1. The modified DETECT is believed to be performance well in capturing the dimensional structure of a test when three conditions are hold. The third condition which stating that there is no-isolated dimensions. How to examine whether this condition is hold or not?
2. My second questions is related to the Point 1. In Corollary 1, it states when conditons 1 and 2 are satisfied, if the items associated with isolated dimensions are in the same booklet, the modified DETECT will perform well. However, if one wants to check whether items associated with isolated dimensions are in the same booklet, first of all, one should decide which item is associate with isolated dimensions. How to?
3. Whether the percentage of disordant items will affect the performance of proposed modified DETECT? Considering that only about 45% items pairs had concordant signs in real data and the author also warrant the optimal partitions.
4. I was surprisng to find that though LAMP test was by design to measure three domains and the D3 indicated a better solution than D2, it was concluded that two-cluster partition is a perferred alternative.
5. I think using CAT as a context of using the modified index is a little misleading. Multistage testing is more appropriate for the context.