Objective. In environmental justice research, different-sized units of analysis have generated mixed results, begging the question of what constitutes the most appropriate unit of analysis. Grappling with this question raises both conceptual and methodological issues. In this article, it is argued that, conceptually, community should be the most important unit of analysis and that, methodologically, units should be selected through an appropriate sampling procedure. These issues are addressed through an investigation of the locations of industrial hog farms in Mississippi. Methods. A geographic information system procedure was used to identify community areas and to select the sample of units. The identified community areas were compared to counties, ZIP code areas, census tracts, and census block groups through a bivariate and multivariate logistic regression. Results. Race was the only variable that was consistently not found to be significant across all units. For the remaining variables—income, education, and industry—results were mixed across units. Conclusions. This research confirms that units of different size generate different results. We argue that the decision about the most appropriate unit of analysis should be conceptually rather than statistically determined and that community is the most important unit of analysis because it holds both legal and social authority to raise concern about environmentally controversial facilities.