I think John K said this:
To have some chance of characterising an amplifier I have seen the following list of measurements being cited as necessary:
"S/n ratio, frequency response, IMD/multi-tone distortion (for all levels, all loads, full bandwidth), channel separation, distortion of crosstalk signal, load tolerance, output impedance, stability margin, supply rejection, EMI susceptibility, ..."
Add in phase response and you got a pretty good measurement suite. And add in if the device or distortions all null to something less than say 75 or so db on music, then dang near nobody, nowhere can humanely hear a difference. Now, we can all decide if "what" we hear is close to the real thing, but thats not measurements, measurments are simply revealing the replication of what signal is input. Tone per our ears is per our ears, and not the same thing.
That null test is very hard to argue against, just the db threshold that humans stop hearing things at is arguable.
Do "standard" tests measure everything, no, they don't. No true, educated, objectivist would say so. Is there a point where a null test can prove that one is hearing things in their imagination, yes, at what level down does the null test have to be is debatable, but IMO null testing is absolute! Notice that CARVER used that test to match two different amps to the point they sounded the same. There you have it, if you want it.