Measuring students' assessments on university course quality using mixed-effects models

In this work a generalized Item Response Model has been considered in order to analyze students’ perceived quality of university course. The approach allows the author to pursue several tasks in the estimation of item and person parameters. Namely: (i) it allows to jointly model students’ answers to the whole set of indicator items; (ii) it considers the dimensionality of the items composing the test by explicitly specifying more than one latent trait; (iii) it evaluates the effect of significant characteristics on students’ perception of the latent traits; (iv) it takes into account the ‘intra-cluster’ variability which affect ratings at course level. Moreover, by moving from the posterior estimates at individual and course level, it has been possible to make pairwise comparisons between courses and students on the bases of their levels of perceived quality