Originally posted by bm75 that said, they have a scientific basis in their tests, so I don't want to be unfair towards them.
Many faults, though.
a) They don't publish exactly how they do the tests. Tests could be faulty, are unverified, have not been reproduced, and we don't know how many lenses they used for each test (because of sample variation, this could be a significant problem).
b) They test lenses on various cameras, which means you cannot compare the lenses between different cameras (not even from the same brand or with same sensor). Even if you look at the same lens tested on K-5 and K-3II, there are differences - and not just the ones you would expect. One lens gets much better, another only a little better, some get worse.. its weird. They should test lenses on an independent sensor (or extremely high resolution film - these do exist) so we could then compare lenses across brands. This should be fairly simple and everyone could reproduce the test
c) Same thing for camera tests - they get tested with some lenses, but I am not sure they use the same exact ultra high resolution lens for all camera tests, which would be required for objective camera comparisons. Some cameras get tested in special hires modes, but others don't. Some cameras have NR and sharpening built-in into the "raw", which also skews results.
d) Finally, not reviewing lenses on K-1, not publishing 645Z results (despite doing the test!), not using Pixel shift mode correctly with the K-1 review (but they
do use advanced features like PS when testing their own camera!),... either these are big blunders, or purposeful bias. I don't know why, but I assume its because Pentax doesn't have a big budget to pay for adverts and positive reviews (unlike Nikon, Sony, and Canon)
When looking for lens info, I look at other websites, as well. LensTip, Photozone,.. and user reviews on this forum. You can find sample photos on 500px, Flickr, in threads here.. relying purely on some numbers from DXo is a bad choice