ALL BLOG POSTS AND COMMENTS COPYRIGHT (C) 2003-2016 VOX DAY. ALL RIGHTS RESERVED. REPRODUCTION WITHOUT WRITTEN PERMISSION IS EXPRESSLY PROHIBITED.

Sunday, November 13, 2016

Even his hindsight isn't 20/20

Nate Silver takes great pride in being less completely wrong than some of the other pollsters, in an article entitled "Why FiveThirtyEight Gave Trump A Better Chance Than Almost Anyone Else (Except the LA TIMES/USC and IBD/TIPP Tracking, Who, Unlike Us, Actually Got It Right). NB: I added the bit in parentheses. At no point does Silver mention any polling organization, or individual, who did correctly predict the election results.
Based on what most of us would have thought possible a year or two ago, the election of Donald Trump was one of the most shocking events in American political history. But it shouldn’t have been that much of a surprise based on the polls — at least if you were reading FiveThirtyEight. Given the historical accuracy of polling and where each candidate’s support was distributed, the polls showed a race that was both fairly close and highly uncertain.

This isn’t just a case of hindsight bias. It’s tricky to decide what tone to take in an article like this one — after all, we had Hillary Clinton favored. But one of the reasons to build a model — perhaps the most important reason — is to measure uncertainty and to account for risk. If polling were perfect, you wouldn’t need to do this. And we took weeks of abuse from people who thought we overrated Trump’s chances. For most of the presidential campaign, FiveThirtyEight’s forecast gave Trump much better odds than other polling-based models. Our final forecast, issued early Tuesday evening, had Trump with a 29 percent chance of winning the Electoral College.1 By comparison, other models tracked by The New York Times put Trump’s odds at: 15 percent, 8 percent, 2 percent and less than 1 percent. And betting markets put Trump’s chances at just 18 percent at midnight on Tuesday, when Dixville Notch, New Hampshire, cast its votes.

So why did our model — using basically the same data as everyone else — show such a different result? We’ve covered this question before, but it’s interesting to do so in light of the actual election results. We think the outcome — and particularly the fact that Trump won the Electoral College while losing the popular vote — validates important features of our approach.
Translation:
  1. I'm a Gamma and I can't admit that I'm wrong without explaining how being wrong only proves that I was right to do what I did. 
  2. Almost anyone else means anyone not KellyAnne Conway, Scott Adams, Nassim Taleb, Mike Cernovich, Vox Day, LA Times, IBD, or TPP Tracking.
  3. A 29 percent chance of winning is practically a near certainty. I mean, sure, you might have interpreted that to mean that Hillary was probably going to win, but that just shows how you don't understand polling as well as I do. The fact of the matter is that we were closer to getting it right than everyone else who didn't get it right.
  4. And by "29 percent", I of course mean 28.6 percent.
  5. And by "such a different result" what I mean is "exactly the same result as everyone else, except those other guys who actually got it right and whom I will carefully refrain from mentioning."
We strongly disagree with the idea that there was a massive polling error. Instead, there was a modest polling error, well in line with historical polling errors, but even a modest error was enough to provide for plenty of paths to victory for Trump. We think people should have been better prepared for it. There was widespread complacency about Clinton’s chances in a way that wasn’t justified by a careful analysis of the data and the uncertainties surrounding it.
Translation:
  1. We strongly disagree with the idea that I could have been wrong. The Secret King is never wrong, by definition! You just don't understand how the appearance of being wrong only shows that I was mostly right, and that just goes to show how much smarter I am than you. Still undefeated!
  2. Next time, don't pay any attention to what I say before the election. Just wait until it is over, and then I'll explain what I meant and how that proves I am right. Always.
Remember, at one point, Nate Silver and 538 gave Hillary Clinton an 87.3 percent chance of victory as recently as October 19. The good news for Trump, the Alt-Right, and even the Republicans is that these hapless morons are too proud to admit or learn from their mistakes, which means they are going to screw up just as badly, or perhaps even worse, in future elections.

Labels: ,

163 Comments:

Blogger Stg58/Animal Mother November 13, 2016 3:18 AM  

When will Nate Silver dry up and blow away?

Rhetorical question, I know.

Anonymous Ron November 13, 2016 3:23 AM  

LOL

Blogger S. Thermite November 13, 2016 3:24 AM  

Earlier tonight, Milo posted to Facebook a photo of Hillary signing an early "Madam President" Newsweek cover.

I'd like to thank Mr. Silver for his contribution to the Hubris Clinton Campaign and the resulting lolz at their expense.

Anonymous Chris M. November 13, 2016 3:25 AM  

After being as wrong as he was this year, he should change his name to Nate Participation Trophy.

Blogger al November 13, 2016 3:27 AM  

We need to start taking corporate scalps. I would propose the CEO of Pepsi be targeted by not only boycotting their products, but by arranging a boycott of all restaurants that carry Pepsi instead of Coke. It's an easy boycott to do since there are many fewer Pepsi restaurants than Coke anyway. So is it viable to use the alt-right's social media presence to boycott Taco Bell, KFC, Pizza Hut, and Arby's just to put the CEO's head on a spike as a warning to others?

Blogger CaseyD November 13, 2016 3:29 AM  

So is not being able to admit you're wrong a strong gamma tell?

Blogger bob kek mando ( I are Spartacus ... and you can too! C'mon, give it a try, these crosses are way more comfy than they look ) November 13, 2016 3:45 AM  

5. al November 13, 2016 3:27 AM
It's an easy boycott to do since there are many fewer Pepsi restaurants than Coke anyway.



sonofabitch.

BWs is a Pepsi chain.

in other news, Nate Silver seems to be the male version of Louise Mensch.

you can see why she miscegenation with a Jew.

Blogger dh November 13, 2016 3:55 AM  

I have spent some time figuring out where I went wrong. I am coming around to the idea that the national polling was within MoE, but that most state-based polls were well outside of their state margin of errors.

1. The national polling was pretty tight, with relatively small samples. I stopped following the national polls closely after the 2nd convention, because I don't believe they have much utility. In this election, if I had of paid more attention to the polling data on the national level I would have noticed that the state polls and national polls had a deviation higher than any US election in recent memory. That should have been a big indicator that something was truly wrong with state-based polls.


2. The main aggregator I spent time following was not 538, but Princeton Electoral Consortium. After looking at all the models out there, I liked his for three reasons:

a. It boiled down the state races to a single indicator of how much the national polls would have to move to make the race a tie, by averaging all state-based polls. This was then used to run monte carlo simulations of all possible outcomes and assign each one a probability. This should be highly accurate based on what we know about probabilities, and the only variable that goes into is the error rate of individual polls, which should be more accurate after averaging.

b. The source code was available, and it had no major errors I could (and the operator of the site accepted several of my patches that fixed slightly computational and coding errors).

c. The model was developed in conjunction with a detailed analysis of the last several modern elections and has a sound statistical basis.

In the end, PEC put the Clinton percentage at >99% on election day. It was wrong. I was wrong.


3. State-based polling most likely has serious quality control problems, and it's not without possibility that there was actual fraud in state based polling. In several swing states the aggregation that everyone used was based on as few as 5 polls. A single or two polls done fraudulently or incompetently could easily make the state appear stronger than it was.

4. I think it was Stilicho or maybe Dixon who was very close to exactly correct about turnout models. The sheer number of people who voted in 2008 and 2012 was staggering. 2008 saw a total of 132 million voters. In 2012 it was around 130 million. 2014 mid-terms was 83 million. We are looking like ending up in the ballpark of 122 million votes this time around. Literally 8+ million people who voted in 2008 and 2012 simply stayed home.

The staggering volume of voters who did not come home to vote is mind-blowing. I was fully expecting to get to 132 or even 134 million again, giving that between 2008 and 2016 the country added roughly 18 million new eligible voters.

Given this, I have to push out my projection that Democrats have a strong chance of a long-term black-latino-white woman coalition that could deny the White House to the GOP long-term. There is still hope.

Blogger OGRE November 13, 2016 3:59 AM  

An actual random sample of voters would provide a better result than their 'models.' Everyone seemed to get the raw data right, it was just how they put it together that screwed up the results. So why even use a model? Just up the sample x3 and go with the raw result; I'm sure it would more closely tie in to reality.

Obviously the reason for the models is to maintain the ability to manipulate the data to achieve the desired result .

Nate's just posturing here in the small hope that he might be relevant in the next go round.

Anonymous Eric the Red November 13, 2016 4:00 AM  

Polling expert Bill Mitchell:

Not only was there oversampling of Democrats in the polls, but there was undersampling of the Monster vote...

https://www.youtube.com/watch?v=dpaiuXNEWUE

Blogger Franz Lionheart November 13, 2016 4:06 AM  

And we took weeks of abuse from people who thought we overrated Trump’s chances.

So they caved in to the establishment, but proudly caved in a little bit less than the others.

"Yes I accepted the post in Hitlers government, but you see, I did it to observe and amend from within, where I could influence so much more and temper the worst. Anybody else would have been just so much worse."

Anonymous dagwood November 13, 2016 4:06 AM  

"coalition that could deny the White House to the GOP long-term."

Long-term, there will be neither a GOP nor a White House. This election is a speed bump, delaying the inevitable. There will ultimately be partition along racial lines, and there will be a separate white ethno-state. The really funny thing will be seeing all the (((various))) anti-whites banging on the fences trying to get in, to once more experience the joys of raaaacism and white privilege. Oops, sorry guys, no: now you run along and enjoy your diversity.

It'll make The Walking Dead look like a cotillion.

Que viva Aztlan!

Anonymous SaltHarvest November 13, 2016 4:07 AM  

Now if only these organizations could appear to underpoll California secession...

Speaking of which, #Calexit is getting some buzz right now.

Blogger al November 13, 2016 4:11 AM  

I believe that Clinton might have won a couple Rust Belt states if she hadn't been so associated with gun control which is one of the best motivators for rural white turnout. Also, lol at them proposing a black muslim for DNC Chair.

Blogger Cail Corishev November 13, 2016 4:11 AM  

What a load of obfuscatory bullshit. I'd heard this guy's reputation but hadn't paid attention to him before, so all election I kept thinking, "This is it? This is all he does?" Here's the Nate Silver Strategy for Winning:

* Guess one election right, to get a reputation.
* Let other people do the polling.
* Average their results and guess at odds between 0% and 100%.
* Claim that this is complicated math, because people don't grok math anymore, so they respect it as a dark art.
* Make lots of predictions, sometimes multiple ones daily, so no one will remember a particular one.
* As soon as results are in, get out in front of the explaining cycle, so you can be the one explaining how the others were wrong.

Good grief, that doesn't require statistics, political acumen, or anything except some talent for spinning BS fast enough that people can't tell you're not saying anything.

On election night, when things were close and signs started to appear that it was going Trump's way, he released a snake graph of the states purporting to show the electoral votes in a useful way. It might as well have been captioned: "Look at this cool graph! Isn't it weird, like a snake? Only real mathemagicians can make that! Do not look at all my incorrect predictions of the past year! Look at the snake graph! Look at it!"

Blogger OGRE November 13, 2016 4:17 AM  

@8 dh. re: the black, hispanic, white woman coalition.

black population has been on a slight decline as part of the overall population. Hispanics are the fastest growing demo by percentage. White women voted in favor of Trump as a group.

And while the Dems have been using this 'coalition' thus far, it won't last. The Dems have increasingly been pushing for a fringe-left, social justice position that is not inherently favored by blacks or hispanics. They've simply been able to construct this narrative that Repubs are the KKK and want to deport all latinos to Mexico. Maintaining that narrative has become increasingly more difficult given the rise of the internet and the collapse in trust of the MSM, and both of those factors are likely to become more amplified.

Blacks and Hispanics are natural enemies, not allies; and neither are really in line with the policies of the social justice left. While blacks and hispanics are all for voting themselves food stamps and health insurance, they do not look favorable on the trans/homo/pedo perverts that have been setting the Dem agenda and is predominately supported by females and gays. Eventually that will come to a head. Identity politics will result in a destructive fracturing of this coalition.

Blogger wreckage November 13, 2016 4:30 AM  

Plenty of people within the Punditsphere called it correctly, Nate, you puke. The WaPo:
https://www.washingtonpost.com/news/the-fix/wp/2016/10/28/professor-whos-predicted-30-years-of-presidential-elections-correctly-is-doubling-down-on-a-trump-win/

An established think-tanker:

http://legalinsurrection.com/2016/11/research-guru-saw-trump-phenomenon-coming-before-anyone-else/

Complex mathematical models frequently give results no better than either a simple regression, or a guess from a good guesser.

Modelling, by and large, is a moderately useful tool that has gained an authority and respectability far in excess of its usefulness.

Blogger wreckage November 13, 2016 4:38 AM  

@15 the Left simply cannot give up on the notion that they have a biologically determinate constituency that is incapable of voting any way but Dem.

Even accounting for fraud, Pinochet's rule was voted in by roughly two-thirds of Chileans.

Presumably they were all non-hispanic whites?

Anonymous JAG November 13, 2016 4:41 AM  

I'm sensing some strong gamma tells coming from dh who still refuses to accept that the vast majority of the polls were manufactured propaganda that had no bearing on reality.

Most of those polls reflected liberal fantasy that they wished would come true, IMO.

Blogger Bernard Korzeniewicz November 13, 2016 4:57 AM  

IMHO you Americans have 4 years tops to prepare for the Machine counteroffensive.
Next time it will not be the old delusioned hag, but nice, intelligent 40 with nicer boobs. And you lose. The System people (not the NYT sheeple, but their wolf leaders) are evil but not stupid.
President

Blogger DEPLORABLILITY November 13, 2016 4:58 AM  

Maybe they let NASA do their math they can't seem to get anything right either.

Blogger wreckage November 13, 2016 5:07 AM  

I boldly predict that the Left are currently, socially/structurally incapable of promoting a hot, smart woman to that much authority.

They just can't do it organically. It would have to be imposed by a strong faction within the party. I suppose that leaves it open for the Clintons to have a go by proxy?

Anonymous Steve November 13, 2016 5:10 AM  

dh - It boiled down the state races to a single indicator of how much the national polls would have to move to make the race a tie, by averaging all state-based polls. This was then used to run monte carlo simulations of all possible outcomes and assign each one a probability. This should be highly accurate based on what we know about probabilities, and the only variable that goes into is the error rate of individual polls, which should be more accurate after averaging.

That's cool, but I reckon the main flaw in all this polling geekery isn't so much the number of jigawatts going into the flux capacitor or the valency of methamphetamine hydrochloride molecules in the Heisenberg compensator.

It's this: voters are human beans, with all that implies.

And we can broadly and roughly predict what people will do, but it's more art than econometrics. Which is fine, because politics isn't science, it's more like sales or seduction.

What we do know is this:

* People vote for people.

Trump was the most personable candidate. That alone should've given the pollsters pause, even before considering that after 8 years of Democratic presidency they'd struggle to muster enthusiasm for more of the same.

If our models predict that the sick, horrible old woman whom nobody likes is going to mop the floor with the charismatic celebrity, maybe our models are wank?, said no pollster ever.

Nah, they were With Her and willing her into the White House. Too emotionally invested to notice data points such as holy fuck, Trump draws yuge crowds everywhere he goes! and bloody hell, Hillary is campaigning against a cartoon frog now! That doesn't look like winning..

Poor Nate Silver. I predict his next gig will be reading tarot cards in a flea market.

Blogger al November 13, 2016 5:11 AM  

I'm not sure about that Bernard, they already have Elizabeth Warren's amazing tits. Too bad she's covered in small pox scars from the infected blankets we perfidiously provided her proud people.

Blogger Noah B aka Carlos Danger Close November 13, 2016 5:13 AM  

Nate Silver might want to read a copy of the New York Times, which has already acknowledged its anti-Trump bias and promised its readers it will get its act together Real Soon Now. But of course they haven't fired anyone over it, so they can't be very serious.

Blogger Franz Lionheart November 13, 2016 5:16 AM  

incapable of promoting a hot, smart woman to that much authority. They just can't do it organically. It would have to be imposed within the party. I suppose that leaves it open for the Clintons

Hot smart daft ugly duckling.

Blogger Franz Lionheart November 13, 2016 5:21 AM  

An even "prettier" picture.

Anonymous Tutor November 13, 2016 5:22 AM  

I continue to be confused by this feigning of ignorance about Trumps popularity. Considering emails came out where it was openly discussed about creating bullshit polls. Wouldn't this alone disqualify anyone from treating the polls as though they were meaningful?

I just assumed everyone was pretending to be surprised by the polls being wrong. But now I am starting to fear that they are actually surprised. Shouldn't the people who make up the propaganda, not then actually believe it to be true? Since they understand it to be a fabrication.

Blogger OGRE November 13, 2016 5:22 AM  

Theres been some posts on Twitter campaigning for Ivanka Trump to be elected as our first female president.

Now thats next level trolling right there

Anonymous John Scalzi Unfinished Asimov Project November 13, 2016 5:22 AM  

"Well, Mr. Scardin, it looks like you're making a bigger fool of yourself than I thought."

"I have nothing to say to you," stoically replied the Secret Mayor of Transminal, "you--you filthy misogynist racist cisnormed rhetorically challenged douchecanoe."

"Well well," replied the fiend, "it looks like you're gonna need more than one lawn treatment. And you're gonna GET more than one lawn treatment."

"You don't worry about me, VoxGaytyz! You don't worry about me! I'M JAHN HARDIN SCALZ--er, SCARDIN! I'M NOT SOME CHEAP VANITY PRESS SELF-PUBLISHED RACIST, TRYING TO ESCAPE THE CONSEQUENCES OF HIS DUMPTH!

I'M GONNA SEND YOU TO TOR TOR, GAYTEZY!! TOR TOR!! TOR!!! TOR!!!"

Blogger MATT November 13, 2016 5:29 AM  

Vox, what sort of rhetorical response do you suggest in order to show someone how ridiculous their anti-Trump stance is? I know you say it's damn near impossible to teach the virtually unteachable, but I'm talking about my brother.

Have you had to deal with cognitive dissonance in your siblings, or are they on the same page on most issues?

Blogger kh123 November 13, 2016 5:30 AM  

"...voters are human beans"

In California anyhow.

Anonymous 5343 Kinds of Deplorable November 13, 2016 5:44 AM  

We think people should have been better prepared for it. There was widespread complacency about Clinton’s chances in a way that wasn’t justified by a careful analysis of the data and the uncertainties surrounding it.

This is hilarious. People should have been "better prepared" for it? Precisely: by rejecting both Silver's numbers and analysis.

Strangely, this is what everyone who got it right did.

Blogger VD November 13, 2016 5:45 AM  

So is not being able to admit you're wrong a strong gamma tell?

Absolutely. Even stronger is attempting to spin being wrong as "actually being right" if only you had known X, Y, or Z. Remember, the Secret King is never wrong.

I'm sensing some strong gamma tells coming from dh who still refuses to accept that the vast majority of the polls were manufactured propaganda that had no bearing on reality.

No, because he has admitted he was wrong. He's just in the position of a numbers guy who has a little trouble accepting the paradigm shift that screws up his numbers. As I said before the election, IF the models were correct, THEN I would be wrong. I was relying on a paradigm shift from ideology to identity, and that's what we saw.

Blogger James Dixon November 13, 2016 5:51 AM  

> We strongly disagree with the idea that there was a massive polling error. Instead, there was a modest polling error, well in line with historical polling errors,

Actually, this is exactly correct. The polls this time mirrored almost exactly the polls for Nixon, Reagan, and Bush the Elder. The polls *always* predict a massive win for the democrat early on and narrow at the end. And even with the narrowing, they never get the actual support for the republican right. The only exceptions within my lifetime were the Obama elections.

Anonymous 5343 Kinds of Deplorable November 13, 2016 5:53 AM  

Given this, I have to push out my projection that Democrats have a strong chance of a long-term black-latino-white woman coalition that could deny the White House to the GOP long-term. There is still hope.

@8. @15. Agreed, and agreed.

I think the white-women-for-Trump percentage was 52 or 53, which (as OGRE says) suggests the Dem's coalition will not hold long-term, particularly if they're running a male candidate and especially if the importance of identifying as white with your vote **from a safety perspective** is pushed for the next four years.

Fear is the biggest motivator for women with families, and they're the biggest group of the three in the coalition. That's the drum to beat regularly.

Blogger Derrick Bonsell November 13, 2016 5:55 AM  

I know that adding four points in the polls to Trump wouldn't have explained PA or WI, but it would have gotten you roughly the (meaningless) popular vote total, which can be explained as not a Trump win because CA, NY, and IL were so overwhelmingly Democrat they immediately tipped the scales and Republicans in those states stay home because they have no chance of influencing the results of any major race.

So Democrats can just run up the numbers there even though they have no further bearing on the results. No doubt the same phenomenon exists in deep red states like Kansas and Nebraska, which goes to show that a direct election would completely change our political system in ways that are impossible to predict.

Blogger James Dixon November 13, 2016 5:56 AM  

> 4. I think it was Stilicho or maybe Dixon who was very close to exactly correct about turnout models.

That must have been Stilicho, dh. I didn't make any specific predictions about turnout except that the Democrats would be down and the Republicans would be up. I'm not sure the latter even turned out to be true.

Blogger Cataline Sergius November 13, 2016 5:56 AM  

I feel a need to do a little preening over my own predictions on the polls.

The polls were correct for Romeny/Obama '12 and McCain/Obama '08 but both of those elections had couple of things going for them that the current election does not.

1. Political stability. Both of those periods took place in times of political stability. That is absolutely, definitely, definitively and in all ways not the case this year. We are in a state of potential rebellion and if you read this blog you know that isn't hyperbole.

2. Right Wing Political Apathy. That was very much in play in '08 and even more in play in '12. Romney could have tried to stoke the middle class uprising known as the Tea Party but he was fundamentally repelled by it. His attitude was clearly one of, if they vote for me fine but our relationship ends there.

The Right Wing is not apathetic at all this year. We are on the march

Anonymous 5343 Kinds of Deplorable November 13, 2016 5:58 AM  

dh who still refuses to accept that the vast majority of the polls were manufactured propaganda that had no bearing on reality

I think dh's contributions during and after the election have been quite useful, and I've found him ruthlessly honest rather than gamma-self-justifying. Rah-rah-ing yourself into a delusionary state is a mark of Leftism, or it should be. I would hate to discourage principled analysis of the facts from different perspectives around here.

Blogger GFR November 13, 2016 6:08 AM  

Elizabeth Warren does NOT have nice boobs. (Or anything else).

Blogger James Dixon November 13, 2016 6:09 AM  

> Next time it will not be the old delusioned hag, but nice, intelligent 40 with nicer boobs.

They don't have anyone matching that description.

> I know that adding four points in the polls to Trump wouldn't have explained PA or WI, but it would have gotten you roughly the (meaningless) popular vote total,

My off the cuff rule has always been to add 5% to the stated Republican support. It's usually worked pretty well. As I said, the exceptions were the Obama elections.

I'm trying to figure out a meme pushing Joe Manchin as the new leader/candidate of the Democrats. I know the odds are against it, but they'll never think of it on their own.

Blogger James Dixon November 13, 2016 6:13 AM  

> I think dh's contributions during and after the election have been quite useful, and I've found him ruthlessly honest rather than gamma-self-justifying.

Agreed. Dh, is honest. He goes where the numbers lead him. I'd say his problem is that he thinks polling is a science, when it's actually an art. There's a good comparison to be made with computer programming in that regard, though it's not exact.

Blogger GFR November 13, 2016 6:14 AM  

Why would you send out a 37 page e-mail attachment on how to rig polls, if you weren't planning on rigging polls?

Blogger Derrick Bonsell November 13, 2016 6:16 AM  

If they try to run a minority like a Tulsi Gabbard they're in for a bad time. Not the fact that they're running a minority, simply that anyone in position to know how to win is a white Democrat who's getting up in years.

Though the possibility of things getting so messed up they just waltz into the White House is still there. If Trump can reassure voters in "Red States" and increase his "Rust Belt" vote he'll easily be reelected (if he chooses too. He might just find that he doesn't want the job anymore.)

Blogger James Dixon November 13, 2016 6:19 AM  

Oh, and the reason for pushing Manchin? Well, aside from his being a native West Virginian, it's actually in the public interest to have a mainstream functional opposition party. We've seen where 8 years of full Democratic control can get us, and full Republican control will have similar problems if extended for too long a period. Yes, Trump will need a full 8 years to roll back the Obama agenda and get the country back on track. But once that's done we actually need the give and take between two parties with opposing ideas. Someone like Manchin could possibly lead the party back to the mainstream and away from the extreme Marxist path they've been on.

Blogger GFR November 13, 2016 6:20 AM  

What logical reason could there be for ALWAYS predicting a massive win for the democrats? Other than the desire of democrat pollsters to demoralize Republicans and energize democrat voters?

Blogger al November 13, 2016 6:25 AM  

Manchin has already offered to defect to the GOP to save himself in West Virginia where the Dems are collapsing. Also, he needs to maintain the influence to help his crony capitalist daughter loot medicare and force schools to overpay for Mylan Epipens. Trump must make an example of Mylan especially for their corporate inversion.

Blogger GFR November 13, 2016 6:27 AM  

They'll probably run Moochelle Obama.

Anonymous Bz November 13, 2016 6:28 AM  

It seemed fairly clear that most of the polls were for propaganda and/or entertainment purposes only. And as Taleb wrote, there were some fundamental misunderstandings of probability there.

Well, you can't blame Silver for trying to salvage what's left of his little operation by spinning like crazy. Or maybe you can. I recall that Silver admitted to plain propaganda against Trump in the primaries, and after that I didn't take him seriously. The underlying problem is, 538 is not seeking the truth.

Blogger James Dixon November 13, 2016 6:28 AM  

> What logical reason could there be for ALWAYS predicting a massive win for the democrats?

The polls are propaganda, not predictions.

Blogger GFR November 13, 2016 6:31 AM  

He doesn't take into account that the polls are intentionally rigged to manipulate the electorate emotionally. That is a remarkable oversight.

Anonymous Bz November 13, 2016 6:31 AM  

Maybe Nate and 538 should branch out to fact checking.

Anonymous SciVo de Plorable November 13, 2016 6:36 AM  

This isn’t just a case of hindsight bias.

I love that. Is there a term for unconscious self-awareness via explicit denial of obvious truth? I assume German.

Blogger GFR November 13, 2016 6:37 AM  

That's so cute. They want to genocide us and you want to strengthen the dynamic interplay of the two-part system.
.
Nuke them from orbit. It's the only way to be sure.

Anonymous SciVo de Plorable November 13, 2016 6:43 AM  

@ CaseyD: So is not being able to admit you're wrong a strong gamma tell?

Yes and no. It could be an omega tell. False humility is a stronger gamma tell. In fact, I hypothesize that gammas are just omegas that are better at hiding it.

Blogger GAHCindy November 13, 2016 6:46 AM  

@30 I don't know what your family is like, so this might be useless to you, but I'm in a similar quandary over my sister. I'd love to think of that magic thing to say that will make her see how insane her political thinking is. I've determined, unfortunately, that the reason she overwhelmingly prefers the anti-white, anti-Christian, anti-family party is that she is actually doing what she has always done: rejecting her parents, her heritage, and now sabotaging her nieces' and nephews'future. She has always chased the novel, preferred the company of strangers, ignored her family obligations, and swallowed the feminist lie, hook, line, and sinker. No kids, pathetic gamma husband, messed up personal life, and no hope for a non-government-supported, productive life. She's still working on a degree in, God help us, cultural preservation, at the age of 36. While she's studying preserving everybody's culture but her own, I'm actually preserving ours by having 8 kids and raising them myself. We're on good terms, superficially, and always have been, but she's (I'm afraid there's no other explanation) actually very hostile to me and mine. She has always defined herself in direct opposition to who she actually is, and has the psych history to prove the mental toll that takes. I can't red pill her. She's a perfectly brainwashed tool of the school system. She literally wants to see her family heritage eliminated...though in the future, through immigration and religious persecution, so that she can always pretend it wasn't her doing. Nothing I can do about that but pray she repents and keep the family peace the way I always have: by pretending I don't know how she really thinks about these things and subtly showing her the truth through my own happiness. I hope someday I'll discover that I'm wrong, and there's actually a magic word I can say that will wake her up to the wrongness of her thinking, but I fear she's already fully aware of what her choices mean.

Blogger James Dixon November 13, 2016 6:46 AM  

> They want to genocide us

The parties as current composed want to genocide us. Trump is remaking the Republican party. I'm hoping someone can do the same for the Democrats.

Anonymous Johnnyboy November 13, 2016 6:47 AM  

@3
Dewey Defeats Truman in our time! Now there's a picture worth spreading far and wide.

@5
What's Pepsi gone and done?

Blogger Mr.MantraMan November 13, 2016 6:50 AM  

Go with Vox's explanation and hope the tard establishment stays in its bubble

Blogger GFR November 13, 2016 6:54 AM  

Mine was like that but she red-pilled herself in her forties. It had something to do with student loans.

Blogger Stilicho November 13, 2016 6:54 AM  

Simply adjusting the pro hillary polls (eg ABC) for historical voter affiliation (for early November of election years) produced an accurate prediction.

Anonymous Bz November 13, 2016 7:01 AM  

By the way, the Princeton Electoral Consortium apparently had Hillary at 99% to win. Here is their leader, professor Sam Wang, eating a bug in penance:

http://www.unz.com/isteve/forecasting-especially-the-future-is-tough/

Blogger GAHCindy November 13, 2016 7:06 AM  

Also, I apologize for being inappropriately long and off-topic. It's weighs heavy on my mind, especially on Sundays. It's not primarily a political problem, but a spiritual one. "This kind (of demon) only comes out by prayer and fasting" would have been a better, shorter answer.

Blogger GFR November 13, 2016 7:06 AM  

Someone CAN.
.
We should genocide them first by supporting Hispanic democrat politicians over White democrat politicians.
.
They are perfectly willing to sell the White middle class down the river, but they aren't willing to go there themselves.

Anonymous hardscrabble farmer November 13, 2016 7:07 AM  

I read through all the comments first to make sure I didn't repeat this, but did anyone see Nate Silver on election night coverage? He makes the Train Is Fine guy look like James Bond.

I'm a farmer so for me everything is genetics or physiognomy when it comes to evaluating livestock and human beings. Nate Silver is a poor specimen of a human being beyond the rotten crooked teeth and prematurely balding pate, stooped posture, nervous eyes, fidgety hands,etc. It's unmistakable and anyone who sees more than a few seconds of this guy should have their sensors wailing. But no, not in America where we look beyond the blubber and the face tats to the unique individual beneath and search out what truly makes them special.

This is a fiction that must be pointed out.

Poor specimens of any species ALWAYS fail to thrive. They never fit in properly with the herd/group, they suffer all kinds of ailments,they have more frequent accidents and incidents, they do not blossom into type but always feature bad hips, patchy hides, poor digestion, etc. With a cow or a pig you simply turn them into meat sooner or cull them earlier, but with humans we spend five to ten times the resources to just keep them at the edge of the social dynamic and then heap fabulist praise upon them when they show up somewhere on time with their clothes on.

It's sad, really.

But try this going forward-judge each new person immediately on how they look as an example of their species and see how it pans out as you get to know them and see if this isn't more statistically accurate than all of their polls put together.

Blogger GAHCindy November 13, 2016 7:08 AM  

@60 ok. Prayer, fasting, and the bills coming due. Lol.

Blogger al November 13, 2016 7:08 AM  

There's so much disbelief on twitter about how completely Trump flipped rural white districts that Obama had won twice. Many of them imply that Trump rigged the election in rural Ohio where Obama won because he beat Hillary there by 30%. They really just refuse to understand that bragging about the white majority dying doesn't ingratiate you to white people. Apparently complaining that the privileged and dying white community isn't "voting for their best interests" is also a talking point.The polarization is working more quickly than I ever imagined. They are just lucky it was only Trump running, since he is basically a moderate Republican with a few populist issues.

Anonymous 0018 November 13, 2016 7:15 AM  

@dh One more paradigm shift that is likely affecting your polls is the societal shift from r to K (h/t Anonymous Conservative). In the recent extreme r environment where non-PC is openly persecuted, r-people (rabbits) tend to over-signal to show that they are part of the herd and to try to bluff the adversary that they are too strong and resistance or counterattack is futile. In contrast, many K-people (wolves) will keep things close to the vest, so that by the time they feel they must signal, it's too late and the fight is on.

Ask yourself how many r people you know that constantly virtue-signal and peer-pressure others to do the same. Conversely, how many of your K acquaintances either don't respond to polls at all or give bland responses?

Blogger Gordon November 13, 2016 7:18 AM  

@56 GAHCindy: It's painful when someone close to you heads down the path to self destruction, even if her path is a long and winding one.

As Vox is known for quoting, Aristotle wrote that some people cannot be instructed. I look at it another way: some people are immune to evidence. As an example, there is a gal in my life, on the periphery. She knows I don't lie, she acknowledges that I have better arguments.

She was going on about poor little Travon Martin being abused by big evil Zimmerman. I pointed out that Travon was not the little kid in the commonly used photo, but a 17-year old thug who was taller and heavier than Zimmerman. I showed her documents to back this up. She admitted the evidence, and instantly went back to believing Travon was the cute little 12-year old in the photo.

Such people are particularly vulnerable to the idea that white people must atone for their sins. She doesn't want to give up her very good job or her comfortable lifestyle, but, you know, white people somewhere should suffer. Probably those bad white people who voted for Trump, because she's virtuous and voted Obama and Clinton.

Blogger Gunnar Thalweg November 13, 2016 7:18 AM  

In college, I had a friend who predicted the student assembly elections consistently correctly with this rule of thumb: Whoever posts the most signs, wins.

During this election cycle, I was hoping for Trump, and the "sign rule" overwhelmingly favored Trump.

On the other hand, Big Data ...

And the results -- were the Democrats stayed home. There was NOT an overwhelming turnout. That actually has me wondering whether something has been afoot as far as voting fraud. This may not have been as close as it appeared. But I'll never know.

I used to come here, even though I disagreed, just to read Vox Day stand firm that Trump would win. This was something that could be empirically tested.

Anonymous Faceless November 13, 2016 7:24 AM  

I guess it's finally time to just go with the sparkling water. They just want to ruin everything Dale Jr once touched this election cycle.

Anonymous SciVo de Plorable November 13, 2016 7:26 AM  

dh wrote:Given this, I have to push out my projection that Democrats have a strong chance of a long-term black-latino-white woman coalition that could deny the White House to the GOP long-term. There is still hope.

1. You're wrong. White women will come home to our tribal coalition, because white men are literally the only people in the world that care about their safety.

2. I hate you. I also like you, for attempting objectivity, but I also hate you for opposing my tribe. I will not speak my wishes.

Blogger Gordon November 13, 2016 7:36 AM  

@72: "White women will come home to our tribal coalition, because white men are literally the only people in the world that care about their safety."

Hm. This is interesting. I want to think about it some, but my first impression is, yes, for some. But those like GAHCindy's sister won't, not until The Day The EBT Cards Stop Working. She'll come running, then, if she survives. But you can never trust her, because once the fear abates, she'll go right back to her old thinking.

Blogger James Dixon November 13, 2016 7:45 AM  

> They really just refuse to understand that bragging about the white majority dying doesn't ingratiate you to white people.

Well, that and promising to put a lot of you out of work. SE Ohio has a large coal/gas base. They knew Hillary was very anti-coal and gas.

Blogger Lovekraft November 13, 2016 7:45 AM  

I have also noticed a consistent, glaring lack, among the sjws, media, policians, of self-reflection. Their arrogance has blinded themselves and will result in more failure.

We get to press our demands, push forward and ignore the obvious obfuscators and meddlers.

Blogger Timmy3 November 13, 2016 7:47 AM  

The polls reflect the bias of the mainstream media. They want Hitlary to win. Of course, I didn't know who to trust. In California, she actually won 2 to 1. Many states, red or blue, are at one side or another. The so-called swing states are closer. Thus the national polls are useless especially when Trump lost very slightly the popular vote although Hitlary didn't win a majority of 50% plus 1. Best to consider each scenario.

Blogger wreckage November 13, 2016 7:47 AM  

@73, women are historically a very conservative voting bloc. They've just been convinced by years of propaganda that white middle class Christian men are the biggest and worst threat to them, so they are cautiously voting that threat out to arm's length.

Propaganda stops working eventually, hell, it even flips: by the end of WW2 Germans were listening to Allied propaganda by choice, because their trust in their own had been totally destroyed.

Right now, in the streets, and broadcast by the Left, the Left themselves are proving that it is they who are dangerous and unpredictable.

Also, let me add: in Australia the best predictor of political affiliation is this: men and women with two kids or more tend to flip to conservative regardless of prior politics.

Childcare and maternity leave concessions might sound like a bad idea from Trump, but they are potentially a useful weapon in the War On Leftism.

Blogger synp November 13, 2016 7:49 AM  

There wasn't a polling error. There is a polling bias. A severe one. Publishing straight poll results would be meaningless because of the bias. So they correct for the bias before publication. That is what the "polling model" is all about.

This is all well and good. The only problem is that they don't have a scientific way of correcting the bias, so it's mostly a guess.

For example: polls are still conducted by calling landline phones. This was always biased, because women tend to answer the phones more than men, but this could be accounted for if you knew from past elections how much women were likely to vote democrat more than men.

These days 40% of households don't have landline phones at all. They don't appear in the polls. These tend to be younger people more than older people, and single people more than married people, and urban people more than rural people. All three of these random variables are correlated with political preferences, so there is bias. But unlike the women vs men thing, we don't know very well how to correct these biases, so all pollsters guess.

I don't know why the LA Times guessed better than others. It could be that they actually have a better model, or that they just got lucky with their corrections.

Anonymous SciVo de Plorable November 13, 2016 7:49 AM  

hardscrabble farmer wrote:But try this going forward-judge each new person immediately on how they look as an example of their species and see how it pans out as you get to know them and see if this isn't more statistically accurate than all of their polls put together.

Well duh. Been doing that my whole life, because physical evidence is objective and useful, and physiognomy is real.

Are you telling me that other people don't do this? Because 1. that would be stupid and 2. it would also explain a lot.

Blogger Lazarus November 13, 2016 7:51 AM  

MATT wrote:Vox, what sort of rhetorical response do you suggest in order to show someone how ridiculous their anti-Trump stance is?

On Wednesday, Charles Johnson, an alt-right troll who calls himself a journalist, was sitting on a Brooklyn-bound F train wearing a Make America Great Again hat. “You support a man who is racist, sexist, and homophobic,” a man standing next to him said,.... “We won—fuck off,” Johnson said....

Blogger Dave November 13, 2016 8:04 AM  

Silver will have to do better than his non-excuse excuse to restore the faith of people like this:

Aaron Levie, CEO of Box expressed his anger with Nate Silver, the pollster of prediction website 538, tweeting:

Nate Silver, we trusted you.
— Aaron Levie (@levie) November 9, 2016

Anonymous SciVo de Plorable November 13, 2016 8:07 AM  

synp wrote:I don't know why the LA Times guessed better than others. It could be that they actually have a better model, or that they just got lucky with their corrections.

What is the difference? How would you define it in a way that you could tell this is the one, and that is the other? And if you can't rigorously define a discernable difference, can you even plausibly claim that one such exists?

I'm quite serious. I actually have a degree in math, but I'm horrible at statistics. I just don't understand it, which frankly makes me think that there is something wrong with the field itself, because I know I'm smart. It's just so arbitrary.

Blogger Josh (the gayest thing here) November 13, 2016 8:12 AM  

NB: Silver is a lambda, not gamma

Blogger residentMoron November 13, 2016 8:16 AM  

As a legal migrant whose already paid hundreds of thousands of Euro in taxes, I am looking forward to next years German election. I suspect the pundits could be wrong about it, too.

Do not read the BBCs article today about the Germans response to Trump winning - if you value your brain cells.

Blogger VD November 13, 2016 8:17 AM  

They don't have anyone matching that description.

I was thinking Jennifer Granholm, but she's 57 now. Too old. If Keith Ellingson becomes DNC chair, they'll run a black, possibly even a black Muslim.

Yes and no. It could be an omega tell. False humility is a stronger gamma tell. In fact, I hypothesize that gammas are just omegas that are better at hiding it.

You're wrong. The strongest Gamma tell is evasion, rationalization, word salad, and finally going radio silent in order to avoid admitting that one is wrong.

Blogger VD November 13, 2016 8:20 AM  

Vox, what sort of rhetorical response do you suggest in order to show someone how ridiculous their anti-Trump stance is?

Given that we had a choice between a sexist racist and a child-abusing Satanist cannibal who was pushing to start a war with Russia in Syria, Ukraine, and Georgia, don't you think the racist was the right choice?

Blogger JaimeInTexas November 13, 2016 8:23 AM  

ZeroHedge had a few articles where staticisians looked at the spread among the polls and concluded that there was deliberate manipulation to achieve a result.
As VD had stated, trends is what mattered. Adjusting for a bias against Trump in polling of 8 to 10 percent, on a state to state basis, and Trump-slide was very much at play.
I bet $100 that Trump would win. I think the bet was made on Saturday. My wife owes me $100.

Blogger synp November 13, 2016 8:24 AM  

dh wrote:Given this, I have to push out my projection that Democrats have a strong chance of a long-term black-latino-white woman coalition that could deny the White House to the GOP long-term. There is still hope.

Any coalitions can only survive until one or more of the components doesn't find it beneficial anymore. White women already defected from this coalition in this election. A third of latinos defected as well. (I'm using "defect" in the game theory sense - not implying any judgement). If they live to regret it, defection rates will be lower next time. If it turns out fine, defection rates increase.

In the next presidential elections, assuming Trump runs again (he will be 74, so nothing's certain) most of the scare tactics employed by the Democrats this time around will not work. The worst scare tactics you can use against an incumbent is "we'll have four more years of this." Unless things are actually awful, this isn't too scary. That's why a sitting president was voted out of office only twice in the last 60 years (Carter and Bush I).

Without the scare tactics, I don't see why women would vote much differently than their social circle, and that social circle also includes men. That social circle is a powerful pull for women and men to vote similarly. Latinos in northern states and in some southern states (not so much in Texas, Arizona and California) tend to favor integration, which may pull them to vote more in line with their neighbors.

Think this is impossible? In this election Orthodox Jews voted for Trump more than in any previous elections. Reform Jews didn't, but it used to be that the Orthodox Jews were just as reliable Democratic voters as their Reform brethren. No more.

Coalitions don't last forever. They are marriages of convenience and should be treated as such.

Blogger Josh (the gayest thing here) November 13, 2016 8:26 AM  

Except the LA TIMES/USC and IBD/TIPP Tracking, Who, Unlike Us, Actually Got It Right

IBD had Trump +2, LA Times had Trump +3.

Clinton 47.7 Trump 47.1 is the current popular vote according to Dave Wasserman.

https://mobile.twitter.com/Redistrict/status/797566062665547776

Anonymous Pops November 13, 2016 8:29 AM  

Trump has just gone back on his promise to repeal Obamacare!

Blogger Josh (the gayest thing here) November 13, 2016 8:31 AM  

This is a really good article by Sean Trende on the 2016 polling:

http://www.realclearpolitics.com/articles/2016/11/12/it_wasnt_the_polls_that_missed_it_was_the_pundits_132333.html

And with the final RCP Poll Averages in a handful of battleground states showing Trump trailing by just 0.6 percent in New Hampshire, 1.9 percent in Pennsylvania, 2.9 percent in Colorado, and 3.4 percent in Michigan, it was pretty clear that Donald Trump had a very real path to the White House.

Blogger synp November 13, 2016 8:32 AM  

SciVo de Plorable wrote:synp wrote:I don't know why the LA Times guessed better than others. It could be that they actually have a better model, or that they just got lucky with their corrections.

What is the difference? How would you define it in a way that you could tell this is the one, and that is the other? And if you can't rigorously define a discernable di>fference, can you even plausibly claim that one such exists?


Repeatability. If their model keeps working for more elections, then it's good. Otherwise, they just got lucky.

I'm quite serious. I actually have a degree in math,
Hey! Me too!
but I'm horrible at statistics. I just don't understand it, which frankly makes me think that there is something wrong with the field itself, because I know I'm smart. It's just so arbitrary.

Lies, damn lies, and statistics?

Statistics just gives you a bunch of numbers from a file with many numbers. As soon as you try to build models, you're out of the safety of maths and into the by-definition uncertainty of science. And unlike physical science where you can repeat an experiment many times to build confidence in your model, you get national elections only once in four years (once in two years if you count mid-terms)

Blogger VD November 13, 2016 8:34 AM  

They got the winner right, Josh. That's the point. Furthermore, neither the Electoral College vote nor the popular vote are complete yet. The very tweet you cited says (millions more still to count).

I really do not understand why you do not simply hold your tongue about anything and everything to do with him. All you do, every time, is remind everyone how completely and stubbornly wrong you were.

You can't spin it. You can't pull it out somehow by correcting someone else about some petty aspect of it. Just accept that you made a mistake, learn from it, and do better next time. This wounded gammatuding won't get you anywhere. Ever.

Blogger E. Burke November 13, 2016 8:35 AM  

The pollsters are kind of accurate, but not as much as they would have the public believe, and hardly more so than in '48, when Dewey defeated Truman.

Blogger Josh (the gayest thing here) November 13, 2016 8:39 AM  

They got the winner right, Josh. That's the point.

That was going to be my next question (probably should have been my first)

What's the best way to evaluate a forecast or a model when they're trying to predict two outcomes (popular vote and electoral college)?

Blogger synp November 13, 2016 8:46 AM  

Josh (the gayest thing here) wrote:What's the best way to evaluate a forecast or a model when they're trying to predict two outcomes (popular vote and electoral college)?

A high rate of predicting correctly? The problem is that there are presidential elections only once every four years, so we don't have enough data points to evaluate their accuracy.

Blogger Josh (the gayest thing here) November 13, 2016 8:48 AM  

You can't spin it. You can't pull it out somehow by correcting someone else about some petty aspect of it. Just accept that you made a mistake, learn from it, and do better next time. This wounded gammatuding won't get you anywhere. Ever.

I'm not trying to spin anything about Trump. I'm trying to point out that when you build a model or forecast based on the public polling data, which is what 538 did, they did a better job than the other folks who built similar models.

Silver spent a lot of time last week pointing out that Trump just needed the polls to be off by a standard margin to win the electoral college.

Anonymous Colonel Farrel's Ghost November 13, 2016 8:53 AM  

The Dem party is shifting hard left after this loss. Their base is becoming radically anti-white. They're in the process of purging moderates and whites. So it's definitely game on: The White Party vs. The People Of Color Party.

Blogger Cail Corishev November 13, 2016 8:53 AM  

The parties as current composed want to genocide us. Trump is remaking the Republican party. I'm hoping someone can do the same for the Democrats.

The left will always want to genocide us, because the whole purpose of the left is to destroy all civilization so they can build a new global order in their own image. Order from Chaos. So the Democratic Party (like the Republican) will stop wanting to destroy us to the extent that it stops being a party of the left.

I don't know what that would take. It's not clear to me yet what policy positions will be available that are neither globalist left nor already staked out by Trump's GOP.

Would it be possible to have two (or more) parties, neither leftist, that compete over rational disagreements on how best to run a relatively benign state?

Anonymous Wiccapundit November 13, 2016 9:05 AM  

@22 Steve:

Dude, I predicted Trump would win, would win 300+ EV, and would win Pennsylvania, Michigan, and Ohio, all by reading the Tarot. I posted these predictions on my website RedStateWitch the night before the election.

What does is say about Silver's black box "methodology" when a witch reading Tarot cards can beat the snot out of his predictions?

Maybe he should move to interpreting chicken entrails.

Anonymous Cinco November 13, 2016 9:07 AM  

what sort of rhetorical response do you suggest in order to show someone how ridiculous their anti-Trump stance is?

If you take any one single issue and flip it on them with rhetoric it works out great, for instance.

The rich should pay their fair share

Rhetorical response: You support starving African babies to death?
Dialectical Backup: Bill Gates donates millions to Africa and they would take some of that away, so more African children would starve.

Open borders/free movement of peoples.

Rhetorical response: Why do you support wiping out what is left of the Native Americans?
Dialectical backup: Can you explain how mass immigration was good for the Native Americans?

It's really hard for me to think like a liberal. In part because I have almost no emotional affect and because I am an INTJ with above average pattern recognition skills and an unshakable belief in social Darwinism.

Someone should really publish a rhetorical response manual with every single sub-part of the liberal agenda and then combine it with a dialectical response.

Blogger Orville November 13, 2016 9:07 AM  

settle down Pops. Trump is all about negotiation jujitsu.

Blogger Cail Corishev November 13, 2016 9:08 AM  

There's so much disbelief on twitter about how completely Trump flipped rural white districts that Obama had won twice.

People in white rural areas may go months without seeing a black person in real life (I do), so their impression of blacks comes primarily from the smart, witty, sophisticated blacks on TV. They saw Obama as one of those nice black(ish) men, and voting for him was a "See, we're not racist" attempt (not that it worked, they might now notice).

On the other hand, they all know at least one shrill, pushy harpy in real life. They might even be married to a Hillary or have one as a mother-in-law. They knew who she was and felt no urge to vote for her, even before you add the war-mongering and satanism.

But media types couldn't understand that, because to do so, they would have to understand how much of Obama's win was driven by virtue signaling, not a recognition of his wonderfulness.

Blogger Josh (the gayest thing here) November 13, 2016 9:10 AM  

The sheer number of people who voted in 2008 and 2012 was staggering. 2008 saw a total of 132 million voters. In 2012 it was around 130 million. 2014 mid-terms was 83 million. We are looking like ending up in the ballpark of 122 million votes this time around. Literally 8+ million people who voted in 2008 and 2012 simply stayed home.

Based on what I linked to upthread, 128.5mm votes have been counted so far, which is only 500m less than 2012. Given that they're still counting votes out west, 2016 will probably surpass 2012 for total votes cast.

Blogger Tex Longhorn November 13, 2016 9:12 AM  

This comment has been removed by the author.

Blogger Cail Corishev November 13, 2016 9:15 AM  

Why would you send out a 37 page e-mail attachment on how to rig polls, if you weren't planning on rigging polls?

But see, that was just about the internal polls that all parties have done, which are intentionally biased for use as propaganda to boost enthusiasm within the troops. That was totally, definitely, absolutely not intended for use in standard polls. Not a bit.

Blogger Josh (the gayest thing here) November 13, 2016 9:17 AM  

A high rate of predicting correctly? The problem is that there are presidential elections only once every four years, so we don't have enough data points to evaluate their accuracy.

Yup. Small sample size theater. Silver got 2008 and 2012 right, I assumed he would get 2016 right. I was wrong.

Blogger Cail Corishev November 13, 2016 9:24 AM  

"The polling numbers were right, but our predictions were wrong" is a lame dodge, because it's the predictions that matter. It's like saying, "I planted the right seeds, so it's not my fault the plants didn't grow, just because I watered them with Brawndo. Look at those great seeds!"

If you took good numbers and produced a bad prediction because you're biased or stupid or out of touch with Americans, you failed. No one cares if you can point to some numbers on your web site and say, "Look, the right info was there, I just missed what it meant."

Blogger Mom November 13, 2016 9:26 AM  

@85 Jennifer Granholm is Canadian by birth and thus ineligible.

Anonymous Nate 538 November 13, 2016 9:26 AM  

The poll is fine.

Blogger Josh (the gayest thing here) November 13, 2016 9:29 AM  

"The polling numbers were right, but our predictions were wrong" is a lame dodge, because it's the predictions that matter. It's like saying, "I planted the right seeds, so it's not my fault the plants didn't grow, just because I watered them with Brawndo. Look at those great seeds!"

If you took good numbers and produced a bad prediction because you're biased or stupid or out of touch with Americans, you failed. No one cares if you can point to some numbers on your web site and say, "Look, the right info was there, I just missed what it meant."


Nobody who created a model using the public polls as the inputs got the election right.

Blogger Josh (the gayest thing here) November 13, 2016 9:31 AM  

My extremely early 2020 prediction is that the Democrats are going to counter Trump by running their own famous billionaire celebrity, so they'll try and draft Cuban or Bloomberg. Maybe Oprah.

Blogger Derrick Bonsell November 13, 2016 9:39 AM  

(then) 78 year old Bloomberg.

Yeah, totally going to happen.

Anonymous Wiccapundit November 13, 2016 9:45 AM  

@103 Cail Corishev

"On the other hand, they all know at least one shrill, pushy harpy in real life. They might even be married to a Hillary or have one as a mother-in-law."

Yep, I stated this a month before the election in a post on my site. I said the election was decided months ago, and not in a tin-foil-hat-conspiracy-to-steal sort of way. It was decided by men. Men who "listen(ed) to Clinton and hear the screeching shrew that was their ex-wife or crazy college girlfriend."

Maybe I should go into the (pseudo)polling business . . .

And by the way, there are indeed some of us witches who are Alt-Right instead of California-Granola-muffinhead-crazy.

Blogger Cail Corishev November 13, 2016 9:45 AM  

Millions of optimistic party faithful predict the election right every time their favorite wins, and wrong every time he doesn't, just like Silver so far -- and they don't need any polling numbers to do it.

Blogger Dave November 13, 2016 9:46 AM  

For Pete's sake, Mom, tell Pops to lay off the sauce already.

Blogger Cataline Sergius November 13, 2016 9:46 AM  

Jennifer Granholm...good ole' Ding Bat Jenny. AKA Transgender Obama in white face. Born in Canada to an American mother. Educated in Berkeley.

An absolute disaster as governor of Michigan, not that the Democrats outside of Michigan ever noticed.

Her governorship was eight years of unrelenting economic misery.

She only got re-elected because the Michigan GOP was in such shambles that it put up a proto-Mitt Romney as a candidate, whom everyone immediately christened "Captain Amway."

She was in all respects a preview of Obama.

To include paving the way for a GOP resurgence. Although lets be clear, Snyder is no Trump.

Blogger JaimeInTexas November 13, 2016 9:52 AM  

National polls are only part of a picture that informs state by state polling. Ultimately it is the states as a whole, through the electoral college, that elects the POTUS.

Blogger James Dixon November 13, 2016 10:11 AM  

The problems with polling:

1) You have no certainty who will vote and who won't in any given election.
2) Reaching a statistically correct sampling of those voters is extremely difficult, especially given 1).
3) You can't be certain what people tell you is actually what they're going to do.
4) The people paying for your poll often want to hear a specific result, regardless of what the actual numbers say.

The combination of the first three mean that even a properly run poll is going to have problems making a correct prediction. Simply put, their margin of error is significantly greater than they pretend it is.

When you add in the fourth, well, you just threw the baby out with the bathwater.

Blogger James Dixon November 13, 2016 10:18 AM  

> Would it be possible to have two (or more) parties, neither leftist, that compete over rational disagreements on how best to run a relatively benign state?

That should be the goal, yes. Is it possible? Not with the state we have now.

Blogger James Dixon November 13, 2016 10:26 AM  

> Silver got 2008 and 2012 right, I assumed he would get 2016 right. I was wrong.

What was the commonality in 2008 and 2012 that was missing in 2016 and could have skewed the polls?

Blogger Josh (the gayest thing here) November 13, 2016 10:34 AM  

What was the commonality in 2008 and 2012 that was missing in 2016 and could have skewed the polls?

That's a good point, but not the whole picture.

In the Midwest there were a lot of rural white working class counties that flipped from Obama to Trump by yuge swings.

I'll have to look, but I remember reading several articles this week pointing out that the polls were more wrong in the Midwest than other regions, particularly with white working class voters.

The Sanders win in Michigan should have been an indication of that. I certainly missed it. IIRC, you didn't.

Anonymous Cadwallander J November 13, 2016 10:41 AM  

First - congratulations to Vox on reversing his historically abysmal presidential prognostication, and calling it quite early. Get that man a blue lagoon and tiny umbrella.

Secondly, the biggest problem with the polls was the stated 3rd party support which evaporated to historical norms on election day. It tells us people misled the pollsters. That means the jig is truly up - if random polled Americans have that level of mistrust of the MSM, their propaganda monopoly is at an end.

Blogger seeingsights November 13, 2016 10:44 AM  

A couple of points:

1. Nate Silver started in sabermetrics, which is the statistical analysis of baseball. I've been involved a little in sabermetrics. The math is at high school algebra level. Some in sabermetrics think that Nate Silver is over-rated.

2. From what I infer, Nate Silver's method in politics consists in taking polls, and weighing them each in terms of their allegedly accuracy, and then averaging out these weighted polls.

3. There is a desire in humans to look for someone with knowledge of the future. Centuries ago, people looked to soothsayers and prophets like Nostradamus. In science fiction, one of the most famous fictional characters is Hari Seldon, who plotted out the future corse of history. In financial markets, people use the phrase 'financial guru', people who suggest what investments to buy. Nate Silver is playing the guru role.

Blogger Francis Parker Yockey November 13, 2016 10:49 AM  

@Franz Lion heart
You spelled Lenin wrong.

Blogger Francis Parker Yockey November 13, 2016 10:55 AM  

@dh
TL;DR:
It doesn't matter how elegant the model is, or how solid the math, if the underlying assumptions are wrong. Gee, it almost seems as if I've read that here before.

Kind of like econometrics and "free trade" that way.

Anonymous Drummergirl November 13, 2016 10:57 AM  

@8 dh - I've read on a couple of sites that states don't count their absentee ballots if it is not necessary based on the number of absentee ballots vs the margin of victory. Do you know if this is true? If it is, do you think it's possible that there were about as many total votes this year as in previous years, but more people voted early via absentee ballot and those votes aren't being accounted for? I just wonder because I am surprised that there were about 10 million or so fewer total votes this year than in the last 2 elections.
It also seemed to be suggested that more people voted early this year than in previous years. I really don't know, I just wondered what your (or anyone else's) thoughts were.

Blogger Josh (the gayest thing here) November 13, 2016 10:59 AM  

I just wonder because I am surprised that there were about 10 million or so fewer total votes this year than in the last 2 elections.

There aren't 10 million fewer votes, right now it's about a half million fewer votes.

Blogger Francis Parker Yockey November 13, 2016 11:02 AM  

@Bernard
The problem with that strategy is that a large part of the Left's base is single, unattractive women (i.e. feminists). They hate attractive women with a passion. A parallel-- why do you think that no black man married to a white woman ever gets anywhere as a political candidate (at least not in any district with a significant number of black women voters)?

Anonymous Drummergirl November 13, 2016 11:24 AM  

@128 - Thanks Josh - Just saw you're post above.

Blogger James Dixon November 13, 2016 11:29 AM  

> That's a good point, but not the whole picture.

No, it's only a first level approximation, but it's a pretty good one.

> The Sanders win in Michigan should have been an indication of that. I certainly missed it. IIRC, you didn't.

I predicted Trump would reform the Reagan democrats coalition, yes. I didn't know if that would be enough for him to win or not.

Blogger praetorian November 13, 2016 11:38 AM  

This was then used to run monte carlo simulations of all possible outcomes and assign each one a probability. This should be highly accurate based on what we know about probabilities, and the only variable that goes into is the error rate of individual polls, which should be more accurate after averaging.

dh...

I...

Monte Carlo?

Truly, nobody knows anything.

Blogger praetorian November 13, 2016 11:49 AM  

Josh (the gayest thing here) wrote:Except the LA TIMES/USC and IBD/TIPP Tracking, Who, Unlike Us, Actually Got It Right

IBD had Trump +2, LA Times had Trump +3.

Clinton 47.7 Trump 47.1 is the current popular vote according to Dave Wasserman.

https://mobile.twitter.com/Redistrict/status/797566062665547776


Loads of trump voters stayed home in reliably blue states. Here in CA we didn't even have a republican senator on the ballot: it was a Dindu-Hindu female vs Hispanic female. I haven't bothered to look up the results.

Blogger praetorian November 13, 2016 11:50 AM  

Modelling, by and large, is a moderately useful tool that has gained an authority and respectability far in excess of its usefulness.

</thread>

Blogger Dave November 13, 2016 11:50 AM  

Even Bubba saw the handwriting on the wall:

Donald Trump: Bill Clinton Told Me Our Campaign Was ‘One of the Most Amazing He’s Ever Seen’

Blogger RobertDWood November 13, 2016 11:57 AM  

Good summary thank you dh

Blogger kurt9 November 13, 2016 12:11 PM  

Some friends and I had a discussion about people like Nate Silver early this year when Trump was starting to make his mark on the campaign. My friend commented that many of these "pollsters" kept proclaiming that Trump would loose because he would not say and do the things that the pollsters (who are all liberal, BTW) themselves wanted him to do, and specifically singled out Nate Silver as an example. It was our opinion that these pollsters are quite frankly mentally ill because it is mental illness to project in such a way that a particular candidate had to loose because they won't do what you want them to do.

All of these pollster, Nate Silver included, are nothing more than liberals projecting their feelings.

In a related manner, the above is the reason why I thought Robin Hanson's PAM (policy Analytics Market) was such a brilliant idea. It would force all of the liberal (and conservative) pundits to put their money where their mouths are and would rapidly discredit those who predictions failed to match reality.

Blogger Teri November 13, 2016 12:23 PM  

Dh, did you look at Axxiom Strageties polls? They choose a county in each of the battleground states to poll. I think they probably came closer than most of the other polls I saw.

Blogger Teri November 13, 2016 12:25 PM  

Here's the link (clearly I can't spell this morning):

http://axiomstrategies.com/abc/

Blogger Cail Corishev November 13, 2016 12:29 PM  

1. Nate Silver started in sabermetrics, which is the statistical analysis of baseball.

I was just thinking about the comparison to sports. In sports, experts pick games (and other bets) every day. They do this based on massive amounts of information about a small number of athletes and their coaches. They have tons of data on the past performance of the players, coaches, and teams, which can all be cross-references against factors like opposition, location, weather, etc. And with all this information about a fairly small set of players, and the ability to test and correct your assumptions weekly (daily in some sports), if you can pick winners at a 60% rate, you're one of the best and can get rich from it.

In politics, we have no data on most of the "players" (voters), and there may be millions of players we don't even know are playing until the game is over. We're mostly guessing (through our biases) about the factors that will affect the "performance" of the players. Opportunities to test our assumptions against results only come every 2-4 years.

So, why would someone trying to analyze this far less specific, less defined, less testable data be expected to do as well as the 60% of the best sports bettors, or even better than a coin flip? Is there any reason to think they'll do better than someone watching changes in the Narrative and in memes being passed around, and making as objective a guess as he can?

Blogger Francis Parker Yockey November 13, 2016 12:42 PM  

Your problem is that you thinkbthe legacy media is reality.
Ricky Vaughn (@ReturnofRV) did.

Anonymous Just another commenter November 13, 2016 12:43 PM  

@22 - So the bean-counters got it wrong?

@56 - Not sure there is one. But you might try this: "You say we need more non-white immigrants. How do you think whites will react when they finally realize that you are actively trying to make them a powerless, marginalized minority in their own nation? They have a history of being extremely good at organized violence, you know. Are you aiming more at genocide for whites, or immigrants?"

@101 - Excellent.

"Are you pro-choice? Yes? Ah, then you must be racist and hate blacks. You are aware that Margaret Sanger founded Planned Parenthood explicitly as a eugenics program to eliminate blacks by reducing their birth-rate, right? 55 million would-be babies later..."

What else...? Ah. "Are you pro-immigration? Really? Why do you hate African Americans so much? I mean, it's low-wage, low-skill entry-level jobs that illegals from Mexico and further south are taking, keeping blacks here in even more desperate poverty and inciting turf wars between them. Oppressing blacks under the guise of kindness to browns. Sneaky."

Pro no-fault divorce : "Oh, so you really do favor breaking up families. Interesting."

Blogger Teri November 13, 2016 12:47 PM  

Couple of thoughts. One of the reasons that pollsters are wrong is because they don't study American history. They really couldn't think past the 2000 election. (And we know that some pollsters outright lied). I used to read those Theodore White "Making of the President" books back in the day. You can learn a lot about how people behave if you study history.

Pollsters tend to ignore personal reports. I see them as trends. We had people standing in line for hours at a Trump rally. There were reports from people that hadn't voted in years, saying they supported Trump. There were reports of folks in hospice determined to vote. All of that points to the "broken glass voter". There were many reports here and on the Net of people praying for Trump and his family. How many people do you think were praying for Hillary.

Mainly, I was convinced it was a change Election. You can't have polls that say a high percentage of people think the country is on the wrong track, then think they will vote for more of the same. The signs were all there. Pollsters just didn't read it right.

Anonymous BGKB November 13, 2016 12:52 PM  

Would anyone worth talking to ever believe polling after reading the pdf pamphlet WikiLeaks released from Pedosta's email showing how to manipulate polls?

56. GAHCindy

Try to work on the husband /kids.

Blogger kurt9 November 13, 2016 1:08 PM  

Nate Silver is the product of a school system that teaches that 70% is passing and that you get partial credit as well as credit for trying. In the real world, 95-100% is passing, there is no such thing as partial credit, and you don't get credit for trying.

Blogger modsquad November 13, 2016 1:38 PM  

White vs black vs Chinese vs Japanese vs clans vs castes vs past vs present vs me vs you.

God has a sense of humour about the whole thing. The only winning move....

https://www.youtube.com/watch?v=eeOHEU7Ykyg

Blogger S. Misanthrope November 13, 2016 1:51 PM  

You perfectly describe the vast majority of "big data" promoters here. Love it!

Anonymous FP November 13, 2016 2:03 PM  

James Dixon wrote:Oh, and the reason for pushing Manchin? Well, aside from his being a native West Virginian, it's actually in the public interest to have a mainstream functional opposition party. We've seen where 8 years of full Democratic control can get us, and full Republican control will have similar problems if extended for too long a period. Yes, Trump will need a full 8 years to roll back the Obama agenda and get the country back on track. But once that's done we actually need the give and take between two parties with opposing ideas. Someone like Manchin could possibly lead the party back to the mainstream and away from the extreme Marxist path they've been on.

No. As someone who lives 50 miles from rioters who voted for Hillary and are now rioting in their own town, as the cops are told to stand back and do nothing while innocent drivers are attacked in the street. As socialist DA's charge people with crimes for daring to brandish their firearm against angry protestors. As the GOPe cucks worry about how bad burning an effigy of the governor by a right wing gun group looks as the dems/sjws/commies riot in the streets and make death threats (and then wonder why they can't get their own people to support them).

Wipe them out. All of them. They give us no mercy. We give no quarter. They will not leave us alone.

Anonymous Tomas Nau November 13, 2016 3:16 PM  

@15 Agree 100%. We'll see a split between either a Diversity-establishment dem types and Bernie types creating a national green party getting 5-20% of the vote depending on election or perhaps an even more complex mess. Establishment dems, a smaller version of my bernie/green party and a hispanic party that's basically center-left, voting for tough on crime that isn't drug or illegal border crossing related and lots more welfare.

Blogger tublecane November 13, 2016 4:08 PM  

They still love in bubbles, even after they get pierced by reality. Now their bubbles are deflated, like balloons a week after your birthday, but they're still in there.

Maybe I've spent too much time in Progressiveland, but I can appreciate the lefty logic. They got crap for not pumping Hillary enough, and now they deserve a treat. Not a whole cake, because Trump did win, but at least a cupcake.

Anonymous JustAnotherPairOfEyes November 13, 2016 5:17 PM  

VD: "I'm getting very tired of losers who INSIST on responding to rhetoric with dialectic."

This is truly useful advice.

Do you use logic on your children? Really? How'd that work out for you?

How about logical arguments with your pets? You don't do that? Then why are you using logic with liberals?

Anonymous Kevin November 13, 2016 5:29 PM  

Granholm is too far out of politics. If I missed someone else saying it, my guess is someone like Kamala Harris - but that is such an obvious idea it was on Huffington. She certainly has been corrupt enough to do it.

Democrats tend to be more radical in their choices so whoever it is might not even be a known name yet. It does not matter who they choose. Trump has razor thin margins of likability and won against the worst candidate ever. If he does a good job he will win in a landslide as his favorable increase as conservatives come to accept him and normal Americans see that nothing crazy happened and he is not the coming Hitler. If he does an average job he probably has tough job but if he does a bad job he loses in a landslide with primary challengers. So really, what matters is Trump doing a good job as President. I know - it's all obvious.

Anonymous dagwood November 13, 2016 5:45 PM  

"That should be the goal, yes. Is it possible? Not with the state we have now."

It's more like, Not with the population we have now.

Anonymous SciVo de Plorable November 13, 2016 8:03 PM  

VD wrote:You're wrong. The strongest Gamma tell is evasion, rationalization, word salad, and finally going radio silent in order to avoid admitting that one is wrong.

Point granted. We've certainly seen it enough times.

Anonymous SciVo de Plorable November 13, 2016 8:25 PM  

Josh (the gayest thing here) wrote:My extremely early 2020 prediction is that the Democrats are going to counter Trump by running their own famous billionaire celebrity, so they'll try and draft Cuban or Bloomberg. Maybe Oprah.

Tantrum Mark is a good bet. Or maybe not, now that I've branded him. But he was a good bet before.

Anonymous SciVo de Plorable November 13, 2016 8:58 PM  

synp wrote:As soon as you try to build models, you're out of the safety of maths and into the by-definition uncertainty of science.

Thank you for the serious answer.

Now I'm going to LKS every lefty billionaire and other possible celebrity candidate I can think of. If I miss anyone, please make an OT request in the newest thread.

Tantrum Mark [obvious and on-camera]

Neopolitan Bloomberg [confusing on purpose b/c then they stop to think, w/Napoleon overtones b/c short]

Gay Oprah [keep it simple and wrong, because if true would be lesbian not gay; make her deal with it]

Charlie Kanye [both nuts, impossible to refute]

There you go. 4.5 political careers stillborn.

Blogger seeingsights November 13, 2016 8:59 PM  

Yeah, I've been wondering which Democrats will run for President 4 years from now. One that come to mind is Elizabeth Warren. She's somewhere, politically, between Bernie Sanders and Hillary Clinton. Another is Bloomberg, as was previously mentioned. A third might be Andrew Cuomo. Looks like one the weakest fields for the Democrats in 30 years.

Anonymous SciVo de Plorable November 13, 2016 9:18 PM  

Rabbit Warren and Napoleon Bloomberg never had a chance. Of course Rich Bernie and Sick Hillary gave it their best already and failed. Which just leaves AC and I've never heard of him. I don't even know enough to construct a LKS.

Anonymous SciVo de Plorable November 13, 2016 9:56 PM  

Nepo Cuomo. So obvious once I looked him up. Everyone hates nepotism that is locked out of that system.

Blogger tublecane November 13, 2016 11:36 PM  

@140-About sports betting, Joe Schmoe could make a living if it was merely a matter of picking winners. It's the odds, spreads, over-unders, and whatnot that require expertise.

Anonymous Tipsy November 13, 2016 11:50 PM  

VD: Reading your "translations" of Nate Silver clearly demonstrates how much of a prevaricator he is. I'm not surprised: He was a Journolister, after all.

Blogger Tom Kratman November 14, 2016 4:42 PM  

Silver's problem was, I think, two-fold. One part is that he had a system that had, so far, with conditions not changing much (all systems, to work, require that conditions not change much), worked. The other part, probably, is that his system was giving him the answers he wanted to see.

It is not impossible that, in the future, he will discount the polls that were most wrong this year, for their dishonesty, account for their oversampling of democrats, account for the rise of the consciously white voter, plus up the value of the polls that were closest, and thus do better. I wouldn't, in any case, count him out entirely until he has a chance to demonstrate that he's learned his lesson and fixed his no longer useful system.

Post a Comment

Rules of the blog
Please do not comment as "Anonymous". Comments by "Anonymous" will be spammed.

<< Home

Newer Posts Older Posts