When testreports are based on jpeg, I have noticed that the choice is often explained by jpeg being camera-dependent only, whereas raw also depends on raw converter software, which again comes in dozens of brands, versions, and with a well of possible pp-options. Raw makes it less trivial to compare different cameras, because it can be difficult to know if you are testing cameras or raw-converters.
Not all raw-converters are available for all cameras, and even if you restricted the test to LightRoom 2.6.1 build 632, with a static set of parameters, it wouldn't necessarily be fair to all the different raw-formats out there.
By sticking to jpeg, the tester omits some non-trivial choices. And it's faster, easier, and less work