Blattman et al. (2016)
The aim of this study is to estimate the direct and indirect (spillover) effects of a large-scale hotspot policing intervention and a large-scale municipal services intervention (broken windows treatment), and to identify any interaction between the two. It is our aim to improve the estimation of direct causal and spillover effects and assess whether hotspot policing (direct) or broken windows (indirect) reduce crime in the aggregate or merely displaces crime. Moreover, it allows to identify how large the bias induced by displacement may be in existing impact estimates at hotspots.
A) Hotspot Policing: Between 756 hotspots were randomly assigned to receive at least 90 minutes of policing per day. B) Broken Windows Treatment: Hotspots most in need for the broken windows treatment were deemed eligible for the treatment, and these hotspots were randomized to receive the broken windows treatment.
A) Hotspot Policing: The remaining 1,163 hotspots were assigned to a control group for which the police station would receive no special instructions but would be free to patrol them as they saw fit. B) Broken Windows Treatment: The hotspots deemed ineligible for the broken windows treatment were included in the study as a comparison group.
Although no data is presented, it is hypothesized that:
1) Treatment will increase policing time in treated hotspots and decrease policing time in non-treated spillover segments;
2) Treatment will decrease crime in segments assigned to hotspot policing;
3) Treatment will increase the number of times a segment is visited by the municipal team, and may also reduce crime, as well as decreases in crime for spillover units; and,
4) Segments receiving both the hotspots policing and broken windows treatments will see larger decreases in crime than segments receiving just one.