Gilles, Jean-LucJean-LucGillesDemetriou, ADochy, E2018-04-302018-04-302005http://hdl.handle.net/20.500.12162/745The confidence degrees technique associated to multiple choice questions (MCQs) makes it possible to bypass the "binary" character of students assessment performance (the selected proposal is either correct or incorrect) provided that a series of methodological rules are followed called "admissible probability measurement procedures" by Shufford & al. (1966). Usually, the confidence percentages which accompany the MCQs answers are used to deliver more subtle feedback on each person's spectral performances. The innovative aspect of our approach lies in the fact that we have exploited the confidence percentages provided by the students to provide spectral information on the quality of the questions (as opposed to information on the quality of students’ performances). Our research thus led to the development of a series of original indices for the analysis of assessments' spectral quality. These spectral indices are intended to be used when the assessor must highlight problematic MCQ and, within those, the proposals which contain anomalies.enType de référence::Communications::Communication scientifique publiée