Here's a way to scientifically test subjective bias: make a cable out of exotic and expensive materials (99.9999% silver, let's say) and make it up to the same specifications that an audiophile cablemaker would. Now, gather a bunch of audiophiles, and explain to them what's in the cable, its market price and random marketing BS that a cable company would likely spout out. Next, make three groups: the control group, who actually weren't briefed on the contents of the cable, and are required to take the test blind. The second group would be given fake "impressions" on the cable - explaining things common to cable reviews, like "lowered noise floor," "opened up treble" and "toned back bass" or something. They would perform a sighted test and write down their impressions afterwards, without talking to each other at all about it. The third group basically does the same test without seeing any "impressions," and afterwards have to write down impressions of their own.
After it all, if the tests all match up (i.e. the second group disagrees with what the "impressions" say and their own impressions match up with the impressions of the first and third group) then it would show that in fact the cable may have made a real difference. If the two non-control groups had similar impressions (due to knowing the materials the cable is made of, marketing and blah blah) and the control group differed significantly, then it would make a strong case that it is simply subjective bias in play. If everything is random all over the place, it would basically just be a null result.
Not sure if I got everything right there though, I just crapped that test in a few minutes.