The purpose of this study is to examine the factor structure of the Korean version of the 20-Item Toronto Alexithymia Scale. The TAS-20 (source of the TAS-20K) has been supported the three-factor correlated model. However, some factor structure studies of the TAS-20 rejected the three-factor correlated model and adopted alternative models.
In study 1, we conducted a comparison study of the alternative measurement models by using CFA. In study 2, we examined scale reliability and gender measurement invariance of the factor structure. To examine the alternative models and scale reliability, we using the bifactor model reliability indices.
As a result, the DIF and DDF factors have a close relationship but the EOT factor has some differences with DIF and DDF. So we adopted a two-factor correlated model with group factor. And the adopted factor structure has partial measurement invariance. Therefore we can compare gender differences of the TAS-20K.
This study has significance that examining TAS-20K’s factor structure and examining measurement invariance in gender.
Citations
The purpose of this study was to examine the factor structure of the Korean version of CES-D scale by applying the bifactor model. Many studies for validating the CES-D scale have supported the four-factor structure. But some studies found that the three factor structure provided a more appropriate solution. Such inconsistency in the number and structure of factors led us to conduct two studies to clarify the structure of dimensionality of the CES-D-K scale.
In study 1, we factor-analyzed the response patterns to the abbreviated CES-D-K scale (11-item, N=12,309) included the Korean Welfare Panel Study. In study 2, we factor-analyzed the data obtained from 223 college and graduate students who responded to the 20-item CES-D-K scale. Correlational analyses were also conducted to investigate criterion validity of the CES-D-K scale with external variables that are theoretically related to depression. The fit indexes of a single-factor model, a four-factors model, and a bifactor model were compared. The Omega coefficients and Explained Common Variance (ECV) were also computed to evaluate the psychometric properties of the CES–D-K scale more accurately.
The results showed that the CES-D-K scale has a high value of Omega-Hierarchical for the total score, low values of Omega-Hierarchical for the subscale scores, and a high ECV value.
Therefore, we concluded that, as the scale with a strong general factor, the use of the scale score can accomplish the goal of measuring individual differences on the target construct of depression with little to no gain from constructing subscale scores.
Citations