Comparison of Stratum Scored and Maximum-Likelihood Scored CATs [electronic resource] / Steven L. Wise.
Outside of large-scale testing programs, the computerized adaptive test (CAT) has thus far had only limited impact on measurement practice. In smaller-scale testing contexts, limited data are often available, which precludes the establishment of calibrated item pools for use by traditional (i.e., it...
Saved in:
| Online Access: |
Full Text (via ERIC) |
|---|---|
| Main Author: | |
| Format: | Electronic eBook |
| Language: | English |
| Published: |
[S.l.] :
Distributed by ERIC Clearinghouse,
1999.
|
| Subjects: |
| Summary: | Outside of large-scale testing programs, the computerized adaptive test (CAT) has thus far had only limited impact on measurement practice. In smaller-scale testing contexts, limited data are often available, which precludes the establishment of calibrated item pools for use by traditional (i.e., item response theory (IRT) based) CATs. This paper introduces an alternative adaptive testing procedure--termed a stratum CAT--that requires no IRT methods for either item selection or proficiency estimation. In two simulation studies comparing stratum CATs to conventional tests and traditional CATs, stratum CATs were found to be substantially more efficient than conventional tests. In addition, with 100 or fewer examinee responses available per item, the efficiency of stratum CATs could match or exceed that of traditional CATs. The stratum CAT may prove to be a useful adaptive testing procedure in situations where limited item calibration data are available. (Contains 3 tables and 12 references.) (Author/SLD) |
|---|---|
| Item Description: | ERIC Document Number: ED429119. ERIC Note: Paper presented at the Annual Meeting of the National Council on Measurement in Education (Montreal, Quebec, Canada, April 19-23, 1999). |
| Physical Description: | 32 p. |