1. Although the results seemed of interest and satisfactory, unfortunately a wrong model was fitted in this study. In Equation 1, involved facets were assumed independent with each other. But as the standard procedure of the Angoff method, panelists could receive the feedback after the first round of judgment, and then they are allowed to adjust the original judgment according to the feedback. Thus the interaction between theta and omega should be considered in Equation 1.
2. Three separation reliabilities were reported on page 398, and luckily they were all .95. However, the values are meaningful if we understand the basic idea of the facets model. Before that, reporting the reliabilities is nonsense, although the values are very high.