FOCUS - Evidence-Based Threat Recognition
Host Training

Pre-Attack Indicators: Why Police Training Has Almost No Research Behind It

Pre-attack indicators, also called pre-incident indicators, are behavioral cues officers are trained to read as signs that someone is about to attack. Nearly every academy teaches them. Almost none have been tested against real outcomes.

Ask almost any police officer what a pre-attack indicator is, and you will get a confident list. The bladed stance; The target glance; Clenched fists, a clenched jaw, the sudden weight shift, the thousand-yard stare. Academies teach some of these cues, and they show up in use-of-force reports every day.

Now ask a harder question: how do we know these cues predict an assault? Which ones, how often, and how often do they show up when nobody attacks anyone?

The honest answer is that we don’t know. After more than four decades of teaching, the published evidence that these indicators actually predict violence fits on a single page. That is not a knock on the officers who rely on them. It is a failure of the profession, the research community and the people who fund both.

The most detailed look at the cues officers trust is a study published in 2014 in the Journal of Police and Criminal Psychology. Researchers surveyed 129 officers from a Midwestern county department, every one of whom had been physically assaulted on the job. Each officer rated 23 commonly cited behaviors on a 1-to-7 scale for how worried each would make them that an argumentative person was about to turn violent.

Three behaviors stood out: a boxer’s stance, invading personal space, and hands in pockets. Every officer rated each at 6 or 7. Clenched hands, verbal threats and glancing around came next. Crying, rapid blinking, hands on hips, and avoiding eye contact rated lowest. Officers assaulted in the past year answered almost exactly like officers assaulted years earlier. Their ratings also closely matched an earlier survey of 178 college students given the same 23 behaviors.

That last finding is the problem. Veteran officers who had been attacked read these cues the same way students with little experience of violence did. The study measured what people believe predicts an attack, not what actually does, and the author says so directly: a perception that a behavior signals violence does not prove it. When the same research team searched for the evidence behind cues sold in popular training books and websites, they found hundreds of sources claiming scientific backing and almost none citing any. They concluded the empirical literature was virtually nonexistent.

The other serious paper cuts the opposite way. In 2017, Kimberly Barsamian Kahn, Jean McMahon and Greg Stewart reviewed 15 pre-attack indicator trainings and materials used by U.S. police, from active training courses to a state DOJ guide. All 15 taught at least one cue that also shows up in people experiencing stereotype threat, the anxiety people feel when they expect to be judged by a group stereotype. Clenched fists, scanning the scene, fidgeting, pacing, and a tight jaw were among them. Not one of the 15 mentioned that a frightened, innocent person might look the same.

A survey of beliefs and a warning about fear. That is most of the published evidence behind a skill set officers stake their lives and their legal defenses on.

Why hasn’t anyone researched police pre-attack indicators?

The knowledge was built backward. Most indicator lists trace to trainers and survivors reviewing footage of officers being assaulted, a tradition that runs back to the 1980 book Street Survival and the seminars built on it. That method only studies the attacks. Nobody reviews the thousands of encounters where someone bladed, clenched a fist or scanned the exits and then complied. Without those, there is no base rate and no false-positive rate. You cannot know a cue predicts violence if you only ever look at violence.

The event is rare and hard to capture. Assaults happen in a small share of police contacts. Catching enough of them on ride-alongs would take thousands of hours. Until recently, most dash- and body-camera footage was too grainy or badly angled to code subtle cues like a jaw clench or a glance.

Surveys are cheap. Validation isn’t. Asking officers what they believe takes a questionnaire and a mailbox. Testing whether a cue predicts an attack takes access to footage, agency cooperation, and staff to code thousands of encounters.

The indicators do legal work. Under Graham v. Connor, force is judged by what a reasonable officer would perceive in the moment. Pre-attack indicators give officers the words to articulate that perception, and agencies, trainers, and defense attorneys lean on them. A rigorous study that knocked out a cue would weaken reports and testimony already on the books. Institutions rarely commission research that could hurt them.

The training market rewards confidence, not validation. Courses sell on dramatic video breakdowns and instructor experience. A vendor who says “we don’t know how often this cue is wrong” loses to one who doesn’t. Meanwhile, academic researchers have had little reason to enter a topic that sits between officer safety and use-of-force critique and draws fire from both sides.

What untested pre-assault indicators cost officers and the public

An unvalidated cue fails in two directions, and both are expensive.

A false positive means an officer reads fear as aggression and uses force on someone who was never going to fight. That produces injuries, lawsuits, prosecutions, and the kind of viral video that sets community trust back years. The stereotype-threat research suggests these errors may not fall evenly across the public.

A false negative means an officer watches for the textbook cues and misses the real one. Veteran trainers already warn that experienced fighters and predatory offenders often show none of the classic signs and instead appear unusually polite and compliant while they close distance. A checklist nobody has tested can teach officers to relax at exactly the wrong moment.

Other professions have done better. Psychiatric nurses and emergency departments use structured tools such as the Brøset Violence Checklist, scored against real outcomes and reported with hit rates and error rates. Officers, who face armed strangers rather than known patients, get lists built on anecdote.

What real pre-attack indicator research looks like

The obstacles that stalled this work are shrinking. Body-worn cameras are now standard in many agencies, and the footage is sharper and far more plentiful than it was even a decade ago. The raw material for a real answer is sitting on agency servers.

A serious research program would do four things:

  1. Code every encounter, not just the assaults. Sample ordinary stops and arrests alongside the violent ones, so each cue gets a base rate and a false-positive rate.
  2. Test cues in combination and in context. Officers are taught to read clusters, not single signs. Research should test clusters too, alongside the known risk factors like intoxication, warrants and prior violence.
  3. Check performance across groups. If a cue predicts differently by race, age or mental state, officers and courts need to know.
  4. Fund it and publish it. Federal and state research dollars, POST boards and agencies should back independent studies, and training vendors should publish what their indicators rest on.

How FOCUS is closing the pre-attack indicator research gap

Some of that work has started. FOCUS (Field Observable Cues for Unknown Situations), developed by Dr. Travis Yates, was built from five years of analyzed officer video footage. The key design choice is the comparison group. FOCUS did not only study attacks. It compared encounters where officers were shot or shot at with encounters where officers went home unhurt.

That comparison is exactly what the traditional indicator lists never had. According to the FOCUS methodology, officers who were shot or shot at faced an average of 4.8 “sentinel cues” beforehand, such as non-compliance, concealment, and verbal statements of intent. Officers who were not hurt faced an average of 1.5. The lesson is not one magic tell. It is that warning signs cluster, and a rising count of them is something an officer can see, act on, and later articulate.

FOCUS turns those clusters into a structured, real-time risk assessment that officers can use in the field and defend afterward. Agencies can learn more about FOCUS training and certification, and the Tactical Behavior Science articles page covers the research behind it.

The next step for FOCUS is to publish the data, open the methods to peer review, and let outside researchers test the findings. Officer safety training should meet the same standard we expect of any tool that decides when force is used.

The bottom line on pre-attack indicators

None of this means officers should stop paying attention to body language. Instinct built on experience is real, and the cues officers trust most may well prove out. It means the profession should find out which ones deserve the trust officers already give them.

We ask officers to make split-second decisions that the law will later judge as reasonable or not. The least we owe them is a tested answer to the question every one of those decisions rests on: what does an attack actually look like before it happens?

Frequently asked questions about pre-attack indicators

What are pre-attack indicators?

Pre-attack indicators are behaviors police are trained to read as signs a person may be about to assault them. Common examples include a bladed or fighting stance, clenched fists, target glancing at an officer’s weapon, sudden weight shifts, and closing distance.

Are pre-attack indicators backed by research?

Very little. Most published studies measure what officers and civilians believe predicts an attack, not what actually does. A 2014 survey of 129 assaulted officers found strong agreement on which cues seem dangerous, but its author noted that perception does not prove validity.

Which pre-attack indicators do officers rate as most dangerous?

Of 23 behaviors rated by officers who had been assaulted, a boxer’s stance, invading personal space and hands in pockets ranked highest, followed by clenched hands, verbal threats and glancing around. Crying, rapid blinking, hands on hips and avoiding eye contact ranked lowest.

Why do pre-attack indicators sometimes fail?

Many taught cues, such as scanning, fidgeting, and clenched fists, also appear in frightened people who pose no threat. Experienced offenders may show none of the classic signs. Single cues are weak; clusters of cues carry more weight.

What is FOCUS training?

FOCUS is a behavioral threat recognition program. It teaches officers to spot clusters of “sentinel cues” and turn them into a structured, real-time risk assessment, based on a comparison of encounters where officers were and were not attacked.

Sources

Tagged , , , ,