Princeton Election Consortium

A first draft of electoral history. Since 2004

It’s All Over Now, Baby Blue: Debate #3

October 19th, 2016, 8:02pm by Sam Wang


If you don’t believe me…yesterday, PaddyPower paid out its Clinton-to-win bets.

Tags: 2016 Election · President

90 Comments so far ↓

  • Lance

    I thought this was a good read from 538 (http://fivethirtyeight.com/features/election-update-why-our-model-is-more-bullish-than-others-on-trump/). Any interest in a similar post/response from the PEC perspective?

    • Sam Wang

      Generally I find their “explanations” to be a bit grating. But I guess I should respond at some point.

      …how much do you care, anyway? The race is basically over.

    • Lance

      I won’t generalize about their posts (I don’t think I read them regularly enough), however, I find it interesting to think about the models used to estimate the election. I realize you believe that “the race is basically over”, and I tend to agree. But that comment is just sidestepping the issue.

      To me, the interesting question discussed in the post is about the underlying structure and assumptions of the predictive model used to quantify the likelihood of election outcomes. Usually, when faced with this type of question, I’d look to find some data and test out the models. But this being the real world of which we have no suitable model, I suppose we’ll never really know. We’re left to then discuss resonable assumptions and attempt to find ways to test those.

      In particular, the assumption that some state outcomes are highly correlated makes some sense to me, and seems to be the major difference between 538 models and others. It seems to me that you don’t believe incorporating this assumption adds value to the models predictive power. I’m curious about the basis of that belief. My apologies if this has already been discussed elsewhere.

    • Sam Wang

      It’s on other threads. Basically it’s included at PEC too, in the t-distribution for movement in the Meta-Margin. That’s the obvious place to deal with it. It’s basically a judgment call.

    • David Elk

      In the other thread someone asked about state correlations…

      The explanation in Mr. Silver’s post involves MI, NC falling if PA falls. If PA (median margin +7-8) falls, of course one would expect a large national opinion swing causing all swing states to fall. The meta-margin captures this by incrementally testing the EC at various levels of national opinion swing, assuming all states move at the same rate. It should in theory test points where PA is R-leaning and therefore other swing states are too, but assign them a very low probability. (PEC trusts polls more than 538, I think?)

      It’s not clear from the article whether 538 correlates states in related clumps or whether they weight states differently based on how they respond to national polls, or what.

      Caveat: don’t know, anything I say could be wrong.

    • Ed Wittens Cat

      No Lance its not a “good read”
      its more armwaving clickbait
      Give me Math or give me Death

      but this is cute –>
      https://twitter.com/republicofmath/status/790683209826787328

  • Ed Wittens Cat

    Word…
    I think the Argent One has lost his last mathematical nerve.
    How did he get from this –>
    https://fivethirtyeight.com/features/donald-trumps-six-stages-of-doom/
    to this?
    https://fivethirtyeight.com/features/election-update-why-our-model-is-more-bullish-than-others-on-trump/
    Nate is a broken man…hes lost his faith in the Maths

  • Mario

    during World War II, Argentina remain steadfastly neutral until the outcome became inevitable, and they formally declared war In March 1945 to join the allies. Is there a quantifiable similar effect in this election, not just for independents but for previously committed Trump supporters? The psychology being that someone does not want to admit to voting for a loser…

    Also, Sam, can you point to East Coast congressional elections that will become the bellwether of democratic control in the Senate or the house? I.e., which elections can we Watch early returns on to get an indication of whether there will be a change of control?

    • Sam Wang

      David Wasserman listed a few key districts to watch at the end of our podcast with him. It included Bruce Poliquin’s district in Florida, Barbara Comstock’s district in Virginia, and John Mica’s district in Florida. You can find these by poking around the Competitive District Finder in the left sidebar.

    • Olav Grinde

      Small correction: Bruce Poliquin is the incumbent in Maine’s 2nd Congressional District – we drive by his signs and the challenger Emily Cain’s signs all the time.

    • Karl Hudnut

      I will not try to answer the question posed. But I will point out, in the “Generic Congressional” at Huffington Post, undecided = 11. Hard to think that will break in favor of R’s. Sam’s number for likely Dem control is at 8. So only need 3+. Dreams can come true? (BTW: I do not dream of killing a real conservative party for ever. But, my opinion, these guys gotta go. Need to be replace with a conservative party that has actual ideas that might actually work.)

    • Sam Wang

      Yeah, I’ve been obsessing over those undecideds. There are a lot of them.

  • Olav Grinde

    Ezra Klein of Vox.com offers a great analysis of the 2016 Presidential Debates as a whole. Hint: the title of his article is Hillary Clinton’s 3 debate performances left the Trump campaign in ruins.

    Klein’s concluding paragraphs:

    “It is easy, now, to assume her victory was assured, to read Trump’s collapse as inevitable. But remember that he triumphed over a talented, 17-person Republican field in debate after debate to win the primary — one-on-one contests are unique, it’s true, but there was no particular reason to think Trump couldn’t use his bullying, blustering showmanship to take over the stage and expose Clinton as inauthentic and out of touch. The reason he didn’t is because she never let him.

    “We aren’t used to this kind of victory. We aren’t used to candidates winning not so much because of how they performed but because of how they pushed their opponent into performing. But the fact that we aren’t used to this kind of victory doesn’t make it any less impressive. Hillary Clinton has humbled Donald Trump, and she did it her way.”

    http://www.vox.com/policy-and-politics/2016/10/19/13340828/hillary-clinton-debate-trump-won

  • James

    Back in 2000, I noticed a mismatch in the odds on two sites. By betting on Gore on one site and Bush on the other, I was guaranteed a profit. That’s what I thought until one of the sites never paid me out. To this day, my junk mail folder fills with spam from online wagering sites, and I’ve even gotten occasional robocalls and texts.

  • Mike

    Slightly off topic, but does anyone have a better name for the online “polls” run by Drudge and Breitbart, and the like? Calling them polls just creates confusion for people who don’t follow these things closely. We need a word that conveys the concept more clearly.

    • Michael

      Online “surveys” seems to be used sometimes.

    • Chuck

      I can think of a few, but they are unsuitable for public consumption.

    • Rob

      Just like junk bonds, I’d call them junk polls.

    • Jerry Wechsler

      “Click-Polls” ?

    • Mike

      @Jerry – I’m liking “click” because it implies that people can click as many times as they want, which is part of the problem with these things. I still wish we could get rid of the “poll” part. Maybe “click tally” ??

    • Olav Grinde

      Click-surveys.

      (I would presume the better click-surveys prevent users from voting more than once.)

    • Chip

      @Olav

      “I would presume the better click-surveys prevent users from voting more than once.”

      Harder than it sounds. To swing online polls people delete cookies and use proxy servers to obfuscate their IP addresses – and can doi it in an automated fashion. They can determine the voting URL (AKA votevote.com/thanks.asp?orgname=charity&vote=true) and post the links on politically friendly sites and Facebook pages and to Twitter followers. And there are even very simple workarounds for polls that require email confirmation. Finally, for those who are serious about hacking online polls there are ‘vote brokers’ who offer their services a la Amazon Turk.

  • Rachel Findley

    Priority: Senate or House? I’m certainly ready to forget The Donald, but I’m not ready to pivot from Senate to House.–I send somebody a small contribution whenever I feel exercised about the state of the nation.
    Two senate races are tied, and PEC has Senate control probability at 50-50. In the Pennsylvania race, McGinty, the Democrat, has come from way behind. Does reversion to the mean predict that she will lose, or what? And if she loses, but the NH Dem candidate wins, who controls the Senate?

    • Sam Wang

      I don’t know. My current thought is Senate, but it’s close. Basically Senate at ~70% is a conservatve-ish approach, and House at ~20% (my guess) is an aggressive approach. Probably you’ll get the most satisfaction from addressing the PA and NH Senate races.

    • MAT

      I think the Senate. With the Senate comes the Supreme Court and even if the House remains R the majority will be small and divided. Dems should have the largest coalition and moderate R’s are likely to find the distance between them and Dems a lot narrower than between them and the Freedom Caucus and Trump faction. Watching the upcoming Speaker election if Rep’s hold the House should be fascinating.

  • Shawn Huckaby

    Btw, I appreciated another Dylan callback in your title, Sam. I sense a theme emerging!

  • Ed Wittens Cat

    Hillary was surprised that Trump raised Mosul offensive, and so was I. This is another example of her inability to be nimble and antifragile.
    There are 5k US troops supporting the Mosul offensive, SOF and advisors, and Trump just denigrated our armed forces.

    • Ed Wittens Cat

      not nimble– i meant *agile*
      that is a better descriptive of Hillary’s problem…she is very workmanlike. Not a bad quality for a president, but hampers her RT debate skillz.

    • Rob

      Well then perhaps that raises the question as to whether good debating skills equals a good president. I’ve encountered bad employees who mastered their job interviews because they grasped the format properly. Maybe being a whiz at a debate isn’t as important as it’s weighted.

    • linda

      Nimble and agile or not, she resoundingly won the debates.

  • bks

    Just for fun:
    Hillary Clinton is the overwhelming winner of an online poll asking readers of conservative news outlet Breitbart whom they thought won Wednesday’s presidential debate.

    As of 12:20 a.m., more than 80,000 votes have been tallied, with Clinton scoring 59 percent of the vote to Trump’s 40 percent.

  • 538 Refugee

    I just watched a split screen of Clinton/Trump during ‘the answer’. At first Clinton is amused. Then she puts on her game face. Then I think it started sinking in just how bizarre/dangerous it was getting. Unfortunately the video clip cut just at her answer.

    After the debate he stayed at his lectern as Clinton went up to shake hands with Chris Wallace. He didn’t even move until Hillary moved towards the crowd. I also saw clip of him going to his vehicle. Very somber mood from everyone in his entourage.

  • Michael Hahn

    If you would like to read a truly devastating column analyzing the outcome of the debate tonight, I can recommend the Anthony Zurcher column on the BBC website.

    http://www.bbc.com/news/election-us-2016-37711218

    I was left speechless when I finished reading!!

    • Michael Hahn

      Not a good night for Trump – Anthony Zurcher, BBC News, Las Vegas

      Donald Trump tried to be restrained. He really did. During the first section of the third presidential debate, when the topic was the Supreme Court, if you squinted you could almost imagine that this was just another presidential race, with two candidates squaring off and vigorously discussing their public policy positions on abortion and gun control.

      Even the immigration discussion started reasonably civilly, until Clinton pivoted to turn a question about WikiLeaks into an attack on Donald Trump’s relationship with Russia and Vladimir Putin.

      At that point, interrupting Trump returned. He called Hillary Clinton a liar and after she called him a Russian puppet, he said that was in fact her. By the time the topic turned to “fitness to be president,” the stage was set for a total meltdown. He said the women who has accused him of sexual harassment were in it for the fame and Clinton campaign stooges.

      Perhaps most remarkable, however, was when moderator Chris Wallace asked Mr Trump to support the election results, no matter how they came out. He refused to do so. He accused the media of poisoning people’s minds. He said Mrs Clinton shouldn’t have been allowed to run for president. It was Mr Trump in a bunker, settling scores and lashing out at enemies real and perceived.

      At that point, the headlines were written and the debate results were locked in. It was not a good night for the Republican.

  • Ed Wittens Cat

    The big thing for me was that Hillary tagged Bernie on her debt free education policy.
    i dont kno if this impacts any other demos.

  • Leading Edge Boomer

    I admire Prof. Wang’s analysis of the national polls and their correlation with down-ballot voting. But I have tons of reservations that an 8% Clinton popular vote majority puts the House in danger for Republicans. Surely they will lose some seats, and maybe (pending much more information) the House can be turned Democratic in 2018. Convince me.

    • Sam Wang

      No, that’s backwards. Even if Democrats get the House this year, there is a good chance they will lose it in 2018. The President’s party usually loses seats at midterm elections. The issue here is whether HRC gets two years of a friendly House, or zero.

    • Scott J. Tepper

      Sam that’s generally true. But will Trump’s candidacy destroy the current Republican Party — which could help the Democrats in the off year?

    • Sam Wang

      I agree that such a scenario is not crazy, and would upend the usual tendency. However, it would also be looking a bit far ahead. I’m just saying the actual ball of wax is now, this year.

    • Bob Wallace

      We might see an all out effort in 2018 with Barack and Michelle Obama leading the charge to get voters to the polls and seize Congress.

      Just because something hasn’t happened before doesn’t mean it never will.

    • Amitabh Lath

      The basic math behind gerrymandering is that you maximize number of districts for a given party at the expense of win margin. That, and the predilection for straight-ticket voting means a large wave can overtop these shortened levees. We can argue about what margin (6%, 8%, 12% …)would flip the house, and if a larger numbers of voters will split the ticket this year, but the basic argument is fairly straightforward.

    • Kevin King

      While the out party usually gains in the midterms, there are exceptions depending on events. In 1998 the Democrats gained seats despite Clinton’s being in office, and in 2002 the Republicans gained while Bush was in office. At this point it’s impossible to say. I’m not writing that it will happen, but imagine a scenario where Republicans keep the House overreach with respect to obstruction and spurious investigations, especially if they also keep the Senate. The Democrats in that case might be able to nationalize the elections and run against a “do-nothing Congress.” It could be similar to what happened in 1998, where part of why the Republicans lost seats was the impeachment.

    • James

      The only way Republicans do even worse in 2018 is if an entirely new party is created to split the conservative vote, and fields viable candidates in a significant number of districts. That takes a lot of work – the two parties have a huge amount of local, state, and national organization. I don’t see Trump doing that. I don’t see any Republic donors putting money into that project since it will only hurt them.

    • liberal

      Kevin King wrote, “…but imagine a scenario where Republicans keep the House overreach with respect to obstruction and spurious investigations, especially if they also keep the Senate. The Democrats in that case might be able to nationalize the elections and run against a “do-nothing Congress.” ”

      Yes, but…

      (1) The Democrats seem less willing to try to nationalize Congressional elections than one might otherwise expect.

      (2) Most voters are quite ignorant of these things and aren’t able to pin the blame on congressional Republican obstruction.

      (3) The media will take something like Republican obstruction of the Zika bill and speak about Congressional inability to pass the legislation, instead of focussing on Republican efforts to hold such legislation hostage by adding stuff to it like defunding Planned Parenthood.

    • alurin

      The thing about gerrymandering is that it has an expiration date. People are constantly moving, demographics are shifting and changing allegiances. By 2018 the 2010 gerrymander will be 8 years old. Democrats could buck the trend.

    • Prehistorian

      James, given that it looks as though there may be sufficient anti-Trump Republicans defecting to the Libertarians to push them over the 5% line, might it not be that with extra funding and credibility they start to take more votes from a warring Republican party?

  • bks

    Earlier today I was going to post a comment that we had reached peak outrage and that Trump couldn’t really do anything more to surprise us. Luckily (for my own self-esteem) I thought the better of it. Preaching sedition, in the words of zombie Jim Morrison, broke on through to the other side.

    • Scott J. Tepper

      There are two rules in dealing with Donald Trump:

      1. TRUMP WILL ALWAYS MAKE IT WORSE.

      and

      2. THE ONLY PERSON WHO GETS ANYTHING OF VALUE OUT OF A DEAL WITH DONALD TRUMP IS DONALD TRUMP.

      I had confidence that Trump would blow this debate. In much the same way he has blown the election. And now I see there is a 99% Bayesian probability that Trump will lose. It’s over. If Trump continues to campaign the way he has, he’ll only make it worse for the downballot candidates. And Trump will violate the second rule – he will not get anything out of this campaign except the tarnishment of his own brand.

      Please proceed, Donald.

    • Ed Wittens Cat

      bks:
      there is no “peak outrage”.
      because of the continuum hypothesis.
      This is a Math site, correct?
      at a minimum outrage is transinfinite.
      :)

  • Bruce Sands

    Yes, this is what we need, a big dose of “we got it won” and, come election day, oh, why bother, we’re up 9%. What could go wrong?

    • Ed Wittens Cat

      yup, we are in the home stretch, dont let up.
      but we should be allowed a little celebration, non?
      Hillary basically formed Voltron and cut the monsters head off.

    • Marybeth

      Are you sure that complacency will take over? Maybe the opposite will occur–Clinton supporters flock to the polls in optimism and relief and Trumpites will stay home in despair.

    • Matt McIrvin

      Only Democrats ever worry that people won’t bother to vote if they say they’re ahead. Republicans figure that people like to come out and vote for winners, to get on that winner bandwagon. And I think they’re right.

    • Chuck

      Speaking only for myself, I have never been complacent. Trump and his cult are an existential threat to America, democracy, and the rule of law. We need every vote we can get.

      And it does appear to be time for a requiem mass for what was once the party of Lincoln and Ike. It’s dead, Jim.

    • Andrew Heil

      First, I see nowhere in this website mission statement an objective of encouraging Democratic voter turnout.

      Second, voting tends to be largely an emotional experience for the majority of voters. Knowing your candidate is very likely to lose is discouraging – people don’t like being a part of that. Conversely, people enjoy voting with the knowledge that they’re pulling the lever for a winner. And this is borne out by election results that more often than not show late fades for the candidate losing badly.

    • liberal

      Agree with the others who think the dominant effect is “everyone loves a winner.”

      Myself, I’m really looking forward to filling out my ballot on that Tuesday in November.

    • Michael Hahn

      Have already filled out my ballot!! LOVE early voting. Now I can sit back and enjoy the show.

  • Ed Wittens Cat

    im visualizing Kelleyanne Conway pulling her hair out in handfuls.

  • Ed Wittens Cat

    Down the rabbit hole– Trump took the Putin bait.
    i can see the SNL skit from my house.

  • Matthew Coons

    The only interesting thing to watch on election night will be to see if McMullin can win Utah.

    • Sam Wang

      Also control of the Senate and House.

    • Matt McIrvin

      Senate control is highly uncertain and extremely important.

    • Jay Bryant

      I’ll definitely be watching for the down-ballot races and for the exact margins of victory. A large margin could make governing a tiny bit easier (I hope).

    • Kevin King

      Not only Congress, but numerous local races, including referenda on cannabis legalization, including in California. Not only that, but can Clinton win a landslide?

    • Charles

      It will be fun as heck to watch if Utah does end up being a three-way race between Trump, Clinton, and McMullin.

    • Matthew Coons

      I was specifically referring to the presidential election. As for the House I see virtually no chance it changes. The Senate will be closer after the election than before and might switch. IMO Congress is more important than the presidency and I’ve been following the Senate races fairly close.

    • Gil

      I for one cannot wait to watch Election Night coverage on Fox News. The schadenfreude will be epic especially Karl “wait, we still haven’t heard from Cuyahoga County!” Rove.

      Oh the look on their pundits’ face … Can’t wait.

      Let the recrimination games begin.

      Please pass the popcorn.

    • Prehistorian

      Gil, Karl Rove has already announced that Trump can’t win, so no schadenfraude there, I’m afraid.

  • Olav Grinde

    Meanwhile, I wonder to what extent this court ruling will influence the result in Ohio:

    A federal district court ruled Wednesday night that Ohio must allow [...] anyone purged since 2011 as well as anyone who has moved within the same county will be able vote.

    …Ohio’s Republican Secretary of State John Husted violated the National Voter Registration Act by purging nearly 2 million voters from its rolls over the past five years.

    https://thinkprogress.org/amp/p/ae4e5d50ae63?utm_content=buffer4ec43&gi=2edc0ecc01c3

    • Phoenix Woman

      Whoa!

      Ohio just dropped into the Dem column.

    • Jeff

      Donald Trump is right. The election is rigged. It’s rigged by Republicans for Republicans. Manipulating voter lists, passing bogus Voter ID Laws, closing down polls in Democratic areas, cutting back or eliminating early voting, preventing voter registration on election day, gerrymandering in states the Republicans control. If all of this isn’t rigging, what is?

    • James

      Two million is yuge number, but it includes people who left the state or died. I can’t determine from the article, but maybe some of the purged voters re-registered. The ones who are left, illegitimately purged voters who didn’t re-register, may be less motivated to actually vote.

    • liberal

      James wrote, “Two million is yuge number, but it includes people who left the state or died.”

      Yep. That’s the main point. Unclean registration lists does not imply unclean vote.

  • Joel

    Is that so they don’t have to take on any more Clinton bets? What happens if Trump wins? They pay out both sides?

    • Sam Wang

      I think that has to be it.

    • Andrew

      I agree, I think you’ve got it.

    • GM

      Yes, and they’ve had to do that at least twice in the past as they prematurely paid-out bets that ended up going the other way. I still hear opera music (although, to be honest, I heard that when Trump was nominated).

    • Dave

      They could have limited the number of additional Clinton bets simply by moving the odds. This probably has more to do with the free publicity.

    • 538 Refugee

      There is a problem with moving the odds if you really believe the odds are zero. Deep pockets can place large bets and still make a decent chunk of change. At some point it becomes prudent for the bookie to just close it out.

      http://www.smh.com.au/news/cricket/out-for-20000–a-punters-nightmare/2006/03/13/1142098402548.html

    • Dave

      (a) I figure deep pockets are also smart pockets and would take the better odds from another bookmaker.

      (b) I dont think anyone at this point actually thinks the odds are actually zero.

      (c) Your link describes the opposite happening, where the “deep pockets” actually lost.

    • 538 Refugee

      Dave, yes the deep pockets actually lost that bet but the point is they normally won’t so the bookie has to make a determination as to whether it is worth it to accept any more bets despite the odds. I’m sure these are software driven decisions. Also, does the bookie want to open up liability for a Trump win payout at astronomical odds? Again, look at the software and make a decision. They probably decided to lock in profits and their own gamble is they are betting against a Trump win.

      That said, does a Trump win get written off to their publicity budget? ;)

  • Andrew

    Well now! Hah ha! :) While I think this is funny as all get out, my inner jaded skeptic is saying this has to be a stunt to get some “free advertising” from the news agencies. Otherwise, what motive do they have? Surely it’s better to sit on $1 million for a few more weeks and collect interest than return it now.

    What I think is even funnier is pink Texas. Oh the possibilities!!

    • Shawn Huckaby

      They actually admitted that was the case. They get the visibility, while still a fairly low risk. They reportedly paid out 2 days early for one of Obama’s victories, but 20 days is pretty amazing!

  • 538 Refugee

    Maybe you could do the bug eating thing on a sliding scale to keep it interesting? There is a local hot dog shop that offers crickets as a topping if that helps any.

    I’m just anxious to see some timely, good quality polling ‘down ticket’ at this point. I think it might be a lagging indicator at this point.

    • Matt McIrvin

      He gave Trump a point spread.

    • JPI

      I’m just anxious to see some bug eating.

    • Chuck

      I once offered to eat a bug on the air as part of a fund-raising gimmick during my weekly show on a local public radio station. Just my luck, we made our goal. I don’t recommend it. (:-)