广义线性混合模型.pdf

第32卷第4期 昆明理工大学学报(理工版) V01．32No．4 2007年8月Journalof ofScienceand and KunmingUniversity Technology(ScienceTechnology)Aug．2007 dezGeneralizedilareneLinearGsleMixeddodMModels GAN Xiao—rong of ofScienceand (FacultyScience，KmmaingUniversity Technology，Kunming650093，China) oftheadvancedstatisticalmodelsiS LinearMixedModels Abstract：One studied：Generalized useoftheGLMMismuch data encoun- (GLMM)．rIIle wider，fromlongitudinal／clusteredmodeling tered inrandomizedclinicaltrialsto data oftenin and commonly spatialmodelingarising epidemiology estimation for the offixedeffectsand a鲥culture．Theprocedureobtainingpenalizedquasi—likelihood variance intheGLMMis useofthefunction the statistical components presented．，I’}le GLMMPQL()in Risdemonstrated arealdataset ina clinicaltrial．Itisconcluded package throughanalyzing arising thattheGLMMisa statisticalmodelfor clusterednon—·Normalandnon——linearda·· powerful modeling ta． Keywords：fixed linearmixed effects；generalizedmodels；variancecomponents Document Article CLC：0212 code：A ID：1007—855X(2007)04—0107—07 广义线性混合模型 干晓 蓉 (昆明理工大学理学院．云南昆明650093) 摘要：探讨了现代统计学的重要模型：广义线性混合模型．该模型的用途极为广泛，例如可用于临床学中 的纵向数据分析及流行病和农业中的空间数据分析．讨论了该模型中参数的估计方法．并用统计软件R 展示了怎样用广义线性混合模型去分析实际数据．其演示表明广义线性混合模型是一种处