离散型广义非线性纵向数据模型中偏离名义离差的检验及其功效模拟(英文)  

Testing for Departures from Nominal Dispersion and Power Simulations in Discrete Generalized Nonlinear Modelswith Longitudinal Data

作　　者：林金官[1] 韦博成[1] 傅珏生[2] 

机构地区：[1]东南大学数学系 [2]苏州大学数学科学学院,江苏苏州215006

出　　处：《应用数学》2006年第2期342-347,共6页Mathematica Applicata

基　　金：SupportedbyNSFC(10371016),SSFC(04BTJ002),NSSPFC(LXZ0415)andthePostDoctorialGrantinSoutheastUniversity

摘　　要：离散型广义非线性模型包括Poisson,二项,负二项模型.本文讨论离散型广义非线性纵向数据模型中偏离名义离差的检验问题,得到了检验的score统计量,并利用MonteCarlo方法研究了检验统计量的性质.最后,利用杀虫剂数据说明了检验方法的应用.Discrete exponential family nonlinear models include Poisson, binomial, and negative binomial models. This paper discusses the test for departures from nominal dispersion in the framework of discrete generalized nonlinear models with random effects for longitudinal data. The score test statistic is constructed, The properties of test statistic are investigated through Monte Carlo simulations. We illustrate our test method by the insecticide data.

关 键 词：纵向数据 广义非线性模型 随机效应 名义离差 SCORE检验 模拟功效 

分 类 号：O212.2[理学—概率论与数理统计]