Princeton Election Consortium

A first draft of electoral history. Since 2004

On CNN around 9:25am Eastern w/Smerconish

November 12th, 2016, 8:12am by Sam Wang


Post-mortem. Plus, I keep my word about the bug.

Tags: 2016 Election

83 Comments so far ↓

  • 538 Refugee

    Business Insider is quoting Sam’s tweets in an article about Jill Stein raising money for a recount:

    http://www.businessinsider.com/jill-stein-wisconsin-michigan-and-pennsylvania-2016-11

  • Emigre

    In a preliminary statistical analysis Nate Silver concludes that education is a better predictor for a Trump voter than income:
    http://fivethirtyeight.com/features/education-not-income-predicted-who-would-vote-for-trump/
    Considering the ignorance among students to sort out real from fake news does not provide a cheerful outlook for the future:
    http://www.npr.org/sections/thetwo-way/2016/11/23/503129818/study-finds-students-have-dismaying-inability-to-tell-fake-news-from-real?

  • LMB

    Sam, what do you think about this comparison of outcomes in paper vs. electronic voting machines’ outcomes?

    (Computer Science) “Experts Urge Clinton Campaign to Challenge Election Results in 3 Swing State”

    http://nymag.com/daily/intelligencer/2016/11/activists-urge-hillary-clinton-to-challenge-election-results.html

    As the Verified Voting Foundation has said for years, unless there is a paper confirmation of each vote, electronic machines have a number of ways in which their results can be tampered with. And with the machines’ proprietary software, it’s hard to prove that a malicious (foreign) actor couldn’t successfully target some counties in swing states.

    I wish I knew how to add my vote to those urging Clinton to demand a forensic audit ASAP, this week, before one of the deadlines for challenge pass.

    (I do understand how Obama would shudder at the thought that Trump – who was so clueless and far behind in transition planning – would be derailed by a failed challenge.) But at the very least we owe it to ensure that each vote IS counted accurately.

  • gumnaam

    Kevin Drum weighs in, and assigns the blame for Trumps victory on (in this order): millenials, Comey, and working class voters. I agree, but want to add that Comey is the only thing that Hillary did not try to address.
    http://www.motherjones.com/kevin-drum/2016/11/why-clinton-lost-bitter-bernie-crooked-comey-and-wounded-working-class

    • Alan Cobo-Lewis

      Really enjoyed the reader comment telling you that you shouldve used medians.

      Uh, no

  • GEinNY

    Here’s an interesting article:

    http://www.rawstory.com/2016/11/something-stinks-when-exit-polls-and-official-counts-dont-match/

    A tidbit from that article:

    “SR: Let’s talk about what you found this week. I’m looking at your 2016 presidential chart. I’m looking at North Carolina for example, where it says the exit poll margin was 2.1% ahead for Clinton, but the final vote count showed Trump with a 3.8% lead. You have similar 4.4% Clinton lead in Pennsylvania but then losing by 1.2% to Trump, a 5.6% shift. You have Florida where she was ahead in exit polls by 1.3% and ends up losing by 1.3%, a 2.6% shift.”

    • Jay Sheckley

      This is the very info I’m hoping Sam will write about. What can we learn??
      As for the bug, Sam, of course kept kept your word.
      Thanks for all you did to get us through this.
      <3

    • George H

      So conclusions such as 30% Hispanics voted for Trump are anything but conclusive.

  • Alan Cobo-Lewis

    Sam,

    The analysis at https://www.washingtonpost.com/news/the-fix/wp/2016/11/17/how-america-decided-at-the-very-last-moment-to-elect-donald-trump/?hpid=hp_hp-top-table-main_fix-last-minute-830a%3Ahomepage%2Fstory suggests that undecideds broke late for Trump. They speculate about breaking against incumbent. This is essentially what you hypothesized in 2004 (a hypothesis we know was disconfirmed). How confident are you in their explanation?

    If there’s a need to smear out predictions, I’d think it should be a symmetrical smearing–but should it depend on % of undecideds or not? And if it should, in theory, depend on % of undecideds, if that a parameter that can be estimated reliably enough to make it more predictive to include that in the model?

    • gumnaam

      There were no Comey letters in 2004. We can spend years soul-searching about this election, but the real difference between a Clinton victory and a Trump victory were those letters.

  • Ash

    I want to believe when President Obama said that Trump was not an idealogue.

    http://www.dailykos.com/stories/2016/11/10/1595617/-The-architect-of-the-most-racist-law-in-modern-American-history-has-been-named-to-Trump-s-team?detail=facebook

    But news like above makes me think otherwise. He might be thin on policy, but he definitely certain about Making America White Again.

  • Tom_b

    Not likely to happen, but if just 36 GOP electors flipped to, say, Romney, it would go to the House. The “faithless electors” would, at most, face a fine. They have basically nothing to lose. Romney would easily win in the House; he’d get all the Democrats and all but the most insane Republicans. American democracy would be rescued from the scrap heap of history. The deplorables would be annoyed, but they are sheep; they always fall in line.

    Just putting it out there.

    • Seth Gordon

      All the executive power that Clinton would have had in this article’s “worst case” scenario, Trump will have in the real world, starting on January 20.

      And all the legislative power that the Republicans would have had in such a scenario, they already do have.

      Given a magic wand, I’d happily wave it in order to make that “worst case” scenario come true.

    • pechmerle

      I would far prefer more gridlock to the enormous damage that the Trump administration plus control of both houses of Congress is going to inflict on domestic policy. For just one example, the Republicans — after waiting for years for their chance — are going to open millions of acres of previously protected public lands (and seas) to more mining, drilling, and logging. The damage done will be either irreversible or take many generations of recovery. The “long term” at stake here is not just until 2020 — it extends far beyond that in many policy areas.

    • Bulgakovs Cat

      So…data over drama except when the data is bad?
      I waded thru some of the “Dear Liberals” tweetstorm and what this is really about is revenge– Trump is redbrain royalty– a giant poke in the eye to the culture thats has disenfranchised and humilated them: music, academe, media, college educated, intellectuals, elites, environmentalists, scientists. I read the history of Brietbart media– its really about taking back culture– it started as a complaint abt the blueing of Hollywood– the next 4 years are going to be a brutal battle against the “libtards poisoning the childrens minds”. Goodbye teachers unions.
      I think the musicians’ celebrity endorsements of HRC actually pushed more blue-collar whites toward Trump.
      Welcome back to the culture wars.

  • Seth Gordon

    A month ago, if someone had asked me what I liked about the PEC model, I would have given two reasons:

    (1) It makes the right call—better than 538, as measured by Brier scores.

    (2) It is parsimonous—relying entirely on state polls, without dragging in a whole bunch of other parameters regarding the reliability of each pollster, demographic characteristics of states, and so on.

    In retrospect (and looking forward to 2020), are there changes in the model that could recover more of the first quality while preserving the second? In particular, can the model be updated to account for a higher risk that undecided voters in each state will break in an asymmetric fashion?

  • Bulgakovs Cat

    Assuming that the underlying structure of reality is gaussian works great for physics.
    Foe social science not so much.
    I’ve spoken before abt the hypothesis that the underlying structure of reality is fractal.
    https://www.bloomberg.com/view/articles/2016-11-11/why-science-couldn-t-predict-a-trump-presidency

    • Scott H

      I think the idea is that that even if the underlying structure is fractal, measurements of that reality still follow a Gaussian distribution when there are a large number of measurements. That’s what makes polling hard. What is an estimate of the true measurement when you have a small and incomplete dataset? You have to add the missing information in as best guesses based on previous or ancillary data-sets, which is where things can go wrong.

    • Bulgakovs Cat

      thats why we need transparency both in polls and aggregators, as Dr. Dedeo points out.
      and i dont agree– im a heretic.
      I think its Self Organizing Criticality all the way down
      and i will likely either either be burnt at the stake or vindicated some day :)

    • Lorem

      I don’t think I understand what you mean by that. Or, more precisely, I don’t understand what predictions we can make from that or how we would reconcile that claim with the central limit theorem.

    • Bulgakovs Cat

      Lorem, u will have to wait for the preprint.

  • Nancy DiTomaso

    Former Princeton political scientist, Larry Bartels, now at Vanderbilt, also showed the very close relationship between the 2012 election results and those in 2016. Bartels argues in a recent piece in Washington Post Monkey Cage that given the fundamentals of the election and the state of the economy, that this election, like 2012, was a toss-up, with Romney winning by a small margin in 2012 and Trump winning by a small margin in 2016. There were only about 110,000 votes that separated Clinton and Trump in Wisconsin, Michigan, and Pennsylvania combined, suggesting indeed that by election day, the outcome was a toss-up that could have gone either way. The fact that all of the polls were off in the same direction and consistently across the campaign is surely disconcerting, but we should not draw the wrong lessons from the outcome of this election.

    According to exit polls from NBC and NYT, Clinton won 37% of the white vote, 88% of the black vote, 65% of the Hispanic and Asian vote, and 54% of the female vote. This is only narrowly divergent from what the Democratic Party usually needs to win presidential elections, which has been 40% of the white vote, 90% or more of the black vote, 60%+ of the Hispanic and Asian vote, and a majority (with more the better) of the female vote. See Black and Black, Divided America, for example, where they also argued that presidential elections in this period depend primarily on the white working class vote in the Midwest, which it did in this election as well. The exit polls also show that Clinton won a majority of those who were most concerned about the economy, but lost by a wider margin those who were most concerned about immigration and terrorism. Hence, this was not about the economic conditions of the white working class, as much as about race and sex.

    Importantly, the discussion about what the Democratic Party should do from here has so far been misplaced (once again). It is not that the Democratic Party forgot about the issues of the white working class, but rather that the Democratic Party since the Civil Rights Movement has also had to address issues of race and ethnic minority voters, and since Reconstruction these votes have often been trade-offs. The more attention to the black vote (race issues), the more white votes the Party is likely to lose, and the more attention to the white vote (class issues), the more blacks may stay home. The white working class has been leaving the Democratic Party because they feel that they are in competition with blacks for jobs that pay a living wage, and they have blamed the Democratic Party for policies that they feel benefit blacks at their expense. The Party needs to find a way to bridge the conflict between the white working class vote and the nonwhite vote in ways that both groups can feel that they will benefit without it being at the expense of the other.

    White collar professionals (and students) who came into the Democratic Party in reaction against the influence of white Christian conservatives have played a role, initially in supporting civil rights issues over the issues of the white working class, but in the Sanders campaign, they did the opposite, and supported class issues (i.e., supposedly white working class voters) while arguing that Clinton was winning “only” because of the minority vote. The more the news media emphasized her support by race/ethnic minorities and women, the more votes she was likely to lose among working class whites (including working class white women who voted with their husbands).

    I have an extensive analysis of these issues in my 2013 book, The American Non-dilemma: Racial Inequality without Racism (Russell Sage Foundation), which reports, among other things on my interviews with whites in three parts of the country, including Ohio about their views of government, taxes, welfare, affirmative action, immigrants, and inequality. What I heard from my interviewees, whom I interviewed when the economy was doing well, was very consistent with what we are hearing now. This is not just a post 2008 issue and not a response only to current economic troubles. The white working class in the Midwest felt the same way and talked about the same issues when I interviewed them before 2008. They claimed to support civil rights but blamed government for supporting “them,” by which they clearly meant race and ethnic minorities. The tension between class politics and race politics has to be addressed by the Democratic Party in order to win in the future.

    • Josh

      “The tension between class politics and race politics has to be addressed by the Democratic Party in order to win in the future.”

      This seems to be in direct conflict with analyses from people like Ruy Teixeira and David Leonhardt who point out that demographic change will diminish the political power of the white working class steadily over time. Do you have any evidence to suggest that this isn’t the case?

  • Chris Burg

    I know I posted this in another thread and I apologize for repeating myself but it seems really important.

    A month ago there were numerous newspaper stories about the Russians hacking voting machine vendors in twenty states.

    Then this bizarre anomaly occurs.

    We know Russia would hack a US election. We know the machines are vulnerable.

    What are the statistical chances these events are unrelated?

    Is there a way to tell from the data if it has been tampered with?

    http://www.chicagotribune.com/news/nationworld/ct-hackers-target-election-systems-20160930-story.html

    • Eli

      Chris, you’ve misread that article: it says nothing about “voting machine vendors” being hacked, and in fact emphasizes that none of the targeted systems had anything to do with casting or counting votes.

  • MNP

    The problem is that when all the votes are counted, HRC may in fact equal Obama’s 2012 result. She overperformed her polls in blue states (where those votes do no good). But the key is that in the Midwest enough 2012 Obama voters flipped. A collapse in the Midwest means that you have to wait until NC, Texas, Georgia and Arizona become truly purple for a democrat to have a chance. Is this a single flip, or is this an outright expansion of the Southern Strategy into the Great Lakes and Midwest? If the latter, it is even darker than it seems.

  • Bao

    Josh // Nov 12, 2016 at 3:35 pm
    If, however, people do feel like their lives are improving, then it’ll be 2024…

    I can’t imagine Trump staying for a second term. He’ll either get bored or impeached. Once he sees the slow pace of Congress and the numerous roadblocks that will be put in place by Democrats and lobbyists, he’ll know why Obama was able to accomplish so little. With a determined resistance, the do-nothing Congress will once again rear its ugly head. Either that or he really will grab some woman by their p#$$y and get impeached.

    • Josh

      Remember that the House votes to impeach and the House currently has a big GOP majority. I think Trump would have to do something really, really bad for the House to be willing to impeach him.

      I do think you’re right that Democrats and lobbyists will fight tooth and nail in any way possible to prevent the removal of some Obama initiatives. (Dodd-Frank, for example, is probably a goner; Obamacare is a big question mark.) They won’t be able to stop all of those changes but through horse trading with McConnell and some strategery they’ll probably be able to frustrate the pace of change, especially since McConnell likely will not nuke the filibuster.

      My prediction is that Republicans spend the next 2-3 years methodically rolling back or removing safeguards put in place by Obama or earlier presidents while pursuing relatively little novel legislation of their own. Meanwhile, we’ll still have tons of immigration, mostly legal but some illegal; people in the hardest hit places in America won’t magically get good, high-paying jobs; ISIS will still exist, etc. Eventually, at some point, we’ll probably hit a recession–11 straight years of economic growth would literally be the longest expansion on record. Basically, it’ll be a reprise of the George HW Bush scenario–make a bunch of popular promises, fail to deliver, get hit with a recession just as you’re going into an election year–but this time, a much more diverse and left-leaning electorate will head to the polls in 2020.

  • pechmerle

    @Bulgakov’s Cat:
    He picked Priebus over Bannon for Chief of Staff — a small but positive step away from the edge of the far-right cliff.

  • silence dogood iv

    No one, no database, no statistical analysis could have predicted a coup with three major players. Russia via Assange/WikiLeaks, day after day, pounded Hillary and the Dems with millions of emails and no one could have ever predicted the FBI-Alt Right/Comey would break the law ten days before the election with another insinuation about Hillary’s emails and stuck Anthony Weiner in it. If you don’t understand this was a alt-right international coup you will never figure out how to stop it in upcoming elections.

    This is not a conspiracy theory – just go back and look at all the things Trump and his surrogates dropped on the press re this treason. The comments about rigged were not against Hillary. Trump was signaling what they were doing!!!

  • Alan Cobo-Lewis

    So, regarding the online panel poll with open-source data hat tracked the consensus but had a median shift toward Trump…was that more less accurate at the national level? (It was obv more correct if applied to state polls.) More to the point, if poll results are so sensitive to assumptuons (so the panel poll could be “unskewed” to the consensus) then how can we bring data to bear on the problem? Is an overall error really unobservable until election day, or are there methodologies that can actually make a better *measurement*?

  • Alan Cobo-Lewis

    Sam, really nice job using your platform while you have it.

  • gumnaam

    We now know that the EPA is toast. Any word on the fate of NSF and NIH yet? Is Ben Carson going to be secretary of Health and Human Services?

  • Five Tool Player

    I was dismayed, but not necessarily shocked by the level of support Trump’s candidacy gave rise to. Far more appalling to me, however, was the mindnumbingly irresponsible, pathetic, inexcusable failure of Anti-Trump people to vote. People who stayed home on election day are responsible for the results. We have no one to blame for the election of Trump than ourselves. I for instance live less than five miles away from the swing state of Michigan — a state that Clinton lost by a total of less than 12,000 votes — and spent a total of zero minutes in the get out the vote effort in that state. We need to do more. I commit to doing something everyday, however small, for the next four years to elect a Democrats in 2018 and 2020. Coming to this site and intellectually masturbating to the beat of Bayesian probabilities does little good aside from reinforcing how high our IQs might have been election day ante.

  • pechmerle

    I hate how this turned out as much as anyone (not named Clinton), but let’s not lose our objectivity.
    Trump has already said he’s in favor of keeping health insurance coverage for “children” up to age 26 in place. The man is not stupid; he knows that has been one popular feature of Obamacare.
    He’s also already said that he favors keeping the rule against denying coverage to people with pre-existing conditions. That one is a very difficult economic proposition without a mandate to require the young and healthy to be in the insured pool, so the insurance co. lobbyists will be all over it.
    But at this stage — campaigning over, actual policy jockeying beginning — it’s worth shifting to paying attention to what he is actually saying.

    • Bulgakovs Cat

      Pech, good points– and i think the GOP elites are back in control of Trump now– will only allow him a short leash on policy. Its not Trumps party now– if he had lost, sure, he could have fought the GOP for control. But hes been coopted
      As we saw over the past 8 years– the presidents only real power is to make war, spy on ppl, and veto legislation.
      Trump wont veto GOP legislation.
      I expect More War given the particular demands of Trumps psychology.
      the GOP establishment is back in control– Trump is just a sock puppet for Ryan and McConnell now. And eventually the base will see this– when Trump cant build the wall, when the jobs dont come back…its just a matter of time.
      The most interesting poll that failed (to me) was the scholastic poll– only wrong twice before.
      Dr Wang is good company.
      Trump is truly the president America deserves– like Chappelle just said, we just elected an internet troll to be the President of the United States.

    • Brian

      If he keeps those two provisions whilst eliminating the individual mandate, the individual insurance market will go into a death spiral. If he keeps the individual mandate, it gets much harder to look like he really “repealed and replaced the disastrous ACA”. This sounds like a tougher decision for Trump and the GOP than it really is. These people never cared about universal health coverage and they aren’t about to start now. It’s going to be up to the states to craft Romneycare systems, all-payer price controls, and new sources of funding for expanded Medicaid. We’re going back to a patchwork. In many states that will mean underfunded high-risk pools and adverse selection.

    • Bulgakovs Cat

      Can i retract my comment? it looks like there may be a fight btwn the Brietbarts and the GOP elites after all.
      https://twitter.com/TeaPartyCat/status/797841509194276864
      we will see i guess if Trump picks Reince over Bannon.
      https://twitter.com/RogerJStoneJr/status/797388379071672320

    • RDT

      He’s got a white nationalist as his “chief strategist.” That’s all I need to hear. If he doesn’t gut Obamacare I’ll breath a sigh of relief, but I won’t be any less appalled by him.

    • Rhina

      i think the GOP will impeach him and get Pence as president so they can get what the want:
      repeal and repeat Obamacare
      Eliminare medicare, medicatr
      spcial security to stock market
      repeal Rie vs Wade
      repeat gay marriage
      who knows what else.
      Scary ti.es, scari times!!!

  • Nick

    I hope you keep doing what you are doing Sam; you provide a valuable service, in this case the polls were off in an unprecedented way (national pretty close, but a number of states weird) … with time it would be interesting to see some more analysis of what you think might have happened with the polling and with this election in general. Pretty impressive popular margin for an electoral defeat.

  • Walter Manny

    Outstanding entomological integrity.

  • CS

    David Rothschild on his website Predictwise.com writes in a blog that his experimental private polling using display poll on MSN and mobile-only on Pollfish indicated that WI, MI and PA were leaning towards Trump. But he kept them hidden because he did not want to look foolish.

  • Bulgakovs Cat

    Dr Wang…the horserace media put Trump over the finish line, didnt they?
    the Horserace became more important than policies or truth.

  • Professor Woland

    It was a good stunt for television. But the fact that nearly every poll got Trump’s election wrong means that Trump has no reason to trust polling forecast or indeed ‘experts’ through his four years.
    The question is, did he win on an intuition of these forgotten class or did his data scientists see a different picture? It is definitely concerning if our numbers begin taking on liberal or democratic interpretations.

  • Tammy

    Dr. Sam,

    I’m pretty new to politics, only began paying attention in 2008 when President Obama was campaigning. After his 2012 win, I was drawn to David Axelrod, David Plouffe and Nate Silver’s methodology and prediction/forecast. During the final weeks of this current 2016 campaign, I saw an interview you did with one of the mainstream media outlets and from there, I googled you and viewed this site countless times. I respect you and your willingness to admit your error. I admire your position this morning while imploring Smerconish viewers to cultivate a deep interest in policy issues rather than titillating trival issues that completely derail our focus. I have now become a part of the community of citizens who view this site regularly for perspective.

  • Bill Miller

    Thanks again Prof Sam,
    Sorry about the bug. I have been thinking about the electoral college vs popular vote controversy. My first reaction was to suggest that a college team that looses the game by one goal might argue they really won by yards gained or passes completed? With more research I found http://www.nationalpopularvote.com/, so this could be the last electoral vs popular mismatch in 2016? …If the states pass the compact. I know if my vote weighed as much here in TX, I coulda stayed home to get out the vote instead of my many trips to Ohio, or PA, or NC, or Fla! (I’ve visited WI and MI too but never expected this!) It’s interesting that Newt even supports the idea. This means that this rule change might not necessarily benefit urban over rural as some have suggested, because the strategy of both parties would adapt to the new rules set.
    Unfaithful elector scenarios are interesting to contemplate, but HRC would need more of these than have appeared in the entire history of US prez elections! Faithless is not a desirable adjective these days.
    I’ve enjoyed your efforts here since 2004 when I thought Kerry would win. In Ohio in 2012 I was astounded when the results did not match what we were seeing at the Romney rallies. This time I wasn’t sure, but I was still surprised!

  • truedson

    The Garland comment was right on. I doubt it would happen but….interesting thought.

    • Arun

      It won’t. He has no reason to appoint Merrick Garland. George W Bush won in 2000 with a minority of the popular vote, but governed like he had a mandate. I expect Trump to do the same. There is nothing in his character that indicates he would do anything else.

    • Sam Wang

      Should I have just used the time to talk about poll aggregation some more? That seemed like a worse use of the time.

    • truedson

      Of course he has no reason…that wasn’t the point….but Trump is so unpredictable and unbeholden to the GOP that it sort of the thing he might do…..Bush was very predictable….and beholden to the party chiefs.

    • Arun

      Sam, my post wasn’t to tell you how you should have used your time on CNN. But I do think anyone assuming he will appoint Garland will be disappointed. I just don’t see it happening.

    • Bulgakovs Cat

      good grief
      poll aggregation cant detect undetectables.
      its like software– u can prove there are no detectable bugs in your program, but u cannot prove there are no undetectable bugs.
      time to move on.
      Bad Data.
      re: Garland
      If Trump doesnt repeal Obamacare –>
      http://www.opposingviews.com/i/society/donald-trump-will-consider-amending-obamacare-not-repealing-it
      And cant bring jobs back or destroy ISIS (impossibilities)
      what promises does he keep to his base?
      Trump promised his base that he would find a Scalia clone–
      thats like the theoretical minimum

    • Bulgakovs Cat

      in case u didnt get my point– if Trump repeals Obamacare thats 20 million people that will vote against him in 2020 plus the parents of 20.5 million college students who will be kicked off their parents insurance.
      ;)

    • Sam Wang

      Sure, if people make the connection, and if there is no other major issue to dominate the conversation.

    • Bulgakovs Cat

      its gunna be pretty obvi to parents who have to buy insurance for their collage age kids or to ppl kicked off their plans.
      not to mention painful.
      What were u thinking abt as a distraction Dr Wang?
      WWIII? Trisolarian invasion?

    • Bulgakovs Cat

      Maybe Obama cant appoint Garland– but he could pardon Snowden.
      Thats what he should do ;)

  • Rex

    You’re a man of your word. But this is very unsatisfying.

    Here’s to 2018

  • WDR

    You are a man of your word. Thank you for your analysis and what you continue to do. I didn’t fully appreciate your comment about the site being for resource allocation, not reassurance, before Tuesday.

    • Alan Cobo-Lewis

      Clinton campaign used Ada for resource allocation. But Ada was not robust to violations of its assumptions. I think silveresque smearing of the probability dist would have led to safet resource allocation (sub-optimal if assumptions were correct but not catastrophic “in the event of an actual emergency”, as the emergency broadcast system used to say)

  • Thomas K Barthel

    Professor Wang, I watched your pre-election predictions (along with 538, The Upshot, and PredictWise) on a daily basis. At first I took comfort in your predictions, as they showed a high probability of a result that I would like. However, as a fellow scientist, I became concerned with your seeming cockiness in your own predictions as you used phrases like “the paint in drying” and “the cake is baked”. I also found your dismissiveness of Nate Silver’s predictions more subjective than objective. In the end, I think your model failed to account for the uncertainty posed by the large percentage of undecided voters, ignored the spread in the polls, and suffered from overconfidence caused by treating each state as an independent variable. You need to do more than eat a bug. You need to eat crow on this one. Like most scientists, I’ve been there before. Better luck next time.

    • Bulgakovs Cat

      Sometimes the Cake really IS a Lie.
      Already bored by the postmortem– its isomorphic with Brexit– it was close, but polls didnt detect it.
      mirable dictu.
      Im interested in periodicity today…will the pendulum swing back in 2020 or will it take until 2024?

    • Josh

      If the first Trump Administration isn’t a success, I’m betting on the pendulum swinging back in 2020.

      -Demographic shifts continue to give Dems a bigger and bigger advantage in national races. No GOP candidate has ever won more than 62 million votes in a national election (Bush, 2004), while Obama received 69 million in 2008 and 66 million in 2012. For the right Dem candidate there are probably upwards of 70-75 million possible votes in 2020, while Trump will have difficulty breaking the 60 million vote barrier. Enormous upside potential…

      -HRC was an eminently qualified but supremely flawed candidate. If the Dems find someone with grassroots potential and fewer chinks in their armor, he or she should have a much easier time convincing sporadic, marginal and first-time voters to turn out.

      -How bad will Trump be? If his government is a redux of the Bush II years, I think he’s out in 4. If he turns out to be an idiosyncratic leader who’s willing to compromise with Dems even against the wishes of a favorable Congress, then who knows?

      Ultimately, I think it’ll come down to whether or not the economic fortunes of the country’s lower-middle to upper-middle class citizens–the people whose wealth sits between the 20th and 80th percentiles–changes for the better or worse. If there’s a recession (we’re probably overdue–it’s been 7 years since the last one), or if the rising tide of the economy doesn’t end up ultimately lifting all the boats in the Rust Belt, then Trump’s a goner. If, however, people do feel like their lives are improving, then it’ll be 2024…

    • Joeff

      This has probably been said, and I certainly don’t mean this personally, but the old maxim, “garbage in, garbage out” certainly seems to fit here. It appears that numerous polls of critical states missed in the same direction, by the same amount, for the same reasons. (see today’s NYT Upshot). All you were doing was taking the public poll results and running a lot of simulations. No one seemed to be seriously concerned about your methodology from a pure statistical standpoint. It all stemmed from systemic errors in polling.
      To quote another maxim, it’s a feature, not a bug.