Professuren für Statistik und Ökonometrie


Here we collect some information on software that can be used to perform different kinds of regression analyses described in our book. The list is of course by no means considered to be complete but mostly reflects our own experience with different types of software. Suggestions, additions and corrections are of course welcome.


  • BayesX has been developed as a specialized software for (Bayesian) inference in structured additive regression. It supports fully Bayesian inference based on Markov chain Monte Carlo simulations as well as empirical Bayes inference based on a mixed model representation of structured additive regression models. It also contains an implementation of an efficient stepwise model selection algorithm.
  • The most important model classes and regression effects included in BayesX comprise generalized additive models, generalized additive mixed models, geoadditive mixed models, Bayesian ridge and Bayesian LASSO.
  • Supported response types include exponential family responses, categorical responses (both ordered and unordered), duration times and multi-state models.
  • A connection to R is provided in the package R2BayesX.
  • Homepage: BayesX


  • R is a free general-purpose statistics software that combines a comprehensive set of regression functionality with a complete programming framework and highly adaptive graphics facilities.
  • The base R distribution supports linear models, generalized linear models, mixed models (packages nlme and lme4) and generalized additive mixed models (package mgcv). More specialized packages also comprise functionality for Bayesian inference, categorical regression models, quantile regression, etc.
  • To identify suitable packages, consulting the R-site search is a good starting point.
  • In addition, R Task Views collect packages that are relevant in a specific area such as spatial statistics or Bayesian inference.
  • Homepage:


  • Stata comprises functionality for all basic regression methodologies ranging from linear and generalized linear models to categorical regression. More specialized solutions for example for multi-level random effects or generalized additive models are available as add-ons (see for example here or here.
  • Homepage:


  • WinBUGS is a general-purpose software for Bayesian inference. It comprises a model specification language that allows the user to write down a model specification in terms of a directed acyclic graph that is then internally converted into an MCMC sampler. This yields rather high flexibility but also limits the control by the user.
  • A connection to R is provided in the R package R2WinBUGS.
  • Homepage: