The application of the empirical likelihood method to a generalized linear model with missing responses at random is considered. Firstly, a class of empirical likelihood ratios for the unknown parameter are defined such that any ratio in the class is asymptotically chi-squared. Secondly, a class of estimators for the parameter are constructed and the asymptotic distributions of the proposed estimators are obtained. Our results can be used directly to construct confidence intervals and regions for the parameters of interest. Finally, some simulations are conducted to illustrate the proposed methods.

empirical likelihood; generalized linear models; missing data; confidence region