Likelihood-Based Influence Analysis in Linear Mixed Modelsfor Longitudinal Data
基于似然函数的纵向数据线性混合模型影响分析

Linear mixed models (LMM) have been widely used inlongitudinal studies. This paper is devoted to discussion ofinfluence analysis in LMM in terms of case-deletion method. Theinfluence measure employed here is a generalized Cook's distance whichis based on likelihood. Two generalized Cook's distances arepresented to detect the influential individual. Realdata analysis and simulation study show that our method is efficient for influence analysis.