In this year's race, national polls show a tie, while state polls show a decisive Obama advantage. The discrepancy has interested many commentators on the right: Karl Rove, Ross Douthat. Also Andrew Sullivan. Today at the Princeton Election Consortium, I offer a resolution to the apparent problem. I suggest that the difference may arise from the fact that the same systematic pollster errors can have different effects depending on whether they occur in national vs. state surveys.

Based on past elections, national poll aggregates differ from election results by as much as 2.5%. During the same period, state-poll aggregation has been considerably more accurate. Even if state polls have the same accuracy as national polls, races at that level are usually decided by larger margins, leaving room for aggregation to remove the effect of the error. I suggest that the Meta-Analysis of state polls provides a more accurate poll-based prediction of next Tuesday's outcome than national polls.

In the Wall Street Journal, Karl Rove surprises basically nobody by predicting a Romney win. His reason? He cites a Romney lead in some national polls. This has become a rallying cry for the right. But is 'the math' correct? Um, no.

Here at the Princeton Election Consortium, the Meta-Analysis of state polls points toward an Obama electoral victory. The median outcome is Obama 303, Romney 215 EV, with a Meta-Margin of Obama +2.2%+-0.5%. To put it into plain English: If state polls are accurate on the whole, then Obama will win.

However, national polls give a different result. National polls since October 14th give a tied median, 'Obamney' +0.0 +- 0.3% (n=44 polls, median +- estimated SEM). Indeed, the discrepancy with the Meta-Analysis has been over 2.0% all season.

What is going on? Nate Silver chewed it over yesterday. Let's go through some possible reasons using PEC's approaches.

Do differences in national and state poll methods account for the discrepancy? If we only accept polls from organizations that survey both the national race and individual states, we will have an apples-to-apples comparison. The result is the same: a national-poll median of Obamney +0.0 +- 0.6% (n=10 pollsters, 1 poll per organization). Dropping automated phone polls (PPP, Rasmussen, Gravis) gives Obama +0.5%, still not enough to account for the difference. Answer: no.

Are state polls slow to catch up? State polls take 10-12 days to reach a new steady state, even when the change occurs in one day, like Romney's 5-point bounce after Debate #1. Could it be that they have not caught up with national polls? This is unlikely for two reasons: (a) In national polls, the race has been stable for the last two weeks - long enough for state polls to catch up. (b) The Meta-Analysis is moving toward Obama - opposite to the direction expected. Answer: no.

Are there hidden advantages in non-swing states? Unlike state polls that influence the Meta-Analysis, national polls sample non-swing states. Could Romney have exceptional support in red states -- or make the race close in blue states? Using polling margins from Pollster.com (and filling in a few missing values using 2008 returns), an average (weighted by 2008 turnout) gives Obama +2.1 +- 0.6%. Sean Trende of RCP has done a similar calculation. That is basically the same as the Meta-margin. Answer: no.

How is the track record of national polls? Here is a comparison of poll margins and final results.

Year Final polling median Actual result Discrepancy
2008 Obama +7.0 +- 0.9 % (n=15) Obama +7.3% 0.3% (0.3 sigma)
2004 Bush +1.0 +- 0.5 % (n=13) Bush +2.4% 1.4% (2.8 sigma)
2000 Bush +2.0 +- 0.9 % (n=15) Gore +0.5% 2.5% (2.7 sigma)

For a bell-shaped curve, the average error is supposed to be 0.8 sigma. Here it's much larger, 1.9 sigma. Evidently, national polls have systematic problems. Aha...here may be our culprit. Answer: national polls do about 2.5x worse at predicting the popular vote outcome than expected if the wisdom of crowds of pollsters was perfect.

How is the track record of state polls? In terms of predicting both state-by-state and overall electoral outcomes, state polls do extremely well. In 2008, I correctly identified the leader 49 out of 51 races. I called two races (Indiana and Missouri) tossups, and those races had margins within 1%. In addition, the 2004 EV median precisely matched the final outcome. In other words, state polls get it 98-100% correct. Answer: pretty darned good.

But if state polls use the same methods, why would they do better than national polls? Well, state polls have three advantages. (1) Most state races, even in swing states, are decided by margins of 2% or greater. So an error that makes a big difference in national polls doesn't matter nearly as much for state polls. (2) State polls target more homogenous populations, which poses fewer technical problems to the pollster. For this reason, the systematic error might be smaller. (3) In critical swing states they are done more frequently. This focuses the data where information is most needed.

BOTTOM LINE: Even if national and state polls have the same flaws, they are consistent with one another. Because state poll aggregation is so powerful, the result based on state polls is likely to be more accurate. That is what I would call The Math.

If you like what we do, visit our ActBlue page. The closest races are where your donations are most effective. Among President/Senate/House, the closest question is who will control the House. Although Senate control is no longer in question, some individual Senate races are in play, including Heitkamp, Carmona, and Tester.

(cross-posted at the Princeton Election Consortium)

#### Tags

EMAIL TO A FRIEND X
You must add at least one tag to this diary before publishing it.

Add keywords that describe this diary. Separate multiple keywords with commas.
Tagging tips - Search For Tags - Browse For Tags

?

More Tagging tips:

A tag is a way to search for this diary. If someone is searching for "Barack Obama," is this a diary they'd be trying to find?

Use a person's full name, without any title. Senator Obama may become President Obama, and Michelle Obama might run for office.

If your diary covers an election or elected official, use election tags, which are generally the state abbreviation followed by the office. CA-01 is the first district House seat. CA-Sen covers both senate races. NY-GOV covers the New York governor's race.

Tags do not compound: that is, "education reform" is a completely different tag from "education". A tag like "reform" alone is probably not meaningful.

Consider if one or more of these tags fits your diary: Civil Rights, Community, Congress, Culture, Economy, Education, Elections, Energy, Environment, Health Care, International, Labor, Law, Media, Meta, National Security, Science, Transportation, or White House. If your diary is specific to a state, consider adding the state (California, Texas, etc). Keep in mind, though, that there are many wonderful and important diaries that don't fit in any of these tags. Don't worry if yours doesn't.

You can add a private note to this diary when hotlisting it:
Are you sure you want to remove this diary from your hotlist?
Are you sure you want to remove your recommendation? You can only recommend a diary once, so you will not be able to re-recommend it afterwards.
Rescue this diary, and add a note:
Are you sure you want to remove this diary from Rescue?
Choose where to republish this diary. The diary will be added to the queue for that group. Publish it from the queue to make it appear.

You must be a member of a group to use this feature.

Add a quick update to your diary without changing the diary itself:
Are you sure you want to remove this diary?
 Unpublish Diary (The diary will be removed from the site and returned to your drafts for further editing.) Delete Diary (The diary will be removed.)
Are you sure you want to save these changes to the published diary?

#### Comment Preferences

• ##### Tip Jar(151+ / 0-)
Recommended by:
ctsteve, JohnAdams999, gwilson, pamelabrown, Ashaman, angry marmot, sewaneepat, valadon, OhioNatureMom, broths, BlueMindState, HudsonValleyMark, housesella, arizonablue, chicagorich, deminva, deep, FishOutofWater, StrayCat, Andrew F Cockburn, willyr, pat of butter in a sea of grits, koosah, pico3, Dobber, karmsy, kefauver, Kimball, because, political mutt, SaintC, kestrel9000, SanJoseLady, bythesea, Rita in DC, binkycat, TFinSF, mconvente, Texdude50, reddbierd, Wee Mama, Obama Amabo, dickensgirl, Dillonfence, Clyde the Cat, BenderRodriguez, Deep Texan, p gorden lippy, pixxer, ItsSimpleSimon, rapala, quell, Alma, DCDemocrat, paul2port, RLF, bluedust, Trendar, tweeternik, Cronesense, Denver11, TomP, poco, Hatrax, 42, brn2bwild, jrooth, be the change you seek, mungley, kkjohnson, GMFORD, Bonsai66, TX Freethinker, mahakali overdrive, caspian, Delilah, Debby, litoralis, TheGreatLeapForward, davidincleveland, salmo, J M F, Xapulin, Sophie Amrain, MeMeMeMeMe, emelyn, anodnhajo, GeorgeXVIII, Just Saying, dmhlt 66, Neosho, quill, Nag, el dorado gal, mbh1023, sfbob, elfling, PennBrian, weidheimer, wishingwell, deboChicago, Hastur, katienne, Don Enrique, jfromga, Vatexia, Jake Williams, Catskill Julie, Philip Vincent, Tony McArthur, BasharH, SneakySnu, Its a New Day, archidem, Donkey Underpants, deha, distraught, sostos, it really is that important, sabo33, elwior, mikeconwell, Yellow Canary, Just Bob, taiping1, araina23, catly, Kristopher, MKinTN, exMnLiberal, TX Unmuzzled, ChrisBix50, dotsright, blackluck, rja, JJG Miami Shores, madhaus, George3, Sylv, SF Bay, bamjack, Jeau, acuppajo, Neon Vincent, Yasuragi, quaoar, pdx kirk, walkshills, JayC, glitterscale, paxpdx
• ##### Just wanted to say thanks(18+ / 0-)

PEC has been great to read the last few months, I enjoy your analysis and response to other's comments on PEC greatly.

• ##### thanks, sam(11+ / 0-)

btw, obama wins virginia

Colorado has a much higher GDP than Utah because it is culturally superior

• ##### And pretty much everything else(2+ / 0-)
Recommended by:
davidincleveland, elwior

C'est fini.

• ##### At the moment(0+ / 0-)

Dr Wang has VA on our side and FL as the only true tossup. I am hoping within the next few days NC edges into the tossup column as well. I want 347!

"I don't want a line in the Sand lines can be moved. They can be blown away. I want a six foot trench carved into granite."

[ Parent ]

• ##### 'The Math' by Sam Wang(19+ / 0-)

...differs substantially from the non-math of Karl Rove, of course.

• ##### Yes ! Karl Rove has failed math class several (2+ / 0-)
Recommended by:
jfromga, elwior

times over.  Sam is at the top of his class.

[ Parent ]

• ##### Karl Rove belongs in prison(8+ / 0-)

how infuriating to think he walks free, makes millions, and still poisons the body politic with his amoral anti american bile while Don Siegleman rots in prison.

• ##### LOVE to see a crying Rove carried awayin handcuffs(0+ / 0-)

Karl Rove should be arrested. President Obama can and should call for unity to solve our nation's grave problems. However, Karl Rove needs to be prosecuted for numerous crimes; illegal coordination between GPS and The Romney Campaign, instructing GOTP activists to PURPOSELY MISINFORM Democrat voters... Oh how I want to see Roves porcine tear-filled face as he is marched of in handcuffs to a squad car. No one likes a loser; the GOTP will turn their backs on him in a flash. They will scurry like roaches because they are are guilty of trying to subvert our elections.

• ##### ERMAHGERD 'Rithmatic!(0+ / 0-)

The wealth of the 1% has taken away the power of the ballot box of the 99%.

[ Parent ]

• ##### I think the divergence is 3 issues(2+ / 0-)
Recommended by:
jfromga, elwior

1. Whether electorate is like 2008 or better in terms of diversity.  Will youth and latino's actually vote?

2. Loose vs. tight screen.

3. Correct weighting of "party ID" (yes, party ID is still being debated--see the recent Q and PPP Ohio polls that were criticized for D 8+ party ID).

Honestly both sides present the arguments convincingly it's hard to know what to believe.

They wild card is OFA.  They know how to target and reach voters.  There was no OFA in 2010 or 2004.

• ##### You want to know how good OFA is? Before early(12+ / 0-)

voting started we were getting calls from OFA to vote, to request absentee ballots and so forth. Not much action from the Republicans.

Tall Papa and I voted on the first day of early voting. No more OFA calls to vote! We do get invitations to volunteer, which we take up as they fit into our schedules.

The funny thing is, now we are getting Republican calls! The oddest ones are for someone called "Paul" on our landline. I have no idea what that is about, unless my father called the Republicans on our landline back in 2008.

Is it true? Is it kind? Is it necessary? . . . and respect the dignity of every human being.

[ Parent ]

• ##### Glad to hear that (2+ / 0-)
Recommended by:
Wee Mama, elwior

OFA wasn't so good about putting me on their "send me less e-mail" list, even when I chose that option twice and then called them and told them that it still wasn't happening. I finally had to choose "send me no e-mail at all"; they complied with that one.

But first, that was earlier in the campaign, and second, being good at honing in on whom to call and whom not to call seems much more important.

• ##### Yes I got several calls from OFA during primary (2+ / 0-)
Recommended by:
elwior, Wee Mama

season. And here in PA, they callled me before and after I voted to see if photo  ID was asked for and if there were any problems.

As in PA , one can be asked but not required to show photo ID this year.

[ Parent ]

• ##### Me & My Wife Are Both(1+ / 0-)
Recommended by:
Wee Mama

Strong Obama supporters... yet we've been getting pro Romney/anti Obama mailers every day. Not just one... but 2 or 3 per day! They're in our names too. I don't know how we got on these mailing lists. I guess that just shows how much money they have to burn.

• ##### Now if we could only get this kind OFA in 2014(0+ / 0-)

We'll be in full control of both chambers of congress for sure.

Doubt we'll get the house this year, though.

--
Make sure everyone's vote counts: Verified Voting

[ Parent ]

• ##### Missing a major issue?(1+ / 0-)
Recommended by:
elwior

WILL our voters in NY, NJ & CT even be ABLE to vote! We need them in large numbers for the popular vote, and I don't see much emphasis on getting the special assets and procedures in place for Nov 6. Bloomberg does not care.

There is no early voting in NY. Unless you are able to get to your BOE to absentee vote (with acceptable reason) you must vote on Tuesday.

Gov Christie minced no words saying he didn't care much about the election in NJ--perhaps since his name is not on the ballot--or whether his largely Democratic people GET TO VOTE in it?

Okay, the Government says you MUST abort your child. NOW do you get it?

[ Parent ]

• ##### Get real re NJ - two things here:(3+ / 0-)
Recommended by:
elwior, ChrisBix50, TheChocolateChips

1. Christie is governor of a state that's essentially been wrecked.  Yeah, the election is not foremost on his mind.

2. I'm getting reports from friends in heavily Republican areas of NJ, and the communities there are wrecked.  Downed trees and power lines everywhere.  If New Jerseyans can't get to the polls on Tuesday, a lot of them will be Republican.

The road to Hell is paved with pragmatism.

[ Parent ]

• ##### Well, this Democrat thinks everyone should be able(0+ / 0-)

to VOTE, even Republicans. Not sure your dismissive sarcasm was entirely necessary, frankly. You related to Christie or something?

It's not just NJ, btw. CT and NYC are heavily blue too and may have serious election displacement. Thank goodness I already voted absentee. I would be beside myself if I could not vote!

Sign the petition

Superstorm Sandy victims must be able to vote!

By Natalie Bareis (Contact)

To be delivered to: The New York State House, The New York State Senate, Governor Andrew Cuomo, The United States House of Representatives, The United States Senate, and President Barack Obama

Petition Statement

Transportation to voting sites and/or alternative methods to vote must be made available to every registered voter the day of the election, whatever the cost.

Petition Background

I am concerned that holding the election on its current date will disenfranchise millions of voters on the Eastern Seaboard who have been affected by Hurricane Sandy. Transportation is at a standstill in one of the most populous cities in the US, and thousands of homes and businesses are flooded in the wake of the hurricane.

Okay, the Government says you MUST abort your child. NOW do you get it?

[ Parent ]

• ##### it may cost Obama(1+ / 0-)
Recommended by:
Catskill Julie

some amount in the popular vote... and that does concern me to some extent... we know how crazy the right wing can be, we don't need to give them any ammunition for their insanity.

• ##### Thanks for keeping us all sane!(5+ / 0-)

Although I admit I was going a little crazy yesterday when your site went down! Phew!

• ##### I can't follow the argument at the end(3+ / 0-)
Recommended by:
jrooth, TX Freethinker, elwior

Your metric for accuracy of state polls isn't the same as your metric for accuracy of national polls. Basically, at the end, you seemed to drop the question of explaining the gap between national and state polls. Instead, you shift to justifying a focus on state polls -- which seems reasonable, but not the same issue.

Election protection: there's an app for that! -- and a toll-free hotline: 866-OUR-VOTE
Better Know Your Voting System with the Verifier!

• ##### State poll accuracy(7+ / 0-)

The metric for accuracy is that races with margins >1% were called correctly. This implies <1% error.

I agree that this can be done quantitatively. When PEC is back online I can attend to that.

• ##### OK -- I'm not sure that it does...(1+ / 0-)
Recommended by:
elwior

but I'm happy to wait for the numbers. I don't have a dog in that "fight."

(Is your standard error of the median reckoned through bootstrapping, or what? I can look for that on the site if I really "need" to know....)

Election protection: there's an app for that! -- and a toll-free hotline: 866-OUR-VOTE
Better Know Your Voting System with the Verifier!

[ Parent ]

• ##### Standard error of the median(1+ / 0-)
Recommended by:
HudsonValleyMark

I calculate median absolute deviation and convert it to estimated standard deviation as described here. Then convert that to SEM. A correction factor must be applied of multiplying that SEM by 1.2, pointed out to me by Prof. Alan Cobo-Lewis (not published).

After that, the median margin and SEM get used to calculate a z-score, which gets turned to a probability using a t-distribution. This is chosen to have a long tail - "black swan" events are thereby included.

• ##### Dr.(2+ / 0-)
Recommended by:
elwior, MKinTN

I have been wondering whether a discrepancy in diminished turnout between swing states and non-swing states may be responsible for the difference in outcomes.  We know that likely voter screens and low turnout tend to benefit the Republican candidate.  It would be possible for turnout to be marginally lower than 2008 in the nation at large but comparable in the contested states, where sustained campaigning, GOTV, and advertising produce a state electorate more in line with national RV numbers, which have shown a consistent palpable lead for the president.

In 2004 for instance, uncontested Illinois cast 5.27 million ballots in the presidential election while Ohio cast 5.62 million despite the fact that Illinois had a larger population.

Perhaps what we're seeing is a national track in line with tighter Republican-benefitting LV screens that will not apply in contested states where campaigning produces turnout that approximates RV numbers, a phenomenon which would benefit Obama.

• ##### Sam, you say that control of the House (1+ / 0-)
Recommended by:
Just Bob

is the closest question.
My question is how close is it?

Strangely, a question this vitally important recieves just about zero coverage. The dream scenario of course is that Nancy Pelosi gets the gavel back and some real work can get done on behalf of the American People.
I just don't know what the chances of that happening are.

"We the People of the United States...." -U.S. Constitution

[ Parent ]

• ##### Likely Democratic seat gain <15 seats...(0+ / 0-)

...not enough to switch control. I'll post on that in the coming days.

• ##### Great work, but please drop the(4+ / 0-)
Recommended by:
litoralis, wishingwell, Hastur, elwior

"Obamney" reference. Ugh! shivers

"The most American thing you can do is to stand up for the dignity of someone else."

• ##### When Does Rove Pull Out His Dry Erase Board...(1+ / 0-)
Recommended by:
elwior

"Every good Christian should line up and kick Jerry Falwell's ass." ------Barry Goldwater

• ##### Rove will do the same thing he did 4 yrs ago on (4+ / 0-)
Recommended by:
elwior, Kristopher, MKinTN, ChrisBix50

Election Night on Fox News. He will keep insisting Obama will lose and then PA and OH are called and then other states and he stands there looking stupid.

[ Parent ]

• ##### Dr. Wang droppin' the geek hammer!(11+ / 0-)

I think Rove's numbers involve -1^0.5

"Who is John Galt?" A two dimensional character in a third rate novel.

• ##### And what's pushing his imaginary numbers(1+ / 0-)
Recommended by:

is evil^0.5, ie, \$\$

--
Make sure everyone's vote counts: Verified Voting

[ Parent ]

• ##### Thanks, Sam!(4+ / 0-)

I'd love to hear your thoughts on Obama's likely share of white voters.  Is the overall percentage being brought down by Southern whites?

Late Summer and early Fall, this percentage was talked up a lot on both sides of the political coin, but I see it mostly trumpeted on the Right now as evidence of an impending Romney victory.

I realize it's closely tied to the question of whether Likely Voter or Registered Voter models will prove more accurate in predicting the winner (i.e., how many younger, non-white voters will vote this time around).

But I also know that Obama is polling very well among white, non-college educated voters in Ohio and some other Midwestern states.

And I cling to the hope that the Republican Party's white/Southern strategy will suffer its last dying gasp in this presidential election, as the electorate continues to skew away from the old white majority.

Nobody puts Baby in a binder!

• ##### this is the last election(1+ / 0-)
Recommended by:
deminva

that is remotely winnable by the GOP with its current all white base.  The percent white in the electorate has fallen from 81% in 2000 to 75-76% in 2008.
What this means is that linear extrapolation approximates the white share of the electorate to be 73.5% in 2012 and 71% in 2016.
So the GOP needs 62% of the white vote this year for Romney to win.
It is possible maybe to get that among white males, but doubtful among white females.
To put it another way, Kerry got 37% of white males and 41% of white females.  Obama gets that this year, he wins handily.
Obama needs 38-40% of the total white vote to deny Romney the election.

However, population growth curves are rarely linear.  And the intrinsic systemic errors in the nat'l pollster sampling methods may be occluding the real number of whites in the population.

So I'm sayin' Obama in the nat'l vote by 4-5%.
;)

"When they ask how I died, tell them....still angry."

[ Parent ]

• ##### A larger fraction of the population is sampled(17+ / 0-)

in state polls.

The very thin sampling involved in national polls may make them subject to a wide variety of systematic errors. Some demographic groups may be systematically undercounted regionally in national polls.

look for my eSci diary series Thursday evening.

• ##### National polls error in the same direction?(3+ / 0-)
Recommended by:
wishingwell, TheOrchid, elwior

I'm having a bit of a problem accepting that the majority of national polls would all error as an undervote for the President.  It doesn't make statistical sense.

Something else has to be going on here.  It sure seems that the blue states aren't quite as blue and the Red states might be a little redder.

I would attribute this to the political advertising in the swing states that expose the Romney lies, and the false equivalency of the media which allows non swing state voters to more easily buy into the Romney lies.

• ##### Mention Cross Roads GPS too.(4+ / 0-)
Recommended by:
elwior, ChrisBix50, blackluck, quell

In PEC, Dr Sam Wang says at the end,

If you like what we do, but oppose who ActBlue supports, then go to CrossroadsGPS, which does just as good a job of targeting knife-edge races, but for Republicans.
We may not like to give free publicity to Karl Rove's outfit. But if Dr Wang said something, we should report it accurately, whether we like it or not.

We are Democrats, we are not afraid of reality. We like to get all the facts, even the ones that are not very pleasant.

So let us not edit out something that is on the record. If Dr Wang himself edited that piece out in DailyKos, then it is ok, I am out of line.

• ##### Yes, but it's a matter of where to end the quote.(2+ / 0-)
Recommended by:
poco, elwior

For the dKos audience, unlike the PEC audience, there's no need to extend the quote to include an alternative to ActBlue.

• ##### I would guess that that was Dr. Wang's editing, as(7+ / 0-)

he knows full well that Daily Kos has the stated goal of electing Democrats. His own site has the stated goal of accurate prediction.

Is it true? Is it kind? Is it necessary? . . . and respect the dignity of every human being.

[ Parent ]

• ##### I think that's a fair point(5+ / 0-)
Recommended by:
Rita in DC, RLF, elwior, Just Bob, MKinTN

Yes, evidently he edited it out himself -- and appropriately, since Daily Kos is not a place where Crossroads GPS contributors should be. But offering the alternative on the PEC site (while not pretending to be indifferent) is pretty cool, I think.

Election protection: there's an app for that! -- and a toll-free hotline: 866-OUR-VOTE
Better Know Your Voting System with the Verifier!

[ Parent ]

• ##### The PEC's findings as of Nov. 1, *8:02 AM EDT*(8+ / 0-)

The findings at the Princeton Election Consortium change several times a day. Here are the PEC's findings as of Nov. 1, 8:02 AM EDT:

As of November 1, 8:02AM EDT:
Obama: 308
Romney: 230
Meta-margin: Obama +2.36%

Probability of Obama re-election: Random Drift 95%, Bayesian Prediction 98.7%

• ##### As of Nov. 1, *9:59 AM EDT,* same findings. n/t(2+ / 0-)
Recommended by:
MeMeMeMeMe, elwior
• ##### As of Nov. 1, *11:59 AM EDT:* (1+ / 0-)
Recommended by:
elwior
As of November 1, 11:59AM EDT:
Obama: 312
Romney: 226
Meta-margin: Obama +2.64%

Probability of Obama re-election: Random Drift 96%, Bayesian Prediction 99.1%

http://election.princeton.edu/
• ##### Sorry! Link correction(1+ / 0-)
Recommended by:
elwior
• ##### comme si belle(2+ / 0-)
Recommended by:
elwior, AspenFern

this says it all

"For a bell-shaped curve, the average error is supposed to be 0.8 sigma. Here it's much larger, 1.9 sigma. Aha...here may be our culprit. Evidently, national polls have systematic problems. Answer: national polls do about 2.5x worse at predicting the popular vote outcome than expected if the wisdom of crowds of pollsters were perfect."

The nat'l polls are WRONG.
Take that Sean Trende!
engagé, redoublet, parré quatre, touché!

"When they ask how I died, tell them....still angry."

• ##### Given your professional knowledge about how we (1+ / 0-)
Recommended by:
elwior

M pick our observations to fit our beliefs, and knowing you are an Obama supporter, how do you counter the argument that you have merely cherry-picked your data points for your analysis?  I happen to believe you, but having some of the same professional knowledge, and certainly the same personal bias, I do worry about blue tinted glasses.

• ##### No cherrypicking, ever(10+ / 0-)

Because my standard policy is to take all the polls. I do not tolerate cherry-picking.

In cases where I take a subset, it is usually to prove a negative point. In this case the negative point is that the difference cannot be explained by taking a subset of polls.

• ##### Thanks. I especially like your second statement(0+ / 0-)

Kind of taking the "null hypothesis" approach.

• ##### this is hardcore nerdspeak(0+ / 0-)

The Central Limit Theorem and assumption of gaussian (normal) distribution should drive all polling to converge by election day.
Unless of course there is systemic error in the methodology of SOME pollsters.
;)

"When they ask how I died, tell them....still angry."

[ Parent ]

• ##### Nerd to nerd .... can you use gaussian assumptions(5+ / 0-)

about essentially non-random (or at least non-static probabalistic) events?  That has always been my uncertainty around all the assumptions we make about poll movement.  We are not watching brownian motion, or six sided die, or two sided coins.  We are watching large numbers of primarily emotion driven primates make decisions....

• ##### great comment!(3+ / 0-)

but I'm a heretic.
I'm a Talebian.  Mandelbrot over Gauss for carbon-based systems.
You should ax Dr. Wang.
;)

"When they ask how I died, tell them....still angry."

[ Parent ]

• ##### Nasseim Nicholas Taleb?(2+ / 0-)
Recommended by:
George3, enhydra lutris

And I wonder what a Mandelbrot "thingy" would look like if you put polling data into its first iteration?

• ##### What happens if you(1+ / 0-)
Recommended by:
elwior

view the state vs. national polls conundrum in light of the bootstrapping (or perhaps monte carlo) methodology for estimating large statistical trends from smaller populations?

In those methods, the one thing that absolutely must be verified before starting is that the subsets are a reasonable approximation of the larger whole.

Please excuse my rusty statistics vocabulary, and I hope that if I used the wrong terminology, you can suss out what I am trying to say...

Can we look at the difference in outcomes between state and national as an outcome of the non-transferrability of population characteristics? In other words, you cannot bootstrap the national vote by repeatedly subsampling localities, because the localities are not statistically similar to the totality...

The only way to ensure a free press is to own one

• ##### But why the obsession with 538 by the right?(3+ / 0-)
Recommended by:
elwior, ChrisBix50, rmx2630

The model is still predicting a 20% chance of a Romney victory - This is not an insignificant number.  And it is after all just a model; if Romney were to actually win given only a 20% chance by the model, 538 would be done anyway.  All a pollster or a modeler/aggregator has to recommend them is their record of accuracy, correct?  I'll admit, if 538 had Romney with an 80% chance of winning, before I slit my wrists I would be looking for a glimmer of light or hope somewhere, and hoping that the model was wrong. Howling at the moon makes a little more sense then howling at, you know, probability and math. (I would be contributing more dollars and helping out with the ground game in swing states!)

• ##### Exposure, surely(1+ / 0-)
Recommended by:
elwior

It's because 538 is of course now on the NYT website, and Nate's been  on TV a whole lot--MSNBC, Daily Show, Colbert, Bill Maher and more. Since those are left-leaning, he's a real target for righty yahoos. You can argue all you want about the NYT, but to the right it might as well be Pravda.

• ##### its just a new front (2+ / 0-)
Recommended by:

in the Conservative War on Science, Math, Reason and Empirical Data.

"When they ask how I died, tell them....still angry."

[ Parent ]

• ##### Romney & Co. will steal a close election(1+ / 0-)
Recommended by:
elwior

it's what they do without compunction.  When rich white guys steal it is not called stealing, it is called winning.  And that is all they care about, without apology.  And their paid puppets in the "media" will turn it all around and make the Democrats look like the cheaters.  All too familiar.

Obama, like Gore and Kerry before him, will accept ithe result in order to protect a process that the modern day GOP has perverted beyond all recognition. The GOP counts on Democratic pols being statesmen in the face of preposterous GOP usurpation of the electoral process.  It is the primary reason the GOP repeatedly feels so emboldened to undermine election after election.

While I certainly prefer Obama winning, I am not his biggest fan.  His continuation of Bush nat'l security policies, minimalist approach to Wall Street corruption (one might say coddling of Wall St.) is in my view why this election is close enough for the GOP to steal. We read here & elsewhere time and again how popular progressive policies are with voters yet the so-called progressive in this race has chosen to at most pursue the most watered down, corporate-approved versions of these policies.  As a result, we have a sputtering recovery & close election.

Close enough for the GOP to again steal.

Hope I am wrong, pray I am wrong, but Wednesday next week is looking to be an awful day.

• ##### cheater detection(0+ / 0-)

im sure the right would like to cheat, but unsure that they have ever been successful at say...flipping the vote in Ohio in 2004....  involves too many other humans who could make big money with a tell-all.
They cheat in little ways, like voter suppression and intimidation, and redistricting.  that stuff flies under the radar.
But stealing an election?
Too much scrutiny involved, too many error components.
Sure if they thought they could get away with it.
But the payoff/risk matrix doesnt work.

"When they ask how I died, tell them....still angry."

[ Parent ]

• ##### 2 things(0+ / 0-)

1) if successful the cheaters control the appartus that would otherwsie be investigating the cheating, ans
2)  Dem opponents do not contest the results for fear of undermining confidence in elections.

and a third for good measure, in light of 1 & 2

3) anyone else who conducts studies or otherwise sets out to prove the stolen election is dismissed as a crackpot consiracy theorist.

I saw th evoter suppression an dresulting long line in OH 2004 first hand while working a precinct in a largely African-Americna community in Cincinnati.  Previous year's mayorl race 5 machines, 2004 Presidential 3 machines 2 of which did not work right.  People left line for work, to drop off kids, take care of parents etc.  I expect the same tactics given current OH GOP leadership in the state.

But Fla 2000 happened, we all watched it live.

• ##### Re: What Karl Rove knows...(2+ / 0-)
Recommended by:
elwior, rmx2630
Sam Wang: Karl Rove predicts that Romney will win - a case of Ro-mentum if ever there was one. On what does he base this claim?
There is, of course, yet another possibility. Renowned Election Engineer Karl Rove knows what’s being cooked in the swing states. In 2010 he engineered the Republican take-over of the House -- but perhaps more importantly, he was also the mastermind behind the GOP winning control of as many swing states as possible.

We know a fair bit about how Ohio’s secretary-of-state Jon Husted fought tooth-and-nail to curtail early voting, even appealing to the US Supreme Court.

On the excellent site where he keeps a running tally of early votes, Prof. Michael McDonald has found it necessary to comment alarming discrepancies in Ohio. I quote:

The Ohio Secretary of State's office Sent me their county level spreadsheet, current as of Oct. 2t6. I do not know what to make of it. There are 17 counties reporting fewer early votes as of Tuesday, Oct. 30 than were reported to the Secretary of State's office on Friday, Oct. 26.
Read that again; it seems that votes are being lost! (my bold emphasis) If this is indeed the case, and it’s not just arithmetic incompetence at Mr Husted’s office, I think we can safely say that Karl Rove has special insights into what is going on in Ohio.

• ##### Oh please(2+ / 0-)
Recommended by:
elwior, ChrisBix50

Karl Rove is not a genius,mastermind or anything of the sort.

• ##### Wouldn't sample size be the main factor ...(1+ / 0-)
Recommended by:
elwior

in the discrepancy in accuracy between state and national polls  ? e.g. OH has about 11.5 M people. Approx 27x less than the US population. A OH poll with a sample size of 500 has a sample size 14x larger than a typical nationwide poll with a sample size  of  1000.  Seems to me that huge discrepancy in sample sizes would be the major  reason why state polls are more accurate than national ones.

• ##### Thanks for all the sexy facts, Sam n/t(2+ / 0-)
Recommended by:
elwior, ChrisBix50
• ##### Polling questions(1+ / 0-)
Recommended by:
elwior

I've worked on local campaigns and have some knowledge about how polling is done for those, but I'm not sure how methods differ for state and national polling...

For the local campaigns, my understanding is that the "likely voter" screen is done via voter records -- ie, people who don't meet the pollers likely voter criteria don't get called.  Is this different in national polling -- most of the discussion suggest that the likely voter screen is based on answers to certain questions?

Also, for our local campaigns, party ID would be based on voter registration info, but most discussions of current polling suggest that party ID is based on self-identification.  Is this correct, and if so, why?

Is there any chance that the methodologies vary for state and national polls?  At the state level its probably fairly easy to get a self-consistent voter database.  It must be harder at the national level.

• ##### another partial explanation: does my vote matter?(1+ / 0-)
Recommended by:
elwior

there is evidence that potential Dem voters are more likely, than Republican ones, to sit out the election. But if they feel that their vote counts (i.e. it's a close races as in swing states) then they are more likely to actually go vote. So I think the closer nature of the presidential election in swing states could explain some of the discrepancy between swing state polls and national polls.

here's some text from a draft diary that I was working on (but I never tried to crunch the numbers in order to test the above theory).

a poll (from USA Today/Suffolk University). It found that among non voters, Obama has a 43-21 advantage over Romney. (This agrees with another question about the 2008 election, where these same non voters said they voted 44% for Obama to 20% for McCain, with 32% not voting.)
The survey identified one extremely persuasive argument. Among Obama supporters, 85% [of non voters] say they would go to the polls if they knew their vote would help swing a close election to the president; 70% of Romney supporters say the same for their candidate.
The poll also found that 69% of these potential non voters were actually registered to vote.

-- Romney on contraception (Planned Parenthood TV ad)

• ##### Will more dems vote this year than in 2008?(1+ / 0-)
Recommended by:
elwior

It is an interesting point Rove is making - why is Dem party ID going up omong likely voters compared to 2008? Are the old white guys dying off?

That said - I think his analysis/case is pretty thin - Rove's cherrypicking national polls to get Obama to 47% (he's really closer to 50 in a lot of polls) and it rests on basically all the polling outfits - other than Rasmussen and Gravis Marketing - being wrong.

I'll take that bet. I wonder why Rove is taking the high risk bet.... I guess he has to justify all the ads he bought with other people's money.

I look forward to Sheldon Adelson and others asking Rove "what the hell - we gave you hundreds of million dollars and you didn't win!"

Hope the prince of darkness and his data minons are wrong... Going to be interesting...

• ##### Tired voters.(2+ / 0-)
Recommended by:
elwior, rmx2630

I think at least some of the discrepancy between swing state and national polling is due to swing state voters who are completely tired of answering polling calls and are just shutting down.  Perhaps only the I-want-to-make-my-opinion-known subset of voters is still answering the calls.

The flip side of this hypothesis is that you would normally expect such people to be randomly distributed across the political spectrum, in which case you'd expect to see an equal number of D and R voters affected.  But that may not necessarily be the case.

The road to Hell is paved with pragmatism.

• ##### Does a national poll(1+ / 0-)
Recommended by:
elwior

measure the popular vote or the majority vote state by state?

48forEastAfrica - Donate to Oxfam> "It is better to light a candle than to curse the darkness." Edna St.V. Millay

• ##### can you add carmona to your act blue page?(1+ / 0-)
Recommended by:
elwior

You mention that  a donation to him is high impact, however you don't have him on your page...

I went there to donate to the other candidates you have listed, but was disappointed not to see him...

"We have always known that heedless self-interest was bad morals; we know now that it is bad economics" - F.D.R.

• ##### Good point BB. The Arizona Senate race is a very(1+ / 0-)
Recommended by:
exMnLiberal

tight one, where we've got an excellent (Latino) candidate, Dr. Carmona, running against a terrible (what else is new?) Republican candidate.

"We the People of the United States...." -U.S. Constitution

[ Parent ]

• ##### Nat'l polls not getting good sample(1+ / 0-)
Recommended by:
elwior

So, essentially the argument here is that while state polls manage to get a good, unbiased sample of voters in their state, national polls are not doing a good job of getting a good, unbiased sample of voters in the whole country?

This makes sense. Getting an unbiased sample for something as unhomogeneous as the US has got to be difficult.

• ##### Yes, it would seem to be a much simpler endeavor (0+ / 0-)

to poll the state of Ohio, Iowa , or Virginia e.g., than a vast,   sprawling nation of 315 million people.

"We the People of the United States...." -U.S. Constitution

[ Parent ]

• ##### I think Colorado(0+ / 0-)

Really is the unknown quantity here. Although most polls favor Obama to eek out a win there, the Republicans still lead by almost 3% in early voting and I have heard Johnson is taking votes from Obama due to his stance on legalizing Marijuana in CO.  Thoughts?

• ##### ROMNEY a socicopath biochemically(0+ / 1-)
Recommended by:
Hidden by:
mindgeek

SO many people wonder about flip-flops, etc and do not understand how the brain works or does not work. My name is Ronald Dishinger- my web site is www.biochemimbal-behavior.com. For many years I have been trying to help schizophrenics, bi-polars and pathogenic (sociopaths) personalities. It seems that many do not even know the word pathogenic-sociopath- meaning the person is out of touch with their feeling- they are  always trying to be someone or something other than who they really are. There are many who are pathological liars or change the personalties very fast to protect themselves as is now called "flip-flops". So an additional part that is very bad is that they do not feel pain to any degree the way in terms of deep pain when something is horribly wrong- They only feel as a mere change again to something else- They simply can not feel pain very deeply- partly because it is their protective devise and they use it as shield for the pain of searching for their true personality. It seems easier for them and they can and do change anytime they need to so pain is again deflected away from them. We do not know where it starts but certainly in very early childhood and continues for years and even an entire lifetime. And others simply see it as that persons’s personality. There are many, many ups and downs trying to live with a pathogenic-sociopathic personality. The farther extension many can easily see such as a sociopath who kills people, cuts up their bodies and then goes to supper and a night out on the town. The sociopath felt no pain for his deeds. So now we have politicians who are also sociopaths now and have had some for some time in our government- large or small. Sociopaths are not as rare as bi-polars or schizophrenics but are not the ones I would vote for ever if I had a choice- Then I do not know what I would do.Why does the mass media not know this?? Thousands of mormons are trying to get out of the Church and can not because of various controls put on them even when the people decide they do not wish to be a mormon any longer- What is control issues if not this.WHY HAS THE MASS MEDIA NOT EXPOSED ROMNEY?? DEALING WITH SOCIOPATHS IS VERY DIFFICULT SINCE THERE IS NO TRUTH WHATSOEVER SO NOTHING TO HOLD ON TO FOR OTHERS TO TRY TO HELP THE SOCIOPATH OR A COUNTRY. THEY WILL GET WHAT THEY WANT OR DESTROY EVERYTHING OR ANYONE IN THIER PATH. THIS IS A MAJOR BIOCHEMICAL ILLNESS.It is not about who is republican, democrat, independent or seaweed but about how people's brains work or do not work-This has been allowed to go way beyond any scope I thought possible-Ronald Dishinger www.biochemimbal-behavior.com

• ##### check out Helen Fisher(0+ / 0-)

the love guru.
that what we call her in class, because of her work on the kinds of love.
http://www.thedailybeast.com/...

"When they ask how I died, tell them....still angry."

[ Parent ]

• ##### Please comment on Newsweek article(0+ / 0-)

Sam, could you please comment on this article in Newsweek regarding the gender gap - Why the Gender Gap Won't Matter in this Election? It makes me a little (more) nervous.  Thanks.

• ##### Even the Votemaster!(0+ / 0-)

Wow, I must I'm surprised that even electoral-vote.com is saying "it's all tied"!

"No presidential race since 2000 has been this close and there are so many late-breaking factors to consider this year, including the debates, Sandy, and the final jobs report tomorrow, that only fools like Morris or dyed-in-the-wool partisans are daring to make a prediction."

Pretty disappointing, I thought he would know better...

• ##### Just to say thanks and I gave to PEC(0+ / 0-)

I had a blackout about pec until fairly late. Now that I'm away for the two weeks before the election ... In time to the nitr before and Election Day.

Putting you and Nate together provided a. Measure of comfort over the difficult days . But I like the activist approach. I' tapped out or nearly, but felt I had to chip in on the pec list because while they were my biggest priority, it obviously is anyway and it fits. Next time more.

I think this country goes to he'll very quickly without a safety net, social sec and Medicare (I'm retired) but feel just as passionately about science, facts over hype and bs and obviously what you and crew do in your real lives. Oh, and then there's Paul Krugman, the national treasure who called it bs before anyone else had the guts.

• ##### Dr. Linzer Agrees(0+ / 0-)
What do we find? This set of kernel density plots (smoothed histograms) shows the distribution of survey errors among the seven largest survey organizations, and in grey, the distribution of errors among the set of smaller firms. The smaller firms’ error distribution matches that of Quinnipiac, SurveyUSA, YouGov, and PPP. The right-leaning firms – Rasmussen, Gravis Marketing, and ARG – are clearly set apart on the pro-Romney side of the plot.
I left the link to your Rovementum post at 538, pollster, and votamatic.  And I tweeted it to Sean Trende.  I will remind you that I told Nate months ago that he needed to use nonparametrics to estimate house effect because of asymmetrical political behavior.
/slightly smug

And I cannot comment on Facebook.
I'm an Assangian-- Facebook is evil.

"When they ask how I died, tell them....still angry."

• ##### TYPICAL ROVE LIES(0+ / 0-)

The Karl Rove has been involved in dirty politics since 1971. Bugging the office of his own candidate then calling the press to smear him. Rove learned from the watergate crew and doesn't make mistakes. Typical to his lies and record sums of GOP money, he tried to spin a last minute steal of California in 2000 - this year it is Pennsylvania. Karl Rove is a master liar.  Karl Rove is a sociopath who has singlehandedly lowered the integrity of our elections and the belief in government. I don't see any TEA Party members complaining about FEMA in the aftermath of Hurricane Sandy.

• ##### Rove the Prophet(0+ / 0-)

So Rove announces a Romney victory and nobody bothers to mention that a few weeks before the 2006 election he also announced that the Republicans woold retain control of Congress.  And he was such a big shot that all the Democratic leaders were convinced that he had an October surprise.  But he had nothing more than what he always had: a good sales pitch and when it works he's a genius and when it doesn't work everyone forgets that he was wrong.

Rove made himself a "player" in this eledtion by running around the country and telling a bunch of senile billionnaires that if they would trust him with their money he'd buy them a Romney vistory.  What's he going to say now?  That he pissed it all away on a losing cause?