Sept. 30: Romney Down a Touchdown?

By NATE SILVER

You can think of a general election campaign as being divided into four quarters.

The first one consists of the presidential primaries. What matters in this stage is not only who gets nominated, but how well they position themselves for the remainder of the race, through things like fund-raising, adopting (or avoiding) issue positions that might be damaging later, gaining experience on the campaign trail, building out their staffs and receiving at least an initial round of vetting by the news media.

The second quarter begins when the candidates effectively clinch their nominations, and the general election campaign begins in earnest. This period is long and sluggish, and most of the “fans” (voters) will not really have tuned in yet. But it is also a time when a campaign makes important strategic decisions, in terms of its tone and tenor, its allocation of resources, and its issue positioning - decisions that it may be stuck with, for better or for worse. The second quarter culminates when the candidates make their vice-presidential selections.

The third quarter consists of the party conventions and their immediate aftermath. It is a short quarter as far as the calendar goes, but one in which the winner of the election often becomes much more apparent. George H.W. Bush in 1988, Bill Clinton in 1992, and George W. Bush in 2004 moved into leads after the conventions that they essentially never relinquished.

We've probably already been in the fourth quarter for a week or so, because we've already passed the point when a convention bounce (if it was indeed a bounce and not a permanent shift in the conditions) might be expected to wear off. And we'll reach a fourth-quarter landmark on Wednesday, when President Obama and Mitt Romney hold their first of three debates.

Every now and then, the game goes into overtime - in which case things lik e turnout and the Electoral College math suddenly begin to matter a great deal. But let's not get ahead of ourselves: it's early in the fourth quarter and Mitt Romney finds himself down in the race. The question is how far behind he is, and what he'll have to do to make up his deficit with Mr. Obama.

According to the win probability calculator at AdvancedNFLStats.com, an N.F.L. team down by field goal with 10 minutes left to play in the fourth quarter has a 34 percent probability of winning the game. A team down by a touchdown wins just 16 percent of the time.

(A technical note for sports geeks: these cases assume that the trailing team has possession of the football with first down and 10 yards to go at its own 20 yard line.)

If you look at our estimate of Mitt Romney's chances of winning the Electoral College, which are about 15 percent right now in the FiveThirtyEight forecast, the touchdown analogy works best: Mr. Romney has about as much chance of win ning as an N.F.L. team does when it trails by a touchdown early in the fourth quarter.

It might be surprising that a team down by just a touchdown - a close game, by any common description of it - winds up winning so rarely. But there are a few things to consider.

First, a field goal alone won't be enough for the team to come back. It needs something big to happen - or it needs to score at least twice.

Second, although there's still enough time in the game for the trailing team to have multiple opportunities to score, there is also enough time for the opponents to score as well and extend their lead. So the team still has to play defense - it's not purely a two-minute drill.

A third and often overlooked (if completely obvious) point: if the trailing team does score a clutch touchdown, it only ties the game. There are a lot of cases in which it will later lose anyway.

Right now, our forecast says that Mr. Romney has only about a 15 percent chance of winning. But that does not mean that he only has a 15 percent chance of tightening the race - or of making it come down to the wire.

But there are plenty of circumstances in which Mr. Romney has some good things happen, makes the race very close, and then loses - whether because he loses Ohio, or because his turnout operation isn't much good, or because the polls turn out to be slightly biased toward him rather than against him.

As for what might happen this week: the first debate alone will probably not provide an opportunity for Mr. Romney to score a touchdown. Historically, the largest shifts in the polls after the first debate have been about three points in either direction - smaller than Mr. Romney's current deficit in most polls. This would be the equivalent of a field goal.

But there is also another potential scoring opportunity for Mr. Romney: the monthly jobs report, the next-to-last one before the election, will come out on Friday.

My v iew is that we've gotten to the point where the importance of these jobs reports can be a little overstated. That's not because the economy isn't important (of course it is), or because jobs aren't an important economic indicator (of course they are), but because any one economic report won't tell you all that much. Furthermore, both the forecasts of jobs numbers and the government's initial estimates of it are quite crude. Payrolls are expected to have grown by about 115,000 jobs in this month's report, but those forecasts miss, on average, by about 70,000 jobs.

Still, there is about a one-in-five chance (based on the historical distribution of forecast errors) that the jobs estimate will miss by at least that margin to the downside, which means that fewer than 50,000 jobs will have been created last month. If we print a jobs number somewhere in that range - say, just 35,000 jobs added last month - perhaps that counts as another field goal for Mr. Romney.

Here's where the precision of our analogy breaks down. Unlike in a football game, we do not know exactly how far Mr. Romney is behind. Could a clear win for Mr. Romney in the debate - and a distinctively bad (but not subzero) jobs report - be enough to draw the race into a tie? One of those things alone probably won't do it. But if he gets both? It seems possible, at least, but it's hard to know for sure.

Also unlike in a football game, of course, the candidates don't necessarily take turns. We've just described Mr. Romney's dream scenario, an imaginable (if not exceptionally likely) sequence of events that could have the race looking almost even by a week or two from now. But it could be Mr. Obama who scores next instead. If he is perceived as winning the debate, and the jobs number misses by a substantial margin to the upside (meaning that more than about 180,000 jobs are created), Mr. Romney's position could become extremely desperate.

Sunday's Polls

There wer e some interesting polls on Sunday, but not necessarily ones that ought to much alter our understanding of the race.

The most noteworthy number was a national poll from The Washington Post and ABC News. Their top-line number was good for Mr. Romney, putting him just two points behind among likely voters - closer than most other recent surveys.

In context, however, the poll was not quite as encouraging as it seemed for Mr. Romney. One issue is that the trendline did not improve for him. The Washington Post and ABC News national polls have generally shown fairly strong numbers for Mr. Romney; they had him behind by just one point among likely voters in a poll they conducted just after the Democratic convention, and ahead by two points in their poll before either party convention.

Another issue is that the story that the Washington Post and ABC News polls are telling about the election is one in which Mr. Romney has Electoral College problems. Although their national polls have shown a very tight race throughout the year, their recent surveys of Florida, Ohio and Virginia had Mr. Obama with big leads there. For that matter, a breakout from their latest national poll showed Mr. Obama with a much clearer lead in swing states than in the country as a whole. I don't think you should make very much of that - I'd much rather look at actual polls of the swing states than a small-sample breakout that lumps them all together - but since the Washington Post and ABC News polls have shown quite poor numbers for Mr. Romney in the battleground states that they surveyed, at least we're getting a consistent story from them.

There were also two Ohio polls on Wednesday. One, from Public Policy Polling, put Mr. Romney down by four points there. If The Washington Post's national poll was not quite as strong as it looked for Mr. Romney, this one was a little better for him than it might appear, since Public Policy Polling's surveys are slightl y Democratic-leaning and since their other two polls of Ohio since the conventions had shown Mr. Obama with slightly larger leads.

However, this was counterbalanced by a Columbus Dispatch poll, which put Mr. Obama ahead by nine points in Ohio. The Columbus Dispatch poll is highly unusual - it is conducted entirely by mail - a technique that has produced some big hits and big misses in the past. But the trendline is worth looking at even if the topline numbers should be taken with some caution, and it was very favorable for Mr. Obama since their August poll had shown a tied race.

My impressionistic view is that Mr. Obama's polls have not been quite as strong over the past three or four days as they were early last week, but the FiveThirtyEight forecast does not yet record any signs of a turnaround. Mr. Obama's position held steady in the “now-cast,” our estimate of what would happen in an election held today. And he gained slightly in the Nov. 6 forecast, as h e is likely to do on most days in which the polling is neutral or ambiguous, since time is running off Mr. Romney's clock.

But time in a presidential campaign is not linear. By this point next week, two of Mr. Romney's best remaining opportunities to change the game will have come and gone.