Proxying Class in Legal Administrative Datasets with Missing Information on SES

• In Event: Justice and Injustice: Political Science Perspectives on Crime and Punishment Mini-Conference Lunchtime Poster Session: New Questions in the Politics of Punishment

Abstract

Disaggregated legal administrative datasets, such as police report instances or individual court
records, often lack information about the socioeconomic status (SES) of participants; even
though these datasets possess robust information about other descriptive attributes, such as age,
gender, race, ethnicity, etc. This makes it difficult to assess how SES, along with other
descriptive attributes, influences the legal system outcomes recorded by these administrative
datasets. To overcome this empirical dilemma, this paper proposes a new methodological
approach for creating aggregate measure scores that proxy the SES of local neighborhoods.
These aggregate neighborhood scores can then be appended to disaggregated administrative
datasets with geocodable information.
To create these scores, I extract economic information about housing units from the CoreLogic
housing database. Next, I use geographic information system (GIS) software to map the
geocoordinates of these homes and project them to a Cartesian plane. Finally, I partition the
affine subspace of these geocoordinates in order to identify relatively homogenous economic
neighborhood enclaves (low income, high income, etc.). This is achieved by iteratively using
regression trees across different rotated axes. This process creates stable economic neighborhood
configurations while also avoiding making a priori assumptions about the appropriate geographic
unit for analysis.
To test these scores, I use disaggregated legal administrative data provided by the North Carolina
Administrative Office of the Courts. These data possess geocodable information about
defendants which are then overlaid onto the spatial arrangement of the new aggregate measures.
Next, I examine how SES, as proxied by the new measures, influences several different legal
system outcomes related to lowlevel
traffic infractions, such as Terry stops, voluntary dismissal
by prosecutors, waiving appearance rights before a judge, etc.
Overall, I find that these aggregate neighborhood scores provide an effective way to proxy SES,
especially when compared to alternative aggregate measures, such as census tract derived
measures. In particular, these new aggregate neighborhood scores enable researchers to conduct
research at more refined geographic units and grants new opportunities to investigate how SES
influences legal system outcomes using large scale administrative datasets.

Author

• Marty Davidson, University of North Carolina Chapel Hill