双向分类随机效应模型中方差分量的估计

New Estimation of Variance Components in Two-way Classification Model with Random Effects

双向分类随机效应模型是一类有着广泛应用背景的统计模型,其中对模型中方差参数的一种重要估计方法是方差分析估计。由于方差分析法得到估计的均方误差(MSE)并不是最小的,本文在一类新的估计族中提出了改进的ANOVA估计。结果表明新的估计比ANOVA具有较小的MSE,这种新的估计方法可推广到医学领域中常见的一般模型。

That two-way classification model with random effects is used widely in practice,one of the most important estimation methods for this model is analysis of variance estimate （ANOVA）. Notice that the mean square error （MSE） of the ANOVA is not the smallest,we propose an improved ANOVA based on new estimation family and prove that the MSE of the new estimator is smaller than that of ANOVA. The estimation method is extended to general models,which are used mainly in the field of medicine.