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

A Test of the Polarization Hypothesis

October 30th, 2016, 3:10pm by Sam Wang

So, on the day that I wrote in the New York Times about how the race is so emotional that no minds will change, people get into hysterics over the Comey/email story. The poetry of that is awesome.

This is actually a good thing. This is an opportunity to apply lessons from this season to predict what happens next.

First, the starting points:

  1. For twenty years, polarization has made voters increasingly emotional and less likely to change their views. Donald Trump represents the culmination of this trend.
  2. On time scales of a week, journalists get bored with a storyline, and look for ways that the trend is being violated. Until Friday, the developing story was “Clinton is coasting to victory.”
  3. Whichever major party you support, your optimal strategy as a citizen is to focus on knife-edge cases, i.e. cases where the outcome is in doubt.

From these, I suggest the following consequences:

  1. The national race will not change meaningfully. This is not a story that changes anyone’s mind. Maybe the margin (national or Meta) between the two candidates will move by 1 percentage point when aggregated…2 points max. It doesn’t change the high likelihood of a Clinton win.
  2. Journalists and pundits will continue to feed hysterics by fussing over the Comey story. They may even attempt to use polling evidence to justify their coverage. However, note that national polls had already tightened by 1-2 percentage points, even before Comeygate.
  3. Keep your eye on the ball, which is downticket. In the Senate, key races in Pennsylvania, New Hampshire, Nevada, and North Carolina will determine control of that chamber. Small changes mean a lot: early voting, get-out-the-vote, bad weather…maybe even the Comey story.

Also, in the House, the Republican majority will narrow, but we don’t know by how much. Apply your energies accordingly. This District Finder App will find a competitive House district near you.

Oh, I’ll go out on a limb on one last item: there is time for one more weird twist in the campaign. Considering the life cycle of journalists’ hidden thought processes, I’ll say it is Donald Trump’s turn for the next adverse story.

Tags: 2016 Election · House · President · Senate

93 Comments so far ↓

  • Some Body

    An alternative hypothesis:

    Most polls find Trump attracting fewer Republican voters than usual, and likely voter screens are unusually kind to the Dems this year. This means that, precisely because of polarization, Trump has some low hanging fruit to pick up, and the Comey story, simply by virtue of making Republicans (and Republican-leaning independents) feel Trump’s chances are higher, could create more than 2% of a slide in the margin (almost entirely coming from Trump’s numbers going up, rather than Clinton’s down).

  • A

    Obama’s job approval numbers keep going up and up, Sam.

    Shouldn’t Hillary’s numbers follow that trend, especially considering how closely they are campaigning together?

  • A New Jersey Farmer

    We already have one piece of evidence that the race will not change significantly, and that’s in the betting markets. Clinton maintains her advantage there. I’m thinking that by Tuesday or Wednesday we’ll have evidence to support your hypothesis.

    Most of Trump’s poll creep in the last week was Republicans coming home. To roost, I imagine.

    • John Jones

      Indeed. I have in fact been encouraging conservative friends who are so very certain that Trump will pull through to put their money where their mouth is a post public proof of their large bets on Trump. After all, if they’re right, at 7/2 odds they only stand to make a tidy profit, right?

      Strangely, none have taken me up on that yet. Apparently talk on social media is just that.

  • (((CassandraLeo)))

    Glad to hear this… which is pretty much what I’d already predicted you’d say, and matched my own assessment of where things were likely to head as well. I’ve thrown some more money at downballot anyway, and will probably volunteer some this week to help with the GOTV operation. I’ve been far more nervous about this election than any other one in my life, not because I think Trump is likely to win it, but because the consequences if he does are likely to be so utterly disastrous in such an unprecedented way. Remembering that this is still unlikely to happen is probably the only thing that’s kept me from being institutionalised at this point.

    • 538 Refugee

      My fear is Clinton doesn’t get a Congress she can work with. Mainly the Senate blocking nominations with even more vitriol.

    • Ed Wittens Cat

      she will have some leverage, 538
      if the dems have the senate, Bernie Sanders becomes chair of the Budget Committee–
      while all legislation must originate in the House, nothing gets to the Senate w/o approval from Budget– room for some horse trading–last time the GOP caused a gov shut-down it burned them pretty good.

      Dr. Wang, you are a light in great darkness.
      And so is Katie Mack.

  • Tony

    Hi Dr. Wang I am curious do you know of a polling outfit called people’s pundit daily? are their projections really as accurate as they say they are?

    • Sam Wang

      Based on who they follow on Twitter, they are probably a fake organization for the Republican side pretending to be neutral.

      Anyway, considering that there is no shortage of sites that do neutral aggregation (us,, FiveThirtyEight, Upshot, 270ToWin, RealClearPolitics…) there is no reason to pay attention to these people for poll analysis.

  • BillSct

    From HuffPollster

    Republican Party Favorable 34%/Unfavorable 57%
    Democratic Party Favorable 46%/Unfavorable 48%

    These numbers have been pretty stable since March. Pair this info with Obama’s favoriability ratings.

  • Martin Schiffenbauer

    Dr. Wang: Do the polls take ito account early voting stats? If so, how do they do it? Thanks for a great website!

    • 538 Refugee

      My guess is an early voter would be counted as ‘likely’ and their preference recorded. They might be scrutinized a little more closely to adjust/confirm demographic models.

  • SF

    Some Body, do you have evidence that likely voter screens have been unusually kind to Dems? I haven’t seen any articles about it.

  • M. Leo Cooper

    As many of you have probably already read, there is at least one news organization that will call winners in swing states while the polls are still open —

    How will this affect voting later in the day? If the election is effectively called for Hillary by mid-afternoon, there could be be consequences for the downballot races if discouraged voters stay home.

    • Lorem

      Elections have been called many days in advance by prediction sites (such as this one) for the last couple of cycles – I don’t anticipate day-of predictions to be significantly more accurate or to make any more difference. Or at least that is likely to be the case while the predictions are just being published on Slate.

      If this were done by all the major media outlets, I agree that it could have the effect of dampening turnout for the expected loser, with potential downballot consequences.

    • Kurt

      I’d agree with that, with Lorem on this. Early voting changes the dynamics of reporting significantly, since you can at least infer likely votes from registration patterns and precinct districts. This is a good thing, as it reduces the time zone effect that traditionally came from a single day registration where the West Coast was still voting while the East was already reporting.

  • A New Jersey Farmer

    And let’s not forget Hillary’s superior GOTV operation, which is already paying off.

  • Tim in CA

    Another reason Donald Trump is cooked: total US voter registration continues to increase, and recently surpassed 200 million registered voters for the first time.

    Although polls take different sample sizes and can only measure their results in terms of percentage point margins, our presidential elections are of course decided by who wins the greater number of total aggregate votes cast in each state. As a result, I find the most informative numbers from recent elections are to be found in the total aggregate votes cast, rather than by percentage point margins, which can suggest that the democratic-republican margin might be more flexible than it really is.

    For instance, compare the total number of votes won by Republican and Democratic presidential candidates from 2004 through 2012:

    2004 – 62 M
    2008 – 60 M
    2012 – 61 M

    2004 – 59 M
    2008 – 69.5 M
    2012 – 66 M

    The biggest takeaway is that the GOP has failed to grow their coalition in Presidential elections since 2004. In fact, GW Bush’s 62 Million votes still stands as the most votes ever received by a GOP presidential candidate.

    Are there any indications that Trump could expand the GOP coalition beyond what was achieved by Bush, McCain, and Romney? Not really. National polls have consistently shown him with among the lowest percentages of support of any recent major party nominee (high 30s and low 40s). Given that there have been at least 15 M new voter registrations since 2012 (see TargetSmart), and that registered voters turned out at roughly 85% rate (or higher) rate in recent presidential elections, we can estimate using conservative assumptions that total voter turnout in the 2016 will be in the vicinity of 140 M. Even if Trump garners 45% of the vote in the general election, he will still wind up in the vicinity of 63 M votes, little changed from other recent GOP nominees who have garnered vote totals in the low 60 Millions.

    Will Trump be able to hold Hillary to less than 65 M votes in order to retain a chance of winning the electoral college? This is not likely, in my opinion. According to TargetSmart, 42.6% of the (at least) 15 M new voter registrants since 2012 lean Democratic, as opposed to only 29% that lean to the GOP. That suggests that the Democrats have added somewhere in the neighborhood of 6.5 M new registered voters since 2012, when Obama earned 66 M votes, whereas the GOP has only added (probably less than) 4.5 M new voters.

    I find it hard to believe that Mrs. Clinton will garner significantly fewer than 67 M votes in this election. If we give Trump an optimistic 63 M votes (45% of 140 M voters), and if we give third party candidates a very optimistic 10 M votes (7% of 140 M voters), which would be the most votes for third party candidates since 1992, we still find a floor for Mrs. Clinton around 67 M votes (48% of 140 M voters). And all of these assumptions are conservative. For Trump, 45% is right at the very high end of what he has received in national surveys, whereas Mrs. Clinton has routinely received numbers of 48% or higher throughout the election season.

    None of this should be surprising. As Sam has detailed exhaustively on this site (and in his excellent piece in today’s NYT), we have an increasingly polarized electorate in which very few voters are truly up for grabs by either side. At the same time, the American electorate continues its inexorable shift toward the Democratic party. Since 2004, Republicans have been unable to grow their voter base in max-turnout Presidential elections, whereas Democratic voter base has continued to grow. 2016 is not shaping up to be any different in this regard. The reasons behind these trends are not difficult to ascertain. Voters enter the American electorate primarily by turning 18 and by immigration. Both constituencies lean significantly Democratic. Whereas, the primary way that voters exit the electorate is through death. GOP votes skew heavily toward older demographics, and thus they must contend with a relentless drip as many of their voters age out of the electorate altogether. It is for this reason that the max-turnout GOP voter coalition in presidential elections appears to be running in quicksand: they are struggling to just to add enough new voters to their coalition to cancel out the ones who are passing on. On the flip side, the Democratic party adds millions of voters to their coalition every four years, rendering it harder and harder to understand how they could possibly lose another Presidential until/before the GOP drastically alters their political strategy and tactics.

  • N Merton

    The +2 map is quite worrisome. If this becomes a referendum on Clinton rather than Comey, the magic bullet of Fla., NC, NH, and ME-2 could put that man in office.

    • Jay Sheckley

      Yes. Maybe the plus 2 map is what we need to link in social media to get people to get off their kick about protest votes and protest nonvoting.

  • Roger

    Somewhat related: One of President Bush’s chief ethics attorneys and other attorneys have filed official complaints with ethics oversight commissions that Director Comey likely violated the Hatch Act.

    After filing the report the attorney wrote an op-ed piece. Some highlights:

    “The F.B.I.’s job is to investigate, not to influence the outcome of an election,” Painter wrote in the op-ed. “And that is why, on Saturday, I filed a complaint against the F.B.I. with the Office of Special Counsel, which investigates Hatch Act violations, and with the Office of Government Ethics.”

    In the complaint, Painter acknowledged that its is unclear “whether the Director of the FBI personally wanted to influence the outcome of the election” based on his actions with respect to the Clinton email investigation, including sending the letter to lawmakers on Friday. However, Painter argues specific intent is not required because “the Hatch Act and ethics rules are violated if it is obvious that the official’s actions could influence the election, there is not another good reason for taking those actions, and the official is acting under pressure from persons who obviously do want to influence the election.

    Absent extraordinary circumstances justifying it, a public communication about a pending FBI investigation involving a candidate for public office that is made on the eve of an election or, as in this case, while voting is in progress, is very likely to be a violation of the Hatch Act. It is also a misuse of official position. The fact that politically motivated members of Congress want the communication to be made publicly only enhances the seriousness of the violation; it is not an excuse.

    • Michael Coppola

      Was Comey’s communication public? Or was his private correspondence made public by individuals not subject to the Hatch Act?

    • Elizabeth Lawley

      Comey specifically said in his email to FBI employees “I also think it would be misleading to the American people were we not to supplement the record.” And since the letter was not marked as classified, it is a public government document.

    • Ed Wittens Cat

      everyone chillax and read this
      the emails are remarkably similiar to Benghazi– something the GOP base will be furious abt forever which has no prosecutable basis.
      Polarization and the GOPs collective bunker mentality in the face of demographic and cultural shifts are creating an extremely dangerous time for our republic– hope it survives.

    • Ruth Rothschild

      Below is a link to an article in the Huffington Post that you may find interesting based on your post. And, it may give some perspective to the question that Michael Coppola asked and to Elizabeth Lawley’s statement in response to your post. I think that it comes down to whether or not this search of Weiner’s emails was properly limited to the reason for the search (i.e., investigation into Weiner’s sexting and nothing more) and maybe not relevant to whether or not the communication was public. Even if Comey wanted not to mislead the American people with his Friday letter and even if the letter wasn’t marked as classified, there’s still the 4th Amendment to the Constitution to consider as explained in the link below.

      I also think that the link to the New Yorker article that Ed Wittens Cat posted gives an interesting perspective about Comey’s revelation, whether or not it violates either the Hatch Act or the 4th Amendment, in that this new email finding is out in the open before the election so perhaps there won’t cause the stir and accusations that might result if it was withheld until after Nov 8th. However, Comey needs to clarify because his vague letter raised more questions than anything else. And, I agree with Sam’s statement about some dirt coming out about Trump before 11/8 (since this appears to be a consistent cycle in this election), as this current Comey stuff ages and becomes less interesting and the media finds the need to generate “interesting” news. It’s getting to be like “Peyton Place”.

      Best thing right now is to GOTV for both the presidential race and the down-ballot races since voter complacency or apathy because they don’t like the candidates could be more damaging than these news stories.

  • JPI

    The last bit about it being Trump’s turn next rings very true. It seems like the national media just has to dole out the negative stories in equal measure, or they feel that they haven’t done their jobs. I’m actually surprised that Trump hasn’t done the thing that he has always done during this campaign when things have gone his way: said some absolutely insane and horrible thing to sabotage himself.

    • Matt McIrvin

      He’s carried on saying insane and horrible things just like always (the latest thing is telling his own supporters that poll workers will tear up their mail-in ballots), but they only intermittently get major traction.

      The pattern I’ve noticed is that it actually damages him are when he gets baited into a bigoted or sexist attack on individuals who are not politicians. As long as he’s railing against authority figures or against vast, nebulously defined groups of people, he’s fine. But when he went after Judge Curiel, the Khans, and Alicia Machado, that actually hurt him.

    • Seth Gordon

      “Let’s think of ways to cast suspicion on the Clintons” has been a journalistic pastime since the 1990s. In this campaign season, the media reverts to that pastime whenever Trump can spend three days without chewing on his foot.

    • JohninDenver

      Trump’s turn could well be Kurt Eichenwald’s Newsweek article today:
      “Donald Trump’s Companies Destroyed Emails in Defiance of Court Orders”

      Mr. Eichenwald tweeted about it coming out on Monday, and said he had written it before last Friday. For those concerned about destruction of email … they just got undercut in a bigly, big league, yuge way.

    • llywrch

      Donald Trump’s chief problem is that he wants every headline to be about him. He’s actually said that he believes there is no such thing as bad publicity.

      Consider last summer during the Democratic convention, Trump actually had an interview. There’s been an informal custom or understanding that when one party has its convention, the other would suspend its campaigning so not to distract attention. This custom has been honored as much out of civility as that it provides a badly-needed break from the stress of campaigning — either to rest or to plan. However, Trump was jonesing for attention, so he contacted a friendly talk show host to talk about basically…nothing.

      So I expect if the Comey story gets too much attention for too long, Trump will do something to bring the spotlight back on him. Most likely, say something outrageous, which will undo the help Comey’s announcement has given him.

      Trump’s lack of impulse control & need for attention is what has made this election so unpredictable. Had the GOP selected someone else as their candidate, the news stories would have settled into much the same rut the polls have — although the polling results doubtlessly would have been far different than they are now.

    • Ruth Rothschild

      Trump may have already done one thing to sabotage himself this past weekend, but it won’t make much of a wave in the media in light of Comeygate dominating the airwaves: Over the weekend, Trump had an African American man, who is actually a Trump supporter, thrown out of a rally in North Carolina when the crowd pointed at the guy. Trump initially said the guy was fine, then in the very next breath reversed course and had him thrown out without even first trying to get the facts straight as to what was going on and whether the man was a dissenter. Turns out that this African American Trump supporter just wanted to deliver a note to Trump (a note of support) and wasn’t a dissenter. Trump’s actions shouldn’t be surprising since this has been his MO since the start of this election cycle, dating back to the primaries. And now, Trump is singing high praises of Comey after months of bad-mouthing him when things weren’t going Trump’s way. But, I doubt that neither this incident at his NC rally nor the fact that one of his supporters at another rally over the weekend yelled “Jew-SA” without Trump calling the person out on the spot for this bigoted epithet will be enough to make this an it’s-Trump’s-turn situation. However, allegedly someone has some inside info. about Trump’s alleged dealings and links with Putin and Russia, and there’s been some noise about a possible revelation about that sometime this week. If this latter is indeed true and if done later this week, I believe that this could be the it’s-Trump’s-turn moment that could eclipse Comeygate in the media. But, I seriously doubt at this point in the campaign/election cycle that any of this, including Comeygate, will swing the meta-margin and EV significantly toward Trump for the following reasons:

      1. Most voters are now strongly entrenched in one of the candidates’ camps and these sorts of Peyton Place dramas aren’t going to sway them away from their candidate this late in the game (i.e., for those who haven’t already voted)

      2. Many people had already voted before Comeygate.

      3. There appears to not be many undecided voters remaining. So, while these sorts of dramas could possibly sway undecided voters, it would seem that a sway toward either side/candidate in this small undecided group wouldn’t be a high enough percentage to be statistically significant that it would cause a huge move in the current EV and meta-margin.

      One advantage to Comeygate is that it actually may motivate Clinton supporters, who figured she had the election in the bag and didn’t need their vote, to get out and vote and support her. Hopefully, this will also mean that these voters will also vote for down-ballot candidates who could help swing the races toward a Democrat majority in both houses (or, in the case of Congress, add enough more Democrats to not have an overwhelming Republican majority there).

  • Jeffrey Lilly

    Dr. Wang, I’d first like to thank you for your hard work and excellent, informative site. Regarding your last point: considering that the media would rather hype e-mails (with nothing to go on, really) than even talk about the fact that the GOP nominee for president has a fraud case pending in November and a civil rape case pending in December, that he recently was accused of $17 million in insurance fraud, and that more women continue to come out and accuse him of heinous personal crimes, I don’t think there’s a snowball’s chance in hell that the media pivots back to attacking him. They have their horse race narrative to flog, and they’ll be damned if the truth is going to take that away from them.

    • Robert M

      My guess as to why Sam’s “life cycle of journalists’ hidden thought processes” leads to a final episode of damaging Trump stories: as much as they need the horse race to make the election “interesting,” plenty of horse race journalists are (justifiably) terrified of a Trump presidency.

  • Debbie lefkowitz

    Thanks for this rational and accurate assessment. It’s amazing to me that otherwise rational people can swing from “the house is ours” to “trump is going to win” in 15 minutes. And don’t get me started on the click bait nonsense that is 538 this weekend. Really sloppy work on their part.

    • Camhilfan

      538 really was a massive joke all weekend. And you basically had Nate tweets being followed up, and essentially undermined, by Harry’s tweets…

      “Hey, here’s how Trump can win (insert state he can’t win)” — Nate

      “He’s got no shot there…” – Harry.

      It was crazy. I was out knocking on doors, so all I saw was twitter activity…that was enough for me.

    • jculb69

      Agree. “Here is what the Electoral map would look like, if only men voted.” What a ridiculous waste of time.

    • Jonathan H.

      Well, he’s got a business to run now, so gotta get those clicks. I’m thankful that there’s at least one person like Prof Wang who does this stuff because he likes it, and as a sort of public service. Thanks again.

    • Sam Wang

      I actually don’t think it’s that. The two questions are (a) why the model is so uncertain over there, and (b) why he says what he says. Possible answers:

      (a) The model got established early in the season. After being burned in the primaries, they put in extra uncertainty. The most justifiable source is undecideds+Stein+Johnson voters.

      (b) He has to talk about his model as if it captures ground truth. See elsewhere on this thread – it was pointed out that Enten apparently does not feel the same obligation.

    • Shawn Huckaby

      Harry Enten from 538 was on an episode of Jacob Weisberg’s excellent ‘Trumpcast’ podcast recently. It was a very balanced discussion not only of where he thought the race stood, but also of the differences in the uncertainty of their model vs. the unnamed others. The primary driver seems to be their conscious choice to model historic uncertainty prior to the last four cycles, even though he admitted everything is much more stable and polarized now. In fact Enten took great pains to say he expects the same outcome (including an electoral total north of 340).

      It was a strange mix of defending while simultaneously undercutting their choices. Worth a listen…

    • Sam Wang

      I am so pleased to hear that assumption laid out clearly. Obviously, I made a different choice, and I think the dynamics this year are consistent with what I chose. But if that’s what they did, it is defensible.

      He is a close watcher of this year’s data. Of course what he sees (and what we all see) is something that is more like post-1996 polarization. So he’s acknowledging that in his discussion. As I said, once a model is launched, you are mostly stuck with it.

    • Richard

      I know there’s a tendency to belittle Nate Silver on this site, but I think it’s misplaced. Most folks commenting on this site seem to prefer Sam’s approach to modeling, but it comes with various assumptions, which however plausible are always subject to empirical verification. I like to think about modelers analogously to how the modelers think about polls. Aggregate, aggregate. We have models from Sam, Nate Silver, Drew Linser at Daily Kos, NY Times Upshot and Huffpollster (though the latter simply gives the current state of the race rather than a prediction as to outcome). I think it’s instructive to look at the range of predictions, think about the various assumptions and view each model as reasonable input for understanding what might be happening. Let’s not cherry pick models in advance but be open to see which ones work best. One of the reasons Sam’s site is so great is that he is extremely transparent in his model assumptions. I think Linser was in 2012 as well (I haven’t checked if there is any changes in his approach this year). Nate is less transparent but certainly has described many of his assumptions.

    • Richard

      Oops, on my previous comment I incorrectly said that Huffpollster only gives the current state. They have a link for their forecast. They are at 98%, in the middle of Sam’s random drift and Bayesian models.

  • Matt McIrvin

    Two points of Meta-Margin loss from here, combined with a significant systematic polling error or lag in the numbers, could be enough to elect Trump. We’re talking about extremes compounded on extremes, though.

    • Jonathan H.

      And if your aunt had wheels, she’d be a school bus.

      Guys. Volunteer to GOTV. If you don’t have time to do that, make sure that you at least vote. Try to persuade friends of yours who seem amenable to your position who may be on the fence. And avoid certain other blogs that contain clickbaity histrionics (hint: rhymes with Schmive shmirty shmate).

  • Bob McConnaughey

    my greatest hope…any where my dollars have gone are to Roy Cooper for NC Gov and Deborah Ross to replace the deplorable Burr in the senate. BTW our Clinton/Kaine sign was stolen while the Ross/Cooper signs were left up. 18 miles south of Chapel Hill, NC.

    • Bob McConnaughey

      btw.. a NC analytical site has done a excellent job evaluating the impact the NC St Leg’s attempt to restrict voting has had on early voting *in our state* so far:

    • Jeff

      In Pittsboro? Someone is probably using your sign as part of a quirky art project, or to make artisanal whiskey somehow.

    • Tom_b

      I’m with you Bob; my first ever campaign donations were to Cooper and Ross this year, and I’m canvassing for the HRC headquarters.

    • cleek

      that’s why i’m not putting out any signs. that, and nobody on our dead end road will see them.

      (fellow P-boro resident)

  • Marcia

    “Donald Trump represents the culmination of this trend.”

    In my fervent dreams, “culmination” would mean that in the future we would begin to move away from this, that he in fact would be the “rock bottom” that we could recover from. Judging from the pledges to investigate Clinton for four years, I can see it will remain a dream.

    • Ed Wittens Cat

      Trump is the possible culmination of the Tea Party/Sarah Palin/Freedom Caucus movement within the GOP.
      If we think of the GOP as a quasi-stabile BTW sandpile, Trump is the biggest “avalanche” to date.
      It maybe the sandpile collapses at the point of the 2016 election.
      The soft sciences are complex.

    • Ed Wittens Cat

      And i say *possible* because because there could be a further instantiation of the Trumpist model– only openly advocating fascism and white nationalism instead of just dogwhistling it.
      It depends on what happens after next week when the GOP establishment tries to regain control.

    • GC

      Trump is the antithesis of the Tea Party, assuming you were a “true believer,” and not some vaguely angry populist who jumped on board without understanding its original goals, e.g., restoring federalism, returning to a currency backed by gold, adhering to the Constitution, etc. Though you can probably find Tea Party-leaning groups that embrace Trumpism, considering its a pretty ragtag movement.

      I’m not trying to stump for the Tea Party, but there’s a lot of conflation in your statement. Trump is a big government egomaniac who doesn’t seem to really care about the Constitution or what it embodies.

      If anything, this circus shows the segment of the Republican Party that strongly believes in having a decentralized constitutional republic isn’t the largest plurality (Utah is perhaps an exception); the GOP is clearly more identity-based than ideological now.

    • Ed Wittens Cat

      The “purpose” of the Tea Party was to invalidate the election of a black man to the high office.

    • Sam Wang

      Actually, I think that GC and Ed Wittens Cat are saying the same thing.

    • Ed Wittens Cat

      i may be gulagged for this, but GC is voicing a large part of the problem. Federalism was just a head fake to cover the racist/white nationalist substrate of Palinism, the HBD movement, the Alt-Right, the Tea Party, and now Trumpism–it is the same strain running through all the above.
      It is a hyuge problem in our national discourse that GC and ilk cant acknowledge this.
      I refer the commentariat to the excellent Podcast #13 on the Alt-Right with Rick Perlstein.

      Even in Siberia there is happiness.
      –Anton Chekov

    • Ed Wittens Cat

      and we are NOT saying the same thing at ALL!
      im talking about podcast 13 with Rick Perlstein and the underlying common animus of the GOP base– the same animus that motivates the Tea Party, the Alt Right, Palinism, Trumpism, the hbd guys, and David Duke and Alex Jones.

  • Greg Gross

    Ouch. Wikileaks publishes emails from Donna Brazile to HRC campaign tipping her off about debate questions….

    • Veronica

      Actually, I’ve stopped reading articles that mention the Wikileaks. Its leader has been known to alter the e-mails to make them more scandalous and salacious than they really are.

  • Mogen David

    Do any of the models incorporate early voting or do they all assume that everyone votes on election day?

    My gut says it matters but my head says the impact is unimportant given the stability we see in recent elections.

    I could envision a model that parses out vote percentages as early voting numbers come in but it’s not a random sample so things would get messy quickly…kinda fun to play with though.

  • wayne purdom

    If you’ll eat a bug if HRC wins Texas, will you eat a vegetable you dislike if the margin is under 3%.

    I’m agnostic on whether this site or 538 or one of the others is right about the magnitude of uncertainty, but one thing that seems clear is that the 538 model is defective this year in being too bumpy.

    From everything we know about elections, especially the Obama campaign polling in 2012, few of the polling shifts correspond to genuine shifts in voter sentiment as opposed to responsiveness and fewer to probable voting at the end.

    • ravilyn sanders

      @wayne purdom // Oct 31, 2016 at 1:41 pm:

      Defective? It is remarkably *effective*. It does what it is designed to do: Generate lots of traffic. Some would even say it has degenerated to a click bait, “you would not *believe* what the swing statistics mean!” “250,475,485 people clicked on this link, and don’t you want to know why?”

  • A

    Nate Silver on Twitter today: “Esoteric hot take: If polling swings are exaggerated by non-response bias, that suggests a significant risk of systemic polling error.”

    • Jeremiah

      Nate Silver’s “predictions” are intellectually dishonest. His “Now-cast” has a lower probability than his “Polls-plus” forecast which doesn’t make any sense. If we take the probabilities of just North Carolina (60.3% for Clinton), Florida (54.2%) and Nevada (59.5%) then the probability of Clinton losing all 3 of these is 7.4% which is the probability of Trump winning them. I know there are more permutations than this but surely Trump needs at least these states and more to win therefore the “Now-cast” probability should be lower for Trump (and higher for Clinton) he is currently at 25.8%!!

    • Seth Gordon

      Jeremiah: Your calculation assumes that Clinton’s winning NC, FL, and NV are statistically independent events. Silver’s model assumes that there might be some trend or systematic polling error that moves all three states in the same direction. This is why, throughout the campaign, 538 has been more pessimistic than the other poll aggregators.

    • Sam Wang

      Actually, that’s not right. Many of you say this, but it is wrong.

      The PEC calculation also has a built-in assumption of uniform swing. It’s actually at the heart of the Meta-Margin concept. The key is how large that swing is.

    • Robert

      They’re becoming a significant outlier amongst all the poll aggregators. I’ve mostly stopped looking as I doubt their insight plus I can’t stand the click bait articles.

    • Brian S.

      @Jeremiah: Your analysis assumes that those 3 states are independent coin flips. As far as I can tell, the reason the 538 model is so “bullish” on Trump compared to other aggregators is that it assumes state-to-state correlations, with the correlations driven by demographics, and has a wider error band for polling than a lot of other sites. Betting markets such as Betfair tend to fall more in line with 538’s projections. In part this could be because the betting markets are aping info from 538, so they’re not fully independent.

    • Sam Wang

      Yes, it’s the error band. I am pretty sure this is the reason. The state-to-state correlations…if anything, they may be underestimating those correlations.

  • Mike Harrison

    While Clinton remains strong here, and less so at that other site, the Senate forecast is the reverse. A 50-50 Senate here with Kaine to break ties, vs a 52-48 Dem Senate there. I would think there would be more correlation between Pres and Senate votes. Anyone have thoughts on why not?

    • GC

      No, I don’t automatically expect a super strong correlation this election.

      I suspect there’s a contingent of “moderate Republicans” who detest Trump and decide to split their ticket: give the presidency to Clinton, but ensure the GOP controls Congress as a check on her presidency.

      The election results could give us an idea if this happens, e.g., Clinton receives unexpectedly strong support in a state that elects a GOP senator. Then compare these results to correlations from previous presidential election years.

  • TeddyVienna

    Where are you all seeing betting markets that were at 90 percent for HRC, dipped slightly to 87, then went back to 89? I’m seeing breathless stories that Trump’s odds against have “halved,” which I’m sure is incorrect but can’t find the data to refute.

    Meanwhile, there are tons of negative Trump stories out today — everything from destroying emails in a fashion that was most definitely illegal to refusing to pay his pollster $750,000 — but I don’t see them gaining traction. Why? Are they not credible? Are people grasping at straws to slam Trump with something new, or are people just tuning out anything that isn’t about the HRC emails?

    • Matt McIrvin

      I think the “odds against halved” stories are about overseas general betting sites, not political prediction markets.

    • Michael Hahn

      @Teddy: Where are seeing these negative Trump stories? I had not seen any today.

    • RDT

      The media thinks Trump is nuts, so is unsurprised by the stories, but is fascinated with the idea that the race is close and he could get elected despite all the bad stuff we know.

      I think now the Clinton e-mail coverage isn’t really about the e-mails, but about whether the dynamic of the race is so weird that they matter.

    • TeddyVienna

      @Michael: Washington Post, Slate, Newsweek. All the places die-hard GOP people will ignore. But the race isn’t about them at this point.

    • Matt McIrvin

      It’s infuriating when the media go up their own butthole with these meta-meta-stories, because people react more to the headlines than to the story content, and anything with “Clinton” and “emails” in it makes them think there’s new crookedness being reported on, when it’s actually just somebody speculating about whether Troubling Questions will raise Troubling Questions.

      On the bright side, the stories about whether or not the FBI is out to get Clinton might keep some wingnut agent from fabricating an actual scandal to drop late this week.

  • Michael

    To quote Sully in “Sully,” “Everything is unprecedented until it happens for the first time.” Are we to conclude that what pundits have been calling “firsts,” really aren’t that, or that first or not, they have not caused any effect that is a “first?”

  • anonymous

    Here are some predictions for the election and thereafter:
    1. Trump will lose.
    2. Democrats will regain Senate.
    3. Republicans will maintain the House.
    4. Jim Comey will resign.
    5. The Senate filibuster will end.
    6. Sam Wang will keep hinting at stopping this site, and yet keep on blogging.

    • Robert

      Re: 5, I assume the lame duck GOP may move to put Garland in before she takes over.

  • Deb

    Nailed that on “Trump has the next turn.” NYT story this evening about Trump’s alleged tax dodge in the 90’s – debt-for-partnership interest swap, to avoid COI (cancellation of indebtedness) income. IMO, probably the biggest reason he’s been under audit for so long.

    • Harry

      About 2 months ago Sam wrote an article about how really mundane the race has become. The polls of 2016 aren’t that much different than 2012. The only reason he wrote that it feels weird is because Donald Trump is bizarre. But the polls themselves, especially in the blue states have been solid for months now. In 2012 on 11/1 NBC/WSJ had Obama up by only 3. His average polling was about 3. Hillary is between 4(fivethirtyeight) and 6(huffpollster). With 7 days to go little will change.

  • Koos van Blerk

    Although Sam obviously attempts to keep the site somewhat unbiased, the blog seems overwhelmingly D leaning. Does that prove that D supporters are smarter and more analytical (and also less subjective) than the opposition? Thanks Sam, this is my daily goto corner for the election updates.

    • Sam Wang

      Part of it is the fact that Clinton is ahead. During the primaries there were more Republican-leaning commenters.

    • Ed Wittens Cat

      I cant name a single public mathematician or physicist that supports Trump/GOP at this point. If u look at the “dangerously bright” of the AltRight, they are either polysci or geneticists. Historically there are SM and BSM physicists and “stringy types” that comment here– and neuroscience types. Dr Wang’s simple mathematical approach is personally very appealing– I prefer him to even Dr. Gelman “the King of the Abaci”.

    • Koos van Blerk

      EWC: It explains R attitudes towards climate science, the Religious Right (one of Sam’s earlier postings), etc. It’s a sad commentary, really, that science can be ignored, but I think it gives Dems an advantage, since they can approach the election more scientifically, and hence effectively, by only campaigning and funding where it matters. This has been Sam’s purpose and emphasis. It also helps to actually being able to find the credible polling aggregators, and not just the one that favors your opinion. But as DT has stated: “I love the poorly educated”. Implying that it’s easy to lead them up the garden path.

    • Central Scrutinizer

      When one party is using internet polling to justify declaring “Our candidate won the debates!”, you won’t find a lot of Stat geeks fawning over him.

      This is especially true because one could easily see a Trump Administration (God forbid) using the exact same method of baked poll results to justify policy decisions. Hey, if it worked in the election….

  • Danny

    Sam–I’m curious–do you still view Clinton’s chances of winning as being in the 97-99% range?

    • Andrew

      Sam has correctly forecasted 2004, 2008, and 2012. There’s really no reason to doubt why it would be wrong now. He also predicted back in December/January that Trump would likely be the nominee.

  • Rudy

    This Wikileak thing is a daily occurrence with not just HRC and WJC involved but Huma, Cheryl, Donna, John P, Tanden… Doesn’t the press have a right to challenge the ethics of purloined email that could be tampered with to display a version of truth. Or have we reached a point where the new normal is drip-by-bloody-drip of something salacious?

  • Rick Howard

    Clinton’s campaign plans to begin airing its first television ads of the general election in Michigan and New Mexico and will return to the airwaves in Virginia and Colorado after a months-long hiatus, a campaign official said.

    “This is a critical time because early voting is the preferred method of voting for Democrats,” said Michael McDonald, an expert on the early vote who runs the U.S. Elections Project. “They are certainly mounting up a lead in the early vote. It’s just, is that lead going to be good enough?”

  • Mike Hais

    Sam, I was going to ask you to explain what accounted for the difference in volatility between your projections and Nate Silver’s on 538. Yours have been extremely stable throughout the course of the general election campaign and Nate’s move with much greater (may I say “extreme” volatility?)

    After reading through many of the comments and your responses on this site, I think I have a better sense of the methodological reasons for the difference. However, I would appreciate any additional details you are able and willing to offer about that difference.

    I do agree with your polarization hypothesis about the stability of partisan attitudes and behavior within the electorate. There is no empirical evidence suggesting that the Comey matter is going to cause many, if any, of those who intend to vote for Clinton to switch to Trump. However, isn’t possible that the matter could depress the turnout of potential Clinton supporters, especially Millennials and African-Americans who appear to be less firmly committed to Clinton than they were to Obama in 2008 and 2012? Comments?

    • David L

      Dr. Wang, I have been a follower of your analysis since 2004. I admire your thoroughness, and your use of mathematical logic, as opposed to a ” media-driven” volatile aggregation approach (a la 538). How do other methodologies justify the volatility, that has Hillary Clinton’s chances of election drop from 88% to 71% in one week? Does that approach to polling analysis have any basis in fact, or is it purely “event-driven” hysteria?

Leave a Comment