Polling the Pollster Pollers: Obama Still Strong

Update: If you're looking for conservative polling, click here.

I've been meaning to put together a post for a while now on the metanalysis sites: the top-feeders one level above pollster.com, realclearpolitics, electoral-vote and the like (who aggregate and average recent polls)–outfits that run statistical crunchers on the polls, and spit out what everyone wants to know:

What are Obama/McCain's odds of winning?

These three sites use similar methods, but as you can see–especially in these post-convention/Palin times–their conclusions diverge.

fivethirtyeight.com. Current odds of an Obama win: 52.1%. These guys actually try to project what will happen in November. (Not easy: the odds here a week ago were 70%.) Founder Nate Silver is also an excellent blogger.

Princeton Election Consortium. They seem to have been discombobulated by the convention/Palin bounces, so I'm not quite sure what their probability is today. (There's a post suggesting that the electoral-vote estimator is not their current prediction.) It was nearly 100% a week ago.

presidentforecast.andreamoro.net. ShHe's got the Obama odds at 85%, and includes some interesting conditional probabilities:

But if Obama loses …
… Colorado, Florida and Virginia 47.7%
… Florida, Ohio and Virginia 65.3%
… Colorado, Florida, and Ohio 54.0%
… Colorado, Florida, Ohio and Virginia 47.5

Less statistically sophisticated, but more graphically sophisticated, check out this cool chart that my friend Mike just put together. (This is a live link, so it will look different from what I'm looking at as I write this; IOW, it may make this post's headline ridiculous.)

It's based on the state-by-state poll numbers from electoral-vote.com. Click for an expanded version on Mike's site.

What it says about today: Obama would win by just taking those states where he's currently polling ahead.

McCain would have to take all his strong, weak, and barely states, plus all the ties, plus all the states where Obama is leading by less than 5%.

This supports what I've been guessing, and what Nate Silver said today about the latest polls:

The theme here is simply traditionally red states coming home to John McCain in a big way

McCain's national polling numbers are up, but if it's just energized Republicans deciding to answer the phone–and that in states that McCain's already winning–it doesn't translate into many electoral votes.

Also, note how the number of "weak McCain" states has been growing over the last month, while the "weak Obama" states have been shrinking. (Until the last day or two.)





3 responses to “Polling the Pollster Pollers: Obama Still Strong”

  1. Mike Avatar

    Hi Steve, Thanks for the link. Here’s one interesting bit of polling data… For all the talk of McCain’s post-convention “surge” (the way the media describes it, it’s much more than a bounce!) McCain’s only gained 7 EVs in the last week. Even more telling, he’s actuall 31 EVs below where he was on Aug. 21, just a few days before the Democratic convention started. Since that same date, Obama has actually gained 17 EVs.

    Some surge, huh? Kind of reminds me of this other surge…

  2. Andrea Moro Avatar

    BTW, I’m a guy. Thanks for linking

  3. Steve Roth Avatar

    Andrea: “BTW, I’m a guy.”
    Oops. Sorry. One of those elitist european-type names.