收回来的数据发现观测变量不符合正态分布,看到类似研究也遇到了同样的问题。他们用heterogeneous kurtosis estimation method来估计共变数分布,以解决这个问题。文献报告用的是EQS来处理,不知道Amos里面该怎么解决。哪位朋友遇到过这类问题啊,求助了~~~ 文献里这么说的:CFA was conducted on the covariance matrix using EQS6 (Hines, Chiu, McAdams, Bentler, & Lipcamon, 2002). Due to significant kurtosis for many of the items (Mardia’s normalized multivariate kurtosis was greater than 3), we used the heterogeneous kurtosis estimation method to estimate the distribution of covariances in the evaluation of all models (Bentler, 1995; Bentler, Berkane, & Kano, 1991). |