Empirical likelihood for quantile regression models with longitudinal data |
| |
Authors: | Huixia Judy Wang Zhongyi Zhu |
| |
Institution: | 1. Department of Statistics, North Carolina State University, Raleigh NC 27695, USA;2. Department of Statistics, Fudan University, Shanghai 200433, China |
| |
Abstract: | We develop two empirical likelihood-based inference procedures for longitudinal data under the framework of quantile regression. The proposed methods avoid estimating the unknown error density function and the intra-subject correlation involved in the asymptotic covariance matrix of the quantile estimators. By appropriately smoothing the quantile score function, the empirical likelihood approach is shown to have a higher-order accuracy through the Bartlett correction. The proposed methods exhibit finite-sample advantages over the normal approximation-based and bootstrap methods in a simulation study and the analysis of a longitudinal ophthalmology data set. |
| |
Keywords: | |
本文献已被 ScienceDirect 等数据库收录! |
|