基于正则化的探索性中介分析:原理与应用

探索性中介分析被定义为从变量集合中筛选潜在中介变量的方法,该方法能在缺乏理论基础的情况下帮助研究者从数据中挖掘潜在中介机制,提供模型构建上的指导。本文介绍了一种基于正则化的探索性中介分析方法XMed(exploratory mediation analysis via regularization)。相比于传统探索性中介分析方法,XMed具有检验力更高、所需样本量更小、能高效地处理高维数据等优点,在认知神经科学、临床心理学等心理学领域有较大的应用潜力。本文主要介绍XMed的原理和实现过程,并通过实例分析展示该方法的应用。

The Principle and Application of Exploratory Mediation Analysis via Regularization

Mediation analysis is common in social science.Exploratory mediation analysis is defined as a series of data-driven methods for identifying potential mediators from a set of variables.It offers insights into the potential mediation process from data and provides guidance on model building.This article introduces the approach of exploratory mediation analysis via regularization(XMed).Compared to conventional exploratory mediation analysis approaches,XMed has higher sensitivity and needs less sample size.Moreover,it can handle high-dimensional data efficiently,which endows XMed a great potential for application in fields including cognitive neuroscience and clinical psychology.This article focuses on the principle and implementation of XMed.An empirical analysis is included to demonstrate the application of XMed.