1. The effect of local dependent has been modeled in bifactor model. Then, the single item consisting of weak LD from each content area were selected out to form a nearly unidimensional test. I don’t understand what is worthy. In practical, the number of content area is few. So the number of item of the unidimensional test is few as well.
2. It seems the application of ECV depends on the bifactor model. I’m not sure if it can be used when exploratory bifactor model was adopted.
3. The results of Table 4 and 5 comparing the parameter estimates from the two competing model provide a rough eyeballing to us. Maybe a statistic of comparison is required to serve us more meanings.
4. I’m still wondering if there is any condition about attitude or personality assessment assumes unfolding perspective. If so, bifactor unfolding model is promising.