First of all, if you had read the post above, you would have noted the call for level matching. The example in my reply to MM was just that, an example, and merely used to illustrate the dichotomy betwen the "night and day" comments and the "sounds subtle" comments, and what context it occured in.
Secondly, the expectations would not be capable of 'creating' this new revelation of a heretofore never before heard aspect of the music. I take it without saying so directly, that the trypical audiophile/music lover would have had many opportuniities ot be excited or anticipatory regarding other component swaps or auditions, and never before heard the 'new' aspect of the music, despite all the anticipation in the world.
[ I do wonder that you complain about DBTs done by the clubs who have stricter standards than you do, to wit, level matching within 0.1 dB and a 5% chance of error. ]
Perhaps you did not read part 1, part 2 and part 3? Those club DBTs all had/have serious and terminal flaws and errors, that render any results totally and completely useless. As far as any true scientist should be concerned, they weren't even a good null, but a trivial one, due to the poor controls and procedures.
I do recommend level matching IF appropriate, and audio interconnects (in general) are one component that does normally not need such adjustments. You still double-check anyway.
As for the value chosen for p, the 90% confidence level is not some sort of super lax criteria, it is merely another variable in the experiment, and as such, would require more consistent overal test results from several independant runs before being accepted as a valid positive by most. That does not automatically render the results from any single run, or any small series of runs suspect, or otherwise less than useful as what it actually is: a 90% chance of being a valid positive.
Jon Risch