Leth-Steensen Craig, Gallitto Elena
Carleton University, Ottawa, Ontario, Canada.
University of Ottawa, Ottawa, Ontario, Canada.
Educ Psychol Meas. 2016 Apr;76(2):339-351. doi: 10.1177/0013164415593777. Epub 2015 Jul 7.
A large number of approaches have been proposed for estimating and testing the significance of indirect effects in mediation models. In this study, four sets of Monte Carlo simulations involving full latent variable structural equation models were run in order to contrast the effectiveness of the currently popular bias-corrected bootstrapping approach with the simple test of joint significance approach. The results from these simulations demonstrate that the test of joint significance had more power than bias-corrected bootstrapping and also yielded more reasonable Type I errors.
为了估计和检验中介模型中间接效应的显著性,人们已经提出了大量方法。在本研究中,进行了四组涉及完全潜变量结构方程模型的蒙特卡罗模拟,以对比当前流行的偏差校正自助法与联合显著性简单检验法的有效性。这些模拟结果表明,联合显著性检验比偏差校正自助法更具功效,并且产生的第一类错误也更合理。