Get started
Work

Predicting the Results of the 2010 Midterm Elections: Judgment, Econometrics, and Prediction Markets

Year: 2010
Type: article
Abstract: In “The Comparative Accuracy of Judgmental and Model Forecasts of American Football Games,” ChiUng Song and colleagues (2007) compared the predictions of two consecutive seasons of American football games on the part of 70 experts and 32 statistical models and found that neither method could beat the betting line. The line is the spread in points by which one team is expected to win. It is set in Las Vegas six to seven days ahead of the game. Here I compare forecasts of the number of House of Representative seats that incumbent Democrats would lose in this year’s midterm elections. The Democrats emerged with a total of 193 seats, a historic 64 fewer seats than they took in 2008. I look at how closely three renowned experts, six statistical models from political scientists, and the Intrade prediction market came to predicting this outcome. The three experts I chose for this study are Charlie Cook, Stuart Rothenberg, and Larry Sabato, all of whom have solid reputations as political analysts and election prognosticators. Indeed, the final forecast of each, issued just a few days before the election, missed the actual count only by between four seats (Rothenberg) and nine seats (Cook and Sabato). Their projections are built from the ground up, so to speak, with a district-by-district analysis that is then supplemented with observations of national trends, such as presidential approval ratings and the generic congressional ballot. For specifics on how I reconstructed their forecasts, see Note 1 at the end.I also examined six forecasts by political scientists to represent the econometric or statistical method. These are shown in Table 1. Half of the models are purely structural, based exclusively on predictor variables such
Source: Foresight
Language: English
Open Access status: green
OpenAlex ID: W2116305574
FWCI:
Cites: 6
Cited by: 1
Related to: 10