Sadly (or, in truth, thank God!) comparisons of arrangers by using converted styles is not a real apples to apples comparison.
The sound itself makes the performance, which becomes the style. Change the sound, and the performance no longer plays to it's strengths. So take a style, carefully optimized for one arranger's sound set, and translate it to another arranger, and you will always have a not so level playing field. And as more and more arrangers add proprietary techniques (mega Voices, 'Live' loops, Guitar Modes etc.) the gap only widens.
I feel that the TRUE comparison test of an arranger's sound quality is simply the comparison of the instrument's ROM styles, which one hopes are optimized as much as possible to the arranger's sound set. From this, you should get a good feel for how the styles are written (busy or simple, supportive or dominant, etc.) and how the sounds 'sit' together, and whether the instrument has a 'home' or a 'live' feel to it.
POSSIBLY, one could convert one set of styles to another format, and gain that instruments strengths, but usually at the cost of what makes them so good on the original arranger. I for one have never heard a translated style sound as good as it did on the original, unless you are talking legacy styles from back before Mega voices and multi-velocity drum samples became the norm.
And would one actually REALLY want to do that to a whole set of styles? The styles AND the sounds are pretty much integral to each arranger. I think it is better to compare the arrangers pretty much OOTB. It's how the manufacturers, at least, feel they sound their best...
So cassp's same song, similar style comparison method is VERY informative. Perhaps with a similar choice of RH sounds it could be even closer, but I still feel it is FAR more informative than a translated styles comparison, which always favors the original.
JMO, yada yada yada...
_________________________
An arranger is just a tool. What matters is what you build with it..!