I totaly agree with you on the first bit - a double blind study is near impossible to conduct to test the NSD because of the variation factor, however, this same statement would explain why NSD doesnt work to the same effect on every AS suffer because of the extreme variation of its application.