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


August 2nd, 2016, 1:00pm by Sam Wang

Three months before the general election, why would a candidate claim that the election will be rigged? Either to foment unrest afterward, or to claim that he was robbed. In either case, the remark suggests that he expects to lose.

The high accuracy of poll aggregation acts as a safeguard against blatant fraud. In the last few weeks of the Presidential campaigns of 2004, 2008, and 2012, poll aggregates (including those served up here) gave the correct winner in all states where the poll median was more than 1% for either candidate. Polls for individual Senate races and national Congressional popular vote also do very well in Presidential years – though less so in off-years like 2010 and 2014.

At the Princeton Election Consortium, we use state polls only, which means that it takes some time to catch up with changes in opinion. The aggregate above includes only three post-convention polls: one each in Pennsylvania, Nevada, and Georgia. At the noon update, Georgia should go to a median of Trump +1%. In 2012, Mitt Romney beat President Obama by 8 percentage points. Previously I pointed out that Trump runs about 9 percentage points worse than Romney in Republican-leaning states. The persistence of that gap suggests that the Republican convention, and Trump’s continuing focus on the late Captain Khan’s parents, have failed to bring his party’s voters home.

Tags: 2016 Election · President

48 Comments so far ↓

  • A New Jersey Farmer

    Given all of the talk about Trump’s problems with Hispanics, why is Nevada so competitive? Or do you expect that it won’t be in another week’s time?

    • Amitabh Lath

      In such cases I worry about the Likely Voter models used by various pollsters. Often they filter by previous voting record, which would undercount first time voters (be they working class whites for Trump or hispanics for Clinton).

      Specifically Nevada is easy to get wrong because a large swath of the population works in the 24/7 hospitality industry. If you require people to be in their homes in the evening to answer the pollster’s call, you will get a biased sample.

      Weighting a known biased sample is doable but never as easy as having an unbiased sample to begin with.

  • Son of Some Guy

    Yes, about Nevada. Catherine Cortez Masto is showing to be trailing by 5% in Nevada polls. But isn’t Nevada about 15% Latino, and isn’t the state’s population overwhelmingly reside in urban settings? What gives? Does Masto have some serious baggage? On the other hand, I remember that Harry Reid was polling about -3% down in his last election, but he won the election handily. Is Nevada polling typically skewed?

  • Heavenly Blue

    Trump has been claiming, since he first entered the race, that everything about the election: his opponents – Lazy Jeb, Little Marco, Lyin Ted, and now Crooked Hillary – the Democrats, and even his own party’s system are rigged/liars/cheaters/corrupt. Claiming the general election is rigged is hardly a sign that he believes he’ll lose; it’s his modus operandi.

  • Bill

    I think a black swan in this election may be what Wikileaks releases in their subsequent releases. Do you think your method of computing the error bound is sufficiently robust to include such an event and its impact?

    • Phoenix Woman

      You mean Putin Leaks?

    • Daniel

      Equally one might question whether the model is sufficiently robust to deal with three more months of Trump insulting everyone in the US who isn’t closely related to him.

      More seriously, there is always the possibility that something might happen to turn the race over to Trump, but that’s a significant chunk of what the 20% estimated probability of a Trump victory represents. The possibility of unforeseen future events is already accounted for in the model. If they weren’t, the Clinton win probability would be higher at this point.

  • Nielen

    This is fascinating reading every day, especially if one has a little statistics background that allows a deeper understanding of the results. I am often transfixed by the results, especially in this election where we are dealing with a bizarre and dangerous demagogue that could upset peace and tranquility in our country. Thanks Sam. Your work is more critical than ever to allow me to sleep at night (so far).

  • Will Hutchinson

    Maybe he’s been reading this website and seeing just how unlikely a Trump victory is.

  • Nielen

    I’ve been outmoderated in the past, but just wanted to thank Sam for providing such a purely scientific analysis of the polls. The current state is reassuring while we face the prospect of having a dangerous demagogue as president.

  • Olav Grinde

    Anyone care to compute the probability that Donald J. Trump will make a concession speech if he gets fewer Electoral Votes?

  • 538 Refugee

    I’m not sure what margin would restore my faith in the system given that we will have a high fraction that will vote for someone who seemingly is unqualified by just about any measure to hold the office.

    Last night I asked a Republican if he was ready to vote Libertarian this year. This is a guy with a decent government position that requires some education. He feels Trump is better than Hillary because she is a liar. (Pot meet Kettle?) He brought up Whitewater, Benghazi and the email. I didn’t get a chance to point out after hours and millions of tax payer dollars the first two went nowhere. As for the email he said he would be fired if he gave secrets to the enemy. ??? I just dropped it. He admits he will be holding his nose when he votes for Trump at least.

    An older survey shows people that listen to NPR are the best informed on issues. FOX? At the bottom. I know one person in the discussion listens to NPR. I can guess the other.

    Is this site the NPR of aggretators? ;)

  • anonymous

    Trump is not beholden to norms of human behavior. Being called a loser is probably the worst insult for him, and he is not beyond contesting crystal clear election results where he is indeed a loser. Arguing poll aggregates with him is like trying to explain to a charging bull how a gun pointed at it works.

    We will see how far the pendulum swings to the right for the R party before it slows down and reverses. It already seems to have reached a ninety degree angle with no signs of stopping though.

  • david c mace

    if the R’s can win it may well come down to FL

    how better to lay ground work for FL 2000 re-dux

  • Kevin Moore

    Just a quick layman’s question for Sam or anyone else who can answer it. I understand that the “median of quadrillions of simulations” is why the banner has a different number than the daily electoral map (e.g., today it’s 324 for the banner vs. 308 for the map), but I can’t create any feasible combination of states (using 270towin) that results in one candidate getting 324 EVs. Is that just a by-product of the definition of “median”? Wouldn’t there be a way to jigger the equation so that it calculates the quadrillion median results on a state-by-state level so that the banner number has a chance of being right on the nose? It would be pretty cool if the banner number precisely matched the final result.

    • Ryan

      It is possible for Clinton to win exactly 324 electoral votes.
      Here’s one semi-plausible map:
      here’s another:

      The way the top-line number of electoral votes is reached is: given the odds of each state voting for Clinton, calculate the probability that Clinton wins 0 electoral votes. Then calculate the probability Clinton wins 1 electoral vote. Then 2. And so on until 538.
      Once we have all these probabilities, the median outcome (today, 324) is the number of electoral votes with 50% of the probabilities above it and 50% below. That doesn’t necessarily mean that it is a highly likely outcome in itself. But it’s the best guess in that it is in the middle; it neither underestimates nor overestimates the number of votes Clinton will get.

    • Sam Wang

      I believe you are asking about the mode, i.e. the single most common outcome. It can be seen by clicking on the map on the right sidebar. The editable version is at

    • Bela Lubkin

      324 would be an exact match of today’s map — — with FL and OH won by Clinton; or; or the 2012 result modified by NC:D, IA:R, NH:R, VA:R

      This seems like an unlikely result to me (VA and NC are similar, but VA is bluer + Tim Kaine should have a bit of influence); but it is certainly not impossible.

    • Letteredwolf

      The number of electoral votes in the banner is a possible answer. See the histogram of all possible outcomes. Notice the % of any of them being correct is very low currently with the amount of possible outcomes. I created 2 maps to show they are possible with current poll numbers in the states.
      The first:
      It has all the toss up states, those with less than 60% probability either way going, (Iowa, New Hampshire, and Florida) for Mr. Trump with Georgia, the red state with his lowest lead, going to Secretary Clinton.

      The second:
      It has one toss up state, Florida, going to Mr. Trump with Iowa and New Hampshire going Secretary Clinton. Also Utah a state Mr. Trump is no up by much as you would expect going to Secretary Clinton.

      I expect there are other combinations you could create with the states that aren’t above 95% for either candidate currently so don’t take these two as the only ones or even the most possible ones.

    • Ken Schulz

      Don’t forget that Maine and Nebraska award 2 and 3 electoral votes (respectively) by Congressional district winner, and two additional (the ‘Senatorial’ allocation) by statewide winner; therefore split votes are possible.

  • Adam

    Regarding his tax returns…I fully don’t expect him to release them, because I think something really awful is in there. We already “know” about his paltry charitable contributions. But if he did release them and they were clean, I would be equally as suspect at this point.

    • seeker

      He says he has no investments in Russia. But how many Russian oligarchs have “invested” in his projects. Especially relevant because apparently banks want nothing to do with him.

    • Sean

      I think there is probably a lot of money coming from people connected to Putin.

    • Michael Levinsohn

      George F. Will has said that Trump has calculated that whatever would be revealed in his taxes is worse than the negative effect of the speculation that goes along with intentionally not revealing them. That effect will depend on how it’s exploited by Hillary’s campaign, no?

  • Amitabh Lath

    Ira Glass had an episode of This American Life a while ago called “fiasco”.

    Description: Stories of when things go wrong. Really wrong. When you leave the normal realm of human error, fumble, mishap, and mistake and enter the territory of really huge breakdowns. Fiascos. Things go so awry that normal social order collapses.

    There are four episodes in the show. We are currently experiencing the fifth.

  • seeker

    I wonder if he even wants to win. Assume he entered to add :luster” to his brand. Assume further that he is too lazy to do the work actually necessary to be Pres.

    Might he be laying the groundwork for keeping his name in the news when he does lose?

    • Phoenix Woman

      It does look like it, doesn’t it?

      To proclaim that the system is rigged is to tell one’s supporters “Don’t bother showing up at the polls”. That puts a monkeywrench in the Koch brothers’ plan to get non-Trump voters to still show up to vote for the Republicans further down the ballot.

    • Sean

      I think he ran solely for publicity but he underestimated how gullible a lot of Americans really are.

    • DaveM

      Of course, the credulous audience for the “it was rigged” response narrows the more lopsided the defeat. If the wheels come flying completely off and the electoral vote gets, say, to 400+ for Clinton, “it was rigged” sounds pretty hollow.

  • deb

    So the “rigging” is only raised as a Nov possibility when the polls are unfavorable? This is just the bounce. Based on recent elections it may go back to 3% or so in another month. Then it wont be viewed as “rigged” maybe?

  • Neb

    Remember Trump’s psychology. He’s the Platonic Ideal of an insecure narcissist. If Trump loses, he needs a narrative that he can turn to by which his image remains glorified, which means he had to have been treated unfairly. A rigged election myth is an excellent basis for that claim.

    Keep in mind also that from a game theory perspective, Trump not a repeat player in the political sphere, unlike pretty much every other candidate from either party. So he does not care about his future as a party elder, or even about such vital concepts as the “institutions of democracy.” If he loses, expect him to tweet for an armed uprising based on the myth of rigged polls, as he did in 2012 (though now much more seriously):

  • Ed Wittens Cat

    i guess im just not data-based enuff to make the cut
    but the reason Project Ivy is *important* is non-linearity.
    because of gerrymandering, even when more dems turn out and vote they can wind-up with less seats, like house of representatives in 2012.
    so its not possible to say (for example) if Clinton is 5 or 7 points in national lead the House will flip with any degree of confidence.
    but even those red-leaning umm…tiles? within the district boundaries have some blue voters within the tile– that could be targetted and persuaded with Ivy– like OFA sent us to houses within neighborhoods within tiles within districts that had a high probability of being persuaded to turn out and vote for Obama.
    optimizes the ground game with micro targetting.
    thats all i wanted to say.

  • Michael Hahn

    Georgia made it onto the “Power of One Vote” list!! WOW!! Call to arms here!!!! Need to keep pushing!!

  • Joseph

    What’s new? Many elections back to and including Bush/Gore have been rigged to one degree or another – by the Republicans!

  • LondonYoung

    As you have told us, pollsters do not simply take n random samples and compute the support for candidate x as (preference for x)/ (n people contacted). Before the results are presented pollsters correct for all sorts of biases they have seen or can think of. So, if too many people answer the question “are you registered as a democrat” compared to the actual percentage that they know are registered as democrats, then they down-weight those respondents in their reported results. Similarly, they attempt to correct for past errors in predicting individual states.

    So, pollster results protect the integrity of the actual results from nefarious behavior introduced in 2016, but if Trump wants to claim that state Y has been cheating by m percent over the last few presidential cycles the pollster results are not helpful in refuting that.

    So, when Trump loses (which is what he expects) he is going to say “look at how I won almost all the states with Voter ID laws and lost almost all the ones that don’t have them – the system is rigged”. While that will be false, it will be hard to refute in a soundbite.

    • 538 Refugee

      ” Before the results are presented pollsters correct for all sorts of biases they have seen or can think of. ” I wouldn’t call weighting the sample to match expected turnout “biases”. Remember, random samples don’t guarantee a representative sample. They model turnout based on historic record and current response rate. I think bias is an unfortunate choice of word here?

  • Richard

    “…why would a candidate claim that the election will be rigged? Either to foment unrest afterward, or to claim that he was robbed. In either case, the remark suggests that he expects to lose.” I’m not convinced Trump expects to lose. Trump’s campaign has been all about inciting anger and derision against his opponents, from Little Marco to Lyin’ Ted to Crooked Hillary — Lock Her Up. But he has a huge problem in that Democrats have won 5 of the last 6 popular votes for president, and his strategy has led to offending pretty much every constituency of that consistently successful coalition. Only way to do that and still win is turn out more of those who have been on the losing side (i.e. angry white right-wing racist, misogynist males). Saying the election is going to be rigged simply might be calculated to increase the turnout he needs. Of course, it could be that he expects to lose. But I would caution against wishful thinking.

    • Ron

      I thought your comment was stellar, till the very end. That Trump saying it because he expects to lose isn’t wishful thinking at all, in fact that comment is part of the narrative that Dems should be “”afraid”” of Trump… Every Dem I talked to thought he was the worst candidate (thus the one we wanted to face) because he would destroy them downballot.
      I’m not saying you are wrong to think, or write, what you did. But its time to start realizing what a lot of us have thought for a long time, that its great that he won the Primary, and he is a crappy candidate. So be it.

  • Kevpod

    I have a suspicion that he is setting himself up for withdrawal from the race.
    After all, it’s “rigged,” and yesterday he said he could be having a great life if he wasn’t running.
    This would pave the way for a Ryan-Rubio ticket.

    • Jaymes Winn

      I don’t see him dropping out. It’s all about his ego, quitting would make him the biggest “loser” in political history. Going to the end and then claiming “Crooked Hillary” stole it is the most face-saving way, even if it seems ridiculous to us, his followers will buy it.

  • LeRoy Bob

    I was just flipping channels and paused for a minute on Fox News. They had a graphic up that said 82% of respondents believe the election is rigged. It seems they will really be pushing that narrative.

  • Olav Grinde

    In 2012, many Mitt Romney supporters tried to unskew the polls.

    With Donald J. Trump, 2016 may be the first time in history a Presidential candidate tries to unskew an election!

  • Erik

    I would think that it is increasingly important to monitor the progress of third party candidates. Trump sure seems to be scoring own goal after own goal.

  • Virginia poliwonk

    Dr Wang: Trump is not the first to claim the 2016 election process is “rigged.” The “BernieOrBust” hard-core Sanders supporters have claimed the primary process was rigged by the DNC. Your “crowd wisdom” post indirectly addressed this (as polls enabled your prediction). Assuming Hillary wins, there will now be two groups who will continue claim that her win was rigged in both stages of the process. Have you analyzed the data re the pre-primary polling for Hillary vs Bernie as compared to the outcomes in each state primary? Especially (or at least) in California? I know that Huffpost Pollster’s aggregates consistently showed HC ahead of BS, but it would be nice to definitively rebut these claims (I guess what I’m hoping for is something tied up in a bow when a BoB tells me Clinton only won because of the machinations of the Democratic party, but I suppose they’ll believe what they want anyway).

  • John Gilbert

    What do you think about this topic now that news has come out that someone has hacked the Arizona (and possibly other states’) voting information? I’m wondering if Paul Manafort may have set Trump up to make his comments, perhaps with inside information that Putin might hack a couple of states (Pennsylvania? Arizona?). If a couple of results were clearly out of line (two or three sigma) with predictions of sites such as this, would that build a case that those states had been hacked? Or would rational analysis be drowned out by angry voices?

Leave a Comment