An Evaluation of a Social Interaction Training Program to Reduce the Use of Force and BuildLegitimacy

Statement of the Problem: As identified in the President’s Task Force, we lack empirical evaluation concerning police training effectiveness, particularly for de-escalation and use of force reduction programs. We propose to examine such a program that stems from a DARPA initiative—"Tact, Tactics, and Trust” (“T3”). This training has components grounded in procedural justice, trust, and legitimacy. Our study will have implications for police use of force knowledge/training and advance our theoretical understanding of how procedural justice and related strategies can be trained and whether they impact police outcomes.

Subjects: The research subjects will be patrol officers in two police agencies.

Partnerships: The project involves partnerships between UofSC, UTEP, Polis Solutions, and the Fayetteville
(NC) and Tucson (AZ) Police Departments. Members of the research team and Polis Solutions have worked with both police agencies in the past.

Research Design and Methods: A randomized-controlled trial will be used to test whether T3 is associated with reductions in officer-level use of force, related injuries, and citizen complaints. Officers will be randomly assigned to high dosage, low dosage, or no T3 training. Outcomes of interest will be tracked 12 months prior
to and after the intervention. We expect high dosage experimental officers to have greater declines in outcomes of interest. We will also use officer pre- and post-test surveys and expect experimental officers to
have greater willingness to use procedural justice and believe the public has greater trust in them compared to control groups. In-depth officer interviews will allow us to contextualize the findings. This mixed-method evaluation will take place over 27 months.

Analysis: Difference-in-difference tests will serve as a preliminary comparison of the experimental and control groups across the various outcomes of interest. Interrupted time-series will be used to assess trend changes in the dependent variables across the groups. Within-officer and between-group change over the observation period will be assessed using multilevel models to leverage the longitudinal data and experimental design. Multivariate models will account for potential confounding variables (e.g., officer productivity). Appropriate regression models will be used in the pre- and post-test survey analyses. ATLAS.ti
will be used to help analyze the qualitative interviews.

Products, Reports, and Data Archiving: In addition to the final report, we anticipate that this study will result in a number of publications, both in peer-reviewed journals and practitioner-based outlets (e.g., The Police Chief). All de-identified data will be archived in a manner consistent with NIJ’s requirements. ca/ncf