So how many possibilities are in play here? I make it 3:
1. Control and A sound the same, B sounds different
2. Control and B sound the same, A sounds different
3. Control, A, and B all sound the same
That about right? So, 33.3% chance of guessing correctly?
[EDIT: Edited because I skimmed the OP too quickly. The test as set up doesn't attempt to determine which cable is the ESP and which is the lamp cord, only to identify any difference(s) between the samples.]
I know this would complicate things, and bongomania has already put a fair amount of work into this, but it seems to me that for accuracy, you'd ideally want to have a number of these tests, with the same people guessing. Some of you guys here would have a clearer idea than I do about the number of tests you'd want to carry out to get near statistical significance.
If someone gives the right answer to a single test, or even 2 or 3, that could represent actual perception, or it could just represent lucky guessing. (Guessing right 3 times in a row isn't farfetched; I'm sure we've all had the experience of flipping a coin and getting heads or tails several times in a row.) The more tests *each person* undergoes under proper testing conditions, the less likely it is that any differences in the ratio of right answers to wrong ones are due to chance, and the more likely it is that a significant difference, if noted, represents a real phenomenon.
The same would be true for wrong answers. If somebody gets it wrong on a single test, that doesn't necessarily mean that there's no perceptible difference and this difference couldn't be perceived on subsequent tests; it could also mean that the listener just missed that one.
That said, I dunno if anybody would want to do this 10, or 20, or 50 times....
