Here Craig Depken plots the relationship between state unemployment rates and tradesports prediction prices. He finds a weak negative correlation (-0.09) between unemployment and price, implying that states with above average unemployment have a slight tendence toward Kerry.

So what else might be driving state-by-state voting patterns? An obvious possibility is social or religious issues. Hence I obtained data on the share of each state's population that attends religious services regularly and calculated the correlation with the tradesports prices. The result, 0.14, indicates that states with above average church attendance lean toward Bush.

One concern with this calculation is the religious attendance data. I've used data from the same source for an academic paper and have some concerns that the data are a bit noisy. (For example, the dataset shows church attendance rates greater than 100% for three Alabama counties used in my previous study.) A second issue is the likelihood that using overall church attendance rates obscures important differences across denominations. (I suspect that Southern Baptists are more important to the Bush vote than are Unitarians.)

A third concern is with the use of tradesports prices. On a hunch that the tendency of prices to tend toward the extremes (0 for states like MA that Kerry will carry and 100 for states like TX that Bush will carry), I obtained polling data from electoral-vote.com and recalculated the correlation coefficient. The resulting correlation of 0.20 between church attendance rates and Bush's standing in the polls is indeed larger thereby confirming my suspicion. Still, I would have expected a larger correlation and suspect that the religion data issues discussed in the previous paragraph are salient.