Such contaminated tests are not too meaningful. If the situation was reversed, so too would be the result.
Even with DBT's the listening panel needs to be selected randomly, or at least pseudo randomly to avoid bias.
I think a big point being missed with DBT tests per Harman's method is that all listening is done in Mono.
Stereo listening can mask problems. With the complexity of modern surround sound tracks, much of what we listen to now is monophonic, just spread through many channels simultaneously. Speakers need to be able to stand on their own, period.
To your point Art Vandelay, Harman also has a listening program that they require their listeners to score highly in, 'Harman - How to Listen'. It trains you to be a critical listener and have the ability to separate sounds of different frequency bands and decide if one sounds better than the other.
Floyd's tests have also proven that listener's have the ability to listen 'through rooms'. Above the transition frequency of a given room (200-300hz) a speaker that has been positioned for a specific listener, would be rated similarly in any room. Floyd did this through 5 or more rooms with the same set of speakers.
He also called BS on his own engineers when he first arrived at Harman. "They were all professionals, I was told." So Floyd put them through both blind and sighted tests of the same speakers to prove the influence of bias. This is covered in Floyd's CIRMMT video on youtube.
My point is that Floyd and Harman went to great lengths to design products based on what their listening tests have proven, what people want to hear. B&W products do not perform well in their tests. The time I was in the speaker shuffler, whatever the B&W bookshelf speaker was in the $1500-$2000pr range, I rated it the lowest of the 4 I listened to. In sighted stereo tests, of B&W's larger models, the sweet spot was enjoyable, but the sound changed as soon as I moved my head. To me, that's not acceptable at $30k!