This article was originally published on LinkedIn on 25 February 2026. It has been moved to the riskfacilitator Insights library so the website remains the permanent source.[1]

Every few years, I hear the same claim recycled in safety circles:

“Accident investigations don’t lead to learning.” “They’re linear.” “They focus on blame.” “They’re outdated.”

These arguments often come from people who have never conducted, led, or relied on a serious, no-blame investigation in a high-risk environment.

A recently published peer-reviewed study challenges those assumptions head-on.

Research led by Kym Bills at Edith Cowan University involved several years of negotiated access to 12 government Safety Investigation Authorities (SIAs) worldwide, including Australia, the United Kingdom, the United States, Canada, France, Japan, New Zealand, Norway, Sweden, Finland, and the Netherlands. These authorities are responsible for investigating the most serious aviation accidents in accordance with ICAO Annex 13.

What they found matters well beyond aviation.

What the research actually shows

The study examined what investigation analysis methodologies are really used by professional investigators when the stakes are high and the scrutiny is intense. Not what is written in manuals. Not what appears in simplified diagrams. But what investigators actually use to make sense of complex events.

The findings were clear:

  • Every SIA studied used multiple investigation methodologies, often within the same investigation.
  • Most used Rasmussen-based systemic approaches such as AcciMap.
  • Many used Reason-based models, including evolved and dynamic forms, not simplistic “Swiss cheese cartoons”.
  • Several used more recent systemic methods, including FRAM and CAST, where appropriate.
  • Many SIAs had developed bespoke methodologies, refined through years of real investigations.
  • Linear, single-cause thinking was explicitly avoided.

In other words, professional investigators are not stuck in the past. They are pragmatic, adaptive, and systems-focused. This directly contradicts the claim that Safety Investigation Authorities are somehow unaware of “modern safety thinking”. The reality is the opposite.

Choosing methods to suit the problem

One of the most important insights from the research was that there is no “one true” investigation method. Investigators choose and combine approaches based on:

  • The nature of the event.
  • The maturity of the system being examined.
  • The quality and limits of available evidence.
  • The learning value expected from the investigation.
  • The resources and time available.

Several SIAs explicitly warned against over-complex tools that look impressive but consume time, distort focus, or drive analysis down unproductive paths. Others cautioned against overly simple methods that miss critical organisational, regulatory, and systemic contributors. This is exactly what experienced practitioners already know.

Good investigation is not about applying a fashionable model. It is about building an accurate understanding of how work was actually done, under real conditions, across the whole system.

A necessary correction to the claim that investigations inhibit learning

Some contemporary safety narratives argue that we should move away from accident investigation entirely and focus only on “what goes right”.

The data in this study offers a measured rebuttal. Professional investigations, when done properly:

  • Do not default to blame.
  • Do not assume linear causality.
  • Do not ignore success or adaptation.
  • Do not operate in isolation from systems thinking.

They already incorporate:

  • Work-as-done, not work-as-imagined.
  • Organisational drift.
  • Trade-offs and competing pressures.
  • Regulatory and governance influences.
  • The interaction between human, technical, and organisational elements.

The problem is not investigation. The problem is poor investigation.

Why this matters beyond aviation

While this research focuses on aviation, the implications are much broader. The same arguments about “outdated investigations” are often used in:

  • Major events and crowd safety.
  • Construction and infrastructure.
  • Emergency services.
  • Media and production environments.
  • Maritime and offshore operations.

If we undermine investigation as a discipline, we don’t become more progressive. We become less informed. Learning requires evidence. Evidence requires disciplined inquiry. Disciplined inquiry requires structure. This research confirms something many of us already know but rarely see articulated with evidence:

  • Serious investigators don’t argue about models. They use whatever helps them understand the system well enough to prevent the next failure.
  • That is not old thinking. That is professional thinking.

If we are genuinely serious about learning, resilience, and prevention, we should stop dismissing investigation and start asking a better question:

Are we investing enough in doing it properly?

This paper was flagged by Michael Tooma in the context of broader conversations about officer due diligence and the quality of investigation and learning following serious events. Having worked inside many of these environments, the findings resonate strongly with what professional investigators actually do when the consequences matter.

Based on: “Aviation accident investigation analysis methodologies used by 12 government Safety Investigation Authorities”, Journal of Safety Research (2026). https://www.sciencedirect.com/science/article/pii/S002243752600023X?via%3Dihub[2]

References

  1. Paul Chivers, Accident Investigations Aren’t Broken. Our Assumptions About Them Are., LinkedIn, originally published 25 February 2026.
  2. https://www.sciencedirect.com/science/article/pii/S002243752600023X?via%3Dihub, source linked in the original article, accessed 10 August 2026.