DCM (Discrete Choice Models) is a package for estimating a class of discrete choice models. DCM implements a wide range of discrete choice models including standard binary response models, with notable extensions including conditional mixed logit, multinomial probit, and random coefficient ordered choice models.
New developments in v2 include a contraction mapping that facilitates the estimation of highly disaggregate models, both logit and mixed logit, over a choice set with thousands of choices. Endogeneity of attributes is handled via an inversion procedure that casts the endogeneity problem within a linear model.
DCM provides estimates of parameter values, marginal effects and elasticities for a wide class of models. The current version can handle cross-section and static panel data. DCM is also flexible in accomodating revealed and stated preference data formats.
DCM requires the Ox Matrix Programing Language, and specifically Ox3.2 or a more recent release. The Console version is available free of charge, and can be accessed using the Download button above. An extended version including graphical capabilities is available through Timberlake Consultants