Comparing Traditional and Spatial Segregation Measures: A Spatial Scale Perspective¹

Abstract

Measuring the level of segregation often encounters two methodological issues: measures are sensitive to changes in the geographical scale of the data and the effectiveness of the measure in reflecting spatial segregation. Several spatial measures have been suggested to measure spatial segregation, but whether they are more or less sensitive to changes in spatial scale has not been investigated, while some spatial measures are relatively scale-insensitive. Using the 1990 Census data of 30 selected U.S. metropolitan areas, this paper demonstrates that these spatial measures, similar to the aspatial measure, report higher levels of segregation when smaller areal units are used in the analysis. Some spatial measures are even more sensitive to scale changes than aspatial measures. Certain patterns of the scale sensitivity were identified, but no general rules can be formulated. A preliminary explanation of the scale effect on spatial segregation measures is offered. [Key words: the MAUP, scale effect, dissimilarity index, spatial segregation measures, multigroup comparison.]

Residential Racial Composition, Spatial Access to Care, and Breast Cancer Mortality among Women in Georgia

“…Some researchers suggest that census tracts may be too arbitrary in terms of their scale and boundaries to explain the mechanism through which local racial composition and segregation impact health outcomes. 36,[63][64][65] While census tracts are unlikely the ideal measure of experienced residential environment, the high correlation of tractlevel exposures and health outcomes 66 supports their use as neighborhood proxies. Second, our study relied on the GCCR, a pre-existing database, and as such our analyses were subject to the limitations of the original dataset.…”

School segregation in Jefferson County and the affiliated Louisville metropolitan area, USA

“…Previous studies of racial/ethnic segregation have also contended that the magnitude of segregation in a specific urban area is sensitive to the delineation of the geographic boundaries of that area and to the unit of analysis to which demographic data have been aggregated (Reardon & O'Sullivan, 2004;Reardon et al, 2009;Reardon, Mathews, O'Sullivan, & Lee, 2008;Wong, 2004). The last several decades have witnessed substantial declines in degrees of segregation in nearly all central cities especially those with mandatory or voluntary school desegregation policies, however, the overall levels of segregation measured from a perspective of the entire metropolitan area have been on the rise in recent years (Bischoff, 2008;Frankenberg, 2009;Logan et al, 2008).…”

“…Until recently, most studies of segregation have used the dissimilarity index, an estimation of the evenness between two demographic groups across sub-areas within an urban community, as the primary measure of segregation (Logan et al, 2008;Wong, 2004). However, critics argue that classic segregation measures such as the dissimilarity index are aspatial because they merely capture a statistical summarization of racial/ethnic disparities in a study area but fail to take the spatial aspects of different demographic groups into consideration, thereby neglecting the geographic dimensions of segregation (Brown & Chung, 2006;Gilliland & Olson, 2010;Kaplan & Douzet, 2011).…”

“…Among various innovations in GIS-oriented spatial measures of segregation, the S index not only summarizes the statistical degree of segregation in an urban community, but also allows the visualization of the geographic and longitudinal characteristics of different racial/ ethnic groups across the same study area (Wong, 1999;Wong, 2004).…”

Progress in Spatial Demography

“…While the Index of Dissimilarity was originally applied to a comparison of two different population groups, most often whites and blacks, recent studies have extended this measure to the multi-group context (see Reardon and Firebaugh 2002 for an overview). Geographers have extended the Index of Dissimilarity in a different way by explicitly incorporating the spatial dimension (Wong 2002(Wong 2004). An emerging preference in the race/ethnic segregation literature is another evenness measure, the Entropy Index (or the Information Theory Index).…”