Optimism Bias in Safety: 6 Myths That Make Supervisors Underestimate Risk

A supervisor can walk past the same unguarded edge for weeks without feeling careless, because repeated exposure makes a hazard look normal before it makes the work safer. This article separates six myths that let optimism bias quietly lower the standard of a safe decision.

Optimism bias in workplace safety is the tendency to believe that a negative event is less likely to happen to us, our crew, or our shift than to someone else. It does not mean that a supervisor is indifferent. It means that familiarity, recent success, and production pressure can make weak evidence feel reassuring.

Across more than 25 years leading EHS work, Andreza Araújo has built her safety message around a practical distinction: a safe culture is learned through repeated decisions, not through declarations. That distinction matters here because optimism bias is rarely corrected by telling people to be more careful. It changes when the work makes risk visible, gives people authority to challenge it, and verifies what happened next.

Why optimism bias survives experienced teams

Optimism bias survives because the immediate evidence often appears to support it. A task was completed yesterday, the same route has been used for months, and no one was hurt during the last shift. Those facts can be true while the exposure remains unacceptable.

James Reason's work on active and latent failures helps explain the gap. A worker may make the final visible choice, yet the conditions around that choice can include weak design, unclear ownership, schedule pressure, or a procedure that no longer matches the task. Treating the choice as a personal attitude problem leaves the stronger causes untouched.

For supervisors, the practical question is not whether the team feels confident. It is whether confidence is supported by current evidence, a credible control, and a decision route that still works when conditions change.

Myth 1: Experience protects people from optimism bias

Experience can improve hazard recognition, but it can also make familiar exposure feel ordinary. A supervisor who has seen a shortcut succeed many times may read that history as proof of reliability even when the task, equipment, staffing, or weather has changed.

Daniel Kahneman's work on judgment shows why repeated outcomes can influence confidence without improving the quality of the underlying evidence. Familiarity reduces mental friction. The decision feels easier, so the person may stop asking whether the control still fits the hazard.

Experience should therefore trigger a sharper review, not an automatic permission. Ask what changed since the last comparable task, which control carries the consequence if it fails, and what evidence would make the team stop. The answer should be visible before work begins.

In Safety Culture: From Theory to Practice, Andreza Araújo connects culture with the habits that leaders repeat under pressure. One useful habit is to make experienced people explain the evidence behind confidence, because expertise becomes a control only when it remains testable.

Myth 2: A clean record proves the control works

A clean incident record shows that a recorded event did not occur. It does not prove that the hazard was removed, that the barrier held, or that the organization would detect a weak condition before harm.

This distinction is important in high-consequence work. The absence of injury can reflect effective control, limited exposure, luck, or incomplete reporting. These explanations require different decisions, yet a single good outcome can make them look identical.

Replace the question “Has this caused an incident?” with “What would show that the control is available, understood, and dependable today?” For a lockout, that may include isolation verification and field observation. For a lifting task, it may include ground conditions, lift-plan review, and confirmation that the exclusion zone can be maintained.

The control is not proven by its presence in a procedure. It is proven when the organization can show how the barrier changes exposure during the work that matters.

Myth 3: Near misses always correct overconfidence

A near miss can interrupt optimism bias, but it does not correct it automatically. Teams may describe the event as a lucky escape, blame an individual, or close the review with a reminder to pay attention. The organization then keeps the same conditions that allowed the event.

Andreza's book Luck or Capability, known in Portuguese as Sorte ou Capacidade, supports a more demanding reading of precursor events. The question is not whether the team was lucky. The question is whether the organization can identify which layers were weak and change them before the next exposure.

Supervisors should ask what made the event possible, which warning was available but ignored, and who had authority to change the condition. They should also test whether the corrective action changes equipment, planning, staffing, sequencing, or decision rights rather than adding another instruction.

A near miss becomes a learning signal only when the work looks different afterward and the people who raised the concern can see what changed.

Myth 4: Confidence is the same as psychological safety

A confident team may speak freely about routine matters while staying silent about a decision that could delay production or challenge a senior person. Psychological safety is not a mood of optimism. It is the practical expectation that relevant information can be raised, received, and used without retaliation.

This matters because optimism bias can spread through a group. When the first person says that a risk is minor, others may withhold a different view to preserve speed, status, or team cohesion. Silence then looks like agreement even though the evidence is incomplete.

Supervisors can test the difference by asking for a dissenting view before approving a high-risk task, inviting the newest or least senior person to identify what could go wrong, and reporting back on how the concern affected the decision. The quality of the response matters more than the number of people who spoke.

A team does not need unrestricted tolerance for every choice. It needs a clear route for technical challenge, a fair response to good-faith concerns, and firm boundaries around deliberate violations.

Myth 5: More training is the fastest cure

Training can improve recognition and skill, yet it cannot compensate for a control that is unavailable, impractical, or routinely defeated by the way work is organized. When a worker is trained to follow a sequence that production planning makes impossible, the system is asking memory to cover a design problem.

The training response also creates a comforting story. It gives leaders a completed action, a certificate, and a message that the workforce has been reminded. Those outputs may be useful, but they do not establish that the exposure changed.

Before assigning retraining, compare the intended behavior with the real task. Is the equipment designed for the safe sequence? Can the supervisor stop the job without a penalty? Are conflicting instructions resolved before the crew starts? Does the procedure define what to do when the plan and field conditions diverge?

Guide to Behavioral Observation: ¿VAMOS A HABLAR? points toward dialogue rather than inspection theater. A useful conversation identifies the pressure behind the choice and then asks which condition the organization can redesign.

Myth 6: The safest message is “nothing can go wrong”

Messages that promise perfect outcomes can make people hide uncertainty. If leaders present safety as proof that nothing will go wrong, workers may feel pressure to preserve the image by minimizing warning signs, delaying escalation, or describing a weak control as adequate.

The stronger message is precise confidence. Leaders can say that the task may proceed because the hazard is understood, the critical control has an owner, the stop route is available, and the verification point is defined. That statement is less comforting than a slogan, but it gives the team something real to rely on.

Andreza Araújo's Illusion of Compliance, the English gloss of A Ilusão da Conformidade, is useful for this distinction. Compliance can show that a requirement was completed, while culture shows what people do when the requirement collides with time, convenience, or status.

Supervisors should replace absolute reassurance with a bounded decision. State the known exposure, the control that must hold, the condition that invalidates the plan, and the person who can authorize a change.

Comparison: optimism bias versus evidence-based confidence

The difference is not pessimism versus optimism. It is unsupported confidence versus confidence that remains open to challenge. The second pattern gives a team permission to stop, revise, and continue when the evidence supports the change.

What supervisors should change on Monday

Choose one familiar task that the team describes as routine and review it at the point where exposure actually occurs. Ask which assumption has become invisible, which control would fail first, and what the crew would do if the plan no longer matched the field.

Then make one decision visible. Record the condition that must remain true, name the owner who can change it, and set a verification that workers can recognize without opening a spreadsheet. If the control cannot be verified in the field, the organization has not finished defining it.

That practice reflects the central lesson of Safety Culture Diagnosis. Diagnosis is useful only when it leads to a change in work, not when it becomes another score that reassures the people who already believe the system is safe.

Conclusion: confidence must stay answerable to evidence

Optimism bias becomes dangerous when repeated success, experience, training, and silence are mistaken for proof that risk is under control. Supervisors reduce it by making current evidence, dissent, ownership, and verification part of the decision itself.

Safety is about coming home. Explore Andreza Araujo's safety culture and leadership resources when your operation needs to turn better decisions into safer work.