Using advanced data analytics to assess the spatial causal effects of policing policies and practices

Brief outline of the research agenda
The theory of procedural justice is built on the idea that when people evaluate their interactions with the police, they are primarily focussed on whether or not the officer makes objective and neutral decisions and treats them in a fair and respectful manner. When people are treated in a procedurally just manner, they tend to find the authorities morally appropriate and give consent to their actions and demands even when they disagree with them. In turn, when people view legal authorities as proper and just, they feel a normatively grounded duty to comply with the law and cooperate with the police and criminal courts. The concept of legitimacy lies at the heart of democratic policing, in that in a democratic society police must seek and maintain public support by acting impartially, using coercion proportionately and persuading the citizenry that they are an institution that is entitled to be obeyed.
Yet, in the procedural justice literature, most of the empirical evidence gathered so far is observational in nature, and rely on the interpretation of statistical associations. In fact, there is a dearth of research systematically assessing the causal claims made by the theory. In a recent review of the literature, Nagin and Telep (2017: 18) voiced their concern: "What has not been established is whether these associations reflect a causal connection between procedurally just treatment and perceived legitimacy and compliance." However, without empirical research demonstrating robust causal relationships, it is difficult to devise successful policy initiatives.
Thus, my principal aim with this fellowship is to test and advance theoretical understanding of some core causal claims of the policing literature. Specifically, I will scrutinise neighbourhood-level and location-based police effects. There is a substantial heterogeneity in the citizens' experiences and views regarding police officers but it is yet unclear to what extent this can be attributed to varying policing strategies in different neighbourhoods. By using geo-coded administrative police data, and merging it with public attitudes surveys, my research can identify policing practices that work best in particular neighbourhoods, to provide tailored recommendations to police forces. To identify causal effects, I will use state-of-the-art causal inference techniques, multilevel matching and location-based regression discontinuity designs. In a nushell, the current proposal plans to address one crucial aspect of procedural justice policing: how the effects of policing initiatives vary across neighbourhoods with different characteristics?