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

Post-debate pundit spin lives up to its promise

September 28th, 2016, 12:12am by Sam Wang

I have little to say about last night’s debate, except to point out that based on polls of undecided and independent voters, Clinton was seen as scoring a convincing win against Donald Trump. This feeling is echoed by one-third of Republicans, which is pretty bad – as well as Rudy Giuliani, Brit Hume, Megyn Kelly, and other GOP bloviators. Oh, and Stormfront. Seems clear-cut.

What will happen next? Before the debate, I used regression to the mean to suggest that in the absence of other information, opinion is expected to move toward Clinton. We should have some national and state polls by Friday. Until then, ponder what value is added by pundit opinions.

The starting point for future comparisons is Clinton 303 EV, Trump 235 EV, Meta-Margin = Clinton +2.1%.

Tags: 2016 Election · President

104 Comments so far ↓

  • A

    Great point, Sam. In fact, as I recall, you said they’d likely point to Hillary being “revitalized” or something along those lines.

    And today, by golly, they were saying she appears “revitalized.”

    However, all of that being said–I do think that Hillary just so happened to do a phenomenal job kicking Trump’s behind at the debate last night, and I wouldn’t want to take away from that by saying this was preordained…

    I think if Trump had somehow fared better, the regression to the mean might have taken a bit longer…

    As it is, I’m calling for a 3-4 point bounce in the polls over the next week or so. She’ll land at a meta-margin of +5 or +6, but I believe that bounce will begin fading.

    The pundits will declare Trump to have won or tied the second and third debates–whatever they need to say to keep things “close”, and by the home stretch we will be looking at Hillary being back down to +2 again…

    And she will win the election similarly to Obama/Romney margins.

    • TZX4

      My concern is that now that Mr. Trump has established an extremely low bar, he now has two more debates to establish a comeback narrative in the sensationalist MSM.
      Might is be an echo of the Obama Romney debate series?

    • Jeff

      2nd debate is a Townhall. Hillary actually does better when speaking directly to voters whereas Trump has a hard time being challenged and will feel the need to correct. My instinct is that Hillary will win 2nd debate and 3rd debate will be too late for Trump as Hillary is already banking votes.

    • Camhilfan

      Yeah, I don’t think Secretary Clinton needs to worry about whether the media declares her the winner of these debates. A lot of people that don’t follow politics until the debate watched that. They saw Secretary Clinton expose Trump badly. That cemented their vote. I think he’s got pretty much all the votes in the bank he’s going to get, unless something catastrophic happens that’s never happened before.

      All she really needs to do to wrap up the election is continue to be Presidential. Continue to prepare. Continue to have a great handle on the issues people really care about. She can bank a lot more current STein or Johnson voters simply through these debates, regardless of how the media spins it. that’s all she needs to do. She will be asked about the Clinton Foundation, and private servers, and emails in these upcoming debates. She needs to handle those questions, and move on quickly. No five minute, bizarre, tangent fill explanations. “I made a mistake. I take full responsibility for it. I’m sorry that the issue has distracted us from the real issues that will impact our country…issues I’m ready to get solved with your help…”

    • Phoenix Woman

      Camhilfan: The Clinton Foundation is a real charity that saves lives. The Trump Foundation is a cash cow and tax dodge for the Donald.

    • Camhilfan

      I know the Clinton Foundation is a real charity that save lives. And you know that. But the portrayal has been that it’s a slush fund for the Clintons. In addition, most people that only heard bits and pieces of those stories because they don’t pay close attention to politics until right about now essentially heard that people donated to the Clinton foundation to gain access to the Secretary of State.

      This is a Presidential election. Perception is as important as substance. So, yes, she’s going to need to handle those questions easily and move on from them.

  • ItsSupercar

    My prediction all along has been that the debates would act like a repeat of the conventions. In both cases, Clinton’s numbers dwindled during the gap between major events, but seem to reset toward an advantageous equilibrium when she gets an equal share of a bright spotlight.

    Still a few anxious days before we actually get to see the needle move, though.

  • TJHalva

    Looking through the debate in the context of an earnings report I think can be constructive (and interesting!).

    With stocks, there is an implied expectation around the outcome of a quarterly report. This can be measured in option volatility. Without getting too deep, the “true” value of an equity is generally at its most unknown right before earnings because you have the market attempting to price in expectations before actual hard data of the companies financials becomes real. This uncertainty is the volatility.

    Polling volatility has been relatively low and steady this cycle. NH is my go to, Trump has led in exactly one poll, yet is considered a “toss-up” by some. The debate can be looked at as an earnings report, but the company reported inline and there were no surprises. Based on media reports Trump lost.

    Volatility was right, as it usually is. In order for volatility to continue to be correct it will remain low which implies a regression back into a higher Clinton margin.

    This is sort of a convoluted explanation, but I think its a fascinating relation that supports the regression to the mean theory from a different perspective.

  • Ed Wittens Cat

    I’m encouraged that Clinton is going to campaign with Sanders this week– she missed an oppo to tell Trump in the debate that she didn’t change her mind on TPP– Sanders did. That would have been brilliant and appealed to Sanders supporters.
    Trump’s Achilles’ heel is his pugnaciousness, and Clinton’s Achilles’ heel is her hubris– Colin Powell pointed this out in his leaked emails.
    She did say the “gold standard” thing, and of course the media gave her a Pinnocchio– she just can’t carry this kind of thing off.

    • LBS

      She said it was a “gold standard” in 2012 and listed the reasons that it was. After the details were negotiated it did not meet several of the criteria she mentioned at the time she said “gold standard” and she therefore no longer backed it. Please read her entire gold standard statement.

    • snolan

      ‘she missed an oppo to tell Trump in the debate that she didn’t change her mind on TPP– Sanders did’

      I thought the same thing. Would have helped to warm her up to the BernieOrBusters… AND deflate Trump a bit.

    • Ed Wittens Cat

      LBS, that is simply untrue
      Clinton changed her position *after* Sanders forced her to.
      That’s a GOOD THING imho.

    • JFCC

      While it may be true that it was Sanders who pressured her to switch on TPP, I think a better answer – and not really an untrue one – would be to say she changed her mind because of what the voters were telling her. They told her by voting for Sanders, of course, but I think saying “voters” is better. (Maybe Trump would respond “Yeah, by voting for Sanders” though, which would be a good laugh line, so maybe she could say “through my arguments with Senator Sanders and listening to his supporters” or something like that.)

    • bks

      No one has a clue what TPP is about let alone what its effects might be.

    • Ed Wittens Cat

      anyone with sufficient substrate can unnerstan TPP
      for example i do– read Petersen Institute and Naked Capitalism
      the problem is older blue-collar white men can neither unnerstan or relate– and thats Trumps base–
      the “Left Behind” of the globalism economy

  • A New Jersey Farmer

    Thanks Sam. The Senate races were moving towards R with Trump. Do you think they’ll follow Clinton back to the mean or does it look like voters will be ticket-splitting?

    • Sam Wang

      My guess is they’ll move back. The Senate Meta-Margin seems to move by similar amounts as the Presidential Meta-Margin, which argues against ticket-splitting as a major phenomenon.

    • Olav Grinde

      Republican megadonors are investing large amounts of money in downticket races.

      I am wondering to what extent the strong drift towards continued GOP control that we have seen is a result of enhanced Red downticket spending.

    • Sam Wang

      No, it looks like everything is moving together.

    • Phoenix Woman

      I’ve noticed that the better Obama does, the better all other Democrats do. If his approval rating goes back above 3%, we win the Senate.

  • W.Dow Rieder

    I don’t think the debate will change the minds of many people who have decided to vote for T or C. Might affect undecided or 3rd party a bit. But those have been breaking about even, so I think the changes we’ve seen in polls have mainly been noise and differential response–and differential response isn’t ‘sticky’ the way actual preference is. I think MM will go back about to long term mean, then drop a bit, because T did get a point or two from holdout Rs deciding he wasn’t unacceptable after all.

    • Jinchi

      I don’t think there are many people who are deciding between voting for Clinton or voting for Trump. Those populations are almost exclusive. Undecided voters are typically those who are deciding whether to vote at all. That is a large percentage of the population in any year and probably larger than normal in a year with two unpopular candidates. I think Clinton helped herself with those left-leaning undecideds. I think Trump hurt himself among right-leaning undecideds. I’m not sure how well polls pick up changes in the likely-voter population, my guess is that the information is too subtle and subjective.

    • Matt McIrvin

      Clinton’s task at this point is mostly to bring home liberal-leaning voters, especially younger Bernie supporters, who don’t trust her and are flirting with voting third-party or not voting. They seemed to come back to her during/after the D convention and then got dissatisfied again in September, especially after Pneumonia Weekend.

  • Steven Fondo


    Election guru extraordinaire, David Plouffe says Trump has no viable path to 270. Media is imagining contrived horse race.

    • Camhilfan

      I watched him with Halperin and Heilemann yesterday. He believes the path for Trump always includes Florida, which he doesn’t believe, with the electorate there, Trump can win. Same thing in Colorado, and probably Nevada.

    • InmanRoshi

      Yes, I’ve heard Plouffe on various sources say he doesn’t see the path to 270 for Trump. Plouffe is working as an outside advisor to the Clinton campaign and evidently has seen their data modeling. (FWIW, the Clinton campaign manager, Robby Mook, is a huge data driven guy, and was in charge of 3 state contests for Hillary in 2008 in which she beat Obama, and Plouffe said Mook was the one guy in her camp that they felt did the best job). Again and again he points to the underlying demographic makes and their data modeling of their electorate.

      I also heard Jim Messina on the Political Wire podcast this week say that there is no question Hillary is winning. Along with his own international consulting group, he’s co-chair of the Priorities USA Hillary SuperPAC. Biased? Yes, but he also knows the data. He said every morning and soon as he wakes up he looks at the results of 6,200 permeations of their latest data modeling simulations.

      On the other side of the aisle you have Stuart Stevens, Mitt Romney’s 2012 Campaign Manager. He says the GOP have no chance of winning another Presidential Election until they change the demographic make up of their electorate because they’ve reached peak angry white male voter, and the idea that there are untapped white voters out there that Trump can draw upon is a myth. He laid this out very well in this article.

      All this to say, these are the professionals. The elite of the elite of their craft, who’s job is to crunch these numbers and to know them backwards and forwards. And when they’re saying, on both sides of the aisle, that the election will probably be close but ultimately the demographic numbers don’t work for Trump, then it means more to me than the latest Phlegmghazi or Basket of Deplorables pundit driven horserace newscycle.

    • Matt McIrvin

      Plouffe is a Clinton campaign guy and has good reason to talk his candidate up; it’s true that he’s probably got better information than the public pollsters, but he’s also a very interested party. I do not believe that Hillary Clinton is going to win Iowa, and he’s claiming that.

      Sam’s current Trump +2 map puts us in toss-up territory, and the question is just which of the gray states he gets in that scenario.

      I think the “Pennsylvania path” they’re talking about is viable, if some event causes a national swing just a little larger than the Phlemgate/Basket of Deplorables weekend. Only my hunch is it wouldn’t be with North Carolina, it’d be Nevada. That would get him to exactly 270; probably it’d actually be 271 from ME-2.

    • Some Body

      Frankly, I fail to see the point of convincing ourselves in advance that Trump can’t possibly win. In six weeks or so we will know who won (or will be waiting for the Supreme Court to weigh in ;-) ). If it’s Clinton, all the happier we will be. If it’s Trump, we’d be better prepared to respond (by reorganizing politically, or fleeing the planet, or whatever), and save ourselves months of paralysing disbelief. I honestly don’t see what good can shutting your eyes to the perfectly real possibility of a Trump win and engaging in wishful thinking do you.

    • InmanRoshi

      Plouffe has said out of all swing states , Iowa and Ohio

      As far as Trump winning PA, now that’s a far reaching longshot.

  • Betsy Teutsch

    My 28-year-old daughter has not paid close attention to the race. she was in for Bernie, dispirited by the primary outcome, and intended to vote for Clinton but not with any excitement. She was nauseous during the debate, and frightened. She had ignored Trump up til now. If she is typical, a main takeaway from the evening is young people flirting with a Jill Stein vote–seeing what that would cost.

  • Bill

    What’s up with RAND this year. Their poll in 2012 provided quick responses to changing trends even if their absolute values were off.

  • mediaglyphic

    Are the UPI/CVOTER polls included in the PEC metrics? I ask because the meta margin and bayesian metrics moved when the ipsos reuters polls came out but did not seem to move back on UPI/Cvoter (which looks like a move back towards Drumpf)

    • C Hart

      Yep, believe so. The polls are downloaded from Huffington Posts poll repository and the UPI polls are in there.

    • Matt McIrvin

      There was a tiny tick down at some point after the bump up.

  • Michael

    Isn’t the “meta-debate” — the post-debate spin and news cycles — more important than the debate itself? By that measure, Trump’s sticking with celebrity feuds and blaming the moderator and microph0ne may have squandered whatever benefit he got in the first few minutes, it would seem.

    • OldenGoldenDecoy

      In this thread Michael // Sep 28, 2016 at 10:57 am said, ” Trump’s sticking with celebrity feuds and blaming the moderator and microph0ne may have squandered whatever benefit he got in the first few minutes, it would seem.”

      Yes… That was his weak attempt to turn the media pundits attention away from issues and not focus on his following list of BS answers… Climate change • Jobs • National debt • Clinton on Trump ‘rooting’ for the housing crisis • Crime • Taxes • Federal Reserve • His bankruptcies • His support for Iraq, Libyan interventions • Guns • NATO • And America’s nuclear arsenal.

      You know, real policy issues.

  • Frank

    But what about Lichtman’s Ke…just kidding!!!

  • Cassie

    I’m encouraged by the words from the experts above. I seem to be less worried than the typical panicked democrat, but still nervous a bit. But I am trying to channel this energy in helping on HRC’s campaign, and doing calls. Also can someone explain to me why most pollsters(I assume are) are using the 2000 and 2004 voter base when we know first of all demographics have shifted since then? Yes I know some consider Obama an outlier, but it still perplexes me. Also is there any reason for me not to think minorities continue to be under polled? Lastly what on earth has happened to Nate Silver? He seems to really have gone off the rails.

    • 538 Refugee

      Quick question Cassie. It never occurred to me until a few weeks ago that campaign call centers could be leveraged as part of their internal polling. Do you ask and record call data that could be interpreted as such or is it mainly aimed at party faithful?

    • Jinchi

      “Lastly what on earth has happened to Nate Silver?”

      He confidently predicted that Trump would collapse throughout the primaries and was repeatedly proven wrong. He is probably surprised that Trump has pulled closer to Clinton than he though possible, and he has seen several high-profile polling failures recently (like Brexit). He may be overcompensating, but at least he’s trying to learn from past mistakes.

    • Sam Wang

      Factcheck: Brexit was not a polling failure. Polls succeeded more or less – pundits and opinionmakers failed.

      As to how one can learn from failure: I would suggest (1) stick with polls when being a data pundit, and (2) be transparent.

  • Bram Cohen

    The debate results are of course part of the regression to the mean. Trump had been unusually out of the media right before it, and voters are back to being reminded what they think about him.

  • Olav Grinde

    “It looks like everything is moving together.”

    The latest Reuters/Ipsos polls certainly support your observation. It has:

    Clinton +6% (Trump +2 a week ago) (movement: 8%).
    Generic Congressional, Dems +6% (R +1% last week) (movement 7%)

    • Matt McIrvin

      Interestingly, that, like the small recovery in the Meta-Margin, seems to be a pre-debate bounce, which dovetails with Sam’s general thoughts.

  • A

    I do think we can agree that the media’s addiction to the “horse race” narrative makes it likely for them to call almost anything in the next two debates a “win” for Trump, or somehow frame the election as closer than it might otherwise be.

    They will need to have some kind of hook to keep people tuned in and watching the shows. This is, after all, their big “holiday blowout sale.” Every four years they get Christmas and Black Friday rolled into one, and they need to exploit that for their ratings bonanza.

    So unless they could somehow pitch the idea of a huge blowout that’s NEVER BEEN SEEN BEFORE, I think they will be pitching a narrowing contest with Trump coming back to win one or both debates (regardless of how pitifully he performs).

    But as Sam has tried to point out, much of this is noise and the regression to the mean should play out due to voter polarization.

    If his theory holds, then by the time the election rolls around and the various bounces wear off, we will arrive at the true margin of victory.

    Likely to be Hillary somewhere in the 2-3 range in terms of the meta-margin, but I’m hoping she out performs her polling and beats Trump more handily than that, as Plouffe has suggested.

    • Violet

      What’s interesting in worries about media pushing a narrative of a Trump win, if, as happened Monday, he is so bad and she is so good, they simply can’t push that narrative with a straight face. Maybe if Trump had been more sane Monday, and Hillary had been weaker, they could have made a narrative, but the candidates will make that impossible.

      Unfortunately, the bar is high for Hillary to be so good that the media can’t pretend she lost. Luckily, she is able to pass that bar pretty consistently.

    • Roger

      I am not sure about Media doing that. I think the Hotel tour -slash- very short Birther admission was the turning point, by the Media, to treat Trump seriously and critical as it would any candidate.

    • Perry

      Like many people commenting here: I don’t like Mr. Trump.

      That said; given his past behavior, I didn’t think he did so bad on this last debate. It seems like the media is really pushing this as a huge win for Mrs. Clinton. I’m not so sure…. I’m of the thinking that she should win. She is a professional politician and does homework!

      My assumption going forward is that Mr. Trump will actually put some real effort into preparing for the next two debates. So how will the media spin his improved performance (assuming improvement)?

    • Perry

      Another observation/thought: From my point of view; Mr. Trump seems incompetent/unknowledgable….. But, I’m not sure if my perception matches reality. After all, 40% to 45% of the electorate seem to like him. Maybe there is more political competence there than meets the eye?

      I do think Mr. Trump has considerably more “charisma” than Mrs. Clinton. She has knowledge/experience/competency(?). She just doesn’t seem very interesting/exciting.

      This is where Mr. Trump is potentially very dangerous to the Clinton campaign. I think Trump supporters are more excited about their candidate (sometimes in unfortunate ways). Whereas, Clinton supporters are like: “ok…. Hillary, I guess”. Sort of more of winning by default.

  • Gil

    Guys I’d like to hear your thoughts on a perception I have as a new voter (this is my first presidential election as a US citizen), which means I don’t have preconceived notions of her from the 90s. I find that the more I see her live on tv, the more I like her. She’s solid, unflappable, calm, etc. But because Trump sucks up all the oxygen, she is not able to drive the narrative, she can only respond and use his shortcomings against him. What I’ve noticed is that after a good event (convention), she goes up, then slowly fades, meaning, voters need constant reminders that she’s in fact the best in this race. Same thing after this debate. She will get a bump, then start fading, until the next major event. I wonder if there’s anything her campaign can do to grab the narrative, maybe major policy speeches every few days or so. Unfortunately we live in an era of instant gratification so votes need a little meat to keep paying attention.


    • Sam Wang

      That’s interesting. I am under the same impression.

      Conversely, I wonder if the Trump campaign was purposely keeping the candidate out of the spotlight in August/September, to avoid negative news stories.

    • JFCC

      I knew this Vox article from May would be prophetic as soon as I read it, and I still think it’s a large part of what’s happening:

      Nature abhors a vacuum, and the modern media abhors a boring race.

    • Violet

      Here’s how I see it, Gil. There’s a caricature of evil Hillary on which her critics have been beating the drum, and there’s the real Hillary, who I’ve met personally, who is warm, funny, extremely smart, and totally unflappable in the face of enormous pressure.

      When more people see her, they see the real Hillary, not the dying, corrupt, inauthentic cartoon media has been pushing. So I’m glad to see you bearing out my theory. I’m hoping with more appearances, interviews and debates, people will come over to her.

    • Matt McIrvin

      My mental model of the race is the same as yours, but I think the fade time of these pro-Hillary events is important, because we’re running out of time.

      The Democratic convention (and Trump’s berserk reaction the week after) got her five really good weeks (much longer than most pundits expect from a convention bounce), and five weeks is all we have left before Election Day. If this debate has an effect like that, Hillary is a shoo-in.

      On the other hand, there are two more presidential debates to go, which might not be under conditions quite as favorable to her (number two is a town hall; number three has a moderator from Fox News who may well be in the tank for Trump). In 2012, Obama recovered from a bad first debate with strong performances in the other two. I don’t quite believe that Trump can manage what he did, but you never know.

    • Matt McIrvin

      …Correction: there are six weeks until Election Day.

    • (((CassandraLeo)))

      On one hand, town hall meetings are one of Clinton’s strongest environments, and the short-fingered vulgarian has been at his weakest when he has to interact with the general public. So he’s going to have a difficult time recovering at that debate. He also seems to have learned the wrong lessons from his loss on Monday; he doubled down on attacking Alicia Machado, blamed his microphone and the moderator for his poor performance, and said he was “generous” by not bringing up Bill’s affairs. I honestly expect his performance at the second debate to be even worse than his first one.

      The third one is worth worrying about. Fox News is not exactly noted for its journalistic integrity. Wallace does seem to care somewhat more about being perceived as a real journalist, though, and actually even fact-checked Cheeto Benito at one point, iirc, which was considered one of his weakest debate moments. On the other hand, he’s also said he’s not going to fact-check this debate. Then again, if he’s perceived as being biased in favour of the weapons-grade plum, that could end up backfiring the way Lauer’s terrible performance a few weeks ago seems to have done.

      It’s difficult to handicap these things in advance, and as our host has pointed out on his Twitter feed, doing so is almost useless anyway. However, I’m not too worried about the upcoming debates. After Monday, I’m pretty convinced that Clinton’s got this. The biggest danger by far is if we get complacent and don’t turn out, but Clinton also seems to have a much better GOTV operation than the Orange Fascist does, so even there she’s got something of an advantage.

    • Gil

      JFCC: wow, the Vox article is stunning, absolutely eerie in how it predicted what would (and did) happen.

      Absolutely spot on analysis.

    • Olav Grinde

      I remember reading that article when it appeared. Ezra Klein’s has long been one of my go-to sites for intelligent news coverage.

      Another is the Christian Science Monitor, which both in this election cycle and the last one has had a strong focus on the ongoing attempts at voter suppression.

      And then, of course, it can be beneficial to look at the USA from abroad: The Guardian, the BBC, and occasionally Al-Jazeera.

    • Alex Coolchien

      I agree with your observations about the near term ups and downs. BTW, I also intend to vote for the first time as a voter, though I’ve been a citizen for over a decade. I did live in the US back in the 90’s though, and I think if you look at the trends over the decades, a lot of her negative perception has something to do with how she is in the public eye, and there are probably subtle gender expectation here. At the beginning of the first Clinton administration, she was absolutely vilified. There were outlandish rumors left and right, and you know why, it was because she stepped into a politician’s role by pushing for healthcare reform. Later on, I think she withdrew a bit from those core political fights, though still kept on championing women’s issues, e.g., going to Beijing for the world women congress. Those played well for her because that was expected of a first lady’s public role. Then, in the second Clinton Administration, I think she won quite a bit of sympathy because of what Bill did, and the impeachment, etc., which again had her in an expected role of an aggrieved wife. I think her Senatorial sting was not that much controversial because she kept a bit of a lower profile, but then into the last 8 years, her negatives picked up again, as she was first expected to and then truly running for President. So net-net, I think the country is still learning to live with a woman at the center of national politics

    • Amitabh Lath

      While I can see the effect I cannot think of how to engineer this “drop in polls proportional to time from last major election event”. How does this work at the voter level? Does a voter get all enthusiastic about Clinton watching the convention, then after a few weeks, meh?

      If I constructed an observable that displayed this sort of “time decay” behavior I would suspect my methodology.

      Could this be an LV filter issue? If there were a significant population of voters right at threshold, falling in and out of the LV pool with the tiniest nudge, one could get this sort of behavior. Perhaps instead of a binary LV determination, there should be an LV weight applied.

    • Marc

      @Amitabh Is it possible that existing supporters become more likely to respond to the survey following a large positive event for the candidate? I think I remember hearing that theory somewhere.

    • Ed Wittens Cat

      Jeez Amit– its the Second Law– entropic decay
      w/o a perceived “NEED to vote”
      eg: convention, debate
      the required energy to vote disappates

      that said, the media is furious over Trump scamming them on his “birtherism presser”
      they fully realize it was an infomercial with 39 seconds of actual birtherism content…
      and they are pissed
      even FOX wont support the (hashtag)TrumpWon eumeme
      i kno PEC is dedicated to things-that-remain-the-same-in-large-samples (Gauss) but im dedicated to things-that-change (Per Bak)
      sry for the rant :(

    • InmanRoshi

      I think this is right. Trump’s campaign is consulted and run by Roger Aisles and Breitbart. It’s managed to dominate the newscycle, which dominates the conversation in the 2 month lull between the conventions and the debates.

      When the media get elbowed out of the conversation, and it’s just a contrast of the two candidates, the contrast is starkly tilted in Clinton’s favor. That’s why it still wouldn’t surprise me if Trump backed out of the remaining two debates. When the election is a matter of manipulating the newscycle, and less about making a binary choice between the candidates, it’s playing on his turf.

  • Matt McIrvin

    If anything could convince me that our analyses are all wrong and Trump is going to win huge, it’s this Gallup poll about voting intentions, which suggests that this is going to be an unusually low-turnout election, and enthusiasm among Democratic constituencies is particularly low. But presumably the polls with LV screens are already building these assumptions in.

    • Joeff

      Dems are said to have a vaunted turnout machine, which does well during presidential years. I’d be surprised if this turns out to be accurate.

    • Matt McIrvin

      I also find the results from September 2000 interesting: more Americans intended to vote, but far fewer said they had spent a lot of time thinking about the election. It was a different time…

    • Matt McIrvin

      Anyway, I think it’s clear that if the election returns turn out to be way off of the polls in Trump’s direction, it could be less because of a Bradley Effect than because of differential turnout.

    • Brian Gray

      Just remember Gallup is the same organization that all the way till November of 2012 that the demographics would drop to the level of 2004(just 22% non-white) when it actually ended up something like 26-28% nonwhite..

    • Sam Wang

      Maybe, but it is better to avoid getting too wrapped up in past errors. Any indication of low turnout would surprise me…but it bears watching.

    • Brian Gray

      Sam I agree one shouldn’t focus too much on the past sometimes when people get it wrong they are the ones making proper adjustments .IMHO if it’s close people will get out to vote…

  • Marco

    I always check the Pollster site (it seems every 15 minutes or so. I am THAT scared). From yesterday to today, the margin Clinton-Trump went from +3.5 to +4.5%. When including the other two, the margin went from 1.8% to 2.7%. Just noise, or the beginning of the bump? The meta-margin should follow suit in the next 12 hours.

    • Some Body

      The MM is based on State polls alone, so you’ll have to wait at least a good few days, not 12 hours.

  • Kurt

    My expectation is that the debate bounce for Clinton will be (forgive me) HYUGGE, and while it will fade it won’t completely decay. Clinton’s supporters are not wild about their candidate (though her numbers are actually quite respectable for self-proclaimed Democrats) but they are VERY nervous about Trump. So I agree with Sam that the bases are stable.

    Clinton showed herself as warm, calm, friendly and with a sense of humor that many people who have not been following her closely haven’t seen. The Clinton shimmy – when she scored a clear point against Trump and did a little shimmy dance with her shoulders – very briefly turned a 68 year old grandmother into a teenager, and I think that one moment has made her approachable to at least undecided women. Trump’s boorish performance did not help there easily – I was picking up a lot of anecdotal comments from the Bernie-sphere especially that those that may have been thinking about Jill Stein have probably moved to Clinton after this.

    Trump’s income tax statement didn’t help him either … nobody likes paying income tax, but to brazenly admit that you are successfully cheating the system will likely have soured many small to mid-sized business owners who do not have that luxury.

    Finally, the media’s fragmenting. The debate was played up for ratings, and it was the largest watched debate in history, by a significant margin. The VP debate might move the needle some, but neither Kaine nor Pence are anywhere near as well known. The remaining debates will be much less heavily watched, the media knows it, just as they knew that this last debate was the Superbowl for them. My guess is that Trump is going to get considerably worse press moving forward, even as the media jockey for spots in a Clinton briefing room.

    So, my expectation is that there will be a major bounce, and some decay, but nowhere near the constant negative drumbeat against Clinton that it had been for weeks.

    • KevF

      “…brazenly admit that you are successfully cheating the system…” Zero taxes can be perfectly legal without any cheating at all. So the assumption of cheating is a rush to judgement that I would expect from right wing propagandists if the roles were reversed.

    • Jinchi

      “Zero taxes can be perfectly legal without any cheating at all.”

      To me this is the bigger point. Trump isn’t able to pay zero taxes because he’s smart. He’s able because he is a billionaire. The system is designed to his advantage. The rest of us don’t have that option. He’s happy to have the middle class carry his burden for him, just as he’s happy to cheat contractors out of the pay he owes them.

    • Matt McIrvin

      There’s some evidence that Trump directed money owed to him to the Trump Foundation, and used Trump Foundation money for his own expenses. The first is legal as long as you paid income tax on the money, but you can’t use it as a tax dodge; Trump’s people seem to be stonewalling on whether they actually did that. The second definitely isn’t legal, though it may not be a criminal violation.

    • Commentor


      You seem to have hit upon a very specific reason why one wouldn’t want to release their tax returns.

      Makes you say, hmmm.

    • Kurt Cagle


      Yes it is certainly legal, but it was also an ill-advised statement for him to make. Trump is pushing for a populist appeal, but the tax admissions serve to highlight how different the world is for the ultrawealthy. Coming the same week as the Wells Fargo disclosures was especially bad timing, because it emphasizes that privilege, especially with the working class base. When your primary message is how bad the country has become but you’re eating caviar off gold trimmed china, there’s a dissonance that is “bad optics” this close to an election.

  • Olav Grinde

    A big point has been made about certain developments only changing the likelihood of survey response.

    Well, perhaps – just perhaps – there might actually be a correlation between likely to respond to the survey and likely to vote.

  • A

    New Public Policy Polls out: Headline, Clinton Leads in Key Battlegrounds; Seen As Big Debate Winner

    She’s up in battlegrounds around 4 points on average.

    Rasmussen has Clinton up 1 nationally, but it had Trump up +5 on their previous poll I believe. Thats a swing of 6.

    The bounce is real. If she gets a big enough bounce, hard to see it wearing off in time for Trump to truly make a comeback.

    And then we can hope for the media to try and form the “blowout” narrative based on “Trump choking” in debate #1.

    Time will tell…

    • Adam

      I was coming to remark on those same polls out this morning. I am curious to see the correlation with the Meta Margin increase. Given the states involved, I would assume a few decimal points with the noon update.

  • Kevin

    I think Nate Silver is calling you out on Twitter (again) this morning. He used the word “sophomoric.”

    • Kevin

      I don’t know where the methodological difference in calculating standard deviation comes from–is it the use the two candidate margin versus absolute movement? RealClearPolitics average versus 538’s adjusted numbers?

    • Sam Wang

      I’m going to do everyone a favor and not read that. Others, let me know if it is actually specific.

    • George

      From your link I didn’t see anything specific to Sam. You’d have to provide more links or a more detailed explanation.

    • Daniel

      I agree that there’s nothing specific to Sam there. To me, it reads like it’s aimed at horse-race commentary in the press, not at polling analysis.

      Either way, given the way 538 covered the Republican primaries, I’m not sure he has a lot of room to criticize people for ignoring polls.

    • Slartibartfast

      Kevin said: “I don’t know where the methodological difference in calculating standard deviation comes from–is it the use the two candidate margin versus absolute movement? RealClearPolitics average versus 538′s adjusted numbers?”

      Just a guess, but if Sam is using the SD in the meta margin and Nate is using the SD from his model (or even the RCP polling average), then I would expect Nate to get a higher value.

      Of course if Nate did what Sam does and used a transparent methodology, then we could just see what caused the discrepancy ourselves and wouldn’t have to guess.

    • Kevin

      His first tweet goes to the volatility of the polls this year in compared with previous cycles. He has a table measuring the standard deviation, arguing that 2016 is less stable than 2012, and that 2012 is an outlier year. I read Silver’s follow up tweet about “sophomoric analysis” as referring to the differences between elections modelers as to how much uncertainty to build into the November prediction, where PEC and Silver represent the opposite poles. Looking back, I think I read the second tweet incorrectly. I am still interested in why different analysts seem to be measuring the volatility of the polls differently.

  • Ben Alpers

    But the metamargin and e.v. count up top just moved toward Trump. Explanations?

    • George

      Saw that and I had the same question. Old numbers falling off? Close states being polled? Is there in fact a link to the raw state poll data in spreadsheet format where we can watch and track what is coming on and going off to create the new number?

  • Christian

    How did the meta margin just fall? Was there a significant poll release after the PPP state polls? They seemed quite strong for HRC.

    • Sam Wang

      Yeah, not sure about that…maybe some medians dropped?

    • CW

      Florida. The latest poll of Clinton +2 aged out two polls of Clinton +5 and Clinton +2. That moved the median from Clinton +2 to Clinton +0.5.

      Florida was the big driver of the MM lift off it’s recent low, and Clinton’s lead was probably a overstated due to randomness (I think it jumped from a tie to Clinton +4% which didn’t really seem plausible). This is just a correction to that randomness. My guess is more polls will cause Florida to settle in at a median of Clinton +2%

    • Sam Wang

      Thank you! This is exactly what people want to know, including me.

    • Kari Q

      I’m not sure either, but isn’t it too early for any debate bounce to show up? Today would be the earliest we could really expect to see it in national polls, so it should be another week before we see something in state polls, right? In any case, it doesn’t seem to be worth worrying about at this point.

    • Camhilfan

      Is that South Carolina Winthrop University poll correct? They don’t poll a large sample of likely voters, but only a 4 point lead for Trump?

      There’s no way it’s THAT close, but if she’s making up even some ground in that state, then the Florida, North Carolina, Nevada, Pennsylvania, etc numbers should look a bit better soon.

  • Louis

    Odd, the Meta fell as today’s state polls should help Clinton. I’m guessing Pollster has not yet uploaded them into their batch. Today’s polls:

    Colorado: Clinton 46%, Trump 40%, Johnson 6% (PPP)

    Florida: Clinton 45%, Trump 43%, Johnson 3% (PPP)

    North Carolina: Clinton 44%, Trump 42%, Johnson 7% (PPP)

    Pennsylvania: Clinton 45%, Trump 39%, Johnson 6% (PPP)

    Virginia: Clinton 46%, Trump 40%, Johnson 7% (PPP)

    South Carolina: Trump 42%, Clinton 38% (Winthrop)

    • Sam Wang

      HuffPollster’s feed is in the lower left sidebar, and a scraped version of their data is in the lower right sidebar. The scraped data is further processed to remove duplicates between 2-way and 4-way and apply PEC’s rules.

      In this case, which PPP result was used depends on the order in which PPP reported the results. I agree that the results seem favorable to Clinton…however, less than half of a percentage point may still be noise. The possibility exists that Clinton got no lift from the debate…but let’s wait for a basket of multiple pollsters to report before reaching that conclusion.

    • 538 Refugee

      I was starting to wonder about the generic house after looking at the results manually for the past couple days but see it is finally moving back up.

  • Joeff

    Your stable model has been a bit of a bumpy ride of late. I’ll venture a guess that because a number of big/medium states are on the bubble, the numbers jump around a little more than one might expect.

    • Sam Wang

      Oh, please. You’re complaining about changes of in the snapshot MM of 2% below the year’s average, and changes of <15% in the probability since its peak. Really???

  • Michael

    Most reports have women at 53% of the electorate, and overwhelmingly opposed to Trump, if not completely in favor of Clinton. Is there a data as to what percentage of a demographic basically seals the deal one way or another? (The same question applies to Hispanics, African Americans, and others, as well.)

  • Joeff

    Ack! Not complaining–observing. Also I’m paying more attention to EV numbers than MM.
    PEC is my rock and salvation. Would love to see 2012 and 08 for comparison purposes.

Leave a Comment