Fitting latent class mixed multinomial probit (LCMMNP) models to simulated or empirical data. Estimation takes place in a Bayesian framework using a Gibbs sampler. The number of latent classes can be updated within the algorithm on a weight-based strategy. For a reference on the method see Oelschl\"ager and Bauer (2021) < https://trid.trb.org/view/1759753>.