Problems of the Equivalence of the Parallel Test Forms

Abstract

The importance of correct usage of the parallel test forms is evident not only in the field of educational diagnostic testing. Especially it applies in “high stakes” examinations, where the important efforts for honesty should not violate the principles of equal opportunities of students. Some simple methods of the development of parallel test forms are investigated in this article. The results showed, that in some cases simple changes to the item forma t (e.g. changing the order of distracters, changing coefficients and similar) causes statistically significant differences in parameters between the parallel items. Such differences have no effect on the average scores, however they may become significant in other parts of the ability scale. Some applications of the Item Response Theory (IRT) methods have been used to compare parallel test forms.