2023 Spatial Data Science Symposium

The use of granular spatial data to examine geospatial mobility in social science research

Panel description: Human mobility, including physical travel of people and goods and virtual transactions of information, plays a significant role in (re)distributing uneven resources in cities and (re)organizing urban demographic, social and economic structures. The recent explosion in the availability of granular spatial data has made it possible to capture movement across space at very refined spatiotemporal levels for a large sample of cities across the world. This has allowed researchers to expand our understanding of social processes that are inherently linked to human mobility, including disease transmission, accessibility to amenities and resources, segregation dynamics, and exposure to environmental hazards, beyond what has been discovered using traditional forms of data. This panel discusses the use of granular spatial data in studying geospatial mobility, focusing on its application in social science research, types, sources and methods, and its theoretical and methodological advantages and limitations.


Noli Brazil, Associate Professor, Department of Human Ecology, University of California, Davis

Jennifer Candipan, Assistant Professor, Department of Sociology, Brown University

Speaker Lineup

Peter Rich, Assistant Professor

Jeb E. Brooks School of Public Policy, Cornell University

Eligible but not accessible? A new method for measuring neighborhood educational access in the era of school choice

In an era of expansive school choice, it is increasingly uncommon that a child’s residential address will directly determine the public school that they are eligible to attend. Nearly half of all families live in areas where they are free to choose from an array of eligible public school alternatives. But is more choice a source of empowerment when attractive schools require long commutes, administrative hassles, or competitive lottery processes? Where are marginalized populations able to access high-quality schools without incurring these added burdens? Although prior work has demonstrated the salience of various practical constraints on family school choice decisions, there has been little effort to incorporate these factors into spatial measures of access educational opportunity. This study introduces a new methodological approach that identifies and summarizes the characteristics of all public elementary schools that residents of a given census block are eligible to attend, weighed by each school’s relative accessibility. Our approach leverages observed enrollment data from student administrative records in Michigan that include information about where children live and where they attend school. We estimate a discrete choice model to identify the institutional, geographic, and infrastructural factors that shape these children’s school enrollment outcomes, while also controlling for the influence of family preferences for school quality and composition. As we show, the enrollment probabilities predicted from this model provide a new and intuitive way to summarize the quality and diversity of school options that families encounter—whether in rural and suburban environments with sparse options or in urban locales where the choice sets are dense. This approach can be used to describe neighborhoods by any sort of measurable school-level attribute, such as the average test performance level, the average test growth score, per pupil funding, peer racial and socioeconomic composition, etc. Furthermore, the estimated model parameters can be used beyond the Michigan student sample to predict neighborhood educational access anywhere with suitable publicly available data.

Brian Levy, Assistant Professor

Department of Sociology, George Mason University

Concentrated Disadvantage, Isolated Affluence: How Neighborhood Mobility Networks Compound Residential Segregation by Socioeconomic Status

Recent research identifies socioeconomic segregation in everyday mobility between neighborhoods as an important predictor of neighborhood vitality, with impacts that compound disparities by residential segregation. The level of socioeconomic (dis)advantage among a neighborhood’s residents is correlated with the level of (dis)advantage in its routine mobility network, but the relationship is hardly deterministic. So, what predicts whether the residents/establishments in an impoverished or affluent neighborhood visit and receive visits from similarly impoverished or affluent neighborhoods? Do features of the neighborhood or the broader commuting zone play a greater role in socioeconomic mobility segregation/integration? This presentation uses 2019 data from SafeGraph and the American Community Survey to answer this question for neighborhoods across the United States.

Joel Han, Assistant Professor

Quinlan School of Business, Loyola University Chicago

Gasoline Prices and Income Disparities in Daily Travel Patterns

We study the effect of gas price fluctuations on regular travel patterns – specifically the frequency of travel and the mean distance traveled – for 10 major cities in the U.S. and the time period March 2018 — May 2021. Our focus is on how the effect of changing gas price differs across neighborhoods by income level. Using aggregated cell phone geolocation data, we calculate the total number of outgoing visits and the weighted mean home-destination distance for every home neighborhood (census block group) in our sample. We merge this data with gas price data from the U.S. Energy Information Administration. We document a significant difference between the pre-pandemic and pandemic periods: During the pre-pandemic period, rising gas prices result in a slightly reduced gap in outgoing visit frequency between high-income and low-income neighborhoods. During the peak pandemic period, rising gas prices instead increase this gap by a larger amount. With mean distance, we find small differences in the response to gas price by neighborhood income, both before and during the pandemic. This implies that effects on mileage traveled are similar to those for the frequency of visits. We supplement these findings with an analysis of the short-run impact of changes to state gas tax rates, which affected three cities in our sample. Using synthetic control methods, we find that increases in the gas tax rate have ambiguous effects on the visit frequency gap, suggesting that the impact of policies affecting gas price could have varied impacts on travel-related inequality depending on local city factors.

Jonathan Tollefson, Ph.D. Candidate

Department of Sociology, Brown University

The environmental dimensions of urban inequality, 1880-1930

This research investigates the role of urban environmental hazards during the initial formation of neighborhood-scale racial inequality in the late 19th and early 20th centuries. Neighborhood-scale inequality is a relatively recent feature of U.S. cities: Racial segregation was primarily a street-level phenomenon in the late 1800s, and the color line only grew to encompass entire neighborhoods in the first decades of the 20th century. Research on urban environmental inequality, however, is left-censored to about the 1970’s, primarily due to a lack of comprehensive environmental site data prior to the establishment of the environmental state. In response, this research leverages an original computational pipeline to identify and geolocate sites related to a particularly noxious source of early industrial pollution. These sites are paired with geolocated historic census data to measure the social stratification of industrial exposure in six US cities over the 1880 to 1930 period. Results suggest that proximity to environmental hazards emerges as a key variable in the reorganzation of urban space, as the iterative movement of people and industry across the city produced new patterns of environmental exposure over time – providing the first empirical assessment of the environmental dimensions of this key moment in the spatial transformation of urban inequality.