The aim of this study was to construct a recognition test, using the Rasch model, composed of 64 items to be recognized, 32 items for study and 32 distracters. The pairs of items comprised written words or figures, they were either modal or cross-modal; half of them belonged to the same semantic category and the other half to a variety of other categories. The pairs were grouped according to these features and it was predicted that they would follow a given order of difficulty and that their mean of difficulty would be significantly different. The test was collectively administered to 231 students from a private university. The outcomes indicated a very good item fit as regards infit and outfit and also a good index of precision. The expected orders of difficulty were obtained empirically, although the means of some groups of items were not significantly different.
Memory; Recognition learning; Item response theory
