数学物理学报 ›› 2013, Vol. 33 ›› Issue (3): 523-534.

• 论文 • 上一篇    下一篇

基于M估计的线性混合效应模型自相关和随机效应检验

孙慧慧|林金官   

  1. 盐城师范学院 数学科学学院 江苏 盐城 |224002;东南大学 数学系 南京 |210096
  • 收稿日期:2011-08-26 修回日期:2012-11-18 出版日期:2013-06-25 发布日期:2013-06-25
  • 基金资助:

    国家自然科学基金(NSFC11171065, 11202180)、江苏省自然科学基金(NSFJSBK2011058)和盐城师范学院自然科学基金(11YCKL019)资助

Testing for Autocorrelation and Random Effects in Mixed Effect Linear Models Based on M-estimation

 SUN Hui-Hui, LIN Jin-Guan   

  1. School of Mathematics and Science, Yancheng Teachers University, Jiangsu Yancheng 224002; Department of Mathematics, Southeast University, Nanjing 210096
  • Received:2011-08-26 Revised:2012-11-18 Online:2013-06-25 Published:2013-06-25
  • Supported by:

    国家自然科学基金(NSFC11171065, 11202180)、江苏省自然科学基金(NSFJSBK2011058)和盐城师范学院自然科学基金(11YCKL019)资助

摘要:

对纵向数据的线性混合效应模型, 用Fisher得分法得到了参数的M估计(稳健估计), 给出了其渐近性质, 研究了M估计下模型的自相关性和随机效应存在性检验问题, 得到了检验的Score统计量, 并通过葡萄糖数据的实例进行了分析论证.

关键词: 纵向数据, 线性混合效应模型, M估计, 自相关性, 随机效应, Score检验

Abstract:

In this paper, the Fisher scoring method is applied to get M-estimator(robust estimator)in the mixed effects linear model for longitudinal data. And its asymptotic property is given later. Then tests for autocorrelation and presence of random
effects in the models are studied based on M-estimation. The score test statistics are obtained and illustrated by the grape sugar data example.

Key words: Longitudinal data, Mixed effect linear models, M-estimation, Autocorrelation, Random effects, Score test

中图分类号: 

  • 62J20