Or, I should say, the graphs on “Income, Religion, and Voting” Andrew Gelman posted on 538.com. I like this one:
No matter how churchy, really poor people strongly favor Democrats. No matter how rich, people who don't go to church VERY strongly favor Democrats. Generally, the higher the income, the more church attendance matters to votes, which as Andrew points out suggests that people care more about “social issues” the richer they get. But the relevant “social issues” is very different for religious and non-religious people. But what's going on toward the top of the income scale? At all levels of church attendance, having more money increases the chances of a Republican vote until you hit a relatively high level of income — looks like something near the border of the upper-middle and lower-upper class — at which point additional income starts to make a Republican vote a bit less likely. This is true even for very frequent chruch attenders. Superrich very-frequent churchgoers are evidently pretty damn Republican, but less so than almost-rich very-frequent churchgoers, and that's interesting.
What do you suppose explains that?