Interpreting Regression Coefficients Through a Spatial Scale Lens

• Activity Number: 344
• Type: Contributed
• Date/Time: Tuesday, August 5, 2014 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics and the Environment
• Author(s): Garritt Page*+ and Brian Reich
• Companies: Pontificia Universidad Católica de Chile and North Carolina State University
• Keywords: Spatial regression ; spatial scale ; CAR model ; areal data

Abstract

We consider spatial regression models for areal data and investigate settings in which a covariate's influence on a response depends on spatial scale. To this end, we develop a methodology that is flexibly able to estimate covariate effects as a function of spatial scale. We show that the eigen-value decomposition of neighborhood structures commonly used in areal data models can be employed to shed light on how covariate estimates can change as spatial scale moves from global to micro. We demonstrate the methods utility using synthetic and real data.