"normanstrong" wrote in message ...
"Mkuller" wrote in message
news:BXb0b.183627$YN5.135766@sccrnsc01...
(ludovic mirabel) wrote in message
:
4) that reporting as the end-outcome the majority vote of ABX
untrained/untalented and ignoring the capable minority who DO
hear-exactly the way Greenhill had done (see Marcus quotes)is a
perversion of "testing".
(Bob Marcus)
Date: 8/17/03 4:57 PM Pacific Daylight Time
Message-id:
wrote:
"Ignoring the minority who DO have fewer and less severe colds
after
megadosing vitamin C is a perversion of testing."
See how silly that argument is? So stop using it.
Bob, your analogy comparing DBTs to medical research is wrongheaded.
If one
person in the Greenhill group, or any other group reliably
identifies a
difference - then it proves a difference exists! The fact that the
others or a
vast majority are not able to reliably identify it under THAT test
and THOSE
test conditions means next to nothing. It's a null - 0 -
meaningless.
That simply isn't true. All it proves is that the individual in
question got results that can be expected only once out of 20 tries
relying on guesswork alone--approximately 8 correct out of 10 tries.
What conclusion would you draw if an individual got 8 out of 10 wrong!
Norm Strong
Norman, in this particular case (Greenhill) he attempted to establish
some acceptable criteria of statistical validity compatible with
real-life restraints. Basing himself on the "binomial distribution
table" proposed by the ABX developer Carlstrom in his official
"Oakland" ABX website he, Greenhill, said that 12 out of 15 corrects
ie 80% in each one of his 6 tests will be considered "correct".
This means that all his subjects had to undergo 90 tests. Say 6
minutes minimum for each A, then B then X sequence. 9 hours for 6
tests. Nonstop or breaks for R&R, lunch, dinner and sleep?
To me the ABX is part of the problem- not the solution. That largely
explains why, indeed, most individuals do get your "8 out of 10 wrong"
and in my opinion always will. Even if someone finally gets around to
testing full-range speakers against each other. By MUSIC not train
whistles and pink noises
After that there will be nothing left to confirm that "they all sound
the same".
In the meantime please explain how you would improve on your
brother "objectivist" Greenhill's methods.
Ludovic Mirabel