It is unfortunate that in the midst of clearly explaining what blind testing is, he states this:
"I’ve seen this effect a lot when doing my Golden Ears seminars (I publish a set of audio ear training CDs called “
Golden Ears,” and often present ear-training seminars using them). Listeners asked to identify the difference between two versions of the same recorded excerpt will have real trouble, at first, hearing that one version is 3 dB louder than the other. Once they are told and shown that such a difference exists, they find it “obvious.”
What does this mean? Was the test double blind? What was the second part where listeners said it was "obvious?" That sounds like a group poll in classroom and not blind. And what is the track in question so that we can try to replicate? Who were the listeners?
As it is, it reads to me that the group failed the test in blind, and then told the answer and they said, "of course I hear it now." That second phase is not reliable because we don't know if they really are hearing the difference or are going by bias created by the presenter.
Regardless, perception of level change is content spectrum dependent. Here is David Clark in his Journal of AES paper:
"High-Resolution Subjective Testing Using a Double-Blind Comparator:"
This is not an exhaustive test but roughly shows that perception of level difference tracks our hearing sensitivity which is the best in mid-frequencies. 3 dB difference there is well above the thresholds in the above study.