There are fundamental problems that make blind testing audio entirely different (and therefore IMO, invalid) to blind testing as used in science labs where it is absolutely the only way to conduct proper experiments.  They were nicely outlined by joel's post back a page or two:

Michael.