纵向多分类数据的广义估计方程分析

Generalized Estimating Equations Analysis of Longitudinal Multi-categorical Data

广义估计方程(GEE)是分析纵向数据的常用方法.如果响应变量的维数是一,XIE和YANG(2003)及WANG(2011)分别研究了协变量维数是固定的和协变量维数趋于无穷时,GEE估计的渐近性质.本文研究纵向多分类数据(multicategorical data)的GEE建模和GEE估计的渐近性质.当数据的分类数大于二时,响应变量的维数大于一,所以推广了文献的相关结果.

Generalized estimating equation is widely adopted in analyzing longitudinal categorical data,and the large sample properties of the corresponding unknown parameter estimator have been established for the repeated univariate responses.Longitudinal multi-categorical responses commonly arise in biomedical and social science studies,but the corresponding large sample properties have not been established in literature,the aim of this paper is to fill in this gap.