Semiparametric Approach for Handling Partially Missing Covariates in the Linear Transformation Model

• Activity Number: 321
• Type: Topic Contributed
• Date/Time: Tuesday, August 5, 2014 : 10:30 AM to 12:20 PM
• Sponsor: Committee on Applied Statisticians
• Author(s): Samiran Sinha*+
• Companies: Texas A&M
• Keywords: estimating equations ; linear transformation models ; multiple imputation ; non-monotone missing data

Abstract

Missing covariate data often arise in clinical studies, and analysis of the such data ignoring the subjects with incomplete information may lead to inefficient and possibly biased estimates. Although there are many methods for handling missing covariates in the Cox proportional hazard model, there is a need of an easy to use method for handling multiple missing covariates in the Cox proportional hazard model and beyond. I will talk about a method for handling missing covariates in the Linear transformation model. The novelty of the proposed method lies in the modelling of the partially observed covariates distribution through a dimension reduction technique, and estimation of the finite and infinite dimensional parameters of the linear transformation model by solving a set of Martingale estimating equations. I will outline the distributional property of the proposed estimator for large sample. I will talk about a simulation study to assess the finite sample performance of the method. Finally, for the purpose of illustration, I will talk about the analysis of the SEER data on breast cancer.