A commission publishes its final report. The concluding section aligns about a dozen findings, half a dozen contributing factors, a few recommendations — and not a single cause. The next day, the press headlines « the cause of the accident, » the civil party’s lawyer highlights a paragraph, and in the preparation room, someone remarks that the report « doesn’t dare say anything. » It says a lot, but in a language where every word has been weighed for what it entails, and which is read according to its own rules.
The word « cause » does not belong to safety analysis
A safety investigation has a single objective, written in black and white in the international convention that establishes it and repeated by the European regulation: to prevent future accidents. It is not intended to determine faults or establish responsibilities. This sentence is not a stylistic precaution; it is the condition for the operation of the system: without it, no one would speak to the investigators, and the raw material of the work — crew statements, workshop logs, an operator’s internal notes — would dry up in a season.
Now, « cause » is first and foremost a legal term. In law, causality is what links damage to a generating fact and opens the way to compensation. A report that wrote « the cause of the accident is the non-compliance with procedure by the captain » would provide, ready-made, the demonstration that a court seeks to build, for a use that the investigation does not pursue. The lexical shift toward « contributing factors » is therefore, in part, an assumed defensive maneuver.
But reducing this choice to legal caution would miss the essential point. The vocabulary has changed because the accident model has changed, and a systemic model makes the singular « the cause » factually false. This is the second aspect, the one that interests the practitioner: it is not just that we do not want to name the cause, it is that there is not one.
Safety reports don’t look for a single cause because accidents usually happen due to multiple factors working together. Saying “the cause” would be like saying one domino knocked over the whole line – but in reality, many things had to go wrong at the same time.
In the 2018 Lion Air 610 accident, investigators found multiple factors: faulty sensor data, maintenance procedures, pilot training, and design flaws in the MCAS system. No single one caused the crash – they all combined in a specific sequence.
Four registers, four writing commitments
A well-constructed report separates registers that do not have the same level of proof or the same effect. The finding, or established fact, is verifiable and sourced: a recorded parameter, a measured wear mark, an entry in the logbook. It contains no interpretation, and that is why it is rarely contested. « The high-lift device was in the clean position at the time of power application » is a finding.
The contributing factor adds a demonstration to this finding: it shows by what mechanism the fact weighed on the course of the event. It does not claim either necessity or sufficiency. Saying that an interruption of the checklists contributed means that it made the outcome more likely or more severe, not that it produced it alone.
The cause, when a commission still writes it, adds an assumed hierarchy: among the factors, these are designated as determining. Some commissions maintain this section, others have abandoned it in favor of a single, explicitly non-ordered list. The difference in practice is not anecdotal: it changes what the reader is entitled to conclude.
Finally, the safety recommendation follows a distinct logic from all the previous ones. It does not mechanically follow the heaviest factor, but the most modifiable one. A major but unchangeable factor — a weather condition, the physics of a stall — will produce no recommendation; a secondary but correctable factor with a stroke of the pen in a manual will produce one.
| Registre | What it takes to write it | What it does not assert | Use for the reader |
|---|---|---|---|
| Constatation | A physical or documentary source | No effect on the outcome | Enforceable factual basis |
| Facteur contributif | Demonstration of an effect on the sequence of events | Neither necessity nor sufficiency | Hook for a barrier |
| Cause | A deliberate ranking between factors | Not legal liability | Quick reading, often misleading |
| Recommandation | An addressee able to act | Not a judgement on the importance of the factor | What directly concerns you |
The test that separates a circumstance from a factor
How does a commission decide? Through a counterfactual reasoning, applied fact by fact: one mentally removes the element from the sequence, leaving everything else unchanged, and asks whether the event would have occurred in the same way. If yes, the fact is a circumstance: true, documented, with no demonstrated effect. If the answer is no, or if the outcome would have been significantly mitigated, one has a contributing factor.
This test has a known weakness, and it must be kept in mind to read correctly: it favors the proximal. The further back one goes toward the organization, the less conclusive it is, because distant conditions are substitutable. Remove the action of a crew at the decisive moment and the outcome changes; remove a policy of rotation scheduling and nothing changes that day, since another fatigued crew would have been in command the following week. Pure counterfactual reasoning thus mechanically brings the focus back to the last minutes, and that is why commissions complement it with an analysis of latent conditions, which does not follow this test.
The trap: extracting a factor from its complete formulation. « Non-application of the interrupted approach procedure » cited alone becomes a crew error; the same factor, as the report writes it, often reads « non-application of a procedure whose three steps were not executable in the available time with the installed equipment. » An internal memo that copies the first version and not the second has reversed the meaning of the report, and you will not see it before the next event.
Why the bundle replaced the chain
The sequential model — a first deviation triggers the next until the accident — long structured the reports, and it remains appealing because it tells a story. It has a dreadful corollary: in a chain, each link is necessary, so each can be designated as « the » cause, and in practice, the one closest to the outcome and easiest to name is chosen, i.e., a person.

A modern aeronautical system does not work that way. It is built with superimposed defenses — design, certification, procedure, training, control, alarm, simulator training — and each is supposed to stop alone what the others let through. An accident there therefore assumes the unlikely alignment of several partial failures, none of which is sufficient on its own. In this configuration, writing « the cause » is not only imprudent: it is arithmetically incorrect, since no element removed in isolation explains the outcome.
This leads to a practical consequence that is often underestimated: the deeper a factor is in the system, the more far-reaching the correction it calls for, and the slower and more costly it is to obtain. Recalling a point of flying technique during the next periodic check can be decided in a meeting and protects a fleet for a cycle; rewriting a manufacturer’s procedure, modifying a certification requirement, or changing a scheduling principle takes years and protects everyone durably. No report will tell you which to choose: it gives you both options and leaves the decision to whoever holds the means.
Reading a concluding section without falling into traps
Four reflexes are enough to read correctly. The first concerns the modality: « it is established that, » « it is likely that, » « it has not been determined whether » are not read the same way and are not used the same way in an internal note. The second concerns the order: some commissions specify at the beginning of the section that the list is not hierarchical, others do not specify it and the order is purely chronological. Check the introductory note before assigning weight to the first item cited.
The third reflex is pairing: match each factor with each recommendation. A factor cited without an associated recommendation is the most instructive part of the report, because the explanations almost always hold in three — the factor is not modifiable, it is outside the scope of the recipients that the commission can address, or an action is already underway. The three concern you differently, and the third tells you that a text will change.
The fourth concerns findings with no demonstrated effect, those that are skipped in quick reading. They appear in the report because the commission judged that they could be useful elsewhere: another file, a trend, a hypothesis not concluded here. This is raw material deposited for the sector’s attention.
| Report wording | What it means | Legitimate internal use | Common misuse |
|---|---|---|---|
| It is established that… | Physical or documentary evidence | Can be quoted as is in an analysis | Aucun |
| It is likely that… | Best-supported hypothesis, not demonstrated | Basis for precautionary action | Presented as a fact in a briefing |
| May have contributed… | Plausible mechanism, effect not quantified | Point to watch | Justification for a sanction |
| Could not be determined | Data are missing, permanently or not | Signal about the traceability available | Read as an absence of any problem |
What this vocabulary changes in your own reports
The subject is not academic: the same grammar applies to the event analyses produced by your safety management system, and that is where it degrades the fastest. The natural tendency of a rushed internal analysis is to write « crew error » or « non-compliance with procedure » in the « cause » box, then conclude with a reminder. This pair has a unique merit: it closes the file. It produces no barrier, it does not survive the departure of the people concerned, and it teaches the rest of the staff that reporting exposes them.
Two writing rules are enough to correct this. First, describe an observable behavior rather than a mental state: « the altitude announcement was not made » can be verified, « the crew forgot the announcement » assumes a content of consciousness to which no one has access. Then, only retain as a factor what you can link to a targeted barrier. If no barrier appears, you have not identified a factor: you have formulated a reproach.
The right reflex: require every analysis to include a « targeted barrier » column, filled in before validation, and check that it never contains the word « reminder. » A factor that only leads to a reminder of a rule is almost always poorly formulated: rephrase it one level higher — why was the rule applicable in one case and not in another — and a concrete measure will appear. The exchanges between safety officers on forums.fl330.com largely revolve around this column.
What you still need to check in your own documentation
Three points deserve control in your own documentation. The event analysis form, first: if it still has a box titled « cause, » you mechanically produce conclusions with a single term, regardless of the quality of the analyst. The requalification process next: who has the right to transform, between entry and annual synthesis, a factor into a cause or a finding into a factor? This step is rarely written, and that is often where the vocabulary slips. Finally, the fate of external reports: a received investigation report must enter the safety management process through an identified door, with someone in charge of deciding on its applicability to your fleet and your sites.
There remains a gray area that no one has resolved: the preliminary report. It contains, by construction, only established facts at this stage, without analysis or conclusion. Taken outside, it becomes a cause in twenty-four hours, and an operator finds themselves commenting on an explanation that the commission never formulated. The only tenable response is to draw no operational conclusions from it before the final report, except for immediate safety recommendations — which, they, exist precisely for cases where one cannot wait.
For those preparing for a qualification, a selection interview, or a position in flight safety, the exercise boils down to a verifiable skill: to present a known case without ever using the word « cause, » by naming the factors, the barrier that failed, and the one that held. A recruiter immediately hears the difference, and the FL330 preparation modules make this a mandatory step, because it is also the language in which the reports you will be expected to write are drafted.
The applicable text, your operator’s approved documentation, and the reports published by the competent commission prevail in all cases over the reading guidelines proposed here, which have no other purpose than to help interpret them.

Leave a Reply
You must be logged in to post a comment.