摘要
目的探索重复测量0过多计数资料的统计模型。方法将重复测量中的个体当作随机效应引进ZIP模型,建立随机效应ZIP模型。结果模拟研究和实例分析表明,由于随机效应ZIP模型既考虑了个体不同测量值的相关性,又考虑了0过多的问题,结果优于随机效应Poisson回归和固定效应ZIP模型。结论在重复测量计数资料0过多时,宜采用随机效应ZIP模型。
Objective To explore the statistical model for repeated measured count data with extra zero. Methods Introduce the individual of repeated measurement into zero-inflated Poisson as a random effect,Build up random effect zero-inflated Poisson model. Results The simulation study and example analysis indicate that because the random effect zero-inflated Poisson takes the correlation of individuals and extra zero into account,the results are better than that of mixed Poisson and ZIP.Conclusion It could model the repeated measured count data with extra zero using random effect zero-inflated Poisson model.
出处
《中国卫生统计》
CSCD
北大核心
2009年第4期344-346,349,共4页
Chinese Journal of Health Statistics
基金
社区卫生服务综合评价指标体系研究(200558)
山西省自然基金(200901105-2)
山西科技创新基金(01200822)
