小样本下岭PLS-SEM与岭CB-SEM的比较  

Comparison of Ridge LS-SEM and Ridge B-SEM Under Small Samples

作　　者：王新芸 袁克海 唐加山[1] 温勇[1] Wang Xinyun;Yuan Kehai;Tang Jiashan;Wen Yong(College of Science,Nanjing University of Posts and Telecommunications,Nanjing 210000,China)

机构地区：[1]南京邮电大学理学院,南京210000

出　　处：《统计与决策》2024年第10期46-51,共6页Statistics & Decision

基　　金：国家自然科学基金资助项目(31971029)。

摘　　要：目前主要有两种结构方程模型(SEM):CB-SEM和PLS-SEM。当样本量较小时,CB-SEM常常会出现不收敛的情况,使用岭方法可以改善这个问题。文章主要研究将岭方法运用到PLS-SEM中,对比岭方法下PLS-SEM与CB-SEM的表现。研究表明,岭CB-SEM和岭PLS-SEM估计量总体上比常规的CB-SEM和PLS-SEM估计量更精确,但偏差没有明显改善;当样本量较小时,PLS-SEM、岭PLS-SEM估计的精确性都优于CB-SEM、岭CB-SEM;对于受到中介变量影响的内生潜变量来说,岭PLS-SEM在估计其他潜变量对它的影响(路径系数)时最精确。There are two main structural equation modeling:CB-SEM(covariance-based SEM)and PLS-SEM(partial least squares SEM).When the sample size is small,the CB-SEM method often fails to converge,which can be improved by using ridge method.This paper mainly studies the application of ridge method to PLS-SEM,and compares the performance of PLS-SEM and CB-SEM under ridge method.The research shows that CB-SEM and PLS-SEM estimators are generally more accurate than con-ventional CB-SEM and PLS-SEM estimators,but the bias has not been significantly improved.The estimation accuracy of PLS-SEM and ridge PLS-SEM is better than that of CB-SEM and ridge CB-SEM under small samples.For endogenous latent variables affected by intermediate variables,ridge PLS-SEM is the most accurate in estimating the influence(path coefficient)of other latent variables.

关 键 词：结构方程模型 岭方法 岭CB-SEM PLS-SEM 蒙特卡洛模拟 

分 类 号：F064.1[经济管理—政治经济学]