TEN days before the election, the race for the presidency is essentially where the median political-science forecast issued two months ago projected it would be: a statistical dead heat in the popular vote, as measured by the latest tracking polls. In the RealClearPolitics poll of polls, Mitt Romney has crept into a narrow lead, 47.9%-47%*, over Barack Obama. The Pollster.com composite poll, which uses a slightly different method of aggregating poll results, shows Mr Obama up by a tenth of a percentage point, 47.2%-47.1. It is probably fair to say that except for some political scientists, almost no one envisioned this state of affairs a month ago, as both candidates prepared for the first presidential debate. Then, Mr Obama led in the RCP polling average by 3.1 points, down only one point from Mr Obama’s peak post-convention lead, obtained on September 30th. Indeed, in both the RCP and Pollster.com composite polls, Mr Obama led Mr Romney by 3 points or more for almost the entire month of September.
Mr Romney’s gain at the national level has also raised his state-level polls, including in the key battleground states. But because Mr Romney is coming from farther back in these swing-state polls, Mr Obama’s aggregate battleground advantage has been reduced, but not eliminated. Today, two of the most widely viewed state-based electoral-college forecast models, by Drew Linzer and Simon Jackman, both show Mr Obama still holding an edge in electoral-college votes, even as the national tracking polls are dead even. Mr Linzer, whose Votamatic website estimates the likely electoral college results by combining a structural forecast model based on past elections with current state polls, projects Mr Obama to win 332 electoral-college votes compared to 206 for Mr Romney. Mr Jackman, who uses a similar state-based polling method, expects Mr Obama to win somewhere close to 302 electoral-college votes, though he says that “290 is the single most likely outcome (observed with 13% probability)”.
In assessing the remaining battleground states, it is clear that Ohio is the main pillar in the Obama firewall. Even if Mr Romney wins North Carolina, Florida, Colorado and Virginia—the four battleground states in which he is polling the best—he will still fall short in the electoral college without also winning Ohio. (Mr Romney trails in the three remaining toss-up states—Nevada, Iowa and New Hampshire—by more than he does in Ohio.) How likely is it that in the remaining week and a half Mr Romney can close the gap in the swing states, particularly Ohio, and breach Mr Obama’s electoral-college firewall? Not very, according to Mr Linzer, who points out that Mr Romney’s post-debate “surge” gained him perhaps two points. Without a similar, widely-viewed focusing event, Mr Linzer finds it hard to conceive of Mr Romney winning another two points in a rapidly diminishing pool of undecided voters.
Are Mr Linzer and Mr Jackman right to be so bullish regarding Mr Obama’s electoral prospects? The answer depends in part on one’s explanation for why Mr Romney trails in the electoral college, despite running even in the national polls. One answer is that Mr Obama’s support is distributed more efficiently; where he leads in the battleground state polls, he does so by a small margin, while Mr Romney has racked up larger polling leads in a few states. Note, however, that this relative distribution of votes bucks historical trends; Democrats in the period 1932-2008 who get 50% or less of the popular vote have not usually won the electoral college.
Moreover, according to Mr Jackman, “on average, 50% of the national two-party vote translates into 247 electoral-college votes”—that is, 23 electoral-college votes short of the 270 needed to win. If current national polling trends showing a 50-50 popular-vote split hold, then, Mr Obama will have to buck the historical odds to squeak out an electoral-college victory. But Mr Jackman’s calculation comes with a very wide confidence interval, indicating a great deal of uncertainty regarding the true relationship between the popular and electoral-college vote.
As the presidential race enters the homestretch, partisans on both sides are debating whether the race has entered a new, stable equilibrium that will endure to election day, or whether Mr Romney still has room to add to his post-debate momentum. If the former, Mr Obama, protected by the electoral-college firewall, may yet hold on for victory. But if the dwindling number of undecideds—now only about 5% of the electorate—breaks decisively for Mr Romney, even that firewall will not be enough to save the president.
* Corrected from 49.7%-49%. Sorry.