How do Techradar's testing procedures differ from DxO? According to Techradar, the D800 has and edge in DR, but it's more like 1/2 to 1 stop instead of the 3 stop advantage quoted by DxO. This seems more in line with my own worthless subjective impressions of what I'm seeing in the files from both bodies. The D800 ,definitely has the edge, but it just doesn't look like a full 3 stops, which represents an 8-fold increase in light.
Secondly, Techradar's tests show the 5DIII's raw files as having a solid 2 to 2.5 stop DR advantage over the 5DII. According to DxO, the 5DIII has inferior DR to the 5DII. DxO also rates the D800 on top of the 5DIII in terms of noise, but Techradar shows the exact opposite. These mixed and conflicting results don't do much for proponents of judging IQ based on lab tests.
Kinda looks to me like (1) there is some buggy "testing" software floating around or (2) there are some buggy testing software users floating around or (3) there are some buggy camera users trying to test floating around.
Just remember folks, it ain't "science" until the results are repeatable...