New Research Validates Why a Data-Driven, Predictive Early Intervention System is Critical

---

 

 

Threshold-based early intervention systems don’t work the way they should. That’s not just our position; it’s a fact borne out by University of Chicago research, validated by Benchmark across the world’s largest multi-jurisdictional officer performance database. Now, a peer-reviewed study published in February 2026 in the Journal of the American Statistical Association (JASA), one of the most respected statistical journals in the world, independently confirms these findings.

The study, authored by Greg Ridgeway of the University of Pennsylvania, introduces a statistical model that measures an individual officer’s propensity to escalate force compared to peers present at the same incident. His conclusion is blunt: EIS thresholds that count incidents “are not working effectively” because they ignore the context in which force is used.

For agencies already feeling the limitations of their current early intervention system – or for those who aren’t using one at all, this research supports our evidence-based approach and explains why, and it points toward what a better approach looks like.

Why Threshold-Based Early Intervention Systems Fail

All early intervention systems in policing today flag officers who exceed a set number of force incidents, complaints, or other events within a defined time period. The approach is simple. And that simplicity is the problem.

Think about the officers working in high-crime areas, on night shifts, or in specialized units like warrant service. They’re going to encounter more situations involving force. A typical threshold system treats all of those encounters the same. The predictable result? Threshold-based flags disproportionately identify the most active officers (false positives) while missing genuine risk among officers with lower activity levels (false negatives). Researchers and practitioners have found this pattern over and over again.

Ridgeway’s study also highlights a distinction that threshold systems miss entirely: the severity of force relative to the situation. An officer who consistently uses more force than peers in identical circumstances presents a very different risk profile from one whose force is proportional to the circumstances. Benchmark Analytics’ patented approach to capturing force data was designed with exactly this distinction in mind, emphasizing proportionality as a core measure. Threshold systems can’t make that distinction. They only count.

What the New Research Confirms: Context Eliminates Confounding, Just as Our Data Has Shown 

Ridgeway applied his model to seven years of use-of-force data from the Seattle Police Department: 4,821 incidents involving 1,503 officers. Instead of counting incidents, the model compares each officer’s force choices against peers who were present at the same scene. Because every officer on scene shares the same environmental context (location, subject behavior, lighting, conditions), the model mathematically eliminates those situational factors and isolates each officer’s individual tendency to escalate.

Out of 1,503 officers, the model identified nine with a statistically high probability of escalating force beyond what peers would use in identical circumstances. That’s less than 1% of the force. At the other end of the spectrum, it found 13 officers present at numerous force incidents who almost never used force themselves. That could reflect strong de-escalation skills or supervisory roles, but it could also signal reluctance to act when the situation demands it.

Pay attention to this next part if your agency uses a threshold-based system. One of the flagged officers had fewer than 10 incidents across the entire seven-year period and never exceeded Level 1 force. No threshold system would ever flag that officer. But the model showed that in nearly every incident, this officer was the one who used force while peers did not. Meanwhile, another officer with 15 Level 2 and three Level 3 force incidents was not flagged, because their force was consistent with what other officers used in those same high-severity situations. Context changed the picture entirely.

How First Sign Early Intervention Solves This Problem

We welcome research like Ridgeway’s because it validates the approach Benchmark Analytics has been building for over a decade. First Sign^® Early Intervention was developed in partnership with the University of Chicago and has been validated across the world’s largest multi-jurisdictional officer performance database (over 80 million records). It was built specifically to solve this problem.

Where threshold systems rely on raw counts, First Sign uses predictive analytics that account for context, behavioral patterns, and each officer’s full activity profile. The system evaluates over 90 model variables across 5+ data sources, including officer attributes, arrest history, use-of-force, internal affairs, and CAD/RMS data. It achieves 85% model precision, a level of accuracy that threshold-based approaches simply can’t match.

The overlap with Ridgeway’s findings is hard to ignore. His research identifies a small percentage of officers accounting for outsized risk. Our analysis of a subset of our partner agencies, employing nearly 10,000 sworn officers, tells a consistent story: approximately 5% of officers generate 66% of injuries and disproportionate use-of-force incidents. But First Sign doesn’t stop at identification. It connects directly to C.A.R.E.^® (Case Action Response Engine), Benchmark’s research-based case management platform that gives supervisors real intervention tools and a structured plan to get officers back on track.

If Your Current EIS Isn’t Working, Here’s Why

If your agency deals with false positives that flag hardworking officers, missed risks among those who fly under the radar, and eroded trust in the system as a result, you’re not alone. And it’s not your fault. This is both a methodology and a technology problem: faulty methodology baked into the wrong tools.

When you move from counting incidents to understanding patterns in context, three things change.

1. You identify the right officers: not the busiest ones, but those whose behavior meaningfully deviates from peers in similar situations.
2. You protect your high performers, who should be recognized rather than flagged.
3. And you build trust, because officers who see that the system accounts for their assignment, their calls, and their environment are far more likely to view early intervention as support rather than surveillance.

Agencies that have made the shift to First Sign have seen real results: a 13% reduction in use of force, a two-level reduction in the severity of force, and a 48% decrease in citizen complaints, all while arrest activity remained essentially stable. One agency case study showed reductions of 31% in force severity, 72% in officer injuries, and 52% in citizen complaints during the observation period following the implementation of Benchmark’s platform.

That’s exactly what Benchmark Analytics has pioneered. No other technology partner does what we do. First Sign and C.A.R.E. represent the only integrated platform purpose-built to identify officers who truly need support and provide a structured, research-based pathway to elevating officer performance, making early intervention smarter, fairer, and more effective for the agencies and communities that depend on it.

Frequently Asked Questions

What Is a Threshold-Based Early Intervention System In Policing?

A threshold-based early intervention system (EIS) flags officers who exceed a predetermined number of events, such as use-of-force incidents or complaints, within a set time period. While widely used, research shows these systems produce high rates of false positives. They disproportionately flag active officers while missing genuine risk patterns among less active ones. Read More >>>

Why Don’t Threshold-Based EIS Systems Work Effectively?

They don’t account for context. Officers in high-crime areas, on specialized units, or working night shifts encounter more situations that involve force. Counting incidents without adjusting for assignment, location, and what’s actually happening on scene means the system can’t distinguish between a hardworking officer and one with a genuine pattern of escalation.

What Is a Data-Driven or Predictive Early Intervention System?

There is no system like First Sign Early Intervention from Benchmark Analytics. It is the only solution that uses predictive analytics and machine learning to identify officers who are truly at risk, drawing on patterns of behavior, context, and multiple data sources rather than raw event counts. Developed in partnership with the University of Chicago and validated across the world’s largest multi-jurisdictional officer performance database. Notably, while Ridgeway’s study focuses exclusively on use of force, First Sign draws on a far broader set of data sources including internal affairs, CAD/RMS, training records, and arrest history. This enables richer analysis and more accurate identification.

What Did the 2026 Ridgeway Study Find About Police Use-Of-Force Analysis?

The 2026 study by Greg Ridgeway, published in the Journal of the American Statistical Association, introduced a conditional ordinal stereotype model that compares officers’ force choices against peers present at the same incident. Applied to seven years of Seattle Police Department data (4,821 incidents, 1,503 officers), it identified 9 officers with elevated propensity to escalate force and 13 with unusually low force usage. The model eliminates environmental confounding that threshold systems cannot address.

How Does First Sign Early Intervention Compare to Threshold-Based EIS?

First Sign Early Intervention from Benchmark Analytics goes beyond threshold-based triggers by using over 90 model variables from 5+ data sources to identify officers who truly need support. It achieves 85% model precision. When an officer is identified, First Sign connects to C.A.R.E. (Case Action Response Engine), a research-based case management platform for building individualized support plans. Across a subset of our partner agencies, First Sign implementation has been associated with a 13% reduction in use of force, a 2-level reduction in severity, and a 48% decrease in citizen complaints.

Reference:

Ridgeway, G. (2026). A Conditional Ordinal Stereotype Model to Estimate Police Officers’ Propensity to Escalate Force. Journal of the American Statistical Association, 1–12. https://doi.org/10.1080/01621459.2025.2597050

Ready to move beyond thresholds? Learn how First Sign Early Intervention and the Benchmark platform can transform your agency’s approach to elevating officer performance. Contact our team to schedule a conversation.

---