Incident Investigation: 5 Myths That Make Evidence Look More Certain Than It Is

The first explanation after an incident often sounds complete because it arrives before the evidence has been organized. A worker missed a step. A guard was open. A supervisor signed the permit. The report then repeats that sentence until a hypothesis looks like a fact.

That habit creates a serious investigation problem. When the story hardens too early, the team may correct the visible symptom while leaving the conditions that made the event possible. OSHA’s Incident Investigation Guide for Employers, published in 2015, recommends a systems approach that looks for underlying causes, and HSE’s HSG245 guidance follows the same logic through information gathering, analysis, risk control, and action implementation.

Andreza Araújo’s investigation philosophy reaches the same conclusion from a practical safety-culture angle. In Sorte ou Capacidade (Luck or Capability), an accident is treated as the late result of barriers that failed, consistent with James Reason’s work on latent failures. In A Ilusão da Conformidade (The Illusion of Compliance), the investigation is useful only when it helps the organization understand what happened without turning the search for a person into the end of the analysis.

Why evidence quality determines the value of an investigation

An incident investigation is not a courtroom performance in which the team must produce one neat explanation. It is a decision process that tests competing explanations against the conditions that existed before, during, and after the event.

That distinction matters because the strongest evidence is not always the most vivid evidence. A photograph may show the final position of equipment, while a maintenance record shows a recurring defect. A witness may describe what they saw, while a schedule reveals that the task was changed twice during the shift. A completed permit may confirm that authorization occurred, while field observations show that the work boundary was never verified.

The practical standard is simple. Separate what is known, what is reported, what is inferred, and what remains unknown. The investigation becomes more credible when each conclusion can be traced back to more than one relevant source.

Myth 1: The person nearest the event is the cause

This myth survives because the final action is easy to describe. Someone bypassed a guard, entered a zone, skipped a check, or moved before the signal was clear. The action may have contributed directly to the event, but direct contribution is not the same as a complete explanation.

OSHA’s 2015 guide tells employers to investigate programs and systems rather than focus only on behavior. That direction does not erase personal responsibility. It prevents the investigation from confusing proximity with causation.

Ask what made the action reasonable or available in that moment. Was the control difficult to use? Did production pressure change the sequence? Was the procedure written for the actual equipment? Did the supervisor know that the task had changed? Was the worker expected to resolve a conflict that management had not resolved?

When the report names only the person, the organization loses the opportunity to change the setting. The better finding describes the behavior, then explains the conditions that shaped it and the barrier that should have prevented the event from reaching the worker.

Myth 2: The first witness account is the cleanest evidence

Witness statements matter, especially when the scene changes quickly. They are not recordings. Position, noise, stress, attention, prior expectations, and conversations after the event can influence what a person remembers and how confidently they describe it.

That does not make a witness unreliable. It means the account needs context. The investigator should record where the person stood, what they could see, when they noticed the event, what they did next, and whether they discussed it with others before the interview.

Use open questions before presenting a theory. “What happened next?” produces different evidence from “Did the operator ignore the procedure?” Compare accounts with photographs, equipment condition, access logs, work instructions, and the timeline of decisions. If two accounts differ, preserve the difference instead of forcing a quick consensus.

HSE’s HSG245 approach places information gathering before causal analysis for a reason. An investigation that interprets before it collects evidence can make every later fact serve the first story.

Myth 3: A completed procedure proves the control was present

Paperwork proves that someone completed a field, signed a form, or recorded an inspection. It does not automatically prove that the control existed in the work as performed.

A permit may describe an isolation that was never verified. A checklist may record a guard inspection without showing the condition of the guard. A training record may prove attendance without proving that the worker could recognize a changed hazard. The document is evidence of an administrative act, not a substitute for evidence from the task itself.

This is where James Reason’s distinction between active actions and latent conditions helps. A visible error may be the last opening in a chain whose earlier weaknesses include design, planning, supervision, maintenance, or decision rights.

Test the document against the field. Check whether the equipment, boundary, sequence, staffing, and control conditions match what the record claims. When they do not match, the investigation should examine why the documentation process allowed that difference to remain invisible.

Myth 4: One root cause must explain everything

The phrase “root cause” can create false pressure to find one final answer. Serious incidents often develop through several contributing conditions that align, which is why a single-cause conclusion can be less accurate than a small set of well-supported findings.

OSHA’s root-cause analysis guidance explains that investigators should look for fundamental, system-related reasons and recognize that there may be more than one. HSE’s investigation model also separates immediate causes, underlying causes, and the controls needed to reduce recurrence.

The useful question is not “Which cause wins?” It is “Which conditions had to be present, and which barriers failed to stop the sequence?” That question keeps the team from choosing one convenient explanation while ignoring staffing, design, maintenance, communication, competence, or management decisions.

Use a causal map, timeline, or barrier review when the event is complex. Then mark the strength of evidence behind each finding. A well-supported contributing condition is more useful than a dramatic root-cause label that no one can verify.

Myth 5: Closing corrective actions means the risk is controlled

Action closure is an administrative milestone. Risk reduction is a field result. The two can occur at different times, and sometimes they do not occur together.

Replacing a component, revising a procedure, delivering training, or issuing a communication may be appropriate. None of those actions proves that the exposure has changed unless someone checks the new condition where the work happens.

HSE’s HSG245 workbook includes action planning and implementation as part of the investigation process, while OSHA’s guidance emphasizes controlling underlying causes rather than stopping at immediate correction. The implication is practical. Assign an owner, define the changed condition, set a due date, and specify how effectiveness will be checked.

For example, if the finding concerns weak verification before restart, the effectiveness check should observe restart work, review the verification record, ask the supervisor to demonstrate the decision rule, and examine whether near misses show the same gap. A closed task in software is not enough.

What a stronger investigation does differently

A stronger investigation begins by protecting evidence and building a timeline before assigning meaning. It then compares statements with physical and documentary evidence, separates immediate actions from underlying conditions, and records uncertainty rather than hiding it.

• Preserve the scene and identify what may have changed before documentation.
• Build the timeline from task preparation through the event and the response.
• Test the first explanation against at least two independent evidence sources.
• Describe the conditions that shaped behavior, not only the final action.
• Give every corrective action an owner, due date, control objective, and effectiveness check.

This discipline also changes the conversation with leaders. Instead of asking who failed, the team can ask which barrier was expected to work, why it did not, and what management decision will make the next task safer. That is the practical bridge between incident learning and safety culture.

Andreza Araújo’s Safety Culture: From Theory to Practice is useful here because it treats culture as visible behavior under pressure, not as a poster or survey score. The investigation becomes a culture test when leaders protect honest evidence, fund the needed correction, and return to the field to verify the result.

Conclusion: A credible finding is tested, not merely repeated

An incident report earns trust when its conclusions survive comparison with the evidence. The person closest to the event may have taken the final action, a witness may have seen an important detail, and a form may show that a process existed. None of those facts alone explains how the event was constructed.

Investigate the sequence, the barriers, the decisions, and the conditions. Then correct what allowed the event to progress and verify that the change works in real work. Readers who want to strengthen the next stage can continue with the incident-investigation gap review and the near-miss reporting guide.

Safety is about coming home, and an investigation is valuable only when it helps the next person do that.