Beyond “Unsafe Behavior”

Pop quiz – are most accidents caused by unsafe behavior or unsafe conditions?

Have your answer? Good, keep it to yourself for a minute because rather than talking about your answer, we should spend some time talking about the question. It’s an age old question in the safety profession, perhaps first asked officially by Heinrich (who told us that 88% are caused by unsafe acts), and a question that is hotly debated to this day.

But here’s the thing – the question itself is problematic because it assumes so much! There are at least three assumptions made in the question that many in the safety profession take for granted:

1. That we can know the actual cause of accidents (i.e. cause is something that exists in reality and can be found);
2. That there is such a thing as “unsafe behavior” that is a distinctly identifiable category; and,
3. That any cause of an accident can be clearly attributed to an “unsafe behavior” or an “unsafe condition”, and there is no significant overlap between the two categories (i.e. they are mutually exclusive).

Unfortunately, even though many in the safety profession accept all of the above without question, each of the above assumptions is, at best, hotly debated, and, at worst, shown to be inaccurate, in the safety science and social science literature. This alone speaks to problems within the safety profession (which we've talked about here and here), and many volumes have been written on the assumptions, which we just cannot do justice to in this post. However, we want to spend some time to at least challenge the thinking regarding “unsafe behavior” and present some new ways of understanding human behaviors.

The first question that should spring to our minds when we speak of “unsafe behavior” is – “unsafe” by whose measure? After all, let’s not forget that in the safety profession there is not a clearly agreed upon definition of what “safety” is! And often when we talk about “unsafe behavior” we’re speaking about it in retrospect, usually after an accident (i.e. “because of his unsafe behavior the employee fell off of the ladder”). This often leads to circular arguments – the employee fell off of ladder because of his unsafe behavior. How do we know his behavior was unsafe? Because he fell off of the ladder.

Even if we take one of the most widely agreed upon definitions of “safety”, that safety is “freedom from unacceptable risk”, we still run into problems. Why? Because we’re defining whether the risk was acceptable or not in a world that is objectively different than the world was before the accident. We know the true cost of the behavior (assuming that we can say that the accident was caused by the behavior, which is another very problematic assumption to make. More on that later). The person did not know the true cost. Sure they probably knew that the potential was there, but there’s a difference between a potential loss and an actual loss. What if the person decided that the potential for an accident was an acceptable risk? Must we then say that the behavior was “safe” because the person accepted it? It seems silly, but this is the sort of question we must ask if we accept that there is a clear distinction between “safe” and “unsafe” behavior.

Furthermore, and this is very important, the distinction between unsafe behavior and unsafe conditions assumes that you can draw a meaningful separation between behavior and conditions or context. But here’s a challenge for you. Assuming for a second that we are able to tell when a behavior is safe or unsafe. Name one behavior that is always “unsafe” without also making a reference to a context or a condition.

Go ahead, we’ll give you a minute…

Times up. Got one? We sure couldn’t. We’ve thought long and hard and we simply cannot come up with any individual behavior that, by itself, regardless of any condition or context, is always unsafe. Whenever we think of an “unsafe behavior” it always has a contextual element associated with it. Some common examples – work near an unprotected leading edge more than 30’ off the ground (the only behavior identified is the work, the rest is context). Working on energized equipment without appropriate controls in place (again, the behavior is the “work” and the rest is context). Even the quintessential unsafe act, running with scissors, could be considered “safe” in certain contexts (i.e. in a medical emergency with trauma scissors).

This means that the debate between “unsafe behavior” and “unsafe conditions” is a false flag. It’s not an either/or scenario. We shouldn’t be focusing on one or the other, but rather on both and how each effects the other. Systems theorists teach us that interactions and relationships between parts are more important than the parts themselves. Why aren’t we focusing on that instead?

And this leads us to some thoughts for alternative ways of understanding behavior. Erik Hollnagel presents an interesting and informative model for understand people’s behavior as it relates to safety. Rather than thinking in bimodal terms (i.e. it is “safe” or it is “unsafe”), people’s behavior is more about variability – i.e. people adjust their performance to match the conditions they are in to help them achieve success in a complex, resource constrained world with competing goals. And it’s this performance variability that creates failure, which we then call “unsafe acts”. However, this performance variability is also the reason that work is successful most of the time (i.e. accidents are rare and the job gets done the overwhelming majority of the time). So, rather than working to eliminate “unsafe behavior”, we should instead help people make better adjustments to their environment.

This requires two shifts for the safety professional:

1. We must understand how work works. Typically we create rules separate from the work environment, then we get frustrated when people violate our rules and get into accidents. We think that this must be a problem with the people involved, but often the problem is that we design work in a way that just doesn’t work. If people are constantly adjusting their performance to help them achieve success then we need to get out there and understand what’s really happening on the ground floor and why. This means that your employees have just as much or more to teach you about how to do their jobs safely as you do to them.
2. There’s just as much, or more, to learn from what goes right as there is to learn from what goes wrong. We spend a lot of time learning from failures, which makes sense, but if that’s all we do then we get a skewed version of the world. We start to think that what we’re finding is unique to failures (e.g. procedure violations). But if we start to learn why our organizations are succeeding then we might start to realize that success and failure often have the same causes. This presents all sorts of unique opportunities for us to learn about how our organization creates safety on a daily basis, and why that sometimes doesn’t work.

It’s time though that we stop hanging onto the crutch of “unsafe behavior”. The concept is misleading and doesn’t really lend itself to good interventions for the safety professional. Further, there are better concepts out there that more accurately explain behavior and present innovate interventions for the safety professional.