A lively discussion in the comments of yesterday’s Snapshot update, along with the release of a handful of new public polls today, warrants a pre-Labor Day update to the Snapshot.
Today, I am only showing numbers with Rasmussen polling included (I think it reads easier that way, plus the snapshot with Rasmussen polls is the "official" projection). Also, Charlie Crist’s chances of victory are now evenly distributed between Democrats and Republicans, rather than calculated on their own. Here is the update:
Senate Snapshot, September 3rd
I do not publish polling averages for any campaign that is further out than 18.5%. I would prefer not to publish any polling averages for campaigns that are further out than 11.0%, given that 10.56% is the largest error ever recorded in the 145 general election, statewide polling averages I looked at to produce this methodology. (The 10.56% instance was the 2006 Idaho Governor’s campaign.) The fewer campaigns displayed, the easier the chart is to understand. Showing real blowouts, such as Indiana, Iowa, and Oregon, just increases the signal to noise ratio by prominently displaying blowout campaigns. They are displayed here to see if people agree with that assessment.
Other commenters were upset that candidates trailing by only 1% or 2% in the polling averages were projected with small chances of victory, such as 36% or 22%. There are two reasons to justify those low percentages. First, because the win percentages are based on if the election were held today. Second, it is because simple poll averaging is actually very accurate. For all the distrust in polling, for all the worries about machine fraud, for all the complaints about how certain polls do not include a large enough sample of one demographic or other, and for all the complaints about certain likely voter models, polling in general election, high profile, statewide campaigns has proven to be a very accurate predictor of final results. The median error for the methodology I use is a minuscule 1.76 from final predicted polling margin to final vote results.
Finally, in response to those commenters who object that certain candidates being ahead even though they suck, such as Dino Rossi and Carly Fiorina, well, I don’t know what to tell you. Polling is the only factors I include in the snapshot. Qualitative factors are not considered. We should all be very worried about Senators such as Patty Murray, Barbara Boxer, and Russ Feingold.
Please, keep your questions and comments coming. Your feedback has dramatically improved the Senate Snapshot.
Notes
--This is a snapshot, not a forecast. All of the odds presented here are based on if the election were held today. It is not a prediction of future trends.
--If a campaign isn't listed here, then it is not currently as close as any of the campaigns listed here.
--A * indicates that the candidate has a primary challenger, but is the favorite to win the nomination.
--All polls used in the averages are taken from Pollster.com.
--A complete description of the methodology behind this forecast, along with all the research and a FAQ, can be found here.