ABX test
A double-blind listening test in which X is randomly either A or B and the listener has to say which — the only way to show that a difference is actually audible.
The method came out of an American hi-fi club around Arnold Krueger in the late 1970s and was written up by David Clark in the Journal of the Audio Engineering Society in 1982. The listener hears A and B knowing which is which, then X without. Since nobody in the room knows the answer, expectations can't leak into the result.
A single guess proves nothing; the statistics do the work. Twelve correct out of sixteen leaves roughly a 4 % chance of lucky guessing, which is usually accepted as evidence. ABX testing was essential to tuning lossy codecs such as MP3 and Opus, and has an unfortunate habit of showing that expensive cables and high-resolution formats can't be told apart from cheaper alternatives.