Using state polls to forecast presidential election outcomes in the American states

This paper presents a simple forecasting model for state-level presidential outcomes, based on statewide preference polls and a lagged vote variable. The analysis illustrates two important points. First, the cadidate who is leading in a state in September usually goes on to win that state in the November election. Second, the combination of pre-election preference polls and a lagged dependent variable generates highly accurate estimates of presidential election outcomes in the states. The limits of using statewide preference polls are also discussed.