Bias-Reduced Tobit Regression

Overview

The R package brtobit provides bias-reduced estimation of so-called Tobit models: Regression models with a Gaussian response variable left-censored at zero, constant latent variance, and a latent mean that depends on covariates through a linear predictor. As an alternative to plain maximum likelihood estimation, the adjusted score equations of Kosmidis and Firth (2010) are utilized to obtain bias-reduced estimates of the model parameters.

Reference

Köll S, Kosmidis I, Kleiber C, Zeileis A (2021). “Bias Reduction as a Remedy to the Consequences of Infinite Estimates in Poisson and Tobit Regression.” arXiv.org E-Print Archive, arXiv 2101.07141. doi:10.48550/arXiv.2101.07141

Installation

The latest version of brtobit is available from R-universe:

install.packages("brtobit", repos = "https://zeileis.R-universe.dev")

License

The package is available under the General Public License version 3 or version 2