NotesFAQContact Us
Search Tips
ERIC Number: ED441024
Record Type: Non-Journal
Publication Date: 2000-Apr-27
Pages: 15
Abstractor: N/A
Reference Count: N/A
Scaling an Introduction to Clinical Medicine Examination.
Fisher, William P.; Suttkus, Ramona; DiCarlo, Richard
This paper shows that a substantial degree of invariance can be attained with an examination not explicitly designed to do so, provides an example of how invariance can be demonstrated through plots, and dispels misconceptions concerning the rigidity of the definition of invariance. Responses of 177 examinees to 94 items of a final examination from an introduction to clinical medicine course were obtained. All data analyses involved fitting a dichotomous two-parameter model via unconditional maximum likelihood estimation and were performed using the WINSTEPS program. Even though the examination was designed by persons untrained in educational measurement, the test item orders achieved a substantial degree of invariance across samples of examinees. The measurement reliabilities closely approximated those predicted by Rasch generalizability theory. The plots of the item difficulty estimates from the separate samples and of the examinee ability estimates obtained from the separate subtests support the use of probabilistic conjoint measurement models and the search for invariance. The study shows that a carefully designed test can achieve a high degree of model fit and thus provide a scientific basis for the practical and convenient inferential advantages that follow from the use of sufficient statistics. (Contains 1 table, 8 figures, and 17 references.) (SLD)
Publication Type: Reports - Research; Speeches/Meeting Papers
Education Level: N/A
Audience: N/A
Language: English
Sponsor: N/A
Authoring Institution: N/A