A final unskewing
Mother Nature is our best teacher and the only one who is always right. – Viktor Hamburger, biologist In yesterday’s Los Angeles Times...
Brief notes
In case you missed it, Rick in Miami provided a Senate update, showing the likely seats at 53-55 D/I – from no change to a gain of two seats. Check it out.
On the House side, here’s a guide for swing districts near you for GOTV.
Today at 2:00pm ET, I’ll be on KCRW-FM in Los Angeles with Mark McKinnon, Sasha Issenberg, and Ron Brownstein. A few other media events here.
Open thread while I gather my wits. The hours are ticking down!
Topics:
Sam,
About what time tomorrow (I assume tomorrow) should we expect the ‘final’ prediction?
I keep asking but no one answers- IS there a final prediction up top before the polls [election booths] close? UNtil a winner is called to PEC’s satisfaction or Mitt gives up [as IF!] does the EV estimator keep putting out a black line of poll-based EV estimates, right past the dashed line?
I have been to the archives but I can’t tell what the header and charts looked line on each day, even from the comments.
Please? I _will_ stick around, but what do we know about how PEC changes? Does something stop tonight? OR what?
@Ms. Jay Sheckley:
Yes. He does do a final prediction. He does some kind of smoothing, so that he doesn’t get caught by the last statistical jitter, and sets that as his final prediction. Last year it was early early morning on election day.
What I don’t know is, if his median and his mode are different, which will he say is his prediction — the likeliest single outcome, or the one that is much less likely to be spot-on, but is in the middle of his range?
SUSA gives McCaskill a 15-point advantage in MO today. Clearly an outlier, but pretty remarkable, all the same.
Isn’t that just the sort of thing we at PEC ignore? Give us a reason to respect these figures.
Sam,
Nate Silver has said that for Romney to win there has to be serious error in the state polling.
I believe there probably is, and Romney will pull this out.
I hope you are right and I am wrong.
In any case…as the election comes to a head I want to thank you for being a steady mooring in the storm.
What has led you to believe there is systematic state polling bias? Not confronting you, just hope we can help talk you back from the ledge!
@ Ohio voter…
Oh…there’s no rational basis for that belief…but this country voted for George W. Bush twice…consequently I have no faith in the electorate to do the smart thing.
Rafalca, you are a 15-year old Oldenburg Mare. You cant even count to ten, let alone do statistical analysis.
Then again, Clever Hans was able to do some pretty impressive things.
Also, Rafalca must be using an awfully large keyboard. Just a thought.
Let me save you the suspense. You are wrong.
Yes…a very large keyboard indeed.
Again…I hope PEC is spot-on…I just have an awful feeling in my gut that Romney’s BS will fool just enough voters.
We will see soon enough.
Well, “this country” did the unthinkable and elected a black man as their President in 2008.
I think Big O’s got this…
Otis, as Garry Trudeau phrased it in a memorable Doonesbury strip i 2008: “He’s half white, you know.” 😉
@ Otis…my man…
A valid point.
However Romney is a much accomplished grifter than McCain.
Frankly my concerns have no place in a PEC thread…but even paranoids have real enemies.
Rafalca, no man can fool the CLT.
Ohio Voter: Please look at the YouGov site. This is a conservative British organization. They give it to Obama.
Most Republican pundits are already making excuses for Romney’s loss. Get used to it.
Rafalca has a point. Go to RealClearPolitcs Video page and listen to conservative analysts call the election for Romney.
George Will, M. Barone, D. Morris, K.Rove, C.Krauthammer cain’t all be wrong can they?
http://www.realclearpolitics.com/video/
@ Piktor…
Well, usually, yes…they are all almost always wrong.
(which is surely your intended snark)
I hope I am too.
The credible arguments for flawed polling are here:
battlegroundwatch dott com
Ahhh. The equine large keyboard vote.
Sam just said on the radio PEC has that factored in.
[I have a friend who repairs those keyboards. This sure is the season! Pull out the hay, and see the damage an angry horseshoe can cause to a circuit board. He’ll be soldering all night, just to see the thing cracked to bits again. Damn shame how the “Clever Hans” crowd treat a keyboard!
Worse though was the guy who kept spilling his trough of Coke.]
Here’s more info on insider advantage FL poll that is torquing FL average. Looks like baby Mason Dixon poll to me. Can’t find link to full data but looks like they are using pre-determined view of electorate (like Mason Dixon) rather than just basic demos. They state Obama’s advantage on Hispanic FL vote as 54-45. Latino Decisions puts it at 69 – 31. Interesting that FL Times Union would publish this now… Here’s more details
http://members.jacksonville.com/news/premium/florida/2012-11-04/story/romney-has-5-point-lead-florida-times-unioninsider-advantage-poll-says
TIMES-UNION PRESIDENTIAL POLL
Overall
ROMNEY: 52%
OBAMA: 47%
Undecided: 1%
18-29
Obama: 67%
Romney: 32%
30-44
Obama: 40%
Romney: 59%
45-64
Obama: 44%
Romney: 54%
65+
Obama: 40%
Romney: 57%
With Florida’s much watched Hispanic population, Obama holds a 54-45 advantage. Even so, Towery said, Romney’s areas of strength will be tough to overcome for Obama.
The Times-Union/Insider Advantage poll was conducted of 437 likely Florida voters and has a margin of error of plus or minus 4.6 percentage points.
You will wake up wednesday freaking out.
There is no way Obama will not win by all the best predictors of the electoral vote.
I think Obama will win over 300 electoral votes over Wrongney…
Obama will win BIG!
You can ignore any poll showing Romney pulling 45% of the Hispanic vote. The poll screwed up.
In Florida? Not so much. Obama won FL Hispanics 57-42, and Bush won them both times.
I think that all FL polls are in trouble. For one thing, none have really looked at third party effects. For another, anger is a strong motivator in politics. About 1/2 of Romney’s voters are voting against Obama rather than for Romney, for example. GOTV is key in close races. Scott just made every Democrat and half the independents in FL angry.
There seems to be a lot more missing demographics than included ones in this poll
Where is the break down of numbers of White, black, hispanic? where is the breakdown by sex.
I see percentages by age group, but no numbers indicating total amounts polled in each age group. was it land line only.
towery is (was a hack for newt) the times union is a faaaaar right rag. they have endorsed romney .obama wont get a break from this outfit.
Am I correct in thinking that the east coast states to watch tomorrow night are NH and PA. If both are won by one candidate, then that candidate will win, and we need not stay up too late. At the moment, polls indicate NH is +2% for Obama and PA is +3% for Obama, and the national meta-margin is between them at 2.28% for Obama.
That is, if Romney wins PA that can mean that polls are skewed 3% or more in Obama’s favor, and thus the 2.28% meta-margin is insufficient for Obama to win. On the other hand, if Obama wins NH that can mean that polls are skewed less than 2% for Obama (or are unbiased or are skewed for Romney), and thus the 2.28% meta-margin is sufficient for Obama to win it all.
@Lee
Well, the margin is similar enough in NH (O+2), PA (O+3) and OH (O+3) that an Obama victory in both NH and PA does not guarantee an OH victory. I will certainly stay up late anyway 🙂
VA is the state I’ll be looking at early. Polls there close at 7:00 (an hour earlier than PA or NH), and if Obama wins there Romney is pretty much sunk.
OH polls close at 7:30, and apparently a number of experts think that the race there is important.
Watch PA, NH, and Virginia. If he gets all three, it’s probably going to be a good night.
(But if 2008 is any indication, it might take longer to read Virginia than Ohio.)
If Obama wins Florida, go to bed, it’s over. If he doesn’t, Ohio is still the key.
@Froggy: I remember Virginia taking a long time to call in 2008, because the returns from Fairfax County and environs were slow coming in. The early returns were all Republican, but everyone knew that wasn’t the whole story.
Until at least Ohio is called, there are still many paths for Romney. According to NYTimes’ 512 Paths to the White House (which does not consider PA to be swing), even if NH goes to Obama, there are still 31 ways for Romney to win. If VA goes blue, that reduces to 6, and if OH does as well, then Obama wins.
My expectation would be that several of the states would be in the ‘too close to call’ column for MSM, meaning we’ll be staying up late. Also, Jon Stewart and Colbert will be on live, and who wants to miss that?
It all depends on when they actually call the various states. I actually wouldn’t be surprised if Ohio was called not long after, or even before, Pennsylvania.
VA closes at 7:00, North Carolina and Ohio at 7:30. Those are the states that you should be watching first.
One problem is that large states don’t report results in a uniform manner on Election Night. For example, in 2008 I was upset to see Virginia coming in five points below my pre-election estimate. The next morning it all got sorted out.
I would watch New Hampshire, a small state that counts fast and is heavily polled. That would help anchor the whole polling dataset and tell us whether there is “unskewing” that needs to be done. I’ll write on that.
And you don’t need to wait until a state is called to know which way the wind is blowing. That’s especially true when looking at county-level data – for instance in Pennsylvania I always look to Bucks County first, and then Centre and Luzerne. I’ll take a look for Ohio and Virginia bellwethers (and NC I guess, though I don’t think Obama has a chance there).
@Craigo: In Virginia, it comes down to Northern VA (Fairfax, Loudon, Arlington, Alexandria and Prince William County), Richmond, Charlottesville, and Hampton Roads/Norfolk.
@ Suja P.
Those are indeed Democratic strongholds, but mostly not bellwethers. Obama won Arlington by nearly 3 to 1.
Looking down the list, it appears that Loudon, Manassas, and a few others were reasonably close. I haven’t checked for 2004 or the state elections yet.
NY Times has a very cool “Paths to the White House” aopplica
Check out the NY Time “paths to the White House” app. Very cool and fun.
@Craigo: If you’re looking for a bellweather, maybe some place like Virginia Beach, which was very close in the previous election. Manassas, Manassas Park and Loudon should’ve drifted more blue in the past four years, but the counties just beyond that might work for your purpose – Fauquier, Rappahannock, Clarke, Culpeper, etc.
Ultimately, turnout in the Democratic strongholds will matter, and hopefully, siphoning red votes to third party candidates will too.
They will matter quite a bit – Obama won’t win without them! – but you’re correct in that I’m looking for counties and cities that tend to fall along the state mean.
About NH… Does it have a large enough minority population to test the “polls are off because minorities won’t turn out” theories?
Good question. About 95% of voters in NH are white, so I don’t think so. No instant gratification for us.
Yeah, New Hampshire is a really, really white state, which used to be hardcore Republican and has become a purple/blue state as a southern piece of it transforms into a Boston suburb.
I’ve been wondering, actually, if that trend might be at or past its high-water mark now and about to recede. The real-estate crisis hit the Boston area early, and the end result of it seems to have been that a lot of commercial and residential development contracted closer in to the city core. There was some migration of business all the way out to 495 during the boom, but now the hot spot for tech industry is Cambridge and Boston. Those Democratic-voting greater Bostonians in Nashua and Hollis may move back to Massachusetts.
NH could actually be a test case for my counter-theory, that enthusiasm among richer, white cultural liberals has dropped off more than among minorities.
The KCRW show seems to be ‘To the Point” one of my very favorite shows ever. It is hosted by Warren Olney and airs at 1pm PST (not 1pm ET).
Anyone wanting to listen live to the show, I found it on the stream at KPBS – http://www.kpbs.org/
Sam’s not on quite yet, as of 1:20 pm cst. It’ll be over by 2pm CST. Check it out if you can.
the listed guests seem to be on now. Have I mussed Sam??
Typo. I lack the power to muss him. Am listening to Olney. He said.”big government”.
He’s on just now …
Been lurking with this site for sevreal weeks (more accurately – obsessed) now. Love the objective approach.
My hope is that your numbers nail the result. I would love to see an unimpeachavble validation of such a rigorous approach. Maybe in the long term it will induce change in the way election campaigns are executed. It’s hard to argue with success.
One nagging fear is that there is some systemic flaw in the sampling methods resulting in a skew towards Obama. Of course this could be systemic both for and against Obama as well as systemic across all the polls. But if the result falls outside the margins, the methodology will not look good in the media , even if it is sound.
The ground game of the two parties seems to be a confounding issue as well. Already the legal battles are starting, with the suits filed in Ohio, Florida shutting down early voting even with people waiting many hours, and rumors of electronic voting anomalies.
Thanks for all your hard work! The next 48 hours should be a very interesting.
Those arguing for a systematic anti-Romney bias in the polling base their claims on party identification (“oversampling of Democrats”). Strangely, they base their arguments on Gallup national polling, yet ignore this article from Gallup:
http://pollingmatters.gallup.com/2012/09/the-recurring-and-misleading-focus-on.html
–bks
p.s. I predict that the number of visitors to this site will drop dramatically on Wednesday.
What an amazing prediction. I don’t know how you arrived at that.
We are descending into the fog of election time, in the midst of which it will be easy to get frightened or enthused by ghosts and mirages. Most of these will dissipate in the light of day on Wednesday, unless the election is close enough to be seriously disputed.
Its the fog of battle, Matt.
But Pallas Athena, the goddess of metis (wisdom) will shield us with her penetrating gaze.
“Achilleus in amazement turned about, and straightaway / knew Pallas Athene and the terrible eyes shining.”
@ Sam Wang
I’m still here, El Guapo!
I think the anxiety has to do with races where the polling was off and a surprise happened – Honey Badger in Nevada in 2010, the Clinton primary win in New Hampshire in 2008 and to some extent Gore overperforming the national polls in 2000.
I do think, however, with the huge number of polls in so many different states and with so many different pollsters with different methodologies the likelihood of a surprising result is a lot smaller. In fact, I bet 2000 would have been less surprising had guys like Nate Silver and Dr Sam Wang had their website up and running and crunching the numbers in a systematic way. Also, there are a lot more state polls to chew over than there were prior to 2004.
@ wheeler’s cat Let’s hope Pallas Athena wasn’t holding Obama by the heel, eh?
@EL: Achilles never had secret service protection.
One thing I’ve been mulling lately is that there may be a surprise to the upside for the Democrats using Sam’s methodology; in fact I would argue that that is more likely than the other direction.
By taking all polls at face value, Sam’s model gives disproportionate weighting to the outliers that are at least tacitly admitted to be partisan efforts. I assume the thinking is that both sides are doing it, and so will cancel each other out, but I don’t necessarily agree. We’ve seen some very transparent attempts by the GOP to influence the momentum narrative, including through polling, with (at least to me) very little evidence of a corresponding level of effort by the Dems.
Possible?
I think the first wave of interest in electoral-vote poll aggregation in 2004 was largely caused by 2000. Everyone was deeply aware after that that it was the Electoral College that counted.
That whole fiasco was what gave us the notion of Red State America and Blue State America. Of course, the divisions had been there before, implicit in old-fashioned hippie-kicking and all of Reagan’s paeans to the heartland, but in 2000 it got incredibly stark.
The unofficial color coding of the US parties we all know today didn’t even exist until the long twilight struggle that was the aftermath of Election Day 2000, when everyone was tossing state-level and county-level maps around.
Obama only needs Ohio, Wisconsin and Iowa to win. He has a ton of States EV to get to 270.
As of Wrongney, if he loses Ohio, he has almost not route to 270 unless he wins ALL the other swing States or Pa.
Do you really think Wrongney will win PA?
@Tony, I am all with you as far as our hopes and aspirations for tomorrow night. Where the agreement ends is when the use of the word “Wrongney” emerges. I come to PEC precisely to get intelligent, well thought out statistics and commentary. The use of “Wrongney” (And come on, that isn’t even funny) has a better home on TigerDroppings.com or Rush Radio.
Agreed – most of us share your enthusiasm and your preference (are you making phone calls today?) but we don’t really get into the hyperpartisan stuff here. Plenty of space on the web for that.
Time for me to put on my voter protection hat, and do my part to ensure that every registered voter on my turf in Montgomery County, PA who wants to vote is permitted to vote.
Don’t sit on the sidelines. Vote. Also, do what you can to help others vote.
Exactly. Before there are issues to be decided, the vote has to be protected and exercised. I’ll be an election judge tomorrow, and I’m glad to see that others will be out there encouraging and protecting the franchise!
Thank you, councellorben
Another thing to keep in mind is that Obama doesn’t need Ohio like Romney needs Ohio.
Obama can STILL win if Romney wins, Ohio, Virginia, and Florida by taking PA, WI, NH, CO, IA, and NV. Now granted, I don’t want it to get to that point, but that’s, as of the polls right now, a better likelihood than Romney getting to 270.
Let’s hope it doesn’t come down to Colorado.
The Republicans have a 2% edge in early voting, much to my surprise.
Dem 34.6%
Rep 36.6%
None/Oth 28.8%
http://elections.gmu.edu/early_vote_2012.html
Olav, I think that if you look at the voting distribution among the early voters in terms of who people actually vote for and not strictly on the basis of registration (it was in one of the other threads), O should have a 3%+ advantage (that’s according to my math, so take it for what it’s worth).
According to the Obama campaign polls, they have a larger advantage than that, but I don’t know how trustworthy that would turn out to be.
I agree that if Obama wins FL it’s game over and goodnight. I’ve no idea when FL will be called, though, and I expect Romney to win it anyway.
So, that takes us to NH – which is insignificant, but if Obama wins it this makes things slightly harder for Romney – and VA. If Obama wins VA, I shall breathe much more easily, but I won’t be going to bed.
As some above have noted, OH may be called before VA anyway. However, if it comes down to counting provisional ballots (ie OH is very close) we won’t be finding out the winner tomorrow.
Spare a thought for me, as I’m five hours ahead of EST – you get to know the result in the small hours, whereas I have to wait until dawn!
Suja P. is just talking about the electoral-vote math, and is correct. The analysis says nothing about how probable those various outcomes are, so it’s a crude indicator of likelihood, but it’s true as far as it goes.
Eyeballing the polling right now, I’m getting more and more confident that Obama has the advantage in Virginia unless vote suppression/fakery somehow messes things up. It looks like he’s breaking away there much as in 2008.
But not in NC or, I think, Florida.
The one state I really don’t think I can call is Colorado. The stories about early voting make me want to call it for Romney, but it’s really hard to tell.
Many people seem incredulous that Colorado could be less Democratic than Virginia, but they’re just assuming this election will be like past elections. I think minority voters are currently way more motivated for Obama than rich white liberals, and though both states obviously have their share of both types, that could well be the marginal difference between Virginia Democrats and Colorado Democrats.
Matt
Some data we have on Colorado was that the legacy polling methodology failed there in 2010, Rasmussen was off by 5%, and Buck and Tancredo (predicted by Ras) both lost.
I doubt that Ras has changed his methodology.
Virginia and Colorado are demographically quite different. why would you expect them to be alike?
Ah, so that’s what the problem was. Thanks, Matt.
I think that third party candidates will have some small role in determining outcome in Virginia and Colorado, and will help Obama win Virginia and lose Colorado or win by a narrower margin (courtesy of marijuana decriminalization).
Yeah, I can’t imagine many Virginia Democrats voting for Gary Johnson over pot. Colorado? *Boulder?* Don’t bogart that joint, dude.
Could be, although since the pot issue is already a stand-alone issue (constitutional amendment) in CO, it’s possible that segment of the electorate would not necessarily feel they had to vote for Johnson. It’s also possible that he could pull from the Ron Paul libertarian side of the republican base too.
@ Matt
I think the legacy is that people regard Virginia as the seat of the Confederacy. And large parts of the state are just that. But NOVA is basically an extension of the very blue northeast corridor.
On the other hand, Colorado has blue Denver, but deep red Colorado Springs.
Virginia is a really, really complicated, really weird state, and I say this having grown up there. It’s the old capital of the South and not-the-South at the same time; it’s got tidewater aristocrats and Appalachia mixed in as well.
Lots of minorities, of more types than even when I was a kid. Many immigrants from all over the world in NoVa. Lots of federal government employees and government contractors, but many of those are military. There are even a few remnants left over of the pre-civil-rights ultra-conservative Southern Democratic party.
VA went for Obama in 2008, but whenever Republicans are in control of the state government, they are still as hard-right as you’ll find anywhere. The notion of it as the capital of the Confederacy still has a lot of truth to it, but it feels like the place is changing rapidly.
Matt, the way I see it, NoVA is more of an extension of DC than it is Virginia, really. We should secede from the rest of the commonwealth.
Why secede when we can run this place? If we formed our own state, we’d cede two reliably GOP Senators to the rest of the state. 1/3 of Virginians are Northern Virginians. Half of them live in FFX Co. When combined with Richmond and Greater Hampton Rhodes, we can dominate state politics if people just remember to show up for our off-off-year elections.
Washingtonian had a cover story about this a while back, on the eve of the 2008 election. Rabble rousers. http://www.washingtonian.com/articles/people/will-northern-virginia-become-the-51st-state/
Greetings from ground zero in Alexandria…
Mason, I’m not the Princess of Patience! I want us to keep the “communist part of Virginia” all to ourselves (there, I said it; I must be a commie pinko liberal)
“People’s Republic of Northern Virginia,” LOL.
I had to go way back to find this one.
http://www.washingtonpost.com/wp-dyn/content/article/2006/11/15/AR2006111501395_pf.html
She was just as weird before and then during the Civil War, which is why West Virginia exists as a state today.
I think a lot of the robo polls with no cells numbers called are underpolling on Hispanics
What if latinos are underrepresented in polls to date? Seems more likely than bias against Romney. Hard to predict turnout for a new and growing group. Seems like a huge latino surge is more likely than an “undercurrent” in the other direction.
If this plays out, FL VA NV and CO may go more Obama than they are showing.
Well, I think there is some truth to that. (See Nevada and Colorado two years ago.) But I don’t think it’s going to make as much difference in Florida, where Hispanics tend to be much more Republican than average (due to the Cuban community). Obama won FL Hispanics by 15 pts in Florida, which is way below his margin nationwide.
I feel like boosting turnout from a group where you are +15 would have significant impact. Not as significant as when you are +30, but significant nonetheless.
I think this is the only way Obama wins Florida. If latinos come out big, Obama in a landslide.
PS – GOP should watch out in 2016. Next state to flip is AZ.
That’s a reasonable point – but remember that Obama did better with the Cuban vote than any Democrat since…well, since before Batista fell. Unless Obama significantly revs up turnout among the Puerto Rican vote, I don’t think he can count on 15 points again – probably the 10-12 point margin that gore and Kerry managed.
Blech, I mean to say closer to the Gore/Kerry results – Bush won both times. 55-45 sounds reasonable, and a split wouldn’t be out of the question.
Latino Decisions says only 31% of FL Latinos support GOP. The Latino Vote is also 17.4% of the electorate, according to them. See: http://www.latinovotemap.org/map/
Cubans make up 31% of Florida electorate. See: http://www.pewhispanic.org/2012/10/01/latinos-in-the-2012-election-florida-2/
Many of them are GOP (but not all). In Nevada, Colorado,Latino support of GOP is more like 20% but are a smaller share of the electorate (around 12%).
So, under counted Latino vote in FL could be a big deal and reason why polls are off.
Of course Latinos in FL tend to be more Republican-leaning in general (because many are Cuban-Americans) but the fast-growing part of the Latino population in FL is non-Cuban. Thus, if Obama benefits from an unexpected surge in Latino votes in NV, CO, etc., he is also likely to benefit in FL.
The question is not whether he’ll benefit at all, but whether expected benefit (or alternatively, expected sampling error) is enough to overcome the a small deficit in the polling. (I took the unweighted Pollster mean and median of the last 10, 20, and 30 FL polls, and they show a persistent Romney lead of about 1 point, and the most recent ten are actually most pro-Romney).
I think Scott has helped Obama in FL on the Latin and Black turnout tomorrow. Several people that the news has talked to who didn’t get to vote early are planning on voting because of the fiasco. Some people who weren’t going to vote are angry. Things could change. Romney holds a razor thin lead.
Tony, consider re-reading what I wrote. I said that MSM will likely not call several of the swing states early (to string the populance along, to avoid what happened in Florida in 2000, etc.), if there is not a blow-out.
I personally think that Obama will win Virginia, Iowa, Wisconsin, Nevada and Iowa, which will be sufficient to get to 270. Ohio and anything else he wins will be icing on the cake.
You typed Iowa twice…
Aw hell, I spelled ‘bellwether’ incorrectly as well. Probably a bunch of other stuff too. Where’s the edit button when you need it?
Anyway, I think that the math still adds up without counting Iowa twice.
The State to watch is Vermont. At 7PM there are lots of red states closing. On the East Coast the red states close the polls earlier and the blue states like New York stay open later.
Vermont is the exception. Vermont polls close at 7:00 Eastern and will be the first state to declare for Obama.
Georgia, Indiana, Kentucky, South Carolina will give Romney a quick lead and his only one of the night. Virginia closes at 7 but will not be called at 7PM. 8PM includes lots of big blue states so the President should be close or tied by then. If so then he wins because New York and California are waiting in the wings at 9PM and 11PM Eastern.
Vermont will keep the President from sitting at 0 votes from 7PM to 8PM. Vermont may also beat out Hawaii to give the President the largest margin among the states.
Our Democratic Governor, Senator and Representative will all be elected in a walk. So I would say that Vermont is the state to watch at 7PM! By 8PM our time in the sun will be over and the President will be on the way to victory.
My memory is that the networks don’t make any calls until the polls have closed in California. Am I remembering that incorrectly?
@weichi
You are right. Last election it was completely obvious that Obama had the votes to win, but they did not call it until the polls closed on the Pacific Coast.
@weichi
Yes, they don’t call the presidency to anyone before the California polls close, but they of course call each state separately before that
If Romney wins Vermont, that’d definitely be an indication of something we don’t understand…
Assuming Obama wins (and I don’t have to go back to the UK) what I’d be most interested in seeing is how the polls performed over the entire course of the last 3 monthsd.
Can Rasmussen claim to be unbiased if they skew Republican for every day but the last and then switch to an Obama win? Or will someone (Sam?) call them on it?
Feeling a bit let down by that sudden drop in your EV (because of NC I guess) yesterday when there weren’t really any significant polls I am aware of to justify that drop. Could it be Wang is just trying to fall in line now just like some of the other pollsters are doing? I thought his model was less volatile than this and better at predicting further out rather than the day before kind of thing.
PEC is the leading edge of the waveform– not the wake.
There’s not much that Professor Wang can really ‘do’ to pull his model in line, as his model relies on the “pollsters” polls to feed into the EV and MM.
The difference between 294 and 347 is CO, FL, and NC, which are all looking like true toss ups.
So the projection is lumpy.
The mess in Ohio regarding provisional ballots still has me worried.
http://www.thenation.com/blog/171011/eleventh-hour-gop-voter-suppression-could-swing-ohio#
Provisional ballots only matter in an extremely close election. I mean it could matter, but I’d be much more worried about it if Obama wasn’t clearly ahead in Ohio.
@Donn – This provisional ballot business seems much overblown to me, with regard to potential impact on the outcome. (Important for the integrity of voting rights, to be sure.)
Consider: Pollster.com currently puts Obama ~3% ahead in their polling average. Assuming that margin of victory and turnout of 6M, that means a margin of about 200,000 votes. In 2008, the counting of provisional ballots and late-arriving absentee ballots altered the election-day count by a net of 57,000 votes, in Obama’s favor.
And here we’re only talking about the number of provisional ballots that might be incorrectly disqualified, not all the provisional ballots. So it would have to be just extraordinarily close to affect the outcome. And of course the legalities should be fully aired and resolved, but there really isn’t significant concern that this will flip the outcome.
Thanks, Matt and James! I appreciate your thoughts!
I would love to see a post-election ongoing effort to remove any and all elected officials, right, left, or center who participated in any fashion in disenfranchising voters.
If we could put forth the same effort nationally as they did in Wisconsin recently, maybe it would send a clear message going forward that we are indeed a democracy, and not a 3rd world banana republic. Maybe then other countries would respect us a bit more when we send in election monitors to criticize them.
As usual, on average,* the Republicans will win the over/under states:
Under populated
Under educated
Over weight
*by national average
You could remove two of those and just go by education. It’s really all about critical thinking skills.
@Fred “Could it be Wang is just trying to fall in line now just like some of the other pollsters are doing?”
No, Dr. Wang uses pure polling for his estimates. He doesn’t weight the polls. As such, his MM will respond very quickly to any changes in the race when compared to other models (e.g. Silver’s 538). However, this also means that it’s a bit more sensitive to noise.
If you look at the MM over the last couple weeks, it’s actually gone up and down fairly dramatically (from ~1.5-3.6) from day to day. It’s best to look at the overall trend, which shows an Obama bounce and then reverted to the mean.
This is my interpretation. I could be wrong though! Anyone else can chime in.
What explains today’s big drop in the meta-margin? In your commentary yesterday, you said focus on the stable meta-margin, not the EV prediction, and pointed out that the meta-margin was stabilizing right where you’d expected months ago. Now it has dropped a lot.
Looks like Pennsylvania is under-polled by the more reputable firms. Some marginal polls dropped PA to +3 and this dropped the MM.
@ John F.
I mentioned this in another thread – independent/Democratic firms generally have Obama/Casey leading by several points, while Republican firms* (and one university) have it a tie race or a 1-2 Obama/Casey lead.
*Susquehanna, McLaughlin, Wenzel, GS Strategies, Rasmussen/Pulse
The university poll is Quinnipiac, who is much better than the firms I just listed. They have a good history in PA.
I think John is referring to the Susquehanna poll and probably the Pulse Research, “Let Freedom Ring” poll. On a histogram of polling results, Susquehanna occupies all the most pro-Romney results. It’s pretty remarkable, actually.
Yes, sorry – I have a previous comment in moderation right now. Basically, five firms plus the Q-poll comprise nearly half the PA results over the last two weeks, and the bivariance appears to be stronger in PA than anywhere else right now.
Sam,
Do you think there is any legitimate, statistics-based reason that predominantly right-wing surrogates and/or pollsters are so confident about a Romney victory? I understand your methodology, but the only reason I can find for the Republican party’s view is posturing. Are we close enough to suggest that a Romney victory (especially of a sizable margin) would be the result of ‘skewed’ national AND state polls?
Thanks for all of your hard work putting together everything and then explaining it in ways that reasonable people can understand.
Question:
I would hazard a guess it’s about posturing, and protecting turnout for down-ticket candidates and initiatives, but that’s just my “gut” feeling (very much frowned upon, especially by me!).
Shawn
it doesnt have to be “gut-feeling” . Its cheater detection and evo theory of cooperation from evo bio and evolutionary theory of games.
for example, i can make a 2×2 payoff matrix that shows it maximizes payoff for Rasmussen to cheat over doing honest polling.
😉
Pardon me, may I vent?
YES! EXACTLY!
I’m curious about NC. According to Prof. McDonald who runs the US Elections Project site, 62.9 % of people in NC have already voted (ID = DEM(47.7%), REP(31.4%), OTHER(20.9%).
So do the recent Gravis or PPP polls(for example) still have their full impact on Prof Wang’s model?
cheers,
gryff 🙂
Sam, a little confused. The last few days the Meta-margin has dropped. ~.7 the the probabilities remain about the same. In your post you say watch the Meta-margin. Is the drop in MM, not significant, or am I missing something?
The point is that whether the MM is at 2.2 or 2.7, it is unlikely in the extreme that opinion will shift, in a day or two, by the amount needed to change the predicted outcome. Therefore the win probability remains essentially unchanged.
Thanks JamesInCA
You’ll notice that the MM has been riding in the upper part of the red zone the past few days (just as the EV number is riding low in the red zone right now) and the “correction” today has simply dropped it back to the center of the red zone (likewise we would “expect” a reversion to the mean on the EV calculation, that is, something higher than the current 303). In that sense, the shift was not surprising, though there are all sorts of technical reasons why the previous numbers might have been high and the current number (and strike zone) low. In any case, if today’s number does represent a reversion to the mean, it suggests that the race changed in October from roughly MM O+3% to MM O+2.25% as the basic ground of the race. At least that’s my read.
I think the O +3 was more of the upper end of a “jittery” meta-margin. Nature of the calculation. Also, a few polls from Republican house pollsters and strong leaners in that direction have dropped the meta margin in some states.
The other big change is that the NC meta margin was lagging the polling for some reason. It has been at +2 by my count all weekend, although it was reported as toss up when the MM was +3.
I am worried about CO, as well. However, if you really think about it, people in CO don’t need to vote for Johnson to legalize pot, they can do it themselves
@bks “p.s. I predict that the number of visitors to this site will drop dramatically on Wednesday.”
Hopefully PEC slate is wiped clean and we have the start of the Meta-Margin for 2016! Lots of hand wringing for those ‘upcoming’ elections! :}
Needless to say, it’s been wonderful to hang out with all of you here. Entertaining, enlightening. Have not enjoyed (though sometimes dismayed to the nth degree) an election built up like this ever. PEC forum has been a satisfying addiction.
PEC and 538 have Florida as the closest race. What are your thoughts on which way Florida will swing?
http://www.mayoclinic.com/health/relaxation-technique/SR00007
If I didn’t have a professional reputation to uphold I would be downtown on the corner acquiring a certain green colored, leafy schedule 1 controlled narcotic.
Does anybody have any thoughts or information on the intense swing in Indiana? Indiana’s been deep red for a long time, went to Obama by 1% in 2008, and is now polling deep red again (Tanenbaum shows an 11% margin). As far as I can tell, no other state has been nearly that volatile during that stretch. The Wikipedia article (link below) does mention that Obama visited Indiana 8 times and outspent McCain by a large margin. Still, I don’t see how to account for the huge swing. Any thoughts?
http://en.wikipedia.org/wiki/United_States_presidential_election_in_Indiana,_2008
Maybe, in response to going blue the last presidential cycle, the Indiana republican party focused more on GOTV efforts to avoid future embarrassment for them? I haven’t looked at any figures so this is just speculation on my part.
If any of the following are called for Obama, its over
VA, OH, IA, FL
If anyone the following are called for Romney, its over
PA, WI, MI
Sam, your swing districts info looks like it’s from three weeks ago… do you have anything more up-to-date?
Thanks!
Can I make a request? A desperate plea really?
Prof Wang… now that you seem to be getting increased media exposure, if you get the opportunity could you please explain to any mainstream media figures the difference between a poll allowing for the possibility a race is tied since that result falls somewhere inside it’s MOE… and a poll saying the race IS “statistically tied”? Every time I see some talking head on television report “New polling shows candidate X is leading by 3 points. Which means the race is statistically tied since that is within the margin of error!” I have to fight the urge to throw things at the screen.
He could give hosts an info sheet suggesting this question, and also the question,”Is there anything else you’d like to add to clarify media confusion about polls?” or some such.
Sam, you’ve done some great work keeping liberals off the edge (I suspect) and I’ve been impressed by your levelheadedness in your writing and your interviews.
I have a crazy conspiracy theory question though: if, with the level of certain that you’ve exhibited, Obama isn’t re-elected, would you give any credence to the possibility that election fraud was involved? I have to say, as a Canadian (we still use paper ballots) I find it beyond strange that your country seems to be moving towards methods of determining winners that leave no paper trail and depend on proprietary software that so many different bodies have shown is, at best, unreliable.
Any chance that this “narrowing of the polls” narrative is the prelude to a massive fraud?
The weather forecast is decent for Cleveland tommorrow. Yea, Weather Gods. Watching the video of lines in Florida and Ohio made my blood boil last night. To all the many decent activists reading this site, election reform ought to be our “meta movement” and that includes election reform in the US Senate!
I am so happy to have found this site after hearing you on Sci Fri. Thank you, thank you, thank you. Your voice of reason is so needed.
VOTER TURNOUT QUESTION,
I hear alot about the importance of voter turnout, and comparisons between elections, is there any data to suggest what percentage +/- turnout will be needed by either candidate to really turn the elections in the battleground states?
Romney needs a lot of Democrats to stay home. If the numbers come close to 2008 Obama wins in a landslide.
watching msnbc this morning, they are talking about early voting in OH, IA and other swing states, and it appears the Republicans have been successful in reducing the number of Democrats who have been able to early vote. the Margins are much closer than they were in 2008. I would discount these numbers if they were coming from Rove, but they are coming from Chuck Todd
@Jon V
Democrats have a very significant lead in Nevada and Iowa. Ohio doesn’t report numbers on party registration, so it’s impossible to know what number of folks are turning out there by any partisan measures.
Colorado has seen Republicans make up ground, but they’ve lost a lot of ground in NC and Florida is somewhat close to 2008. It should be interesting to see how it unfolds.
why would you trust Todd anymore than Rove? I wouldn’t, they all are part of the same scum.
Believe me…the demographics of Virginia have changed considerably. It is not the old confederacy any more.
You know I just had something of a revelatory thought on voter suppression. I came to the conclusion that it’s very easy to slip into the mindset that the general population is this thoughtless mass swinging like a pendulum to the different political masters. And then I realized that people are probably MORE likely to wait long hours in line if they think someone is trying to take their vote–especially groups that are traditionally oppressed (e.g. minorities).
I am surely hoping the Republican suppression efforts backfire!
Just ‘tooting my own horn’, but I got polled last Friday by SurveyUSA and the results are out! Obama +5 in Ohio.
Well done. Have you considered taking your show on the road?
This pretty much sums up Ohio right now:
http://i.imgur.com/2JsCX.jpg
I live in NH. It is 95% white.
If Obama wins VA+NH he won’t need OH.
Obama+Clinton drew 14k in Concord. Largest rally ever in our small state. Two 3rd party candidates may draw a few thousand from Willard.
Predicting NH will go PBO by 1 or 2 points.
So there were some Vegas odds that came out today with Obama at -400 and Romney at +300. Clearly they have not been following this blog or 538, but where would they be getting their information? With money on the line, how are they in such direct opposition to most of the data out there?
Here is the link – http://sports.bovada.lv/
These are “money line” or “American” odds. A minus sign means a win is considered more likely; it’s the amount more that you have to bet to win $100 if you’re right. A plus sign means you win more if you’re right, because a win is less likely.
They think Obama is much more likely to win.
Well that makes much more sense. I guess this shows my total lack of experience in sports gambling – in my mind anything with a minus sign that relates to money would be “bad.”
But thanks for this – I will actually sleep better tonight as a result of your reply!
-400 to win $100 you would have to bet $400. obama heavy favorite.a week ago it was -165 .
I don’t know anything about sports gambling either, but I’d seen people cite odds like that before, once or twice, so it gave me something to go on when searching for more information.
In Va, the only swing back to R would be the coal counties of the west. They were reliably D until the last election. McCain barely carried them and we will see how R does. NoVa is the blue part of the state, as well as the major cities (except for Va Beach). The divide between D and R here is Route 15. West is red and east of it is blue. You can see it play out on election maps. But count on the fact NoVa is the primary carrier of this state for O.
Ok, no one will tell me me when the PEC predictions at top stop or freeze on this one.
Let me ask this then: Does/will the PEC Median EV Estimator line continue PAST the dashed line now shown vertically through the red/yellow bar?
Listening to Sam and others on the radio. This is irritating like watching the first debate. Sam looks good though -no prob there [not like where the Prez looks at the studio audience as told to while the camera shows him looking down]
But I hate the framing- “let’s ask Sam, who knows. Hmm, surprising. Hmm. Now what do _you_ think, Random Careerist?
Now theyre talking abut Congress with no clue that Sam knows about the House. Be sure and tell em your history beforehand next time youre on a show with others…
They’re conflating the terms Republican, conservative and reasonable.
Ah! They let sam close the event with his forecast. Proper! I’m pleased.
My, you are a fast one to comment.
Yes, I was there for minor quantitative commentary. I was all set to talk about the Senate and House, but nobody asked. At least I got to correct Mark MacKinnon on his contention that Sandy is the dog that ate Mitt Romney’s homework.
Congratulations, Sam.
You got real info out there, and though this shouldn’t matter like it does, you looked great doing it. A neat trick on radio.
You left the host wanting more, and gave enough info for the folks to find you.
Congratulations, Sam.
Can you tell me when the final predictions here post at top? Is it days after the polls close? Does the Median EV estimator line continue past the dashed line.
[Radio keeps blathering: these are _middle class_ people who are homeless! As a person who gives out sleeping bags to people sleeping in puddles in doorways I can only think, O geez, _that’s_ the tragedy- not that theyre newly homeless human beings, but what class they are.
May I scream now?
Hey Dr. Wang,
I’ve been assigned to read “The Whole-Brain Child” by Siegel and Bryson; perhaps after I finish it, I will look into “Welcome To Your Child’s Brain.” Thanks for running a first-class site here.