Princeton Election Consortium

Innovations in democracy since 2004

Outcome: Biden 306 EV (D+1.2% from toss-up), Senate 50 D (D+1.0%)
Nov 3 polls: Biden 342 EV (D+5.3%), Senate 50-55 D (D+3.9%), House control D+4.6%
Moneyball states: President AZ NE-2 NV, Senate MT ME AK, Legislatures KS TX NC

State polls – fresher than you think

October 21st, 2008, 2:24pm by Sam Wang

As you can see, the Meta-Analysis has taken a downtick. This suggests that the race has leveled off – or may drift toward parity a bit. Notably, state polls now lag national polls by only a few days, making them an excellent source of current information about the race.

The state polls used by the Meta-Analysis for Florida, Missouri, North Carolina, Ohio, and Pennsylvania were all conducted in the last week (see our data source at Indeed, our rule of “last three polls and all polls in the last seven days of polling” is well justified by the proliferation of polls: 6 in Ohio and 8 in Florida!

Considering that national polls have an error of at least one percentage point, at this point you are better off following the Meta-Analysis, whose effective error is much smaller. The tradeoff between speed (national) and accuracy (Meta-Analysis) is gone. At this site you now get both.

At this point I am not expecting the outcome to move more than about 15 EV from the current median. What I’m really watching for is some fresh Senate polling data, especially in Mississippi and Kentucky.

My posting will be fairly light for a day or so. I have a few pressing projects: giving a midterm exam, hosting a Supreme Court Justice, some pressing data analysis, and writing on the neuroscience of being undecided. But I’m still around, and the Meta-Analysis will continue to be updated.

Tags: 2008 Election · Princeton

23 Comments so far ↓

  • Evans

    I think light posting can be tolerated for hosting a supreme court justice :)

    Thanks for your continued insights!

  • Ginny Mayer

    Good luck with your midterms.:)

  • Bill B

    Thank you for your important work… on this site and elsewhere.

  • Michael S

    Be good to Justice Ginsburg.

  • Paul

    Have fun! Ah, midterms….

    To ponder when you get back: Why on earth is McCain still pushing hard in PA? Does he know something we don’t?

  • Michael

    McCain is pushing hard in PA because he has no choice. Kerry states + CO, IA, & NM = 273. McCain can’t win CO, IA or NM, so he absolutely has to flip at least one Kerry state to have any plausible path to 270. It doesn’t make much sense to push hard for any Kerry state other than PA. Also, CO + IA + NM = 21 EV = PA. One uphill battle is cheaper and easier to wage than three.

  • Phil

    Are you giving midterms, or taking them?

    (PU ’70)

  • Paul

    Michael: The most coherent explanation I’ve heard. Hadn’t realized that CO was less close to tipping than PA. Thanks.

  • Pete D

    Personally, I wonder how much, if any, Colin Powell’s endorsement of Obama will impact any of the elections (e.g., flipping MS, KY and/or adding to Obama’s EV total). Of course, Powell’s remarks are not the only bit of anti-GOP sentiment being spouted these days. Some have said we’ve reached the end of what’s mapped out in the book “Nixonland” (a book I recommend). At least one columnist referred to the GOP as being in a circular firing squad. Myself, for the first time ever, am proud to let other Republicans know that I’m of the New England establishment. Anyone to the right of Nelson Rockefeller has NEVER been okay in my book.

    With that said, there’s only 2 weeks left. The book “Nixonland” raises the possibility that the rest of the country became more “Southern” around the time of Watts (August, 1965). Is it possible that the South is becoming more “Northern”, now that we’ve arrived at 2008?

    Well, keep an eye on those races in Dixie, Sam, and once again, thanks for your good work!

  • Bruce W

    I’m not sure why anyone is worrying about what McCain does anymore–he knows he’s going to lose, and his campaign is just putting on a show. They had a bit of a bounce the past few days, but it’s over and he’ll fall off again until a few days before the election. He can’t win in PA, and it now looks like MT, IN, and GA could be in play. This is going to be epic.

    The Republicans–a party among which I used to number myself–have completely lost perspective. They need to be crushed in this election so someone sane and realistic and progressive and decent can pick up the pieces and move the party into the 21st century. In the meantime, Obama is someone who will actually lead the country forward, and is a good, thoughtful, intelligent, and passionate American. A *real* patriot: someone who wants the best for this country and its people.

  • Matt McIrvin

    One uphill battle is cheaper and easier to wage than three.

    And, yet, there’s a huge opportunity cost: if he pushes hard in PA, that’s money and effort that he can’t spend on holding nearly-tied Bush states like Ohio or Florida or even North Carolina.

    McCain’s in really bad shape.

  • Michael

    if he pushes hard in PA, that’s money and effort that he can’t spend on holding nearly-tied Bush states like Ohio or Florida or even North Carolina.

    True enough, but remember that they’re cutting back in New Hampshire, Wisconsin, Colorado, Maine and Minnesota. Those are the resources that will be going into the push in PA and holding the vulnerable Bush states. They’re also a message issue — playing to fears of socialism (and maybe another ism or two) actually could work in PA, OH, FL, NC and VA, but it likely hurts McCain in places like CO, MN and WI.

  • Phil

    A downtick in the median doesn’t bother me too much when I see the lower limit of the 95% confidence range remains unchanged.

  • Michael S

    Hans, the y-axis scale is different in the two histograms. The scale difference enhances the spike for the most likely scenario.

    As mentioned earlier (by me, cough!) the graph is leaning towards a single spike as we get towards election day.

  • emil ruebe

    Sam, what about postponing the days of “light posting” until after Nov4? I do not think we can afford to miss your analysis before that date.
    Just tell the supreme court justice you get back to her later on ;)
    Midterm exams? Nobody will miss them, will they?

  • Mark

    Nope, thumbnail is still different from big plot.

    Currently we have 4 big spikes, which I assume represent Obama winning both OH and FL, one of the two, and neither.

    Also the EV estimator went down but meta-margin went up. Does this mean that OH and FL are tossups now, but Obama has a larger lead in other states?

  • Sam Wang

    Mark, regarding the uptick in the Meta-Margin and the downtick in the EV estimator: yes, your interpretation is correct.

  • Mikerw

    Great analysis. I suspect that your students wouldn’t object to skipping the midterm in the interest of keeping the political junkies fed.

  • Glenn

    Was over at HuffPost and caught an update concerning the new AP numbers… Apparently they’re flawed, nearly doubling the number of evangelical/born again christian voters from 23% of likely voters in 2004 to 44% today. That’s totally implausible.

  • Sherry Havelka

    Sam, does your current analysis show any hope for McCain in PA?

  • Sam Wang

    Sherry Havelka: no.

  • Bill Dalrymple, Princeton Health Service 1961-1977

    Last Thursday, a busload of Black HS seniors drew up at a Ft, Lauderdale voting site and 38 piled out and joined the very long line presumably to vote for Obama. I bet your estimate of 1% downtick of his probable vote in the polls is a low estimate.

  • Sam Wang

    Note that pollsters are now asking whether people have voted early, and if so, how they voted. Therefore these 38 people should be sampled in polls.

Leave a Comment