Hurricane Sandy appears to have lifted Barack Obama to a very small lead in the national polls. Two weeks ago, Mitt Romney led in the Real Clear Politics national poll average by a bit over 1%. Today he trails by less than 1%. Barack Obama’s ratings as a strong national leader improved a few points from his four days off the campaign trail, as he played the role of executive in charge for the first time in a while. The hug from New Jersey Governor Chris Christie bathed the president in the warm glow of bipartisanship, which is always helpful with independents, especially women independents.
Whatever momentum Mitt Romney carried into the week dissipated as he was largely off the national news media during this period. Were the election held over this past weekend, I believe Obama would have won with a fairly comfortable margin in the Electoral College, perhaps 303-235, with Romney holding all the McCain states and picking up Indiana, North Carolina, and Florida.
This may still be the national result on Tuesday, though there is a bit of evidence in the PPP and Rasmussen national tracking polls that Obama’s numbers may have peaked over the weekend and since slipped a bit in the latest surveys. Obama’s numbers have also fallen in a few state surveys, including Florida, Virginia, Ohio, Iowa, and New Hampshire, while the president seems to be doing a bit better in Colorado.
There is also a new state that may have come into play — Pennsylvania. The latest polls in the Keystone State show the president with a lead in the 3-4 point range, perhaps a point bigger lead for Obama than in Ohio. In the past three presidential elections, Ohio was 9, 5, and 6 points more GOP-friendly than Pennsylvania. The fact that the two states are polling in synch with each other is evidence of the success of the Obama effort in Ohio and/or its complacency in Pennsylvania. It would be a bitter pill to swallow if Obama held Ohio but lost the presidency due to a shocking defeat in Pennsylvania, a state where Democrats enjoy a voter registration edge of over a million.
For Romney to win, he may need to win the popular vote by at least 1%, according to New York Times blogger Nate Silver. Silver’s model has Romney with more than a 60% chance to win if he wins the national popular vote by 1%, but only a 40% chance if he wins by 0.5%. With a 2% national margin, he would be the overwhelming favorite in the Electoral College. A tie in the national popular vote results in Obama winning the Electoral College 80% of the time in Silver’s simulations.
Could Silver’s model be wrong (he gives Romney a 13% chance overall to win) and Romney pull off a big upset? The blogger made his reputation with his successful picks in the 2008 presidential and Senate races. He picked 49 of 50 states right in the presidential contest (I had 48 right), missing only Indiana. I missed Indiana and North Carolina, which Obama carried by 0.3%. We both picked McCain to win Missouri, which he did by 0.1%.
The three states of Indiana, Missouri, and North Carolina were really the only states in doubt on Election Day 2008, so no one who was looking closely should have gotten fewer than 47 right. I would dare say that even Silver would admit that getting North Carolina and Missouri right was due more to luck than to his model’s precision accuracy. Indeed, his model’s accuracy was enhanced throughout the campaign by access to all the Obama campaign polls, which could be used to tweak his own model.
Similarly, there was only one nail-biter in the Senate races in 2008, the Franken-Coleman race in Minnesota. Silver was right on that one and I was wrong (our only disagreement in Senate picks in 2008), but a strong case can be made that Franken’s victory by 300 votes after a long recount may have been due more to superior lawyering than to an actual win in the votes cast. When the result is effectively a tie (e.g., the Bush-Gore race in 2000 in Florida, the Franken-Coleman race in 2008, or the 2008 presidential contest in Missouri or North Carolina), anyone who claims credit for a good pick is kidding himself.