454 – Advances in Spatial and Spatio-Temporal Methodology with Applications to Official Statistics
A Model Selection Study for Spatio-Temporal Change of Support
Andrew M. Raim
U.S. Census Bureau
Scott H. Holan
University of Missouri - U.S. Census Bureau
Jonathan R. Bradley
Florida State University
Christopher K. Wikle
University of Missouri
Spatio-temporal change of support (STCOS) methods are designed for statistical inference and prediction on spatial and/or time domains which differ from the domains on which the data were observed. Bradley, Wikle, and Holan (2015) introduced a parsimonious class of Bayesian hierarchical spatio-temporal models for STCOS for Gaussian data through a motivating application involving the American Community Survey (ACS), an ongoing survey administered by the U.S. Census Bureau that measures key socio-demographic variables for various populations in the United States. Importantly, their methodology provides ACS data users a principled approach to estimating variables of interest, along with associated measures of uncertainty, on customized geographies and/or time periods. In this work, we revisit use of the STCOS methodology to capture median household income at the county level. We use the Deviance Information Criterion (DIC) to study variations in the prior and in basis function specification, and to select a reasonable final model. This work makes use of an R package which is currently under development, whose aim is to make STCOS methodology broadly accessible to federal statistical agencies such as the Census Bureau, the ACS data-user community, and to the general R-user community.