Via Ezra Klein via Andrew Gelman, check this out:
Check the R2 on this mutha.
This is something that poli sci is good for. The chart doesn’t say, exactly, that campaigns don’t matter–why did Eisenhower thwomp Stevenson so badly in 1952, for example? (This model is also unusual in using only economic growth to predict outcomes; most predictive models also include war/peace, and presidential approval, and if we were using one of those, Gore’s loss in 2000 would be the outlier.) What it does say is that it’s damn hard to escape entirely the broader historical phenomena that define the contours of electoral politics at any given moment. Obama currently is probably ahead by 4-8%, and he probably still will be on election day.
By the way, despite what some say, 4-8% ain’t nothing to sneeze at. Except around the time of the Swift Boats/GOP convention in 2004, neither candidate ever had a lead any bigger than that. Dukakis lost by about 8% in 1988, and is remembered as something approaching the platonic ideal of “loser.”
