CATDEV: Stata modules for interpretation of categorical dependent variable models

There are several methods that can be used to effectively interpret the results of regression models for categorical dependent variables. Each of these methods requires the analyst to complete post estimation computations of the estimated parameters. Generally speaking, the estimated coefficients cannot be meaningfully interpreted without additional computations. This set of additions to Stata (in the form of ado files) that facilitate the interpretation of results of the following models: binary logit, binary probit, ordinal logit, ordinal probit, multinomial logit, Poisson regression, negative binomial regression, and tobit. These commands make it easy to do the computations for a variety of methods of interpretation: predicted outcomes and plots of these outcomes, discrete changes in predicted outcomes, partial change in predicted outcomes, standardized coefficients for variables based on a latent variable, factor changes in odds ratios for logit models, and factor changes in mean counts for count models. Details on each of these methods of interpretation can be found in my book Regression Models for Categorical and Limited Dependent Variables (Sage Publications).