized estimating function. Under the joint model-design framework, we introduce the penalized survey-weighted quadratic inference estimator and obtain sufficient conditions for the existence, weak consistency, sparsity and asymptotic normality. To illustrate the finite sample performance of the model selection procedure, we include a limited simulation study.