不平衡两因素套设计模型中方差分量比的区间估计  

CONFIDENCE INTERVALS ON RATIOS OF VARIANCE COMPONENTS IN UNBALANCED TWO-FOLD NESTED DESIGNS

作　　者：李新民[1] 

机构地区：[1]山东理工大学理学院统计系,淄博255049

出　　处：《系统科学与数学》2010年第1期72-78,共7页Journal of Systems Science and Mathematical Sciences

基　　金：国家自然科学基金(10771126);山东理工大学科研基金(2006KJZ01)资助

摘　　要：给出了次样本容量相等时,不平衡两因素套设计模型中方差分量之比的广义区间估计.为了研究所得区间估计的优良性,并与Burdick,Birch and Graybill给出的修正大样本(MLS)方法进行对比,进行了模拟研究和分析.结果表明所得方法优于MLS方法,特别是在极端不平衡情形.In the paper, the unbalanced case of the two-way nested designs with equal subsampling is considered. Using the method of generalized confidence intervals introduced in Weerakandi（1993, 1995）, a new method is proposed for constructing confidence intervals on ratios of variance components. To compare the resulted intervals with the Modified Large Sample （MLS） intervals by Burdick, Birch and Graybill （1986）, a simulation study is conducted. The results indicate that the proposed method is better than the MLS method, especially for very unbalanced designs.

关 键 词：广义置信区间 覆盖率 方差分量 

分 类 号：O212.1[理学—概率论与数理统计]