It is very clear that a camera which incorporates a sensor with about 4 Micron pixel pitch reveals more noise than one with a 6 Micron pitch - to compare the performance of the firmware and electronics of the camera one must compare comparable pixel dimensions.
Second - every test is a subjective temporary verdict about one or few items - and not a general statement about anything else than: What I have seen at that time with that specific item applying specific testing methods which I have selected for that purpose - which means you can test something and another tester does the same test and gets different results. That's why you should never rely on any singular test result - read as many as you can find and then draw your own conclusion. I am writing test reports myself and use DXO Analyzer as well as the IE Analyzer and to top it also the system of Norman Koren - all three come usually up with some differences in performance. My approach is then to compare those results and come to a conclusion by integrating all results as good as possible.