ABSTRACT
The statistical analysis of spatial data forms the subject matter of “spatial statistics.” The academic subject matter of Geographic Information and Analysis comprises three principal components: geographic information systems (GISs), spatial statistics, and classical spatial analysis. GISs constitute a powerful new technology that can address many information needs of decision makers working with geographically (geo-) referenced data—data that are tagged, or identified, by locational coordinates. Cordy and Griffith found that, in general, the need to take spatial autocorrelation into account in variance estimation for geo-referenced data tends to negate advantages due to computational simplicity affiliated with the use of ordinary least squares estimation. For the common case where the spatial autocorrelation parameter needs to be estimated, some of the potential gains in efficiency by employing spatial statistical estimators are not realized. A spatial scientist needs to properly analyze the variablility of data over space at an appropriate scale.
