Evaluating a city-led intervention under severe data constraints: A frontier matching approach to analyze resident reporting behavior
Using a novel matching method with a large, matched sample (n=77,955), we present estimates on the effect of a city-led intervention—which pairs community-oriented policing with citizen engagement and highly visible service provision—on reporting behavior of traditionally underreported 911 and 311 incidents. We find that treated households were 32% more likely to have a drug-related crime reported than untreated households within 3 months of the intervention (risk ratio = 1.321, 95% CI [1.034, 1.687], p=0.026) and 42% more likely within 5.5 months compared to untreated households (risk ratio = 1.415, 95% CI [1.162, 1.723], p=0.0005). In examining calls related to services provided during the intervention, treated households were 9% more likely to have an intervention-related service need reported than untreated households within 3 months of the intervention (risk ratio = 1.090, 95% CI [1.007, 1.180], p =0.034), with similar results within 5.5 months (risk ratio = 1.088, 95% CI [0.959, 1.055], p=0.014). We describe strategies to navigate various challenges in the use of city administrative data for research and offer potential interpretations for these findings, which may be driven in part by increased trust in institutions and increased visibility of classically “submerged” governmental service provision.
Date: 24 May 2023, 14:00 (Wednesday, 5th week, Trinity 2023)
Venue: Manor Road Building, Manor Road OX1 3UQ
Venue Details: Skills Lab
Speaker: Eleanor Dickens (University of Oxford)
Organising department: Department of Economics
Part of: Novel Ideas: MPhil Seminar Series
Booking required?: Not required
Audience: Members of the University only
Editors: Emma Heritage, Daria Ihnatenko