This paper introduces the generalized beta (GB) model as a new modeling tool in the educational assessment area and evaluation analysis, specifically. Unlike normal model, GB model allows us to capture some real characteristics of data and it is an important tool for understanding the phenomenon of learning. This paper develops a contrast with the normal model, making one to observe that there are situations in which the most common assumption is that the normality of data is not always the best. The theory of educational assessment should begin to open to new statistical tools offered and add new models in order to capture one of the best features of the data and reject strong assumptions like symmetry.