NotesFAQContact Us
Search Tips
PDF pending restoration PDF pending restoration
ERIC Number: ED133331
Record Type: Non-Journal
Publication Date: 1976-Apr-9
Pages: 7
Abstractor: N/A
Reference Count: N/A
Can Rasch Scaled Scores Be Predicted from a Calibrated Item Pool?
Doherty, Victor W.; Forster, Fred
In order to verify that a test made of items selected from a pool produces the same Rasch scaled achievement scores as would be predicted from previous Rasch difficulty calibrations for those items, the following experiment was conducted. Seven math tests with similar content but graduated difficulty were administered to students in grade seven. The tests overlapped each other so that each test included about 20 items that were also included in one of the other six tests. The items in these seven tests were calibrated for difficulty using the Rasch procedure. Through the linking data provided by the common items, each of the items in all of the tests was adjusted to the same scale of difficulty. These items then formed a single pool of items. These items were listed in ascending difficulty order and were divided into seven level tests each with 30 items. These new tests were administered to about 1500 students in the seventh grade. Rasch item calibrations were again computed, and linking data used to adjust items to a common achievement scale. Rasch scaled achievement scores were computed for each possible raw score for each of the level tests and for the same items as calibrated in the original pool. The comparable scaled achievement scores for the tests were established on the basis of the actual performance of students on the second administration. Tables present the results. (RC)
Publication Type: Reports - Research
Education Level: N/A
Audience: N/A
Language: N/A
Sponsor: N/A
Authoring Institution: N/A