Sign up FAST! Login

Either all the polls are wrong, or Romney has no viable path to win the election.

Stashed in: Awesome, Obama!, politics, Republicans

To save this post, select a stash from drop-down menu or type in a new one:

New York Times ran 512 paths to victory. It looks very bad for Romney.

Five Thirty Eight now gives Obama 92% chance of winning, with him likely to take 315 electoral votes and also the popular vote.

Romney's best remaining chance at victory includes winning the three big battleground states that are still in play this cycle: Ohio, Virginia, and Florida. With Colorado and New Hampshire, Romney would walk away with 279 electoral college votes. That's very unlikely.

Here's a bunch of what-ifs.

It really comes down to the ground operations now...

$100 that Obama's ground game is much better than Romney's. I've seen it first-hand today. It's weird/creepy how effective their ground game is. Romney is doubly the worst candidate; people are mobilizing against him because he's so robotic. 

God have mercy on our souls if a VC becomes President.

 Can you say more about the ground game you got first hand exposure to?

 It's definitely all going to be demographic turnout... I think that the polls are particularly wrong about CO, and NH.  Both states had thoroughly ineffectual GOP state parties in recent elections, but have gotten on their game this time around... This is particularly true of NH, where a vicious civil war between the 2 GOP factions (which operated as sub-parties) only ended in 2010 and resulted in a massive landslide in the state House, Senate, and Exec Council.

Romney pulled out of NV, MI, and MN late last week and you could say they pulled out of OH too, except something occurred there that I don't think has ever occurred in politics before: literally no more ad space to buy even if they wanted to.

PA is going to be the wild card, due to the storm. There are still about 30k w/o power, mostly in the Mainline counties, and the storm still dominates the press there (tho the imagery is mostly the debacle in NYC).

Also PA is a pseudo-VoterID state. Poll workers are still required to ask for ID, but Poll judges (yes, they're elected separately in PA) aren't allowed to restrict a voter without one. However, even the question being asked will deter some.

Fascinating that after months of battling in 10 states, we're coming down to the ground operations in a handful of them.

I'm wondering if all that money spent on ads makes any difference at all.

Cramer predicts a huge Obama victory.

 I've long advocated that it doesn't... However, gotv doesn't make the campaign consultant any money (they usually take 15% spiff of ads), so it's common for a campaign to completely rely on volunteers.

@Belich Do you live in PA? PA is not a wildcard. It is for Obama. OH is not a wildcard. It is for Obama. The storm destroyed the last vestige of hope Romney had. Nobody wants a vulture capitalist as President; if there weren't irrational concerns about him being a mormon from the religious right, his comments on Detroit, FEMA, the 47% and a litany of other gaffes have put the nail in his coffin.

@Panda I got home; there was a printed thing-a-mabobber hanging on the doorknob. But it wasn't low-quality crap like most people print; this was high-quality, 300-pound paper.

It had my name; local address, time, polling date, website. 

Nice glossy photo of Barack Obama with a phone number to call. Also what Senator to vote for.



"YOU DO NOT NEED An ID TO VOTE. IF YOU ARE A FIRST_TIME VOter, PLEASE VISIT" *OK this sentence wasn't in caps but the others were. 

Phone calls, emails, get out the vote type stuff; I mean, if you intend on voting one way or the other, I think they're doing a phenomenal GOTV campaign. 

 1. Calm down.

 2. I worked in PA politics for 5 years and have several friends there who work on both sides, incl an old friend of mine who is likely to win a seat in the state House.  The PA Dems are worried but confident and the PA GOP is amped up over actually mattering for once.  The storm made Eastern PA a huge unknown and people I know on the ground there acknowledge that. Operatives don't like unknowns.

Gartman still believes Romney will win handily.

 I really really want a tie... but i'm a bastard like that.

The omens seem to point to romney (i.e. Shrum, Redskins, the Flip Rule) tho not all (i.e. Dick Morris), and the demographic fundamentals <i> as we have traditionally understood them</i> seem to point to Romney.  He _should_ be up by at least 7 or 8 b/c of the economic numbers alone.  

It is entirely possible that the pollsters have gotten the demographic ratios completely wrong and there will be a surge of minority support for Obama that actually meets or exceeds 2008, but Latinos are not particularly relevant in the battleground states (but are in NV and could be in AZ) and Black voters are not as large part of the electorate as they were in 2008 (tho will certainly turn out in numbers that would otherwise be historic).  Romney smacks Obama around among independents and is at near parity with educated white women.  Under pre-2008 demographics, he should be up at least 10pts, and if you're basing your turnout demos on that, then yeah he's gonna landslide.

What is absolutely certain, and I seriously doubt the GOP comprehends this yet, but this is the last election where it's even remotely possible to win with only white voters.  The nutters will absolutely have to change or risk becoming a rural rump party, like has happened in CA.

 Another omen.... Dixville notch tied..... in 2008 it was 15-6 Obama.

It's weird that they would try to win with only white voters.

Seems like a huge, unnecessary handicap.

Brace yourselves. Half the country is going to be very unhappy.

Brace yourselves voting is coming

So, could all the polls be off due to systemic error?  Yes - no ideological bias required.  Nate Silver

has an excellent, advanced projection model, but it has a weakness that

he himself has pointed out and that is completely outside of his

control: It depends on the quality of the input polls.  Silver is OK

with using the state polls because they have a history of doing well. 

However, they didn't do as well in 2010 as they had in the past.  And,

what if that underperformance was a leading indicator of the effects of

some basic challenges to the state polls that indicate one can no longer

compare them to the historical state polls because the differences are

differences in kind?

What's changed enough to undermine the assumption of quality in the

state polls?  A few things (and none of them depend on any ideological

biases or preferences on behalf of the pollsters).  Those changes are

driven by technological change, crowding of the field, tighter funds,

and the polls using 2008's anomalous turnout as a base. 

The technological challenge for pollsters is caller ID.  It's

obviously been around for a while, but caller ID has become free and

ubiquitous - essentially a default.  People can see who's calling - or

who is calling and declining to ID himself. And if the voter doesn't

wish to deal with another political call, he can simply not pick up.

That's an attractive option for me in all cases, but one can imagine

that even someone who typically agrees to be polled might get tired of

it.  Which leads us to the second, interplaying factor.

Overcrowding of the field in the swing states has become a much

bigger problem over the last couple election cycles.  Traditionally,

there were only 1-3 established polling organizations doing swing-state

polling, but that number has exploded to (I believe) 13 in Ohio (as a

representative swing state).  So you have many more pollsters trying to

conduct their polls.  And, you also have a lot more money being thrown

at political calls of all sorts: GOTV, push polls, fundraising, etc. I

don't live in an official swing state, and about half of my calls over

the last 2 weeks have been political calls.  And I have anecdotal

evidence that for someone in OH with a landline, it's along the lines of

28 of her last 30 calls.  She is not answering those calls, BTW.

And that leads us to why those two factors matter.  Pollsters need to

survey proper samples to get usable numbers that they can project over

their target populations. That is tough when a lot of people refuse to

pick up the phone.  And that also plays to the last factor: contacts are


Pollsters utilize call centers to contact voters and get their

opinions for their polls. Each contact costs money. If a pollster gets a

person to answer the phone, and that pollster is trying to control

costs, that pollster will not want to discard that person's response. 

However, pollsters also know that there is a hierarchy of responders in

terms of quality in projecting electoral results: Likely voters >

registered voters > adults.  At this point in the election cycle,

everyone wants to report on Likely Voter numbers and no one cares about

Registered Voter numbers, much less adults.  But putting people who pick

up the phone through a Likely Voter screen will inevitably result in

rejecting some of those people - and the more rigorous the screen, the

more likely the respondent will be rejected (and you can make

assumptions about the likelihood of voting as compared with the

likelihood of taking the time to respond to a pollster, which will

likely vary based on the age of the respondent).

So, that hasn't mattered a whole lot historically, because you had a

couple of polls sponsored by big news organizations who were willing to

foot the bill to get good results.  However, as newspapers and even TV

have faced technological disruption and budget cuts, the budgets for

polling have been cut.  And as a lot of new polling organizations have

come into existence to try to get in on swing-state polling, you have

more competition for fewer dollars.  What is one very easy way to reduce

the cost of running the polls?  Loosen the Likely Voter screen and

reduce the number of rejected respondents.  The effect of doing so will

move the results of nominal Likely Voter polls closer to the Registered

Voter polls. 

This would have a systemic effect of underreporting GOP support,

based on the historic gaps in the Likely Voter and Registered Voter

polling numbers (and note that electoral results track better with good

Likely Voter polls historically).

That's the first factor: weaker LV screening combined with voter

fatigue w/ polling and calls would lead to GOP performance being


The second factor is sample weighting for Dem./Rep./Ind.  Pollsters

weight their samples based on party ID to attempt to reflect the

population that will turn out to vote. Most, if not all, the state polls

are basing their D/R/I sample weights on the 2008 exit polls (or, if

they are a bit more meticulous, on actual final voting numbers). 

However, 2008 was an anomaly that isn't explained by demographics. 

Democrats had a very good turnout; GOP had an historically depressed

turnout.  2010 showed what could happen with an opposite effect.  But

2000 and 2004 are likely much better examples of what we should expect

in 2012.  This would mean slightly lower D turnout, and significantly

higher R turnout, than in 2008.  There doesn't seem to be any evidence to assume an anomalous turnout like 2008; while we don't know precisely how early voting tracks election day voting, the early voting numbers seem to indicate 2008 is the wrong model:

This would mean that the weights being assigned to D/R/I turnout

based on 2008 are overstating Democrat results and understating GOP

results for what we should expect in 2012.

So, you therefore have two major factors that don't rely on any

ideological bias on behalf of the pollsters that may be resulting in

systemic under-projection of Romney's swing-state performance.  The

national polls, which have shown much better strength for Romney, don't

have the same Likely Voter problems as outlined above, but they still

tend to have the weighting issues. 

There may be other factors at play - which may be enough to cancel

the above.  But, you don't need any nefarious explanations if Romney

outperforms Nate Silver's model tomorrow, which is based largely on the

assumption that those state polls are accurate projections of how the

voting will turn out.

Right, all the polls COULD be wrong.

That's why we determine the election on actual votes, not projected votes.

At this point it comes down to who has the most effective ground operation, who gets the most voter turnout, how many voters get turned away because the don't have the right identification, etc.

Half of Californians now vote by mail. But most of the U.S. that votes has gotta show up to the voting booths sometime today.

 True - hopefully we'll have it all sorted by tomorrow morning (who knows with Ohio, though...).Also, I want to note the possibility of a "Sandy Effect." 

Obama saw an uptick in his national numbers, which was caused almost

solely by a huge uptick in his numbers in the Northeast.  However, I

think it's fair to assume that the people pollsters could reach in the

Northeast would be happier and more likely to support the incumbent than

people whom the pollsters could not reach due to evacuations, power

outtages, etc.  Assuming those people were unpollable but still will

vote, that national bump may vanish - and there may be some effect on

Pennsylvania and Ohio (I assume there's no chance Romney wins New Jersey

or New York irrespective of how many people are upset and unpollable).

The combined effect would need to be about 3% undersampling Romney support to make it likely Romney would win, at least under these scenarios.

The Republican party is hedging it's bets by getting Marco Rubio ready to run for President in 2016.  

 In all likelihood the next GOP nominee will be a governor who is actually successful and popular.. Jindahl is known to have these ambitions.. Christie maybe.. and i'm hearing things about Susana Martinez of NM... of course they fucked the last NM governor out of seriousness, Gary Johnson.

You May Also Like: