1. I cannot figure out why more negative PRC slope is, better person fit it is. Where is the baseline? Is any quantitative index for it?
2. The subscript of β should be without j, right?
3. How to compute EB?
4. Step 1 and 2 in table 1 looks a bit bizarre. They both did not fit the dataset to full model (should be considered ‘more useful’ in the researcher’s mind) but two constrained models to test u0j and u1j respectively.
5. Example 1 shows the two-stage estimation. What is the ‘true’ model we use? Why not use full model directly? I guess full model cannot be easily identified.
6. How many replications in example 1? It seems no replication.
7. Example 2 says it calculate the correlation between threshold and probability? What does it mean?