Princeton Election Consortium

Innovations in democracy since 2004

Sep 18: Biden 352 EV (D+5.3% from toss-up), Senate 52 D, 48 R (D+3.2%), House control D+2.0%
Moneyball states: President NV AZ PA, Senate MT KS AK, Legislatures KS TX NC

Why did the polls seem so variable this week?

October 27th, 2016, 8:45pm by Sam Wang

Why do the polls seem so variable this week? The basic answer is that there were a lot of them. Outliers are an inevitable consequence.

Trump is going on about three polls that he likes best. They are all favorable to him – IBD/TIPP, L.A.Times/USC/Dornsife, and Rasmussen. They show a tie or a small Trump lead. Meanwhile, for some reason he is ignoring polls that show Hillary Clinton 12 to 14 points ahead. Huh, it’s so weird that he would do that.

Anyway, this amount of variation is totally natural. It’s kind of like watching dogs in the park. If you wait for ten of them, you’re more likely to see a big or small one. But what we want here is the median dog.

The more polls there are, the wider the range of outcomes that you’ll see. With the election so near, we’re getting megadoses of polls: I count around 17 in a seven-day period. With that many, of course you will get outliers. As a rough rule of thumb, when the frequency of polls goes way up, you should expect the overall range of outcomes to double, more or less.

In this case, imagine that among pollsters, the standard deviation of the Clinton-Trump margin among pollsters is 3.5%. If there are 5 polls available, about half the time* one of them will be more than one standard deviation away from the median. Under current conditions, a typical range would be Clinton +2% to Clinton +9%.

With 17 polls available, the range will be more like 2.2 standard deviations in either direction. In that case the likely range is a whopping 16 points – Trump +2% to Clinton 14%. Which is just about what we saw this week.

Don’t be like journalists who run off after the most extreme report – that is ridiculous. The only honest thing to do is to take a median or average of all the polls you can get your hands on. Right now, the collected wisdom of all the poll is that Hillary Clinton is ahead by a median of 6 percentage points nationally. Not 4 points, and not 8 points. Her lead is 6.0 +/- 0.9 % (n=12 polls over the period October 20-26, median +/- estimated one-sigma standard error).

Do I have any criticisms of those pollsters? No. They are experts in their field (except maybe Rasmussen – they need a Truth and Reconciliation Commission to repair their methods). There’s nothing wrong with being an outlier – every year it has to be somebody. The consolation prize is that lots of people pay attention to an outlier.

*For total nerds: I’m using a t-distribution with parameter 3.


I’ll be honest – sometimes I get a little impatient with readers who show up here mainly looking for reassurance in the Presidential race. I understand this is an attraction of the site. But occasionally when I run into people who say they come to my site and feel relieved, I detect a hint of implied complacency. That would be antithetical to why I do this!

I view PEC as a tool for directing your positive energies to make a difference. Of course we are interested in the Presidential race…but there is little that can be done to affect it. It is basically decided. A good way to work off any anxiety is to work on races where we genuinely don’t know what will happen. Senate control is totally up in the air. There is Nevada, North Carolina, Pennsylvania…so much to do. Donate your time and money to the side that you favor. Or, use the Competitive House Race Finder in the left sidebar.

Tags: 2016 Election · President

146 Comments so far ↓

  • Matt McIrvin

    At HuffPo it’s about 7 to 8 points on the two-way question, about 6 on the three- or four-way. Which of those is more relevant… who knows?

    • RonL

      The difference in the two-way and four-way polls, I suspect, will be less consequential than the impact of the Democratic GOTV effort. Estimates are that the Clinton camp’s attentive approach to voter outreach (vs. the Republican inattentiveness) will = 2 percent additional margin for the Democratic slate. A 2 percent potential not included in any polls of likely voters or registered voters.

    • Sam Wang

      Lots of people around here are talking up a bonus 2% for Democrats because of differential get-out-the-vote activity. Could be, but it would be an error to assume this is the case.

      Early-voting could be some indication…but even that is not a guarantee, since it’s just banking votes in advance. For example, who cares if Republican GOTV is inadequate if those voters show up on Election Day anyway?

    • Matt McIrvin

      Yeah, I’ve been wondering if Democratic early voting is up because Democrats are fired up to vote… or if the Democrats who were going to vote anyway are just more inclined to vote early, because they’re afraid of some violence or chicanery mucking up Election Day.

    • Matt McIrvin

      (Personally, I voted early for no real reason but pure curiosity, because it was my first ever chance to do it here, and I was wondering how it worked.)

    • Matt McIrvin

      Usually there’s an assumption that both sides are going to at least make a reasonable effort. This year, the Republicans aren’t doing a lot–but, on the other hand, they may need it less, because they have the demographic of elderly white people, who are incredibly reliable voters.

    • Kevin King

      It seems like the 2-way median is more like Clinton +6.5 right now, and the polls including Johnson are at 6. I’ve been keeping track of the national polls for about a little more than a month, taking the medians of polls that sampled within a week of the ending date of the most recent polls. I have found the following: (1) starting right before the first debate the polls including Johnson have almost always stayed around Clinton +6%. Occasionally, the median briefly drifts away from that margin, but otherwise, Clinton’s lead has been rather consistent. (2) The 2-way polls have almost always stayed between .5% & 1.5% more in favor of Clinton than the polls including Johnson, and they have been somewhat more variable. And (3) indeed the national polls with Johnson generally lead the meta margin by a week or two, generally showing Clinton 1% to 2% higher in the national polls than the meta margin. It’s uncanny how it follows! For instance, when the meta margin was at around 5% or a little less, the national polls were rather steady at 6% in favor of Clinton. Maybe a week ago, the national polls rather suddenly kind of took a little bit of a nosedive until they matched the meta margin, for around a day or so. Then, the meta margin suddenly dropped to around 4%. Now, the last couple days, the national polls (including Johnson) came back to Clinton +6%, but weakly, meaning the median was the last poll with 6%, with several at 5% right underneath it. Perhaps the meta margin will drift back up to 5% again,. or else the national polls will come back down.

      FWIW, I personally don’t come here for any kind of reassurance, but because I want to know what’s truly happening and because it’s fun to follow. It’s also interesting to see the difference between the real state of the race and the media’s portrayal of it. The Washington Post keeps on writing, for example, that Trump’s campaign is spiraling downward and dragging other races behind it. Yet it’s nothing of the sort; yes, he’s behind, but the bottom’s not falling out, as we can see: the race for Congress is slightly favoring Democrats in the Senate, as it has been the entire campaign, and the Democrats have struggled to keep a margin over 5% for any length of time.

      Finally, I am kind of keeping my eyes open for the “natural experiment” of whether the “GOTV” operation truly makes a difference. If so, perhaps we will see Trump consistently underperform his polling numbers in the states where the Clinton campaign is making an effort. I am agnostic on the issue until we see what happens.

      I will also note that the Trump campaign is using marketing techniques and analytics of a sort: they are using targeted marketing via social media, both to rally his supprters and to attempt to suppress turnout from likely Clinton voters. Their methods just aren’t the sort Clinton is using. We shall see what happens!

    • Shawn Huckaby

      I have to cop to being one of the people thinking Clinton would outperform her predicted margin, through a combination of factors–including her superior field organization.

      After Sam’s skepticism I tried to find any sort of empirical research into the quantified impact of a good ground game. Wikipedia quotes a study by Donald P. Green and Alan S. Gerber from 2004, titled ‘Get Out the Vote: How to Increase Voter Turnout’, which attributes about a 1/2% gain to a well organized GOTV effort.

      ‘The Monkey Cage website (
      has an interesting analysis of various facets of the 2012 Obama campaign, and they conclude, “Other things equal, Obama’s vote share was about three-tenths of a point higher in counties where Obama had one field office and six-tenths of a point higher in counties where Obama had two or more field offices.”

      Additionally, there’s this: “We simulated a counterfactual in which Obama had no field offices but nothing else changed. In that scenario, we estimate that he would have lost about 248,000 votes nationwide. Given where those votes were located, we estimate that Obama would have lost Florida by a very narrow margin in this scenario. This estimate parallels estimates from the 2008 election. Joshua Darr, Luke Keele, and Matthew Levendusky have estimated that Obama’s 2008 field operation won him about 275,000 votes in total and could have been responsible for his victory in North Carolina. A similar analysis by political scientist Seth Masket found that Obama’s 2008 field operation may have been consequential in North Carolina, Florida, and Indiana.”

      I can’t vouch for any of their methodologies, but it is clear that campaigns expect some value from these efforts or they would spend their money elsewhere.


    • Shawn Huckaby

      And FWIW, I have been following Sam’s advice since the conventions. Clinton’s many texts have gone unanswered while I try to push that Act Blue thermometer through the roof. ;)

    • Andrew

      I am hypothesizing that GOTV efforts are mostly or entirely baked into the polls already. If polling reflects a convolution of voter preference and voter enthusiasm (e.g., Gelman et al., The Myth. Swing Voter), shouldn’t the polls already reflect the opinions of people who have received a GOTV message because they more likely to answer a phone call from a pollster?

    • Kevin King

      My apologies for all the posting. This is the last one, I swear: Andrew, great point. Perhaps it depends on the likely voter screen. From what I gather, some simply ask whether a citizen intends to vote. Others asked more detailed questions such as whether one has voted before. Polls that use the the fomer “soft” voter screen might bake get out the vote efforts into the polls, whereas other polls that ask whether the citizen has voted before, etc., might not bake as much of these efforts into the polls. I wonder if some pollsters ask whether the citizen has been contacted by a camaign and how many times, as part of the likely voter screen. This could make for some interesting research if pollsters are willing to share their data.

    • Dell Martim

      OK, if the GOTV advantage is 2%, or if it is virtually nil, as Sam seems to believe, then that will rewrite all the Poli Sci textbooks…what is a party supposed to spend hundreds of millions of dollars on if not advertising or GOTV? And if GOTV makes virtually no difference, why bother to contribute to or work for a national candidate?

      I know that the answer is: Work locally. Fair enough..but such a small effect on a national race would certainly invalidate everything that we seem to believe about money and politics. Plus, let’s not forget that donations to the national party or candidate often get spread to down-ballot races.

    • Andrew

      Thinking about this last night a bit, I realized that one thing that could create a deviation from the polls is if the lop-sided GOTV changes the likely voter models. According to the WaPo, the best likely voter models are often based off of determining who voted in previous elections. If GOTV on Clinton’s part convinces some voters who don’t normally vote to vote, and that effort isn’t reciprocated by the GOP, well that would skew things. Early on there was some noise about the increased number of voter registrations, especially among latinos, but I’m not sure how significant those were. On the other hand, early on there was also much discussion about Trump getting people to the polls who don’t feel they fit into one box or the other, so that could potentially shift things for him as well.

    • Andrew

      P.S. Whoops, Kevin, completely missed your post! Yeah, you are right, it depends on the likely voter screen and how accurate it is.

    • Michael Ralston

      Don’t forget there’s GOTV efforts that *can’t* reflect in polls, because the ‘visible’ part of them happens on the day of – calling people on the day to remind them, bussing people to the polls, and so on.

      That seems pretty valuable, intuitively, but of course it’s hard to say since in past elections both sides have been pretty even and we don’t see any major effects.

      (but race is close, emails etc.)

  • A

    Great post. Thanks for the reminder (which sadly, we need you to write about once every two to three days).

  • Dennis Enderson

    I read today that “Rasmussen” isn’t even Scott Rasmussen’s anymore. He left his company after his epic 2012 election miscall to create some type of news oriented website that has subsequently folded. Rasmussen is also the same outfit that calls itself “Pulse Opinion Research” and floods the media with slanted polls that tilt averages towards Republicans all over the map. I doubt that even a “Truth and Reconciliation Commission” could redeem that so- called “polling” firm. They must have thick calluses on their thumbs.

  • A New Jersey Farmer

    There are a couple of House races here in NJ that could use some attention too.

  • Ebenezer Scrooge

    I’m not sure I agree with Sam on the insignificance of early voting. Weather on election day makes a significant difference AFAIK–bad weather typically favoring Republicans. Early voting reduces the significance of possible bad weather, or bad colds, or car breakdowns, or whatever. An early voter has a probability of voting of 1, even if their probability of voting on election day is 0.8.

    • Sam Wang

      Sure…but I am addressing some manic talk around here about hidden bonuses for Democrats. I’m not buying that yet. It has more support than the shy-Trump-voter idea, but that isn’t saying much.

    • GC

      “Manic talk” as in the hand-wavy stuff about Clinton’s GOTV effort (which is “better” than Trump’s) yielding an extra 1-3% on Election Day? I use “better” because while I’ve seen this stated several times, I’ve yet to see it quantified or modeled. Doesn’t mean it’s false, though.

      I mean wouldn’t the impact of Clinton’s ground game already be captured in the polls?

      WRT early voting, my suspicion is that it becomes more of a game changer as 50%+ of the electorate does it. So you’d see a situation where the outcome is baked in weeks before Election Day, simply because there aren’t enough “day of” voters left to sway things. This is all speculation, of course.

      But in my lifetime, I wouldn’t be surprised if early voting reaches at least 60% during presidential election years.

  • Pieter Van Tol

    Sam —

    How do pollsters deal with early voters? If they get a hold of someone and the person says “I already voted,” shouldn’t they ask “for whom”? (Or “for who”?) If the pollsters ignore the early voters, it seems like it will mess up their model, especially if early voting is 30-35% of the electorate. Sorry if this is a dumb question; I am good at that!

  • Charles M

    Thanks so much for this site and all the work you do.

    While I agree that journalists should focus on poll aggregation, it is worth remembering that polls are expenisve. So if a newspaper or network has paid for a poll, it makes some sense that their news story will highlight that poll’s results.

  • Robert M

    It’s not just journalists; the campaigns like to highlight the worst polls to scare you into donating. And thanks Sam for the kick in the pants about focusing downballot – I’ve been guilty of coming here for reassurance, but now I’m getting involved with the Catherine Cortez Masto campaign in NV

  • Michael

    What is to be made of the “voter suppression” strategies of Trump’s data team? Bloomberg reports they admit they know Trump’s substantially behind, so they’re trying to go at three of Clinton’s main demographics to suppress their turnout, making Trump’s projected 45% enough to eke out a win. Any history of such a thing working?

    • Ed Wittens Cat

      A Trump win isnt really the purpose of Project Alamo.
      Its designed for November 9th and beyond.
      This is the purpose– to cleave the GOP 40% right down the middle– to take the hardcore 20% into Breitbart’s alternative universe.
      In the Frankenstein movies I always rooted for the Monster.
      This is poetic justice if ever i saw it.

  • RonL

    1% is not interesting to me. Far more interesting is the down-ballot races, as Sam suggests.

    And I’m resident in one of the tight races in the House of Reps — NY 22nd — a race in which national monies are quite evident. Republican candidate Tenney is openly advertising herself as a bulwark against Clinton. So it’s simple for me to put my efforts where they have max. value.

  • Amitabh Lath

    I suspect GOTV does make a difference because many do not pay much attention to elections and voting is not high on their priority list. They need the push.

    But it’s been difficult to gauge magnitude of the effect because both sides have been pretty well balanced. Maybe this will be the year of asymmetry that allows us to measure this effect.

  • JerryA

    Regarding news reporters or editors picking an extreme poll- getting people to read their articles (and look at ads) is what pays the bills. Of course they will pick an outlying poll to get people worked up, instead of saying “Yup, X politician is 6 points ahead of Y politician *again* this week.” Mind you, there are a lot of so-called journalists who would likely misspell “median” and “standard deviation” and not understand the math, never mind grasp the implications for the election. I know this is true based upon the number of fools who happily repeat verbatim whatever misrepresentations and lies campaign spokes-weasels tell them, many in quite prominent newspapers/websites.

    • TeddyVienna

      I’m a journalist, and this might be the most accurate criticism of journalists I’ve ever seen.

      CNN’s revenues have gone through the roof because of Trump. So while Twitter argues about whether they’re “liberal,” the truth is it’s not in their financial best interest to declare this campaign over.

      And yes, I’ve twice had to explain to fellow journalists that 1/4 = 0.25. The conversation usually ends by producing a quarter from my wallet and asking how many cents it is.

    • pechmerle

      Thanks for the link, Kevin. Fascinating look into how they’ve worked on securing a solid core base through social media data.

  • Bruce Sands

    Yes, and in 2014 the Dems were bragging about their GOTV efforts, 2% effect is dangerous thinking IMO. I am worried about complacency in the democratic electorate.

    • Amitabh Lath

      Where does that 2% estimate come from? Is it just from tallying up reports from all the door-knockers after the election is over? In which case it isn’t worth much because who knows how many would have voted anyway?

    • xian

      if you’re worried about complacency do some phonebanking

    • Prehistorian

      Regarding Amitabh’s point, there is a large literature on various aspects of the effect of GOTV efforts. One paper I found interesting was the recent one by Maskett and others on field offices in ‘Political Communication’.

      From the abstract: “We find that the presence of
      Obama field offices was associated with greater Obama vote share at the county level,
      although we cannot detect a similar relationship for Romney field offices.”

  • Ruth Rothschild

    What I like about this site is the logical and scientific approach that Sam takes to making sense of general trends in this election, both on the presidential and down-ballot polls, and the statistical tools that are used to show these trends. It’s an objective, no-nonsense, non-emotional approach to making sense of where things generally stand at any given moment in time during an election cycle, both down-ballot and from the presidential aspect. I’ve found the information and explanations of the various statistical analysis methods (e.g., explanation of various statistical measurements such as meta-analysis, regression to the mean, etc, etc) to be very educational and insightful. You could go crazy watching the polls on a daily basis because, as noted in this article, they vary day-to-day. But, PEC gives a snapshot with a logical approach so we can make better sense of thing. The bottom line for me is that I’m learning alot about polling techniques, but the most important thing is to get out and vote for the candidates of your choice– both presidential and down-ballot– because each vote counts. I’ve also found the various articles to be very insightful and educational. And, I also enjoy reading peoples’ comments and the ensuring discussions on the various articles.

    Thank you, Sam, for hosting this website.

  • Greg Hullender

    Shouldn’t it be possible to compare early voting this year vs. early voting in previous years to get some sort of estimate as to what the final turnout will be? Or are the early-voting rules too inconsistent from year to year?

  • A

    Anyone can make a couple of phone calls and knock on a few doors–the notion that someone has neither time or money to spend is probably rare indeed.

    Or even donate $1 or $5 if you can’t afford more. Every bit of time and money does help, even if it’s just a few minutes or a few dollars. Nobody is telling you that it must be exorbitant amounts of money or hundreds of hours of time or it’s not meaningful…

    Seems to me that Sam is talking about a particular type of anxious person who watches polls and clicks refresh (I might resemble this at times) and doesn’t actually help and get out and do stuff to really make a difference.

    Making his comment seem as if it’s about someone’s so-called privileged position in life is disingenuous, I think.

  • Kevin King

    I’ve been fascinated by the Utah situation. Looking at, it certainly seems like Ipsos/Reuters and UPI/CVOTER’s not including McMullin is affecting the medians. In point of fact, that race could very well be even closer than 8%. I’m not sure how the situation might properly be accounted for, other than what is happening. At least Ipsos has 30% undecided, much of which I would gather is McMullin’s share. It’s an unusual situation and bears watching, IMHO.

    I really hope McMullin pulls it out, somehow, and goes into the history books: how interesting to have a non-major party candidate win electoral votes for the first time in almost 50 years!

  • jharp

    So why is it preferable to look at the median dog instead of the average dog?

    And many thanks to anyone who might take the time to respond.

    • Slartibartfast

      Because one lone wolf can move the mean outside the pack while the median is always in the center of the bunch.

      Consider the data set: {-1,1,30}

      mean: 10
      median: 1

    • Josh

      Outliers skew averages but not medians.

    • T S

      Because average dogs are mean….

    • Owen

      Or another example – if you live in a small county with lots of poor folks and one retired billionaire, the average household income might be 1 million dollars. The reality might be that most people are below the poverty line, but that one guy skews the average enough to appear that everyone is well off. The median, on the other hand, would give information that is meaningful due to elimination of this outlier.

    • Steven

      It also is a lot more work to add up the size of all the dogs and divide by the number of dogs you see. Might as well look for the median dog.

  • Veronica

    I’ve registered voters and donated some dough.

  • Field Worker

    As someone who has worked as both a volunteer and paid staff member in three of this year’s contested states, I think that the analysis of GOTV above is somewhat incomplete. I feel that GOTV is most effective in smaller states with lower traditional media exposure, and has the most effect on medium-propensity voters. Such voters often need to be contacted several times to maximize the chances that they indeed cast their ballots.

    I think it is also noteworthy that the swing states of Iowa and New Hampshire, due to their small size and outside influence in the primaries, tend to be more receptive to traditional GOTV efforts. Keep in mind that a surprising number of voters in these states often personally meet members of the (final) presidential tickets (especially in Iowa due to the caucus system).

    As a staff member in Iowa, I was extremely proud of the fact that Iowa substantially outperformed most polls and aggregators (by much more than the fractions of a percentage point mentioned above). President Obama carried Iowa by six points in 2012.

  • Some Body

    Silver just had a post on polling variability as well, also comparing it to previous election cycles. His conclusion: the variability is just above the confidence interval (if you ignore any error that’s not pure sampling error). I think he was using a normal distribution for this, though. It’s an interesting post, and an interesting comparison (and complement) to this one. It’s here:

    And on the PS part, well, I don’t think I’m legally allowed to contribute time or effort to US political campaigns (or, more precisely, that US political campaigns are legally allowed to accept money or efforts from me), so I’m fine with keeping everyone out there in the Western Hemisphere out of trouble, and with devoting most of my time and resources to pursuing radical political change in the little hellhole I’m from. US poll watching is more of a hobby :-)

  • A

    A little levity on Twitter tonight:

    LOLGOP ‏@LOLGOP 11h11 hours ago
    Thank goodness someone invented a medium where strangers could argue with neo-Nazis about poll numbers.

  • Anja Fare

    I’m very intrigued by the…what should we call them? Uncaptureds? this year.

    Texas alone has over 1 million new voters which is almost as many as Romney won the state by in 2012. And many Registered Republicans are voicing an inclination to stay home, to not vote at all for president, or even to vote for Mrs. Clinton.

    I don’t think our methods of polling are good at measuring the kinds of radical shifts we’re going to see in this election.

    The huge number of new Latino voters just aren’t being sampled well nor are married white women who are caroming away from Trump in droves.

    It will be interesting to get the postmortem in the days after the 8th, especially in the swing states and in Texas.

    I don’t think this is the beginning of a national trend, but it is certainly riveting.

  • anonymous

    The web statistics probably show that a majority of Sam’s readership is mostly interested in the Presidential contest. Not sure why this should be annoying. Everyone should vote, but everyone does not need to be financial or time contributors to electoral campaigns. This site is attractive to me because I see a different and accurate model backed up by clear explanations, and because the resulting understanding enriches my mind. There are plenty of other sites now that do electoral predictions that could provide me the same fix, but I am really looking for new blog posts when I come here, not just the Presidential contest graphs. I make no apology for checking the graph though, not doing so would require me to be half Vulcan.

    • Sam Wang

      “half Vulcan”…I do view the examination of Presidential polls as a political version of pon farr. Every four years, a weird passion grips me and for several months, all I can do is calculate medians and talk about them. Thousands are drawn in at the same time.

    • GM

      Sam, that’s probably the best forum reply in the history of PEC. I tip my hat to you.

    • alurin

      I think it’s perfectly logical to check the Presidential contest graphs. The binary outcome of the Presidential race is not the only important outcome. As Sam has noted many times, the downballot races are actually correlated with the top of the ticket, and the Presidential race probably has better time resolution. The bigger Hillary’s lead, the more likely a Democratic takeover of the Senate and House becomes. That is especially true now that the Clinton campaign is spending some of its resources to boost downballot candidates.

  • Josh Soffer

    “Making his comment seem as if it’s about someone’s so-called privileged position in life is disingenuous, I think.”
    Of course, you might agree that there may be other legitimate ways to characterize Sam’s comment that are neither disingenuous nor complimentary.

  • Olav Grinde


    When I went to college, a study showed that the average professor spent an average of 63 hours per week on teaching, research, conferences and other job-related work. That exceeds the workload of most Americans.

    I would be surprised if the workload is less for a neuroscientist and brain researcher at Princeton University.

    And yet, in addition to his “day job” (which I am sure often extends into the wee hours), he finds time to run this excellent election prediction website. I have been following PEC since 2004 – in these 12 years I have found none better.

    You and I have 168 hours per week – just like Sam and everybody else. I don’t know what sort of hours you work, or how many hours you sleep or watch television, but I bet that still leaves a few.

  • Ed Wittens Cat

    on GOTV:
    heres a paper that postulates GOTV effect of field offices at increasing up to 1% vote share per county–>
    heres monkeycage on 2012
    i worked on OFA in 2012 and the emphasis was on decentralization and shoe leather
    ive done 2 weekends on Project Ivy now and the emphasis (at least here) is centralized phone banks & emails
    Project Alamo is interesting in that it pushes digital via facebook– the Breitbart model
    But the goal of Alamo isnt really GOTV– its securing the hard right half of the GOP base for post election ops.
    I wonder if any Alamo staffers got fired for the voter suppression comments to Bloomberg– the word “suppression” is pretty toxic in the general.
    The choice of Alamo as a name is interesting– i remember a kiddie cassette that had old disney songs on it from my childhood…
    “In the sand he drew a line
    with his hardy saber
    out of 189
    not a man crossed that line
    you may ne’er see your loved ones
    Travis told them that day
    those who want to can leave now
    those who fight to the death let them stay”
    the Alamo was fought against Santa Ana and the 10,000 man strong mexican army.
    seems appropriate.

  • Rudy

    Sam, you make a simple and persuasive case that it is “the median, stupid”. But what I see pernicious is I see that Trump and his cohorts pick the outlier polls to argue that things are looking up and, simultaneously, the main body as evidence the election is rigged.

    • Destiny

      or maybe, people have just witnessed the rigging first hand…In North St. Louis, incumbent Penny Hubbard won the 2016 Democratic primary for Missouri’s 78th House District by 90 votes. Her challenger, Bruce Franks Jr., contested the results, citing a lopsided absentee vote tally that heavily favored Hubbard. Franks had carried the in-person vote. District Judge Rex Burlison determined that a sufficient number of improper absentee ballots had been cast to change the results of the election, and ordered a second election. Franks won the redo by a margin of 1,533 votes.

    • Jeremiah

      @Destiny I don’t know why the lopsided absentee vote would necessarily indicate any fraud. Fraud indicates fraud and in this case many irregularities were discovered concerning how absentee ballots were processed.

  • Matt McIrvin

    About GOTV, I think those quotes are right: there’s a real effect, but we’re talking about sub-1% differences. Those can be crucially important in a close race, like Ohio or Iowa, or some of the key Senate and House contests. It’s not going to turn a 6% national lead into some enormous blowout.

    • alurin

      I don’t think there’s much data to support a conclusion either way. In fact, this election will be an interesting test, since it’s the first time there has been a substantial asymmetry in GOTV efforts.

    • xian

      Well, if you count voter registration as part of GOTV then the effect can be much larger on the state level. For example, the Obama campaign reported registered enough Dem voters in Nevada in 2012 to increase their net share of the electorate in that state by 7%.

  • Phoenix Woman

    Re: the ground game – Trump showed that ground game doesn’t matter in the GOP primaries if the candidate is willing to be openly racist. Once that was established, he was in like Flynn – and set to be destroyed by a general election audience that is far less white than the Republican electorate.

    Having said all of that, ground game can matter quite a bit. It got Rick Nolan elected by insuring more Democrats turned out than Republicans in Minnesota’s Third District.

    By the way, here’s a good side by side chart of HRC and Trump fundraising for the past few months:

    • Rieux

      Er, Nolan represents Minnesota’s Eighth District, not the Third.

    • Dave Wink

      I am not at all convinced that fundraising impacts the Presidential election, currently. That is something worth examining after this election. If the headline of this election is that polarization creates a situation where the map is set in cement, then it seems that media sources give you all the advertising that you need. Who is not aware that the two Presidential candidates exist? Romney spent a lot on the last election and got nothing for it.

      I do not extend that idea to down ballot races.

    • alurin

      I disagree that Romney got nothing for his money. He didn’t win, but he basically cut Obama’s margin in half, relative to 2008. Incumbent Presidents running for their party’s second term in office typically expand on their margin (see Bush 2004, Clinton 1996, Reagan 1984, Nixon 1972, Johnson 1964, Eisenhower 1956, Roosevelt 1936). So I think Romney did pretty well, possibly because he had a lot of money to spend.

  • Prairie Pundit

    FWIW, I am guilty of coming to this site three or four times a day for a sanity check on the state of the presidential race, but I also draw on PEC’s valuable information about down-ticket races when deciding where to donate my walkin’-around money. If not for PEC’s assurances that the presidential race is so stable, my donations would probably go to Hillary instead of to McGinty, Ross, etc. So, for this faithful reader, PEC is working as intended. I reckon there are many more like me.

    • Marlene Snyder

      Yes, I too come here several times a day just to get my breathing under control. But my money has gone to Nevada–I actually grew up in Utah, not a lot of point in donating to any Blue races there, but the Nevada senate race is important and is the happy (I hope) recipient of my available money this fall, well worth it! (I now live in Canada, and I have already voted).

  • Ravilyn Sanders

    @Sal // Oct 27, 2016 at 11:38 pm

    Given the stature and track record of Sam, he is giving up millions of dollars in personal pay to be a prof in Princeton.

    If Sam is willing to do what someone else wants him to do, (i.e. working on project not personally interesting and motivating for him), he would be heading some quant dept of some hedge fund, raking in millions. I am sure his *students* take up these careers and *start* at a salary higher than his at the univ.

    I know of professors who routinely write rec letters and provide references to their grad students who go into the “industry” with a starting salary on par with tenure-track positions. Heck, my own starting pay was not too far below my Guru’s pay in the univ.

    You know his stature and guess the caliber of his grad students. Extrapolate and see how much he values his freedom to work on “interesting” projects over money.

    Heartfelt thanks for Sam and his grad students and others toiling behind the scenes to keep this site up. This site loads and works faster and better than many professionally maintained sites.

  • TeddyVienna

    This may be a dumb question, but why is the meta-margin down to 4.1%?

    I come here somewhat for reassurance but also because I’m interesting in learning more about data-driven journalism. The difference between the numbers here and at 538 is worth exploring. (Though, frankly, I think it all comes down to Silver feeling nervous after underestimating Trump in the primaries, so I think he’s added a lot of wild variables to his simulations.)

    • Matt McIrvin

      Clinton’s MM has run about two points behind her national lead all year, with some time lag. I think the reason is that she’s doing unusually well in deep-red states where those votes don’t really buy her anything, because they won’t flip the state.

  • M. Leo Cooper


    Please forgive me for asking once more about the effect of early voting on the polling. Do the pollsters actually ask whether the respondents have voted, or even for whom? Do the pollsters thus separate the results into two categories — those who have voted and those who have not yet?

    It seems to me that early voting might have the effect of decreasing the effective sample size of at least undecideds. And, on election day itself, early voting trends plus exit polls should make it possible to call at least some of the swing states with pretty good confidence before the polls close. At least one news organization has already hinted that it will do so.

  • Andy

    I think the one thing that early voting does for Dems is it helps circumvent GOP efforts to suppress the vote by closing poll sites or limiting the numbers of machines. The more people who vote early in Dem-leaning districts, the more manageable the lines are on Election Day, meaning that fewer people will get frustrated and leave without voting.

    I really hope that HRC as POTUS puts fixing our messed-up voting system near the top of her list.

  • Michael Coppola

    I’m a bit confused… If Presidential, Senate, and House races largely track together, and an HRC landslide is needed to flip the House, then what’s so wrong with keeping one’s efforts focused on the top line race?

  • LBS

    About Hillary’s ground game: I have been travelling once or twice a week to WI, where I grew up and went to grad school, to volunteer for Hillary and Russ as I did for Obama both times. Midterms I work for Congressional Democratic candidates in IL, where I live. My anecdotal observation this year is that the door knocking to urge voting among likely supporters is of special value with new voters (most on my list have been Latino/a) and sometime voters (do not vote in every election). The most frequent questions I get from them are: Where do I go to vote? and What hours are they open? Have also gotten questions about having moved and therefore having photo ID with different address than their home. It is quite eye opening. Another observation: Many of the women I talk to seem more committed than past elections. With WI looking pretty solid I am thinking of spending the rest of run up to the election in IL 10. Although I do contribute to close Senate and Congressional races and I do work, I am guilty of clicking on this site at least four times a day to stem my anxiety and get a reality fix. HRC margin and probabilities calm me down but the Senate makes me more anxious! As a retired psych professor, it has been interesting to observe my own behavior (mostly rational) and emotional state (mostly irrational) this election.

    • Roger

      Thank you for doing the work you do.

    • Ettaroo

      I would be curious to hear about reactions during GOTV post-Friday’s new email wrinkle. Personally, I don’t think this news should have legs, but its being treated by most of the media as a crisis/disaster for HRC, i.e. “optics.” I’m curious if this actual impacts any of the voters targeted by GOTV efforts.

  • Froggy

    A dog party? A big dog party! Big dogs, little dogs, red dogs, blue dogs, yellow dogs, green dogs, black dogs, and white dogs are all at a dog party. What a dog party!

    And I found this while looking for that, totally off topic and totally worth reading:

  • GC

    “This site loads and works faster and better than many professionally maintained sites.”

    Preach it! I love how this site doesn’t have ads or crapware, and that Sam shares the code behind the model.

    Btw, this site encouraged me to get three people to register to vote. As someone who’s voted in every election since I turned 18, it’s still surprising how many people — even well-informed ones! — need prodding in order to get out and vote.

  • Mike

    I come here for reality. I understand (some of) the statistics and (most of) the concepts used. But I couldn’t do (any of) it myself. For me, reality is comforting this year and therefore I do come here for reassurance.

    If the presidential race was close, I would have donated to HRC. Since it isn’t, I gave some money to Doug Applegate hoping against hope that he can beat that prick Darrell Issa.

    Thanks, Sam.

  • Michael

    Sam, Please try not to judge too harshly the anxious readers who come here for reassurance. This election has brought on more jitters than most, what with all the talk of “fraud,” and “revolution,” and not accepting the results. This site is cheaper than therapy, and probably more reality-based. Do not underestimate the service you provide.

    • Greg Gross

      Hear hear! Your site’s followers can be both in need of reassurance AND committed to helping in the process by donating, phone-banking and otherwise promoting their candidate(s).

      Perhaps you were referring to the median site visitor, in which case, I may see your point. Otherwise, please don’t pigeonhole.

      Thanks for all of your reassurance AND advocacy for having us play an active role in the political process.

      On that topic: Perhaps the thermometer should once again grow a bit taller? More than 10 days to go!

    • (((CassandraLeo)))

      Gotta second this. I constantly freak out over the race because Trump is a candidate of unprecedented authoritarianism and awfulness. I know very well that he has a very slim chance of winning, but the fact that that chance is anything greater than zero is absolutely terrifying. I honestly don’t think I’ve ever been this afraid of anything before in my entire life.

      That said, I’ve also thrown quite a bit of money at the ActBlue fund etc. and have volunteered for the Clinton campaign registering voters (in Florida) when time has permitted. One can freak out about the presidency and still try to do what one can to affect the outcome of downballot races.

  • Amitabh Lath

    There will never ever be 100% certainty of anyone winning, even on the eve of the election. The tools used to measure have statistical and systematic uncertainties, and known and unknown correlations.

    Richard Feynman relates an anecdote about questioning Space Shuttle engineers after the 1986 Challenger disaster, asking what they thought the chance of (1-) failure was. All gave numbers in the high 90%’s but one guy said 100%. He was management. Feynman called him a weasel in his memoir.

    To soothe your freakouts, here are links to the most awesomest of all freakouts, Andrew Sullivan post Obama-Romney debate.

    and again

  • Kanwaljit Singh

    OCTOBER SURPRISE! FBI is reopening their investigation into Clinton emails. I think it’s safe to say, the race just got scrambled into a toss-up category, at the minimum, if not an outright Trump win.

    • Matt McIrvin

      Look at the evolution of the Meta-Margin and EV count the last time Comey dominated the news. There was a drop but not to anywhere near a Trump win. I find it hard to believe that Comey is going to recommend criminal charges against Hillary Clinton in the next 10 days, which is the one thing I could see being worse than what went down then.

      This is why you run up the score.

    • CJM

      Pete Williams’ reporting on this issue is leading me to believe this is going to be much ado about nothing.

    • Jon Kanders

      Trump has -30 favorability, recent polling indicates he’s underwater with women (the largest demographic in the electorate) by at least -15, has demonstrated he has no command beyond the most superficial understanding of any issue, and has insulted every other demographic. He’s hovered at 38-40% in polling for months, it’s his ceiling. Frankly, I think Hillary wins even if she’s indicted in the next 10 days, which is extremely unlikely.

    • RonL

      Wait an hour before you gloat. The emails nearly “discovered” are related to the Weiner sextext investigation & are not from Clinton. = Other than among the alt-right hystericals: a shelf life of less than a couple days. Apparently no real news here. Less impact.

    • A New Jersey Farmer

      I think it comes down to whether enough women decide that they can’t vote for Hillary because of this news. That it seems to be about Wiener will only reinforce the notion that men behave very badly. Hillary still wins.

    • Kanwaljit Singh

      To RonL: I am not gloating. I am a Clinton partisan. I am distressed actually.

    • Greg Gross

      Interesting perspective on Comey’s new disclosure to Congress about emails possibly relevant to the Clinton probe: a former DOJ spokesman slams Comey’s move, saying it violates DOJ rules, and that this is a blatantly political manuever by the Repub FBI Director:

    • Rob

      Then Kanwaljit, take a breath. If Clinton were running against a likable Republican then I might agree but she’s running against Trump and I suspect the hit may be less than he hopes for.

    • Kate H

      This story feels less like an attempt to achieve a Trump presidency than an attempt to lay out the battle-lines of the Clinton presidency.

    • Roger

      Compiled betting odds dropped from 90% Clinton to 87% initially on the news, and have recovered to 89% a few hours later:

    • Kevin King

      I don’t have a link handy because I’m on my phone in bed, but I also saw that the stock market took a hit on the news.

    • Rob

      It did take a hit but then gained it all back as it became apparent that maybe this is not as big a deal as the GOP would like to make it.

    • Rob

      Oops, it gained nearly all back, the Dow was down 8.49 today, no sweat.

    • Chris

      There is no factual basis for your statements. If you’re a regular PEC reader you should really know better.

    • OldenGoldenDecoy

      Hello NJFarmer. From your deep impression of this ongoing kerfuffle relating to the female voters you seem like a fairly in-the-know sensible and well educated fella. Now one question. Can you tell us exactly how many “communists” were in the left-wing of the Democratic Party faithful back in 1972? I was 26 at the time. How about you?


  • Sal

    You may not have meant it like this, but people who are “relieved” by following your site may not have the possibility of activism, nor do they have the luxury of a well-paying tenured job at Princeton University with health insurance, a retirement plan, etc to be able to rush out and become activists. If you genuinely despise those who merely read your site instead of becoming activists, why bother to publish data at all, instead of a website full of uninterrupted exhortations to activism?

    • Rob

      Wow, you seem touchy. How did you take Sam saying he’s “a little annoyed” to actually despising people? You’re leaping to places his language doesn’t support. If I interpret Sam correctly, he’s suggesting we need to remember that who we elect as president doesn’t “fix” anything if we also end up with a congress that’s opposed to that person. That means if you’re a Hillary supporter and she’s left with a GOP House and Senate it’ll be 4 years of meaningless frustration, nothing but gridlock, refusal and occasional government shutdowns thrown in for fun so we need to remember to worry about those races too because if all you have is the White House, you don’t have enough to get real.

  • Gopa

    Sam, do you or your assistants selectively screen and block comments? Earlier you said this was only about data. I see all kinds of comments here, dogs, party dogs, etc. Do your assistants mean what they say and say what they mean, or like Trump at his rallies throw dissenters out?

    • Mike

      This is not a commercial site. I don’t think Dr. Wang is selling our data to digital marketers. We are not being bombarded with advertisements for goods or services for/from the last commercial site we visited. We are not customers in any sense of the word. We have been invited to participate in a community of interest.

      Or was this satire? I hate it when I don’t get satire.

    • Ruth Rothschild

      Not sure if you’re response was to my post, so decided to clarify why I asked Sam the question. I know that this isn’t a commercial site and also know that Sam isn’t selling his data to other digital marketer sites. However, I do know, based on reading the various articles he has posted on this site, that his analyses for any of the races, presidential/senate/house, are based on data he gets from various sources– i.e., Huffington’s databases and various other pollsters. On his Meta-Analysis page at this site, Sam explains the methods he uses to derive the statistical data (e.g., meta-margin, electoral vote, removing noise from the various polls, etc). So, my question about the disparity in the national House race between Huffpost’s data, other pollsters’ data Sam may use, and his own meta-analysis has nothing to do with his selling data to digital marketers but rather is because I don’t understand why there is such a disparity between his statistical data and that of other pollsters’ data that he may be using in his meta-analysis. I’m seeking to understand this better. Perhaps you can explain this to me?

  • anonymous

    Great, just as we were talking about ignoring the Presidential contest, a large perturbation 11 days out from the election, which might take longer than 11 days to manifest in state polling. I feel you, Paul Krugman.

    • (((CassandraLeo)))

      It would be more worth worrying about if the story hadn’t fallen apart literally within hours of being reported. If this affects the election at all, it could be in the form of backlash against what appears to be a partisan attempt to manipulate the election results shortly before the close of voting. However, I’m predicting no measurable effect on the race.

    • TeddyVienna

      If people were rational and took the time to look beyond the Wall Street Journal’s headline, this would be no issue.

      But I’ll be coming here for reassurance even more now, I’m afraid. Sorry, Sam.

    • Kevin King

      This isn’t going away that easily. We shall see what if any affect it will have on polls. Annoyingly to my ears, the media had already been breathlessly reporting a “tightening” of the polls before the email matter. They certainly seem to be trying to create a closer race

  • Philip Weiss

    I’ll be honest – sometimes I get a little impatient with readers who show up here mainly looking for reassurance in the Presidential race. I understand this is an attraction of the site.

    Perhaps you should re-design the site and offer some statistical analysis that is drives people toward Senate or House races then. There’s information in the meta-margin for the presidential race that isn’t offered anywhere else. But a national look at local races doesn’t really interest me. I know doing the equivalent of meta-margins for individual senate races would be a lot of work, but that’s the sort of thing that would be needed to get me coming here for down-ticket information.

    And at a minimum, to accomplish that you really need to redesign the site so that what you do offer for down-ticket stuff is better presented.

    I don’t think what you do present is wrong. But it doesn’t really seem to fit with your stated goal of getting focus on the down-ticket.

  • David Elk

    In case you missed it – speaking of medians, I think today’s Ohio median should be -4.0 going by the huffpollster feed, but it’s 0.0 in the archive.

    It looks like will skip polls with an ending date of “today” since it filters by < day instead of <= day.

    But maybe that's intentional, I was just confused trying to reproduce the results.

    • Sam Wang

      Thank you for flagging this. What line of the script do you think would be changed to prevent this?

    • David Elk

      Line 354, I believe.


    • Sam Wang

      OK, we have changed it. I do find it odd that any pollster would release a poll on the same day as its completion…but for now we will do it this other way.

      FYI, the general idea here is to have a rule that allows regeneration of the EV history at a future time, once the election is past.

Leave a Comment