Paper Summary
Share...

Direct link:

Item Exposure Control in Cognitive Diagnosis Computerized Adaptive Testing

Sun, April 19, 4:05 to 5:35pm, Marriott, Floor: Sixth Level, Northwestern/Ohio State

Abstract

This paper introduces a restrictive progressive method in conjunction of the G-DINA model discrimination index (RP-GDI) as an item selection method under cognitive diagnosis computerized adaptive testing, and investigates the efficiency of the index in terms of item exposure rate. The efficiency of the index was compared with the posterior-weighted Kullback-Leibler index (PWKL) using a simulation study. The results showed that the index was promising to estimate the attribute patterns for each examinee correctly. Moreover, it produced smaller exposure rates compared to the PWKL using the restrictive progressive method without sacrificing too much attribute classification accuracy. Moreover, the impact of different factors was also investigated.

Authors