"Cub Driver" wrote in message
news

Are you saying the whole thesis in The Killing Zone is based on such
an elementary methodological error?

Not entirely. But you are dealing with such small numbers here that
the only way to make them valid would be to keep the dead pilots alive
and let them keep on flying.

I don't think that's correct. The percentages are small, but with hundreds
of fatalities per year, the magnitude of the "killing zone" difference would
pass a statistical-significance test. It's just that the author isn't
measuring what he claims to be measuring.

Fatalities are only caused by people who crash. Once you're killed,
you don't get to play any more, or in this case to accumulate more
hours. Even if fatalities were distributed at random among the entire
pilot population, the survivors would necessarily have more hours than
the ones who were killed.

Not by a noticeable amount. The fatality rate is only 0.05% per year. If the
fatality rate per hour of flight time were constant as a function of total
hours flown, the distribution of flight-time hours among the dead would be
virtually identical to the distribution among the survivors. (If the
fatality rate were, say, 20% per year, then your point would apply.)

--Gary