OK, [puts on beta-tester gloves]
I tested this during the beta, but came to no conclusion. At the time, revolts weren't working very well anyway. And, defection and independence weren't complete yet. Later, there were more important things to test.
But, if someone wants to get some serious beta testing experience to warm up for HOI or CK, then this is how I would test it:
A) The documented affect is that rebellion delay can be changed from -6 months to +6 months. What the heck is that?
B) One possibility is that the onset of increased revolt risk from war exhaustion is modified. I forget the exact details, but I think war exhaustion starts to occur after 2 years and increases every year (by 1% in core provinces and 2% in non-core provinces). So:
TEST - Play England in 1419. Do not bother moving units and such, you will do quite well, anyway

D). I used England a lot in beta testing for this reason.
* Use the robespierre cheat and set all DP sliders to the middle position. Set the Decentralization slider to the far left for one test, and to the far right for another test. Note, I always neutralize other DP sliders when testing to ensure only one slider affect is under test.
* You are already at war. Make peace with everyone but France to get rid of the irritating peace offers. Don't worry about losing battles, provinces, etc. These are not at issue in the test.
* Monitor a province or two for revolt risk every month. Note when an increase occurs. Keep testing until it is obvious that the revolt risk modifier does not work this way, or until you see that it does work this way, and the modification pattern is known. That is, the revolt risk modifier might only change when the initial war exhaustion occurs, or it might modify every subsequent change in war exhaustion, or both.
* If the modifer works this way, retest with a couple of other slider positions to ensure that the incremental modifications work correctly. Strictly speaking, you should test every slider position. I did that for every DP slider for a first test during the beta, but only did spot checks after that (it takes too damn long!).
C) Ok, it's not B. Perhaps the revolt modifier affects when a province declares independence or defects? The documented rules on this (as stated by God - Johan) is that an independence check is performed after 1 year, and a defection check is performed after 5 years. I have seen variable defection frequencies and other people have mentioned this as well, so this may in fact be what the slider modifications affect.
TEST - Play England in 1419. As before do not do anything.
* Use the robespierre cheat and set all DP sliders to the middle position. Set the Decentralization slider to the far left for one test, and to the far right for another test.
* You are already at war. Make peace with everyone but France to get rid of the irritating peace offers. Don't worry about losing battles, provinces, etc. But, take some French provinces to increase the test cases, particularly for independence.
* Eventually you will start getting revolts. Note when they occur. Note when an independence or defection occurs. See if the pattern of an independence is now occurring after 1/2 year and 4 1/2 years for one DP extreme and 1 1/2 years and 5 1/2 years for the other DP extreme. As an aside, your government will fall in about 20 years, so testing time is limited. Again, if it is working this way restest with a couple of other slider positions to verify the incremental modification.
D) Damn, it wasn't C either. One last possibility. Once revolt risk for a province is positive, revolt checks are presumed to occur every month. Perhaps the modification "shifts" when this revolt check occurs somehow or sommink... I can't describe it because I don't know what I'm talking about.
TEST - who knows? Perhaps a detailed tally of all revolt occurances over several hands off games where some specific countries have their DP slider set to the extreme position for the first set of tests, and the other position for the second set of tests. Then having run all these tests you do some slick statistical analyses on the results.
E) So it wasn't B or C, and even if you figured out a way of doing D you didn't do it because you are not crazy.
ACTION - write a bug report.
Welcome to the world of hard-core beta testing. I'm kind of interested in the results if someone would do the testing...