The mistake a lot of the people in this thread seem to be making is that
zeroing in on the flight test and trying to use the results of the
flight test to establish an ACTUAL quality level for the pilot at that
point in time doesn't equate. All the flight test does is establish that
the pilot being tested has met a MINIMUM STANDARD.


Correct as stated, but it is not unreasonable to make statistical inferences.
For example, assuming any reasonable (such as gaussian) distribution of pilot
abilities at flight check time, a higher =average= pilot quality will translate
into more passes and fewer fails. Assuming a similar distribution among pilots
who take accelerated vs standard training, the set of pilots with the highest
level of fails is likely to have a lower mean than the set of pilots with the
lowest level of fails.

I do grant that (and this is what I think you are getting at) one can correctly
infer nothing about the shape of the pilot distribution from the pass/fail
ratio, and even that given a distribution (such as gaussian) one can correctly
infer nothing about the sharpness of the peak from the pass/fail ratio, nor
about the ability of any individual pilot from his pass/fail result. But that
is not necessary to address the underlying issue.

Jose


--
(for Email, make the obvious changes in my address)