A Measure of Spatial Segregation: The Generalized Neighborhood Sorting Index.
Paul A. Jargowsky, University of Texas at Dallas; Jeongdai Kim, University of Texas at Dallas.
Download paper here.
The measurement of spatial segregation has often used measures such as the Neighborhood Sorting Index (NSI) or the Index of Dissimilarity (D) that do not take full account of space. We propose a Generalized Neighborhood Sorting Index (GNSI), an extension of NSI, in which each neighborhood is part of a broader community, the extent of which can be easily changed by increasing the order of contiguity or the allowable distance between neighborhood centroids. The GNSI has desirable characteristics as a valid spatial measure of segregation and significantly alleviates the checkerboard problem while NSI and similar measures are insensitive to the spatial arrangement of the geographic units in the analysis, such as census tracts. An application of GNSI to the ten largest metropolitan areas reveals that the GNSI gives a more complete account of the extent of economic segregation and the changes over time, and ranks the metropolitan areas differently in term of the level and changes in economic segregation.