非线性随机效应模型的异方差性检验  

TESTING FOR HETEROGENEITY IN NONLINEAR MODELSWITH RANDOM EFFECTS

作　　者：林金官[1] 韦博成[2] 

机构地区：[1]江苏教育学院数学系,南京210013 [2]东南大学应用数学系,南京210096

出　　处：《系统科学与数学》2002年第2期245-256,共12页Journal of Systems Science and Mathematical Sciences

基　　金：国家自然科学基金资助课题

摘　　要：随机效应模型广泛应用于刻画重复测量数据的特征．在该模型中,随机误差的方差包括受试群体内部及受试群体之间两项方差．Ｚｈａｎｇ和 Ｗｅｉｓｓ ２０００年研究了线性随机效应模型的异方差检验,本文对非线性随机效应模型,分别讨论了群体内、群体间和多变量的异方差性的检验问题,得到了检验的ｓｃｏｒｅ统计量,并讨论了三种情形下,相应的ｓｃｏｒｅ函数之间的关系．最后给出一个数值例子说明上述方法的有用性．Abstract Random-effects models are widely used to characterize repeated measurementdata. Random-effect models for longitudinal data include both within-and betweensubjectvariances. Zhang and Weiss (2000) studied heterogeneity tests for linear models with randomeffects. In this paper, three types of heterogeneity of variance are discussed in nonlinear modelswith random effects. They are variance heterogeneity of within, between and within-betweenmultivariate subjects. Several score test statistics are obtained and the relations of three scorefunctions are presented. A numerical example is given to illustnde the methodology

关 键 词：非线性模型 随机效应 异方差性 似然比检验 SCORE检验 score函数 

分 类 号：O211[理学—概率论与数理统计]