Application of Beta Regression in Empirical Research

DOI: 10.5176/2251-1911_CMCGS42

Authors: Siew-Pang Chan and Priscilla Yang

Abstract: A regression model is developed based on the beta distribution bounded within (a, b), where. The model is applied to forecast the U.S. presidential election results. The forecast is compared with that of ordinary least squares, robust regression and generalized linear model with gamma distribution. In view of its flexibility and theoretical justification, the proposed beta regression should serve as an alternative tool to the more familiar regression models.

Keywords: beta distribution, beta regression, GLM