The science has been settled on that for many decades. Most of what remains is in the realm of psychoacoustics and human psychology.
We have dozens of members on ASR that prefer the sound of a sharp rock dragging through plastic over digital, so I would indeed say that the area of psychoacoustics ala Toole's research on speaker measurement preference or Olive's Harman target curve preference (far from flat) for headphones is needed. And it could go far beyond that.
edit: I think people are being far too kind to even vintage McIntosh. I would bet good money that those amps, Sansui, Fischer, HK, etc would measure very poorly.