论文部分内容阅读
纵向研究可用于精神卫生及其服务领域的科研中。纵向数据分析的主要方法是广义线性混合效应模型(GLMM)和加权广义估计方程(WGEE)。虽然这两个模型已被广泛应用,也有大量文献发表,但是人们并没有清晰地描述这些方法间的差别以及方法本身的局限性,缺少相关的文献记录。遗憾的是,有些差别和局限性会明显影响对研究结果的报告、比较和解释。本文回顾了纵向数据分析的两种主要方法,强调两者的相似之处和主要差别。我们使用真实数据和模拟数据着重比较这两类模型的假设、对参数的解释、适用性和局限性,并讨论了将这两种不同的方法用于真实数据研究时的注意事项,提出了相关的警示。
Longitudinal research can be used in research in mental health and its services. The main methods of longitudinal data analysis are Generalized Linear Mixed Effects Model (GLMM) and Weighted Generalized Estimation Equation (WGEE). Although these two models have been widely used, there are a lot of literature published, but people do not clearly describe the differences between these methods and the limitations of the method itself, the lack of relevant documentation. Unfortunately, some differences and limitations can significantly affect the reporting, comparison and interpretation of research findings. This article reviews two main approaches to longitudinal data analysis, emphasizing the similarities and major differences between the two. We use real data and simulation data to emphasize the hypothesis of the two types of models, explain the parameters, applicability and limitations, and discuss the precautions when using these two different methods in real data research. Warning.