Statewide Trial‐Heat Polls and the 2000 Presidential Election: A Forecast Model*

Objective. We provide an examination and update of a presidential election forecasting model that we have previously developed to predict state‐level presidential election outcomes. Method. Our model consists of September statewide trial‐heat polls in 1992, 1996, and 2000 along with a prior vote variable. We use this model to generate predictions for both state‐level and national‐level outcomes. Results. Although our model generated reasonably accurate point estimates of state‐level outcomes in 2000, it still incorrectly predicted a Gore victory in 2000. Conclusions. We discuss possible explanations for the 2000 misprediction and present updated coefficients to be used to generate a forecast for the 2004 election.