I posted an experiment awhile back where I tapped the lens hood of a 300mm lens in the middle of a one second exposure. I used a 16mp camera and a 12mp camera for the test, both DX. I repeated the experiment 4 or 5 times with each body. The 16mp camera showed more smear than the 12mp. That's the only test I've ever seen on this issue. Everything else was conjecture.