Robust empirical likelihood inference for longitudinal data

This paper introduces the robust empirical likelihood (REL) inference for the longitudinal data. We propose the REL method by constructing robust auxiliary random vectors, and employ bounded scores and leverage-based weights in the auxiliary random vectors to achieve robustness against outliers in both the response and covariates. Simulation studies are conducted to demonstrate the good performance of our proposed REL method in terms of both robustness and efficiency improvement. The proposed method is also illustrated by analyzing a real data set from epileptic seizure study.