Quote Originally Posted by 3db
Good read
Here's another more techie version by the same author: AES Report. I think it is important to understand what the finding mean - and what the findings of other blind tests don't mean.

1. This was a preference test of speakers with easily measured tonal differences.

2. The purpose of the test was to calibrate the trained listeners so that in house experts could be used for future product development instead of costly trials using larger audiences. Note the comment concerning the consistency of the preferences between the trained and untrained listeners. That information was gold to them.

3. You'll note that considerable effort was made to access the *performance* of the listeners.

"The combination of training and experience in controlled listening tests clearly has a positive effect on the listener's performance"

Indeed. Anywhere from 3 to 27 times better! Some even had perfect scores requiring some statistical fudging.

4. I think everyone is in agreement that the room makes a difference. With some speakers, more than others. Here is their "speaker shuffler"

While each speaker is rotated to the same position, this device really only works fairly with monopole box speakers. Dipoles will not get an optimum environment since their rear radiation contributes to their sound. Distance to back wall and the amount of reflections must be carefully optimized.

5. Null results in blind testing are the absence of a conclusion. Despite this, many argue that null tests *prove* one thing or another. This they do not. Nor should the results of tests between components "A" and "B" be automatically extrapolated to "C", "D","E", "F", etc. or - components that didn't even exist at the time of the test. This fallacy is used by Roger Russell with his wire links.

rw