Conference paper
Comparing and Combining Dichotomous and Polytomous Items with SPRT Procedure in Computerized Classification Testing
Distributed by ERIC Clearinghouse
AERA Annual Meeting (San Diego, California, USA)
1998
Abstract
The purposes of this study were to: (1) extend the sequential probability ratio testing (SPRT) procedure to polytomous item response theory (IRT) models in computerized classification testing (CCT); (2) compare polytomous items with dichotomous items using the SPRT procedure for their accuracy and efficiency; (3) study a direct approach in combining these two types of items in CCT; and (4) study a simple method to control item exposure rate in CCT. The Monte Carlo simulation technique was used, manipulating several independent variables. Item parameters from the 1996 National Assessment of Educational Progress (NAEP) science assessment were used, resulting in 247 dichotomous items and 266 polytomous items for the 3 NAEP grades. The results of using the three item pools (polytomous, dichotomous, and combined) were generally consistent. Items selected according to item information at the cutting theta resulted in the best classification accuracy and efficiency. Results show that polytomous items work well with SPRT in CCT. SPRT was also found to be a good procedure for mastery decision making. (Contains 2 tables and 14 references.) (SLD)
Details
- Title: Subtitle
- Comparing and Combining Dichotomous and Polytomous Items with SPRT Procedure in Computerized Classification Testing
- Creators
- C. Allen LauTianyou Wang
- Resource Type
- Conference paper
- Conference
- AERA Annual Meeting (San Diego, California, USA)
- Publisher
- Distributed by ERIC Clearinghouse
- Number of pages
- 16 pages
- Comment
- Paper presented at the 1998 AERA Annual Meeting in San Diego, California, April 1998.
- Language
- English
- Date published
- 1998
- Academic Unit
- Center for Advanced Studies in Measurement and Assessment
- Record Identifier
- 9984631938502771
Metrics
1 Record Views