1. They manipulated the size of sample, number of items, magnitude of c parameter, or the magnitude of correlation in multidimensional model. Is there any expectation of those parameters about how they affecting the type one error and power before conducting simulation studies?
2. The extension of M2 may be derived from multivariate multinomial distribution for polytomous case.
3. For section 5.3, the data were generated from 2PLM. Is it supposed the parameters were calibrated from 2PLM firstly to get estimates. Then treat them as if calibrated from 1PLM (null hypothesis) and finally calculate the every index?