A Comparison of Statistical and Spatial Approaches for Environmental Risk Assessment. Robert B. McMaster, Helga Leitner, and Eric Sheppard, Department of Geography, 414 Social Sciences Bldg.,University of Minnesota, Mpls, MN 55455
Research in the assessment of environmental equity, as related to hazardous materials, has taken two basic approaches over the past decade: statistical and spatial. As applied to census data and, for instance, toxic release inventory sites, statistical techniques utilize classical inferential methods, such as correlation or multiple regression analyses, to identify potential relationships between the hazardous sites and the targeted subgroup of population, such as minorities, those located in public-assisted housing, and / or low-income groups. A finding of several of the more recent studies is that the resolution of these data (for instance block, block-group, census tract, MCD, or county) can significantly affect the result of the correlation or the model. An alternative set of methodologies has applied GIS-based spatial analysis in an attempt to ascertain whether differences exist between the densities or percentages of a given population inside or outside of some search radius (buffer). These studies often test for the effect of buffer size, or use alternative geometries, such as plume-dispersion models, in place of simple circular buffers. This study has completed a risk assessment for the City of Minneapolis using a variety of geodemographic measures and hazardous materials sites, including TRI, Superfund, Petrofund, and Land Recycling sites. Using this data set, we compare the two approaches, statistical and spatial, and discuss problems associated with the application of each.