well yes agreed and agreed I suppose. All other things being equal - which they seldom are (in my experience anyway).
Thinks like the perez-thalmann lens resolution tests drive me a little bit crazy. I think they are GROSSLY misleading (although the fact that the results are all OVER the place prove this to a degree) due to only using a single sample of EACH lens - and then again - each from an unknown origin - in the meantime lens cell spacers are lost, cells are put on the wrong shutters, the variation in manufacturing tolerances can be very significant, etc etc etc...
This is not a comment directly on your discussion of testing - so much as venting about 'limited sampling' - even doing EXTENSIVE tests but only with a single sample can be misleading. It violates the first principle of scientific testing - which is 'take many many samples!'
But then again - who has time?