A model for bimodal rates and proportions

The beta model is the most important distribution for fitting data with unit interval. However, not suitable to bimodal interval data. In this paper, we propose a constructed by using an approach based on alpha-skew-normal model. We discuss several properties of distribution, such as bimodality, real moments, entropies and identifiability. Furthermore, new regression proposed residuals. Estimation performed maximum likelihood. A Monte Carlo experiment conducted evaluate performances these estimators in finite samples discussion results. An application provided show modelling competence when sets bimodality.