Q:
1. In table 3, take item 1 as example we can find that crossing over the threshold 1 to threshold 2 is quite large, -2.008 – 3.094 ~= -5. The absolute distance is quite large in common latent continuum. In other item’s threshold, we can also find this results of estimation. Does it make sense?
2. In p. 816, it says the MNRM can be reduced to PCM via constraint on a parameter. But I can’t figure out why it can be.
3. The design for extreme response by constraint a parameter is interesting. It means the person will respond extremely to either side on the continuum when theta_ERS is getting positively large; the person will select middle categories except the two extreme categories when theta_ERS is getting negatively large. The shortcoming is the magnitude of value of constraint is quite arbitrary. It is interesting to explore the magnitude of value on the estimate of theta_ERS.
4. In figure 1, the results perplex me. The curve for theta_ERS is not symmetrical while using the theta_ERS = 0 as original point. And I do not know how to interpret it in common sense. The plot is derived from real data analysis, so it is interesting to know if it will get the same results of figure 1 in simulation study? However, it is not presented in this article.
5. In table 8, when -1.5 < theta1 < -.5, the MAD is larger than in other middle range of theta when theta_ERS is large. It is a bit curious to know why it happened.