Here's my theory regarding the flips:
If the city has been captured by the opposing armies, the most powerful factor regarding the flip probability is the culture built up in the city by the previous ruler. You can mitigate this most effectively by building your own culture quickly in the city after capture. This is done best by rush building a library then temple after you quell resistors. I don't think that having a garrison makes a difference in the %chance on that turn just for being there, but more for the fact that they get rid of resistors faster (cities with resistors have a greater %chance of flipping) and that the roll to find out how many resistors are quelled can use a bad roll that would have resulted in a flip. The other factors such as location of capital, total culture comparision, government type, and happiness level may play a factor as well, but they are small in comparison.
I've never had a city with low culture flip back after I captured it. However, I've had cities with high culture flip back to the previous ruler even though I've had all of the other known factors in my favor.
A good test that might be useful is to set the resist % for democracy to 100% in the editor (I don't use the editor much, but I know there are options based on government type) and then play a game where you take a high culture city from an opponent. See if you can coax a switch by leaving resistors and keeping the city unhappy.
Any thoughts on this?
If the city has been captured by the opposing armies, the most powerful factor regarding the flip probability is the culture built up in the city by the previous ruler. You can mitigate this most effectively by building your own culture quickly in the city after capture. This is done best by rush building a library then temple after you quell resistors. I don't think that having a garrison makes a difference in the %chance on that turn just for being there, but more for the fact that they get rid of resistors faster (cities with resistors have a greater %chance of flipping) and that the roll to find out how many resistors are quelled can use a bad roll that would have resulted in a flip. The other factors such as location of capital, total culture comparision, government type, and happiness level may play a factor as well, but they are small in comparison.
I've never had a city with low culture flip back after I captured it. However, I've had cities with high culture flip back to the previous ruler even though I've had all of the other known factors in my favor.
A good test that might be useful is to set the resist % for democracy to 100% in the editor (I don't use the editor much, but I know there are options based on government type) and then play a game where you take a high culture city from an opponent. See if you can coax a switch by leaving resistors and keeping the city unhappy.
Any thoughts on this?