优势比下带有测量误差的非线性因果中介模型的效应估计

Effect Estimation of Nonlinear Causal Mediation Model with Measurement Error under the Odds Ratio

因果中介分析用于研究自变量是否通过中介变量对因变量产生影响,阐明潜在因果机制。本文研究当中介变量具有测量误差时,利用优势比方法定义中介效应,对无交互作用和有交互作用分别建立逻辑回归和线性回归模型的结构方程,获得自然直接效应和间接效应的解析表达式。

ausal mediation analysis is used to study whether independent variables have effects on dependent variables through mediation variables and clarify the potential causal mechanism.In this paper,when the intermediate variable has measurement error,the odds ratio method is used to define the intermediate effect.For the two cases of noninteraction and interaction,the structural equations of logical regression and linear regression models are established respectively,and the closed-form expressions of natural direct effect and indirect effect are obtained.