@ Janneman,
as SY pointed out- it could be significant.
Probability to get right answers by chance in these sort of tests is p=0.5 .
If a listener is right on every trial he reaches a probability of p=1 .
So, normally in the analysis we are looking for p>0.5, but if somebody is wrong on every sample he reaches a probability of p=0, and it maybe that he hears a difference in every trial but is unable to give the right answer.
Results like the one you´ve described should in every case get a further examination.
An interesting reading on this topic are Les Leventhals articles in the JAES. He creates the term `Statistical significant poor performance´ for these results.
But unfortunately his articles are a bit hard to read for the stastics challenged. 
But i think well written so definetely worth a try.
Wishes
|
