I've mentioned statistician Nate Silver before (How Nate Silver Predicted Obama's Win; Nate Silver's Oscar Predictions, Reviewed). He's best known for his statistical work on the 2008 US Presidential election, in which he analyzed various publicly available poll results, handicapped them using baseball-style analysis, and did a surprisingly good job of predicting how specific areas of the US would vote. After the election, Silver gave a TED Talk in which he talked about racism as a factor in the election -- he wanted to know whether racism as a factor affecting voting was statistically predictable -- meaning whether some other factor (like geography) could predict whether a given white voter would not vote for a black candidate. Silver's talk is now up on the TED site, and it's worth a look.
Discussed: the 2008 US Presidential election, the "blueing" of America, what's the matter with Arkansas, "rednecks with guns," asking people if race was an important factor influencing their votes, is racism statistically predictable, the General Social Survey, white people with black neighbors, interracial marriage, street grids versus cul-de-sacs, intercollegiate exchange from New York to the South, predictable problems being designable (solvable).
I don't want to ruin the surprise, but Silver's short answer is: yes, racism is predictable (statistically speaking). It's not about urban versus rural, though -- watch the talk to see what factor predicts racism affecting votes.