Look, the most appropriate measurement is, and it should be obvious, that of the human ear, and this is a point many audio objectivists miss. Measuring by electronic equipment is merely for practical convenience, but listening tests are a must for truly meaningful and comprehensive results. In the sciences where humans are part of the system, from drug testing to psychology experiments, testing on subjects is a given. Specifically regarding perception, companies and researchers working on new video and audio codecs know that the final test that must be passed is that of perception rather than some measured statistic.
On the other hand, this does not really support the subjectivist positions since these perception-based tests must be conducted rigorously to be valid, and they are thus inevitably blind tests (better yet, double blind tests). The frequent objection from the subjectivist camp that a negative in a test doesn't prove lack of difference is correct; however, a series of negatives probabilistically implies that. Of course, the assumption here is that such tests are conducted properly, which generally has not been the case in audio equipment testing. For some good discussion of methodologies that try to satisfy both camps see Jon Risch's comments on the AudioAsylum forums.
|