RMSEA與卡方差異法於檢定模式差異之正確性：測量不變性之檢定

This study is to investigate the screening accuracies between competing models by using RMSEA methods and Chi-square difference tests. The experimental design to demonstrate the accuracies contains various magnitudes of sample sizes， factor loadings， and degrees of freedom when multiple group confirmatory factor analysis is used to test measurement invariance. Some conventional practices believed that Chi-square difference test could be sensitive to sample sizes in selecting competing models while RMSEA may not have the problems. Results of simulation studies in this paper show that the beliefs have to be adjusted at least for some specified conditions， for example， some magnitudes of sample sizes， model differences， and model complexity (i.e.， degrees of freedom). Discussions and implications on the interesting results are provided.