|
Activity Number:
|
137
|
|
Type:
|
Invited
|
|
Date/Time:
|
Monday, August 3, 2009 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Statistics and the Environment
|
| Abstract - #302964 |
|
Title:
|
A Closer Look at Statistical Inference with Geographically Weighted Regression Models
|
|
Author(s):
|
Carol Gotway Crawford*+ and Linda J. Young
|
|
Companies:
|
CDC and University of Florida
|
|
Address:
|
1600 Clifton Road NE, Atlanta, GA, 30333,
|
|
Keywords:
|
spatial regression ; spatial random effects ; mapping
|
|
Abstract:
|
Geographically Weighted Regression (GWR) models are local regression models that allow covariate effects to vary in space. Most applications of GWR are focused on visualizing and mapping spatially-varying regression coefficients. In this presentation, we describe inferential methods for geographically weighted regression models, focusing on hypothesis tests involving the spatial regression parameters. We frame these methods around data collected to assess the association between myocardial infarctions (MIs) and ambient ozone levels in Florida for environmental public health tracking at the Florida Department of Health. Using these data, we compare results from hypothesis tests involving spatial regression parameters in the GWR model to those from spatial random effects models and discuss the advantages and disadvantages of both approaches.
|
