University of Washington - Statistics
This talk describes new multivariate regression and model-based clustering methods for statistical inference with multivariate mixed outcomes. We use the term mixed outcomes to refer to binary, ordered categorical, count, continuous and other ordered outcomes in combination. Such data structures are common in social, behavioral, and medical sciences. We develop two regression approaches, the semiparametric Bayesian latent variable model and the semiparametric reduced rank multivariate regression model, for mixed outcome data. In contrast to existing parametric approaches, these models allow us to avoid specification of distributions for each outcome. Similarly, we are able to avoid specification of outcome distributions with our new semiparametric correlated partial membership model for model-based clustering of mixed outcome data. We rely on Gibbs and Hybrid Monte Carlo sampling methods for estimation. We use the multivariate regression approaches to investigate the association between cognitive outcomes and MRI-measured regional brain volumes using data from a study of subcortical ischemic vascular dementia. We demonstrate the proposed semiparametric correlated partial membership model on player data from the 2010-2011 NBA season.