The Bradley Curve Is a Mirror, Not a Map
Why most organizations stall in the dependent stage — and what the original model never quite said.
In 1995, a DuPont employee named Berlin Bradley sketched a curve that has shaped how the safety profession talks about culture for thirty years. Four stages — reactive, dependent, independent, interdependent — plotted against a falling injury rate. It became one of the most reproduced diagrams in occupational health. Search any major EHS software vendor's website today and you will find a version of it. ISN, Veriforce, EcoOnline, Cority, SafetyConnect, and dozens of others use the curve as the spine of their cultural-maturity narrative. The model has aged better than most. It is still useful. But I have watched it get misread so consistently, in so many organizations, that the diagnosis has become its own obstacle.
Most leaders treat the Bradley Curve as a map — a sequence of stops on a journey from chaos to maturity. I want to argue that it is something else entirely. It is a mirror. And the reason most organizations stall in the dependent stage is not that they do not know about the curve. It is that they cannot bear to look at what the mirror is telling them about themselves.
Let me describe the stall, because I have lived inside it more than once over a thirty-five-year career.
In a dependent culture, workers follow rules because rules are enforced. Management commitment is visible. Procedures are written. Audits happen. Training is delivered. The injury rate trends down for a while, sometimes for years, and senior leaders start to believe the system is working. Then it plateaus. Recordables flatten. The next set of investments produces diminishing returns. New initiatives launch, land for two quarters, and fade. The organization spends another decade running in place.
The diagnosis is not subtle. A dependent culture has reached the limit of what compliance alone can produce. Compliance is real, non-negotiable work — it is the foundation everything else stands on. But it is a floor, not a ceiling. The data backs this up. ISN's most recent multi-framework analysis, drawing on more than 178,000 OSHA recordable incidents between 2017 and 2024, found that while overall recordable incident rates have continued to decline, serious injuries and fatalities have not followed the same trajectory. The Edison Electric Institute, in approving its updated SIF criteria in 2024, put the underlying mechanism plainly: "Reducing the rate of low-severity injuries may not lead to a corresponding reduction in SIFs." Translated into Bradley Curve language, this means that the work that moves an organization from reactive to dependent is not the same work that moves it from dependent to interdependent. The first transition is procedural. The second is cultural. And procedural improvements, no matter how well executed, will not produce cultural maturity.
Here is where the mirror gets uncomfortable. Moving from dependent to independent is not a procedural upgrade. It is a power transfer. To get workers to take genuine ownership, leadership has to give some of its ownership away. The frontline supervisor has to surrender the reflex to provide every answer. The corporate safety function has to stop being the singular voice on every decision. The plant manager has to accept that a crew that stops a job they consider unsafe is not undermining production. They are doing the work the organization has been telling them, in posters and town halls, that it wants them to do.
Most organizations cannot do this. Not because they lack the will, but because the operating model is built around control. The metrics reward control. Incentive structures reward control. The leadership pipeline selects for control. Asking a system optimized for control to redistribute it is a structural problem, not a motivational one. The Bradley Curve becomes a place organizations point to in slide decks while continuing to do, every day, the very things that keep them where they are.
I do not say this to scold. I say it because the honest naming of the stall is the precondition for moving past it. When I look back at the facilities I led during my time at GE, Kimberly-Clark, Bristol-Myers Squibb, and Coveris, I can name specific years where I personally was the obstacle to my own organization's maturity. I was the safety leader workers called for every decision because I had been so available, so visible, and so willing to absorb the problem that I had quietly trained an entire workforce to outsource its safety judgment to me. I had built a dependent culture by being too good at being needed.
Reversing that takes deliberate work. Three practices have mattered most.
The first is teaching workers to make decisions rather than ask permission. This sounds obvious, but in practice it requires the safety practitioner to bite his tongue regularly. When a worker brings a question with an answer they already know, the worst thing a leader can do is provide the answer. The best thing is to ask, "What do you think?" — and then live with the awkward pause while ownership transfers across the table. Do this a hundred times and the dependent reflex begins to break. Todd Conklin's HOP work makes the same point from a different angle: when the people closest to the work are the ones designing how the work gets done, the system gets safer not because the procedure improved but because the ownership relocated.
The second is making the cost of staying dependent visible. Most senior leadership teams have never been shown what the plateau actually costs them — in stalled performance, in talent attrition, in incidents that would not have happened in a more interdependent culture. The Bradley Curve is most useful, in my experience, not as an inspirational diagram but as a costed diagnostic. "Here is where we are. Here is what staying here will cost us over the next three years. Here is what moving will require." Leaders respond to clarity, even when the answer is uncomfortable.
The third is recognizing that the destination is not independence. It is interdependence. The original DuPont framing is precise on this point and worth quoting back to anyone who has flattened it: an independent culture is workers taking responsibility for themselves, which is genuine progress, but not the endpoint. The goal is workers taking responsibility for each other — and, with the safety practitioner, for the system itself. That collective ownership is what separates the organizations that reach genuine maturity from those that get partway and quietly settle in.
There is one more piece I want to name plainly, because I have seen it derail more transformation efforts than any other single factor. The Bradley Curve is not something an organization completes. It is something an organization practices, every day, in thousands of small decisions about who speaks, who decides, who is listened to, and what happens after an incident. Cultural maturity is not a destination you arrive at. It is a discipline you sustain.
Which means the right question for any safety leader looking at the curve is not, "What stage are we in?" It is, "What did we do this week that pushed us forward, and what did we do that quietly pulled us back?" Most weeks, you will find evidence of both. Most weeks, the regression came from the leadership team, not the front line.
If the curve is a mirror, then the work is not to flatter the reflection. It is to look honestly, name what is true, and choose, deliberately and uncomfortably, to share authority with the people closest to the work. Until that share is real, the plateau is not a phase. It is the destination.
And no one wants to spend a career staring at their own ceiling.
Sources & Further Reading
• Bradley, B. (1995). The DuPont Bradley Curve. DuPont.
• ISN. (2025). Serious Injury & Fatality Insights, Multi-Framework Analysis 2017–2024.
• Edison Electric Institute. (2024). Serious Injury & Fatality (SIF) Criteria: Approved Changes.
• Conklin, T. (2019). The 5 Principles of Human Performance.
• U.S. Bureau of Labor Statistics. (2026). Census of Fatal Occupational Injuries 2024.
