多水平模型在分层抽样研究中的优越性

目的探讨在分层抽样横断面调查研究资料中多水平模型相对于传统模型的优越性。方法分别采用普通logistic回归模型和多水平logistic回归模型分析"第2次全国人体重要寄生虫病现状调查"中的华支睾吸虫病流行区调查中的广东省数据。结果在传统logistic回归分析中差异无统计学意义的年龄组,在多水平logistic回归分析中差异有统计学意义。结论与传统logistic回归模型相比,多水平模型在处理层次结构的数据时考虑到数据误差的层次性,可以有效地分析具有层次结构的数据。

Superiority of the Multilevel Model in Stratified Sampling Research

Objective To explore the superiority of the multilevel model compared with the traditional model in cross sectional survey research data by stratified sampling. Methods Using the data of Guangdong in “the Survey of the Second National Important Parasitic Diseases”, the paper analyzing the factors of Clonorchis sinensis disease infection with traditional logistic regression model and multilevel logistic regression model separately. Results The age group is not significant in traditional logistic regression model but becomes significant in multilevel logistic regression model. Conclusion Compared to the traditional logistic regression model,multilevel models can efficiently analyze cross-sectional study data, according to error from different levels while analyzing hierarchical structure data.