|
Abstract:
|
Community policing has become a popular program around the world. In contrast to its popularity, little work has been done to rigorously evaluate its effectiveness. In this paper, we examine four policy-relevant causal questions regarding the effect of community policing on Chicago's personal crime rates. To answer these questions, we extend Rubin's causal framework, which defines causal effects as unit-specific differences between potential outcomes under two alternative treatments. By employing Cox's "no interference between different units" assumption, Rubin's framework ensures that each unit has only one potential outcome per treatment. However, this assumption is not valid in spatial settings, where the treatment assignment of one's neighbors may affect one's outcome. We redefine potential outcomes and causal effects, and discuss assumptions that make this framework tractable.
|
