I have spent a career as a political scientist, with much training in the scientific method and statistics, trying to apply this to political questions. In hypothesis testing one needs valid measures of the variables of interest. Although objectivists don't realize it they want measures but really cannot deal with the issue of whether they are valid or not. Subjectivists doubt that any measure available are valid. I am a subjectivist of this type.
Unless measures are seen as valid, they are useless. Total Harmonic Distortion is an example. I remember a seminar at the RMAF where Atkinson had two amps on a desk. One was a Bolder and the other unnamed. He continually showed the Bolder to be superior with different loads in terms of THD. I asked whether the designer of the unnamed amp thought THD mattered to the sound and when we were going to hear both amps on speakers. I knew full-well that we were going to hear nothing. Many in the audience expressed dismay at my question. Mr. Atkinson only said that they did not have the capability to listen to the amps. Since I used to love Bolder amps when they first came out but not any longer, I suspected that the no name amp might sound better. I left the room thinking that THD was an invalid measure.
I have also repeatedly been involved in "double blind" tests. The "same/different version is utterly invalid IMHO. But having the equipment out of sight and listening and rating seems valid. But once when I was involved in such an experiment, a Bosak preamp rated best rather than preamps whose designers were among the raters. I had rated it among the middle group, but borrowed one to take home. My Dyna PAS 3 was clearly better.
I would be a strong objectivist were we to have valid measures. I would suggest that a system's ability to resolve the decay of notes if measurable would be valid.