I don't think they got this one right (again).
Yet another questionable listening test descirption and link to files (http://www.avsforum.com/forum/91-audio-theory-setup-chat/1967658-blindtest-between-sample-rates.html#post33426234)
The killer is that the test doesn't have enough unknowns (just 3) to do much in the way of serious statistics on.
The obvious problem is that 2 of the files are 2496 and one is 24192.
The less obvious problem is that the down sampling has significant aliasing up to 26 kHz.
The source material could be a bootleg given the way the audience clapping sound. And the music, if you call it that... (I must be getting old) ;-).