An Adaptive Test for Learning Objects: item calibration

Learning Objects differ from the usual digital resources because they include both learning contents and assessment. In this paper we propose a methodology that can make an adaptive assessment of the knowledge/ability level achieved by the learner, and is thus less boring for primary school pupils than most of those produced up to now. The methodology, based on the ideas of Game Based Learning, Adaptive Systems, and Item Response Theory, allows tests to be created that can adapt the difficulty level of the questions to the individual learner. The preliminary phase in building a CAT is the calibration of the items in the bank item. This consists of assigning a level of difficulty to each item to be proposed to the student. Different psychometric models are used to scale the items. Usually, the IRT is employed because it uses the same metric to measure both the item difficulty and the student’s level of ability, so the two measures are comparable. Usually, the main problem in building a quiz game is the choice of what question to pose to the student. The CAT theory allows automatic selection of the question to ask, according to both its difficulty and the individual student’s abilities.