Abstract
The study evaluated the reliability of pass and fail classifications for several teacher certification tests. Since these
tests are used in the context of a cut score to classify examinees as pass and fail, evaluating the accuracy and consistency
of these classifications is important. The classification accuracy and consistency statistics were estimated using the RELCLASS
software. Results indicated the following. (1) The 29 teacher certification tests that were examined had a relatively high
classification accuracy (0.827 to 0.999) and consistency (0.760 to 0.999). (2) Both classification accuracy and consistency
increased as the difference between the mean and cut score increased. (3) Classification accuracy and consistency was higher
for multiple-choice (MC) as compared to tests consisting of only constructed-response (CR) items or a combination of CR and
MC items.
tests are used in the context of a cut score to classify examinees as pass and fail, evaluating the accuracy and consistency
of these classifications is important. The classification accuracy and consistency statistics were estimated using the RELCLASS
software. Results indicated the following. (1) The 29 teacher certification tests that were examined had a relatively high
classification accuracy (0.827 to 0.999) and consistency (0.760 to 0.999). (2) Both classification accuracy and consistency
increased as the difference between the mean and cut score increased. (3) Classification accuracy and consistency was higher
for multiple-choice (MC) as compared to tests consisting of only constructed-response (CR) items or a combination of CR and
MC items.
- Content Type Journal Article
- Category Assessment
- Pages 1-10
- DOI 10.1007/s12646-012-0147-9
- Authors
- Gautam Puhan, Educational Testing Service, Princeton, NJ, USA
- Leanne Gall, Educational Testing Service, Princeton, NJ, USA
- Journal Psychological Studies
- Online ISSN 0974-9861
- Print ISSN 0033-2968