How exactly do DxO mark arrive at that headline lens "score"? an average of lens performance across focal lengths/apertures? wide open? stopped down?As JRista explained, DxOmark appears to have a Nikon bias on lens tests - BUT, that's because, for whatever their reasoning, part of their lens test also included a measurement that takes into account the signal to noise ratio of the camera body it's being tested on. Possibly for consideration of final noise quality in printing.
The innately lower noise Nikon bodies thus boost Nikon lens scores..
I can see the logic for testing lens/sensor combinations even it won't give you a measure of lens performance relative to each other across different systems.
What I can't understand is giving one headline score or indeed exactly how this P-Mpix number was arrived at. At least with DxO's sensor marks its clear what the headline scores represent a certain level of performance at a certain setting or settings(even if you disagree with weighing performance in that fashion) and you have the ability to view other settings numbers. Here you just have one number then just an "P-Mpix map" that leaves you guessing based on colour and seems to max out at 12 P-Mpix.
Personally if anything I'm seeing more Canon bias here than I am with the DR figures. I'm no expert at the methodology of these tests but I can see nikon's clear advantage in DR when it comes to real samples. On the other hand I'm not seeing the "equalising factor of superior canon glass" nearly as much, D800 shots clearly look like they capture more detail than 5D3 shots to me.