RprobitB-package: RprobitB: Bayesian probit choice modeling
In RprobitB: Bayesian Probit Choice Modeling
RprobitB-package R Documentation
RprobitB: Bayesian probit choice modeling
Description
Fits Bayesian probit models for binary, multinomial, ordered, and ranked
choices in cross-sectional and panel data.
Details
Model
Decider n faces J alternatives on choice occasion t. Alternative j
has the latent utility U[n,t,j] = X[n,t,j] %*% beta[n] + epsilon[n,t,j],
where the covariate vector X[n,t,j] follows from formula and the error
vector across alternatives is multivariate normal with covariance Sigma.
Unordered choices select the alternative with the largest utility, ranked
choices order all utilities, and ordered choices compare one utility with
increasing thresholds gamma.
Fixed coefficients are identical for all deciders. Random coefficients named
in random_effects vary across deciders and follow a multivariate normal
distribution with mean mu and covariance Omega. Effects named in
latent_class_effects differ between classes latent classes with weights
weight: a random effect then has class-specific means and covariances,
another coefficient one value per class. Log-normal effects apply exp()
or -exp() to the latent normal coefficient. Utilities are identified only
up to level and scale, see the normalization details in fit().
Estimation
Posterior draws come from a Gibbs sampler that augments the latent utilities
(Albert and Chib 1993; McCulloch and Rossi 1994; Imai and van Dyk 2005).
Heterogeneity can follow a finite mixture, a sparse overfitted finite
mixture (Rousseau and Mengersen 2011; Frühwirth-Schnatter and Malsiner-Walli
2019), or a Dirichlet process mixture with Neal's (2000) auxiliary-parameter
allocation update and the precision update of Escobar and West (1995).
Independent chains run through the
future framework, and retained
draws use the posterior format.
Evaluation
predict(), residuals(), logLik(), WAIC(), loo(), and
bayes_factor() evaluate choice probabilities and likelihoods with
choicedata. Panel likelihoods
of mixed models integrate over the random coefficients with the GHK
simulator (Train 2009). Information criteria follow Vehtari, Gelman, and
Gabry (2017), and Bayes factors use bridge sampling (Gronau et al. 2017).
Author(s)
Maintainer: Lennart Oelschläger oelschlaeger.lennart@gmail.com (ORCID)
Authors:
Lennart Oelschläger oelschlaeger.lennart@gmail.com (ORCID)
Other contributors:
Dietmar Bauer dietmar.bauer@uni-bielefeld.de (ORCID) [contributor]
References
\insertRefAlbert1993RprobitB
\insertRefEscobar1995RprobitB
\insertRefFruehwirthSchnatter2019RprobitB
\insertRefGronau2017RprobitB
\insertRefImai2005aRprobitB
\insertRefMcCulloch1994RprobitB
\insertRefNeal2000RprobitB
\insertRefOelschlaeger2021RprobitB
\insertRefOelschlaeger2026cRprobitB
\insertRefRousseau2011RprobitB
\insertRefTrain2009RprobitB
\insertRefVehtari2017RprobitB
See Also
Useful links:
RprobitB documentation built on Sept. 20, 2026, 9:07 a.m.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
| RprobitB-package | R Documentation |
RprobitB: Bayesian probit choice modeling
Description
Fits Bayesian probit models for binary, multinomial, ordered, and ranked choices in cross-sectional and panel data.
Details
Model
Decider n faces J alternatives on choice occasion t. Alternative j
has the latent utility U[n,t,j] = X[n,t,j] %*% beta[n] + epsilon[n,t,j],
where the covariate vector X[n,t,j] follows from formula and the error
vector across alternatives is multivariate normal with covariance Sigma.
Unordered choices select the alternative with the largest utility, ranked
choices order all utilities, and ordered choices compare one utility with
increasing thresholds gamma.
Fixed coefficients are identical for all deciders. Random coefficients named
in random_effects vary across deciders and follow a multivariate normal
distribution with mean mu and covariance Omega. Effects named in
latent_class_effects differ between classes latent classes with weights
weight: a random effect then has class-specific means and covariances,
another coefficient one value per class. Log-normal effects apply exp()
or -exp() to the latent normal coefficient. Utilities are identified only
up to level and scale, see the normalization details in fit().
Estimation
Posterior draws come from a Gibbs sampler that augments the latent utilities (Albert and Chib 1993; McCulloch and Rossi 1994; Imai and van Dyk 2005). Heterogeneity can follow a finite mixture, a sparse overfitted finite mixture (Rousseau and Mengersen 2011; Frühwirth-Schnatter and Malsiner-Walli 2019), or a Dirichlet process mixture with Neal's (2000) auxiliary-parameter allocation update and the precision update of Escobar and West (1995). Independent chains run through the future framework, and retained draws use the posterior format.
Evaluation
predict(), residuals(), logLik(), WAIC(), loo(), and
bayes_factor() evaluate choice probabilities and likelihoods with
choicedata. Panel likelihoods
of mixed models integrate over the random coefficients with the GHK
simulator (Train 2009). Information criteria follow Vehtari, Gelman, and
Gabry (2017), and Bayes factors use bridge sampling (Gronau et al. 2017).
Author(s)
Maintainer: Lennart Oelschläger oelschlaeger.lennart@gmail.com (ORCID)
Authors:
Lennart Oelschläger oelschlaeger.lennart@gmail.com (ORCID)
Other contributors:
Dietmar Bauer dietmar.bauer@uni-bielefeld.de (ORCID) [contributor]
References
\insertRefAlbert1993RprobitB
\insertRefEscobar1995RprobitB
\insertRefFruehwirthSchnatter2019RprobitB
\insertRefGronau2017RprobitB
\insertRefImai2005aRprobitB
\insertRefMcCulloch1994RprobitB
\insertRefNeal2000RprobitB
\insertRefOelschlaeger2021RprobitB
\insertRefOelschlaeger2026cRprobitB
\insertRefRousseau2011RprobitB
\insertRefTrain2009RprobitB
\insertRefVehtari2017RprobitB
See Also
Useful links:
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Embedding an R snippet on your website
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.