Spatially Dependent Predictor Selection for Small-Area Health Modeling

• Activity Number: 461
• Type: Invited
• Date/Time: Wednesday, August 7, 2013 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistics and the Environment
• Author(s): Andrew B. Lawson*+ and Jungsoon Choi
• Companies: Medical University of South Carolina and Medical University of South Carolina
• Keywords: spatial ; small area ; health ; predictor ; selection

Abstract

The association of small area health outcomes with spatially-referenced predictors is often a focus. Geographically dependent random slope models have been proposed and offer a flexible alternative. Discrete grouping (clustering) has also been proposed recently (Choi, et al 2012). It is also possible to consider an extension of these models where it is reasonable to assume that different sub-regions of the study window have different predictor associations. To achieve this a novel spatially dependent predictor selection method is proposed. Initially a Gibbs variable selection (GVS) and the Kuo & Mallick method are considered. Variants can be considered such as spatially dependent SSVS. We will demonstrate the methods within a simulation study and also a case study.

Choi, J., et al (2012) A Bayesian Latent model with spatio-temporally varying coefficients in low birth weight incidence data. Statistical Methods in Medical Research, 21, 5, 445-456