The Boundaries of Spatial Inequality: Three Essays on the Measurement and Analysis of Residential Segregation

Scholars have engaged in a longstanding debate about how best to measure residential segregation, and scores of summary indexes have been developed in response. The dissimilarity index came into widespread use during the mid-20th century and remains the most popular measure. In recent years, the information theory index has become a common alternative and has several advantages over other indexes. Segregation can be calculated for any number of groups and decomposed into the segregation occurring within and between groups or subregions. In the first essay, I demonstrate that although the information theory index can be decomposed into additive components, it is misleading to interpret the results as segregation. I propose a new measure, the divergence index, that improves on existing measures of inequality and segregation. It measures the difference between an empirical distribution and a distribution that represents equality. It is a measure of surprise: how surprising is the composition of a local area given the overall population composition of the city? There is no segregation if all local areas are a microcosm of the city's population. I show that the divergence index is a conceptually intuitive and methodologically rigorous measure of inequality and segregation. It addresses the limitations of existing measures and accurately decomposes segregation across contexts and nested levels of geography.;The dominant focus of quantitative studies of residential segregation has been on how individuals are organized into places, primarily neighborhoods, cities, and metropolitan areas. The methods commonly used to measure segregation are aspatial. They summarize certain characteristics of segregation, such as concentration or clustering, but they ignore the spatial features of segregation patterns, such as how neighborhoods are spatially arranged in a city. As a consequence, many studies find the same level of segregation whether a city has a patchwork of racial or ethnic enclaves, or is divided into large areas with little or no diversity. In recent years, there has been heightened interest in developing methods that capture the spatial proximity of neighborhoods and the geographic scale of clustering. However, recent approaches lack a realistic measure of distance, and do not accurately represent how segregation varies within cities. In the second essay, I introduce a new method for studying the spatial context of residential segregation. My approach reorients the study of segregation from place as the focal organizing unit to space as the medium of organization. I measure the distance between locations and the reach of local environments along the city's road network. Road distance is a more realistic measure than straight line distance (i.e. "as the crow flies"), because it captures the connectivity of the road network and the excess distance imposed by spatial boundaries.;Spatial boundaries, as a focus of analytic interest, have been largely absent from the quantitative segregation literature. In the third essay, I apply my new method and measure in an empirical analysis of residential segregation in U.S. cities. Results reveal that cities previously thought of as comparable, are actually quite different. Even cities with similar aspatial segregation scores, nonetheless can have different spatial patterns of segregation. I demonstrate the salience of city boundaries for structuring segregation patterns, and how the presence of physical boundaries, like a river, affects the composition of local environments. Further, the experience of segregation within some cities is highly unequal. Some residents experience completely segregated local environments, while others live in areas as diverse as the city. In these cities of extremes, local segregation is simultaneously better and worse than indicated by overall city segregation. An emphasis on spatial boundaries and local context reframes our understanding of segregated environments. My results provide deeper insight into the segregation of even the most studied U.S. cities, including Philadelphia, New York, and Milwaukee.;This dissertation provides the first quantitative method for studying how spatial boundaries structure patterns of residential segregation. The approach can be applied to systematically analyze how the presence and impact of spatial boundaries varies within and across cities and over time. Beyond simply marking where one place ends and another begins, spatial boundaries structure political power, physical access to locations and services, and the resources and opportunities available to residents. This work attempts to bridge qualitative insight on the lived experience of unequal social environments and how we measure segregation for city populations. It lights a path forward for studying spatial boundaries as a mechanism of social and spatial inequality and suggests multiple avenues for future work.