Polly's Archives

1. Polly's Page

  • Polly explains the 2004 Pollyvote.
  • Polly also provides PDF versions of the monthly Table 1 March 8 to November 2, 2004.
  • Also available is the final Table 2, including all poll data, by date and source, going into the Pollyvote.
  • In addition, Polly provides 6 different Pollygraphs, charting the changes in the Pollyvote over time.

2. Polly's Data:

Polly's 2004 Presidential Election Data Set (a Microsoft Office Excel 2003 spreadsheet)

3. 2006 Elections Archive:

Will the Republicans retain their majority in the U.S. House of Representatives this year?Read more . . .

Commentary on Election Forecasting

Panels on Presidential Election Forecasting at the ISF 2005 International Symposium

Three panels on presidential election forecasting are included in the program of the 2005 International Society of Forecasters “International Symposium on Forecasting,” which will be held June 12-15 in San Antonio. One is a Roundtable on Ray Fair’s Presidential Vote Equation, where Prof. Fair will respond to critiques of his model. Professors James E. Campbell, Richard Gleisner, and Joe Stone are among the presenters. For summaries of the presentations, click here. At two other sessions, several forecasting models that did relatively well in 2004 will be displayed, along with other papers. For papers by Professors Campbell, Erikson and Wlezien, Jones, Lichtman, Norpoth, and others, click here.

Immediate Post-election Plans

In the immediate future, we will be analyzing our data and writing up the findings for presentation at academic conferences. Among other things, we plan to do the following.

  1. Compare the accuracy of the several components going into the Pollyvote across the forecast horizon, assessing, in particular, the relative performance of the presidential elections forecasting models in 2004.
  2. Carry on a discussion with members of our Panel of Experts relative to the use of the Delphi method for forecasting elections. To the best of our knowledge, we were the first to do so. We wish to do a post-mortem on the project, looking for ways to improve it and obtain more information with it in future elections.
  3. Calculate the Pollyvote retroactively in previous presidential elections, comparing the results with those of 2004 and addressing, among others, the following questions: To what extent did the Pollyvote improve accuracy in predicting the outcome of earlier elections relative to the polls, the IEM, and the quantitative models? Which component methods have been more accurate at various horizons? Might the differential accuracy justify a re-calibration of the Pollyvote weights?

Other Initiatives

We invite scholars doing election forecasting for other races in the United States (be they state governorships, mid-term congressional elections, or local contests), as well as those doing work in other countries, to share with our readers their models, data, and findings. Also, we encourage one and all to experiment with the Pollyvote to see if it yields better predictions than reliance on a single method. Additionally, we will update the bibliography and the links page periodically as new papers, publications, and sites come to our attention.