Nate Silver: The “Iowa bounce” is not so bouncy

I don’t follow polls that much, and I’m always mystified at how many people seem to regard Nate Silver, who runs the famous FiveThirtyEight site, as some kind of prognosticating god. After all, even on the eve of the last election he predicted that Hillary Clinton had a more than 70% chance of winning the election. And that wasn’t the only time his predictions have been misleading and probably based on a bad survey. Nevertheless, since all of us are sitting on our thumbs waiting for the results from Iowa, Silver’s written a column saying that the lack of early results not only screws up his forecasts, but certainly helps some candidates at the expense of others. Click on the screenshot:

He first calculates that, based on the results of how a given state affectsthe results of polls in other places, Iowa has a net worth of 800 delegates—20 times its actual weight of 41 delegates. So if you win there, and thus win early, you get a bounce of 23% in the polls—far greater than that of any single state.

But the screwup in Iowa has taken the wind out of all the candidates’ sails as they must head for New Hampshire for the next primary vote, a state where fully half the Democratic electorate is undecided, and Bernie is tied almost neck and neck with Biden.

Silver:

Everything was a little weird in Iowa this year, however. And there were already some signs that the Iowa bounce — which essentially results from all the favorable media coverage that winning candidates get — might be smaller than normal. Iowa was bracketed by an extremely busy news calendar: President Trump’s impeachment trial both before and after the caucuses, the Super Bowl on Sunday, the State of the Union address on Tuesday. There was not the usual climactic uptick in media coverage around Iowa. From initial indications — to the extent any information at all is reliable at this point — Democratic turnout there wound up being fairly low.

But we weren’t prepared for what actually happened, which is that — as I’m writing this at 3:15 a.m. on Tuesday — the Iowa Democratic Party literally hasn’t released any results from its caucuses. I’m not going to predict what those numbers will eventually be, although early indications are that Bernie Sanders, Pete Buttigieg and perhaps Elizabeth Warren had good results. The point is that the lead story around the 2020 Iowa Democratic caucuses is now — and will forever be — the colossal shitshow around the failure to release results in a timely fashion.

There’s a lot of number-crunching, and the upshot is that Iowa had the potential to help Sanders much more than Biden (“Sanders’s chances of a majority [of final delegates at the Democratic convention] would have shot up to from 31 percent to 58 percent with an Iowa win, Warren’s from 5 percent to 32 percent, and Buttigieg’s from 4 percent to 22 percent”).

Actually, given that it’s still a week until New Hampshire residents vote, I’m surprised that a one-day delay in reporting the Iowa results could screw up the candidates’ standings so much, but that, at least, is what Silver thinks:

. . . The Iowa Democratic Party’s colossal screw-up in reporting results will potentially have direct effects on the outcome of the nomination process. The failure to report results will almost certainly help Biden, assuming that indications that he performed poorly in Iowa are correct, as they won’t get nearly as much media coverage. And they’ll hurt whichever candidate wins the state — most likely Sanders or Buttigieg. (Although if Sanders winds up finishing in second place or lower, he also might not mind a reduction in the importance of Iowa, especially with one of his best states, New Hampshire, coming up next.)

Furthermore, Iowa is typically a state that winnows the field. But with every candidate either having performed well there, potentially having an excuse for a disappointing finish there, or somewhere in between, it might not do that. Delaying the winnowing process would tangibly increase the chance of a contested convention.

It’s not a good situation for the Democratic Party. And it’s already too late for the damage to be entirely undone, even if Iowa eventually gets its act together.

Do we know that Biden didn’t do well in Iowa? How? More important, we shouldn’t care. Anybody with any brains knows that Iowa should not be able to have the disproportionate influence on the nomination process given its smallness and its homogeneity. Things need to change. But I doubt they will.

61 Comments

  1. Mike Anderson
    Posted February 4, 2020 at 1:41 pm | Permalink

    But the screwup in Iowa has taken the wind out of all the candidates’ sails

    As I’ve said: Bloomberg Wins Iowa

    • W.T. Effingham
      Posted February 4, 2020 at 4:06 pm | Permalink

      I’m bracing myself for: tRUMP loses popular vote by approximately nineteen million, wins Electoral College by 76.

  2. Mark R.
    Posted February 4, 2020 at 1:51 pm | Permalink

    This from the L.A. Times:

    After a technical breakdown that threw the Iowa caucuses into chaos, state Democratic Party officials plan to have 50% of the contest’s results later today.

    Release of those results is expected by 2 p.m. PST, according to officials from several campaigns who were briefed on the matter by party representatives.

    Why release just 50%? That seems to make a confusing situation more confusing. They should release 100% once they have them all tallied.

  3. Mike Anderson
    Posted February 4, 2020 at 1:59 pm | Permalink

    even on the eve of the last election he predicted that Hillary Clinton had a more than 70% chance of winning the election

    I don’t understand why you have an issue with this or consider it wrong.

    • Debbie
      Posted February 4, 2020 at 2:19 pm | Permalink

      Exactly. This is an erroneous statement. That’s not how probabilities work.

    • Posted February 4, 2020 at 2:21 pm | Permalink

      I don’t consider it wrong; I consider it to be misleading and probably based on unrepresentative information. I’ll change what I wrote to “misleading” to quiet the complainers. But the point is the same: a lot of people seem to worship Silver and I don’t think he deserves that kind of approbation.

      • Posted February 4, 2020 at 2:42 pm | Permalink

        Nate, of course, got famous by absolutely nailing the 2012 election. He parleyed that into his 538 site, to which I admit a mild addiction. But being a prognosticator is a tough business. The world has a way of proving your predictions wrong.

        • Simon Hayward
          Posted February 4, 2020 at 2:59 pm | Permalink

          I guess it depends on what you mean by “famous” but it was his earlier baseball predictions and the surrounding system that brought him to a lot of people’s notice. As an aside he also got 49/50 states correct in the 2008 election so it wasn’t a fluke.

          • Mike Anderson
            Posted February 4, 2020 at 3:23 pm | Permalink

            Also, weeks before the 2016 election, Silver pointed out the unusually high possibility of Trump losing the popular vote but winning the Electoral College.

      • Simon Hayward
        Posted February 4, 2020 at 3:17 pm | Permalink

        Jerry, if you look at the projection the night before the 2016 election, the model gave Trump a 28.6% chance of winning and also gave Clinton an 81.4% chance of winning the popular vote. So the outcome was well within the likely range of possibilities – in fact 538 was one of the few models that gave Trump any serious chance.

        https://projects.fivethirtyeight.com/2016-election-forecast/

        As the 538 folks say (a lot) these things are all probabilistic, so I couldn’t get even as far as “misleading” on this. Disappointing, demoralizing, disastrous, possibly but thats the outcome, not the prediction!

        BTW Silver is a UofC Econ grad if that’s worth anything to you.

      • CJColucci
        Posted February 4, 2020 at 3:26 pm | Permalink

        A 70% chance of winning means a 30% chance of losing. If we ran elections 10 or 100 times, those would be great odds. But we don’t. With an honest (American-style) roulette wheel, 21 has a 38-1 chance of coming up. Sounds pretty rare, but it comes up all the time, probably a couple of times a night at a given casino. In the 2016 election, a lot of fluky things broke Trump’s way and he won the electoral college, while handily losing the popular vote, because of fewer than 80,000 votes spread over three states. All of this is, or should be, well-understood. If calling Hillary a 70-30 favorite is “misleading,” it is misleading only to people who don’t know what it means to be a 70-30 favorite. That is, of course, a lot of people because a lot of people don’t know jack s**t, but that’s true of just about anything.

        • Posted February 4, 2020 at 3:42 pm | Permalink

          Thanks for telling me I’m one of those who know jack shit, which of course is a Roolz violation. Some people just can’t be civil in their comments, and of course the wages of sin is a tuchas hit by the door.

      • Posted February 5, 2020 at 7:04 am | Permalink

        It didn’t mislead me, although I agree that some people were misled by Silver’s numbers. I remember people saying “85%: it’s a sure fire thing” and me responding by pointing out that it is about the same chance as not blowing your head off in a game of Russian Roulette and would they try that.

        I don’t think Silver’s data was faulty either. He doesn’t do his own surveys but aggregates lots of polls weighted by their perceived quality (presumably judged by past performance, sample selection and methodology). What I do have a problem with is that, afterwards, he claimed the model worked pretty well, but given that we only have one real life datapoint to judge his model against, I’m not sure what his justification for claiming that is.

        • Posted February 5, 2020 at 9:43 am | Permalink

          I suspect that polls, Nate Silver’s included, are now simply running up against the inherent limitations of the polling process. They try to adjust for the fact that they can’t really randomly select voters to take the poll, make them report their feelings honestly, tell whether they will really show up to vote, or something won’t happen to change their mind between the poll and their voting.

      • Jay
        Posted February 5, 2020 at 7:20 pm | Permalink

        Nate has the most sophisticated election forecast model in the world (or if anyone has a better one, they’re keeping it a secret). When no other forecast model gave Trump more than a 5% chance of winning, Silver’s model predicted he would have a 30% chance. It is not all surprising that an event that is given a 30% chance occurring, occurs. The model that says coin flips are independent equally likely events gives the result TH of two consecutive coin flips only a 25% chance of occurring. Yet you would not be the least bit surprised if you flipped a coin twice and got that result, much less think the model was wrong.

        • Jay
          Posted February 5, 2020 at 7:27 pm | Permalink

          To expand a bit, if candidates given a 70% chance of winning won every time, then the model would be wrong, because empirically those candidates had close to 100% chance of winning.

          • tomh
            Posted February 5, 2020 at 11:38 pm | Permalink

            The problem is, with just a one-off trial, it’s difficult to prove empirically.

            • Jay
              Posted February 5, 2020 at 11:50 pm | Permalink

              What you do is compare the predictions of the model with the frequency of the actual outcomes over a large number of predictions. If, 10%, 20%, … , 90% of events that the model predicted would occur with 10%, 20%, … , 90% probability, respectively, then the model is well-calibrated.

            • meisnerman
              Posted February 7, 2020 at 2:16 am | Permalink

              Yes, if Nate only gave a probablalistic forcast for a single election it would be pretty much impossible to prove whether his projected odds were accurate. If you looked in isolation only at the Trump/Clinton result in the 2016 General election you might conclude that there’s is no value in Nate’s models. Luckily though, Nate predictions have given odds on hundreds of individual elections over the last several cycles. Given that nuymber of results it becomes possible to calibrate his accuracy (which is REALLY quite good.)

              Put another way, if Nate predicts only one race with 70-30 odds and the underdog wins, it might look like he’s a hack. But if he predicts TEN races with 70-30 odds then the underdog MUST win about 3 of those times or else he’s not doing it right.

              Humans are really bad at thinking probabalistically, but 538 keeps tilting at windmills hoping to explain that a 30 percent chance for Trump is NOT a certain loss. It’s actually a pretty decent shot! (Actually closer to 50% than to 0%!)

  4. Posted February 4, 2020 at 2:12 pm | Permalink

    There were already those “mysterious” circumstances last time that worked against Sanders. Every time he seems to get closer, there seem to be some more “mysterious” circumstance. It really does look like “Democrats” name may be a misnomer, as it seems they’re not honouring a democratic process they pretend to have.

    The Nathan Robinson comments in the Guardian:

    “But Biden caught a lucky break. With the party not releasing the actual result, his campaign sent a letter demanding that the result be suppressed until such time as the “quality control issues” were resolved. If it takes long enough to get the official count, Biden may hope that Iowa is old news, or that the issues surrounding the caucus are discussed far more than the actual result.”

    Lucky break! What a remarkable coincidence, again. He wrote “If you’re a Sanders supporter, you have reason to be suspicious.” But why only Sanders supporters? Isn’t anyone in a democratic process presumably interested in a fair game?

    “For Sanders supporters, being denied a rightful victory in Iowa gives feelings of déjà vu. In 2016, Sanders may well have won Iowa, possibly by a lot, but the state party did not release the vote totals. Instead, it only released delegate numbers, which showed Bernie narrowly losing the state “701-697” to Hillary Clinton.”

    All of these shenanigans get Trump re-elected in the end.

    • Posted February 4, 2020 at 3:39 pm | Permalink

      Sigh. At least the superdelegate b.s. has been reduced to nearly harmless levels.

    • Saul Sorrell-Till
      Posted February 5, 2020 at 9:05 am | Permalink

      Do you really believe that this colossal fuck-up was intentional? It damages the entire party, all of the candidates, and conceivably might tip the election towards the GOP if it’s close.

      If Trump, Putin, etc. could have intervened*, they would have done exactly this.

      *Assuming they didn’t.

      • Posted February 5, 2020 at 9:50 am | Permalink

        Sounds like it was completely an own goal affair. Nevada was evidently planning to use the same app but now have announced they’re dropping it.

        It’s quite possible they are unfairly throwing the software company that made the app under the bus. On the other hand, even if the app itself wasn’t at fault, they should have worked harder to make it a success. On yet another hand, a small software company can do only so much.

  5. allison
    Posted February 4, 2020 at 2:19 pm | Permalink

    I would disagree that Silver was “wrong” in 2016 when he gave Clinton a 70% chance of winning; he was actually closer than nearly all of the other prognosticators. I believe the Princeton Election Consortium had Clinton at “99%+” likely to win. (If Silver had said “Clinton will win”, you could say that he was wrong.)

    If I say that rolling a die is only 16.7% likely to generate a six, and a six is rolled, the prediction was not “wrong”. Unlikely events happen.

    • Posted February 4, 2020 at 2:24 pm | Permalink

      Yes, yes, I wrote too quickly and have explained what I meant above. However, the fact that everybody else got it MORE WRONG doesn’t mean that we should have confidence in his results. Which a lot of people, by the way, do.

      • meisnerman
        Posted February 7, 2020 at 2:29 am | Permalink

        I would agree perhaps that many people were misled by Nate’s prediction, but that’s not exactly the same as his prediction being misleading. I’d argue that calling them “misunderstood” is more accurate. Silver took great pains to explain (over and over to anyone who wouuld listen) that his odds for the race predicted a very real chance for a Trump victory. Right up until the results became apparent, 538 was the target of derision by almost every other consolidater for rating Trumps chances way too high.

        It is true that people have unwarranted confidence when they see something like a 70% chance for Clinton, but that has more to do with the way our brains think try to create shorthand for probablities than with any shortcoming with Silver’s work.

        As a fan of Silver’s and a longtime fan of your own work, Jerry, I rather suspect that you might very much appreciate Silver’s rational approach to Poll consolidation and his rigorous insistence on following the data – should you ever be inclined to take a deeper dive into his work.

    • Posted February 4, 2020 at 2:29 pm | Permalink

      If memory serves, I believe the much loved owner of a certain website was offering favorable odds to anyone who was willing to bet that Clinton would lose. Fortunately for him, nearly all of us were certain she would win, and missed out on easy money.

      • Saul Sorrell-Till
        Posted February 4, 2020 at 2:57 pm | Permalink

        FWIW, in the comments for PCC’s infamous pre-2016-election ‘TRUMP HAS LOST’ article, I wrote a hefty post explaining why I thought Trump was going to win(and was pounced upon – sort-of understandably – for saying so).

        Unfortunately I didn’t have much money at the time so I didn’t take up the available bet.

        …I’m rather hoping PCC writes a ‘TRUMP HAS WON’ article in the days before this election, and work his reverse-magic again, this time on the microdextrous berk in the WH. There would be a nice symmetry to it.

        • Simon Hayward
          Posted February 4, 2020 at 3:22 pm | Permalink

          If I thought that would work I’d endorse the idea heartily! Unfortunately the lightning may strike from the other side this time ;(

        • Posted February 4, 2020 at 5:39 pm | Permalink

          I don’t think anyone is going assume a Democratic win like we did in 2016. In fact, the way the Dems are doing right now, I’m thinking Trump will win.

          • Posted February 4, 2020 at 7:00 pm | Permalink

            In one week, the Democrats make a complete fool of themselves, Trump gets to make a triumphant SOTU speech, and Trump gets acquitted. A perfect Trifecta.

    • Bruce Cochrane
      Posted February 4, 2020 at 2:34 pm | Permalink

      I totally agree with Allison – in fact Silver came very close in his prediction of the popular vote in 2016. What did him (and all of us) in was a ~70,000 vote margin in three states (Wisconsin, Michigan and Pennsylvania), which no polls could possibly call.

      Going further, while some of Silver’s (and other’s) prognostications about the impact of this mess may be speculative, there’s no question that it’s a setback for the Democrats. The entire premise of the campaign was that Iowa would somehow provide some clarity. It has not. even when results are announced (today? tomorrow? Who knows?), their credibility has been seriously damaged.

      Finally, I’m more optimistic than PCC-E that there will be change in the future, at least if the Democratic Party wants a nomination process that actually produces winning candidates. Carter and Obama were able to build momentum in Iowa that took them to the White House, but very few others have done so. Perhaps now that will sink in.

    • Posted February 4, 2020 at 2:40 pm | Permalink

      “However, the fact that everybody else got it MORE WRONG doesn’t mean that we should have confidence in his results.” Silver may not have gotten it “wrong” at all, and his prediction may not have been “misleading” at all. Maybe – apart from the “we have no free will and everything that will be is preordained” perspective :-> – Clinton really did have a probability of winning of 70%. And then the dice got rolled, and Trump won. It’s fine not to like Silver, or to think his abilities are overestimated, but the fact that he predicted a 70% chance of winning for Clinton and then Trump won is not any kind of evidence at all that he was “wrong” or even “misleading”. You’ve walked your initial wording back a bit, but I think you need to walk it back more.

      • Posted February 4, 2020 at 3:49 pm | Permalink

        Sorry, but please don’t tell me what to do. Yes, I understand the commenters and agree with them, and perhaps my rancor, and my unfortunate choice or words is due to my disappointment with how the election went.

        That said, you’re being extraordinarily patronizing to tell me that I need to “walk my wording back more.” Would you like to rewrite that sentence for me?

        Jebus. You once characterized another commenter as “rude”. Look in the mirror.

  6. Posted February 4, 2020 at 2:23 pm | Permalink

    It is perverse that, thanks to the insatiable appetite of the media, a scrap of information coming from a small, non-representative state is given such grossly inflated importance, simply because Iowa goes first. Why should the order of primaries even matter in a rational world? Obviously, a herd effect. Why do not states all agree on a national primary day and end this nonsense?

    • Mark R.
      Posted February 4, 2020 at 2:51 pm | Permalink

      Agree…and why can’t they at least do some kind of “Super Tuesday” for the 1st primary…do Iowa and 4 other more representative states on the same day.

      I also don’t think a caucus allows for greater participation in a primary.

  7. Posted February 4, 2020 at 2:37 pm | Permalink

    This is the absolute last thing the DNC need.

    They were already caught with their thumbs on the scale last time. This is already going to be a tough race, you can’t have shenanigans going on, or even suspected right now.

    • Mark R.
      Posted February 4, 2020 at 2:56 pm | Permalink

      There weren’t any shenanigans as far as I have heard/read. All the votes are there, they just had a problem counting them in a timely manner. I also don’t think any Iowans who participated in the caucus thought it was unfair.

      But I agree, it is not what the DNC needed.

  8. GBJames
    Posted February 4, 2020 at 2:38 pm | Permalink

    sub

  9. Randall Schenck
    Posted February 4, 2020 at 2:42 pm | Permalink

    It is fitting that a simple new system of sending in results via the phone instead of just phoning it in would cause all the confusion. The media and all the pundits should all go spend a few full years in Iowa to understand how ridiculous all of this is. Of course they are not going to do that because, who the hell lives there anyway?

  10. Posted February 4, 2020 at 2:53 pm | Permalink

    I think Nate Silver’s fame is based on his engagement with the community and open science approach to polling. He is up-front about it being an inexact science and constantly seeks to improve things. When his predictions are wrong, he figures out why (a theory at least) and shares it with others. I don’t know if that fame is unfair but that’s my impression of why he has so much mindshare.

  11. Ken Kukec
    Posted February 4, 2020 at 3:21 pm | Permalink

    I’m always mystified at how many people seem to regard Nate Silver, who runs the famous FiveThirtyEight site, as some kind of prognosticating god. After all, even on the eve of the last election he predicted that Hillary Clinton had a more than 70% chance of winning the election.

    Hell, even the “Murderers’ Row” Yankees of Ruth, Gehrig, et al. in the 1920s didn’t win every World Series.

    FiveThirtyEight was less wrong about the 2016 presidential election than the rest, and Nate Silver personally wrote at some length before the election cautioning that Trump had a potential path to victory. He also had the best run of anyone ever in accurately predicting elections from 2008 through 2014 — a run that included two presidential, two midterms, and several special elections.

    FiveThirtyEight‘s tracking poll still represents the gold standard in politics.

  12. Ken Kukec
    Posted February 4, 2020 at 3:43 pm | Permalink

    The Dems need to revamp how they go about picking a presidential nominee — and perhaps the boondoggle in Iowa will serve to spur that change. Opening day for Iowa’s new delegate selection system last night was a bomb on the magnitude of Ishtar or Heaven’s Gate.

    Iowa has also had its problems on the GOP side. I recall when Mitt Romney initially “won” the 2012 caucuses, only to have Rick Santorum declared the victor a week later, after the New Hampshire primary, thereby robbing Santorum of any chance of “Iowa bump” (a tough break that couldn’t have happened to a more deserving fellow).

  13. Eric Shumard
    Posted February 4, 2020 at 3:44 pm | Permalink

    Nate Silver’s statement that Trump had a 30% probability of winning in 2016 is not misleading although it is apparently misunderstood. A more complete statement is that within the set of all possible universes that are consistent with the polling results (including their uncertainties), the subset of universes where Trump wins was 30% of the entire set. Silver’s methodology is to run a large number of Monte Carlo computer simulations to estimate the fraction of polling consistent universes that have a particular outcome. While adulation might be over the top, Nate Silver knows what he is doing.

  14. Posted February 4, 2020 at 3:45 pm | Permalink

    Nate might have underestimated the Comey chill on Clinton’s chances, as his infamous announcement occurred 11 days before the election. But how was he to know? I remember railing againstComey when he did this, but what do I know? Anyway, I’m certain that tRump was hoping to pull a ‘Comey-esque’ announcement on Biden, using the Ukraine prez as proxy.

    • Mike Anderson
      Posted February 4, 2020 at 4:31 pm | Permalink

      Nate might have underestimated the Comey chill on Clinton’s chances

      Nate’s numbers come from polling data, not current events. After the election he analyzed the polling data and concluded it was Comey’s last minute announcement that tipped the scales to Trump.

      • Ken Kukec
        Posted February 4, 2020 at 5:05 pm | Permalink

        As I recall, throughout the 2016 campaign, FiveThirtyEight gave three versions of it tracking poll — “polls only,” “polls plus,” and a third set that included 538’s proprietary “special sauce” (which took into account things like economic indicators). As the campaign wore on, those numbers converged.

  15. Posted February 4, 2020 at 4:27 pm | Permalink

    I believe each state sets the time of its primary. The date may have to be approved by the DNC or the RBC. I don’t think either national committee can tell each state when its primary has to be held.

    A one day national primary held on the first Tuesday in June would be one way to do it, but then we would not have the process of slowly weeding out some candidates who don’t do well early. Our present process lets the last state primaries chose between the remaining candidates but lets the early states weed out the ones they don’t like. Not sure whether the early ir late states gave more power.

    Each process has its pro and cons.

    • Mike Anderson
      Posted February 4, 2020 at 4:36 pm | Permalink

      A one day national primary held on the first Tuesday in June would be one way to do it

      I agree, although I suspect both the DNC and the RNC want to preserve this slow motion style of primary election, that they might be able to steer if it goes too far off an expected track.

    • Posted February 5, 2020 at 7:16 am | Permalink

      But what if a poor showing in Iowa (41 delegates) or New Hampshire (26 delegates) causes the candidate that New York (274 delegates) would have voted for to drop out of the race?

      It seems to me that the ideal way to do it would be to have all the primaries on the same day and use a single transferable vote system.

      • Posted February 5, 2020 at 9:47 am | Permalink

        How about just a national primary without delegates? That would draw attention to the unfairness of the electoral college and perhaps nudge us towards eliminating it. They can still have their party convention, though I suppose it will take some of the drama out of it.

  16. Posted February 4, 2020 at 4:36 pm | Permalink

    Political parties need to do polling, and the for the public it’s probably valuable to get the occasional indication of which why the winds are blowing, just to reduce anxiety or activate emotional coping mechanisms.

    Apart from that, they wind up pulling the public into debates about things political parties used sort out behind closed doors, (which candidates presents themselves the best, how much will this faux pas affect their electability, etc); and pulls the public into partisan debates about which demographic sector will vote for which candidate — in a manner that would be considered racist, sexist, ageist, bigoted, selfish, misanthropic and stupid in any other context.

    And in the case of Trump, polls that (rightly) showed him at more than 40% gifted him a legitimacy as a candidate than he would have gained by getting him to talk in detail about policy.

    And asking people what they think, and then reporting it back to them should not count as news. And certainly not important news.

  17. Posted February 4, 2020 at 4:53 pm | Permalink

    I would like to see something done to shorten the selection process. Prohibiting primaries from being held prior to June or July of the election year I believe would help.

    • tomh
      Posted February 4, 2020 at 5:03 pm | Permalink

      I don’t see how that would help. Candidates would be much slower to drop out and we would have crowded debates even longer.

    • Torbjörn Larsson
      Posted February 4, 2020 at 5:16 pm | Permalink

      The newly published model (see my comment below) implies prohibiting primaries entirely will help, it would make voting more of a stable median voting system.

      Another recommendation would be to do like Australia and make voting mandatory, it would remove low voter turnout which is a factor that enable the instability of negative representation.

      The third factor of polarization is not easy to do something about, especially in a two-party system. But maybe taking care of the two other factors would stabilize elections again.

      • Mike Anderson
        Posted February 4, 2020 at 5:52 pm | Permalink

        Another recommendation would be to do like Australia and make voting mandatory

        That would be hard to achieve here because the Republican party would be negatively impacted by high voter turnout, so they fight to make voting as difficult as possible.

  18. Torbjörn Larsson
    Posted February 4, 2020 at 5:06 pm | Permalink

    I don’t think Silver does bad statistics. It is rather likely, as I commented on earlier today [ https://whyevolutionistrue.wordpress.com/2020/02/03/cant-we-please-get-rid-of-political-polls-and-our-own-political-poll/#comment-1807294 ], that US – due to factors such as a primary election system, strong polarization and low voter turnout – has entered a chaotic regime. No matter how well Silver polls opinion, the outcome will behave stochastic:

    Adapted by SciTechDaily; R^2 = 0.86

    Things need to change. But I doubt they will.

    From Ars commentary:

    The model predicts that, in the presence of instability, the amount of polarization will increase in a characteristic way. The real-world data supports that contention.

    In short, no matter what your political opinion, prepare to be overstressed and depressed by election results for the foreseeable future.

    • Torbjörn Larsson
      Posted February 4, 2020 at 5:17 pm | Permalink

      [Links to the research and commentary in my comment.]

  19. Posted February 7, 2020 at 11:36 am | Permalink

    I’ve looked very briefly at what has been claimed by Motherboard to be the application used to send in votes from the caucus members. What a piece of overengineered nonsense. (Does one *really* need to display icons at native speed on a vote submitter, for example?) Assuming they have the real deal, I am not surprised at all that there was trouble.


%d bloggers like this: