One of the most intriguing findings in Red State, Blue State (one that is echoed in my own work) is the how income and religion interact in vote choice and partisanship. Many commentators (think Thomas Frank) have assumed that the rise of a religious cleavage in American politics has somehow suppressed the role of class and income. But the findings in Red State, Blue State and Polarized America show that not only is this not true, but that religion may actually exacerbate the role of income in party identification and vote choice.

Below I have reproduced a figure from Polarized America which shows the Republican identification for white "born again" or "evangelicals" compared to other whites at a number of income levels. Note that all the data used in this figure are whites so there is no direct role of race. Although as Eric points out, there may be an indirect one.

The basic story, confirmed by Andrew and his colleagues, is that low-income conservative Christians are a slightly more Republican than other low-income whites but high-income conservative Christians are a LOT more Republican than other high-income earners. So the role of income in shaping partisanship is greater for this religious group that whites as a whole.

Some commentators (think Kevin Phillips) have noticed this phenomenon and have attributed it to the "prosperity gospel" preached in many evangelical churches. But before we grab on to such a U.S.-centric explanation, it is worth noting that John Huber and Piero Stanig have compiled data showing that poor religious voters in Europe are more likely than low-income secular voters to support parties promoting economically liberal policies (in the European sense of the term). So any explanation of the phenomenon ought to have a passport.

Recently there has been a lot of theorizing among political scientists and economists as to why religious belief or practice might systematically lead to greater support for the economic agenda of the Right.

In an important study, Ken Scheve and David Stasavage show that religiosity is associated with lower levels of social spending on a cross-national basis and that religious voters are consistently less like to support social spending than the non-religious across advanced democracies. Their explanation for these findings holds that religiosity directly affects preferences for social insurance. Drawing on literature in psychology, Scheve and Stasavage argue that religiosity reduces the "psychic costs" that the experience of economic shocks such as unemployment or illness entails. Because religious belief serves as a substitute to social insurance, it reduces the demand for social spending.

Similarly, Roland Benabou and Jean Tirole argue that religious beliefs tying rewards in the after-life to industriousness on Earth induce lower support for redistribution. In their model, religious voters oppose redistribution because non-believers also benefit and because it dulls their own incentives to work hard.

Religious adherence may not only have a psychological or normative affect on redistributive preferences. Huber and Stanig argue that religious organizations provide direct material substitutes for state provided redistribution and social insurance.