Presidential Race ‘Tight’ in Meaningless National Polls [UPDATED]
A week out from the election, President Obama is a heavy favorite to win re-election. But the major press continues to pretend otherwise.
A week out from the election, President Obama is a heavy favorite to win re-election. But the major press continues to pretend otherwise. The New York Times headline “Obama and Romney in Exceedingly Close Race, Poll Finds” is but the latest example.
President Obama and Mitt Romney enter the closing week of the campaign in an exceedingly narrow race, according to the latest poll by The New York Times and CBS News, with more voters now viewing Mr. Romney as a stronger leader on the economy and Mr. Obama as a better guardian of the middle class.
The president is holding his coalition together with strong support from women and minority voters and is supported by 48 percent of likely voters nationwide, the poll found, while Mr. Romney holds a wide advantage among independents and men and is the choice of 47 percent.
The race for the White House, which has been interrupted by the aftermath of Hurricane Sandy’s deadly assault on the East Coast, is heading toward an uncertain conclusion. The president was set to stay off the campaign trail for a third straight day to tour storm damage on Wednesday with Gov. Chris Christie of New Jersey, a Republican. Mr. Romney was set to resume a full schedule in Florida.
In the final days, the most intense competition between Mr. Obama and Mr. Romney has narrowed to seven states, but the national poll illustrates why the Romney campaign is working to expand the battleground and seize upon the deep concern in the electorate about whether the president should win a second term.
Not a single person covering politics for the New York Times believes this nonsense. The national horserace numbers have zero bearing on the race; what matters is the disposition of a handful of swing states, with Ohio almost decisive among them. And not a single reputable poll has shown Romney ahead in Ohio this cycle.
It would be as if the New York Times sports page reported on last Sunday’s game between the New York Giants and Dallas Cowboys and focused on the fact that the Cowboys had 28 1st downs to the Giants’ 11, dominated total yards 434 to 293, and were two-and-a-half times more efficient on 3rd down and neglected to point out that the Giants scored 29 points to the Cowboys 24. To be sure, all of those statistics are interesting in understanding the game. But the object of the contest, as Herm Edwards reminded us a decade ago, is to win the game. And you do that by outscoring your opponent, not by amassing more total yardage.
Amusingly, WaPo both acknowledges this yet ignores it in reporting on their latest poll:
Most likely voters — including both Democrats and Republicans — say the winner of the presidential election should be decided by popular vote, not the electoral college, according to the latest release of the Washington Post-ABC News tracking poll.
In the new national poll, 49 percent of likely voters back Republican Mitt Romney, while 48 percent support President Obama. It’s the fifth straight day that a single percentage point (or less) has separated the two candidates.
The presidency, of course, will be decided at the state level — with the overwhelming focus on a small set of swing states where polls also point to an extraordinarily competitive contest next week. But most voters wish that weren’t so: 56 percent of all likely voters say they would prefer the one who gets the most votes across the country to be the next president; 37 percent would want the one with more electoral votes to prevail.
But we’re not going to have the election people wish we were having. So why bother at this late stage of the race with a survey based on a non-existent contest to see who would get more votes if we put them all in a single bucket?
There are enough state polls out there showing close races in key states—including states that one would think would be locked up by now—to give Republicans some hope of pulling off a victory and to make Democrats worry. And, as I noted in a comment thread yesterday, it’s possible that polling as we know it doesn’t work anymore in an era of cell phones, call screening, and frustration over constant interruptions at home. But, if we’re going to assume polling is accurate for the purposes of the story—and it’s perfectly reasonable to do that until we have a 1948-like failure—we shouldn’t pretend that they state-level polls are all wrong.
UPDATE: To further drive home my point, here’s today’s PollTracker electoral map at TPM:
With four toss-up states, Obama is leading 264 to 206. The toss-up states are Wisconsin, Michigan, Virginia, and Florida. Obama has a slight lead in all four of them. If Obama takes only Wisconsin, he gets to the magic 270 threshold. If he takes only Michigan, he gets to 276. If he takes Wisconsin and Michigan and Romney takes Ohio—currently in Obama’s column—he gets 272.
The latest AP analysis (29 October) at Time’s Swampland Electoral Calculator has Obama with 271 Electors to Romney’s 206 on a map that doesn’t assign Florida, Virginia, Colorado, Nevada, or New Hampshire. If Romney were to somehow sweep all of those, he’d finish with 269. And nobody seriously thinks Nevada is in play, do they?
RealClearPolitics currently has it at 281 to 257 for Obama. And that’s giving Romney Florida, Virginia, and Colorado. For Romney to win, he’d need to take 12 electoral votes out of Obama’s column, with Ohio being the most probable scenario for doing that. Of the 9 polls on Ohio in RCP’s average, only Republican-biased Rasmussen has Romney ahead. And even Rasmussen only has Romney up by 2.
Similarly, Scott Elliot’s Election Projection projects Obama getting 290 Electoral Votes to Romney’s 248.
And, yes, Nate Silver’s FiveThirtyEight has it 299 for Obama to 239 for Romney. That’s with Obama taking Ohio, Virginia, and Colorado along with Ohio.
At some point before the election, I’ll put up my Electoral College prediction. Right now, I’d give Romney Florida and probably Virginia but would give Obama Ohio and Colorado. That would be a comfortable 290 to 248 win. Shifting Ohio to Romney’s column takes it to a slim 272 to 266 win. So, Romney would need to also take either New Hampshire or Colorado.
Could any of those scenarios play out? They could! But they seem damn unlikely to me less than a week away. And with a lot of votes already banked.
Grifters gotta grift.
I do think the national numbers serve as a sanity check on some state polls. A tied national race means that Obama is not ahead by 5-6 in Ohio (he’s definitely ahead, but not by that margin) and that Romney is not on the verge of taking Michigan, Pennsylvania, or Minnesota–three states Bush lost while winning a national majority while also winning neighboring Iowa and nearly winning Wisconsin. And Bush was actually contesting those states.
But, if the narrative reporters went by numbers instead of drama, who would read them?
Grifter respect…
🙂
Next Wednesday will be quite interesting to behold. And by “interesting” I mean in the train wreck sense.
@Geek, Esq.:
This doesn’t logically follow. A ‘tied national race’ depends entirely on the location and composition of the sample.
A ‘national race,’ polled by Rasmussen, that doesn’t contact cell phone only households and weights ever-so-slightly heavier in the South, could very well show a tie, whereas state polls in places like OH and VA could show a candidate to be well ahead.
Mixing apples and oranges.
I think the press likes to whipsaw their readers, alternating the electoral story with the popular vote one. It makes more of a horse race. Seemingly the story changes, again and again.
Mitt Romney clearly cares about his DVOA.
@Cycloptichorn:
I ignore Gallup and Rasmussen when forming opinions about the race.
Other legit polls–CBS/NYTimes, ABC/WaPo, Pew, all show a dead heat nationally.
Ohio has traditionally run about 0-3 points more Republican than the general electorate. I can buy it being 2-3 points more Democratic than nationally due to the superior Obama campaign there, but not 5.
Similarly, if the national race is virtually tied, no way is Romney going to win PA or MI or MN.
The important question, of course, is that if Obama wins next week, with or without a popular vote majority, how will the mainstream media and our political establishment respond to the total freak-out on the right. Everybody from Tim Russert on down fell in line behind Bush the Younger in 2000, but will Fox do so this time around?
Mike
Just once I’d like to see an honest headline, such as: Romney continues to run up the score in the Confederacy but it isn’t going to help him any in the EC.
@MBunge:
Mike, why would they? They didn’t in 2008.
The Princeton Election Consortium has it as between 95 and 99% likely to be Obama…and there are no likely scenarios for Romney to win. They said a week ago that the cake is baked.
Nate Silver is close to the same.
Meanwhile…the President of GM is charachterizing the claims Romney is making, in an attempt to take Ohio, as being from an alternative universe…and Romney is posing fake relief efforts…while Chris Christie is praising Obama’s response. Then you have an employment number coming out on Friday…which is predicted to be just a bit stronger than last months.
But yeah…it’s close.
We elected G. W. Bush twice. We might as well elect a weak imitation with better hair.
Romney’s got something like a 1 in 3 chance of winning. Which, as a number of people have recently pointed out, is a pretty decent chance. He absolutely could win. He’s just less likely to win than Obama is.
Neither result should be a big surprise. What will be interesting, from the point of view of the modelers and pollsters, is how the actual results mesh with the pre-election polling, particularly in the swing states.
@Rob in CT: “Romney’s got something like a 1 in 3 chance of winning.”
What is this assertion based on?
Couple thoughts:
1. While the swing state numbers are close, the key thing is that most, in particularly Ohio, have also been relatively stable across multiple polling cycles. While there might have been some tightening, the lack of Romney leads means that things like “margin of error” can be ignored.
That’s not a good thing for Romney considering that Obama has a clear path to victory even if Obama loses the three more volatle swing states (Florida, Colorado, and Florida).
2. Even conservative leaning meta-pollster’s formula’s continue to give Obama a commanding lead in the EC. For example there’s http://www.electionprojection.com. While the proprietor has been predicting a Romney will win (and that, at any moment, state level polls will start pointing to that), his own algorithm continues to show Obama with a commanding lead in the EC.
3. As I wrote earlier this week, the split between popular vote and swing state votes might attributed to the targeted marketing approach of this campaign. Outside of swing states, neither campaign is making any real appeal to independent voters. So it’s not surprising, given their make up, that outside of swing states, that “independents” seem to be breaking for Romney.
4. The real surprise of Karen is that it may contribute to a Popular Vote/Electoral Vote split. Any drop in turn out due to the storm won’t be enough to flip the (solidly blue) affected states. Nor do I think, like some liberals, that Obama’s response (or Romney’s past talk of cutting FEMA) will change many votes. But the disaster condition left in Karen’s wake could create enough of a reduction to give Romney the edge in the popular vote.
And, personally, I’d love to see a Popular Vote/Electoral Vote split this year (provided Obama wins the EV) — count me on team “Taylor”, I think Electoral College reform is long overdue, and a Republican Win/Loss might make all the difference in getting it done.
Duh, I keep writing “Karen” when I meant “Sandy.”
@Me Me Me:
Intrade odds give Romney about one in three; Silver gives him about one in four; Princeton gives him about one in twenty. But as people keep saying, if somebody told you that the airplane you were getting on had a one in twenty chance of crashing, you probably wouldn’t fly today. So if you live in a swing state, make sure to vote.
Good update James. It demonstrates the first point I was making above.
Since you’ve given you’re prediction, here’s mine: I’m going with the current RCP no toss-up map you posted: 281 to 257 for Obama (giving Romney Florida, Virginia, and Colorado). Obama might get one of those three states, but given recent polling cycles and looking at the electoral history of those states, I don’t think he’ll get two or all three of them.
I expect Obama to squeak out a popular vote win, with less than a 2% gap.
I am going for Obama to end up winning 332-206 (i.e., he gets all 4 of those states marked toss-up). That has always been the most likely outcome in Nate Silver’s maps since he started this year’s projections, and currently it’s the result in about 16% of his runs (or roughly 2/3 of the chance he has for a Romney win of any size).
I seem to remember the Republican primaries were obviously over about three months before the media would admit they were over. The general has been over since sometime in September.
@gVOR08: The conventions and debates legitimately gave Romney a chance to make up ground. But he totally botched the convention while Obama nailed his. Other than bringing his A game for the first debate while Obama phoned it in, he really hasn’t done anything to help himself.
@Me Me Me:
538 model, which I didn’t check for the most up-to-date numbers. Maybe it’s 1 in 4. Which is still means him winning shouldn’t be some shocking result.
I’ve a contrary question for both you, James, and Doug. If Romney wins, and wins big, will you finally start to question the sources of (election and) news you have been relying upon for months (more likely years)?
@bains: I’d question the viability of the polling model that’s proven astonishingly accurate over the last several decades but that has some rather obvious flaws in the era of Caller ID, cell-only houses, etc. But the polls are all we have to go on right now; all else is anecdote and spin.
@James Joyner:
The first debate definitely helped Romney, giving some umph to a surge that appears to have begun the week before.
Unfortunately for him, Romney started that slide from his lowest point — a combination of a botched convention and the reaction to the release of the 47% tape. And to your point, his campaign does not seem to have capitalized on the momentum where it counted (i.e. the swing states).
All that said, one has to wonder the condition Obama will be in if he wins this. If Romney wins the popular vote (due in part to the fact that most of the country has been largely ignored by the two campaigns and most of the super pacs) it’s going to be another difficult four years for Obama as the Republicans (and in particular the Senate) will probably take that as a sign that they weren’t obstructionist enough to fully sink his presidency.
@bains:
If Romney wins, and wins big, will you finally start to question the sources of (election and) news you have been relying upon for months (more likely years)?
Finally? What reason has there been to think that polling cannot be a useful predictor of the election outcomes (or, more accurately, the present state of the race)?
If the only reason is you don’t like what the polls say, that’s no reason at all. Various models averaging polling data predicted the last presidential election with a high degree of accuracy.
As James said, we’ll know the problems we suspect are growing with the polling models are real when those models fail. They haven’t failed yet, and certainly haven’t been replaced by anything more accurate. You act like they’ve been failing for years and the rest of us have been pretending otherwise.
Exactly. If you look at the trend lines for the EV (the only thing that matters) on both the Princeton site and at 538 you’ll see that Obama was consistently trending up and Romney consistently down…right up until the Denver debate…Romney then got a signifigant boost…but after that shake up (and just before the Biden/Ryan debate) the trends reverted to the pre-Denver trajectory and have been steadily in Obama’s favor since.
Which is not to say that the stupidity of the American Electorate will not rear it’s ugly head on Election Day. I personally would not buy a used car from Romney…but much of America is dumb enough to.
@James Joyner: I should have qualified my statement to – The general has been over since September barring completely unforeseeable occurrences like Obama being caught over a dead body with a smoking gun, a hurricane in NY, or Romney winning three debates against Barack Obama. (And I suspect the hurricane helps Obama. The GOPs will find or fabricate some scandal in the federal hurricane response, but likely not by next Tuesday.)
@ C. Clavin…
Also it’s imprtant to note that Obama has never…never…been behind. Not even in the RCP Average…which is borderline worthless.
@bains:
If Obama wins with an electoral college vote in the 280 – 300 range, which most sane prognosticators are predicting, will you finally start to question the sources of (election and) news you have been relying upon for months (more likely years)?
That is all fine and dandy James… but all the polls you have promoted, and all of the analysis of said polls come from… partisan sources. NPR, Time, NYTimes, TPM… come on, are you trying to etch out of John Cole market?
@bains:
How about the recent newsmax/zogby poll that also shows Obama leading in Ohio? http://www.newsmax.com/Newsfront/zogby-ohio-romney-obama/2012/10/31/id/462197
Or is newmax now a leftist site because it delivers bad news.
And did you know that a recent NPR poll had Romney ahead nationally?
Oh and how do you account for the fact that Scott’s – a self identified conservative republican – algorythmic analysis of the polls continue to show Obama the winner. One would expect he’d discount the liberally weighted polls.
Tell me, is it the duty of republican and conservative bloggers to go all Limbaugh and support thier candidate (in the face of contradictory evidence) no matter what (even if it means lying thier audience)?
@bains:
NPR, Time and NYTimes are all non-partisan. In no way can they remotely be considered partisan news sources by anyone with a firm grip on reality.
TPM is left leaning, but their poll-tracker accepts almost all publicly released polls. They do exclude internet (Zogby) polls but don’t exclude R-leaning polls as a matter of policy.
Romney = 330
That pronouncement brings a big “ha ha ha idiot” from the audience.
Look at that again on Wednesday, November 7.