For example, two speakers could be inaudibly different, or they could be audibly different. Now just finding out that they are audibly different does not say which one is prefered, that takes another type of test. But just saying that one speaker is prefered, does not say which speaker is more accurate.

All too often listening tests are performed badly and the results are questionable, but then they can be performed correctly, but the results don't say anything about preference or accuracy. Its trival to test for differences, difficut to test for preferences and an extremely difficult test to test for accuracy.

People will most certainly not always pick the most accurate speaker as the one that they prefer. Preference is not accuracy.

Subjective space is very shaky ground - thats hwy I prefer the objective which is far less prone to these pitfalls. But then whenever you quote the objective data someone says "Yea, but it sounds good to me."