A Comparison of Two Equipercentile Equating Methods for Common Item Equating

Abstract

Two general procedures have been suggested for equating form X to form Y when an internal anchor test V with nonequivalent groups is used. In one method, X is equated to V by using equipercentile equating on one group of examinees. Then V is equated to Y by employing equipercentile equating on the other group of examinees. From these two equatings one can arrive at an equating of X to Y. This procedure is referred to in this paper as the Design V (Angoff, 1984) method. The second method is referred to as the frequency estimation method. The frequency estimation procedure requires consideration of the joint X, V frequency distribution for one examinee group and the joint Y, V frequency distribution for the other examinee group. The Design V method requires consideration of only the marginal distributions. Braun and Holland (1982, pp. 39-42) indicated that there are theoretical problems with the Design V method. However, the data requirements for the Design V method are many fewer than those for the frequency estimation method, and the Design V method is computationally much less burdensome. Computer storage requirements and computational burden are important considerations in the development of transportable computer systems for equating. The purpose of this study was to compare results for the two methods to determine whether or not they produce similar results in actual equatings. The results indicated that in the examples studied the two methods could not be considered interchangeable. These results have practical implications for the development of transportable computer systems for equating in practical settings.