Computerized adaptive testing for the rasch testlet response model with ability-based guessing

Sheng Yun HUANG, Wen Chung WANG

Research output: Contribution to conferencePaper

Abstract

In this study, we developed a new model to incorporate ability-based guessing into the Rasch testlet response model, implemented corresponding CAT algorithm, and compared the performances of three item exposure control methods through a series of simulation. The results show that the CAT algorithm and the three item exposure control methods for the new model have been successfully developed and implemented. The simulations suggested that the smaller the testlet effect, and the longer the test, the smaller the RMSE will be; the SH online method and the SH online with progression method can maintain a well-controlled item exposure rate as their pre-specified rate, and the progressive method can achieve a higher bank usage, without substantial loss in measurement accuracy.
Original languageEnglish
Publication statusPublished - 2010

Citation

Huang, S.-Y., & Wang, W.-C. (2010, May). Computerized adaptive testing for the rasch testlet response model with ability-based guessing. Paper presented at the Annual Meeting of American Educational Research Association: Understanding Complex Ecologies in a Changing World, Denver, CO.

Fingerprint Dive into the research topics of 'Computerized adaptive testing for the rasch testlet response model with ability-based guessing'. Together they form a unique fingerprint.