Princeton Election Consortium

Thanks to returning and new readers Washington Monthly, The Economist, Angry Bear, Chimpanzee Tea Party, and Matt McIrvin. McIrvin’s capsule history of the 2004 Meta-Analysis is particularly good.

A reader James asked about the extreme values that the “Voter Influence” measure was taking on. His comment and my reply are here. This quantity (seen in the right sidebar) expresses the marginal influence that individual voters have on the final outcome probability. Per his feedback, I’ve modifed it… [Read more →]

The comments on the last thread were quite instructive, and led me to look over Silver’s methods documentation in detail. Wow, that’s quite a complex procedure he has. I should probably address your questions about it before commencing with further description of the Meta-Analysis (which is not a prediction). [Read more →]

Today I’d like to outlline the basic contrasts between this calculation and a popular resource, FiveThirtyEight.com. That site, run by Nate Silver, a sabermetrician, is a good compendium of information and commentary. However, both our goals and methods differ on several key points. The biggest difference is that this site provides a current snapshot of where polls are today, while he attempts a prediction. His approach also has a conceptual problem…

[Read more →]

Check the right sidebar for some cool interactive maps. The state-by-state probabilities are in this file; the first number on each line gives the percentage probability that Obama is ahead. The bars indicate electoral votes for which win probabilities are currently greater than 95% for either candidate.

The question arises: how good is the Meta-Analysis in comparison with a more conventional measure such as an average of multiple national polls? Answer: Today, the Meta-Analysis is over four times as accurate as an average of recent national polls. Furthermore, it gives an estimate in the units that matter - electoral votes. Here’s why.

Averaging is an excellent way to see past variation in individual data points. A standard measure of how well you know the true average is the standard error of the mean (SEM), which can be thought of as a multiple-poll version of the famous Margin of Error. Today, the SEM of the last 6 polls is 2.3%, a fairly typical value.

But the Meta-Analysis uses dozens of polls at any given moment from states in contention, and over 100 polls in all… [Read more →]

For those of you who followed my analysis in 2004, welcome back. Just as I did then, I’ll be providing meta-analysis of polling in the 2008 Presidential race. My central goal is to reduce hundreds of state-level polls to simple statistics that will show you the state of play. The methods will be transparent, and with the help of Andrew Ferguson, automated and more visually accessible.

We’ll also do more than just report on polls. The reason is simple: compared with 2004, this area has blossomed (here are several great resources). [Read more →]

After a hiatus to do real work, I am back. This site really needs to be reorganized, and perhaps written up as a proper print article. In the meantime, I promised to describe the pre-election Fox News interview… [Read more →]

Election results can be displayed in many ways. The cartograms used on this site are a way of displaying electoral votes accurately. To see displays done by population or on a county-by-county basis, see these interesting maps.

Several interesting and good articles about exit polls. Here is a piece in the Washington Post about demystification of exit polls. Also, in this week’s New Yorker is a great article by Louis Menand on a meeting that took place at Stanford at which pollsters discussed the interpretation of this year’s results. Essentially, they think the values talk is a misinterpretation of a bad question. The article argues that it was essentially terrorism (i.e. 9/11) that swung it for Bush. The article is print only - go get it. The December 6 issue.