wrote in message
oups.com...
[snip]
As to why ABX does not work in audio component comparisons? I
hate speculations. (Medicine had nothing but before it originated a
double blind evidence-based methodology)
But the lure is irresistible- so here goes.
If you look at the results of Sean Olive's double- blind
widely representative loudspeaker panel you see that majority performed
very badly when asked to differentiate between speakers. But the same
majority went unhesitatingly for the speakers with smoothest response.
Perhaps cerebral cortex performs badly when asked to
concentrate on differences between the complex musical signals
succeeding each other but has much less of a problem when asked simply
: which one do you like better?
This difficulty would be compounded with an ABX protocol in
place of simple DBT. There you have to memorise A, then B, and then
decide if X is more like A or B. It may work very well in audio
research on simple tasks (phase differences, codecs and such) but
complex musical signal may be beyond its scope.

This is very similar to my inclination. As pollsters know, how you ask the
question influences the answer that you get.