Optimum Number of Strata in the a-Stratified Computerized Adaptive Testing Design

In the a-stratified method, a popular and efficient item exposure control strategy proposed by H. Chang (H. Chang and Z. Ying, 1999; K. Hau and H. Chang, 2001) for computerized adaptive testing (CAT), the item pool and item selection process has usually been divided into four strata and the corresponding four stages. In a series of simulation studies, researchers examined the optimum number of strata by systematically varying the number of strata, pool size (200, 400, and 800 items), item characteristics (0., 0.5 correlation between difficulty and discrimination), and item selection method (largest information, matching estimated ability with difficulty). Results show that quite independent of the item pool size and the correlation between item discrimination and difficulty, ability estimation deteriorated while the number of overand under-exposed items decreased with an increase in stratum number. There is a diminishing return in that dividing the pool into too many strata can also be problematic because when the stratum is too small, there are not any items of close difficulty for each particular examinee. The results are in general agreement with the speculation that too few and too many strata may not provide optimum efficiency and balanced item pool utilization. It is shown that the ideal and optimum number of strata to be used in each specific application depend on the item pool structure, test length, and other testing conditions. (Contains 8 figures, 24 tables, and 8 references.) (Author/SLD) Reproductions supplied by EDRS are the best that can be made from the original document. Optimal Strata in CAT g.1 Optimum Number of Strata in the a-Stratified Computerized Adaptive Testing Design Kit-Tai Hau, The Chinese University of Hong Kong Jian-Bing Wen, The Chinese University of Hong Kong Hua-Hua Chang, University of Texas, Austin Note. Paper presented at the American Educational Research Association Conference, New Orleans, April 1-5, 2002. Please address all correspondence to Professor Kit-Tai Hau, Faculty of Education, The Chinese University of Hong Kong, Shatin, N.T., Hong Kong (email: KTHAU@CUHK.EDU.HK). U.S. DEPARTMENT OF EDUCATION fice of Educational Research and Improvement EDUCATIONAL RESOURCES INFORMATION CENTER (ERIC) /1"his document has been reproduced as received from the person or organizaiion originating it. 0 Minor changes have been made to improve reproduction quality. o Points of view or opinions stated in this document do not necessarily represent official OERI position or policy. PERMISSION TO REPRODUCE AND DISSEMINATE THIS MATERIAL HAS BEEN GRANTED BY