In article ,
"Peter Duniho" wrote:

If pilots made for better drivers, then average pilots should have fewer
accidents in airplanes, relatively speaking, than average drivers do in
autos.

You can't make that conclusion or assumption.

Sure I can. I just did.

ok. sure, you can. But it isn't valid.


The only valid test of pilots making better drivers is to look at some
means
of putting pilots through drivings tests vs non-pilots. Looking at the
statistics of drivers involved in auto accidents to see if there is a
statistical diffence between pilots/non-pilots would be relevant but not
conclusive.

Even your proposed driving test would not be conclusive. Tests have biases
and inaccuracies too.

Well, of course a flawed test would be useless. But not all tests have
meaningful biases or inaccuracies.


That said, statisticians make a pretty good living discovering interesting
facts about the world through nothing more than simple study of the existing
numbers. If you really care, you might want to read the book
"Freakanomics", which has lot of interesting case studies in statistical
conjecture.

Discovering a correlation doesn't prove cause and effect, a mistake
way too many people make. Absent proof of cause and effect, these statistical
"facts" are generally just (potentially) interesting trivia.


Sure, it's technically conjecture, but there's very little in the world that
can actually be *proven* -- there is always a non-zero chance that the
attempt at the "proof" is flawed -- and statistics, when applied in a
careful manner, can reveal all sorts of interesting truths.

None of which supports your orginal thesis or even validates your approach.

--
Bob Noel
Looking for a sig the
lawyers will hate