So far not even a single link to a paper describing a DBT (for audio) with statistically valid positive results was posted here.
I am having trouble with your statement, but before I go any further, can I please ask what you mean by "positive results"?
Please do not talk about "statistically valid". I think most of us understand what that means ... just let us know what you mean by "positive results".
By "positive results" I mean that the difference in the sound with a change in (the system, the source, etc), was actually reliably detected rather than not detected in a double blind test. Only a "positive" result could provide a step in establishing the sensitivity of the procedure.
Cheers
Alex
To take this on face value, it would seem that you are saying that the only way a DBT could be considered valid is if there were positive statistically significant results identifying a difference that we are saying cannot objectively exist. If this be the case, then - by definition - it cannot ever be demonstrated.
However, in trying to place what you are saying into a scenario that
can achieve the desired result, I have this proposal, assuming we have the ability to switch configurations in a suitable manner:
- That instead of changing just one parameter - the audiophoolery device being in or out - we have a second parameter that is also changed. The change in this second parameter would need to be subtle, but one that would be
expected to be detectable, especially in a statistically significant sample.
- The test would be run, varying both parameters and the results analysed.
- For the results to be valid, the variation of the secondary parameter should be clearly identifiable. If not, then the whole test should be dismissed. With a valid test, if the audiophoolery product were actually making a difference, then this, too, should be clearly identifiable - but if there is no statistically significant differentiation, then the audiophoolery product can be said to have no effect.
@Alex Nikitin - is this the sort of thing you meant?
There are two questions, though. The first is "What should this secondary parameter be?". To be appropriate to the test, I imagine it should probably produce an effect that is the same as the audiophoolery product claims. That, in itself, would seem to be a challenge - but considering the claimed effects of these products are so indefinite, does it really matter?
The second question is: Do you have a 3 state test or a 4 state test? In a 3 state test, the configurations would be: (i) Vanilla (ii) with Audiophoolery device (iii) with secondary parameter change. A 4 state test would have one extra state: (iv) with Audiophoolery device
and secondary parameter change. (Not sure which way to lean on this one.)
Just putting this out there...