Forecasting Models and the Presidential Vote

Kenneth Wink
Content Type
Journal Article
Political Science Quarterly
Issue Number
Publication Date
Spring 2019
Academy of Political Science
FOR DECADES NOW, political scientists, journalists, campaign managers, and pundits have sought to predict the outcomes of elections well in advance of the day the votes are cast. As the ultimate office in the U.S. political system, the presidency has been the focus of much of this activity. Since the late twentieth century, a number of prognosticators in the discipline of political science have used forecasting models to predict presidential elections. These models have become visible in pre‐ and post‐ election coverage, and a sort of competition has emerged to produce the most accurate model whose variables offer the greatest amount of forecasting lead time before the election. Once considered “recreational political science,” forecasting presidential elections has become a cottage industry. Furthermore, the attention paid to the accuracy of these models has led to better explanations of election outcomes and allowed interested persons to see patterns in elections that are stable from year to year and to identify outlier elections and the factors that led to unique election outcomes.
International Affairs
Political Geography
Global Focus