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

Midnight Suspense Theater: The Senate

October 31st, 2016, 9:32pm by Sam Wang

Around here, trick-or-treat just ended. Or so we think. Actually, there’s one more trick: the Senate.

Everyone’s up in arms over this Comey/email thing. As I suggested would be the case, it’s not affecting the Presidential race in any meaningful way. That cake is baked. However, like any good thriller, there’s a fake ending – the election next Tuesday. After that comes actual governing.

The odds overwhelmingly favor a Hillary Clinton victory, as you can see above. However, think about how the opening stages of a Clinton Administration will go. For Cabinet and federal court confirmations Senate consent is necessary, and in our polarized age is easily withheld.

Remember I said that a Clinton Meta-Margin of +3% is the breakeven point for a Democratic Senate? That’s just about where the race is. Either party could easily end up in control – things are on a knife edge.

Earlier this month, there were four toss-up seats (New Hampshire, Nevada, Pennsylvania, and North Carolina). Then Missouri entered the mix. Now there’s a sixth seat: Indiana. Evan Bayh (D)’s lead over Todd Young (R) has narrowed to a median of 2% (n=3 polls). Use the RCP counter to see that starting from 47 near-certain Democratic/Independent seats, these six races can make control go either way. Democratic control is favored, but just barely.

The optimal use of your resources is not to work on the Presidential race. Instead, anyone interested in leveraging their get-out-the-vote efforts should look at The Power Of One Vote on the right sidebar. The scores in the third column show how much a single vote is worth, in terms of moving the probability of Senate control toward Democrats or Republicans. New Hampshire and Nevada are on top because they are small states, and each person’s vote counts for a lot. Right after them are the four other states I listed.

Another target for activism is close House races, which you can look up here. Many districts are doubly important because they have close Senate races. Around here, the juiciest target is Bucks and Montgomery County in Pennsylvania, home of the 8th District. There may be a close district near you!

Tags: 2016 Election · House · Senate

97 Comments so far ↓

  • Percysowner

    I’m going to canvass this weekend, but really, I’m in Ohio and there is practically no chance Portman will lose. I checked the House seats that are in play and there are none in Ohio and only one within 200 miles. Still, I registered voters and phonebanked already, so I’m giving it my all.

    • Susan Sullivan

      I’m also in Ohio. Result seems baked in to me, at least in my deeply blue area. However, I donated $300 to Hillz this month to do my part.

    • Sam Wang

      Imperfect resource allocation! Donating to the Presidential race at this point, while satisfying I bet, is not the optimal use of money.

    • Sam Wang

      Go to Indiana or Pennsylvania.

    • TJHalva

      If you’re not in a swing state, the best resource allocation of all may be to local candidates at the state house and senate level. They’re basically running the GOTV operation at this point and trying to turnout people in a small area. Something like $300 goes a long way in a district of 20k people. That’s roughly 4000 printed mailers. If you’re not a mailer fan, it buys food for everybody knocking.

  • Lorem

    Unrelatedly, I glanced at the Power of One Vote: presidential edition, and the entire population of NJ voting is worth about 0.3 voters in Colorado? I guess it’s because by the time everyone in NJ turns out for Trump, the election will be long over in all other states, but it’s still an amusing comparison.

    And that’s quite a spooky pumpkin there, by the way.


    Trump Companies Destroyed Emails In Defiance of Courts

    October 31, 2016By Taegan Goddard2 Comments
    “Over the course of decades, Donald Trump’s companies have systematically destroyed or hidden thousands of emails, digital records and paper documents demanded in official proceedings, often in defiance of court orders. These tactics—exposed by a Newsweek review of thousands of pages of court filings, judicial orders and affidavits from an array of court cases—have enraged judges, prosecutors, opposing lawyers and the many ordinary citizens entangled in litigation with Trump. In each instance, Trump and entities he controlled also erected numerous hurdles that made lawsuits drag on for years, forcing courtroom opponents to spend huge sums of money in legal fees as they struggled—sometimes in vain—to obtain records.”

    “This behavior is of particular import given Trump’s frequent condemnations of Hillary Clinton, his Democratic opponent, for having deleted more than 30,000 emails from a server she used during her time as secretary of state.”

  • Matt Sharp

    “The optimal use of your resources is not to work on the Presidential race.”

    This depends how you’re defining ‘optimal’.

    Your resources may be more likely to change the outcome of who controls the Senate rather than the Presidency. However, the impact of who wins the Presidency may be so great (compared to who controls the Senate) that there is still a higher expected value in working on the Presidential race.

    For example, if you think one candidate is much more likely to start a nuclear war, then it’s probably still worth supporting their opponent rather than looking at the Senate.

  • 538 Refugee

    Any networking nerds, computer geeks around? Trump had a server that seemed to be connecting to a Russian server regularly. The really odd thing is that when it was exposed as being tracked through DNS queries it disappeared and reappeared under a new name and the communication resumed. There would have had to have been some type of data exchange for the Russian side to know the name of the new server. They could have hidden this connection the second time around by hard coding the IP address. It is a simple matter of putting the name/ip address in your host file. Amateur hour.

    The people tracking this have some high level Internet access, but not the means to intercept traffic like the FBI or CIA might have.

    Could this be the “explosive” information Harry Reid hinted at? FBI says no, at least publicly.

    • Avattoir

      The words you read into what the FBI has said to these NYT reporters are not in the NYT report.

      Among the concepts below “conclusive” – a word the FBI DOES use according to this report – are suspicion, reasonable (a.k.a. known as reasoned) suspicion (all that’s necessary to satisfy the 4th Amendment to gain a warrant), and probability (all that’s strictly necessary to support an indictment).

      I have no doubt whatsoever that federal law enforcement officials are in fact investigating Trump in relation the Trump Foundation, financing from Russian sources of entities that Trump controls or acts through, and connections between the Putin regime and Trump, directly or indirectly, and the Trump presidential election campaign.

      As Director Comey testified to the House Oversight committee this past summer, all it takes for the FBI to consider initiating such an investigation is a request from Congress or a Congressional committee or some official making the request on behalf of either (He didn’t exclude requests from individual elected officials in Congress), or, of course, someone in the executive branch making such a request with sufficient authority.

    • Ed Wittens Cat

      Brief survey of the infosec community– PwnAlltheThings and InfosecTaylorSwift say probably no:
      Chris Sogholan agrees:
      evacide and Morgan Mayhem just link the Slate article w/o comment…
      and the Intercept has been investigating for months without finding anything definitive..

    • Ed Wittens Cat

      what I think would be more productive is linking the Trump campaign (Roger Stone) to Wikileaks disbursement of Russian hacker data targetting Clinton.
      Wikileaks doesnt hack– they publish leaked data.
      Someone is feeding them material.

    • Matt McIrvin

      Reid is willing to bluff on these things–remember when he said he’d heard Mitt Romney paid no income taxes? Wasn’t true, but it got Romney to release his returns. (With Trump, of course, all you get is “that’s because I’m smart”.)

      I think I agree with Josh Marshall’s take on this: we know Russia strongly wants Trump elected, and we know Trump consistently favors their interests in his policy preferences (such as they are) and his public statements. Does it really matter much whether they’re actively colluding?

    • Joel

      @Matt McIrvin

      There’s also the Telegraph story that didn’t make much of an impact — but it establishes that the Trump campaign is more than willing to solicit illegal influence from foreign parties.

    • anonymous

      Harry Reid did not make up the no taxes thing about Romney. He was supposedly told this information by Jon Huntsman. For some reason, Romney and Huntsman hate each other. In the Comey situation too, Reid’s claim is seems to be based on some credible information (that subsequent press reports seem to be supporting).

    • alurin

      I don’t think Putin is that interested in electing Trump per se, though that would be a bonus. His primary aim is to discredit American democracy… and Trump is doing a fantastic job of that!

    • Shawn Huckaby

      It strikes me as more likely that it’s the MoJo story from an FBI confirmed veteran spy (non-US) claiming intel on an official Russian plan to cultivate, compromise, and then blackmail Trump. This combined with the official announcement of a probe into former campaign manager Manafort’s international travel is very interesting…

  • Zed

    A little off topic, but do you have any explanation for why the histogram of possible outcomes is currently a bimodal distribution?

    • Matt McIrvin

      That’s Florida, I think. Whenever a state is uncertain, you get a doubling of peaks corresponding to that state’s electoral vote count. If there are many doublings, it ends up looking more like a smeared-out lump. But Florida is so big that its individual effect is really noticeable.

  • smartone

    A question the past few weeks there seem to be a serious move away from Trump and yet did not move the needle on Senate races at all.

    Can you explain why this happened?

    • Ruth Rothschild

      I think this may be the split ticket we’ve been reading about in the news for a few months. Republican voters may not back Trump, but they don’t want to see a Democratic party majority in either of the houses because they want some checks and balances on Hillary. So, these Republican voters split their votes by voting for a non-Republican, whether Clinton or a 3rd party candidate, in the presidential race, but still vote along party lines in the down-ballot races (i.e., Senate and Congress). Also, Sam had indicated that with a meta-margin of only a little above 3% for Clinton, the Senate race could go either way at this point, thus his Senate Snapshot of 50-50 at the top of this page. And if I recall correctly, in an earlier article he also stated that Clinton’s meta-margin would have to rise to over 5% or 6% for the needle to move squarely into the Democrat camp for the Senate races. Something I recently read that gives me a little hope is that if one or two of the Senate races in battleground states falls to the Democrats, the other battleground states may follow suit. But, how accurate this assessment is won’t be known until next week. Stay tuned….

  • Shawn Huckaby

    It seems Professor Wang was indeed prescient on the news cycle moving back the other direction. Taxes and Russia seem to be the next two shoes to drop.

    Let’s see if it moves the needle in the Senate!

  • George H

    BWilliam H. Frey@rookings:
    Can “hidden” white working class voters deliver a Trump victory?

    “Scenario 3”: Trump wins by 500K, or 0.3%.

    Improbable, but plausible.

    • Shawn Huckaby

      Hmm. A close reading shows that even with a 22% (!) increase in this demo over 2012 levels, as in scenario 3, it’s still not enough to tip the election.

      As stated in their summary: “What this analysis makes plain is that even in the rosiest of scenarios, an extraordinary turnout of white working class voters, by itself, cannot produce a Trump victory. Thus, the ‘hidden voter’ theory appears to be highly overblown.”

    • Kearny

      On the other hand, are the polls missing the probable/possible increased amount of Latino voters?

      My wife is latino, and I get the impression her friends are all more likely to vote (against trump) than ever before.

  • Di Zhu

    Hi Dr. Wang,

    I have a question about early voting turnout of African Americans in FL. It’s down quite a bit compared with 2012. Did the polls consider this when they weight the demographics? Also, do you think this may be a problem for Clinton campaign?


  • Matt

    Sam–as ever I admire your steely resolve here, and it cracks me up that Nate’s polls-plus is now TWENTY-FIVE % more bearish than you (and sure to go lower still). But, that said…is there something asymptotic about your sense of the meta-margin, such that this most recent and quite rapid 1 pt drop is meh, but dropping another two points, say, is beyond imagining? Or is it just the final 1 pt that’s beyond imagining and thus gives us an unbreakable firewall? You are genuinely unconcerned about for instance a 1-2% systematic polling miss?

    • Tony Asdourian


    • Daniel

      The predictions aren’t being made by anyone’s “sense” of anything. They’re products of the models being employed. FWIW (not much), even just common sense, from looking at the polls, suggests that fluctuations from a 5% to 3% Clinton lead are par for the course, but a fluctuation from a 3% to 1% lead would represent a fundamental change in the character of things.

  • Penn Naym

    Sam, have you heard the news about the public transit strike in Philadelphia? Over 4,700 workers are on strike. If that continues through election day, could it have a major impact on turnout?

    • Sam Wang

      In Philadelphia, nearly everyone gets to the polling station by walking.

    • Stephen

      “If we foresee an agreement will not come to pass, SEPTA intends to seek to enjoin the strike for November 8th to ensure that the strike does not prevent any voters from getting to the polls and exercising their right to vote.”

      In Plain Speak : We are not going to strike on Election Day. Buses will be running.

    • DaveM

      “In Plain Speak : We are not going to strike on Election Day. Buses will be running.”

      Since that’s the word from management and not the union, a more accurate translation might be:

      If buses are not running it will be the union’s fault (or maybe some judge’s), because we’re planning to seek an injunction.

    • Deb

      While people walk to polls they depend on public transportation to get to or from work in time to vote so without that there could be an effect. I’m in the philly burbs and many take trains to work here as well so the strike can effect turnout if it lasts that long.

    • Darryl Stout

      But they could end up getting home from work very late and/or very tired.

    • Kool Earl

      And to add to Sams reply. In any election there are rides made available for those who unable to walk to polling stations. I would be confident that contingency plans will be implemented if the strike lasts through election day. I wouldnt worry about the strike depressing turnout

  • Jim Wright

    What’s the deal with the WaPo/ABC tracker? I readily admit the polls have likely tightened some in the past week, but a 13ish point swing in a week seems incredible.

    I know the WaPo/ABC tracker has been wrong a week out in the past as well, but I’m just wondering what in the methodology makes it 1) subject to such wild swings in a week and 2) makes the poll end up being wrong more than right a week out?

    • Sam Wang

      You are expecting too much from a single poll. Probably this is normal statistical noise, plus some actual movement.

      When comparing two margins, i.e. the Clinton-Trump margin on one date with the same margin at another date, the combined uncertainty is larger by a factor of sqrt(2). In this case they report +/-3%, which means that the change has an uncertainty of +/-4.2%. The margin could have up to twice that uncertainty if it is driven by a flip between Clinton and Trump, hard to imagine but it could be.

      Obviously you should be looking at aggregates, which show that the race has tightened by 2-3 points since Oct. 24.

    • Amitabh Lath

      The Upshot at NYT tabulates all the state predictions for various websites and Silver’s has Clinton systematically lower in many battleground states (NH, CO).

      The 272 line runs between CO and NC. Is Colorado really in play? Trump has never been ahead in any public poll there, and if the Clinton camp had even a wiff of doubt about it (private polls for example) she would be hitting it hard.

      But let’s see what the Clinton campaign actually has going on there. Nov 1 and 2 we have Chelsea Clinton doing GOTV. That’s it for the national campaign.

      They are not scared of losing Colorado.
      And with that, 272 is in the bag. NC, OH, FL are icing.

    • Jim Wright

      I keep coming back to that Amitabh.

      I think the term “firewall” is overused, but that’s essentially what NH+VA+PA+CO is. Three very good polls for HRC in the former three today, those are all but out of reach.

      Agreed with Colorado, I don’t see any signs from the campaign that they are going to lose there.

      You’re still president whether you get 272 EVs or 400

    • Ruth Rothschild

      Here’s an article from Huffington that maybe be of interest to you regarding the ABC’s latest poll showing Trump ahead. I think it nicely puts things in perspective and seems to piggyback with Sam’s response to your post earlier this morning:

  • mediaglyphic

    The Evan Bayh race is interesting. I wonder if the shift is due to the raw polls or a big change in the LV model. The clinton campaign is likely not putting much energy in the GOTV in Indiana and i wonder if this is shifting the LV. There are a lot of variables and a large stasis here!

    • trostsky1960

      I live in Indiana and we are bombarded with negative Bayh ads pushing the “he’s a lobbyist and works for them” meme which I would posit as the biggest factor affecting his polling. Trump up 11 points in latest state poll which isn’t helping either.

    • Ed Wittens Cat

      way cool.
      and transparent just like PEC!
      and maybe in 2020 polling will incorporate MIT Social Machine Project tech.
      but it still wont address Dr. Wangs polarization theory, that emotion beats math for most of h. sapiens.
      im feeling a lot of deja vu over the last election– the missing white voters and poll unskewing is back– redbrain poll aggregators and the Trump campaign still tryin’ to drive up the animal spirits of the GOP base.
      bluebrains still incredulous that any human would vote for Trump.
      im still an intransigent educationist– America is the richest country in the world– cant we gift our citizens with enough math to understand polling?

    • Shawn Huckaby

      We can’t even gift our citizens with enough math to not play the lottery.

    • MarkS

      Playing the lottery (or any game with a negative expected value) is logical if your utility curve has the right shape. And many people assign positive utility to daydreaming pleasantly about winning.

    • TeddyVienna

      Hey — some of us play the lottery to support schools! Or for the fun of dreaming.

      If you know you’re throwing money away, I don’t have an issue with it. The sad part is when people who don’t have the disposable income get addicted to scratch-offs.

  • Bob McConnaughey

    Is there substantive data out there on state gubernatorial races? I’ve given ~ 1/3 to Cooper, 2/3 to Ross having, perhaps naively, assumed that our current republican was doa after HB2 fallout…more to Cooper again?

    • Bobby

      When I donated to Roy Cooper today, his Web site had a little blurb claiming that a Fox News poll within the last few days showed that he lead McCrory by 6 points. I gave nearly identical proportions as you did to Ross and Cooper this morning.

      Professor Wang, thank you for this site! I think, on my own, I would have kept giving money to the Clinton campaign alone. Instead, under the influence of your clear analysis, I have spread my donations with a bit more going to the NC races since that is my home state.

  • Lloyd


    This comment has no bearing on the current discussion. It is regarding the colors used to display the EV and Meta-Margin. It is difficult for me to distinguish between the >80 and >95 percent shades of blue and red in the map. The colors are so closely matched that my eyes cannot tell the difference. I find this impacts the readability and accessibility of the information on the chart. Have you given any thought to implementing a more user-friendly color chart to increase accessability?

    I realize this is a not-for-profit website, and I enjoy the sane and intelligent coverage. Thank you for your hard work.

    • Sam Wang

      Hmmm, open to changing the colors on the cartogram. The 270towin map is out of my control.

    • GC

      Sam, if you’re interested this site has battle-tested color schemes for web maps, including some that are colorblind safe. One option is to use the 3-class red and blue sequential single hues for the cartogram.

    • Sam Wang

      I love it! Might use this or this.

    • Lloyd

      Yes. I will often consult your 270towin custom chart to decipher the finer breakdown by color.

      For your own map, I think either of the color schemes you found would be a major upgrade for the accessibility of information and would lead to an enhanced user experience. Thank you for your consideration and effort.

    • Jenny

      While on the topic, the grey-shaded area on the EV and Meta margin charts is a bit bright for a white background. I did not even see it at first. I’d suggest a cyan shade, as that would make a nice contrast with both the background, the red “strike zone” and the yellow 95% confidence interval.

      Not a priority, I know, but it would make it easier to see.

  • Tom Gavin

    Regarding allocation of energy, isn’t it likely that Clinton donations could help down ballot races through a coattail effect, or do you think that is just inefficient?

  • Marco

    The nail-biter next week will be the Senate. Dr. Wang gives 70% probability to the blue team. All of the sudden, Huffington Pollster give that probability at 83% (they were usually more bearish that Dr. Wang ). Wonder why…

    • wayne purdom

      The biggest reason is simple. Until very recently, they didn’t think there were enough polls in Indiana to make a poll-based prediction. Once they did they moved the Indiana race from 50-50 to a much larger Democratic chance, though ironically, at the same time, the Bayh margin was steadily falling.

  • SF

    Sam, about the color schemes. I think (“this” #1)

    is the best one. It has the most color differentiation.

    • Protanope

      From a color-blind perspective, “this” #1 is worse than either the existing scheme or “this” #2. The three colors in the middle are barely distinguishable.

  • Shawn Huckaby

    It strikes me that we are about to live through as close to a perfect empirical test of the polarization hypothesis as one could devise. If any of the most incendiary stories on either side comes to pass it will prove both instructive and frightening to observe how many people are able to place country before party, and how many aren’t.

    If in fact we as a country have moved to a place where we will stick with our team absolutely and categorically without exception, I’m not sure there is much hope for us to continue as a nation.

    I know this sounds bleak and hyperbolic, but I strongly believe we are very close to being off the map here.

    • david

      The good news is that the obvious national divide appears to be age related. Wait a couple of generations and those Trumpers (along with the man himself) will be gone. Even younger evangelicals are lightening up.

    • Ed Wittens Cat

      Not quite Shawn
      The first thing that will happen
      given an HRC win
      is the war for the GOP…usually the losing candidate (see 2012 Romney) gets blamed–but Trump may be able to cleave the party and take his half (the Brietbart deplorables) with him.
      so that will be the real test of Wang Polarization Theory.
      Will the GOP base abandon the establishment (soi disant small government, federalist, gold-standard, free traders) and throw in with Trump to run a more more polished less yucky Freedom Caucus rep? Or maybe Jared or even Ivanka?
      Will Paul Ryan & the establishment cave to Trumpism?
      Ryan just voted for Trump fyi.
      Will Cruz or or Kasich or Rubio try a run?
      interesting times.

    • BrianTH

      We were really polarized in the late-19th/early-20th Century too. Hopefully it won’t take another Great Depression to cause a realignment.

    • Ed Wittens Cat

      demographic shift and cultural shift will cause realignment…
      but very slowly

    • Slartibartfast


      I agree with you that the events following the election will prove to be a test, although I’m not as pessimistic as you regarding the likely results. Also, the only people that will have the chance to show that they can place country before party are those who turn out to be wrong. The right didn’t do so well at this after President Obama was elected and has utterly failed to take their guns for nearly eight years and I wouldn’t expect them to fare much better if Clinton proved to be a good president or a Trump presidency turned out to be a catastrophe, but I fear that many liberals wouldn’t do much better at owning their mistakes if the situation were reversed.

    • JamesE

      I seem to recall hearing something about mid-19th century political polarization. Eleven states leaving the Union, 750,000 people killed?

    • TheGhostofSimonBolivar

      I, for one, agree with you. In 1987 I was sitting on the East German border as part of a NATO military exercise. And I thought, “The Soviet Union will eventually collapse. But I will never live to see it.”

      Two years later the Berlin Wall fell and two years after that, the Soviet Union was dead.

      When you are living in the times, it is hard to see past the forest for the trees. People don’t want to believe things could change as radically as they can for them personally. And as quickly as they can. Its only in retrospect where you can see clearly the events that lead to the fall.

      Much like the Thomas Jefferson warning that he had heard the knell of the nation after the 1820 compromise. We can see clearly now that he was right. Not dealing with the situation in that moment, lead to a much, much worse holocaust later.

      Sometimes there is nothing you can do. Right now, all we can do…… vote. Talk. Hope for the best.

    • Ed Wittens Cat

      im sorry to repeat myself…but the true test of Wang Polarization Theory isnt the presidential election– its if the GOP repolarizes after Trump loses.
      can evangelicals, the Kochs, Ryan moderates, federalists and goldbugs get in bed with the AltRight?
      i dont know the answer– but we will find out i guess

    • Ed Wittens Cat

      the cover of the American Conservative this month has an elephant armwrestling Trump
      and the title– The Fight for the Right
      heres a good longform article

  • Phoenix Woman

    Speaking of the Donald:

    A Russian businessman boasted of securing “hundreds of millions of dollars” of Russian money for Trump.

  • Markyd

    Sam, can you give us a quick primer as to what the meta-margin, Bayensian % and EV projections were at this time in 2012 (one week out)? I looked through the archive but I wasn’t sure I was looking at the right thing. Thanks!

  • Than

    Other GOTV methods include texting voters, millenials, and minorities, through various organizations. If you can’t do a GOTV bus ride, or donate money, this is an effective and rewarding way to make an impact!

  • Tony

    Silver has frequently been criticizing unnamed forecasters who have Trump with a 1% chance of winning. Since you’re the only one, I imagine he’s referring to you. It seems to me he’s built an outsized level of uncertainty into his model. And today I heard him set expectations low on his podcast. He said that unless there is a big swing on the next 7 days to one candidate or the other, $3/ going to get “a lot of states wrong”. Yet the day before he said he’d love to put a couple grand on the line against some of the “more confident” forecasters. Be interesting to see who comes out on top November 9. In putting my money on you FWIW.

    • Debbie Lefkowitz

      I’m so disappointed in Nate and his team this year. I wonder if Harry Enten stays with 538 after the election. He’s the only one who is saying Trump has no path, the numbers have not shifted appreciably. I wonder if Nate is looking back at his time at the NYT fondly now that the ESPN overlords are forcing him to post what if the Evan McMullin is the next President articles every day. Really appreciate the stats based, rational case Sam makes.

    • Matt McIrvin

      Nate is hedging, as he always does; he shows Clinton ahead so if Clinton wins he’s right enough, but by talking up Trump’s chances he’s also cornered the “Trump wins” market; if Trump wins he’s the genius.

      But that makes his model less interesting as a hypothesis, because he’s not sticking his neck out. Sam’s making the sharpest prediction that is remotely reasonable. If he’s wrong he gets to refine the model for next time, assuming Trump doesn’t have him arrested like that one troll predicted.

    • 538 Refugee

      The problem with Nate even disputing the 1% is that if Clinton wins, he loses the argument. We seem to be immune to “October Surprise” at this point so I’m not sure what he thinks can move the needle.

  • Shawn Huckaby

    Interesting piece on YouGov today about “phantom’ poll swings, and what they’re calling “non-response bias”. They agree with Sam that the race has been basically stable the entire time, but they claim the swings in polling following major events can be explained by this bit of psychology: “when things are going badly for a candidate, their supporters tend to stop participating in polls.”

    Their model’s solution is to weight to past voting—specifically to the 2012 presidential race, which then predicts a Clinton victory by a similar margin to 2012. Perhaps too much weighting if it just duplicates the same result?

  • Richard DuBois

    I’m heading to Las Vegas for a weekend of canvassing.

  • CyclicLaw

    Gotta say, you definitely seem to have gotten into Nate’s head! He brings up the “other models” and 99% practically every day on the 538 podcast.

  • Jared Howe

    If you want to help contact registered voters in battlegrounds states, you can volunteer with Next Gen Climate. Go to

    BTW, long time reader. Just thought I’d add that it’s a lot better to work on GOTV than obsessively checking this site.

    • H-

      I agree. I just signed up with GOTV to go door-to-door to make sure people are thinking about it and have a game plan to get to the polls here in northern NJ. Have to get Garrett out and Gottheimer in!!

      It’s my first time doing it, so I’m nervous but it feels good, too!

      Best wishes to all!

    • Susan Sullivan

      Right, but some of us are elderly and have stronger bank accounts than legs.

    • Victoria Baker

      Susan Sullivan– The site Jared linked to is for working from your home, by phone and computer.

  • Debbie Lefkowitz

    Gave to Applegate in California Congressional race. Booting Issa would be most awesome.

  • Susan Sullivan

    Hillary is campaigning for down ballot candidates, so $$ to her still make sense. And no matter how confident Dr.Wang is, I am still nervous and freaked out.

    • Roger

      She will be in Arizona today and that trip probably takes up a big chunk of her time. Seems like a kneel down move in a football game by the winning QB.

      Supposedly all travel/rallies and ad buys have to be approved by the analytical department of the campaign – most efficient deployment.

      So I find it hard to believe, she would waste most of a day, in the last week, going to Arizona unless the campaign was seeing numbers indicating a comfortable win.

  • Runner

    If you rely upon math, statistics and science, then HRC will without question be our next President. If you rely upon “hidden white working class voters”, skewed polling, “huge” numbers of people attending political rallies, and gut feeling emotionalism, then DJT will be our next President. My nickel will always be on math, statistics and science. It’s a no-brainer.

    I am interested however in whether you think overall voter turnout will be higher, lower or essentially the same in this 2016 Presidential election than it was in the 2012 Presidential election?

  • Don F

    The national GOP gave up on Colorado before the primaries by failing to designate — and thus to fund — a candidate for Senate. Personally I think Bennet, the least charismatic member of that body, was vulnerable. Don’t forget he got there in the first place by burying a vigorous young challenger under piles of corporate cash, and thus has no love from Democratic activists.

    Looks like their decision could swing the nation. It could also even replace a stalwart GOP Congressman, Mike Coffman, bringing Democrats up to four out of seven Reps.

  • Jim

    I am not sure how Nate Silver is still taken seriously by the media and quoted repeatedly by others.

    I find his forecasts diluted and constantly hedging – he was right “1” time in the last election but unable to make bold predictions based on a solid model. I mean I can say without any forecasting knowledge it looks like a 70/30 probability for Clinton (+/- 10 pts).

    I can respect this model for it’s definitiveness.

    I would have like to seen how Evan McMullin could be a potential spoiler to either candidate, I don’t think enough attention was given to this 5th candidate.


    • Evan Grantham-Brown

      Agreed. The thing about Nate Silver is that he’ll do things like call ten races at 80/20. And then in all ten races, the 80% guy wins, and everyone is like, “NATE SILVER IS A GOD.”

      Which is *wrong*. If you call ten races at 80/20, then on average, you should expect upsets (the 20% candidate winning) in two of the ten! If you keep calling races at 70/30 or 80/20, and then the underdog always loses, you’re not actually doing a good job prognosticating.

Leave a Comment