Three analytical traps in accident investigation
Vložit
- čas přidán 28. 08. 2024
- In this video Dr. Johan Bergström introduces how the US NTSB investigation into the accident of Asiana 214 falls into the following three analytical traps:
1: Counterfactual reasoning
2: Normative language
3: Mechanistic reasoning
For those who stand to watch the entire video, there will be a bonus trap at the end of the video :)
www.humanfactors.lth.se
The processes explained here are fundamental to the analysis process and forming reasoned arguments for the investigative process.
Opinions, bias and obvious conflicts of interest are additional traps to negotiate. formed, reason arguments particularly as an IIC on major investigation avoids conflict
The accident final reports are consensus documents in accordance with the comments process in ICAO Annex 13; the language of consensus is tame.
There are exceptions, the BEA are effective at communicating their message and recently the Norwegians dropped the facade of normative language and went for EASA and Airbus over a helicopter accident that should not have occurred.
There's a 4th trap, and that's Logical Fallacies. Either-way, these are well constructed and informative videos which highlight the logic and reason required to form objective rationale.
Very informative videos. I am hopeful that more will be posted. Well done.
Where are the new videos? I've been reading Woods, Hollnagel and Dekkar, but the safety industry is still talking behavior, swiss cheese and pyramids.
I try to watch this often when I'm in the midst of an investigation.
Well communicated. Thank you.
Very interesting... but I'd like to suggest that a portion of these reports are worded specifically for consumption by the general public and living victims for the event. When it comes to assigning blame for a large scale human tragedy there are often political and economic elements that cloud the investigative process.
Exactly. Some reports read more like a defense of what happened. See this thick report here? It's not our fault at all. Maybe a bad part, the pilot was daydreaming, maybe both but we have no control over that. We'll have a mandatory class or something like that. Like Friday said in Dragnet, "Just the facts maam". The real root cause. Then real corrective actions.
It would have helped me better understand why these are traps and are bad if you'd have spent an equal amount of time illustrating some example alternatives to the passages criticized. Much of the language and approach in the example report are typical in air crash investigations.
Stay tuned Shawn, we have that as a plan for an upcoming video.
I'm pretty sure there's a hole in the logic of this presentation. Either you missed the definition of the list (for any number of reasons) therefore possibly impeding your ability to explain how it works or (quite possibly) the FAA "Analytical list of traps" guidelines are flawed. Bold statements? Yes! However, we'll stick to the logic in this presentation in this rant. I am going to have to read the official FAA guidelines before going for them. Correct me if I'm wrong. Anyway, here's how. Assuming a system or person is either right or wrong without knowing personally (before investigation) and not being capable of finding who or what's really responsible after documentation (after investigation and reading the safety guidelines as they currently stand) is observed, renders the entire judgment process, null and void. Impotent. Who decides what the real factors are involved, if no one or nothing is responsible or not under these presented guidelines? The problem isn't what works and was done successfully, it's what didn't and wasn't successful. We're talking about lives here. The Passengers living due to what was done right isn't a factor. Especially, if it's mentioned about what was, wasn't, could, couldn't should and shouldn't have been done in the report. That's all that can be possibly known and surves as a guideline in itself. For experts to rehash a known, predictable, documented system after knowing what worked or not from the investigation is counterproductive and further complicates investigation, communication and resolution and costs lives. Logic is a very rare dicipline these days that most will not follow. Humans aren't so predictable because, of this. A lot of this just sounds like preventing hurting someone's feelings for using apparent terminologies for apparent facts. Feelings never saved lives. Ended them more than not. Empirical and Deductive Reasoning are all you need.
I'm not sure I understand all your points, but your last sentence troubles me greatly: "Empirical and Deductive Reasoning are all you need.: Agggh - no! This is exactly the 100 years of conventional (and fallacious) wisdom what we in the safety profession are now trying to overcome!
Logic is not the same as meaning and context...(which is (I think) the point of Dr. Bergstrom's video). Context determines behavior. Conventional analysis however eschews context and focuses on conditions, decisions and behaviors, without going deeper to understand the 'local rationality principle'...that people make decisions based on what makes sense in the moment, and sense-making is HIGHLY contextual. Empirical and DEDUCTIVE reasoning are exactly what enable investigation biases and heuristics...which tend to confirm already pre-conceived conclusions.
INDUCTIVE reasoning assumes I have a goal in mind (to understand contextual factors involved in an accident), but I don't have a hypothesis yet formed. DEDUCTIVE reasoning requires me to gather evidence and support (or idealistically, disprove) my conclusion, but because of our biases, conclusions usually support our starting hypothesis (especially in complex, adaptive environments).
INDUCTIVE reasoning requires us to seek to understand and gather facts to form an hypothesis, rather than identify independent and dependent variables to support our pre-existing hypothesis.
Over 30 years, I've tried it both ways. Deductive reasoning leads to conclusions that are emotionally satisfying but not preventive. Inductive reasoning leads to incident conclusions that are contextually richer, provide more lasting responses, and identify "adjacent possibilities" for improvement that conventional incident analysis never will.
You are skipping issues related to the investigative process in the US. The NTSB is charged with finding a single cause of an accident. This makes them deal with accidents in a specific manner. We all know that most accidents are caused by multiple factors. You focused on the language they used describing specific events and actions but you didn't discuss whether they were describing the aircraft state accurately. Further, their choice of words is designed to not make a definitive statement on crew mindset unless they specifically have it documented. The listing of operable systems is functional and provides information to those unfamiliar with information on the state of the aircraft. Finally, they are judging. That is their stated role in the US by law.
Then US law needs to be changed, as air accident investigation should never be about judging or apportioning blame. It should only be about learning what happened, how it happened, and what can be done to prevent it or a similar event happening again. Leave the judgement to the courts.