抄録
Currently, test operation using Item Response Theory (IRT) requires test items to undergo parameter estimation using examinee data. Furthermore, after equating, the items may be included in an item pool that can be used for several tests. However, this test operation method contains the probability of item content leakage. Thus, estimating item parameters while keeping the item contents secret would be useful. In this study, to make such a situation possible, a model in which item parameters are estimated using a paired comparison from the perspective of the difficulty of items by a rater familiar with the field is proposed. The estimation accuracy of this model was confirmed in a simulation study, and the feasibility of its use in practical settings is demonstrated using actual data.
