AudioBanter.com - View Single Post - science vs. pseudo-science

(Audio Guy) wrote in message news:4dJdb.455493$Oz4.260164@rwcrnsc54...
In article a0Fdb.611941$Ho3.119231@sccrnsc03,
(ludovic mirabel) writes:
(Audio Guy) wrote in message news:XWldb.603104$o%2.282900@sccrnsc02...

This just shows again that you have an incomplete understanding of
statistics. You state that a hit is 12 out of 15. If it is not a hit,
then it is considered to be within the realm of random chance and so
is not counted any differently whether it is 1 out of 15 or 11 out of
15. That does not by any stretch of the imagination "gives as many
different results as there are people doing it".

I said I will not discuss satististics but I'm always eager to
learn. You are right.
It is all the same. No differences. Or maybe there are SOME
differences but they are not statistical. They are just an illusion, a
puff in the wind, a knock and we are through the mirror in the
Wonderland.

Did you not employ statistical analysis in any of the medical DBTs
you were involved in? Because it's key in analyzing the results of
the test or tests. Without the statistical analysis you cannot tell
which results are just random chance and which are really
significant.

I will agree with your point that when results get very close to
shoving significance the individual in question should be retested.
But without the retesting you cannot state that they did hear
something, only that they may have heard something. But I have yet to
see you make that qualification.

Dear man , I qualified thusly at least ten times in the
last two years, asking why didn't the proctord pay attention to the
only interesting results, namely
those of the exceptional performers and rechecked them SOS. But I
can't expect you to memorise my collected writings. So for your
convenience: Yes they should have done it. Even though Greenhill ran
not just one but six tests and only two of his subjects scored
consistenly well
in five of them.
You don't say how many times they should have repeated it to
satisfy you and Pinkerton. Twice? 3 times? Ten times like Norman
Strong once suggested?
So let's collaborate on an ideal design
Your statistical prowess encourages me. Let us get an ABX
project, you and I together, based on Sean Olive's results. (of course
he did not use ABX but we respect his results, right?).
To do justice to the differences in performance between trained
(the best) , semitrained (in the middle - 3 times worse) and the great
unwashed (us audio consumers- just like the audio students- 27 times
worse) we'll get three groups going.
First the random collection of audiophiles. Get them ABXed on
anything reasonably comparable other than the grossly unlike
loudspeakers. The result almost guaranteed: "No difference, no
preference".
All is for the best in this best of all possible worlds.. Just
what you all would have wanted. The result is accepted without a
murmur just like your predecessors in the
"Stereo Review" days accepted Greenhill, Clark, Masters and so on.
As long as their ABX manipulated results on cable, preamp, amp,
cdplayer, dac were "They are all the same" . Electronics is wonderful
Now group 3, the trained. Some get 80% correct. Panic in the
ranks. This couldn't be! Repeat please. And keep repeating till they
say "uncle" ie. till they are half-deaf and ready to confess that
there is "No difference"- and can I go home, please?
The intermediate group , the salesmen, doesn't count. They
convince easy.
Isn't statistics wonderful too?
Ludovic Mirabel.