论文部分内容阅读
目的探讨多水平模型技术在Ｍｅｔａ分析中的应用价值。方法根据Ｍｅｔａ分析只能得到研究水平数据的特点，探讨了两水平随机效应与混合效应模型的建模方法。结果获得平均“效应尺度”及其标准误的估计，以及与研究结果差异有关因素的估计。结论在Ｍｅｔａ分析中，多水平统计模型不仅可估计平均“效应尺度”及其可信区间，而且可分析与研究结果差异有关的影响因素。
Objective To explore the application value of multi-level model technology in meta-analysis. Methods According to the meta-analysis, only the characteristics of the research level data can be obtained, and the modeling methods of two-level random effects and mixed effects models are discussed. Results An estimate of the mean “effect scale” and its standard error, as well as an estimate of the factors associated with the difference in the results of the study were obtained. Conclusion In meta-analysis, the multi-level statistical model can not only estimate the mean “effect scale” and its confidence interval, but also analyze the influencing factors related to the differences in the research results.