In 2019, NPI, in partnership with George Mason University, was awarded a grant to conduct a randomized field experiment exploring policing strategies to address violent crime hot spots. The experiment was originally designed to test whether an enhanced version of pedestrian stops informed by constitutional and procedural justice principles could significantly affect crime in crime hot spots while enhancing perceptions of procedural justice among those stopped and those living in the hot spot areas. The study was designed in partnership with the Phoenix Police Department (PPD).

Since the original scope of work was developed, considerable changes in policing strategy, police-community relations, and challenges posed by nationwide staffing shortages in police departments have occurred, leading to a change in the project’s original scope.

Under the revised project scope, NPI is evaluating the impact of the procedural justice and constitutional policing training provided to PPD officers using AI-driven research tools. In partnership with Polis Solutions, Inc., we are utilizing a multi-modal AI program to measure the key features of procedural justice exhibited during interactions between police officers and community members captured on body-worn camera (BWC) video. This program was developed under NIJ grant 2020-R2-CX-0010, “Multi-Modal Analysis of Body Worn Camera Recordings: Evaluating Novel Methods for Measuring Police Implementation of Procedural Justice.”

The results will let us determine whether the training changed the behavior of PPD officers in the field and provide additional validation for the software. Using advanced AI tools and analytics to evaluate procedural justice can allow law enforcement agencies to assess training impact and officer performance more robustly and efficiently. These tools can comprehensively process large volumes of BWC footage, making it possible for supervisors to assess the quality of officers’ interactions over more simplistic output metrics like the number of arrests or tickets issued. The insights gained can subsequently help law enforcement organizations inform training programs, policy adjustments, and early warning systems.

A large-scale community sentiment survey of Phoenix residents and an officer pre- and post-training survey will supplement these findings. This project is particularly relevant for law enforcement agencies aiming to enhance accountability and transparency in police-community interactions.