Princeton Election Consortium

A first draft of electoral history. Since 2004

Obama 318 EV popular +3.0% Meta +3.7%; House Dem win 91%; Senate toss-up!

July 8th, 2012, 8:53am by Sam Wang


Update: Presidential charts are live! All credit is due to our own Andrew Ferguson, as well as Pollster.com, whose high-quality data feed we rely on. Stay tuned for Senate/House…

Poll-based “forecasts” add gratuitous uncertainty that make a race look closer than it really is. This leads to underestimates of the EV estimator and the current win probability. For instance, the “Now-cast” at FiveThirtyEight has substantial errors. Although future trends are hard to predict, the most accurate forecast for November is a snapshot of where things are today. Here’s where the national races stand  as of July 8, 2012:

President: Obama vote share 51.5 +/- 0.2 %, Obama 318 EV to Romney 220 EV. Meta-margin: Obama +3.72%. Feels like Obama’s going to win (barring a European collapse). The polls-today snapshot is 7 sigma (huge). November is a ways off, but I would still give 8-1 odds.

House: The generic Congressional ballot, which is fairly reliable, favors D by +3.0 +/- 2.2 % (seven polls in June, each pollster counted once). An election based on this sample would give a Democratic takeover with 91% probability (i.e. 1.4 sigma).

Senate: D 49-52 votes, a knife edge (49=R control, 50 to 52=D control). Close races to watch: MA Warren (D) v. Brown (R) , VA Kaine (D) v. Allen (R), ND Heitkamp (D) v. Berg (R), IN Donnelly (D) v. Mourdock (R). If Republicans win three of these four races, they are likely to gain control of the Senate. The main problem in ND and IN is not enough data (which is where interpolative models add value). Current evaluation: give to Warren and Kaine (or their opponents, take your pick). There’s also a more unusual Senate campaign giving opportunity…

The state of the Wisconsin Senate race depends on who gets the GOP nomination. If Thompson (R), he’s heavily favored. If Hovde (R), then Baldwin (D) is favored. And Thompson and Hovde are neck-and-neck. What to do? Hmmmm.

Tags: 2012 Election

8 Comments so far ↓

  • Sam Wang

    Wu and Matt, thank you. Matt, glad to see you again.

    Whatever one thinks will happen in the next few months is legitimate speculation. As a quantitative matter, at the moment the Obama InTrade price is undervalued at 56%. If your goal in reading this site is to make money, the obvious recommendation at present is to buy that contract — if one thought that the price tends to move toward the actual probability. InTrade stops short of that point, as I have pointed out before.

    If your goal is to maximize your influence on next year’s political landscape, I would give to some of those marginal Senate races. I don’t actually think Warren is the optimal target, on the grounds that the race is already awash in money. But Massachusetts is certainly a more effective place to give than, say, Pennsylvania or Washington State.

  • Matt McIrvin

    …As for this site, I’m glad it’s back and glad that it validated my decision to give most of my campaign bucks to Elizabeth Warren.

  • Matt McIrvin

    On Intrade you’re not really betting that Romney will win, you’re betting that Intrade’s Romney stock will go up at some point that you can sell before the election. On those grounds, it might well make sense to buy Romney, and sell the day after the RNC.

  • Albert

    Based on the number alone I agree that Obama would be the favourite if the election held today – and I think he should win this one. However, I am now long Romney heavy on Intrade for couple of reasons:

    - Romney’s and RNC outraising Obama and DNC by some $35 million just in June alone. This does not take into account the GOP’s Super PAC which I would think haul in truckload of cash as well. Obama has been spending more in television Ads than Romney counterpart – and we shall see Romney returning the favour soon. I think they’re waiting for a good timing to release the arsenal (i.e worsening job numbers)

    - Convention bump. In 2008, the convention brought McCain even slightly ahead than Obama for a very brief period of time.

    -VP Pick. I know it’s been beaten to death that VP makes very insignificant impact, however Portman could very well be the exception since Obama is now ahead by just 2 points there (courtesy of RCP).

    All in all, I think this could get to 50-50 come August / end of August.

  • Albert

    Sam! Finally! :) .. I am a huge huge fan of this website … welcome back! looking forward for your insight and analytic on the election.

  • Sam Wang

    I did notice that McConnell comment. The Senate probability is likely to shift in the coming month, since right now data are scanty.

    The Nate Silver book: erm, do I have to? “Signal vs. noise”…ironic, since there is noise creeping into his calculation, mixed in with the signal. At present, he gives the probability of Obama winning the election if the election were held today (“Now-cast”) at 80.5%. That’s 0.85 sigma.

    In other words, he expresses less confidence in his calculation that one would get from reading a single poll — and he is using dozens of polls at both the national and state level. For a snapshot of today’s standings, a correct calculation that added value would give extremely high statistical confidence.

  • Wu

    Fascinating post. Could recent Presidential elections really all have been called by the middle of the July before the vote?

    Looks like the Senate races will be interesting. Maybe Senate Republican leader Mitch McConnell reads this blog: this morning on CNN’s State of the Nation he said the odds of Republicans taking over the Senate are 50-50.

    Also, later this year I’d love to get a review from you of Nate Silver’s upcoming ‘Signal and the Noise’ book!

  • » Election 2012 coverage – under construction

    [...] Election 2012 coverage – under construction Go to this article [...]