Semiparametric Additive Beta Regression Models

Abstract

In this paper, we study a semiparametric additive beta regression model using a parameterization based on the mean and a dispersion parameter. This model is useful for situations where the response variable is continuous and restricted to the unit interval, in addition to being related to other variables through a semiparametric regression structure. First, we formulate the model and then estimation of its parameters is discussed. A back-fitting algorithm is derived to attain the maximum penalized likelihood estimates by using natural cubic smoothing splines. We provide closed-form expressions for the score function, Fisher information matrix and its inverse. Local influence methods are derived as diagnostic tools. Finally, a practical illustration based on real data is presented and discussed.

Keywords

• beta distribution
• diagnostic techniques
• maximum penalized likelihood estimates
• penalized likelihood function
• semiparametric additive mode