Lens tests only apply to the SPECIFIC Camera model they were tested on. You cannot compare results taken with different cameras.
Are you sure they were taken with the same camera??
If you compare the difference between multiple lenses on each site you can usually see when one is getting significantly different results from the others.
On TDP specifically you can also look at the 200f2 samples for a comparison of all the different bodies.
It looks like all review sites have dud lenses (relatively speaking) in their line-up. Ideally each site would test three or four samples, but that doesn't quite seem to be practical. I always look at as many reviews as possible when considering a lens.
The best example I can think of right now is the Sigma 120-300 f2.8 OS, which scores very highly set to 300mm f2.8 in some reviews, but the TDP sample at that setting looks awful (in the center, the edges are universally bad). Then sometimes lenses that score somewhat poorly on other sites look great on TDP. Again, it looks like there's a lot of variation in both the samples and potentially in the reviewers.