无重复因析试验中一种散度效应的估计方法  

A Estimator of Dispersion Effects in Unreplicated Factorial Experiments

作　　者：王钰[1] 李济洪[1] 

机构地区：[1]山西大学计算中心,山西太原030006

出　　处：《山西大学学报（自然科学版）》2010年第2期186-189,共4页Journal of Shanxi University(Natural Science Edition)

基　　金：国家自然科学基金(60873128)

摘　　要：在无重复因析试验的多个散度效应分析中,现有的许多方法都存在错误识别的现象,即两个显著的散度效应可能在它们的交互列上产生一个错误的(spurious)散度效应.为了解决这种模棱两可性,文章提出了一种基于闭的位置效应集合残差的改进H方法(称为AH方法),证明了AH的估计的无偏性,并通过一个基于实例的模拟验证了此方法.In the analysis of multiple dispersion effects for unreplicated factorial experiments,there often exists the phenomenon for picking up factors spuriously, that is, two active dispersion effects may create a spurious dispersion effect in their interaction column, and the most existing methods are subject to these spurious effects. To resolve the ambiguousness,a adapted H method（called the AH method） was introduced based on residuals from the fitted closed set of location effects. The unbiased condition of the AH estimator was proved,and the simulations based on a real example were used to illustrate the results.

关 键 词：多个散度效应 无重复因析 无偏性 模拟 

分 类 号：O212.1[理学—概率论与数理统计]