Originally Posted by
BoeingDriver99
Well no because if the data was properly analysed using inferential statistics then one off events would not make any difference. Trends over time might make a difference eventually but for example; the data up to present day could result in interesting results - for example; no difference, small difference, large difference. They would be interesting results in their own way.
The unfortunate truth is that trends usually don't give actionable intel towards stopping the next bingle. They look like they should mean something, but most times the next loss or embarrassment comes from outside of the focus at that time. In recent events, the only glaring example that really stands against that is the USS Connecticut's allision with the south china sea. The squadron command had highlighted deficiencies that seemed to be defied by the crew command, and either following being overridden by higher command or a change of heart, the event mirrored the concerns of the squadron command.
No trend picked up the 2 x B777 events in DXB, the take-off one or the GA one, they are system resonant events that come from what otherwise looks like a normal day at the office. The Exeter landing would not necessarily have been proceeded by actionable QAR alerting of stable approach criteria unless the screening looked at the full path of interest, rather than 2 snapshots. Even with screening triggering the unstable approach as an event, unless attributed to a common pilot, their problem doesn't get highlighted, it becomes a blip in the overall scheme of events. There are ways to get a meaningful understanding of the risks that the operation has, but it means a change of the manner by which we do business, and it is incompatible with pathological management teams, which are overrepresented in aviation.