I’ve been tiptoeing around this point, because I don’t think that our forecast model ought to represent the totality of our analysis about the election.
But if you look at the right-hand column of this page, you’ll notice that our forecast has moved toward President Obama over the past several days. It now gives him about a three-in-four chance of winning the Electoral College on Nov. 6.
I’ll explain a little bit more about how the model comes to that conclusion in a moment, but the intuition behind it is pretty simple:
1. Polls usually overrate the standing of the candidate who just held his convention.
2. Mitt Romney just held his convention. But he seems to have gotten a below-average bounce out of it. The national polls that have come out since the Republican National Convention have shown an almost exact tie in the race.
3. If the polls overrate Mr. Romney, and they show only a tie for him now, then he will eventually lose.
The first point is the simplest of all, but perhaps the most important. There is a lot of focus on the bounce that a candidate gets after his convention — that is, how the polls conducted just after the convention compare with the ones taken immediately beforehand.
But the more instructive comparison may be how the post-convention polls track with the actual election result — it’s Nov. 6 that we’re really concerned about, after all.
On average, between 1968 and 2008, the challenging candidate led by 10 percentage points in polls conducted just after his convention. By comparison, the challenging candidate eventually lost the popular vote by an average of three points in these years. That means the post-convention polls overrated the challenger by an average of 13 points.
The good news for Mr. Romney is that this tendency has been growing smaller over time. However, it hasn’t necessarily disappeared. In 1992, Bill Clinton led by more than 20 points after his convention, but his actual winning margin was only about six. In 2000, George W. Bush came out of his convention in Philadelphia with about a 10-point lead in the polls — but he eventually lost the popular vote. John Kerry, in 2004, got very little bounce from his convention. But Mr. Kerry nevertheless came into his convention with a lead, and he maintained it — then he lost the popular vote by about two points instead.
There were also two recent cases in which the post-convention polls did not overrate the standing of the challenging candidate. In 1996, Bob Dole trailed by about nine points immediately after the Republican convention in San Diego, which is the same margin by which he eventually lost to Mr. Clinton. And in 2008, Mr. Obama led by about four points after his convention in Denver, but won by a wider margin, seven points, instead.
Both 1996 and 2008 are slightly odd cases, however. In 1996, the polls had Mr. Clinton ahead of Mr. Dole by double-digits almost the whole way through the campaign, including on Election Day itself; the period right after the Republican convention that year was about the lone exception. So this may have been a case of two wrongs making a right: the polls were persistently biased toward Mr. Clinton, and Mr. Dole’s convention bounce counteracted that temporarily.
And in 2008, it’s not quite clear what would have happened if not for the collapse of Lehman Brothers, which occurred about two weeks after the party conventions. Mr. Obama very probably still would have won, but it might have been by a smaller margin had the economic problems not become so acute so quickly.
Certainly, Mr. Romney could benefit from a parallel to either of these circumstances. The election is close enough that if there is even a modest bias in the polls toward Mr. Obama, the president’s chances will be tenuous.
And the economic recovery is tepid enough that if there is some intervening event, such as in Europe, then some voters who were willing to give Mr. Obama the benefit of the doubt may change their minds.
But that isn’t how you draw these things up. You don’t want your chances to come down to the residual chance of a polling error or an October surprise. You want to be ahead after your party convention — not just tied, something that even Walter F. Mondale had (very briefly) managed to do in 1984.
The forecast model had no expectation of a double-digit bounce for Mr. Romney. This year the polls have been hard to move for any reason. By the model’s logic, that implies that the conventions wouldn’t move them much either.
But since Mr. Romney’s deficit with Mr. Obama was very narrow (he trailed by about two percentage points heading into the Republican convention in Tampa, Fla.) it wouldn’t have taken much of a bounce to put him ahead.
Specifically, the model’s assumption was that the post-Tampa polls would overrate Mr. Romney’s eventual finish by about four percentage points. That’s why the forecast has moved toward Mr. Obama over the last several days. It sees a bunch of polls showing a tie in the race, and interprets them as tantamount to a four-point Obama lead, since they were taken in the midst of what should be a high point for Mr. Romney.
Now that the Democratic convention has begun in Charlotte, N.C., the pressure will be on Mr. Obama. For two weeks or so after Thursday, the model will subtract a point or two from Mr. Obama’s column in the polls.
But Mr. Obama seems to have more control of his own destiny right now. If he carries even a modest bounce out of Charlotte, he’ll remain in the front-runner’s position.
And if Mr. Obama gets a bounce that’s a bit better than modest — say, he leads in the national polls by in the neighborhood five or six points next week, as Mr. Bush did following his convention in 2004 — Mr. Romney’s position will start to look fairly grim.
It may be presumptuous, of course, to assume that Mr. Obama will get any bounce at all. And even a strong speech in Charlotte could seem futile if we get a poor jobs report on Friday. By this time next week, we could be talking about how Mr. Romney’s two-point bounce looks good compared with Mr. Obama’s zero-point bounce.
But we have to evaluate the data as it comes in. Being only tied in the polls immediately after his own convention is unambiguously a bearish sign for Mr. Romney — and probably the most tangible sign to date that Mr. Obama is the favorite.