Princeton Election Consortium

A first draft of electoral history. Since 2004

Meta-Margins for control: House D+1.0% Senate R+4.2% Find key elections near you!

We’re back!

November 3rd, 2012, 1:04pm by Sam Wang

At last, we’re back. Andrew is a hero! Restoring the last few days of contenthere  now… Update: Done, with some modifications. If you like the temporary look see If any temporary-site URL gives an error, replace “election” with “synapse.”

Comments please!

Tags: Site News

266 Comments so far ↓

  • Mark Eirich

    Congrats on getting the site back up. You’re doing brilliant work here.

    I really liked that when the site was down, I could see all four graphs on my screen at once without scrolling or clicking….

  • Vaughan

    I asked a question on a previous thread that probably should not have been posed there, but do you ever factor in opposing ground game abilities on a State level?

    • badni

      Vaughan — He doesn’t factor in anything.

      That’s sort of the entire premise of his model — no factoring in anything (except the state polls): for each state, he just takes the median of recent polls. He chose a particular rule for how many, and how recent, the polls are that he would include, and that is it — apply the rule to determine how old the oldest polls can be that you include, and then take the median of all polls in the state. No tweaks, factors, weightings, or exceptions.

      For discussion purposes he analyzes the results of these median calculations with various statistical tools, but the actual calculation of margins and electoral votes is really as simple as that.

  • colo voter

    Hi Sam, first of all, thanks for all of your work – it’s always a pleasure to read. I have been following your site and several others that do quantitative analysis and/or predictions of elections for quite a while. This seems like the year that your material has finally hit the mainstream though. I guess a couple of cycles of being pretty close to dead-on will do that. I do have a question though. Do you think about possible long-term negative effects to this sort of analysis? If analyses such as yours or Mr. Silver’s are taken as the gospel truth in 2016 or 2020, do we risk them becoming self-fulfilling prophecies? I know you may not realize how much people out there are starting to talk about you guys, but they are. People that have no math or stats background are asking me if I have heard of these sites, because they know my math and stats background. Because of this, I wonder about people becoming disengaged early on when they see that the computers are predicting a big victory one way or the other.

    • Rudy

      True that it may lead to self fulfilling conclusion but to the candidate behind it could lead to greater enthusiasm, more focus on States they are behind, strategic messaging. But then again, the candidates do just that!

  • Pete

    323 means Virginia and Florida and Iowa in the Obama Column. I know that’s not how the model works but it has to refer back to a real life situation to be relevant. Most of the models out there miss this important point (Nate!) 297.2 electoral votes will never happen, period.

  • Robin Colgrove

    Glad to have you back. One funny thing is that your _mode_ electoral vote tally, as well as Nate Silver’s now looks like what Drew Linzer’s votamatic Bayesian prediction (O-332) has sat at basically the whole year. It will be funny if after all the ups and downs of other aggregators the one prediction that has barely moved at all turns out to call it right!

    • Sam Wang

      Well (sniff), it was my prior as well, if you recall. Why do you think the red strike zone hasn’t moved?

    • Ms. Jay Sheckley

      Right on. I notice the prediction up top is the mean or similar of the histogram, not the most likely… Is that to avoid jitter? It does seem to be trending up toward the Bayesian… Oh, Dr Wang, I’m SO tired! I wanted everyone to know you were back, and people on the map thread where I urge folks here are being a bit horrid. They won’t quibble with your algorithms, they just disparage, and one (a fan of the Cato Institute) called me faith based! I’ve tagged you and PEC if you feel like helping r messaging me n any important facts I’ve missed. Though really- Gallup? Rasmussen? Hmpf! :D Since when is fame accuracy?

    • Sam Wang

      I’m chewing over national polls. They show the same movement as state polls – 2.0% upward in the last two weeks. It’s a totally independent measurement since national polls are a different dataset than state polls. This is very interesting – and provides two completely different streams of evidence pointing toward the same conclusion.

      How ironic that your Republican friends think you are the one who is faith-based.

    • Ms. Jay Sheckley

      BTW, I’m talking above about where I posted your map on Facebook. This is not a share per se, it’s a screengrab with text telling people your full site is back and to come here. Any time I use something of yours I tag you as a notification, so look for “tagged you on a comment on her picture” or don’t. Your time is valuable and I thank you and Andrew for everything you do. Proud to say, I HAVE brought you new admirers! And your info is good for my dad, stuck in the blackout with only your predictions to console him. Anyway, though with all these moves, and all those blackouts, the numbers may be lower, but you know for sure there are a whole lot of us who will follow you anywhere.

  • Dgarr

    Ahhhh So glad everything is back to normal. I look forward to following even after the election. This is the best site ever. Thanks Andrew and Sam!

  • Curtis

    Great to see you back up and fully online!!

  • Meg

    The right wing sites are still running with the idea that all the state polls are wrong because they are assuming 2008 turn out models, and turn out is likely to look substantially different (I’d tend to agree that it seems likely the Dem turn out will be slightly less enthusiastic, if only because the 2008 race was truly historic).
    Can anyone clarify this for me?
    I know party id is fluid, and the polls report what they get from the sample.
    Are they also then tilting the model further based on 2008?
    Is there anything to this idea from the right? If not, where does it come from?
    I apologize if it’s an obvious question. I’ve scoured both this site and 538 for a simple answer, and I still don’t get it.
    I only had statistics for liberal arts majors, so perhaps that’s where I’m failing.
    Thanks for any help!

    • orchidmantis

      No, they do not tilt the data to match a 2008 turnout model.

      If you look at graphs of Party ID (here’s one: ) at the end of 2010, the big Tea Party surge, there’s a huge drop in R and corresponding rise in I. i.e. a bunch of conservatives stop identifying as R and start saying they’re I. So the mix of generic conservatives and liberals in the electorate has not necessarily shifted much, but what they call themselves has.

      This also explains Romney’s lead amongst the greatly expanded independent pool, which now contains a lot of conservatives recently frustrated with the Republican brand, but still voting for Romney.

    • mediaglyphic

      i think there is some merit to the fact that democratic turnout might not be as strong, however seth stephens-davidowitz wrote an article in the NYT that used google search data to forecast turnout, and he thinks turnout will be at 2008 levels. Early voting in some states also does seem to be down a little from 2008, but hard to say what this means , as many things could be responsible for this.

    • wheelers cat

      Meg, may i draw your attention to this fabuloso votamatic post?

  • Carlos

    Yaaay Sam and Andrew for putting it all back together!

    Ok, now to some last-minute peeks at Sam’s ActBlue recommendations for the House…

  • SoleburyJim
    This “alternative turnout” spin persists. Don’t I recall a similar narrative out of the McCain camp in 2008, and why would the Romneyites believe they had a better read on turnout than anyone else?

    • Sam Wang

      Also in 2004. It’s always good for a laugh, why not this year?

    • xian

      The alternate turnout model seems to be the key to Romney camp internal polls showing them competitive in PA or Wisc or winning in Ohio or whatever malarkey they’re smoking today.

      It would be sort of funny if they unskewed themselves out of a proper strategy.

  • Froggy

    Ted Frank puts forward what is the probably the best conservative case that the polls are far enough off such that the election comes out as a Romney victory:

    I don’t find it convincing (heck, he doesn’t even find it convincing), but he at least gathers all the arguments in one place, and presents them with absolutely no crazy talk added.

    • mediaglyphic

      a very biased republican article, I think his #3 claim (state polls are assuming a very left leaning turnout), seems tenuous. If he is correct about state poll turnout assumptions en masse being too left leaning, then his argument might have merit

    • mediaglyphic

      sorry i meant unbiased republic…

    • grandpa john

      my understanding is that the people being polled self identify their party affiliation. Other than Rasmussen there is no predetermined turnout or party affiliation ratio.

    • Ms. Jay Sheckley

      Froggy writes: “I don’t find it convincing (heck, he doesn’t even find it convincing)”

      You write goodly.

    • grandpa john

      To me, it reads like a collection of Republican talking points that have been proven to be incorrect numerous time. Same old bull crap that they spout when losing
      If the claims are correct, then why doesn’t the election results differ from the poll predictions?
      Anybody remember a vociferous rebuttal of the polls in 2010, when they all showed them regaining the house? Me neither.

    • E L

      “No crazy talk added” is like a doughnut without the frosting.

    • Ms. Jay Sheckley

      What I don’t understand is, if the GOP has a well-intentioned plan, why would they lie about everything?
      Why aren’t more Christians and even churches recommending voting for the Christian candidate, who is also the one who cares about the poor? Why arent they more concerned about hypocrisy [Mathhew 23], wolves in sheep’s clothing and “the father of lies”?
      What Eisenhower foresaw about Christian zealots has come through bigtime. I have a best friend from my teens who is a serious brilliant fun born again, who has watched mainly only “edifying media” for 30+ years and I read their stuff seeking points of agreement. The rift, which started with the council of Nicaea in AD 300, breaks my heart.

    • Shawn Huckaby

      @Ms. Jay Sheckley:

      Although the most common narrative has most, if not all Christians marching in lockstep down a very rigid and narrow path, there is a strong and growing undercurrent of “progressive” Christians that take a much more New Testament view of things.

      Check out Jim Wallis’s site, Sojourners ( It’s a very different environment with an emphasis on social justice, caring for the least among us, and indeed for all of God’s creation. I am hopeful to watch youth embracing this part of religious tradition, even as the old fire and brimstone congregations decline.

    • Matt McIrvin

      I know Ted Frank from way back; he’s a smart guy, but has been marinating in the Republican idea-sphere for quite a while. I have to admit, that makes a lot more sense than most of the “Romney landslide you libs!!!1” rants.

      I do think it’s amusing that he sees uncertainty as something Nate Silver admits only grudgingly, when it seems to me that Silver bends over backwards to acknowledge it, sometimes to a fault.

    • Ms. Jay Sheckley

      Shawn, So impressed with the Sojourners site I’ve bookmarked to share with others. First time I’ve done that with a religious website.

      Froggy, I’m anxious to read that. I am desperate to hear what they have to say for themselves because it just doesnt add up and harms my compassion. HOWEVER apparently it helps the opposition when we share their talking points, no matter how we think we’re framing them, because we recall assertions better than we recall sources. This is how lies become beliefs. Perhaps we are better at not having this problem, but that’s part of the danger. That said, to be uh fair and balanced, here’s all the other arguments in one place: Crazytalk Only. Out of respect for the power of lies, let’s not mention any of this again. …Feel free to reply of course, but we ought _try_ not to quote what passes for substance…

  • ThatSeattleGuy

    Ah, love to see the monotonically incrementing MetaMargin, day after day. Balm for a worried soul.

    Given that both the EV numbers here and at 538 keep inching up, now well past 300, I think it’ll be very funny if the election ends up at the mode of the outcome distribution – i.e., 332 EVs – and makes Drew Linzer at Votamatic (he of the ridiculously stable forecast at that number) look like a freaking genius.

    But really, I’ll be glad to take 270 and not look back.

  • Allan Marlow

    The probability of an Obama victory given by this site provokes the following question.

    If anyone takes this forecast seriously, what is keeping that person from making a fortune trading on Intrade? That person is NOT making a fortune on Intrade because the price gap persists!

    The usual answer to this question is very simple: The person responsible for these numbers is not willing to put his or her money where his or her mouth is.

    Why not? The answer is almost always because deep inside, they don’t actually believe what they are saying.

    It could also be that whoever generates these numbers is too disinclined to pick up dollar bills from the sidewalk when he sees them lying around. Because if the numbers reported have any credibility, that is precisely what is happening.

    If you REALLY believe what you are saying, please open a trading account with Intrade, you still have time, and ENRICH yourselves!

    • Sam Wang

      I really hate this kind of comment, but I’ll let this one through.

      Briefly: gambling is a vice. Also, vices are best left private. Finally, anyone serious about this would be calculating arbitrage for low downside risk and Kelly criteria for maximum return.

      Now get off my lawn.

    • wheelers cat

      Gee Allan…slumming?
      ;;waves hand mysteriously;;
      these are not the phds you are looking for.


    • Ms. Jay Sheckley

      Intrade, by keeping the odds closer than they ought to be, is selling bets to everyone, profiting on each transaction and keeping it lively. Like the stock exchange itself, _they_ are not gambling.

      Dr Wang, being a neurologist knows a lot about the brain chemistry which leads to the illusion that gambling is worthwhile. Even professional gamblers are not gambling, theyre _working_ as they call it with players less skilled than themselves. If you are interested in delusion, that subject is possibly interesting, but not if your interest is in being deluded, or worse in deluding others. Lost story short, gambling is a vice. Even your vice, in this forum, is best left private.

      Contrastingly, this site is about truth, which is worth more than gold. And how to use our minds and our maths to get there. Why not hang out here and just learn?

    • AlpsStranger

      There’s some billionaire with a bot set to short Obama at exactly 66.6% on Intrade, at least as far as I can tell.

      None of us can overpower Sheldon Adelson’s megabux bot :P

    • Iseeurfuture

      Why? come here to attack? If you don’t like what you see then don’t visit. Simple right?

    • Matt McIrvin

      Speaking personally: Do I actually believe Sam’s top-line 98.2%, 99.2% probabilities? No, I don’t. I think his Meta-Analysis is probably the best of the various poll aggregators, but the *prediction* is a new and experimental thing for 2012, and I don’t trust the width of his tails in accounting for random stuff not accounted for, which makes all the current difference between, say, Nate Silver’s probabilities and Sam’s.

      We’re right in this weird zone in which uncertainty about uncertainty can lead to gigantic differences in stated probabilities. Without putting your thumb on the scale concerning the centroid, you can drive the numbers as close to 50% or 100% as you like just by spitballing about tail widths.

      Right now, I think Nate Silver is more in the ballpark concerning probabilities, though I have less general faith in Silver’s model as a snapshot of the race.

      But this is all gut-level stuff. I think we’re at the point where it’s not much worth arguing about.

      I had a moment of bafflement last night at CO and FL being blue when it seemed like Pollster was saying different, but I think I was just getting a few hours ahead of the PEC updates; I see the tick down now.

    • Curtis McMinn

      Long-time reader, but first time to comment and I’d like to direct my first response to Allan’s thoughts. Dr. Wang, I understand if you weed out my comment.

      Thoughts like this miss the entire point of what Sam Wang is trying to accomplish. Or Nate Silver, Drew Linzer, et al. Forgive me, if you will, for getting a little cheesy. These people are patriots. They don’t want others misled by those motivated by ratings and money. Enrichment comes not from money, but from being able to identify truth in a chaotic process. We have enough people commentating and enriching themselves financially. Maybe Sam Wang and others will make money off of their predictions, but that’s not the point. And just my observation — I don’t think they really care about money. Being right is their compensation. Being wrong – and public humiliation – is a big enough motivator to get it right.

      I also believe Dr. Wang is motivated by teaching others how to understand what’s going on. As someone that casually took statistics in college, this site and others have shown me how valuable that skill set can be.

      No reason for monetary gain – especially through betting – here. Just a love of statistics, and enriching others!

  • P G Vaidya

    Dr Wang,

    After reading your comment about looking at Rand
    poll, I looked at last nights and looked at the shift between candidates. Specifically, I looked at the Results by respondent’s characteristics section. It showed a large defection from Obama to Romney in the Hispanic subset. Is there some barrage of Spanish language commercials? Could this affect, for example, Colorado?

    • Matt McIrvin

      Could be support firming up from conservative-leaning Cubans in Florida. Hispanics are not a monolith.

    • Ms. Jay Sheckley

      When I look at these candidate shifts , the bigger shifts over time seem to be blue. The recent shifts favor Romney the last few days. I guess the percent shown is of all Rand undecideds. So most of those folks just prefer outlier candidates or none?

    • Matt McIrvin

      It seems almost as if the RAND shift graph is a leading indicator: we’re seeing some of what looks like regression to the mean in the Meta-Margin right now, and it started in the RAND a couple of days ago.

  • Daniel Wallace

    Great to see your technical difficulties resolved. This has become my favourite site of the election. Here’s hoping that your vision of a better news media comes true: that pundits discuss policy and clarify voters’ choices, rather than pretending to narrate a semi-spurious horse race.

  • Kathy Davis

    Your return is indeed timely, Dr. Wang. Thanks for keeping us all abreast of the numbers in a modified format in the interim. I wonder why the media at large this afternoon still characterizes the poll numbers in the swing states as within the margin of error while often citing Republican enthusiasm on Nov. 6 as the dark horse which might push Romney over the top. Any thoughts?

  • MarkS

    Sam and Nate both now have 332 as the mode, in agreement with Drew’s long-standing prediction. Among the rational poll aggregators with excellent track records, Gott & Colley are now the outliers at 291; FL and VA are the difference. Someone’s going to lose bragging rights!

  • StatsJunkie

    Sam, I notice that Nate Silver claims that there remains around a 16% chance of a Romney victory, based solely on that same percentage chance of “systemic bias” in Obama’s favor in the polls. To me, that seems very high. Do you have any comments about Nate’s hypothesis?

  • Jewish Steel

    Sam, you predicted that the meta margin would likely return to the + 3 neighborhood a couple of weeks back when Obama was dragging a bit after debate #1.

    Well, well, well. Nice prognosticating there, dog.

  • wheelers cat

    You know who else is likely very happy you are back?
    Sean Trende.
    I have been ….umm…tossing a few bombs at him on twitter.

    Heres some red meat for YOU ma cher.

    • grandpa john

      It appears that basically he doesn’t want to admit that are problems within the national tracking polls that he covers, since he does not include all of them in his averaging
      So maybe if he included all of them there just might not be a problem of divergence, but doing that would not drive the agenda would it.?

    • wheelers cat

      well the big scandal in the twitterverse was that Dr. Linzer SAID “Both Trende and Cost contend the “bad” polls are those that favor Obama.”

      So Sean Trende got all pissy with Dr. Linzer, and Jay Cost and ChuckTodd and various random twitter personalities weighed in.
      it was glorious.

  • E L

    Hooray! Thank you Andrew and Sam. The PEC community lives. I’m 99.8% happy!

  • Kathy Davis

    One more question, Dr. Wang: When does the polling stop? Do the polling groups continue through the afternoon of Nov. 6?

    • Prairie Pundit

      That’s a very good question, Kathy, and one that’s been gnawing at my mind, too. Not only the question of when does the polling stop — I assume pollsters shut down at different times, for different reasons. The question is: When does polling stop to such a degree that credible aggregators like Sam stop aggregating?

    • Olav Grinde

      As I understand it, this year is different — there won’t be exit polls from all the states. I would feel reassured if there were, as it would draw attention to any serious divergence between “what voters say that voted” and “the votes officials/machines counted”.

    • Ms. Jay Sheckley

      agree we need exit polls more than ever. you sure theyre stopping that???

    • Prairie Pundit

      My understanding (developed from half-listening to NBC news as I get ready for work each morning) is that the MSM and the pollsters with whom they subcontract are not quitting exit polling cold turkey. Rather, they are merely foreswearing the bad stuff in heavily blue or red states, with the notable exception of heavily blue or red states, like Illinois and Mass., that have some dramatically interesting connection to the candidates. In other words, reports of the death of exit polling are highly exaggerated.

    • Kathy Davis

      Thanks Prairie and Olav.
      Helpful clarification and concerns.

    • Olav Grinde

      Ms Sheckley, I seem to recall reading that the networks were limiting exit polls to just a few states this year. (I’m sorry I don’t have a reference.) In past elections, I thought this was a pretty widespread practice.

  • jefflz

    Repeat of Facebook post:

    I am much more comfortable, statistically speaking, with Sam’s assessment of probability of an Obama win than I am currently with Nate Silver’s. Nate, who does an outstanding job, still gives a 16% chance to Romney bases solely on what would have to be systemic bias in favor of Obama across all polls. I find that number to be exaggerated and in fact at odds with the concepts underlying poll aggregation. I do however retain real fear that the election in a critical state like Ohio could be hijacked- what is the probability of such an event? It is definitely not zero based on past history.

    • Sam Wang

      What’s interesting is that Nate is arguing vociferously for Obama having the lead. I would never argue vociferously for an 84% outcome. When you roll a die, do you argue that it could not possibly come up as a 6?

    • Joel

      “What’s interesting is that Nate is arguing vociferously for Obama having the lead.”

      Nate’s behavior in television appearances, etc. also seems to indicate that he has more confidence than his model suggests. It does make one wonder.

      p.s. if you eliminate his “systemic bias” calculation, his percentages look eerily like your own.

    • Ms. Jay Sheckley

      ” When you roll a die, do you argue that it could not possibly come up as a 6?”

      LOVE! 86% chance of rain: You bring the umbrella, but just don’t know. I lived in Ibiza fr awhile, sorry for _furious_ U.K. weekend tourists who believed it cannot rain there. It rains everywhere! So sad…

      I am now telling people that PEC’s trusted analysis of probabilities says the president _will_ win- if the election is held, and nothing totally preposterous happens.

      I’m a bit worried about all those voters in the dark… but they are more likely to be talking to their neighbors, sharing news and polling place info, and looking for a well lit, heated place to vote?

    • David Fry

      Sam, when rolling a die, I would argue strongly that the chance of a 6 is not nearly the same as the combined chance of a 1, 2, or 3. And that’s what Nate’s arguing against: the idea that the polls show a tossup.

  • Bruce Wayne

    Hi Sam,

    How does your model account for systematic biases in the polls? Pollsters have to make a lot of assumptions when constructing their sample such as the demographic makeup of the electorate, the likelihood that someone votes etc. Since they all use similar methodologies, they all could be making a single bad assumption that could bias their results in one direction or the other. Such biases wouldn’t be reflected in your “random drift” estimates.

    My point is that there is a distinction between winning a phone survey and winning an election. We often make the assumption that phone surveys are good, unbiased predictors of elections, but that doesn’t have to be true. If there is a systematic bias, all the polls would be off and so would be your model. It seems to me that you are neglecting this possibility in an assigning Obama a 98% chance of re-election.

    • mediaglyphic

      past is prologue, in the past their did not seem to be a systematic bias big enough to effect the polls. If anything the bias seemed to be in the favour of repubs last time. Dr. Wang keeps saying that if the polls are wrong more than they have been in the past, and biased against the republicans, then the model would not work.

    • Neus

      I think Dr. Wang’s alludes to this in the website somewhere. Even crappy pollsters or polls including the ones you are concerned about is one of the possible outcomes in the sample distribution of 2. 3 quadrillion outcomes being tracked here. In my papa’s lingo, ” son, easy…I factored in the pot holes in the driving time to0 “! My two cents……take it to Dr. Wang…he is the guy of ( and for) the moment!

    • Joel

      Statistically significant (98%) is not the same thing as “certain”.

      The 2% accounts for the systemic polling failure that you postulate.

  • Angela

    Thank you Andrew and Sam! You rock; and you help my sanity and sense of reality.

  • Brian C.

    The PEC Chrome extension has been updated to work with the new (old) site. As of yesterday it had been in “temp” mode so that it worked with the new/temp site.
    Huge thanks to Andrew Ferguson for reaching out!

    (requires Google Chrome)

  • Beaucon

    Thanks to everyone who helped get this site back up. Thanks also to all the contributors who make the site even better for their thoughtful comments. What is going to happen to this community after the election?? I will really miss this level of discussion.

  • Larry Pareigis

    Welcome back, Dr. Wang. Your site is balm for an election-weary-yet-God-help-me-I’m-addicted soul. Like someone else posted, 332 would be stellar but 270 up works for me.

  • Zmoneygrip

    Good Doctor, in the increasingly unlikely event you find yourself having to eat a bug, would it be a “true bug” from the suborder Heteroptera?

  • Neus

    “In my [Dr. Wang’s] view, the prediction is the correct probability. [….] over the coming days these probabilities will converge to the same value”.
    Dr. Wang–Does the exact convergence occur on Nov 6 or earlier (sometime on Monday?). I’m so breatheless right now. I take a deep breath whenever I log in to/I type ” Princeton Electoral Consortium” for the MM, RP and BP.
    I feel like like a passenger on a fast-moving plane about to land with the last drop of jet fuel….with a probably “earth quake-force” nature of touch down. The current positive Meta Margin, though, for O is……a God-send therapy. Thx.

    • Ms. Jay Sheckley

      Neus makes a good point. We keep asking about election outcomes, but does anybody know how/when the means [PEC’s pre-election phase] fructifies the end ?

  • MeGo

    Dear Dr Wang,

    Thank you for being.

    Without you and Nate Silver of the New York Times, I might well be going mad.

    Keep up your good work!

    • E L

      @Ms. Sheckley What if the networks gave a party and no one came because the networks conceded that Sam Wang is always correct? Ho Hum… I think I’ll just watch a rerun of Ozzie and Harriet?

    • Ms. Jay Sheckley

      I’m a bit vexed. ” If the networks” etc, so what? sure, you didn’t mean it. But there’s been far too much muddled bleep in this election already. It’s nearly impossible to get out the word out about this blog on the scale it deserves, possibly because billions are being spent attraction attention to bogus “exclusive” polls … ::smoke coming out my ears:: I don’t watch tv not produced by Letterman or Lorre, but ok, let’s watch that kid Ricky Nelson sing. Say, you got Burns & Allen?

  • Michael Worley

    Why isn’t intrade rising? I would have thought one side would flinch.

  • Pui Yee

    I’m worry if minorities do not turn out to vote. The swing States could swing to the other side.

    Are minorities usually not considered likely voters? So are their numbers already computed lower in a likely voter scenario?

    Has pollsters already adding minorities as unlikely voters even if they are saying they’re sure to vote ?

    If that is true, if minorities vote big, these good numbers in these polls could be even bigger for Obama?

    • Suja

      Knowing that anecdote != data, I’m still going to say this. My husband tried to early vote in VA today. At around 4:00 PM, there were a few hundred people in line, a large number of whom appeared to be some sort of ethnic minority. The size of the line and its composition gave me even more hope that we’d go blue once again.

    • Matt McIrvin

      @Suja: The long line could be good or it could be bad. Was it so long because of demand, or because of lack of supply? Did the people in line actually get to vote?

    • Suja


      My husband was the only one I noticed turn back without voting, while we were there, and the line was still growing when we left. I’m sure he’s not the only one, but it did not seem to be a major deterrent.

      If I have to wake him up at the crack of dawn and kick him out of the door to get him to the polls on Tuesday, then that’s what I’ll do.

  • Amitabh Lath

    Hi Sam, Andrew: good to see you back to normal.
    We still don’t have power in the Rutgers area.

    Not to curve fit by eye or anything, but the Medaian EV and Meta-Margin, in the range
    [conventions – now] looks like a sine wave.

    • Ms. Jay Sheckley

      I went to Rutgers summer school… Mt dad is in the dark in Edison. I do have a local source for D batteries if you need em.

    • Amitabh Lath

      Ms JS, good to meet a Scarlet Knight on this board. Hope your folks in Edison are keeping warm.

      We live in Highland Park. Half the town has power. We are in the half that does not.

      Now we’ve self-evacuated to a hotel near the Philly airport. But you bet we’ll get up to town to vote on tuesday.

    • Sam Wang

      I think it’s underdamped. What if the election were in December? Maybe one of the candidates would be at >400 EV.


    • Ms. Jay Sheckley

      I heard just after Sandy hit that some states were thinking of delaying their elections… havent heard that lately. So December is unlikely, but a delay might be more probable than a Romney win. If enough states go blue, doesn’t that effect the House ratio?

      BTW, Sam & Andrew, you write: “For now the temporary site is visible at, which is normally my research website. If any temporary-site URL gives an error, replace “election” with “synapse.”

      I infer [incorrectly?] that if your regular site disappears again, we should look for you at . But perhaps you’re not implying that.

    • Sam Wang

      Oh, good point. Probably that’s a good idea. Belt and suspenders, all that.

  • zenger

    “Ted Frank puts forward what is the probably the best conservative case that the polls are far enough off such that the election comes out as a Romney victory:”

    The Romney +2% scenario still gives Obama 277, leaving NH and VA as tossups.

    Gonna have to be mighty skewed.

    • Jay Bryant

      My concern isn’t the polls being off; as you say, they’d have to be “mighty skewed” for that kind of problem to be an issue. Instead, my concern is election officials (from Secretaries of State down to precinct election officials) making some votes not count and keeping certain people from voting.

      I’m glad the meta-margin is as up as it is, since I think it needs to be a point or two in Mr. Obama’s favor for him to overcome electoral interference. I’m still worried. Tuesday night will be a nail-biter and can’t come soon enough for me.

    • Shawn Huckaby

      How about this in Ohio?

      I don’t have a breakdown of the average historical number of provisional ballots there, or how they broke R/D, but interesting either way…

  • Jay Bryant

    Oh, and thanks for getting the site back up. I won’t use Facebook, so I missed out for a couple days.

  • DaveM

    Well, I’m sure glad to be back to the familiar look-and-feel, but I was mightily impressed with the way the PEC community carried on without dropping a beat. Threads at dKos, threads on Facebook, and I even spent a little time this morning scouring pages from 2004 on the Wayback Machine.

    I must confess that I’m now one of those folks who have trouble waiting from 8 pm to 8 am for the overnight MM update and therefore have taken to loading swing state charts at Pollster so as to form a rough idea of what to expect…

  • Reggie Williby

    Dr. Wang,
    After the election, and you are proven right, will you hold the psuedo-intellectual pollsters accountable for their obvious attempt to discredit polling analysis in general and you specifically. Dr. Wang, I know you are not petty or small-minded, but they need to pay for their actions.

    • Matt McIrvin

      What could he do, make them eat a bug?

    • Reggie Williby

      Mr. McIrving,
      I would suggest those frauds be forced to stand bare foot outside Dr. Wang’s office in the snow holding copies of the PEC!

    • Ms. Jay Sheckley

      Reggie, I’m unsure if you mean pollsters or poll hecklers. I’ve been wrestling those today ::shudders:: but let’s not be punitive because:
      (1) It’s possible that some light may dawn.
      (2) If not, give ’em time. After all, shaming often serves less than no purpose, because, except for dog shaming :D) it can be seriously counter productive. Additionally,
      (3) Retaliatory behavior makes us look bad and (4) feel worse, as part of a rage/shame cycle.
      (5) The gentlemen in question may already be serving a life sentence as idiots.

  • Analytical

    The ground game outlined in Jim Messina’s memo today is astounding (even if only 50% work out). I have seen this kind of organizing from very very up close. It can be quite deadly! If their ground game works out, we may actually see President Obama winning in excess of 332 EVs.

    • Martin

      Apparently Mr. Messina forgot about Colorado where they are losing early voting even at this late stage! A big chunk of Colo. has already voted! at this rate they are getting their butts kicked over there!

      How can that be if they have such a powerful GOTV? Did they not realize that Colorado was a swing state? Did it slip their mind? No 332 without Colorado!

    • Joel


      From the memo:

      “The math is clear: our opponent is losing among early voters in nearly every public poll in every
      battleground state, meaning that if these public polls are right, he would have to win 65 percent
      of the remaining votes in North Carolina, 59 percent in Iowa and Colorado, 58 percent in
      Nevada, 55 percent in Florida and Ohio, and 52 percent in Virginia and Wisconsin.”

      Obviously that’s campaign-speak but there’s this as well:

      Based on the last 3 elections 2006, 2008,2010;Democrats vote between 91-93% according to party affiliation, Republicans in El Paso vote at 85% and statewide 87% according to party affiliation and Unaffiliated lean Democratic 51-54%. (HT to RWN)

      Extrapolating these numbers onto a 1,640,023 voter sampling we begin to see a picture of my unprofessional and roughly estimated voting trends from little old obsessed me.

      91% of 567,569 Democratic votes cast is 522,163
      13% of 605,586 Republican votes cast is 78,726
      51% of 466,868 Independent votes cast is 238,102
      54% of 466,868 Independent voters cast is 252,108

      Total Indy’s @ 51% = 838,991 = 51 % Democratic estimate
      Total Indy’s @ 54% = 852,997= 52 % Democratic estimate

    • Joel

      P.S. the projection based on past behavior that I linked to can be found here:

      It also matches the state-level polling picture (as told by Sam’s voting margin, but also other models, such as Nate Silver’s)

    • Martin


      I had already read the memo. The question is why are they still behind in Colo. early voting even at this late stage if their ground game is so great? Why aren’t they atleast tied???

      I don’t have clue what Messina is talking about when he says 59% percent of the remaining vote for Romney for him to win? THEY’RE BEHIND RIGHT NOW. Only way he’s right is if the independents in e. voting are skewing heavily towards O which is doubtful. Is he reading different numbers than the ones publicly available or what?

    • David

      Martin, what makes you think the Obama early vote is behind in Colorado? Clearly they’re not.

      Romney would need to win 59% of the vote that shows up on Election Day to win. That means Obama is ahead, by their calculation, in early voting.

    • pechmerle

      Re the discussion Martin vs. Joel on where things stand re early voting in Colorado, it does appear that Martin is correct that Republicans currently have the lead in early voting there. That’s if I’m reading the table at George Mason U.’s site correctly:

      Has Messina’s memo got the facts wrong on Colorado? Is he basing on polling as to who has already voted vs. the secretary of state’s tally (his wording seems open to ambiguity on that)?

    • mediaglyphic

      In the memo in question messina is quoting polls

      “The math is clear: our opponent is losing among early voters in nearly every public poll in every
      battleground state, meaning that if these public polls are right, he would have to win 65 percent
      of the remaining votes in North Carolina, 59 percent in Iowa and Colorado, 58 percent in
      Nevada, 55 percent in Florida and Ohio, and 52 percent in Virginia and Wisconsin.”

      Polls are telling him this, he is not tabulating early voting by looking at party id. This could be right or wrong, but these are the facts as far as Messina’s polls tell him.

    • MAT

      I believe Messina is basing his memo data on the campaign targeting model – according to Sasha Issenberg’s ‘The Victory Lab’, the campaign daily extrapolates a probability of an Obama vote to every potential voter in the US based on regressions of demographic data and the enormous (far dwarfing the polls) amount of voter contact they have. The model apparently performed very well in 2008.

  • Hudson Reeve

    Sam, maybe you have posted about this elsewhere, but what is your reaction to the prediction by the Colorado University political scientists Bicker and Berry (mostly based on indirect economic data) of an easy Romney win?:

    • Sam Wang

      Read this for a start. In my view, political science models are not mainly for prediction. Instead, they are experiments to understand the political process better. In this case it has been fairly obvious all season that their experiment completely mismatches polling results. Presumably they will get some attention on Wednesday.

  • Shawn Huckaby

    An interesting, and rather thoughtful handling of the recent Nate Silver attacks on one of my favorite podcasts, Slate’s Political Gabfest:

    It’s in the third segment, although the whole thing is worth a listen. One of the principals, John Dickerson recently graced this very site. Interesting that you weren’t lumped in here as well, Dr. Wang (although maybe obliquely?).

  • kent

    Hey Hudson Reeve, I want to thank you for directing my attention to that website (thenewamerican dot com). Quite an astonishing read. It’s hard to know where to start. At the top of their “List of Friends”: the John Birch Society.

    Anyway, it’s clear that the site is cherry picking anybody who agrees with them. Clicking through these professors’ past work, they cite Reagan’s “comeback” win over Carter (which as has been pointed out many times wasn’t a comeback at all), etc.

    You may also want to check out their original piece, with details on their regression:

    Just at a glance I find their predictions ludicrous. Among others: Democrats running for re-election are massively harmed by unemployment, to the tune of losing 3.3 points on the vote for every point in the unemployment rate. By contrast, “For Republicans running as the
    in-party candidates, the impact of the national
    unemployment rate is not significant.” Yeah, right.

  • Seager Mason

    Terrific website Sam, and a great relief for those of us in New Zealand and most of the rest of the world who live in fear of another republican US government.
    We had a very scary interview on Radio NZ yesterday with Victoria Collier of about vote rigging in the USA.
    I was a first-time election official during our election last year, and was extremely impressed with NZ’s very rigorous and transparent manual voting and counting system (a number of countries sent experts to watch and learn from us). So I was amazed to learn yesterday just how non-transparent and open to rigging the computerised voting and counting system is in the USA.
    I see that 19 states have abandoned exit polling this cycle, fortunately they don’t appear to be the critical swing states. I will still be waiting very nervously for the positive result you are predicting.
    Hopefully the emergence of powerful analysts like PEC, 538 and Votamatic ect will help to put a check on potential rigging.

    • xian

      States don’t do exit polling: news orgs and pollsters do. So the states haven’t abandoned exit polling; they’ve been abandoned.

  • Violet

    Hi, Sam: Long time reader, first time poster. Thanks so much for your hard work, and for making math fun and reassuring for the masses.

    I have a great deal of respect for Nate Silver, but believe that his predicting a 16% possibility of Romney winning lets him hedge in case Romney does win by saying, “Well, I gave him 16%.” I like that you trust your model and go out on a longer statistical limb by actually predicting near 100%, with a real EV number of 323, instead of Nate’s impossible 306.9.

    I don’t think I’ve ever hoped as hard that someone is right than I hope you are, and not just to save you from eating a bug.

    • Ohio Voter

      Two things:

      1) In Silver’s defense,the 15% Romney has is pretty much based on the possibility of a systematic polling bias.

      2) Silver will still make his final predictions as well, he won’t just give odds

  • wheelers cat

    One bizarre phenomena that I observed while PEC was down and I was wandering homeless on the internet was some sort of cris de nerf at Votamatic where everyone was flashing credentials.
    Even PEC’s own Froggy.

    Dr. Froggy, JD, PhD, and future honorary DDS
    October 30, 2012 at 6:47 am

    I found alternative commentariats on the interwebz to be dissatisfactory in the extreme.
    My adventures certainly made me appreciate this one, and all the work Dr. Wang does to shape it.
    Its unique.

    • Froggy

      I also was once named a Hero of the Soviet Union, and in my spare time I am an up-and-coming superhero with powers that make the bad guys wet themselves from fear. (Modesty prevents me from going into greater detail.)

      It’s good to be back here on PEC, where the commenting community is great, and I’m just plain old Froggy.

    • wheelers cat

      For some reason I entirely believe that, especially the super-hero part…
      /salutes froggy with respect
      Truly, the rest of the internet is a vast intellectual wasteland populated by claquers and grifters.
      Does that make me an elitist intellectual snob?

      There have been some awesome discussions at PEC this cycle. I’ll miss it.
      I guess we are Dr. Wang’s garden. Hope we were worth all the work he put in, pruning and feeding.

    • Ralph Reinhold

      I think that was caused by Rightwingut, PhD, the red faced soap boxed political scientist who opined over facts.

    • Matt McIrvin

      It’s because there’s a guy there whose Romney victory predictions are always accompanied by mention of his Ph. D. and his status as a political scientist. They were all dogpiling on him.

    • wheelers cat

      lol, not just him. That Allan Marlowe guy was phd-ing a lot of the regulars too, freaking scolding them!
      And I just dont understand the conservative tendency to privilege the the Market over the Maths. It makes no sense. The Market can be way wrong.
      Consider how wrong Intrade was on the ACA decision.
      They seem to think the Market should inform the Maths.
      Isnt it the other way around?

    • Matt McIrvin

      I often refer to Sam as “Prof. Wang” because in the academic environment I came out of, a Ph. D. and a couple of bucks will get you a cup of coffee, but a professorship is really something.

  • Ms. Jay Sheckley

    you glance at the new EV prediction and immediately know what changed is Colorado.

    • Ohio Voter

      And PA moved from 5 to 4 haha

    • Olav Grinde

      Darned ink shortage!
      They didn’t have enough Blue to color Florida and Colorado a nice sky hue this morning.

    • ChrisD

      Here are all the changes in the Power of Your Vote table:

      IA: O+3 => O+4
      PA: O+5 => O+4
      CO: O+1 => tied
      FL: O+1 => O+0.5
      MN: O+7 => O+7.5
      LA: R+13 => n/a
      NM: O+10 => n/a
      ME: n/a => O+7
      WA: n/a =>O+14

    • DaveM

      Although I’m having trouble finding any combination of Colorado polls as listed in Pollster’s chart which produces a tie. I get O+2 as the median…

    • Froggy

      DaveM, the Colorado numbers have been off for a couple of days now, at least based on what is posted on Pollster. I suspect that the problem is that multiple releases of the Ipsos/Reuters daily tracker are somehow being counted as separate polls (which as I understand it they shouldn’t be). I sent an email to Andrew and Sam about this a few minutes ago.

      I think the same problem is beginning to affect Florida as well. I have it as Tied now, rather than O+0.5.

    • Olav Grinde

      Froggy — so is just having a glitch with their polling machine? Ah, that’s a relief to hear! For a moment I thought Reality was losing its Liberal bias. ;)

    • DaveM

      Thanks, Froggy. I also had Florida tied, as well as North Carolina at R+1.

  • Dean

    From what I could tell in polling aggregation, Obama has persistent leads in northern swing states. This is where he should win the election.

    Right now, Florida appears to be a toss-up or very slightly leaning Romney, whereas Virginia is leaning very slightly to Obama or is a toss-up. Colorado is a toss-up to leaning very slightly toward Obama.

    I would love to see Obama get the north plus at least Colorado or Virginia (or both). I would really love for him to get Virginia. I’m hoping Obama gets at least 281 EV’s and calls it a day.

    • Reason

      Dean. Me too. I am in Va and already voted. The OFA has now called me. Twice. So they are doing their job. I am reaching out to others to get them to vote.

    • Dean

      Great job in GOTV in Virginia. I’m in Chicago. On Tuesday night, I will be watching the election at a friend’s house, a few miles away from McCormick Place (convention center), where the president is planning on being.

      NBC/Wall Street Journal just released its last national poll (O+1) and will release state polls later today.

    • Suja

      I think that he will win Virginia. Virgil Goode would pretty much only pull Romney votes to him, and Gary Johnson would pull more from Romney than Obama. If the number and frequency of calls I receive from the Obama campaign is any indication, GOTV is working well.

      I do wonder what the breakdown is on newly eligible voters (through immigration, migration, aging up) in the Commonwealth.

    • Dean

      I was wrong in saying that NBC/WSJ will release state polls later today. The article says simply that Obama leads in a majority of state surveys, and the the rest of the new national poll (O+1) will be released later today. Sorry about that. I’m in the throes of “OPW” (Obsessive Poll Watching).

  • JustAnotherUndecidedVoter

    We’re Back!

    I don’t understand. Where did you go?

  • ChrisD

    Hmm. Who turned the clock ahead? The site seems to be on Atlantic Time (Canada) now: GMT -04:00.

Leave a Comment