🤖 AI Summary
This paper addresses the vulnerability of conventional GeoRDD assumptions—namely, continuity of potential outcomes or local randomization—to violation in spatial point process data. To mitigate this, we propose a robust causal inference framework grounded in weaker, more plausible assumptions. Methodologically, we develop a general-purpose robust testing procedure that integrates localized estimation near the boundary with a novel adaptive spatial resampling strategy to accurately approximate the null distribution of the test statistic. Applied to evaluating the causal effect of precinct boundaries on arrest rates in New York City, our approach detects a statistically significant and robust policy effect—contrasting sharply with findings from standard GeoRDD analyses. The framework balances theoretical rigor with practical implementability, offering a generalizable tool for causal evaluation of spatial policies in urban governance and related domains.
📝 Abstract
We study variation in policing outcomes attributable to differential policing practices in New York City (NYC) using geographic regression discontinuity designs (GeoRDDs). By focusing on small geographic windows near police precinct boundaries we can estimate local average treatment effects of police precinct practices on arrest rates. We propose estimands and develop estimators for the GeoRDD when the data come from a spatial point process. Standard GeoRDDs rely on continuity assumptions of the potential outcome surface or a local randomization assumption within a window around the boundary. These assumptions, however, can easily be violated in real applications. We develop a novel and robust approach to testing whether there are differences in policing outcomes that are caused by differences in police precinct policies across NYC. Importantly, this approach is applicable to standard regression discontinuity designs with both numeric and point process data. This approach is robust to violations of traditional assumptions made, and is valid under weaker assumptions. We use a unique form of resampling to provide a valid estimate of our test statistic's null distribution even under violations of standard assumptions. This procedure gives substantially different results in the analysis of NYC arrest rates than those that rely on standard assumptions.
