Princeton Election Consortium

Innovations in democracy since 2004

Outcome: Biden 306 EV (D+1.2% from toss-up), Senate 50 D (D+1.0%)
Nov 3 polls: Biden 342 EV (D+5.3%), Senate 50-55 D (D+3.9%), House control D+4.6%
Moneyball states: President AZ NE-2 NV, Senate MT ME AK, Legislatures KS TX NC

KCRW – To The Point with Warren Olney

November 5th, 2012, 4:00pm by Sam Wang

On KCRW-FM Los Angeles (and by podcast), some good discussion about national vs. state polls.

Ron Brownstein (National Journal), Sasha Issenberg (The Victory Lab), and Mark MacKinnon (Bush 2000) were very good on how polarized the electorate is. Sasha Issenberg speculated that “fewer than 10% of minds can be changed.” I’d put that number far lower. If you look at the Meta-Margin graph, the swings of this year’s race could be explained by as few as 3% of people changing their minds. That’s assuming none of the ups and downs come from changes in voter morale on either side. The swing is half what it was in 2008.

History of Popular Meta-Margin for Obama

Also, Mark MacKinnon suggested that Hurricane Sandy “stopped” Romney’s momentum. As I have pointed out here, Ro-mentum came to an end three weeks ago. However, I am starting to see the outlines of a story along the lines of “the dog ate Romney’s homework.” Maybe this dog?

Tags: 2012 Election · President

152 Comments so far ↓

  • Paul

    Agreed — and I don’t even think it’s very many people changing their minds. It’s enthusiasm, which sways likely voter models.

    My gut feel: less than 1% of voters who had opinions in July have a different opinion now.

  • Howard Roark

    I’ve noticed there’s this tendency that “pundits” have to say “well it can’t possibly be a 98% chance” because “anything can happen” and nothing can be “so certain”. Mark McKinnon used this line of reasoning. Why is it so difficult for them to see that yes, we DO understand that “unforeseeable” events could happen that could change the outcome of the election, but based on history & statistics, we can assign a definite probability to the outcome?

    • badni

      I think 98% is a definite probability that excludes massive surprises.

      98% says that with current polls and a reasonable amount of drift, O has a 98% chance. And maybe, the day before election day, that is right.

      But a week out from the election, I would be willing to bet that if we waited through a few hundred election cycles, and looked at all the ones that were this close at this point in the cycle, more than twice in a hundred elections there would be something in the last few days that changes the outcome. Two of those results would be polling errors and random drift.

      The others would be things like war breaking out, an election-day hurricane, a true October/November surprise such as the leading candidate’s tax returns being released and showing he took part in an amnesty for tax shelter abusers. I don’t think the 2% accounts for those.

    • Joel

      “The others would be things like war breaking out, an election-day hurricane, a true October/November surprise such as the leading candidate’s tax returns being released and showing he took part in an amnesty for tax shelter abusers. I don’t think the 2% accounts for those.”

      You are exactly right. Those events (I’ll exclude war, because usually there’s a buildup) are black swan events. Unpredictable but explicable in the aftermath.

    • theDAWG

      There is a 98% chance the Kansas City Chiefs will not win the Super Bowl this year.

      Is it okay to say that? If yes, then we’re just talking about a gradient of probability.

    • DPB

      That’s pretty standard from a journalist. Objective measurement dehumanising the world and removing an essential essence which only the poetic can touch.

    • Paul

      What a muddled article. Gerson doesn’t seem to understand the difference between discussing what *should* happen and discussing what *will* happen.

      The first is all about feelings. The second is all about analytics.

    • Ross C

      Ah yes, the pro-torture Gerson opines on how nasty science intrudes on our humanity. You know, it’s like how climate change can’t be real because it feels wrong, and stuff.

    • Wheels

      I love that he said “tell the college kids there’s a job out there where they’re always 100% correct.” Yeah, sort of like being a WaPo columnist who never has to account for facts or previous statements.

    • jd

      Once you read his bio, you will understand why is anti statistics. You are right is column is laughable

  • Howard Roark

    @rags Thanks for the laugh! I continue to be amazed by pundits who think they live in this fantastical country in which people wake up every morning and change who they’re going to be voting for based on sound reasoning and a tough decision-making process! This is simply asinine and removed from reality. As Sam has pointed out earlier, a vast majority of people (90%+) have very set opinions about who they’re going to vote for in Presidential elections. Whether they are willing to share their opinion is another story.

  • Denis

    Howard, the fantastical America that these pundits live was appropriately coined by Jon Stewart as “Bulls**t mountain”

  • Michael Worley

    Has Sam’s geek guide been posted?

  • CZL

    @rags — shorter M. Gerson: “Numbers are evil and this is an election about feelings, not vote counting.”

  • DelMartian

    That’s a yellow-dog Democrat, if I ever saw one!

    Been lurking since 2008. Love this site, Sam, and the commenters.

  • orchidmantis

    I think we should encourage the “Sandy did it” model as an explanation that leaves conservatives moping about the weather, rather than hanging anyone in effigy. I’d tired of anger and bargaining.

    My hope is that after a good long mope the purge that I expected in 2008 will finally start. That clear-headed people will look at the makeup of the crowds behind each candidate and conclude that the Republican base needs to start looking a lot more like the Democratic base, and figure out the serious policies to get there. But I expect we need one more cycle on the outs.

    • Howard Roark

      I’ve been making the following argument to conservatives around me: “since God is responsible for Sandy, and Sandy is responsible for the coming Obama victory, then God is on Obama’s side”.

      The dumb ones really have a hard time dealing with this idea. The smart ones get a good laugh out of it.

    • Ross C

      Orchidmantis, I’ve been saying that since 2004, when Bush was re-elected. I said, if Bush doesn’t completely destroy the country by 2008, he will have done such damage that the Republican party will never again control the White House or a branch of Congress. He didn’t quite destroy the country, but by 2010, they had the House. If I was Mark Twain or HL Mencken, I’d have something trenchant to say about that. “People are idiots” is the best I can do.

  • securecare

    Purrrr (rub, rub).

  • Billy

    Sam, I was wondering if in future years you can include the poll variance for each state’s median margin. Lately FL, CO, VA, and NC have been bouncing around and that can be due to (1) roll-off of old polls and (2) a large variance between polls. It would be nice to have that embedded into the sidebar since you do report it on some blog posts…

    Or maybe it should be included as a “final state-by-state election prediction” before tomorrow?

  • Martin

    Florida back to R? Really? Today’s polls from FL looked pretty decent for O. I don’t know how much cred you give to Zogby but they had a good number in FL for O and so did a couple of others. What’s the deal Dr. Wang???

    • Joel

      @ Martin: here’s the breakdown.

      Obama’s numbers in the last 6 polls:
      46, 47, 47, 47, 50, 50
      Romney’s numbers in the last 6 polls:
      45, 45, 46, 48, 49, 52

      Average the middle two numbers in each series and you get 47-47. If you want to include “outliers” you have an average margin of O 47.8 to R 47.5. Damn close no matter how you look at it.

  • john

    Dr. Wang, I NEED your geek guide to 2012!!!! Can’t go election night without it! Thanks for all you do. You’re the best!

  • Jay Bryant

    Good of you to write so many posts here in the last days of the election, Dr. Wang.

    Thanks for all the effort.

    (I’ll be checking in after the election, too – just saying thanks because you’ve been very involved with your audience.)

  • Stephanie

    I was looking at the Florida numbers, and it seems to have settled into the pattern that the national poll numbers were last week (almost all the polls for a candidate had a companion poll that put their opponent up by the same margin)- that was pretty unanimously declared a tie for the nation, why is Florida in Romney’s column (barely) now? Why would it not remain a toss-up? I do think Romney has a good shot at winning Florida, but I don’t really think Obama has less of a shot by much given the most recent polling. Is this your way of taking the ridiculous voter suppression tactics into account in Florida into account Dr. Wang? Or am I missing something? I am probably missing something, but I would appreciate an explanation none-the-less.

  • Shawn Huckaby

    I posted here right before the crash that there were already signs of the “Sandy” narrative starting to bounce around the twitterverse at that time. It appears denial isn’t just a hurricane in Egypt.

  • AlpsStranger

    Let’s go boys!

    May history forget the name Mitt Romney after tomorrow.

  • Montykoolaid

    Hey, quick question.

    My office manager (who is a democrat) just told me her prediction.

    It will be way too close to call tommorow night, that we wont know for a few days, is that possible? Do you guys think that will happen?

    Also, thank you Sam, for keeping me semi sane until we finally know.

    • Suja P

      @Monty: Possible in some states, but unlikey to figure out who won the whole thing.

    • Ralph

      Take another look at the polls and consider that Johnson draws preferentially from Romney. It would seem that Goode and Stein draw preferentially from the undecided. (I assume that all draw from Obama but the numbers are small). I don’t think it will be close, unless there is widespread voting fraud at the machine level.

  • Pat

    I know the purpose of your EV estimator is meant to ignore the precise state-by-state wins and only derive a final number based on the win probabilities in each state.
    Still, will you give us your thoughts about how the closest states (CO, VA, FL) will break? Eagerly waiting for the final prediction!

  • GeoLove

    Sam… it’s fun to see you have a sense of humour. Enjoyed your link to the dogs’ Facebook site. In fact, I’ve enjoyed your math (statistics? probabilities?) lessons, too. Wish I had you as a prof for my MBA work at Kent State… would have been a better class I imagine.

  • Joel

    Florida is going to be interesting. Looks like it’s 47-47 (Obama) versus 46-48 (Romney) at the medians.

    • Craigo

      I was about to post the same thing. I really think Romney’s going to pick off NC and FL, but nothing else.

    • Craigo

      Oh, IN, duh. I forgot about it because it never seemed remotely in doubt.

  • Fred

    I see Sam has FL tied. I think we are going to win it but I also think the GOP is going to do their best to try steal it 2000 style.

    I really thought we were going to get NC again but that one is not looking so good.

  • zenger


    James Pethokoukis tweet: “My reporting = those Romney internal polls are the real deal. If Rs want this, they can take this. More to come …”

    He tweeted something similar about Team McCain internals showing him closing fast to win. We know h0w that went.

    Question: Has there been discussion of internal polls here? Would there be credible reasons for such polls to be more accurate?


    • Howard Roark

      According to Nate Silver, internal polls have an average 6-point bias in favor of the person releasing them.

    • Craigo

      Campaigns generally do a very good job of polling – they have every incentive – but they will never deliberately release unfavorable results, and if it’s leaked you have no way of knowing whether they are accurate. They should never be aggregated with public surveys.

    • ChrisD

      Here are those leaked internals, according to Race 4 2012, my go-to right-wing website:

      R+1 in Ohio.
      R+3 in New Hampshire.
      R+2 in Iowa.
      Tied in Wisconsin.
      Tied in Pennsylvania.
      Losing in Nevada.

      Meanwhile, Rasmussen has Obama at +2 in NH.

    • Matt McIrvin

      So this is why Romney is campaigning in Pennsylvania tomorrow?

      Oh noes, he’s flipping it! Better spend more time there to make me cry extra hard!

    • Joel


      The most right-leaning pollsters have had a house effect of what, +5 this cycle? If you apply that lean to Romney’s numbers it makes a lot of sense.

      I will say that if those numbers *were* true, Romney wouldn’t need PA, right? He would just be doubling down his campaign in Iowa and New Hampshire.

    • theDAWG

      Internal polls are like corporate pharm trials; run 10 of them and publish the best two.

    • David

      So if Nate Silver is right about that 6 point bias that would give us…

      R+1 in Ohio. >>> Obama +5
      R+3 in New Hampshire >>> Obama +3
      R+2 in Iowa. >>> Obama +4
      Tied in Wisconsin. >>> Obama +6
      Tied in Pennsylvania. >>> Obama +6
      Losing in Nevada. >>> Obama winning big.

      Those match up to the poll averages pretty well. So it looks like a poll that’s +5 or +6 Romney. I’m guessing it’s their magic feel good poll they give to donors and pundits, as opposed to the bleak one only for the top strategists.

  • Matt

    What I don’t get with all this Romney momentum and what not is that these professional pundits presumably have access to the same information that I do. I mean even simple averaging sites showed the momentum halting weeks ago. Why is it that person with a fulltime job outside politics like myself is well aware of this fact and the pundit class seems completely clueless about it?

    • Jay Bryant

      Because “Mitt’s Dead in the Water” doesn’t put eyes on their content for weeks on end, while “It’s Gonna Be a Squeaker!” does. Remember, nearly all journalism is done by companies for profit and by journalists who want to remain employed by those companies.

      “So much for Objective Journalism. Don’t bother to look for it here–not under any byline of mine; or anyone else I can think of. With the possible exception of things like box scores, race results, and stock market tabulations, there is no such thing as Objective Journalism. The phrase itself is a pompous contradiction in terms.” – Hunter S. Thompson (on whose works I did my Master’s thesis, long ago)

    • Suja P

      They have to blather on about something or the other for weeks on end. And ‘all the math points to this guy winning’ does not make for riveting narrative.

    • Peter Principle

      To quote Upton Sinclair (for about the zillionth time): “It is hard to get a man to understand something if his salary depends on him not understanding it.”

      Simplest explanation of Gramscian hegamony I ever heard.

  • Chris Bixby

    Thank you Dr. Wang for saving my sanity. Can’t wait until Wednesday.

  • zenger


    Motivated reasoning. They all want him to win. They all believe he should win. They cannot believe data that says otherwise.

    So they go with Morris Albert:

    “Feelings! Whoa! Whoa! Feelings!”

    “According to Nate Silver, internal polls have an average 6-point bias in favor of the person releasing them. ”

    Saw that tweet, which made think this must be an area of study. Are internals routinely released at some point?

    • David

      One set of internals are given to favored pundits and donors often. That’s not the set that the campaign works on usually. Although the Romney campaign has struck me as more delusional than most since they let Obama freely define them in June and July.

  • Craigo

    Public service announcement for tomorrow night/early morning: The popular vote count will probably not be accurate until a few days after the election. So don’t tear your hair out or lambast Sam if the Meta-Margin is off as of tomorrow at midnight.

    • Craigo

      As mentioned elsewhere, campaigns have no incentive to report objective poll results. If anyone leaks Obama’s internals tomorrow, you should ignore them as well.

  • SoleburyJim

    Looks like is beginning to creep up for Obama–69.3% as of this writing, up a few points this afternoon.

    • Craigo

      Intraders are mostly conventional wisdom types. The number say Obama, the pundits say tie – so they resort to the golden mean fallacy.

      There was interesting article at Buzzfeed by a heavy investor, who mentioned that in 2008 an investment continually inflated McCain’s price by buying in the middle of the night at high prices. Intrade’s volume is undoubtedly greater now so this is less likely to work, but it’s revealing of how sensitive these markets are to poor information.

  • Brian

    Can someone point me to a good analysis or essay regarding internal polls, and what their value or worth is? Most ly they sound like nonsense. Thanks in advance.

  • RandyH

    Love the FB link Sam thanks,2 paws up!:)

  • wheelers cat

    well…this has been a great season.
    bien merci Dr. Wang.
    my only regret is that i didnt get quite enough Tapen and Amitabh….but then again, who does?

  • RandyH

    So is today the last day of national and swing state polls or tomorrow another dump?

  • piktor


    Your prediction at the Olney radio show was Obama 303 EV. Florida is still in play, so president Obama could win by a larger margin. Is this your final prediction?

  • mediaglyphic

    Gallup is caving. New Gallup tracker has Romney+1 for LV and Obama +3 for RV. This is a +4 move in the LV since the last tracker Oct 28. and +3 for RV.

    THis is a big swing in Gallup.

  • Osso


    According to the “foreign” Press, 90% of voters in Ohio today were wearing Obama hats/buttons.

  • Peter Principle

    No, this dog:

  • Bruce Wayne


    Will you be making an official final prediction?

    My swing state predictions are as follows:
    Obama: OH,PA,IA,NH,CO,WI and NV + usual blue states for at least 290
    Romney: NC + usual red states for at least 206.

    Networks will call it for Obama by 9pm ET or whenever the IA results are projected.

    FL and VA will be decided by less than 1%. FL probably won’t declare a winner until the following day.

    Gun to my head, I say Obama takes VA and Romney takes FL for a final total of
    Obama: 303
    Romney: 235

    • Sam Wang

      Yes, tonight. We are trick or treating here. Halloween was postponed.

    • Reason

      Yep. That is how I called it as well. He wins Va and barely loses FL. I think he also barely loses NC.

    • 538 Refugee

      Sam, you going out as Mitt Romney to scare your neighbors? Hope you get lots of good candy. :D

  • PhilC


    I accept the 99% likelihood of an Obama win and think it a better estimate the 538 blog. I am more skeptical of the predictability of the EV count, despite your stellar record.

    The part that confounds me is how to weigh early voting.

    My uneducated thought is it would seem to require accurate knowledge of the number of votes cast in each state each day, distributed on the unit interval, to use as cumulative weights.

    I also don’t understand if the pollsters include “already voted” as “likely voter” or simply don’t count them, or whatever. Or does it not matter? And if not, why not?

    • Osso

      I totally agree as we need to have two sets of Polls, one for States with Early votes and one without (Nationa Polls contain too much noise)….

      Early vote States ratio to Election day vote States is almost 2:1 and that by some reports Obama is leading the Early vote by 2:1 ratio as well….

      My guess is that most Pollsters will have egg on their face on Wed.

      My prediction: Obama will take 54% of the vote!

  • 538 Refugee

    Listening now. My blood is boiling at the “exit polls were wrong” statement. I think the exit polls were right. It was the counting of votes that was wrong. I used to be of the opinion that pollsters shouldn’t do exit polls until then.

  • E L

    Oh noes! Looks like we scared our pet troll off. Let’s feed him if he comes back. We can rename him Seamus.

  • NickinLA

    I am a liberal and am trying very hard to really understand voter suppression. If a state government is trying to suppress the democratic vote by limiting hours, etc. aren’t they in effect suppresing ALL VOTERS? Or, are they only doing this in inner city neighborhoods?
    Wouldn’t limiting early voting means long lines tomorrow? No matter who you’re voting for?

    • mediaglyphic

      Nick, it depends on the specifics. Less affluent voters have less flexibility in terms of being able to vote (especially if your boss is the koch bros or the florida versailles character). If someone wants to rig the game they can do so, by knowing the lay of the land. But they should expect a backlash and i think the ground game is giving them one this year.

    • 538 Refugee

      It is harder for poor people to vote especially if it is just one day after work. In many cases the polls may only be open for an hour or two past quitting time. If you’ve done hard manual labor or stood on your feet all day this makes a difference. By having weekends and alternate days you increase the chance they can get to the polls.

    • Brian

      Because Florida for example is becoming notorious for squeezing multiple large urban areas into a handful of polling places. You don’t have these problems in rural traditional Republican areas

    • Khan


      They are doing it in Democratic voter regions. For example, in Ohio the SOS increased early voting hours in Republican districts and decreased it in Democrat districts. Obviously the court smashed it into pieces, but to answer you–yes they are only doing it in heavy Democrat voting regions.

    • NickinLA

      Thanks guys. I had a suspicion about this but if it sounds illegal, why hasn’t anyone done anything about it? Why hasn’t the Attorney General involved himself in this? I am not talking about Voter ID law, I am talking about early voting per se.
      If voting is a right given by the federal government then why is the state government allowed to suppress this law? Obama faught the AZ SB1070 law then why not this?
      It’s frustrating!!!

    • Craigo

      No. Urban precincts are more crowded than those in the suburbs or rural areas, so cutting back hours means longer lines in those places (in Ohio, thousands of people were still in line in places like Cuyahoga County when the polls closed).

      Republicans are more likely to be able to take time off to vote – which is especially critical when you live in a precinct with long lines. Early voting helps Democrats whose voters have more trouble getting to the polls on a workday.

      Secretaries of State and election supervisors can exacerbate these problems by allocating too few voting machines to Democratic-leaning precincts – as happened in Ohio in 2004, and this year in Miami-Dade.

      Ohio also attempted this year close the polls early in Democratic counties, while keeping them open later in Republican areas – a blatant violation of equal-protection.

      And there’s good old-fashioned tactics like calling or mailing Democratic voters to give them the wrong date or polling place, both of which have occurred in Arizona this year.

    • Khan

      @ Nick

      The courts are very actively engaged in this. Most of the suppression has been stopped thanks to quick legal action by the Democrats.

      The FBI is investigating a lot of GOP suppression, including one that just happened today in Arizona where Jeff Flake’s (GOP) campaign robocalled Democrats and told them to vote in the wrong places.

      The US AG is a very politically precarious job. He cannot appear to be partisan, even when he has complete legal standing to do something.

    • 538 Refugee

      NickinLA. The SOS in Ohio actually issued instructions that were counter to state law hoping to be able to disqualify ballots on technicalities. Like a wrong or illegible digit when recording a drivers license number. State law says the poll worker does this so there is no question about it. The SOS said the voter must fill that in.

    • Craigo

      Early voting is not a right, it’s just good policy. (Republicans have been waging a rhetorical war to define voting as a “privilege,” despite the righto to vote being mentioned in the Constitutiona more than any other.) And if you provide it to all some within a state, you have to provide it to all, pursuant to the 14th amendment.

    • Craigo

      Jon Husted appears to be trying to set a record for most ass-kickings in federal court this season.

  • Wonder

    I don’t understand what you mean by this: “That’s assuming none of the ups and downs come from changes in voter morale on either side.”

    What is the distinction you’re making between changes in voter “morale” and voters changing their minds? Can you clarify?

    Also it would be interesting to see the 2004, 2008, and 2012 graphs overlaid, to get a sense for the relative variations. Is this graph available somewhere? Maybe I missed it.

    • Khan

      I assume Dr. Wang is referencing the fact that voters who have high morale are much more likely to be engaged and responsive to polling. That translates to higher passing through the LV screens and a more pro-whatever candidate they support poll.

      In that case, the actual break down in the electorate hasn’t really changed at all, only the morale of the voters, which polls cannot accurately interpret. You get things like Romney seeing a 5-10 point swing in the polls after the first debate. 10 points is 15 million people. Obviously, 15 million people didn’t change their mind about their candidate of choice.

    • Wonder

      Khan, thanks for your answer, but this still doesn’t make sense. The only thing I can infer is that the distinction is between (A) voters who switched from being lukewarm or undecided about a candidate to being more strongly behind a candidate, and (B) voters who switched from one candidate to the other. Both of those seem to me to be examples of changing one’s mind.

      If that is, in fact, the distinction that’s being made between a change in voter “morale” and a change in voter “mind,” then the 3% number that motivated my question does not seem particularly meaningful.

  • E L

    Screwin’ with Mitt:
    Obama connected Bruce Springsteen, Chris Christie during call from Air Force One:

    [Sorry, Sam. I couldn’t resist or as Oscar Wilde said “I can resist anything but temptation.”]

  • Princeton Voter


    As a Princeton alum, I am very proud to see such a thoughtful and empirical study of electoral politics done at Princeton.

    Question to Sam or anyone else on this board:

    How do we reconcile the new polls that just came out on party affiliation (showing that more Republicans will participate in this election) vs. poll results based on turnout mirroring 2008 election (more Democrats)?

    Shouldn’t we essentially update pretty much all polls done in the last 2 weeks using a higher weight toward higher Republican turnout, which would essentially wipe out Obama lead in key battleground states?

    Not that I put much trust in Dick Morris’s analysis (there is a reason why he has been called the worst predictor) but here is how he explains this:


    • Khan

      @Princeton Voter

      Most reputable polling firms do not weight by party ID. They do weight by demographics–a la age, race, gender. If a poll shows up Dem +8, that is because the appropriate sample set of the population claimed to be Dem +8.

      The only major polling firm that weights by party ID is Ras. And he is widely criticized for it.

    • Craigo

      To follow up on Khan:

      Age, sex, race, income, and education are demographics, which are relatively stable. Party identification is an attitude, which is relatively fluid. Weighting by party ID is like weighting according to pro-life or pro-choice, or pro or anti marriage equality.

    • Princeton Voter

      @Khan and others:

      I think what Dick Morris is saying is that polls use a wrong representative sample of voters, specifically over-polling Democrats at expense of Republicans.

      For example, the latest New York Times/CBS/Quinnipiac Poll puts Obama +1 yet their underlying assumption is that likely voters are comprised of 37% self-identified Democrats and 30% Republican (Visit and view Cross Tabs for details).

      Mathematically speaking, if this composition was to change from 30%/37% to let’s say 37%/30%, all of a sudden we see Obama +1 change to a statistically significant Romney lead.

      So the thesis here is that major polls are inherently flawed because they use wrong representative samples skewed heavily toward a higher percentage of self-identified Democrats.

      Is there any validity to this argument?

      Thanks again!

    • Khan

      @Princeton Voter

      We are addressing the same thing, just with different terms :)

      Most of the reputable polling firms work as follows. They poll N number of people. They ask their age range, race, gender, and sometimes ask for income range and education level attained. They will then at some point ask how they identify or what they plan to vote as, Dem or Rep.

      They then match their polling sample to the age, gender and race breakdowns of the general population. They do not match party ID because it’s a fluid variable and can change poll to poll.

      So Dick Morris is complaining about how the population is identifying themselves–essentially that there are more Dems than Rep.

    • Anbruch

      “Is there any validity to this argument?”

      Only if you believe that there is some systematic reason pollsters would be undersampling voters who ID as GOP. Personally, I don’t see the mechanism and can think of many more reasons pollsters would undersample voters who ID as Dem.

    • Princeton Voter

      @Khan, @Craigo, @Anbruch and others:

      Yes, we are talking about the same idea essentially.

      The underlying question is: Are samples used by polls truly representative of people who will actually vote tomorrow (vs. 2008 election)?

      E.g. pollsters may say, let’s get 15% of our 1000 voter sample from 18 to 25 age group. On the other hand, tomorrow we find out that only 5% of the people who actually voted come from that age group.

      Well, this 10% sample composition difference can make a big difference in predicting election outcome.

      Since all polls are to some extent”synthetically” constructed to match likely voters, is there a chance that many of them use the same wrong assumption about sample compositions and that tomorrow we will all be surprised by the outcome?

      Thanks again everyone!

    • Anbruch

      I think general consensus is that Gallup’s numbers are so odd because they are using a very unusual demographic weighting compared to other pollsters. There is a nontrivial chance that Gallup could be right, but I still don’t lose sleep over it. On the other hand, response rates to pollsters are so low that it is really a miracle that the variation of those who answer and don’t answer appears to be distributed sufficiently randomly that the weightings can be applied to the poll and get a number that seems to reflect the sampled population. We shall soon see whether it in fact does.

    • Khan

      @ Princeton Voter

      Yes, there is a chance. However much to Dick Morris’ dissatisfaction, if there is demographical skewing it will not be Romney-friendly. If there is a likely place for potentially missing the appropriate demographic weighting it will be with Latino voters–and the skew would be underrepresentation. That would be very bad for Romney. The other potential skew would be over-representing the white vote.

    • grandpa john

      An attempt to validate a questionable arguement by referring to aDick Morris argument, will automatically invalidate the the premise of the argument.

  • Chris

    Thank you for your website. This was my first election following PEC (in addition to 538); I imagine it will take me weeks to get weaned from checking you a half-dozen times a day. Looking forward to 2014.

  • PhilC

    Limiting hours suppresses the votes of people who are only able to vote at certain times, or who historically have voted on certain days. The technique is based on statistics showing one’s own voters are less affected. Cost/benefit ratio calculation. “We lose some voters, but the other side loses more”. All the vote suppression schemes are based on similar logic. Same logic as the historic poll tax.

    • NickinLA

      Thanks Phil.

    • wheelers cat

      Its Min-Max theory. Maximize your votes, minimize the votes of citizens that wont vote for you. Rural voters dont stand in line.
      Its called the Sailer Strategy, after Steve Sailor.

Leave a Comment