Studies are proposed which center around the development and testing of new model-based statistical procedures. Included are applied studies in social, biomedical and physical systems whose analysis motivates certain new models and techniques that are tailored to practical problem-solving. The models to be studied gnerally have complex graphical structures, and the related inferences draw on several sources of statistical and expert knowledge. The statistical techniques include Bayesian and belief function methods that are computationally intensive, thus requiring algorithm development and are to be studied in part through assessment of numerical implementations.
