1. the abstract describes that sample size 1 can be applied using their methd. I think all nonparametric methods can do it as well. the fact is that it is useless to provide any evident information.
2. the proposed method is appealing. however, it does not provide us any inference about how examinee responds to items because no theory was involved.
3. it may device a new index to indicate how much deviation of the specific underlying model was observed.
4. the formula 2 may be defined in another way. that is, the right term in equation is replaced by substraction bwteen probability rather than y minus ita.
5. the Dina model was estimated using miles but the nida model was estimated by em algorithm. is it a fair comparison?
6. the simulation used extreme large sample size to demonstrate its use. but how small sample can be effective fortheir method ? their conclusion is over-confident for the method.
7. the distance measure may be replaced by other method such as spectral clustering.