1. At first, items belonging to which family requires classification by statistical method or expert’s judgments. It looks like a ‘classification’ problem except using model checking method mentioned in this paper.
2. P.142 it says certain combinations of radicals and incidentals may result in invalid item. Even using constraints to avoid the occurrence, how do we know its in validness? From practical or statistical view?
3. Are the beta parameters in Equation 4 estimated?
4. How many items were used in simulation one in page 148?
5. The mu(b) = -0.258 in page 148 looks strange. Do not know why it is so.
6. Compute the family information using MC method is straightforward. Is it possible using numerical integral? Because only three dimensions were used here.
7. It is better to know the practical variance of item parameters, then using them in simulation studies.
8. The Equation 6 requires to know the prior distribution of item parameters in advance. It is hard to know actually. If the distribution form is violated, what will happen to results?
9. Larger within variance in a family will lead to lower test information. What is the criterion to justify what is good or bad results?