Comment by nullc

It's also a reason why double-blind testing is important. If someone doing the setup is expecting one piece of kit to sound better, if it doesn't they'll check the configuration more, and difference in gain can come from many sources. So errors that result in higher gain in favor of the "better" candidate go uncorrected, while ones that favor the worse tends to be fixed.

Point being: it doesn't even require an unscrupulous sales person to get similar results to an unscrupulous sales person! :P