Michael Fremer's article in the February issue Stereophile talks about these types of comparison tests. What the reporters and skeptics fail to take into consideration is that we are listening to MUSIC over a SYSTEM, in a listening ENVIRONMENT, which means there is a world of interacting emotions, thoughts, physical/technological factors and other tangible and intangible variables (e.g, our own personal histories with a particular piece of music) simultaneously interacting that result in a gestalt experience when listenting to music reproduction. Change the variables and you've changed the experience.
In business we talk about findings that are statistically significant and that are managerially significant. Some finding might not reach the point of being statistically significant, but managers might still act on it because it is "meaningful" to them. Conversely, some finding might be statistically significant, but it is not managerially significant, and lacks "meaning". It is the same with listening tests. We can, and do, extract "meaning" from music that goes beyond conscious thought, and certainly beyond statistical results from listenting to unfamimlar music through unfamiliar equipment in an unfamiliar setting where "meaning" is non-existent.
All in all, I learn nothing surprising or of value from the article.