Spatial segregation, segregation indices and the geographical perspective

What could be more inherently geographical than segregation? However, the richness of the spatial variations in segregation is seldom captured by the dominant genre of empirical research. Returning the 'geography' to segregation research, we argue that local areas need to be given considerably more attention, using measures that explicitly reveal the spatial fabric of residential clustering along racial/ethnic lines. We first critique global measures such as the Dissimilarity Index and its spatial counterparts. Attention then turns to local measures such as the Location Quotient and Local Moran's I, applying them to Franklin County, Ohio, the core of Columbus MSA (Metropolitan Statistical Area). Our interpretation of the findings also employs local knowledge concerning neighbourhood characteristics, ongoing urban processes, historical occurrences, and the like. Thus, while local indices based on secondary data expose the terrain of clustering/segregation, follow-up fieldwork and/or secondary data analysis in a mixed-methods framework provides a better understanding of the ground-level reality of clustering/segregation. Tangible evidence of the gain from this approach is provided by our evaluation of conventional frameworks for understanding racial/ethnic aspects of residential patterning - assimilation, stratification and resurgent ethnicity - and in our proposal for a new framework, 'market-led pluralism', which focuses on market makers who represent the supply side of housing. Copyright (c) 2006 John Wiley & Sons, Ltd.