摘要
结构方程模型虽已被广泛应用于社会科学实证研究,但其方法本身仍处于发展过程中。因PLSSEM估计程序被认为具有较好的"普适性"而逐渐流行于市场研究中,在关注结构方程模型应用的几个重要问题的基础上,比较了CB-SEM与PLS-SEM两类估计程序的主要差别;以Wheaton等人研究使用的模型为正确模型,设计了三种模型设定,又在随机模拟的基础上,讨论了在两类估计程序下拟合指标和估计系数对模型结构变化的不同反应,并给出一定的相关建议。
Structure equations modeling has been widely applied in the empirical research onsocial studies, whereas it is far from maturity. PLS--SEM procedure has been considered as a silver bullet and been warmly accepted in marketing research. Author pays attention to some arguable problems about SEM methodology and compare PLS--SEM to CB--SEM, and then discuss their distinct performances on structure alternatives with simulation. Conclusions and suggestions have been finally presented.
出处
《统计与信息论坛》
CSSCI
北大核心
2015年第7期7-15,共9页
Journal of Statistics and Information
基金
国家社会科学基金项目<我国科技政策分析的数据与方法研究>(14BTJ020)
