I posted the same message on AV123 as well.

Just some suggestions:

1. Ideally, you should use people who do not own either brand. Once they own, they will be able to recognize the sound and that will introduce a bias in their selection. This obviously makes the data less valid. We call this "selection bias" in medical studies. In other words, the sample of listeners should be "naive" to the subject speakers.

2. The person conducting the study (?you) should not know which speaker the person is listening to. This may introduce bias in your recording the opinion. This is called "double blinding" and should be easy to do. Once you have everything setup, you can ask your friend who does not know which speakers are what to play and record the results. You can analyze them. You could blind the person analyzing also (called triple blind), but the value is limited.

You are doing a fascinating thing. Let's do this as scientifically valid as medical research.