The investigation report says: “Worker failed to follow procedure.”

The corrective action says: “Retrain the worker. Remind the team about the SOP. Increase supervision.”

Six months later, a different worker at a different site does the same thing. Another incident. Another investigation. Another report that says the same thing in different words.

This is the cycle that defines safety management in organisations that haven’t engaged seriously with human factors safety. And in Malaysia — where toolbox talks, signage, and awareness campaigns remain the dominant safety intervention — this cycle is widespread.

The reason it repeats is not that workers are careless, untrained, or lacking awareness. The reason it repeats is that the organisations are trying to solve a systems problem with a behavioural intervention — and those two things are not the same.

Human Behaviour Is Not a Lever You Can Pull

Human behaviour is not a choice made in a vacuum. It is the output of a system. Every action a person takes on a worksite is influenced by a complex combination of factors operating simultaneously:

• Time pressure — real or perceived — that makes the faster option feel necessary
• Social norms on site — what experienced workers model, what gets laughed at, what gets praised
• Reward structures — whether the organisation actually rewards compliance or just says it does
• What gets ignored — near misses that aren’t reported, shortcuts that are tolerated, deviations that have never caused harm
• Equipment design — whether the tool, workstation, or process makes the safe way the easy way, or makes it harder and slower
• Experience history — what the person has learned from the last 200 times they performed this task without consequence

When an incident occurs, the person involved was responding rationally to this system. Not recklessly. Not carelessly. Rationally — in the way that a person does when the system around them has consistently reinforced a particular approach to getting the job done.

You cannot mandate your way out of that system with a toolbox talk. A one-hour session about awareness does not change the time pressure, the social norms, the reward structures, or the equipment design. It adds information to a person who is already operating in a system that will continue to produce the same behaviour.

Why Administrative Controls Have a Ceiling

Administrative controls — procedures, training, supervision, signage, toolbox talks — have a fundamental characteristic that limits their reliability: they assume cooperation. They depend on a person choosing to comply at the right moment, under the right conditions. When those conditions change — and they always do — the control fails.

This is not a failure of the person. It is a design characteristic of administrative controls. They are built on human performance, and human performance is variable. It varies with fatigue. It varies with distraction. It varies with the pressure of the moment and the culture of the team. Expecting administrative controls to deliver the consistency of an engineering control is expecting something that administrative controls are not designed to provide.

What Human Factors Safety Actually Looks Like in Practice

Effective human factors safety in Malaysia starts with a different question. Instead of “how do we get workers to comply?”, the question is “what conditions are producing non-compliance, and how do we change them?”

This reframes the investigation, the risk assessment, and the corrective action in fundamental ways.

Principle 1 — Design for the worst decision, not the best intention

Principle 2 — Reward structures tell the truth

Principle 3 — Near misses are data, not embarrassments

The Human Factors Safety Implication for HIRARC and Bowtie

Understanding human factors safety also changes how you approach your risk assessment documentation. When you list “supervisor monitoring” or “worker follows SOP” as controls in your HIRARC, you are listing administrative controls that assume cooperation. For low-hazard, routine tasks, this is proportionate. For major hazard scenarios, it is insufficient.

The DOSH Malaysia hierarchy of controls places engineering controls above administrative controls precisely because of this reliability difference. A pressure relief valve does not require a supervisor to decide to act. A gas detector interlock does not require a worker to choose to respond. They work regardless of what the human decides.

For your highest-risk scenarios: identify the human controls in your HIRARC. Ask what conditions could prevent them from working — time pressure, social norms, equipment layout, incentive structures. Then ask whether there is an engineering equivalent that would eliminate the dependency on human decision-making in that moment. That is the human factors safety question that risk assessments in Malaysia rarely ask.

That’s the difference between a safety programme that relies on people being perfect — and a risk system that accepts they won’t be and designs accordingly.

Asyraf Khalil

Asyraf is a Risk Management and Incident Investigation specialist with 15 years of HSE experience across Malaysia. He is the founder of Cikgu Barrier and delivers practical training in HIRARC, Bowtie Analysis, and Tripod Beta Incident Investigation.