ERIC Number: ED420676
Record Type: Non-Journal
Publication Date: 1998-Apr-14
Reference Count: N/A
Effect of Item Selection on Item Exposure Rates within a Computerized Classification Test.
Kalohn, John C.; Spray, Judith A.
The purpose of many certification or licensure tests is to identify candidates who possess some level of minimum competence to practice their profession. In general, this type of test is referred to as classification testing. When this type of test is administered with a computer, the test is a computerized classification test (CCT). This paper addresses the effect of item selection on item exposure rates within a CCT. A testing program's ideal CCT item pool would consist of items that tended to measure the best at the latent equivalent of the passing score (theta p). For this study, it was hypothesized that there would be significant differences in item exposure rates for the extreme items in the pool (those that measured best at high and low values of theta), but that the overall impact of these differences would be negligible. A simulation was designed, using an actual item pool of 1,235 items. Observed item exposure rates were then calculated under two target exposure rate (TER) conditions, one set at 0.20 and the other at 0.10. Both methods of item selection produced about the same degree of classification accuracy, but performance of the CCTs in terms of classification accuracy, declined slightly for TER=0.10, as would be expected when the use of more items with less information is necessary. (Contains one table, seven figures, and six references.) (SLD)
Publication Type: Reports - Evaluative; Speeches/Meeting Papers
Education Level: N/A
Authoring Institution: N/A