Semiparametric estimation in regression with missing covariates using single-index models (Q2330532)

This paper proposes an unweighted mean-score-form estimator for the coefficients of a regression model with the values of some covariates missing at random. The estimation procedure employs a kernel-assisted estimator for the augmentation of the generalized estimating equations by a single-index model. Thus, it avoids the inverse of selection probabilities of small value that may occur in the augmented inverse probability weighted approach. Demonstrations are given that under certain conditions the proposed estimator is as efficient as the existing methods based on standard kernel smoothing. A simulation study and an application to real data of Canada 2010/2011 Youth Smoking Survey (YSS) are presented to illustrate the proposed method.