CC, I think your ruler experiments come out a little funny because the 150mm lens is acting a little shorter than the 100mm. If it had been nikons it would be fine but with the 1.6x crop of canons (not to mention the subtle differences in focal length between different lens models).
The ruler experiments (using macro lenses, shown here) were shot entirely with a single 5D Mk III body. One Canon 100 mm and one Sigma 150 mm, and an off camera flash. Full frame, no crop factor. The first attempt didn’t work as well as it might have because I didn’t actually get the subject to be the same size in both photos. Drawing a subject on the ruler for the second attempt helped me get the size much closer in both shots. Focusing rails would have made life a lot easier. Moving a 5 foot tall tripod back and forth fractions of an inch and having to adjust height to keep the angle the same was a bit of a pain! The least expensive rails at our local stores seem to be CDN$350, and I don’t do enough of the kinds of macro photography to justify getting them. The camera may not have been moved to exactly the correct place, but I think it is close enough to show getting the subject the same size causes DOF to be the same, or at least “close enough for government work”.
In CC’s comparison shots with the painted faces you can also see the bokeh advantage of the full frame camera. It isn’t massive but it does equate to 1 1/3 stops. The bigger sensor will also collect 1 1/3 stops more light leading to better noise performance.
The shots from 10 feet taken with both full frame and crop bodies are very close. If they are displayed at the same subject size, they are quite close. If I had realized I would be putting them up for this discussion, I would have shot the APS-C image with a 30D so the photosite size would be almost the same as for the full frame body. The only difference between the full frame and crop images if all you do is take one body off the lens and put the other body on, as Tony did in his video when he used a 70 to 200 to shoot his wife, should be to have a lot of extra real estate around the edges of the frame. Differences in pixel density and software may throw this off slightly, but not significantly.
Noise performance will probably always be better in a full frame body than a crop body, for the same generation of chips and software, due to the larger sensor typically having more light delivered to each photosite. But, an older full frame body may not deliver noise performance that’s as good as a newer crop body. I still have a 1Ds and I think ISO 3200 is the highest it can be set. My 550D can deliver pretty clean photos at ISO 6,400 if I pay attention to lighting and exposure. At the same time, in bad lighting, my 5D Mk III will deliver about a stop better results than the 550D, I like to keep it below ISO 12,800.