Regression modelling with the tilted beta distribution: A Bayesian approach
Abstract
Beta regression models are commonly used in the case of a dependent variableythat exists on the range (0,1). However, whenycan additionally take on the values of zero and/or one, limitations of the beta distribution and beta regression models become apparent. One recent approach is to use an inflated beta regression model which has discrete point‐valued components. In this article, we introduce a new class of regression models fory∈ [0,1] that is fully continuous. This allows the entirety ofyto be treated as a continuum instead of discontinuously, which appears to be a new development for the literature. We use a Bayesian approach for estimation. We also illustrate the impact of different choices of prior distributions on empirical findings and perform a simulation study examining model fit.
Faculty Members
- Eugene D. Hahn - Department of Information and Decision Sciences Franklin P. Perdue School of BusinessSalisbury University Salisbury MD U.S.A
Themes
- Model fit evaluation
- Beta regression models
- Continuous dependent variables
- Bayesian estimation
- Inflated beta regression