Drones as First Responders and Police Response Time: Evidence From Two Randomized Controlled Trials

Drone as first responder (DFR) programs send drones to 911 calls ahead of, or instead of, patrol officers, yet causal evidence of efficacy is lacking. We report two preregistered randomized controlled trials of DFR. Philadelphia randomized 1,546 calls; Montgomery County randomized 151 days. In Philadelphia, assignment to a drone response reduced receipt-to-closure service time, the featured co-primary outcome under the preregistration’s missing-data provision, by 14.5 minutes (16.6%). The preregistered travel-time estimate, measurable for only a quarter of calls because responding officers rarely log arrivals, was negative but inconclusive. Exploratory decomposition indicates that most of the total-time difference accumulated before first dispatch. In Montgomery County, operating sites reduced median first arrival by 1.33 minutes (10%) across all eligible calls, including those receiving patrol-only response. On attended calls, drones arrived 4.6 minutes after call receipt and reached the scene before patrol dispatch in 40.9% of linked calls. Patrol arrival was also shorter across all in-catchment calls (−0.81 minutes), despite drones reaching only 6.5% of eligible calls. Among noneligible calls, the reduction was clearest for Priority 2 (−2.23 minutes); the overall estimate across priorities was smaller with more uncertainty (−0.58 minutes). Effects persisted under intent-to-treat, remained confined to catchments, and increased with deployment volume. Across the trials, DFR improved response, with patterns consistent with direct response benefits and patrol workload relief, although staffing and fleet capacity constrained benefits.