Since Internet Explorer 8 will have two factors influencing it's choice of rendering mode (from a selection of three), I was curious as to how they interact with each other. Then a comment on some of Shelly Powers' commentary on the subject gave me enough momentum to throw together some tests and see for myself.
I've documented my procedure and the results on my main site, in an article titled "Internet Explorer 8 Doctype and HTTP Header Switching". You can also find the test cases there.
Please let me know if you spot any flaws in the tests or my reasoning.