Is De-escalation Training Actually Backed by Evidence?

Estimated reading time: 11 minutes

Police agencies have been chasing de-escalation training for decades. The push for departments to adopt it started back when Ferguson erupted in 2014, well before George Floyd, and it only intensified after Floyd’s death in 2020. Underneath all of that pressure sat an uncomfortable gap: training is one of the least studied things in policing, and nobody had ever actually scientifically tested whether de-escalation training changed what officers did in the field.

Proponents made a reasonable case. Teach officers to slow down, read a crisis, and buy themselves options, and more encounters end without force. Critics pushed back with an equally reasonable concern: teach an officer to hesitate, and somebody gets hurt, including the officer.

In the winter of 2018, the Louisville Metro Police Department decided to buckle down and actually test a de-escalation training. LMPD executives invited a research team from the University of Cincinnati to evaluate a new de-escalation program, called ICAT, before it even rolled out agencywide, agreed to randomize which officers got trained first, and signed off on publishing whatever the results turned out to be. Nobody was requiring this of them. They asked for it, and by November 2019, they had trained 1,049 officers, 85% of their sworn force.

The Study That Tested Whether De-escalation Training Works

Robin Engel, Nicholas Corsaro, Gabrielle Isaza, and Hannah McManus published the evaluation in Criminology & Public Policy in 2022, out of the University of Cincinnati’s Center for Police Research and Policy.

ICAT stands for Integrating Communications, Assessment, and Tactics, developed by the Police Executive Research Forum (PERF). It combines critical thinking, crisis recognition, communication skills, and operational tactics, aimed at encounters where someone is behaving erratically, whether from a mental health crisis, substance use, or developmental disability, and is unarmed or carrying something less lethal than a firearm. That target matters since over 40% of fatal police encounters involve a suspect who is unarmed or armed with less than a firearm, exactly the population ICAT is built for. The training runs two consecutive 8-hour days of lecture, discussion, and scenario-based practice.

Here is what actually sets this study apart: the researchers didn’t just ask officers how they felt about the training. They measured what officers did. Earlier ICAT research had found real shifts in officer confidence and attitudes toward people in crisis, but attitudes don’t tell you what happens on a call. This evaluation tracked behavior instead: the frequency, severity, and type of force used, and the injuries that followed, pulled straight from LMPD’s mandatory use of force reports.

If “stepped-wedge cluster randomized controlled trial” makes your eyes glaze over, you’re not alone. Here’s the plain version, and why it matters: researchers grouped LMPD’s nine patrol divisions into three batches and randomly decided which batch got trained first, second, and third. Every division started untrained and ended up fully trained, just not all at the same time. That randomization is what makes the study credible. Because the researchers, not the department or the officers, decided who trained when, no division was hand-picked in a way that could bias the results, and the trained divisions stayed comparable to the untrained ones. And because the training was staggered, each phase let the researchers compare divisions that had just been trained against divisions that hadn’t been yet, which is exactly what lets them say the drop in force was caused by the training rather than something that was going to happen anyway.

What They Found: Does De-escalation Training Work?

The core results:

• Use of force incidents fell by 28.1%
• Citizen injuries fell by 26.3%
• Officer injuries fell by 36.0%

Remember the worry that de-escalation training makes officers hesitate at the wrong moment and gets them hurt? This is the data that pushes back on that claim. All three numbers are statistically significant, which in plain terms means these are real, reliable drops, not the kind of random monthly variation you’d expect to see even if nothing had changed.

Force also got less severe, not just less frequent. Comparing 2018 to 2019, the sharpest drop showed up in the hands-on categories: empty hand strikes, kicks, knee strikes, and takedowns, down 21.2%. Officer-involved shootings barely moved, but that’s expected rather than concerning. Shootings make up less than 2% of all use of force incidents at IMPD, too rare for a single-department study to detect movement either way, and ICAT was never built to touch that category since it was only meant to be used on individuals who do not have a firearm.

So what explains a drop this size? The researchers checked the obvious alternative first: maybe arrests were just falling, and fewer arrests means fewer chances to use force. Arrests did fall, about 11.5% overall, but arrests for serious violent crimes, the ones least tied to officer discretion, barely moved. That’s a signal the drop wasn’t just crime going down. And when researchers measured the two declines against each other directly, the drop in force outpaced the drop in arrests, a difference they call marginally significant, meaning suggestive rather than ironclad, but pointing the same direction as everything else here.

Then they went further. Four more possible explanations, tested one by one: a use of force policy change in October 2019, a traffic stop policy change in August 2019, COVID-19 stay-at-home orders in March 2020, and the killing of Breonna Taylor that same month. Each time, they reran the numbers with that period cut out and again controlling for it directly. The result held steady every time.

Speaking of March 2020: that’s the same month LMPD officers killed Breonna Taylor during a search warrant, two months before George Floyd’s death set off a summer of protest. It’s a hard coincidence to sit with. A department that had spent two years quietly building the best evidence in the country on reducing force was, in the same season, at the center of the tragedy that made the whole country furious about force. The researchers address this directly rather than dodge it: ICAT trains officers to slow down with someone who is unarmed and in crisis. It has nothing to do with executing a warrant or a scenario where a firearm is already involved. The training’s success in one arena says nothing about what happened in the other, and the study is honest about that gap rather than papering over it.

Worth knowing before you act on any of this:

• This is one study, of one training program, at one agency. It doesn’t tell you what works everywhere.
• The stepped-wedge design used only three groups, fewer than the five or six typical in this kind of research, which limits some precision.
• The study wasn’t built to test whether ICAT’s effects fade over time, and separate pilot research on ICAT has found some decay in officer attitudes after training, pointing to the necessities of refresher trainings.
• It measured injuries and force counts, not whether officers were actually using de-escalation tactics on scene. An officer could de-escalate a situation down a notch, say from an ECW deployment to empty hand control, and it would still count as a use of force incident here. The real effect could be understated as easily as overstated.

What This Means For Your Agency

Every recommendation here comes with an acknowledgment: training budgets are tight, backfilling two 8-hour days per officer is a real scheduling problem, and “just send everyone to training” is easy to say from a research center and hard to execute from a watch commander’s office. Here’s what the evidence actually supports.

Use de-escalation training, but be specific about which one. This study is evidence for ICAT. It is not evidence for de-escalation training as a broad category, since the field still can’t agree on what that even covers. Plenty of commercial programs claim the same benefits ICAT delivered here without anything behind them. ICAT is the one with a randomized trial. PERF provides the full ICAT training guide and train-the-trainer support at no cost: ICAT Training Guide (PERF).

Training alone will not get you there. The researchers put it bluntly: “we cannot train our way out of this problem.” Training needs reinforcement from policy, supervisory oversight, and managerial support, or the effects erode. First-line supervisors matter more than anyone else in that chain. If sergeants treat the training as a checkbox, officers will too. LMPD’s own response was to start building de-escalation training specifically for its first-line supervisors.Policy can reinforce the same behavior, sometimes cheaply. The researchers cite Dallas PD’s point-and-report policy, which simply required officers to document any time they pointed a firearm, whether or not they fired. That single documentation requirement was associated with significantly fewer officer-involved shootings and fewer cases where officers mistakenly believed they saw a weapon, with no corresponding increase in officer injuries. Small accountability mechanisms can shift behavior without new curriculum or new budget. Here is a summary of that research: Dallas PD point-and-report policy

Measure before you evaluate, not after. LMPD’s use of force reporting system already existed before this study, which is exactly why the evaluation was possible at all. If your agency doesn’t already track force incidents by type, severity, and injury outcome in a consistent way, that infrastructure is worth building regardless of which training you choose next, because it’s the only way you’ll ever know if the training worked.

Bring the community into the process, not just the announcement. The researchers recommend incorporating community feedback into both the design and delivery of de-escalation reform, not just the press release afterward. Your community isn’t asking whether you train on de-escalation anymore. They’re asking whether you can prove it works, and involving them earlier makes that answer more credible.

LMPD seeks public input on updates to use of force policy

A link for the public to provide input will be live for about two weeks, until April 12.

Final Thoughts

Agencies have been under pressure to fix use of force since Ferguson, and that pressure has only grown since. For years, the honest answer to “what training actually works” was that nobody knew. LMPD’s willingness to test their own program, before anyone was demanding it of them, is why an honest answer finally exists for ICAT specifically.

It would be tempting to end there, but Breonna Taylor’s death, at the hands of LMPD’s own officers, happened in the middle of this same story, and it deserves to be named plainly rather than folded into a footnote. She was a woman who was killed during a warrant execution that had nothing to do with the kind of encounter ICAT was built to change. This study does not tell you that a tragedy like the one that took her life can’t happen at your agency too. No training does, and no single program should be sold to you as if it does.

That’s the caution worth carrying into your next training decision. De-escalation is not one thing, and not every program wearing that label has evidence behind it. ICAT does, at least at one agency, for now, and, importantly, that evidence points toward officer safety rather than away from it. Treat that as a strong reason to look closely at ICAT specifically, and a strong reason to be skeptical of anyone selling you “de-escalation” as a solved problem.

For the fuller picture beyond this one study, including which de-escalation programs still have no evidence behind them at all, one of this study’s own lead authors, Dr. Robin Engel, published an updated field-wide review in 2026:

Reassessing police de-escalation training: Evidence, gaps, and policy priorities – Niskanen Center

In the post-2020 policing landscape, a multifaceted picture is emerging regarding law enforcement agencies’ attempts to institutionalize de-escalation training and practices in their daily operations, and the limited evidence available to support these efforts.

Enjoyed this post?

Get more research translated into plain language delivered straight to your inbox.

Subscribe below and let me know which topics you want to hear more about. It only takes a minute.

If you found this helpful, share it with your team or on your socials. It helps more people make sense of research that matters.

Carlee Ruiz

Carlee Ruiz is the founder of PoliceResearchHub, a platform that turns academic research into practical insights for law enforcement. She brings seven years of experience as a crime analyst and justice researcher, with a focus on evidence based policing, gun violence prevention, and public safety strategy. Carlee currently works as a researcher at National Policing Institute, where she supports evaluations of law enforcement programs and criminal justice initiatives across the country. Through PoliceResearchHub, she aims to make research easier to understand and more useful for the people doing the work on the ground.