Multivariate regression has been used extensively to determine if race/ethnicity or socioeconomic status is related to presence of pollution sources, quantity of pollutants emitted, toxicity of emissions, and other indicators of environmental health risk. Most previous studies assume observations and error terms to be spatially independent, thus violating one of the standard regression assumptions and ignoring spatial effects that potentially lead to incorrect inferences regarding explanatory variables. This chapter focuses on the problem of spatial autocorrela-tion in geospatial analysis of environmental justice and explores the application of simultaneous autoregressive (SAR) models to control for spatial dependence in the data. A case study uses both traditional and SAR models to examine the distribution of cancer risk from exposure to vehicular emissions of hazardous air pollutants in the Tampa Bay MSA, Florida. Several approaches are explored to augment the standard regression equation, identify the neighborhood structure of each tract, and specify the spatial weights matrix that accounts for variations in cancer risk not predicted by explanatory variables. Results indicate that conventional regression analysis could lead to erroneous conclusions regarding the role of race/ethnicity if spatial auto-correlation is ignored, and demonstrate the potential of SAR models to improve geospatial analysis of environmental justice and health disparities.
CITATION STYLE
Chakraborty, J. (2011). Revisiting Tobler’s First Law of Geography: Spatial Regression Models for Assessing Environmental Justice and Health Risk Disparities. In Geospatial Analysis of Environmental Health (pp. 337–356). Springer Netherlands. https://doi.org/10.1007/978-94-007-0329-2_17
Mendeley helps you to discover research relevant for your work.