A score test for zero-inflation in multilevel count data

被引:23
作者
Moghimbeigi, Abbas [1 ,2 ]
Eshraghian, Mohammad Reza [3 ]
Mohammad, Kazem [3 ]
McArdle, Brian [4 ]
机构
[1] Hamadan Univ Med Sci, Dept Biostat & Epidemiol, Sch Publ Hlth, Hamadan, Iran
[2] Hamadan Univ Med Sci, Ctr Hlth Res, Hamadan, Iran
[3] Univ Tehran Med Sci, Dept Epidemiol & Biostat, Tehran, Iran
[4] Univ Auckland, Dept Stat, Auckland 1, New Zealand
关键词
POISSON REGRESSION; BINOMIAL REGRESSION; MODELS;
D O I
10.1016/j.csda.2008.10.041
中图分类号
TP39 [计算机的应用];
学科分类号
081203 ; 0835 ;
摘要
The zero-inflated Poisson regression (ZIP) in many situations is appropriate for analyzing multilevel correlated count data with excess zeros. In this paper, a score test for assessing ZIP regression against Poisson regression in multilevel count data with excess zeros is developed. The sampling distribution and power of the score statistic test is evaluated using a simulation study. The results show that under a wide range of conditions, the score statistic performs satisfactorily. Finally, the use of the score test is illustrated on DMFT index data of children 7-8 years old. (C) 2008 Elsevier B.V. All rights reserved.
引用
收藏
页码:1239 / 1248
页数:10
相关论文
共 25 条