NotesFAQContact Us
Search Tips
ERIC Number: ED453282
Record Type: Non-Journal
Publication Date: 2001-Apr
Pages: 21
Abstractor: N/A
Reference Count: N/A
Integrating Stratification and Information Approaches for Multiple Constrained CAT.
Leung, Chi-Keung; Chang, Hua-Hua; Hau, Kit-Tai
It is widely believed that item selection methods using the maximum information approach (MI) can maintain high efficiency in trait estimation by repeatedly choosing high discriminating (alpha) items. However, the consequence is that they lead to extremely skewed item exposure distribution in which items with high alpha values becoming overly exposed while many low alpha ones may never be selected. With some demonstrated empirical success, the alpha-stratified design (ASTR) and its extension, the alpha-stratified with b-blocking method (BASTR), were proposed to tackle these problems simultaneously. While the latter method takes into consideration the correlation between the alpha- and beta-parameters, both stratified designs advocate the more frequent use of low alpha items in earlier stages of testing. The main objective of this study was to investigate the performance of integrating the traditional information based selection and the new philosophy of using low alpha items first in a multiple constrained setting. The performances of MI, BASTR, and their integration, MIBASTR, were empirically compared through simulation. Results indicate that BASTR was the best in utilizing the entire pool evenly, and thus in tackling item security problems. On the other hand, MI and MIBASTR offered high and comparable measurement efficiency. The latter outperformed the former in item exposure and pool utilization. (Contains 2 figures, 4 tables, and 26 references.) (Author/SLD)
Publication Type: Reports - Evaluative; Speeches/Meeting Papers
Education Level: N/A
Audience: N/A
Language: English
Sponsor: N/A
Authoring Institution: N/A