Fisher Paige, Cao Wentao, Yu Qingzhao
Louisiana State University Health Sciences Center, New Orleans, US.
J Open Res Softw. 2020;8(1). doi: 10.5334/jors.277. Epub 2020 Oct 7.
Mediation analysis refers to the process of making inferences on effects of third variables that intervene in the relationship between an exposure and response variable. The relationships among variables can be modelled by generalized linear models (GLM). However, GLM are not sufficient to describe relationships among variables when there are nonlinear relationships and potential interaction effects. A general mediation analysis method was developed using not only GLMs, but also multiple additive regression trees and smoothing splines by Yu and Li (2017). The method is implemented in the R package, . In this paper, we developed SAS macros so that functions in the mma package can be called and the mediation analysis performed in the SAS environment.
中介分析是指对干预暴露变量和反应变量之间关系的第三个变量的效应进行推断的过程。变量之间的关系可以用广义线性模型(GLM)进行建模。然而,当存在非线性关系和潜在的交互效应时,GLM不足以描述变量之间的关系。Yu和Li(2017)开发了一种通用的中介分析方法,该方法不仅使用了GLM,还使用了多元加法回归树和平滑样条。该方法在R包中实现。在本文中,我们开发了SAS宏,以便可以调用mma包中的函数,并在SAS环境中进行中介分析。