The best manufacturers I know use testing as a starting point, and fine tune be ear from there, they remeasure to confirm they're still in the ball park after listening sessions are concluded. The one's I've known well admit that measurement doesn't detect nuance or subtle things well. It's important to note that microphones and the ear/brain detect sound very differently, so relying too heavily on measurement without developing the skill of listening can give a false sense of being right. Hearing is a sense, but listening is a skill that can and should be developed.
The concern of bias can work both ways. Knowledge of a specific measurement can just as easily sway our perception as knowledge of a part change...it's just tougher to argue if objective data is considered truth. Long term listening tends to overcome initial bias.
It takes me extended listening sessions....sometimes days and weeks to decipher the full impact a small change in my system has made. Big changes show up quickly, but small ones do not. That long term view helps overcome day to day variables, and provides ample opportunity to get familiar with a given change. Most blind tests are done with very quick samples, often on a system and room other than our own, and sometimes with music samples we're not familiar with. If the test is done with more than a couple of people, there's the issue of less than optimal seating. Add in the pressure of trying to hear a difference on demand, and it's all too possible for the listening test to prove ineffective....at least for subtle changes, which things like cables, caps, etc., tend to be. I simply don't consider most blind tests to be strong evidence as proof of anything but the most obvious differences.
AFAIK, Danny did do a blind cable test, and correctly identified 8 out of 10 changes. The two he missed with done with music he didn't pick, which highlights the importance of the material used.