My latest politics column for the Boston Globe, on the strengths and limits of using demographics to predict elections is here. An excerpt:
IF YOU spend enough time studying election returns, you can start to see politics as just another branch of science — like quantum physics, but without the humanity. Pundits who prefer to psychoanalyze voters are apt to depict campaigns as wild affairs full of plot twists, but the geography of American politics is surprisingly resistant to change, at least in the short term.
Thomas Frank can ask "What's the Matter with Kansas?" because it votes Republican, but he can't accuse that state's voters of being fickle or impulsive. Kansas has been one of the country's most Republican states for more than 90 years.
It's not just a matter of knowing history; demographic factors such as race and religion can be mercilessly accurate in predicting how a place will vote in a presidential election. Still, there are the occasional aberrations, as well as places where contradictory rules of demography collide, and they're what really make politics interesting. For example, it's a good rule of thumb that the more urban a place is, the more liberal it is. Manhattan votes Democratic, and Mayberry votes Republican. Except that the most rural state in America — not measured by population density but by the more useful metric of what percentage of the population actually lives in small towns — is Vermont, which is solidly Democratic. And Utah, which is actually more urban, if also more religious, than New York state (there are lots of acres but very few people outside the Salt Lake City area) is the nation's most Republican state.