Originally Posted by
engine out
No management pilots. A Fatigue group representative Captain on one, a normal line Captain on one and a AIPA rep the Captain on the other. The FO are all normal line pilots as are the SO’s. Fatigue monitoring for ten days before and after and reaction tests every 2 hours in flight. Yes it’s not a large data set but better than no data set.
A little "data" is more dangerous than zero data.
If it is proposed that IF the "data" supports the fatigue profile, the union accepts this type of flying, then questions ought be asked about probity.
By all means conduct a study, but at least make it robust. The health and well being of union members depends on it.
Other stakeholders are far more skeptical.
From a statistical point of view obvious problems are many.
- Model is not representative of reality.
- Passenger load not commercial
- Flight profile not typical
- Pre and post "sample flight" duties are unknown. What about a situation where the company applies minimum turnaround, an additional duty or the like? This is conveniently excluded.
- Biased sampling.
- Insufficient observations (3)
- Interrogation of sickness and correlated association. Are those pilots/cabin crew who are seriously ill long haul or domestic? Does the union know this? The sick rates may already infer a problem exists.
No matter the argument, a study paid for by an employer, designed to ensure a commercial outcome is axiomatically likely to result in the outcome being pre-determined.
The first and most important rule of statistical model is unbiased sampling. Having achieved that, there must be sufficient observations, over sufficient time to draw meaningful inferences for the generalised form.
This rubbish does neither and is a thinly veiled commercial consultation.