The effect of misspecifying the random-effects distribution in linear mixed models for longitudinal data
The effect of misspecifying the random-effects distribution in linear mixed models for longitudinal data
复制标题
DOI：
10.1016/s0167-9473(96)00047-3
复制
发表时间：
1997-02-06
影响因子：
1.8
通讯作者：
Lesaffre, E
中科院分区：
文献类型：
--
作者：
Verbeke, G;Lesaffre, E
在随机效应正态分布的假设下，线性混合模型中固定效应和方差分量的极大似然估计是一致的，并且是渐近正态分布的，即使随机效应分布不是正态分布。然而，为了得到正确的渐近协方差矩阵，需要对逆Fisher信息矩阵进行三明治式校正。大量的模拟表明，即使在中等样本中，得到的修正标准误差也明显优于未经修正的标准误差，特别是对于随机效应协方差矩阵中的参数。
Maximum likelihood estimators for fixed effects and variance components in linear mixed models, obtained under the assumption of normally distributed random effects, are shown to be consistent and asymptotically normally distributed, even when the random-effects distribution is not normal. However, a sandwich-type correction to the inverse Fisher information matrix is then needed in order to get the correct asymptotic covariance matrix. Extensive simulations show that the so-obtained corrected standard errors are clearly superior to the naive uncorrected ones, especially for the parameters in the random-effects covariance matrix, even in moderate samples.