Trial Phoenix: Drone as First-Responder Study in San Francisco Police Department

Last registered on April 01, 2026
View Trial History

Pre-Trial

Trial Information

General Information

Title
Trial Phoenix: Drone as First-Responder Study in San Francisco Police Department
RCT ID
AEARCTR-0018198
Initial registration date
March 25, 2026

Initial registration date is when the trial was registered.

It corresponds to when the registration was submitted to the Registry to be reviewed for publication.

First published
April 01, 2026, 9:32 AM EDT

First published corresponds to when the trial was first made public on the Registry after being reviewed.

Locations

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information

Primary Investigator

Name
Carolyn Coles
Affiliation
San Francisco Police Department
Contact Primary Investigator

Other Primary Investigator(s)

Additional Trial Information

Status
In development
Start date
2026-04-27
End date
2026-09-01
Keywords
Crime, Violence, & Conflict
Additional Keywords
police, technology, unmanned arial vehicle
JEL code(s)
Secondary IDs
Prior work
This trial does not extend or rely on any prior RCTs.
Abstract
Police agencies are looking to technological innovations as force multipliers. Drones as first-responder programs are increasingly popular options. This study evaluates a Drone First Responder (DFR) pilot program implemented in the San Francisco Police Department to assess its impact on service delivery for lower-priority calls. The quasi-experiment mirrors standard operations to inform command-level decisions on DFR expansion.
Eligible incidents include a subset of lower-priority calls deemed appropriate for drone response. Within the treatment district, drone deployment is randomized among eligible calls to support causal inference. The pilot study is scheduled for 90 days following an initial implementation and stabilization period to ensure data quality.
Primary outcome measures include call wait time, call response time, and call disposition. These outcomes will be compared across three conditions: DFR-assisted calls in the treatment district, non-DFR calls within the treatment district, and comparable calls in the control district. Findings will inform the effectiveness of DFR programs in improving efficiency and operational outcomes in urban policing contexts. Implications for conducting internal quasi-experiments before scaling technological innovations will be discussed.
External Link(s)

Registration Citation

Citation
Coles , Carolyn . 2026. "Trial Phoenix: Drone as First-Responder Study in San Francisco Police Department." AEA RCT Registry. April 01. https://doi.org/10.1257/rct.18198-1.0
Sponsors & Partners

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information
Experimental Details

Interventions

Intervention(s)
Unmanned aerial vehicle (UAV) or drone will be used to clear lower priority calls in one police district during the pilot study period.
Intervention Start Date
2026-05-18
Intervention End Date
2026-08-31

Primary Outcomes

Primary Outcomes (end points)
(1) call time: tracking the amount of time a call takes to clear from call to cleared call (2) call wait time: the length of time a call waits from dispatch to on-scene time (3) call disposition: the difference between the outcomes of the call the UAV responds to and the calls the officers respond of the same type.
Primary Outcomes (explanation)

Secondary Outcomes

Secondary Outcomes (end points)
(4) officer safety
Secondary Outcomes (explanation)
Considering officer do not have to go to certain calls, are they safer as a results. This will be measured by looking at the call codes that produce officer injuries and comparing them to the call codes that have been removed from an officers duty list because of UAV implementation.

Experimental Design

Experimental Design
The 10 district stations were ranked by their number of lower level calls for service. Then that list was cross referenced with the shifts to identify which shift periods carry the largest loads of lower level calls. The top four districts were elevated as options for the pilot study. Command staff talked through the feasibility of each of the four districts and two were identified as the best locations for treatment and control stations. It was determined that the calls for service would be able to be randomized. A list of call codes for the lower priority calls were given to several decision-makers to identify UAV eligible calls. A subset of lower priority calls for service were approved for the pilot study. Those call codes will be used in the pilot and randomized for intervention. Each pilot will flip a coin to determine if a UAV will be deployed to handle a call for service from the approved list of call codes. A tracking document will show which calls for service were treated and which were not.
Experimental Design Details
Not available
Randomization Method
randomization done by coin flip
Randomization Unit
individual calls for service
Was the treatment clustered?
No

Experiment Characteristics

Sample size: planned number of clusters
2500 calls for service per month (during pilots working hours)
Sample size: planned number of observations
2500 calls for service monthly
Sample size (or number of clusters) by treatment arms
2500 calls for service treatment, 2200 calls for service control * 3 months of observation
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
Supporting Documents and Materials
IRB

Institutional Review Boards (IRBs)

IRB Name
IRB Approval Date
IRB Approval Number
Analysis Plan