Andrew Gelman and his co-authors build on results from a recent paper to draw some fascinating maps of what the 2004 presidential elections would have looked like if only (1) poor, (2) middle income, or (3) rich people had been able to vote.
These maps show some obvious class differences (as well as illustrating the importance of poor voters to Democratic electoral prospects), but also suggest (as does the paper) some interesting relationships between how people at different income levels vote in rich and poor states.
For poor voters, there is no systematic difference between rich and poor states. But for middle-income and especially for rich voters, there is a very strong pattern of rich states supporting the Democrats and poor states supporting the Republicans.
In short, rich people and poor people who live in poor states have very different voting preferences from each other. Rich people and poor people who live in rich states have much more similar voting preferences. Gelman et al. don’t have any hard and fast explanation for this (they note that race explains about half of this disparity, but only half). However, their results do suggest that some of the conventional wisdom of American journalists on class, voting and geographic location stands in sore need of revisiting.