广义p值与多元随机效应模型的精确检验  

Generalized p value and exact test in multivariate random effect model

作　　者：杜岩 范永辉[1] DU Yan;FAN Yong-hui(School of Mathematical Science,Tianjin Normal University,Tianjin 300387,China)

机构地区：[1]天津师范大学数学科学学院

出　　处：《哈尔滨商业大学学报（自然科学版）》2019年第6期740-744,共5页Journal of Harbin University of Commerce：Natural Sciences Edition

摘　　要：在多元随机效应模型中,一般采用似然比检验或者其他近似方法检验关于回归系数矩阵的统计假设.这些方法都不是精确的检验.为了得到一种精确的检验,方法是把广义p值应用到多元随机效应模型中.模拟结果显示,在小样本情况下,广义检验可以控制检验水平,且比似然比方法有更好的检验效果.In multivariate random effect model,likelihood ratio test or other approximation methods were generally used to test statistical hypothesis about regression coefficient matrix.None of these tests are accurate.In order to construct an exact test,the method was to apply the generalized p value to the multivariate random effect model.Simulation showed that in the case of small sample size,the generalized test can control the test level,and has better test effect than likelihood method.

关 键 词：多元随机效应模型 似然比检验 回归系数矩阵 精确检验 广义P值 模拟结果 

分 类 号：O212.6[理学—概率论与数理统计]