Oct. 30: What State Polls Suggest About the National Popular Vote

Mitt Romney and President Obama remain roughly tied in national polls, while state polls are suggestive of a lead for Mr. Obama in the Electoral College. Most people take this to mean that there is a fairly good chance of a split outcome between the Electoral College and the popular vote, as we had in 2000. But the story may not be so simple

For both the swing state polls and the national polls to be right, something else has to give to make the math work. If Mr. Obama is performing well in swing states, but is only tied in the popular vote nationally, that means he must be underperforming in noncompetitive states.

But polls of noncompetitive states don't always cooperate with the story. Take the polls that were out on Tuesday.

Mr. Obama trailed by “only” eight points, for instance, in a poll of Georgia that was released on Tuesday. Those are somewhat worse results than Mr. Obama achieved in 2008, when he lost Georgia by five percentage points. But they're only a little bit worse, whereas the national polls are suggestive of a larger decline for Mr. Obama in the popular vote.

Or take the poll of Texas, also out on Tuesday, which had Mr. Obama behind by 16 points there. He's obviously no threat to win the state or come close to it, but that still represents only a 4-point decline for Mr. Obama from 2008, when he lost Texas by 12 points instead.

High-population red states like these, Texas and Georgia, are just the sort of places where Mr. Obama would need to lose a lot of ground in order to increase the likelihood of his winning the Electoral College while losing the popular vote.

Perhaps Mr. Obama is underperforming in deeply blue states rather than deeply red ones? Sometimes you'll get numbers that check out with this assumption: Mr. Obama did get some mediocre polling in Oregon on Tuesday, for instance. But he also got a poll showing him ahead by 23 points in California. Another survey on Tuesday gave him a 31-point lead in Massachusetts.

Yes, I am deliberately cherry-picking a bit. But the discrepancy seems to hold if you look at the data in a more comprehensive way. Nor is it an unusual feature of the FiveThirtyEight model. Rather, pretty much every method for evaluating the election based on state polls seems to hint at a very slight popular vote lead for Mr. Obama, along with an Electoral College one.

In the table below, I've listed the current forecasts at seven different Web sites that use state polls, sometimes along with a modicum of other information like a state's past voting history, to produce predictions of the popular vote in each state.

The first of these sites is FiveThirtyEight. The others, in the order that they're listed in the table, are Electoral-Vote.com; Votamatic, by the Emory University political scientist Drew Linzer; HuffPost Pollster; Real Clear Politics; Talking Points Memo's PollTra cker; and the Princeton Election Consortium, which is run by Sam Wang, a neuroscientist at Princeton. These are pretty much all the sites I'm aware of that use state polling data in a systematic way.

You can see that the various projections strongly agree with another, for the most part, in making “calls” about individual states. The only state where different sites show different candidates ahead right now is Florida, where Talking Points Memo gives Mr. Obama a nominal 0.2-percentage point lead while the others (including FiveThirtyEight) have Mr. Romney slightly up instead. There are also four states - New Hampshire, Iowa, Colorado and Virginia - in which some methods show an exactly tied race while oth ers give Mr. Obama the lead.

Although I hope that this chart serves as a useful reference point - and as a reminder that other data-driven sites that look at the polls with the same philosophy that FiveThirtyEight applies are achieving largely the same results - I'm more interested in looking at this data in a macroscopic way.

Suppose, for example, that you take the consensus forecast in each state. (By “consensus” I just mean: the average of the different forecasts.) Then you weigh it based on what each state's share of the overall turnout was in 2008, in order to produce an estimate of the national popular vote.

Do the math, and you'll find that this implies that Mr. Obama leads nationally by 1.9 percentage points - by no means a safe advantage, but still a better result for him than what the national polls suggest.

What if turnout doesn't look like it did in 2008? Instead, what if the share of the votes that each state contributed was the same as in 2004, a better Republican year?

That doesn't help to break the discord between state and national polls, unfortunately. Mr. Obama would lead by two percentage points in the consensus forecast weighing the states by their 2004 turnout.

Or we can weigh the states by their turnout in 2010, a very good Republican year. But that doesn't help, either: instead, Mr. Obama leads by 2.1 percentage points based on this method.

(In each of these examples, you'd get almost exactly the same outcome if you used the FiveThirtyEight forecast alone rather than the consensus. We're on the high end and the low end of the consensus in different states for Mr. Romney and Mr. Obama, but it pretty much balances out over all.)

Whether the state polls or the national polls characterize the election correctly could well determine its outcome.

Mr. Obama's lead in the Electoral College is modest, but also quite consistent across the different methods. The states in wh ich every site has Mr. Obama leading make up 271 electoral votes - one more than the president needs to clinch victory. The states in which everyone has Mr. Romney ahead represent 206 electoral votes. That leaves five states, and 61 electoral votes, unaccounted for - but Mr. Obama would not need them if he prevails in the states where he is leading in the polls.

But perhaps national polls tell the right story of the race instead - meaning that the state polls systematically overrate Mr. Obama's standing?

It's certainly possible. (It keeps me up late at night.) If the polls in states like Ohio and Wisconsin are wrong, then FiveThirtyEight - and all of our competitors that build projections based on state polls - will not have a happy Nov. 6.

With that said, our decision to cast our lot mostly with the state polls is not arbitrary. In recent years, they've been a slightly more unbiased indicator of how the election will play out.

Bias, in a statistical sense, means missing consistently in one direction - for example, overrating the Republican's performance across a number of different examples, or the Democrat's. It is to be distinguished from the term accuracy, which refers to how close you come to the outcome in either direction. If our forecasts miss high on Mr. Obama's vote share by 10 percentage points in Nevada, but miss low on it by 10 percentage points in Iowa, our forecasts won't have been very accurate, but they also won't have been biased since the misses were in opposite directions (they'll just have been bad).

In a previous article, I examined the history of bias in public polls based on whether they've tended systematically to overrate the standing of the Democrat or the Republican. (The answer is that they don't exhibit either bias on a consistent basis, as long as your using likely voter polls; registered voter polls will tend to overstate the vote for the Democrat.)

This article also contained a comparison between state and national polls in the presidential race: which have been more free of bias?

In recent elections - since state polling data became more robust - it's the state polls that have done a bit better. This was especially so in 1996, when national polls implied a double-digit victory for Bill Clinton over Bob Dole (and Ross Perot) but state polls were more in line with the single-digit victory that he actually achieved. In 2000, state polls provided an accurate portrayal of a too-close-to-call race, while national polls missed high on George W. Bush vs. Al Gore.

There have been other years like 1992 in which the national polls did a bit better. But on average since that year, the state polls have had a bias of 1.1 percentage points - half as much as the national polls, which have had a 2.1-point bias instead.

We're approaching the point where Mr. Romney may need the state polls to be systematically biased against him in order to win the Electoral College. And that certainly could turn out to be the case: if Mr. Romney wins the popular vote by more than about two percentage points, for example, he'll be very likely to cobble together a winning electoral map, somehow and some way. (And he'll be a virtual lock if the results are in line with Mr. Romney's best national polls, like the Gallup survey, which put him four or five points ahead.)

But the historical evidence weighs in slightly more heavily on behalf of the state polls, in my view, when they seem to contradict the national ones. If the state polls are right, than Mr. Obama is not just the favorite in the Electoral College but probably also in the popular vote.

Tuesday's Polls

Mr. Obama made gains in the FiveThirtyEight forecast on Tuesday, with his chances of winning the Electoral College increasing to 77.4 percent.

A fair amount of this boils down to Ohio, where three polls released on Tuesday gave Mr. Obama leads by margins ranging from three to five percentage points. Two of the polls, from Grove Research and the Mellman Group, generally show strong results for Democrats, which give them less impact in the forecast after applying our adjustment for pollster “house effects”. Still, the three polls taken collectively were enough to widen Mr. Obama's projected lead in Ohio to 2.4 percentage points from 2.1 on Monday. Given how central Ohio is to each candidate's electoral strategy - and how little time remains in the race - this was enough to improve Mr. Obama's Electoral College chances. (The forecast does not yet account for the poll by Quinnipiac University for The New York Times and CBS News, which had Mr. Obama five points ahe ad in Ohio but which was released after we had run the model for the night.)

Another poll that received a lot of attention on Tuesday was one by Glangariff Group Inc. in Michigan, for The Detroit News. That survey had Mr. Obama ahead by only 2.7 points in Michigan.

There has been some odd polling in Michigan this year, but the Detroit News polls have not been a big part of the problem. Instead, its surveys have usually come pretty close to the polling consensus in the state. Furthermore, this survey suggests tightening in the race in Michigan since earlier this month, when a poll by the same firm had Mr. Obama ahead by 6.7 percentage points instead.

Nonetheless, Michigan is probably not as close a s two or three points right now: most polls released after the first debate in Denver suggested a lead for Mr. Obama in the mid-to-high single digits. Usually, states do not shift all that much relative to others in their region. The fact that Mr. Obama's polling has held up reasonably well in Ohio and Iowa, for example, is reason to suspect that some of the movement in the poll represents statistical noise, even if it comes from a good polling company.

Perhaps more important, we're at the stage in the race where getting a relatively good poll does not matter all that much: the question is which candidate is ahead outright in enough states to secure 270 electoral votes. Michigan deserves to be monitored over the final week of the campaign, but in all probability Mr. Romney's more likely paths to victory will run through Ohio instead.

Mr. Obama had a somewhat above-average day in national polls on Tuesday, which had him up in the race by about one percentage point on average. Part of this is because the Gallup poll, which has shown very poor results for Mr. Obama, did not publish results on account of Hurricane Sandy.

Perhaps the most intriguing result from this group is the poll from Google Consumer Surveys. (Yes, Google has begin to conduct surveys online.) That poll had Mr. Obama ahead by four percentage points, an improvement from a roughly 1-point deficit for Mr. Obama in their prior survey last week.

The Google survey could be an indication that the effects of the hurricane will play somewhat to Mr. Obama's political advantage. But it will probably be Thursday or Friday, once power and some of the national tracking surveys that have been discontinued have come back online, before we can say so with much confidence.

In the meantime, the state polls continue to hint that Mr. Obama remains the favorite to win the Electoral College - and if the state polls are right, he may well be the favorite in the popular vote as well.

Oct. 29: Polling Slows As Storm Wreaks Havoc

The effects of Hurricane Sandy on next Tuesday's election are hard to predict. But the storm is likely to have an impact on the volume of polling in the meantime.

Three of the eight national tracking polls - those from Gallup, Investors' Business Daily and Public Policy Polling - have announced temporary suspensions in their polling. Further delays and cancellations are likely over the next few days, especially in the Northeastern states.

Still, we were not completely without polls on Monday.

President Obama had a mediocre day of polling in the battleground states. In particular, a Rasmussen Reports poll of Ohio showed him trailing Mitt Romney there by two percentage points. It was the f irst poll to show to him down in Ohio since a poll by a Rasmussen Reports affiliate, Pulse Opinion Research, on Oct. 15.

Is this the sign of a shift toward Mr. Romney in Ohio?

It's probably premature to conclude that on the basis of this poll alone. With so many polling firms active in Ohio, any sustained trend toward Mr. Romney in the polls there should be reasonably apparent by late this week, although there may be some delays in the interim. (The storm's reach was broad enough to produce power outages in the Cleveland area, as well as in parts of southeastern Ohio.) Mr. Obama's projected lead in Ohio decreased to 2.1 percentage points from 2.2 points on the basis of the Rasmussen survey.

In Florida, a CNN poll had Mr. Romney up by one point against Mr. Obama , or tied with him when third-party candidates were included on the ballot. Another CNN survey, earlier this month, had also shown Mr. Romney one point ahead in Florida. The poll is consistent with how the FiveThirtyEight projection views Florida, with Mr. Romney ahead by 1.2 percentage points in our forecast there.

An American Research Group poll of Colorado had Mr. Obama one point behind there, although that reflects an improvement for him from their prior poll of the state, when Mr. Obama trailed by four points.

Mr. Obama got reasonably favorable numbers in polls of New Hampshire and Nevada, although both were from Democratic-leaning firms and so did not affect our forecast much.

Arguably Mr. Obama's best poll of the day, instead, was an Elon University survey of North Carolina, which had a tied race there. But North Carolina may be far enough way from the electoral tipping point that it is unlikely to figure all that prominently in next Tuesday's math. Mr. Obama could win North Carolina - the forecast model gives him about a 20 percent chance of doing so - but most of those outcomes involve cases in which Mr. Obama will already have secured enough electoral votes in states like Virginia to have clinched another term.

National polls out on Monday, including some of the tracking polls that released a last set of results prior to suspending their polls, showed a tied race, on average.

On the surface, these numbers look reasonably favorable for Mr. Obama, since he made gains on average (by 0.7 percentage points) from the previous edition of the same surveys.

A poll by Pew Research, for example, had a tied race - a better result for Mr. Obama than their last national poll, which had him trailing Mr. Romney by four points instead.

The FiveThirtyEight model views the Pew poll somewhat negatively for Mr. Obama, however. The reason is that, although a comparison of the trend against the most immediate prior release of the poll is important, the model also evaluates each survey in the context of all other polls that the firm has released over the course of the year. Pew Research had shown very strong results for Mr. Obama earlier in the year - for example, it had him up eight points among likely voters in a survey they conducted after the Democratic convention in September, so a tie there is still a middling result for Mr. Obama by comparison.

Mr. Obama's position declined slightly in the forecast on Monday, with his chances of winning the Electoral College decreasing to 72.9 percent from 74.6 percent on Sunday, and Mr. Romney's increasing to 27.1 percent from 25.4 percent.

Storm's Effect on the Popular Vote

On Monday, I mentioned the possibility that Hurricane Sandy could increase the chance of an split between the Electoral College and the popular vote, on the theory that it might reduce turnout in the blue-leaning states of the Northeast.

After running some numbers on this, I am less convinced that the storm is all that likely to induce such a split.

The FiveThirtyEight forecast of the popular vote is based on adding up the projected results in each state, and then weighing them based on their projected turnout. (Turnout estimates are based on the state's turnout in 2008 and 2010 and its population growth over the past four years.)

As of Monday, this projection showed Mr. Obama ahead in the popular vote by 1.5 percentage points nationwide. (Why does this result differ from what national polls seem to say? That's a great question, and one we have addressed previously: there are systemic differences between the way that state polls and national polls seem to be perceiving the race this year.)

How would the projection change if turnout were reduced by 10 percent in Connecticut, Delaware, Maryland, New Je rsey and New York?

Mr. Obama's projected margin in the popular vote would decline to 1.2 or 1.3 percentage points, meaning a shift against him of only 0.2 or 0.3 percentage points. Instead of having a 6 percent chance of winning the Electoral College while losing the popular vote, as the default version of the FiveThirtyEight model gives him, those chances would increase to about 8 percent.

This is, of course, a fairly crude assessment: a fuller one would require a consideration of exactly which counties were affected, and whether they were Democratic or Republican leaning. It might also need to consider the effects of inland flooding and blizzards on states beyond the coasts, for example, in West Virginia.

But the United States is a big and resilient country; a disaster in one part of it may have a surprisingly small impact when viewed in terms of its entire economy, or its entire population.

Nor do I necessarily think that it can be taken for grante d that the storm will reduce turnout all that much, even in the worst-affected regions. Although the storm's after-effects may make it physically harder for some people to vote, disasters can also increase civic-mindedness and patriotism, attitudes that make voting more likely.

Oct. 29: Polling Slows As Storm Wreaks Havoc

The effects of Hurricane Sandy on next Tuesday's election are hard to predict. But the storm is likely to have an impact on the volume of polling in the meantime.

Three of the eight national tracking polls - those from Gallup, Investors' Business Daily and Public Policy Polling - have announced temporary suspensions in their polling. Further delays and cancellations are likely over the next few days, especially in the Northeastern states.

Still, we were not completely without polls on Monday.

President Obama had a mediocre day of polling in the battleground states. In particular, a Rasmussen Reports poll of Ohio showed him trailing Mitt Romney there by two percentage points. It was the f irst poll to show to him down in Ohio since a poll by a Rasmussen Reports affiliate, Pulse Opinion Research, on Oct. 15.

Is this the sign of a shift toward Mr. Romney in Ohio?

It's probably premature to conclude that on the basis of this poll alone. With so many polling firms active in Ohio, any sustained trend toward Mr. Romney in the polls there should be reasonably apparent by late this week, although there may be some delays in the interim. (The storm's reach was broad enough to produce power outages in the Cleveland area, as well as in parts of southeastern Ohio.) Mr. Obama's projected lead in Ohio decreased to 2.1 percentage points from 2.2 points on the basis of the Rasmussen survey.

In Florida, a CNN poll had Mr. Romney up by one point against Mr. Obama , or tied with him when third-party candidates were included on the ballot. Another CNN survey, earlier this month, had also shown Mr. Romney one point ahead in Florida. The poll is consistent with how the FiveThirtyEight projection views Florida, with Mr. Romney ahead by 1.2 percentage points in our forecast there.

An American Research Group poll of Colorado had Mr. Obama one point behind there, although that reflects an improvement for him from their prior poll of the state, when Mr. Obama trailed by four points.

Mr. Obama got reasonably favorable numbers in polls of New Hampshire and Nevada, although both were from Democratic-leaning firms and so did not affect our forecast much.

Arguably Mr. Obama's best poll of the day, instead, was an Elon University survey of North Carolina, which had a tied race there. But North Carolina may be far enough way from the electoral tipping point that it is unlikely to figure all that prominently in next Tuesday's math. Mr. Obama could win North Carolina - the forecast model gives him about a 20 percent chance of doing so - but most of those outcomes involve cases in which Mr. Obama will already have secured enough electoral votes in states like Virginia to have clinched another term.

National polls out on Monday, including some of the tracking polls that released a last set of results prior to suspending their polls, showed a tied race, on average.

On the surface, these numbers look reasonably favorable for Mr. Obama, since he made gains on average (by 0.7 percentage points) from the previous edition of the same surveys.

A poll by Pew Research, for example, had a tied race - a better result for Mr. Obama than their last national poll, which had him trailing Mr. Romney by four points instead.

The FiveThirtyEight model views the Pew poll somewhat negatively for Mr. Obama, however. The reason is that, although a comparison of the trend against the most immediate prior release of the poll is important, the model also evaluates each survey in the context of all other polls that the firm has released over the course of the year. Pew Research had shown very strong results for Mr. Obama earlier in the year - for example, it had him up eight points among likely voters in a survey they conducted after the Democratic convention in September, so a tie there is still a middling result for Mr. Obama by comparison.

Mr. Obama's position declined slightly in the forecast on Monday, with his chances of winning the Electoral College decreasing to 72.9 percent from 74.6 percent on Sunday, and Mr. Romney's increasing to 27.1 percent from 25.4 percent.

Storm's Effect on the Popular Vote

On Monday, I mentioned the possibility that Hurricane Sandy could increase the chance of an split between the Electoral College and the popular vote, on the theory that it might reduce turnout in the blue-leaning states of the Northeast.

After running some numbers on this, I am less convinced that the storm is all that likely to induce such a split.

The FiveThirtyEight forecast of the popular vote is based on adding up the projected results in each state, and then weighing them based on their projected turnout. (Turnout estimates are based on the state's turnout in 2008 and 2010 and its population growth over the past four years.)

As of Monday, this projection showed Mr. Obama ahead in the popular vote by 1.5 percentage points nationwide. (Why does this result differ from what national polls seem to say? That's a great question, and one we have addressed previously: there are systemic differences between the way that state polls and national polls seem to be perceiving the race this year.)

How would the projection change if turnout were reduced by 10 percent in Connecticut, Delaware, Maryland, New Je rsey and New York?

Mr. Obama's projected margin in the popular vote would decline to 1.2 or 1.3 percentage points, meaning a shift against him of only 0.2 or 0.3 percentage points. Instead of having a 6 percent chance of winning the Electoral College while losing the popular vote, as the default version of the FiveThirtyEight model gives him, those chances would increase to about 8 percent.

This is, of course, a fairly crude assessment: a fuller one would require a consideration of exactly which counties were affected, and whether they were Democratic or Republican leaning. It might also need to consider the effects of inland flooding and blizzards on states beyond the coasts, for example, in West Virginia.

But the United States is a big and resilient country; a disaster in one part of it may have a surprisingly small impact when viewed in terms of its entire economy, or its entire population.

Nor do I necessarily think that it can be taken for grante d that the storm will reduce turnout all that much, even in the worst-affected regions. Although the storm's after-effects may make it physically harder for some people to vote, disasters can also increase civic-mindedness and patriotism, attitudes that make voting more likely.

In Pennsylvania, the Democratic Lean Is Slight, but Durable

We continue our Presidential Geography series, a one-by-one examination of each state's political landscape and how it is changing. Here is Pennsylvania, the Keystone State. FiveThirtyEight spoke with Terry Madonna, a professor of public affairs and director of the Franklin & Marshall College Poll; and Marc Meredith, an assistant professor of political science at the University of Pennsylvania.

New reports indicate that the 2012 presidential campaign is coming to Pennsylvania. After a spate of advertising during the summer, Pennsylvania - in a break from tradition - has largely avoided the volume of campaign commercials that states like Ohio and Virginia have seen.

But beginning Tuesday, Restore Our Future, a “super-PAC” supporting Mitt Romney, will blanket Pennsylvania with about $2 million worth of advertisements. President Obama's advisers greeted that news on Monday by saying that the Obama campaign would also spend advertising money in the Keystone St ate between now and Election Day.

Pennsylvania has been a swing state in presidential elections since the 1950s. In the last 60 years, the candidate who carried the state has also won a national popular vote in every election but one. Over that time, Republicans have carried Pennsylvania in six elections, and Democrats have carried it in nine.

But while Pennsylvania has swung between the two parties, its relative partisan bent has remained remarkably consistent: slightly Democratic.

In fact, Pennsylvania is the only state in the nation that has been unfailingly Democratic-leaning relative to the national popular vote in every presidential election since 1950. Wi th the exceptions of the landslide elections in 1964 and 1984, however, Pennsylvania's leftward lean has been fairly narrow, between one and five percentage points.

The fact that Pennsylvania is just slightly left-leaning and worth 20 electoral votes, tied for the fifth largest haul with Illinois, makes the state an attractive target for Republicans. Pennsylvania is not necessary for Mitt Romney to reach 270 electoral votes, but it would provide him with more flexibility, allowing him to lose two of the three smallest battleground states - New Hampshire, Nevada and Iowa - if he carried North Carolina, Florida and Virginia.

In addition, Pennsylvania has a lot of white, working-class voters who have never been especially enamored with President Obama (Hillary Clinton bested him by 10 percentage points in Pennsylvania's 2008 Democratic primary). Pennsylvania is also relatively old, with the fourth largest share of residents 65 years and older. The Republicans took the state's governor's mansion and a United States Senate seat in 2010.

But Pennsylvania may be fool's gold for the Romney campaign. The state is relatively inelastic; it has few true swing voters, and turnout tends to be the final deciding factor. In other words, the state's Democratic-lean isn't severe, but it is hard to reverse. Yes, Republicans have carried the state six times in the last 15 presidential contests. But in each of those wins the Republican won nationally by at least seven percentage points, a margin that is unlikely this year no matter who wins.

It has also become slightly harder for Republican presidential candidates to put together a winning map in Pennsylvania. The last time a Republican carried the state was in 1988, when President George H. W. Bush defeated Michael Dukakis. Mr. Bush carried the Philadelphia suburbs - Bucks, Chester, Delaware and Montgomery Counties - which have often been the pivotal swing vote .

“The Republicans won elections here in the 1980s because they were winning the Philadelphia suburbs,” Mr. Madonna said.

In contrast, the single time Pennsylvania was carried by a candidate who failed to win the national popular vote was in 2004. President George W. Bush won re-election by 2.5 percentage points but lost the Keystone State by 2.5 percentage points. Mr. Bush lost the all-important Philadelphia suburbs.

Those suburbs have become less-hospitable to Republican presidential candidates. Bucks and Chester Counties are still competitive, but Delaware and Montgomery Counties are reliably left-leaning in presidential elections.

Much of the eastern wing of the state has become more Democratic-leaning, but this has been counterbalanced somewhat by a trend toward the Republicans in western Pennsylvania.

In 1992, the campaign strategist for Bill Clinton, James Carville, described Pennsylvania as Philadelphia in the east, Pittsburgh in the west and Alabama between the two cities (that's not the exact quote, but it conveys the sentiment),

There was an element of truth to Mr. Carville's assessment at the time; the state was bookended by two urban pockets and the rest of Pennsylvania was more rural. But in the 20 years since Mr. Carville made the statement, the state's political landscape has shifted considerably.

Now, a better breakdown of the state is between east and west. Most of central Pennsylvania is still rural, but the eastern third of the state has become Democratic-leaning and culturally and politically Northeastern. Western Pennsylvania has a more Midwest feel and has trended toward the Republican Party.

The regi onal difference is evident in the state's two biggest cities. Philadelphia is heavily African-American and overwhelmingly Democratic. Pittsburgh is less diverse, more blue-collar and less overwhelmingly Democratic. The Republican candidate for governor, Tom Corbett, carried Pittsburgh's Allegheny County in 2010.

The difference is also apparent in the suburbs and smaller cities. The Northeast corridor - stretching from Philadelphia's suburbs in the south up through the Lehigh Valley and into Scranton's Lackawanna County - has become more left-leaning over the past two decades. Part of the shift toward the Democratic Party, particularly in the Philadelphia suburbs, has been driven by women, as the Republican Party became increasingly associated with social issues like opposition to abortion and same-sex marriage. The realignment occurred throughout the Northeast and New England.

The northern part of the Northeast Corridor, in Lehigh Valley and Scranton, is more blu e-collar and less left-leaning than the Philadelphia suburbs. But white college graduates and minorities - groups that skew Democratic - have increased as a share of eligible voters in the Philadelphia suburbs, the Lehigh Valley and the Harrisburg-York-Lancaster region in south-central Pennsylvania.

The south-central region, specifically York and Lancaster Counties, is the beginning of Republican territory. There's a large Amish vote in Lancaster County, and the region tends to be more socially conservative, Mr. Meredith said.

Western Pennsylvania is culturally Midwestern, more socially conservative, and has moved towardthe Republican Party. Politically, it looks like Ohio, with a solidly but not overwhelmingly Democratic city, Pittsburgh, surrounded by heavily Republican suburbs in Westmoreland and Butler Counties.

Over all, the state's leftward lean has increased just barely because as parts of the Philadelphia suburbs moved left, the old mining and mill towns in southwestern Pennsylvania moved right. The southwest corner of the state was historically Reagan Democrat territory, Mr. Madonna said, and has become more Republican-leaning in recent years.

The Bellwether: Bucks County

Bucks County is very likely to provide an early clue as to how Pennsylvania will vote. It has been an almost perfect bellwether, just one percentage point more Democratic-leaning than the state in 2008 and exactly matching the statewide vote in 2004 and 2000.

Southern Bucks County, around Levittown and Bensalem, is solidly Democratic. To the northwest, Bucks County becomes more politically competitive in suburban communities like Yardley (where this i ntrepid FiveThirtyEight writer happened to grow up) and Newtown. Northern Bucks County, around Quakertown, is more Republican-leaning.

The Bottom Line

Mr. Obama is a 94 percent favorite in Pennsylvania, according to the current FiveThirtyEight forecast. The state is hard to move independent of a shift in the national political environment and is unlikely to vote Republican in an election that is so close nationally.

The new Restore Our Future ad buy notwithstanding, the Romney campaign has not seriously contested Pennsylvania, a state that is hard to move without an all-out effort. In the last 30 days, Ann Romney, Mr. Romney's wife, campaigned once in Pittsburgh, and Mr. Romney visited Philadelphia for a fund-raiser.

The state's partisan makeup has changed just slightly since 2008. The Democratic voter registration advantage in Pennsylvania doubled to a little over a million in the run-up to the 2008 election, Mr. Madonna said.

“Since 2008,” Mr. Madonna added, “the voter registration numbers have remained remarkably consistent.”

If Mr. Romney wins nationally by three or more percentage points, Pennsylvania could come along also. But in a closer contest, the Keystone State is likely to remain blue for Mr. Obama.

Impact of Hurricane Sandy on Election Is Uncertain

I'm not sure whether I render the greater disservice by contemplating the political effects of a natural disaster - or by ignoring the increasingly brisk winds whipping outside my apartment in Brooklyn. Still, I thought it was worth giving you my tentative thoughts on how Hurricane Sandy might affect the runup to next Tuesday's election.

We may see a reduction in the number of polls issued over the coming days. The Investor's Business Daily poll has already announced that it will suspend its national tracking poll until the storm passes, and other cancellations may follow. And certainly, any polls in the states that are most in harm's way, including Virginia, New York, New Jersey, Delaware, Maryland and Pennsylvania, will need to be interpreted with extreme caution.

But what about the national polls that remain in the field? Could the storm affect their results?

Imagine that 15 million people are essentially off-limits to pol lsters because of the hurricane, because they are without power, displaced from their homes or otherwise are well-adjusted human beings who are more interested in looking after their families than in answering a political survey. The Northeast is Democratic leaning, of course: imagine that these voters would prefer Barack Obama to Mitt Romney by a net of 20 percentage points, on average.

Fifteen million Americans represent about one-twentieth of the American population. If one-twentieth of Americans, who are 20 points Democratic-leaning, are unable to reply to surveys, Mr. Obama's standing in the polls would be negatively impacted by a net of one percentage point as a result.

That calculation assumes, however, that pollsters are reporting their results verbatim. Instead, almost all polling firms weight their polls to cover non-response among different demographic groups. Some also weight their polls by geography, or might begin to do so because of Hurricane Sandy . This might mitigate the effects, although perhaps in unpredictable ways, since the weighting algorithms that different pollsters use are as much an art form as a science.

Some analysts have also expressed concern that the storm could depress turnout along the Eastern Seaboard on Election Day itself. Since the affected states are Democratic-leaning, and since many of them are so Democratic-leaning that they are likely to vote for Mr. Obama even in a low turnout, it is thought that this might reduce Mr. Obama's national popular vote without hurting his standing in the Electoral College much, potentially increasing the risk of a split outcome.

This is a plausible argument, but let me offer a pair of cautions against it.

First, the Northeast is a wealthy party of the country, and wealthier regions have better infrastructure than impoverished ones, allowing them to recuperate more quickly after a disaster. Were the hurricane expected to hit at the same time ne xt week, it would almost certainly be profoundly disruptive to the election. But the effects might be more modest a week from now.

Second, although the storm surge represents the most immediate threat from the hurricane, inland areas are under considerable risk as well. Hurricane Sandy could potentially flood riverbanks and other low-lying areas, both because of the storm surge carrying forth into them and then because of the potential for large amounts of rainfall. Moreover, these inland regions may be less well prepared to deal with the storm's effects, especially given the news media's tendency to focus its alerts on the impact to major, coastal cities and then to ignore the impact of a storm once it passes through them. (Hurricane Irene in 2011 produced more deaths in landlocked Vermont than in New York City.) Thus, Sandy's after-effects could be felt in red-leaning areas like central Pennsylvania and West Virginia, along with others that are more Democratic-leanin g.

Along the same lines, it is probably unwise to anticipate what affects the storm might have within particular states, such as whether it might affect the Democratic parts of Pennsylvania more than the Republican ones. Hurricane Sandy is just too large a storm, and has such unpredictable destructive potential, to make reliable guesses about this.

The storm, of course, will also affect the plans of the campaigns and the tenor of news coverage about them.

Academic studies on the effects of natural disasters on elections have produced somewhat ambiguous results, but don't contradict the intuitive notion that a disaster response that seems well managed could help an incumbent, while a botched response (especially if the storm damage is severe) could harm him. However, most of these studies seek to evaluate the political effects of disasters on elections held months or even years later, so their utility for understanding the immediate political consequences o f a disaster may be limited.

More important: if you are in an affected region, take the weather forecast very seriously and make sure that you're safe and secure. Public officials, from mayors to presidents, unfortunately may have their incentives corrupted by political considerations, and will not always provide the best guidance to the public as a result. Err strongly on the side of caution; FiveThirtyEight will be here when your power is back on.

Oct. 28: In Swing States, a Predictable Election?

The conventional wisdom about this year's presidential race is that it has broken out of stasis to become wildly unpredictable.

And yet, after a period of polling turmoil following President Obama's convention in Charlotte, N.C., and Mitt Romney's sharp rebound after the first presidential debate in Denver, the polling in most swing states now looks very similar to the way it did for much of the late spring and summer.

When we introduced this year's FiveThirtyEight forecast model on June 7, the closest states were Colorado, Ohio and Virginia, each of which slightly favored Mr. Obama. In Florida and North Carolina, meanwhile, we had Mitt Romney listed as a modest favorite.

Pretty much the same could be said about the race today. In fact, our projected leader in all 50 states is the same as it was at our launch of the forecast in June.

The table below lists our projected res ults in the 18 states (and one Congressional district, in Nebraska) that were predicted to produce a single-digit race as of our original forecast in June. On average, our projection of the popular vote in each state has moved by just one and a half percentage points, in one or another direction, since then.

The change has been especially small in several of the most important battleground states. In June, Mr. Obama was projected to a 3-point lead in Nevada, a 2.3-point lead in Iowa, a 1.3-point lead in Virginia, a 1.1-point lead in Ohio and a 1-point lead in Colorado. The forecast in those states has moved just four-tenths of a point since then, on average; the largest shift has been in Ohio, where Mr. Obama's polling has bee n reasonably resilient and he now has a 2.2-point edge.

Which states have shown more movement? Wisconsin, clearly, became more competitive after Mr. Romney's selection of Representative Paul D. Ryan as his running mate. Mr. Obama enjoyed a more comfortable margin in New Hampshire in June than he does now. And in Indiana and Missouri, two states that are ordinarily red-leaning but which might qualify under a broad definition of swing states, the campaigns have spent few resources and polls there have gradually drifted toward showing a clearer edge for Mr. Romney.

The states where Mr. Obama has made gains since the June forecast are fewer in number (although the fact that Ohio is among them makes Mr. Obama's electoral math stronger). Mr. Obama is running slightly stronger in Florida now than he was in June, although that still reflects a disappointment for him since he was leading in our forecast there before the Denver debate. Mr. Obama has also made gains in Oreg on and New Jersey, counterparts to Indiana and Missouri in that they might have become competitive had Mr. Romney made more of an effort to contest them, but Mr. Romney did not do so.

In blue states, Mr. Obama's numbers are little changed on average from the June forecast. New York is one exception; polls there have shown an especially large lead for Mr. Obama, larger than we originally predicted. And Mr. Obama might have to settle for a single-digit margin of victory in New Mexico, but he is in little danger of losing it.

Change in FiveThirtyEight Forecast - Blue States

It's in deeply red states where the forecast has shifted more. On average, Mr. Romney has gained two percentage points since June between the red states you see in the chart below.

Could the race really have been this stable? Part of it is in the way the FiveThirtyEight forecast is constructed, since it uses nonpolling factors in a state (like its past voting results, relative to the national mean) in cases where the polling is sparse, helping to avoid unwarranted fluctuations based on one or two outlier polls.

But consider this interpretation of events. Prior to the Democratic convention, Mr. Obama's favorability ratings exceeded his approval ratings by several points. The voters who took a favorable view of Mr. Obama but had lukewarm views of his performance may have represented easy targets for the president, and they were brought into the fold by the strong spe eches given by Bill Clinton, Michelle Obama, and Joseph R. Biden Jr., who praised Mr. Obama's accomplishments. After the convention, Mr. Obama's approval ratings caught up with his favorability ratings, and his performance in horse-race polls against Mr. Romney improved as a result.

If Mr. Obama won over his share of Democratic-leaning undecided voters in Charlotte, however, Mr. Romney claimed most of the Republican-leaning ones after Denver. That event precipitated a sharp improvement in Mr. Romney's favorability ratings as voters saw him share the stage with Mr. Obama, and appear competent, confident and presidential.

Some other truly undecided voters may have jumped on the bandwagon along the way, favoring Mr. Obama after Charlotte but Mr. Romney after Denver based on the momentum they perceived in the race. However, their support may be inherently fickle; Mr. Obama now appears to have regained roughly a percentage point of the support he lost after Denver, wh ich may come from this group. (It may also be that the magnitude of the respective bounces was slightly exaggerated in the polls because of the tendency of more enthusiastic voters to respond to surveys.)

The national polls certainly look different than they did in June, but part of that may be artificial as well.

In June, many of the national polls were conducted among registered voters rather than likely ones. In national polls that have reported both registered-voter and likely-voter results throughout the course of the year, Mr. Romney has run about three points stronger in the likely-voter version, on average. (Interestingly, this difference is smaller in state polls, where the difference is about two points on average.)

Some of what looks like movement toward Mr. Romney in other polling averages instead reflects pollsters switching to their likely-voter models. The FiveThirtyEight model, since its design converted registered-voter polls into likely-vo ter ones from the start, will recognize that a change in methodology is different than a new trend in the race.

This change has been less pronounced at the state level, since a larger fraction of the state polls were reporting likely-voter numbers to begin with. We alerted you back in August to the prospect that the different-seeming results between state and national polls would become more apparent as more national pollsters flipped over to their likely-voter models.

Of the remaining gains that Mr. Romney has made in national polls, much of it may have come from his improved performance in deeply red states; that is where our state-by-state forecasts show his numbers improving the most. It might be kept in mind that, during the Republican primaries, Mr. Romney's performance was strongest in states and counties that are Democratic-leaning in general elections, while being weaker in deeply red areas. As highly conservative voters became more comfortable with Mr. Romney, however, he made gains among them.

So sure, there have been some twists and turns along the way - but that is true in every presidential election. Anyone who thinks this race has been especially wild need only look at the polling trajectory in 1976, 1980 or 1992 to get a sense for what a truly variable presidential election looks like.

Over all, instead, this race has been fairly stable relative to most presidential elections. And especially in the state polls, the results we're now seeing are quite consistent with what the economic fundamentals might dictate: a very tight race, narrowly favoring Mr. Obama.

There is always the chance that the race could be disrupted again over the final week of the campaign, perhaps because of the candidates' responses to Hurricane Sandy. And there is the possibility that this will be one of those years where the polls miss the mark badly in one or another direction on Election Day.

There is a pretty good poss ibility, however, that our forecast in every state on Nov. 6 will be the same as it was on June 7. Colorado, Virginia and Florida, being the closest states in the forecast now, are the most likely to switch sides.

Sunday's Polls

The polls released on Sunday were mostly unremarkable. The eight national polls showed about a tied race, on average, with each candidate gaining ground in some polls but losing it in others.

The state polls out on Sunday were slightly favorable to Mr. Obama, with polls by the firm Public Policy Polling showing him making gains in Ohio, Florida and New Hampshire. In another poll of Ohio, from Gravis Marketing, Mr. Obama held a one-point edge, versus a tied race before.

It should be noted that Public Policy Polling is showing a more pronounced shift toward Mr. Obama than any other pollster. Their polls were strongly Democratic-leaning at the start of the year, but then showed a poor run of numbers for Mr. Obama after the Denver debate, with their polls often being slightly more Republican-leaning than the consensus. Now, Mr. Obama has rebounded in their surveys, typically by two or three percentage points in the different swing states that they've polled, to the point that they now seem to be Democratic-leaning again by a percentage point or two.

Our forecasts adjust for these pollster “house effects” - whether pollsters are Republican- or Democratic-leaning relative to the consensus. But one innovatio n that we might consider in future years is to also apply an adjustment for the variance or “swinginess” in a poll. Some pollsters, like Public Policy Polling, American Research Group and Gallup, show more pronounced fluctuations. In others, like Rasmussen Reports and polls for the online firm YouGov, the numbers seem to move much less. There are sometimes logical reasons why these patterns occur - Rasmussen Reports, for example, weighs the results by party identification, which tends to dampen swings (if also potentially missing real changes in the electorate).

Still, there are now so many active polling firms, especially in states like Ohio and Florida, that one or two new polls won't affect the forecast much. (In Ohio, Mr. Obama advanced only to a 2.2-point lead on the basis of the new surveys, as compared with a 2.1-point lead in Saturday's forecast.)

Even in this last full week before the election, I'd encourage you to take a more macroscopic view of the election. We have seen, broadly speaking, a mild recovery for Mr. Obama over the past week or so in the polls. Among polls that have surveyed the race more than once since Denver, his numbers have improved more often than they have worsened in the most recent edition of the survey, and Mr. Obama's predicted probability of winning the Electoral College has improved as a result (to 74.6 percent as of Sunday).

Then again, it may be best not to make too much of these mild fluctuations. There certainly seemed to the the possibility in a brief period after Charlotte that Mr. Obama would run away with the race, although even then forecast expected some reversion to the mean.

But other than in that convention bounce period, the polls have usually told about the same story: that Mr. Obama has a modest edge, but far from an insurmountable one, in the states necessary for him to win him 270 electoral votes.

The forecast model is a bit more confident now about Mr. Oba ma's potential to turn that edge in to an Electoral College victory because there is so little time remaining in the race.

Still, an election held today would probably keep us up quite late before we knew the result with much certainty - and so, in all likelihood, will the one on Nov. 6.

Oct. 27: Minnesota Moonlights as Swing State, but Ohio and Virginia Are More Crucial

With the election so close, Saturday is no longer a day of rest for pollsters. Eight polls were released from potentially competitive states:

This is a diverse group of polls that does not really lend itself to a snappy summation. But let's start in Minnesota, a state that we have given little attention this year.

There were two polls out in Minnesota on Saturday. One, from St. Cloud State University, gave President Obama an eight-point lead. But another, from Mason-Dixon for The Minneapolis Star Tribune, had Mr. Obama with a smaller lead of just three points.

Which of these polls represents the more likely state of play in Minnesota? The FiveThirtyEight forecast projects Mr. Obama t o win there by 6.8 percentage points, meaning that it is somewhat closer to the St. Cloud State poll. Other recent polls, conducted after the first presidential debate in Denver, gave given Mr. Obama leads of margins between 5 and 10 points.

Mason-Dixon is a strong polling firm, but their results have been more Republican-leaning than the consensus in Minnesota and most other states. That does not mean its pollsters are a bunch of partisan hacks; they have a pretty good track record, and it is good for a candidate - in this case, Mitt Romney - when a strong polling firm consistently shows solid results for him.

At the same time, this context - what we call pollster “house effects” - is important to keep in mind. The odds are that if Mason-Dixon were to poll other states that they have skipped, liked Ohio or Virginia, they would also show reasonably strong results for Mr. Romney.

If Mason-Dixon's hypothesis about what the turnout will look like is correc t, then Mr. Obama is likely to have bigger problems than in Minnesota and may be the underdog. But the Minnesota poll does not necessarily present evidence that the state has moved all that much relative to other key states.

In other respects, however, it is surprising that Minnesota has not received more attention. In 2010, Mr. Obama won it by 10 percentage points - the same margin that he had in Iowa, New Hampshire and Pennsylvania, and less than his 14-point win in Wisconsin.

Part of the neglect in Minnesota almost certainly reflects the way the campaigns have been treating the state. There are periodic rumors and reports that Mr. Romney's campaign is dabbling with the idea of buying advertising buys in Minnesota, but he and Republican-aligned groups have spent almost nothing there.

In 2008, by contrast, Republicans held their convention in St. Paul, and John McCain's campaign made a considerable effort to contest Minnesota. It was one of the few states, in fact, where they sometimes outspent Mr. Obama's campaign.

For all those efforts, Minnesota still wound up being a bit Democratic-leaning, since Mr. Obama's 10-point win there exceeded his 7-point margin nationally. The state has gradually drifted from the left wing of American politics toward the center, but its Democratic heritage is hard to overcome.

This year's polls show that Minnesota has been about five points more Democratic-leaning than the country as a whole. Perhaps a late effort by Mr. Romney's campaign could bring it back to within three points of the national average, as it was in 2008. But it is questionable whether that would be a good use of resources, since Mr. Romney would probably win it only if he were performing strongly elsewhere, especially in the Midwest, and that would mean that he probably would have already accumulated 270 electoral votes. Minnesota, despite the new polls, has only a 1-in-500 chance of being the “tipping-point sta te“, according to the FiveThirtyEight model, meaning that its electoral votes would be decisive given how all the other states might line up.

The most consequential polls of the day were probably in Ohio and Virginia.

The Ohio poll was a good one for Mr. Romney. The survey, conducted by the University of Cincinnati for a consortium of Ohio newspapers, showed the tied race, 49-49, with almost no undecided voters left. The same survey had given Mr. Obama a 5-point advantage before the Denver debate.

Some liberals have critiqued the Ohio poll for being out of date - it was in the field between Oct. 18 and Oct. 23, meaning that some of its interviews were conducted before the final presidential debate in Florida.

I think this criticism is probably overdone. There is little evidence that the race has changed all that much since the final debate; the FiveThirtyEight model finds that Mr. Obama has perhaps gained half a percentage point nationally since the n, but probably not much more than that.

And apart from the timing, the poll has a lot going for it: it has a good track record and collected a reasonably large sample size, meaning that it gets a lot of weight in the FiveThirtyEight forecast.

But the poll should not be used to imply that the race is tightening further in Ohio. There have been 12 other polls of the state that also conducted at least some interviews after the Florida debate, and they showed Mr. Obama up by two points there on average, which is about where the FiveThirtyEight forecast now shows the state. If a candidate holds a two-point lead in a state, it is normal for some polls to show him tied or trailing by a point or so instead in contrast to others that might put him four or five points up.

That is pretty much what we see in Ohio right now, with the edge in the polling average remaining with Mr. Obama. The new poll reduced his chances of winning the state to 73 percent from 76 percent in the forecast.

The best poll of the day for Mr. Obama was in Virginia, where a Washington Post survey showed him four points ahead. The result is not a huge surprise since The Washington Post had also shown good numbers for Mr. Obama there in the past, putting him eight points ahead in a survey conducted before the Denver debate.

Other polls of Virginia this week have shown anything from a two-point lead for Mr. Romney to a five-point edge for Mr. Obama. Perhaps, along with New Hampshire, it has featured the least consistent polling of any swing state all year.

One critique I have seen of the poll is that it seems out of step with The Washington Post's take on the national race. Their most recent national tracking poll, conducted with ABC News, gave Mr. Romney a one-point lead.

I agree with this question to some extent: it is unlikely that Mr. Obama will lose the national popular vote by one point while winning Virginia by four.

At the same t ime, The Washington Post is hardly along among pollsters to show better results for Mr. Obama in the swing states than in their national numbers. This is a pattern that we have been observing all year, and it seems to have become more pronounced recently.

There are three ways to resolve the conflicting numbers: they could reflect a pronounced advantage for Mr. Obama in the Electoral College relative to the popular vote. Or there are the more pedestrian hypotheses that get too little consideration, in my view: that the state polls might be overrating Mr. Obama's performance, or alternatively, that the national polls might be underestimating it.

The FiveThirtyEight model accounts for all of these possibilities, in its own way. It does find that Mr. Obama is slightly more likely to win the Electoral College than the popular vote. But our method also lowers Mr. Obama's projection in the states, while raising it in the national polls, because they are somewhat discord ant with each other.

Still, Virginia is a state that Mr. Romney is having difficulty putting out of play. (The other poll published on Saturday, from Gravis Marketing, should not necessarily give comfort to Mr. Romney. It showed a tie in Virginia, but its previous poll had given Mr. Romney a five-point edge instead, having been an outlier at the time.)

Virginia now ranks second on our list of tipping-point states. It has a 15 percent chance of deciding the election - well behind Ohio's 49 percent chance, but ahead of everything else.

Suppose that Mr. Romney were to win all the states where he is currently favored in the FiveThirtyEight forecast, including Florida and North Carolina, and that there are no major upsets in the blue-leaning states, with Mr. Obama winning Pennsylvania, Minnesota and Michigan. Then Mr. Obama wins Virginia.

That would give Mr. Obama 250 electoral votes to Mr. Romney's 235, with six states - Colorado, Iowa, New Hampshire, Nev ada, Ohio and Wisconsin - unresolved.

Mr. Obama could win the election by winning Ohio, plus any of the other five states. (Ohio alone would not quite suffice, bringing him to 268 electoral votes.)

The real value of Virginia to Mr. Obama, however, is that it would allow him more paths to victory even without Ohio.

For instance, if Mr. Obama won Virginia but lost Ohio, he could still win the election by carrying Nevada and Wisconsin - where he has a clearer edge in the polls than he does in Ohio - and then any one of the following states: New Hampshire, Iowa and Colorado.

The map below represents one of Mr. Obama's more plausible winning paths that would get him to 270 electoral v otes without Ohio.

Over all, the stronger poll in Ohio for Mr. Romney slightly outweighed the poorer one for him in Virginia. His chances of winning the Electoral College inched up to 26.4 percent from 25.6 percent in Friday's forecast.

Oct. 26: State Poll Averages Usually Call Election Right

The FiveThirtyEight forecast model has found the past several days of battleground state polling to be reasonably strong for Barack Obama, with his chances of winning the Electoral College increasing as a result. The intuition behind this ought to be very simple: Mr. Obama is maintaining leads in the polls in Ohio and other states that are sufficient for him to win 270 electoral votes.

Friday featured a large volume of swing state polling, including three polls of Ohio, each of which showed Mr. Obama ahead by margins ranging from two to four percentage points.

Between Ohio and the other battleground states, Mr. Obama held leads in 11 polls on Friday, against four leads for Mitt Romney' s and two ties. Mr. Romney's leads came in North Carolina and Florida, two states where the FiveThirtyEight forecast already had him favored.

To the extent that there was a trend in the state polls, it was slightly favorable for Mr. Obama. Among the eight polls that had previously published numbers after the first presidential debate in Denver, Mr. Obama gained about one percentage point, on average.

Mr. Romney made gains in four of the five polls that had last surveyed the race before Denver. Nevertheless, his average gain in the polls â€" 2.4 percentage points â€" was less than the 4-point bounce he was seeing in the immediate aftermath of the Denver debate. This suggests that Mr. Romney's bounce has receded some since his post-Denver peak.

The national polls out on Friday were not terribly newsworthy. Mr. Obama had a miniscule lead of 0.2 points between the eight national tracking polls that were published, reversing an equal ly small 0.2-point advantage for Mr. Romney in the same surveys on Thursday.

You can see here my thoughts on reconciling the differences between state and national polls. They may be reflective of a potential split outcome between the popular vote and the Electoral College, but there are other plausible hypotheses as well. Specifically, it could be that the national polls slightly underrate Mr. Obama's position, that the state polls slightly overrate it, or both.

The FiveThirtyEight forecast has Mr. Obama leading the popular vote along with the Electoral College, because it uses both state and national polls to calibrate its estimate of where the vote stands. Also, however, Mr. Obama's s tate polls were adjusted slightly downward because his national polls remain middling.

Still, our state-by-state forecasts are extremely similar to those issued by our competitors. For example, we had Mr. Obama projected to win Ohio by 2.4 percentage points as of Friday. That compares to a 2.3 percentage-point lead for Mr. Obama in the Real Clear Politics average of Ohio polls, a 2.9-point advantage for him in the Huffington Post Pollster model, and a 2.7-point edge for him according to Talk Points Memo's Poll Tracker.

How often does a lead of two or three points in the polling average, with 10 days to go until the election, translate into a victory in the state?

This is the sort of question that the FiveThirtyEight forecast is designed to address. But a simpler method is to just look at what happened when candidates held similar advantages in the past.

In the table that follows, I have attempted to recreate a simple polling average for competitive st ates in past elections, using about the same rules that Real Clear Politics applies.

In particular, I've looked at all states in our database in which there were at least three distinct polling firms that conducted surveys in the window between 10 days and three weeks before the election. Like Real Clear Politics, I used only the most recent poll (the one closest to the 10-day cutoff) if the polling firm surveyed the state multiple times during this period. I used the version of the poll among likely voters if it was available, defaulting to registered voter numbers otherwise.

In the table, I've listed all cases in which the race was within the single digits in the polling average. If you focus on those cases where a candidate held a lead of two to three percentage points, he won the state in all six out of six cases, although the sample size was small.

Historically, this two- to three-point range has been something of an inflection point. Poll leads of 1.5 percentage points or less have been very tenuous and have not conveyed much advantage.

On the other hand, there was not a single instance in the database where a candidate lost a state when he held a lead of more than 3.5 points in the polling average at this point in time. (Bill Clinton, in 1992, lost Texas despite leading George H.W. Bush there by that margin.)

It is possible to generalize these findings by means of a probit regression model, where the independent variable is the candidate's lead in the polling average and the dependent one is whether he won or lost the state.

That analysis implies that a lead of 2.4 percent in the polling average (Mr. Obama's current edge in Ohio in the FiveThirtyEight model) would translate to a win in the state 82 percent of the time. This percentage is similar to, but slightly higher than, the FiveThirtyEight forecast, which gave Mr. Obama a 76 percent chance of winning Ohio as of Friday.

It is important to emphasize that this analysis covers cases in which there were at least three distinct polling firms active in a state; you will find more frequent misses in cases where there were just one or two polls.

In Ohio, however, there are not just three polls: roughly a dozen polling firms, rather, have surveyed the state over the past 10 days.

There are no precedents in the database for a candidate losing with a two- or th ree-point lead in a state when the polling volume was that rich.

Instead, the biggest upsets in states with at least five polls in the average came in 2000, when George W. Bush beat Al Gore in Florida, and in 2008, when John McCain beat Mr. Obama in Missouri. Mr. Obama and Mr. Gore had held leads of 1.3 percentage points in the polling averages of those states.

If you look at the actual track record of state polling averages, it may even seem as though the FiveThirtyEight forecast is being conservative in giving Mr. Obama “only” a 76 percent chance of winning Ohio. I do not necessarily think that is the case.

The state-by-state polling averages have performed very well in recent years, but that is not likely to have been the case in, for example, 1980, when Ronald Reagan substantially beat his polls on Election Day. Years like 1980 are not represented very well in the tables above, because there were few states with rich polling that year. But they are considered by the FiveThirtyEight model, which calibrates its estimates of uncertainty based on the performance of state and national polls dating back to 1968.

Still, it is misinformed to refer to Ohio as a toss-up. Mr. Obama is the favorite there, and because of Ohio's central position in the Electoral College, he is therefore the overall favorite in the election.

Oct. 25: The State of the States

Thursday was a busy day for the polls, with some bright spots for each candidate. But it made clear that Barack Obama maintains a narrow lead in the polling averages in states that would get him to 270 electoral votes. Mr. Obama also remains roughly tied in the polls in two other states, Colorado and Virginia, that could serve as second lines of defense for him if he were to lose a state like Ohio.

The day featured the release of 10 national polls, but there was little in the way of a consistent pattern in them. On average, the polls showed a tied race. Furthermore, among the nine polls that provided for a comparison to another poll conducted after the first presidential debate in Denver, the net result was unchanged, on average, with Mr. Obama gaining one percentage point or more in three polls, but Mr. Romney doing so in three others.

Mr. Obama held the lead in nine polls of battleground states on Thursday, as compared to three leads for Mr. Romney and two polls showing a tied race.

This tally exaggerates the lopsidedness of the polling a bit, since the state polls released on Thursday were something of a Democratic-leaning bunch, some of which had shown strong numbers for Mr. Obama previously.

Mr. Romney's strongest number came in a Fox News poll of Virginia, which had him 2 points ahead there â€" a sha rp reversal from a 7-point advantage there for Mr. Obama before the Denver debate. However, Mr. Romney's worst poll of the day was probably also in Virginia, where Public Policy Polling showed Mr. Obama's lead expanding to 5 points from 2.

Among the 10 polls that provided for a comparison to another poll conducted after the Denver debate, Mr. Obama gained 1 percentage point, on average. The past week of polling suggests that Mr. Romney is no longer improving his position in the race.

Whether Mr. Obama has any momentum of his own, such as because of this week's debate in New York, is less clear. To me, it looks more like a gradual reversion to the mean than anything all that assertive.

At the same time, Mr. Obama has led in the polling averages all year in states that would allow him to win the Electoral College, and that remains the case now.

In the chart below, I've summarized the current FiveThirtyEight forecasts in a rather comprehensive list of s tates in which each candidate has at least a 1 percent chance of winning, according to the forecast. The chart also lists the most likely range of popular vote outcomes in each state, enough to cover 90 percent of all possible outcomes.

There is more uncertainty about the outcome in some states â€" not just because some are closer than others, but also because of a number of other factors that the FiveThirtyEight forecast accounts for in formulating its probabilistic estimates of the potential range of outcomes in each state.

Our research suggests, as intuition might dictate, that the outcome in a state is more certain when there is a higher volume of recent polling there.

In addition, the outcome is more certain when the polls are more consistent with one another. If a candidate leads by almost exactly four points in every poll of a state, that is a more reliable advantage than in a case where some polls have the candidate up by eight points, but others sh ow a tied race â€" even if these disparate polls show a 4-point lead for him on average.

Finally, some states are more “elastic” than others, meaning that they contain more swing voters. New Hampshire, for instance, is notorious for unreliable polling and for voters making up their mind at the last minute. This is probably not just a coincidence; New Hampshire has a disproportionate number of independent voters, and their preferences are more fickle than those of strong partisans. Thus, holding a small lead in the polling average in New Hampshire will not translate into victory as reliably as in another state like Pennsylvania, which has fewer swing voters and where elections are usually come down to a contest to turn out the respective party bases. The FiveThirtyEight forecast accounts for these properties.

These details aside, it is possible to place the states in to several broad groups.

First come a set of blue-leaning states â€" Oregon, New Mexico, Minnesota and Michigan â€" which might theoretically have been competitive but where the campaigns have not spent very many resources. There is little reason, at this point, to expect them to play much of a role in the math on Election Night. If Mr. Romney wins them, or comes within a point or two of doing so, it will probably indicate that he is overperforming his polls across the board and is headed to a clear national victory.

Pennsylvania is somewhat more competitive, but Mr. Obama leads there by about 5 points in the forecast (that was also his margin in a Rasmussen Reports poll of the state on Thursday). With the exception of one strongly G.O.P.-leaning firm, Susquehanna Research, no other polling firm has shown Mr. Romney ahead in the state all year. A 5-point lead in the state with a week and a half to go should translate into a victory for Mr. Obama more than 90 percent of the time, especially in a low-elasticity state like Pennsylvania.

Wisconsin and Nevada come next. While both remain winnable for Mr. Romney, they have featured among the more consistent polling; Mr. Romney has led in just one poll of Nevada since the Denver debate, and none in Wisconsin.

If Mr. Obama wins Wisconsin and Nevada along with the states like Michigan where he seems to have a clearer advantage, he will have 253 electoral votes, putting him 17 votes shy of clinching an Electoral College majority.

Ohio, which has 18 electoral votes, is the state most likely to provide those votes to Mr. Obama. He leads in the FiveThirtyEight forecast by 2.3 percentage points in Ohio, and by a similar margin according to other Web sites that aggregate polls. The forecast gives Mr . Obama about a 75 percent chance of winning Ohio. This figure is, not coincidentally, close to Mr. Obama's 73 percent overall chance of winning the Electoral College. (Ohio has about a 50-50 chance of providing the decisive Electoral College votes.)

New Hampshire and Iowa have featured less consistent polling than Wisconsin, Nevada or Ohio, and both are high-elasticity states that provide less overall predictability. Mr. Obama has about a two-in-three chance of winning each one, according to the forecast. However, these states alone would not suffice for Mr. Romney to win the Electoral College if he also lost Ohio, Nevada and Wisconsin.

Colorado and Virginia appear as though they might be the closest states in an election held today. Mr. Obama arguably has just the slightest edge in Colorado, where three of four polls released on Thursday showed him ahead, and where a fourth showed a tied race â€" but those polls were a Democratic-leaning group, so it is probabl y best to view the state as a tossup. The outcome in Virginia, where the polling has been inconsistent all year and was so again on Thursday, is anybody's guess.

However, the fact that Colorado and Virginia have been the closest states in the polling recently, and that both are fairly essential to Mr. Romney's path to victory while being more superfluous for Mr. Obama, is evidence that Mr. Obama has an overall advantage in the Electoral College.

The forecast model continues to give a slight edge to Mr. Romney in Florida. There, in contrast to several other swing states, it has been the more methodologically reliable polls that have tended to show a clearer advantage for Mr. Romney. Florida is by no means a sure thing for Mr. Romney - Mr. Obama's chances of winning it (35 percent) are larger than Mr. Romney's chances of carrying Ohio (25 percent), according to the forecast. But the polls in Florida have historically done a good job of predicting the result, and it is unlikely to leapfrog several other states and be the Electoral College tipping point on Election Day.

All of this holds doubly for North Carolina, where Mr. Romney leads by about 3 percentage points in the forecast and has about an 80 percent chance of winning.

Beyond North Carolina, there aren't very many states that Mr. Obama has a realistic chance to win, even if he is having a strong night overall on Nov. 6. Arizona probably provides Mr. Obama his best hope, but the forecast still puts his chances there at only about 3 percent.

It's important to keep in mind that the potential errors in the polls between different states are partly correlated with one another. That is, if Mr. Romney overperforms his polls on Election Day in a state like Ohio, he is also somewhat more likely to do so in other states like Iowa, especially if they are demographically similar.

The FiveThirtyEight forecast accounts for this property in its overall assessment of the Electoral College, and it is one reason why our forecast gives Mr. Romney slightly better Electoral College chances than other forecast models that might assume more independence in the state polling. However, we may be approaching the point where the state polls will have to be systematically biased toward Mr. Obama in order for Mr. Romney to have strong chances of prevailing on Election Day.

Auto Rescue and Low Home-State Bonus Keep Michigan Out of Play

We continue our Presidential Geography series, a one-by-one examination of each state's political landscape and how it is changing. Here is Michigan, the Wolverine State. FiveThirtyEight spoke with Bill Ballenger, the editor and publisher of Inside Michigan Politics; and Matt Grossman, an assistant professor of political science at Michigan State University.

The government's rescue of the American automobile industry appears to have given President Obama a slight boost in Ohio, whose 18 electoral votes may very well decide the nation's next president. Even at times when the presidential race has been a dead heat nationally, Mr. Obama has retained a consistent lead in Ohio , a state that has usually been slightly Republican-leaning relative to the country as a whole.

But in Michigan - the home of the auto industry - the political effects of the rescue of General Motors and Chrysler have received less attention. Mr. Obama carried Michigan by 16 percentage points in 2008, suggesting the state's 16 electoral votes were out of reach for Mitt Romney, rescue or no rescue.

There are a few problems with this logic, however. First, Mr. Obama's margin in 2008 was somewhat inflated. Senator John McCain essentially conceded the state, pulling his campaign out of Michigan about a month before the election. Polls showed Mr. Obama's lead quickly ballooning. Had Mr. McCain contested Michigan through the final month, the 2008 margin might have been closer.

Mr. Romney was also born in Detroit and raised in suburban Bloomfield Hills. His father, George Romney, was governor of Michigan from 1963 to 1969. Presidential candidates have historically received an average bonus of roughly seven percentage points in their home state. In fact, the last native Michigander to run for president, Gerald Ford, received that exactly, winning Michigan in 1976 by five percentage points while losing the national popular vote by two.

Nonetheless, the Romney campaign doesn't seem to be seriously contesting Michigan. In the last 30 days, Ann Romney and Representative Paul D. Ryan have each visited the state once, but neither Mr. Romney nor Mr. Obama has campaigned there.

In a campaign with a native son on the ballot, “Michigan has lost its battleground status,” Mr. Ballenger said.

Can Mr. Obama thank the auto rescue for keeping Michigan out of play? Partly, but Mr. Romney is also unlikely to get a big home-state bonus, local analysts said.

Click to enlarge.

Republicans carried Michigan in five consecutive presidential elections from 1972 through 1988. But Michigan wasn't a truly Republican-leaning state; it was a tipping point state. It was reliably Republican because the nation was reliably Republican. During that period the G.O.P. won the White House every year but 1976, and Michigan went red then anyway because of Ford.

As cultural conservatism gained sway in the Republican Party nationally, however, socially moderate voters in suburban Michigan began favoring Democrats. This realignment was not as pronounced as it was in New England or the rest of the Northeast; Michigan has a strong anti-abortion movement. But the Wolverine State was relatively balanced politically - “It was the old Ohio,” Mr. Ballenger said - and a small shift toward the De mocrats tipped the scales.

Democrats have carried Michigan in the last five presidential elections. The state's leftward lean increased from two percentage points in 1992, relative to the national popular vote, to nine points in 2008.

Detroit, though a shadow of its former size, is heavily African-American and overwhelmingly Democrat-leaning. About half of Mr. Obama's margin of victory in the state in 2008 came from Wayne County, where Detroit is. Democrats are also dominant in college towns like Ann Arbor and East Lansing; the state capital, Lansing; and smaller industrial cities like Flint.

The traditional base of Republican support is in the southwest, around Grand Rapids, where many voters are fiercely anti-abortion, influenced by the socially conservative Reformed Church in America (formerly the Dutch Reformed Church).

North of Grand Rapids and Flint, the state is more sparsely populated. The northern part of the lower peninsula is solidly Repub lican. The Upper Peninsula is traditionally blue-collar Democratic, but socially conservative and wary of gun control and environmental regulations, according to The Almanac of American Politics.

The two main political battlegrounds in Michigan are just north of Detroit: Oakland and Macomb Counties.

Macomb County is the birthplace of the “Reagan Democrats,” the socially conservative, white, working-class Democrats who voted for Ronald Reagan. Even after Reagan, Macomb County has been willing to vote Republican. George W. Bush carried it in 2004.

Macomb County is not as blue-collar as it once was, but it still has a lot of workers in the auto and related industries. Had th ere been no auto industry rescue, Macomb County voters - their connection to the Democratic Party already weak - might have moved even further to the Republican ticket. In fact, Mr. Romney may still carry the county; Mr. Obama won it with 53 percent in 2008.

The Bellwether: Oakland County

The other main swing county in Michigan is Oakland County, where Mr. Romney grew up. Oakland County has gone from a Republican bastion to a battleground as socially moderate voters in bustling suburban towns like Farmington Hills and West Bloomfield have trended Democratic. This area has also grown increasingly diverse, as middle-class African-American families have moved there from Detroit.

The partisan shift in Oakland County has made it an almost perfect bellwether for the statewide vote. It was one percentage point more Republican than the state in 2008 and roughly two points more Republican-leaning in 2004 and 2000. If Mr. Romney is having a really good night, and the Michigan vote is unexpectedly close, it will be apparent in Oakland County and Macomb County (itself a decent bellwether), which together accounted for almost a quarter of the vote in 2008.

The Bottom Line

Mr. Obama is a 97 percent favorite to carry Michigan, according to the current FiveThirtyEight forecast.

It is impossible to know what would have happened if General Motors and Chrysler had not been rescued, but Michigan's economy might have been in far worse shape than it is now. The state's unemployment rate is relatively high, at 9.3 percent, but it has dropped precipitously from a peak of more than 14 percent.

In addition, Mr. Romney doesn't appear to be getting a substantial home-state benefit, both Mr. Ballenger and Mr. Grossman said.

“I don't think many voters consider Romney a home-state candidate, especially not compared to Ford,” Mr. Grossman said.

Ford represented Michigan in Congress for more than two decades. Mr. Romney , a former governor of Massachusetts, spent most of his adult life living outside of Michigan and was never elected to any political office in the state.

Mr. Romney didn't appear to receive much of a home-state bonus in Michigan's 2012 Republican primary. He squeaked by Rick Santorum by just three percentage points.

The FiveThirtyEight model currently projects Mr. Obama to carry Michigan by seven percentage points and the nation by 1.5 points. That would make Michigan 5.5 percentage points Democratic-leaning relative to the national average, almost exactly where it was from 1996 to 2004.

It is possible the boost to Mr. Obama from the auto rescue and Mr. Romney's home-state bonus are canceling one another out. Or, perhaps neither is having a substantial effect.

“Michigan has been slightly to the left of center for a while now,” Mr. Grossman said, “and it doesn't seem to be moving a whole lot, in my view.”

A Windows 8 Cheat Sheet

October 25, 2012, 3:08 pm

In my New York Times column on Thursday, I pointed out that Microsoft's new Windows 8 feels like two operating systems in one. There's the traditional desktop Windows, best for mouse and keyboard, and the new TileWorld (as I call it), best for touch screens.

Why “best”? Because desktop Windows has tiny buttons, menus and controls that are generally too small for finger manipulation, and TileWorld is filled with gestures that make sense only on touch screens.

If you install Windows 8, you'll have to learn both environments, like it or not; you can't live in just one environment or the other. So the question arises: how are you supposed to operate TileWorld if you have a nontouch computer?

Answer: There are mouse and keyboard equivalents for the touch gestures.

Surprisingly, you have to dig around a bit online to find out what they are. So here, for the benefit of Windows 8 adopters, is your centralized cheat sheet: all of the most important touch/mouse/keyboard shortcuts for Windows 8.

Open the Start screen. The colorful rectangular tiles that make up the new Start screen are easy to reach.

Touch screen: Swipe your finger into the screen from the right border to make the Charms panel appear (described next); tap Start.

Mouse: Point to the lower-left corner of the screen; when the Start screen icon appears, click.

Keyboard: Press the Windows key.

Many new tablets and laptops have a dedicated Windows-logo button under the screen. Pressing it also opens the Start screen.

Open the Charms panel. The Charms menu is a thin vertical panel of important icons like Search, Share, Start and Settings.

Touch screen: Swipe your finger into the screen from the right border.

Mouse: Point to the top right corner of the screen.

Keyboard: Press the Windows key+C.

You can also jump directly to one of the buttons on the Charms panel.

Share button: Windows+H

Settings button: Windows+I

Devices button: Windows + K

Open the App menu. Programs designed for TileWorld often have a few options, represented as icons in a hidden horizontal bar. In Internet Explorer, for example, this bar shows all your open browser tabs.

Touch screen: Swipe into the screen a short distance from the top or bottom of the screen.

Mouse: Right click anywhere in the window.

Keyboard: Press the Windows key+Z.

Next app. Here's how you jump from one TileWorld app to the next. (The Desktop, and all of its own programs, are represented as one jump.)

Touch screen: Swipe into the screen from the left border.

Mouse: Point to the upper left corner of the screen.

Keyboard: Press and release the Windows key+Tab.

App Switcher. In regular Windows, Alt+Tab (and hold down the Alt key) shows you a little dashboard displaying the icons of all open programs, so you can jump directly to the one you want.

Touch screen: Swipe into the screen from the left border, then back out again. A vertical column of open app icons appears.

Mouse: Point to the lower left corner of the screen.

Keyboard: Press Windows key+Tab, but keep the Windows key pressed.

Or press Alt+Tab (hold down Alt) as you always have. That brings up the traditional horizontal row of open-app icons. This app switcher includes open desktop apps (traditional Windows apps).

Split the screen between two apps. This feature made its debut with Windows 7; it lets you split your screen between two programs' windows.

Touch screen: Swipe your finger slowly into the screen from the left or right border. Or swipe down from th e top edge, then to right or left.

Mouse: Drag a window's title window to the left or right side of the screen until its outline changes to a full-height, half-width window. Release.

Keyboard: Press Windows key plus the left or right arrow key.

Close an app. Here's how to exit a program in TileWorld.

Touch screen: Swipe down from the middle of the top border, almost all the way down the screen.

Mouse: Point to the top of the window to make the grabber handle appear; drag it all the way down the screen.

Keyboard: Press Alt+F4.

Right click. In Windows, right clicking an item summons a shortcut menu - a short menu listing commands relevant only to the object you clicked. In Windows 8, that menu takes the form of a horizontal strip at the bottom of the Start screen, offering options like Uninstall and Unpin (from the Start screen).

Touch screen: Swipe down from a tile on the Start screen.

Mouse: Right click, of course.

Keyboard: Press the little menu key.

Zooming in or out. To magnify or shrink your view of a photo, map or Web page, proceed like this:

Touch screen: Spread or pinch two fingers on the screen, just as on an iPad.

Mouse: While pressing the Ctrl key, turn your mouse's scroll wheel.

Keyboard: Press Ctrl and the + or â€" key.

Zooming fully out. On the Start screen, you can zoom out so far that your tiles become little icons; in this mode, you can group them or move them en masse.

Touch screen: Pinch two fingers on the screen.

Mouse: While pressing the Ctrl key, turn your scroll wheel. Or point to the bottom right, and then click the Summary View icon that appears. (To zoom out again, click any blank area.)

Keyboard: Press Ctrl+the minus key.

Search for files or settings. The new TileWorld Search command requires that you specify what you're looking for: an app, a file and so on. But there are shortcuts for file searches and s ettings searches.

Touch screen: Swipe in from the right border; tap Search; tap Files or Settings.

Mouse: Point to the top right corner of the screen; tap Search; tap Files or Settings.

Keyboard: Press the Windows key+F for files, Windows key+Q for settings.

Search for apps. This one's really best with the keyboard: you can jump to an app on the Start screen, even if it's several horizontal scroll-pages away.

Touch screen: Swipe in from the right border; tap Search.

Mouse: Point to the top right corner of the screen; tap Search.

Keyboard: At the Start screen, just start typing.

External monitor/projector options. Do you want your main screen mirrored on the external screen, or extended onto it? You can open a handy panel that lists your options.

Touch screen: Swipe in from the right border; tap Devices; tap Second Screen.

Mouse: Point to the top right corner of the screen; tap Devices; tap Second Screen.

Keybo ard: Press Windows key+P.