A note on converting calibrated parameters of item response theory models. Issue 12 (1st December 2020)

• Record Type: Journal Article
• Authors: Magis, David
• Is Part Of: Communications in statistics. Volume 49:Issue 12(2020)
• Page Start: 3065
• Page End: 3079
• Publication Date: 2020-12-01
• Subjects: Item response theory -- Polytomous responses -- Difference model -- Divide-by-total model -- Calibration; Mathematical statistics -- Periodicals; Mathematical statistics -- Data processing -- Periodicals; Digital computer simulation -- Periodicals; 519.5
• DOI: 10.1080/03610918.2018.1532515
• Languages: English
• ISSNs: 0361-0918

Abstract

Abstract: Many item response theory models were proposed to estimate ability levels of test takers using their responses to a set of polytomously scored items. Recently, an increased number of calibrated item parameters have become publicly available. Unfortunately, these parameters are often calibrated under some model that is not the researcher's target model. This note presents a straightforward computational method to convert one set of calibrated item parameters to the corresponding parameters of another target model. The core of the method stands in equating the response category probabilities of both models. The accuracy of the method is assessed through several examples.