Ordinal data: a new model with applications

In this work we propose a new discrete probability distribution useful when we work with ordered categorical data. The discrete-beta distribution has a highly flexible shape and it can be either over-dispersed or under-dispersed with respect to the binomial distribution. It has only two parameters, which have a very clear interpretation: the mean and the precision parameters of the beta latent variable. Adding directly covariates on parameters (according to CUB model framework), it is very suitable to regression.