Equivalence of Rasch Item Calibrations and Ability Estimates across Modes of Administration.

The equivalence of pencil and paper Rasch item calibrations when used in a computer adaptive test administration was explored in this study. Items (n=726) were precalibarted with the pencil and paper test administrations. A computer adaptive test was administered to 321 medical technology students using the pencil and paper precalibrations in the item selection algorithms and in the computation of examinee ability estimates. The response data from the computer adaptive test administration were analyzed yielding recalibrated item difficulties and examinee ability estimates. Item precalibrations were compared with item recalibrations. Examinee ability estimates obtained using the item precalibrations on the computer adaptive administration were compared with the examinee ability estimates obtained from using the item recalibrations on the computer adaptive administration. The correlation for examinee ability estimates was 0.99 and for item correlations it was 0.90. Some item calibrations shifted but most remained consistent within the limits of error. Item shift, however, did not affect the ordering of examinee ability estimates. (Contains 1 table, 3 figures, and 23 references.) (Author/SLD)