Well, in all honesty, I wouldn't have done this test to begin with, since I don't really follow the reasoning behind mixing two developers without a clear theory of what's to be expected. I probably would have just tested one developer next to another, and maybe two dilutions for both developers if there's any (even anecdotal) evidence that this would to 'something'. Mind you, I think there's nothing wrong with mixing developers per se, but IMO it's more part of the darkroom discipline of "let's wing it and do a dash of this and a squirt of that and see what we get". In other words, a very informal approach to things, and an approach in which systematic testing is anathema. But that's my personal view, only.
Having said that, for this kind of test, I would have probably set up a scene with a fairly high brightness range (SBR) - something like an interior shot looking outside into the yard or something. Preferably some even surfaces both in the shadow and the highlight areas to judge grain if that's relevant. Do the test on a day with constant lighting conditions, of course. Meter the scene and take some notes so you know where everything is placed, approximately, and what the SBR is quantitatively. Stop down lots, but stay out of reciprocity failure territory of the film. And then bracket for instance a -1.5, +/-0 and a +1.5, and repeat those three shots across an entire roll. That way, there's at least consistency in the exposure and the scene.
Shooting a grey card 10 times or so at different exposures accomplishes something similar, but takes more film in the end. Alternatively, expose the film not in a camera, but e.g. under the enlarger, using a Stouffer step wedge - but I admit I like 'real world testing' (it's a little less soul-sucking), so I'd go with the high contrast scene instead.
Divide the film into test strips and then develop those. As to analyzing, I'd probably slap them onto the platen of a flatbed scanner first, preferably with a step wedge along with it, and scan all tests in one go. This ensures exactly the same exposure and contrast for all scans. Alternatively, I'd contact print them. And then just eyeball in first instance, and see what stands out. Look at how much shadow detail is recorded, to what extent highlights are being compressed (or not), ogle the grain a bit at full magnification. If that gives any clear clues, try to formulate hypothesis and then see if they hold true. For instance, if I get the impression that one sample has more distinct grain than another, I'm going to put them side by side and compare the grain at different density levels, but also critically examine if the contrast in both samples is really the same - if it isn't, I'm probably just fooling myself anyway.
So it's basically a process of iteratively formulating expectations and then trying to falsify those, setting up little experiments (preferably as efficiently as possible) to gather the necessary data. It's in part fairly uncreative and tedious, systematic work, in part intuitive observation.