blind tests? what are blind tests?
A blind test is where you don't know which is which, when you're testing the samples. A blind test of colas would be if I gave a couple of un-marked cups of cola & asked which one you liked best. A sighted test of colas would be if you watched me pour some Coke into the left cup & some Pepsi into the right cup.
Another way to do it would be to pour Coke into both cups. In a blind test you'd likely get a 50/50 split; but if you labeled the test cups Coke & Pepsi your test results would skew towards Pepsi if you gave the test to people that usually drank Pepsi.
People have done blind test of compressed formats vs. uncompressed formats. You can see some results
here.
You can also do your own tests. Rip a track from a good CD with a good encoder (LAME) at various bit rates. Then burn a regular audio CD with the different bit rate versions of the track & include the original wav file.
Until someone can point me to a blind test that finds a difference between 320kbps mp3s and wavs, I'll consider it just an audio myth, like exotic cables, Brilliant Pebbles, and Clever Little Clocks.