高级检索
刘巧兰, 李晓松, 倪宗瓒. Meta分析中多变量多水平模型方法的探讨[J]. 中国公共卫生, 1999, 15(10): 952-953.
引用本文: 刘巧兰, 李晓松, 倪宗瓒. Meta分析中多变量多水平模型方法的探讨[J]. 中国公共卫生, 1999, 15(10): 952-953.
Liu Qiaolan, Li xiaosong, Ni Zongzan. Exploring the approaches of multivariate multilevel model to Meta-analysis[J]. Chinese Journal of Public Health, 1999, 15(10): 952-953.
Citation: Liu Qiaolan, Li xiaosong, Ni Zongzan. Exploring the approaches of multivariate multilevel model to Meta-analysis[J]. Chinese Journal of Public Health, 1999, 15(10): 952-953.

Meta分析中多变量多水平模型方法的探讨

Exploring the approaches of multivariate multilevel model to Meta-analysis

  • 摘要: 为了探讨多变量多水平模型在Meta 分析中的应用价值, 根据Meta 分析中具有两个/ 效应尺度0 的数据, 拟合双变量两水平模型, 估计有关参数以及预测值。结果表明, 在模型中可同时获得两个平均/ 效应尺度0 及其相关系数的估计, 以及其影响因素的估计。Meta 分析中如果具有不止一个/ 效应尺度0, 则多变量多水平模型可获得多个平均/ 效应尺度0 的估计值及其可信区间, 也可分析多个平均/ 效应尺度0 估计值之间的相关关系,探讨有关因素对这一关系的影响。

     

    Abstract: Objective:to explore the applicability of the multivariate multilevel model to Meta Analysis;Methods:focusing on the property of some Meta analysis data with two effect magnitudes,to fit two-level model with two-responses to estimate parameters and to make prediction;Results:with the fitted model,estimates of two mean effect magnitudes with their correlation,and affecting factors were obtained simultaneously;Conclusion:when more than one effect magnitudes exist in data of Meta-analysis,multivariate multilevel model should be employed to estimate the values of all the mean effect magnitudes and their corresponding confidence intervals.Moreover,multivariate multilevel model allows the analysis of correlation among mean effect magnitudes,and the exploration of affecting factors,to be conducted.

     

/

返回文章
返回