Princeton Election Consortium

A first draft of electoral history. Since 2004

A welcome to new readers – and a caution

September 9th, 2008, 9:51am by Sam Wang


Welcome! One-fourth of this site’s traffic has come in the last two days. On Sunday, someone made the 100,000th page view. This post is for new (and returning) readers who need to be brought up to speed.

Also, because of rapidly changing events, the true current snapshot is not what’s listed above. Instead, it’s McCain 300 EV, Obama 238 EV. Why? Read on.

This site takes the most recent available state polls and does a meta-analysis to generate a precise snapshot of where the Electoral College stands (see the Methods and FAQ). This snapshot analysis made a correct prediction on Election Eve 2004, and this year is automated. There are many features here – take a look around. We will also feature election-related commentary by Princeton faculty on topics ranging from voting technology to how false beliefs are formed.

However, right now there’s a problem with state polls: most of them are weeks old. I have implemented a temporary fix to get us through the bounce, which is transient but still of interest.

Usually, state polls give a very accurate estimate of state outcomes, and therefore of the Electoral College. This week is unusual for three reasons:

  • State polling came to a halt for several weeks during the conventions.
  • There have been several opinion-shifting events: the Democratic convention, especially Hillary Clinton’s speech, and the Republican convention, especially Sarah Palin’s speech.
  • Recent national polling shows bounces in both directions, leading to a net gain for McCain.

So I am offering an adjustment to the EV estimate based on the recent swing in national polls. It’s described here, and the most recent update is here. This adjustment gives the best snapshot of what would happen in an election held today. The current national-poll margin is McCain ahead by 2.0 +/- 1.2 %. A commensurate adjustment to the state-polls-only EV estimate given above gives a reversal of the result to McCain 300 EV, Obama 238 EV.

I will update this national poll-adjusted figure for the coming week. After that, the bounce transient may settle out. We will see whether any changes over the last few weeks are lasting. To see examples of how bounces have subsided in the past, see this excellent post by Mark Blumenthal. Hint for reading the graphs: you’ll get a better feeling by looking at the data points, not the curve fit.

Tags: 2008 Election · Site News

No Comments so far ↓

Like gas stations in rural Texas after 10 pm, comments are closed.