Erikson and Wlezien have amassed data from close to two thousand national polls covering every presidential election from 1952 to 2008, allowing them to see how outcomes take shape over the course of an election year. Polls from the beginning of the year, they show, have virtually no predictive power. By mid-April, when the candidates have been identified and matched in pollsters’ trial heats, preferences have come into focus—and predicted the winner in eleven of the fifteen elections. But a similar process of forming favorites takes place in the last six months, during which voters’ intentions change only gradually, with particular events—including presidential debates—rarely resulting in dramatic change.
Great work as always Joe! As I said in the Obama vs. Romney thread the wheels have fallen off the Romney campaign and even many right wing pundits are saying so!
