Safety-I and Safety-II, Human and Organizational Performance (HOP), Behaviour Based Safety and the Bradley Curve: this page puts the common methods and models of occupational safety in context – true to their origins, with sources, and with a clear statement on how SAFETEE combines them in projects.
Safety methods are ways of thinking and working that companies use to prevent accidents and enable safe work. Classic safety engineering (Safety-I) asks what goes wrong and removes causes. Newer approaches such as Safety-II and HOP ask how work succeeds under varying conditions and shape organisation and context. Behaviour Based Safety supplies observations and conversations. Maturity models such as the Bradley Curve describe safety culture – but do not explain it.
The distinction was coined by Erik Hollnagel: in 2013 as a EUROCONTROL white paper (with Leonhardt, Licu and Shorrock), in 2014 as a book. Safety-I defines safety as a state in which as few things as possible go wrong. Safety-II defines it as a system’s ability to succeed under varying conditions. Both perspectives look at the same work – and see different things.
Hollnagel himself describes Safety-II as a complement to Safety-I, not a replacement. That is exactly how we use it: risk assessment, rules and barriers stay – complemented by a view of how your teams actually get the work done.
HOP grew out of the human performance movement in the US nuclear industry (INPO; DOE Human Performance Improvement Handbook, 2009) and was brought to industry by Todd Conklin and Sidney Dekker (“Safety Differently”). HOP is not a method with a form but a mindset with five principles – and one tool that makes it concrete on site: the learning team.
Erring is normal and predictable. Systems must be built so that a mistake does not become an accident – error tolerance instead of error-free.
Whoever looks for culprits gets silence. Whoever understands conditions gets information. Just culture is the precondition for learning.
Time pressure, tools, instructions and role models make behaviour likely. To change behaviour, change the context first.
Organisations do not learn from reports but from conversations with those who do the work – before and after events.
How management responds to an event decides what gets reported next time.
Facilitated conversations between those who do the work and those who design it: what makes this task hard? What helps? The result is better conditions, not new rules.
The DuPont Bradley Curve™ (today dss+) describes safety culture as a journey through four stages along which the accident rate falls: reactive, dependent, independent, interdependent. The picture is catchy – which is why it hangs in many meeting rooms. But it is a consulting product, not an empirically tested model.
The four stages are useful for talking with managers about responsibility. To locate an organisation we use observations, reports and learning teams – not a self-assessment on a curve.
The five-step ladder (pathological – reactive – calculative – proactive – generative) describes how an organisation handles information: are messengers punished or heard? It is a model too, but one with observable anchors.
Observation rate, share of safe observations, lead time of reported obstacles, number of learning teams, management response time: these are figures that show movement before the accident rate does.
Behaviour Based Safety is the oldest of the methods described here and the most contested. Both rightly so.
Its scientific basis is applied behaviour analysis; the first field study is by Komaki, Barwick and Scott (1978). BBS spread commercially from the 1980s through DuPont STOP and consultancies such as Behavioral Science Technology (Krause).
A meta-analysis of 13 studies (Tuncel et al. 2006) found significantly fewer injuries after BBS rollouts – but rates the studies as methodologically weak: almost only before-and-after comparisons without control groups. The effect is plausible, percentages are not.
The criticism – blaming the worker, incentives not to report, the refuted Heinrich claim of 88 % “unsafe acts” (Manuele 2011) – applies to badly designed programmes. Well-designed BBS asks about context at every observation, does without bonuses and gives HOP what the mindset lacks: data from everyday work.
Observation without asking about context turns people into the problem. Context without observation stays theory.
Programme design, observer training, leading indicators – no blame, with a systems view.
None of the methods replaces another. In our projects – HSE management, safety coordination, safety consulting – the situation decides which tool comes first:
Risk assessment, rules, barriers, instruction. Without this basis no other approach has ground under its feet.
Instead of the third warning: a facilitated conversation with those doing the work. What makes the task hard? Usually time, tools or conflicting instructions.
Clarify causes, yes – but also: why does this work succeed on every other day? The answer shows which adjustments the system carries and which it does not.
Not a stage on a curve but observation rate, reporting behaviour, management response time – measured with a BBS programme that asks about context.
Knowing the methods is the start. These services bring them into your operation:
Observation, feedback and leading indicators – programme design through to handover to your multipliers.
Safety culture, organisation and leadership – consulting for companies that want to go beyond compliance.
Training for managers, observers and contractor coordinators under DGUV Regulation 1.
Safety-II is an approach to safety management coined by Erik Hollnagel (white paper 2013, book 2014). Safety is not understood as the absence of accidents but as a system’s ability to succeed under varying conditions. The focus of analysis is therefore the normal case – how work is actually done (Work-as-Done) – and not only the rare accident.
Safety-I asks what goes wrong, looks for causes and removes them; people are seen as a source of error. Safety-II asks why work almost always succeeds and sees people as the resource that closes the gap between plan and reality. Hollnagel regards both as complementary: Safety-I for rules, barriers and legal duties, Safety-II for learning and adaptability.
HOP is a mindset in occupational safety that regards human error as normal and therefore designs work systems to be error-tolerant. It stems from the human performance movement of the US nuclear industry and was popularised by Todd Conklin and Sidney Dekker. Its core tool is the learning team: facilitated conversations with those who do the work.
In the common wording after Todd Conklin: People make mistakes. Blame fixes nothing. Context drives behaviour. Learning and improving is vital. How leaders respond to failure matters. The exact wording varies slightly between sources; the message stays the same.
The DuPont Bradley Curve (today dss+) is a maturity model for safety culture with four stages: reactive, dependent, independent, interdependent. It depicts how the accident rate falls as maturity rises. DuPont says it has marketed the model since 1995; the origin of the name is poorly documented.
No. We know of no independent study that proves the depicted link between culture stage and accident rate. The curve is a consulting product and works as a conversation piece, not as a measuring instrument. Criticism targets the linear stage model, the focus on individual behaviour and the accident rate as a yardstick, which also falls through non-reporting.
In 1931 Herbert Heinrich published the ratio 300 : 29 : 1 (near misses : minor : serious injuries) and the claim that 88 % of accidents are caused by unsafe acts. In 2011 Fred Manuele showed that both figures rest on untraceable insurance files and that the causes of serious accidents do not match those of near misses. As a basis for safety programmes they are outdated.
A facilitated conversation – usually 60 to 90 minutes in two rounds – between the employees who perform a task and those who design it. The aim is not to find blame or a single cause but to understand what makes the work hard and which conditions can be changed. Learning teams are used after events and preventively.
That depends on the starting point and the maturity of the basics. Without a sound risk assessment, rules and barriers (Safety-I) no culture programme holds. Recurring deviations call for learning teams (HOP), observable routine tasks for Behaviour Based Safety, the wish for figures for leading indicators. We clarify this in a first conversation and a short on-site assessment.
Industry, starting point, current concern: with three details we tell you where we would begin – and get back to you within one working day.
