Comparison of Electronic Examinations using Adaptive Multiple-choice Questions and Constructed-response Questions

The use of computer-based examination systems offers advantages related to the reduction of human resource allocation and to gains in objectivity for the scoring process. Multiple-choice questions (MCQs) are widely used in such systems; one of the main reasons that hamper their effectiveness, in comparison with constructed response questions (CRQ), is the influence of guessing. Considering limitations within previously proposed MCQs examination methods and scoring rules, in the present work a novel MCQs examination method is presented, termed 'adaptive' MCQs method. MCQs are divided into 3 categories, per difficulty level. The `path' that an examinee will follow is constituted by 3 phases, wherein a set of questions belonging to one of the three difficulty-categories, is appointed. The exact path followed is selected per the success level of the examinee in the preceding phase. The scoring provided by the adaptive MCQs examination method produced results that were statistically indistinguishable to the scoring produced by a traditional CRQ examination method. At the same time, both the scoring results of the adaptive MCQs examination and the scoring results of the CRQ examination differed significantly from those obtained by a generic `non-adaptive' MCQs examination.