ABSTRACT This paper discusses one of the peculiarities of information collected by rubrics, i.e. the qualitative nature of the categorical variables gathered. The qualitative nature of this data does not allow using conventional univariate statistical techniques such as, for example, calculating means and variances, or multivariate statistical techniques such as correlation or regression analysis. To address this issue, we turned to latent variable models and item response theory (e.g. a Rasch-family model). We apply this methodology to asses the “Written Communication “ learning goal in a Macroeconomic Theory course. Keywords Rubrics learning goal assessment, Item response theory, Rasch model