The Investigator Who Doesn't Know What They Don't Know: Competence, Bias, and Dunning-Kruger in HSE

He had 18 years of operational experience. He had managed HSE departments across three continents. He had conducted, by his own count, over forty incident investigations. He had never received a single day of formal training in investigation methodology. He did not know what Tripod Beta was. He had never heard of AcciMap. He could not distinguish between a precondition and a root cause. He thought "5 Whys" was an advanced methodology. He was not incompetent. He was unconsciously incompetent — the first stage of the Dunning-Kruger competence model. He did not know what he did not know. And because he had been conducting investigations for 18 years without anyone questioning his methodology, he had no signal that his approach was insufficient. His organisation had reinforced his overconfidence by accepting every report he produced without independent quality review. This is not an exceptional case. It is the norm across high-hazard industries. The majority of incident investigators in mining, construction, oil and gas, and manufacturing have operational backgrounds. They are good at what they were trained for. They were never trained for investigation.

THE COMPETENCE PARADOX The Dunning-Kruger effect creates a dual burden: investigators with limited methodology skills lack both the skill and the metacognitive ability to recognise they lack it. They produce reports that look professional, follow a logical structure, and reach confident conclusions — but systematically stop short of systemic causes because they do not have the conceptual framework to see them.

Stage 1: Unconscious incompetence — "I don't know what I don't know"

The investigator has operational experience but no formal investigation training. They apply common sense, experience, and linear thinking. Their reports identify immediate causes but rarely trace causation to organisational factors. They are confident in their conclusions because they have no framework against which to judge their own limitations. Research confirms this: a 2025 Springer study found that many safety incidents are deepened by overestimation of competency and understanding of safety rules.

Stage 2: Conscious incompetence — "I now know what I don't know"

The investigator receives formal training in a structured methodology (Tripod Beta, STAMP, AcciMap) and realises how much their previous investigations missed. This is the most uncomfortable but most valuable stage. Confidence drops as awareness rises. Many investigators resist this stage because it invalidates their prior work.

Stage 3: Conscious competence — "I can do it, but it takes effort"

The investigator applies the methodology deliberately, following the structured process, building evidence maps, tracing causation paths. The analysis is rigorous but effortful. Quality is high when the investigator is focused and has adequate time. Quality drops under pressure or time constraints.

Stage 4: Unconscious competence — "Systemic thinking is automatic"

The investigator has internalised the methodology. They automatically look for organisational factors, question management decisions, and challenge defensive narratives. They no longer need to consciously apply a framework — systemic thinking is their default operating mode. This stage typically requires hundreds of hours of deliberate practice across multiple investigations.

KEY TAKEAWAY Most organisations assume their investigators are at Stage 3 or 4. Most are at Stage 1. The gap between assumed and actual competence is where investigation quality dies — and where the Dunning-Kruger effect is most dangerous.

1. Mandate certified investigation training: No person should lead an investigation without completing a certified programme in a recognised causation methodology. "Experience" is not a qualification. Experience without methodology produces experienced incompetence.

2. Implement investigation competency assessment: Test investigators periodically using scenario-based assessments. Present them with a case and evaluate whether their analysis reaches systemic causes or stops at individual behaviour. Objective measurement is the only antidote to the Dunning-Kruger effect.

3. Require peer review of all investigation reports: Every investigation report should be reviewed by a qualified peer before finalisation. This is standard in academic research, financial auditing, and medical practice. It should be standard in incident investigation.

4. Track the ratio of systemic to individual recommendations: If an investigator consistently produces recommendations targeting worker behaviour, their competence in systemic analysis is insufficient. The recommendation ratio is a proxy indicator of investigation competence.

5. Create a supervised practice pathway: New investigators should work under the supervision of an experienced, methodology-trained lead investigator for their first five to ten investigations. This mirrors the supervised practice requirements in medicine, aviation, and law.

THE DREYFUS TEST Stuart Dreyfus's Five-Stage Model of Adult Skill Acquisition maps the journey from novice to expert. Most incident investigators are operating at the novice or advanced beginner stage but are assigned expert-level responsibilities. The gap between assigned role and actual competence stage is where investigation quality fails.

"The most dangerous investigator is the one who has done forty investigations and learned nothing from any of them — because nobody ever told them their methodology was insufficient."

— Bruno Hounkpati