I don’t see why the analysis proposed by the OP couldn’t work if done properly. I don’t understand **Miller ** and **Glee’s ** objections.
Sure, those who believe in precognition tend to say it is unreliable. Sure, they’re not always going to believe their alleged precognition and may fly/drive etc anyway. But all it takes is for precognition to work sometimes for some people and to cause some of them not to fly and there would be *some * effect and if you studied enough incidents the effect would reach a point of statistical significance if it existed.
Sure, it’d be invalid to compare occupancy levels on a crashed plane on an unpopular redeye flight by one airline to occupancy levels on a highly popular peak hour route by another airline, but no one but a statistical idiot is going to do that. Nor would you study routes that were always full (or over-full). You’d find a crash on a particular less-than-capacity flight on a particular day, and then compare occupancy with the same flight in former weeks over a long period.
Of course, if you did the study you’d find nothing, but to dismiss the possibility of the study on the assumption it would be done in a stupid way is being kneejerk.