Document Type
Conference Proceeding
Abstract
Recently, some high-stake CAT exams revised the design to allow students to change answers. Simulation studies are conducted to evaluate the impacts of answer change on item selection and final theta score estimation. Three factors are conditioned when simulating the response data of answer change in CAT administration: answer change location, answer change pattern, and test length. The findings show a significant impact on the students’ score estimate when the answer change happens at the beginning of the test. The results will shed light on the answer change policy and their implications in the CAT administration.
Recommended Citation
Liu, Jingyu and Deng, Nina, "The Effect of Answer Change on CAT Score Estimation" (2024). NERA Conference Proceedings 2024. 11.
https://digitalcommons.lib.uconn.edu/nera-2024/11
