It would seem to me that we would have a lot more accurate data for subjects like aircraft parts where we use logs that track failures/hours used than we would for parts failures on revolvers where we rely on people to report incidents.
Any rare event is more likely to be over reported than under reported. An event that happens ten times out of 5000 gives us a chance to have 4990 false positives reported and only ten chances for a false negative to be reported.
|