Permeating Segregated Spaces: A Measure of Dynamic Segregation

• In Event: Dynamics of Racial Separation and Solidarity

Abstract

Social boundaries between groups are a central feature of politics across the world. Paramount to many of these studies is how different groups are spatially configured – whether it concerns the ethnic security dilemma, patterns of ethnic patronage, public goods provision, ethnic voting, or the overall resilience of ethnic politics, each of these foundational theories of ethnic politics embeds some assumptions about how different groups are distributed across space. Yet, our tools of measuring groups’ spatial distribution remain coarse. Relying on some variant of the ethno-linguistic fractionalization (ELF) index to measure ethnic heterogeneity or the dissimilarity index to capture segregation, political scientists restrict their purview to diversity and segregation so defined.

Though other criticisms of these types of measure abound, I focus on their inability to capture individuals quotidian movement outside of their residences. Because these demographic measures often rely on census data, they focus primarily on only one component of individuals’ experiences of ethnic diversity: their residential proximity to members of other ethnic groups. Research across a variety of disciplines, have shown that people spend a large amount of their time outside of their residential contexts, traversing a variety of neighborhoods with changing demographics along the way. A variety of new methods and measures have been recently developed to capture this dynamic type of segregation (see Atheys et al. 2021 and Kwan 2013 as examples). However, these methods are quite data intensive, requiring call detail records or individual travel diaries to effectively capture individuals penetration of spaces outside their residence.

I offer an alternative measure that combines open source street network and census data to capture dynamic exposure to other ethnic groups. In particular, I measure exposure to out-groups as a function of a particular area’s location in the broader street network and its connectivity to areas holding non-coethnics. To do this, I rely on a classic measure of connectivity in network analysis, betweenness centrality, which captures the degree to which a particular location is a common link to other locations and has been shown to effectively capture pedestrian flow. Building on this classic measure, I create its weighted counterpart which takes into account the ethnic demography of its connecting locations. Intuitively, the measure incorporates the degree to which an area serves as a bridge connecting two destinations and then weights that connection by the number of people in those destinations that are from the same ethnic group as those in the linking area. I test the validity of this measure using simulated data as well as survey data to show that the measure effectively captures dynamic inter-ethnic exposure and has implications for perceptions of inter-ethnic inequality.

Author

• Paige Bollen, Massachusetts Institute of Technology