Flexible Semiparametric Hierarchical Spatial Models

• Activity Number: 643
• Type: Contributed
• Date/Time: Thursday, August 8, 2013 : 8:30 AM to 10:20 AM
• Sponsor: Biometrics Section
• Author(s): Aaron Porter*+ and Scott H. Holan and Christopher K. Wikle
• Companies: The University of Missouri Deptartment of Statistics and University of Missouri and University of Missouri
• Keywords: Bayesian ; CAR ; Fay-Herrot Model ; Small Area Estimation

Abstract

Spatial modeling has been a topic of extensive research over the past decade with many parametric models having been proposed. Nevertheless, in practice, it is often beneficial to relax a priori modeling assumptions to accommodate the ever-increasing complex data structures that frequently arise. Our hierarchical Bayesian approach specifies a flexible likelihood along with a spatial dependence structure placed at the latent process level of the hierarchy. Similar to the fully parametric setting, this provides substantial improvements in terms of computational efficiency. The effectiveness of our approach is illustrated through simulation and a diverse set of case studies, including small area estimation.