Information and other criteria in structural equation model selection

摘要: This article presents the results of a simulation study evaluating information criteria in conjunction with other well-known for model selection structural equation modeling (SEM). Two sets experiments were performed. In both sets, sample sizes n = 100,400,1000,6000 used and performance 18 was assessed by frequency which each five analytic models selected as best criterion 500 replications. first set correctly specified (noncentrality parameter 0) entertained combination misspecified ones, while second all misspecified. experiments, we found that perform better than overall, but Cudeck Browne's cross-validation index ( CVI) remains an attractive option. Within class criteria, Akaike's (AIC) is to show some overfitting tendency. We demonst...