What proportion of speakers that are manufactured and purchased does the preference score apply reasonably well to?
Probably a lot of them, I never said the preference score wasn't a useful guide for the speakers it does apply to; it is, although it's certainly far from absolute. It functions as a reasonable way to eliminate poor and mediocre speakers from consideration as long as they're all of traditional design. For comparing speakers within 1 or 2 points it's completely useless though. So again, it's an interpretation of some of the measurements.
But that's got nothing to do with whether it applies in this case. It doesn't matter how many speakers the preference score applies to if the one you're discussing is significantly different from the ones used to derive it in the first place. No more explanation is required here.