A strictly positive estimator of intra-cluster correlation for the one-way random eects model
A strictly positive estimator of intra-cluster correlation for the one-way random eects model
复制标题
单向随机效应模型的簇内相关性的严格正估计量
DOI：
--
复制
发表时间：
2011
期刊：
影响因子：
--
通讯作者：
P. Lahiri
中科院分区：
文献类型：
--
作者：
S. Gabler;M. Ganninger;P. Lahiri
集群内相关性的估计通常用于设计和分析大型跨国抽样调查，如欧洲社会调查（见Ganninger 2010：19）。Kish（1962）将群内相关性定义为群间方差与总方差的比值。因此，根据Kish的定义，集群内相关性是一个严格的正参数。众所周知（Wang et al. 1991），标准方差分量方法（如ANOVA方法）经常会产生负估计值，特别是当真正的聚类内相关性和聚类数很小时，这种情况在实践中可能会出现（Killip et al. 2004）。这个问题的一个标准解决方案是将簇内相关性估计值截断为0。（坎贝尔et al. 2005）.本文给出了单向随机效应模型的簇内相关性的一个新的估计，并证明了它是严格正的.我们比较了提出的估计与方差分析估计使用蒙特卡罗模拟研究。
Estimates of intra-cluster correlations are routinely produced for designing and analyses of large cross-nationalsample surveys like the European Social Survey (see Ganninger 2010:19). Kish (1962) de ned intra-clustercorrelation as the ratio of the between cluster variance to the total variance. Thus, according to Kish’s de nition,intra-cluster correlation is a strictly positive parameter. It is well-known (Wang et al. 1991) that standardvariance component methods, such as the ANOVA method, can frequently produce negative estimates, especiallywhen the true intra-cluster correlation and the number of clusters are small, a situation that can arise in practice(Killip et al. 2004). A standard solution to this problem is to truncate the intra-cluster correlation estimate to0. (Campbell et al. 2005).In this paper, we present a new estimator of the intra-cluster correlation for the one-way random e ects modeland prove that it is strictly positive. We compare the proposed estimator with the ANOVA estimator using aMonte Carlo simulation study.