Polly crows: For the second time in a row, Polly called a presidential election with a perfect or near-perfect forecast.
Election decided on Issues, not Race: So much for the Bradley Effect
With Barack Obama as the first black candidate running for president, a lot of attention was given to the question how much race would matter in this year's election. Drawing on examples from earlier elections, political pundits discussed the so-called Bradley effect. According to this theory, black candidates might be overestimated in the polls. When polled, people would state that they would vote for a black candidate out of political correctness reasons - but would then vote for the white candidate. However, since the mid-1990s, political scientists could not observe this bias anymore. In a recent study, Daniel J. Hopkins from Harvard University concluded that the Bradley effect has disappeared. He explained the overestimation in the polls by the well-known effect that polls typically overstate support for front-runners.
Now, seeing the election results, the pundits have been proven wrong. It was not the candidates' skin color that mattered to voters. Our research shows that voters select the candidate they expected to do better in solving the country's problems - and they did so in this election.
In our research, we predict election outcomes based on which candidate the voters expect to do better in handling the issues facing the country. In analyzing polls U.S. Presidential Elections from 1972 to 2004, we found that voters chose the candidate they expected to do the best job in dealing with the issues facing the country on 7 of the 9 elections.
For this year, starting with the first forecast in March 2008, the PollyIssues model consistently predicted Senator Obama as the next U.S. president. For example, Obama was expected to do a better job on economic issues. In the latest update of the model prior to the election, Obama received 52.5% of the popular "two-party vote." This almost hit the bull's eye. For the last four elections since 1996, on average, PollyIssues missed the popular two-party vote shares by less than one percentage point.
In electing Barack Obama as the new president, America has overcome questions about racism ruling the election. Again, the voters have shown that it's the issues that matter when selecting their president.
The PollyIssues paper can be found at http://pollyvote.com.
A special panel session at ISF 2003 in Merida, Mexico (June 18) will provide important evidence that supports Principle 5.7. The session title is: "Damping Works for Seasonality Estimates." Don Miller and Dan Williams will summarize their findings and commentary will be provided by David Findley, U.S. Census Bureau; Rob Hyndman, Monash University; Anne Koehler, Miami University; Keith Ord, Georgetown University; and Benoit Quenneville, Statistics Canada. The session will emphasize steps needed to foster implementation of this important innovation.
We also use this opportunity to announce a new section of forecastingprinciples.com called "Working Papers with Evidence on Principles. " It allows researchers to establish a claim and to gain feedback. You are invited to participate. The Miller/Williams paper is the first contribution to this section.
New Special Interest Group Added: Health Forecasting - Of general interest