GENERALIZED PARTIAL LINEAR MODEL WITH UNKNOWN LINK AND UNKNOWN BASELINE FUNCTIONS FOR LONGITUDINAL DATA

In this paper we develop a generalized partial linear model for longitudinal data. In the model, we allow the link and baseline functions to be unknown. We explicitly express the estimators of regression parameters and the baseline function; hence, the computation and programming of our estimators are simple. We show that the proposed estimators of regression parameters and the baseline function are asymptotically normal with a simple variance estimator for the baseline function. In simulation studies, we demonstrate that the proposed nonparametric method is robust with limited loss of efficiency.