No abstract

No abstract

In this article, the generalized linear model for longitudinal data is studied. A generalized empirical likelihood method is proposed by combining generalized estimating equations and quadratic inference functions based on the working correlation matrix. It is proved that the proposed generalized empirical likelihood ratios are asymptotically chi-squared under some suitable conditions, and hence it can be used to construct the confidence regions of the parameters. In addition, the maximum empirical likelihood estimates of parameters are obtained, and their asymptotic normalities are proved. Some simulations are undertaken to compare the generalized empirical likelihood and normal approximation-based method in terms of coverage accuracies and average areas/lengths of confidence regions/intervals. An example of a real data is used for illustrating our methods.

a b s t r a c tIn this paper, we consider the problem of variable selection for varying coefficient partially linear models with longitudinal data. A new variable selection procedure is proposed based on smooth-threshold generalized estimating equation (SGEE). The proposed procedure simultaneously selects significant variables in the parametric components and the nonparametric components. The approach avoids the convex optimization problem and is flexible and easy to implement. The consistency and asymptotic normality of the resulting estimators are established. Extensive Monte Carlo simulation studies are conducted to examine the finite sample performance of the proposed variable selection procedure. The proposed procedure is further illustrated by an application.

No abstract