the paper considered the available information for theta estimation and the efficiency of use of items. he added two additional criteria for original minimum SE criterion. One is called hyper parameter, the other is hypo parameter. The hyper parameter is trying to monitor the possibility of expected reduction in SE. If it does, additional items are administered to respondents even the SE criterion was met. If the SE criterion was not met yet, the hypo parameter was used to check if additional information is larger an arbitrary level for theta estimation. The two criteria were combined with the existing method MEPV to see how it works for theta estimation and the use of items.
It is natural that the new rule lead to slight reduction in RMSE and increased the item length for administration. The matched and mismatched cases were considered in respect to test information. In overall trait estimation, the RMSE is a little lesser than SE criterion and a bit more items required for administration. For extreme trait level, the new rule gave a larger improvement in reducing the need of items. It saves time and more efficient.
More studies should be taken to give a more general rule to set the value of hyper and hypo parameters for any kinds of CAT conditions.
Qs:
1. The paper was not well-written in English. I have a hard time in understanding his phrase in English.
2. The figure 4 is totally unclear to read. I think it is the most important part in this article.
3. The aid of this new rule is not as much as I think. It only gave slight improvement for the reduction of RMSE in overall condition.
4. The new rule can be used in CAT based on CDM or unfolding model.