There are, that I can think of immediately two basic ABX testing reasons / methods.
Some rules of good ABX testing are that the switch be instant and that the sources are level-matched. Louder sources will almost always sound better, and the brains ability to hold very specific audio detail lasts a fraction of a second. If one does not follow these rules, you are simply listening, not really ABX testing.
If we wish to arrive an a objective conclusion... to test what we think we hear, then ABX testing should be done blind. Knowing which speaker, amplifier, or encoding you are listening to will color your subjective experience.