Nate Silver on MI:Possible Biggest Polling Error in History

[Joe the Revelator]

I said earlier today that I had an intuition Sanders could beat his polling in Michigan tonight, but I didn’t expect things to be quite so close. If Sanders winds up winning in Michigan, in fact, it will count as among the greatest polling errors in primary history. Clinton led by 21.3 percentage points in our final Michigan polling average. Previously, the candidate with the largest lead to lose a state in our database of well-polled primaries and caucuses was Walter Mondale, who led in New Hampshire by 17.1 percentage points but lost to Gary Hart in 1984.

[Motown_Johnny]

1. Translation: I Fucked Up! n/t

[Joe the Revelator]

3. Exactly

[Lucky Luciano]

13. not really

If he has 95% confidence intervals, he is bound to be wrong out of 20 times on average. I don't know what his confidence intervals were though.

[Motown_Johnny]

21. He had a Hillary win at above 99% n/t

[corbettkroehler]

34. Love The .GIF! n/t

[tomm2thumbs]

59. B I N G O ! ! ! !

those are ridiculously WRONG numbers and a complete FAIL on his 'talents'

[BlueStreak]

73. Live by the sword, die by the sword

Polling is getting worse every cycle, for lots of reasons. And in primaries, the polling is doubly bad because it is so thin.

Ultimately the polls may have been right as far as they went. but to get the right answer, you have to also know the turnout of each demographic.

Clinton won the black vote. Sanders has been winning the youth vote. Chances are that Silver failed to account for turnouts when he came up with 99% probability.

The most important point is that Silver's same algorithms are why he forecasts Clinton being the nominee. After tonight, this is not at all certain. Success begets success.

[Nyan]

17. Many times.

[monicaangela]

26. Not the Wizzard.....

Don't tell the Hillary people. We don't want them scurrying around like a balloon with the air escaping.

[Mira]

50. ...

[Adrahil]

37. Well, he doesn't conduct the polls....

In fact, his "demographic model" shows that MI should slightly favor Sanders.

But is is a remarable result! Should make next Tuesday very intersting!

[Motown_Johnny]

40. The way he weighs endorsements along with polls is all him.

He fucked up.

[Adrahil]

44. Well, he has a model, not a crystal ball.

His weighted model still uses polls as the base. But it does count as a miss, no matter what.

[Motown_Johnny]

45. Giving her a 51% chance of winning would have been a miss,

giving her a 99% chance of winning is a fuck up.

[Dem2]

41. Well, to be fair, he doesn't actually do the polling

[Motown_Johnny]

70. Please see post #40 n/t

[Dem2]

72. Bernie was ahead in zero polls before the primary

Even God couldn't have predicted this result. Keep blaming the messenger though, it's kind of entertaining.

[Motown_Johnny]

74. All I did was translate what he said. Besides, this just shows the bias of the polling companies.

They are consistently off by more than their stated margins of error and always in Hillary's favor.


I guess maybe Bernie is smarter than God. At the rally I attended Saturday night Bernie said that he thought he was going to win.


Also, I got a text from the campaign earlier today stating that their internal polling "shows the race is very close".



Shouldn't you be supporting the candidate who is smarter than God?

[Dem2]

75. Yes, there was a post here Monday night that Hillary's internal polling

was showing a much closer race, including the implication that she might lose it.

There were indications of this result, it's still amazing how badly pollsters were off.

I do find the theory about open primaries vs. closed ones interesting.

[Motown_Johnny]

76. So your statement about even god not predicting this result was know to be false when you made it?

Interesting.

[Dem2]

77. Maybe God knew about the internal polling?

It is interesting how much better internal polling is than the dozens of polling companies that do this for a living.

[Motown_Johnny]

78. It would seem that both camps knew pretty much what was going to happen.

It stands to reason that it was something that could have been known.

I stand by my first post. Silver Fucked Up.

[Dem2]

79. I wonder if Nate has access to internal polling?

The whispering of said internal numbers was very close to the primary - at least from what I read.

[randome]

96. Unless you're a statistician, you have no idea if his numbers are good or not.

Clearly they were wrong for Michigan but that does not mean that all his numbers are wrong.
[hr][font color="blue"][center]Aspire to inspire.[/center][/font][hr]

[Motown_Johnny]

97. I didn't say that.

I said that he fucked up. That was in regards to Michigan.


Everyone fighting me on this thread is claiming that he somehow did not fuck up his forecast for Michigan.


They are all crazy.

[mhatrw]

68. What about EVERY OTHER prediction 538 had made outside the South?

Bernie Sanders Has Now Won 9 of the Last 10 States Outside of the South!

This is an undeniable FACT if you include Democrats Abroad. If not, Sanders has won 8 of the last 9 states outside of the South.

Of these 10 states, the only non-Southern state Sanders has lost is Massachusetts, and Sanders supposedly lost this state by just 1.4%.

538 has underestimated Sanders in EVERY state outside of the South. Every single state. More often than not by more than 10%.

[Betty Karlson]

87. "And I shouldn't listen to my corporate sponsors all the time"

Because my work has become shoddier with every time that I 've done so.

[apnu]

100. Props to Nate for being a big boy and owning up to it. (nt)

[johnnyrocket]

2. Yep, this one will make all the polls suspect. Everywhere...

...and make anything possible moving forward.

[think]

5. The polling has been way off. People are going to take note....

[Mnpaul]

62. Their "likely voter" model is off this year

lots of unlikely voters showing up for Bernie.

[PoliticAverse]

84. Indeed. For all the talk about 'math', political polling is about guessing who will actually show up

to vote. And it appears Sanders is motivating people outside the normal guess models.

[BernieforPres2016]

90. There was an enormous disparity between self proclaimed Democrats and Independents

Voters who identified themselves as Democrats were 57-41 for Hillary. Self proclaimed independents were 71-28 for Bernie. The mix was around 2/3 Democrats, 1/3 Independents. There was a big split between white voters (57-41 Bernie) and non white voters (62-38 Hillary). Voters aged 18-44 were 67-32 for Bernie, voters aged 45 and up were 61-36 Hillary.

So if you don't get the mixes on who shows up to vote very close, your projections are going to be garbage.

Here is a link to exit poll data:

http://www.nbcnews.com/politics/2016-election/primaries/MI

[gollygee]

4. It looks like it will at the very least be closer than the polls suggested

However, I've seen a lot of elections in our state and until we have a large percentage of votes from Wayne, Oakland, and Macomb counties, I'm not counting on anything.

[Joe the Revelator]

6. Win or lose, the polling was way off.

[lancer78]

11. Oakland is almost done.

Hillary is barely winning by 2000 votes out of 160K +

[gollygee]

16. And that is promising. :)

I'm hopeful, and I was not expecting to still be hopeful.

[Renew Deal]

7. I thought it was odd that the Clinton and Sanders factions were saying the polls were wrong the last

Few days. They knew something.

[Adrahil]

42. I think so.

A few folks over the last few days got that vibe. Appears they were right.

[Bread and Circus]

8. Yo Nate Silver.... don't underestimate the Bern

[DLnyc]

10. And everyone else, don't overestimate Nate Silver! n/t

[sprts]

49. Yes people - please don't underestimate

the corporate professional propagandist.

[Mnpaul]

65. Oops

Bernie Sanders Could Win Iowa And New Hampshire. Then Lose Everywhere Else.
http://fivethirtyeight.com/datalab/bernie-sanders-could-win-iowa-and-new-hampshire-then-lose-everywhere-else/

[Qutzupalotl]

9. In fairness, it's hard to poll an open primary.

I'm guessing there was some crossover after the Fox News town hall. I thought Bernie presented himself well there and on Sunday. Maybe Rs figure Drumpf has this locked up, or maybe Bernie was more appealing than anyone else.

[SheenaR]

12. >99% chance to win Michigan

I know this thanks to all the posts on Nate Silver this week from HRC's people

[morningfog]

33. Where's that "MATH" poster? Lol.

[Jitter65]

14. Not an error. The more than 15K who switched for the primary is the reason for the difference.

I warned folks about this earlier. The MI election official pointed out the they had over 15K requests to switch parties for this primary.

[Bluenorthwest]

25. Sorry, that math does not work any better than Nate's. Nate did not even think it was close. Nate

should be aware of all such details prior to making his prophecies for profit. I assume that he was. He was simply and massively incorrect, way off base, not at all accurate.

Bernie's current lead is over 26,000. Nate had it all wrong. As do you.

[jeff47]

27. 15k isn't big enough for this shift.

The polls being bandied about were Clinton winning by at least 20. At the moment, it's Sanders by 5. If that holds, 15k is not enough people for a 25% swing.

[democrattotheend]

36. How can one "switch parties" in Michigan?

I thought Michigan had no party registration.

[TTUBatfan2008]

15. No wonder...

...she went with the misleading attack on the auto bailout. Internal polling must have been pretty bad. Would not have needed a cheap attack if she was truly up 20 points.

[arcane1]

28. And the overblown complaining about use of the word "ghetto"

Distract distract distract!

[gollygee]

32. We all remember the time when the bailout was being discussed

And I think everyone in Michigan remembers that it was a difficult and complicated issue. I was put off by how she presented it like some simple and obvious thing. A few years after the fact it might look that way, but I've got to think most of us in Michigan remember that it was way more complicated than that when it was all going on.

[wilsonbooks]

18. Here is a meal for you Nate

[Kalidurga]

82. He can even enjoy a desert

[Bluenorthwest]

19. This cycle is simply dining on the bones of those prone to pronostication...

I saw about half an hour of MSNBC before the polls closed and it was like a wake for Bernie's campaign, what a lot of idiotic shills lined up jawing off the same series of put downs. Chuck Todd ripped into Bernie for being in Florida, where Hillary also is. It was fine for her, he should be in Ohio and Chuck was just disgusted that he was not. Disgusted, shocked. 'Florida?!?!'

Fuck them all they should have no audience left to them, they should survive the remainder of their lives on the cash Comcast has left on their nightstands.

[Arazi]

20. Nothing about this election is normal. People are PISSED!

[jillan]

22. How many polls has he been accurate on during this election?

[Uglystick]

39. So far? Zero.

It's not even close. Nate should quit political polling right now and start guessing for baseball, football, hockey and basketball. Safer money and they don't care.

[Samantha]

23. I thought last night it would be impossible to properly poll Michigan because there were

just too many atypical variables that could sway the numbers at the last minute.

Sam

[Takket]

24. last poll i saw....

on channel 4 Detroit said Clinton was up 62-29 in the polls. i almost didn't bother going to the primary because it was supposed to be a blowout....... this would be a huge win for bernie if he holds on. was up by 19k a few minutes ago, now 24k. lead only getting bigger!

[arcane1]

31. "i almost didn't bother going to the primary because it was supposed to be a blowout"

That is exactly what they're trying to get you to do: not vote. I'm glad you saw through it

[CoffeeCat]

29. There's a pretty cool independent guy who predicted Bernie to win MI 53/46 (article)

It’s a bit unsettling to go against the grain with this forecast. As far as I know, every outlet is projecting a Clinton win tomorrow in both Michigan and Mississippi.

The Sanders campaign must be doing something remarkable in Michigan right now, because the upswing in Sanders popularity among my data sources is undeniable. I am seeing levels of interest in Bernie Sanders in Michigan similar to that of Colorado, Oklahoma, Kansas, and Nebraska. This, along with Michigan’s relatively normal demographic makeup, leads me to personally believe that he does have a chance. It leads my model to estimate that he will win there. Hillary leads every conventional poll, however, which makes me skeptical of these numbers.

Bernie Sanders will be lucky to get above 20% in Mississippi, but I do believe that if he doesn’t win Michigan, the final results will be very close. Here are the numbers:

My official prediction is that Bernie will win Michigan and Hillary will win Mississippi, but in reality Michigan is too close to call with a mathematical model. Elections culminate in a single number after the movement of hundreds or thousands of variables, and as statisticians we can only select a few of those and hope that we account for as much variance as possible. Given the outcome of all the other elections so far this season, the positions of those variables right now in Michigan seem to indicate that a massive upset will happen tomorrow night.
https://tylerpedigo.com/2016/03/08/michigan-mississippi-democratic-primary-projections/

[Jitter65]

38. Yep, he was spent on. nt

[FlatBaroque]

30. It's OK Nate, your new, fat bank account will cushion your hard landing. //

[jalan48]

35. LOL-right-Bernie doesn't have a chance-don't bother voting-get the message?

[billhicks76]

43. ie Polls Rigged Too

[noamnety]

46. I nominate DU's JohnnyRingo for the Crystal Ball Award on this one.

I think he was more correct than he wanted to be. When Hillary wouldn't do the town hall in Michigan, he said "Why would the front runner want more debates? While it's possible she actually has something more important to do, what could she gain in another debate so soon? The underdog candidate always calls for more debates and Town Halls because they have nothing to lose and everything to gain.

Eventually she'll have to face him again, but when Hillary starts calling for more debates you'll know her campaign is in peril."

and right after he posted that, she flipped and said she wanted to do the town hall.

http://www.democraticunderground.com/12511408425#post150

Pretty sure he was onto something.

[PATRICK]

85. Damn right

Politicians actually have a lot of inescapable tells even if the MSM can't see or won't. The whole establishment perception game is a power play against Sanders and especially the agenda he raises. I think we can ignore the garbage discouragement and distraction ploys from the generally ill graced dishonest power brokers. Polls have been more and more a subtle tool used against the very voters they are talking to so scientifically.

And why isn't Sanders lead much larger? The general sea of deception that flows from the local fishwrap, TV and radio and internet outlets is scientifically made to blind a huge chunk of the electorate- but far, far too riskily not to create a tremendous blowback. In place of truth they offer noting of substance. Won't, can't and spit in the suckers' faces even when they accept the game.

[WillyT]

47. Cell Phones... Social Media...

How long will it take them to get it ???

[Ford_Prefect]

48. They missed a simple truth: the world has changed since 2008 and even 2012, and people know it.

The establishment clues don't hold value because they do not represent the majority anymore.

[appalachiablue]

55. Absolutely!

[Kip Humphrey]

51. Nate Silver: never beneath him to explore the depths, twists to correlate Bernie & Gary Hart

[BigBearJohn]

52. Lesson learned: DON'T LISTEN TO POLLS!!! -- Get out and VOTE!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

[JackRiddler]

53. It's not really an error.

1) I'm not talking about the polls are skewed or fixed, although they seem to be.

2) More importantly: Primary season polls are unreliable, if not open bullshit. Turnouts are too small and unreliable. Motivation is everything and cannot be predicted. Of course they're constantly wrong! These problems can't be fixed until we're all wired into brain chips. Which I know many of the data-heads are looking forward to, me not. Nate Silver is not my god. Very impressive his general election accuracy but if he stops promoting himself he'll admit these obvious realities.

[Bluenorthwest]

91. It's too late for that. Nate and many others have presented his work as nearly divine in accuracy

and none of these 'reasons' you now so casually list were ever spoken of by Nate. He never said 'I might be wrong' until he already was wrong. No talk of low turnout making unreliable predictions and more to the point, Nate was correct in Mississippi where turnout was very low and wrong in MI where it was higher.

People who speak with great certainty when they are not really certain will always pay a price in politics. Many DU posters are now cued up to become laughing stocks at best, to be seen as exploitative hacks willing to make use of very important issues to score cheap primary points at worst.

[JackRiddler]

92. Because Nate Silver doesn't admit the problem...

but instead presents these bullshit predictions on data he knows can't possibly do what he is claiming, he is a self-promoting bullshit artist. Or up the bunghole of his past successes. Or skewing for Clinton. But there's no way he doesn't know the limits of primary polls, so he is responsible for misleading.

[Bluenorthwest]

94. Ah, I thought you were defending him but you are saying his 'error' was intentional and that's

very possible. I have noticed that Nate's work claims to be 'THE Math' but it requires lots of language about the numbers to sell his opinion. In my business life I am always skeptical of words being used when numbers are called for.

[JackRiddler]

98. Depends what you are talking about.

Predictions of primaries by polling, especially under 2016 conditions, are obviously unreliable. They are beyond the means of "scientific" polling. Proliferation of samplings and tweaking aggregate models cannot make a difference. It's an object that can't be measured by the proposed tools. This is not something that Silver intends. It is a reality.

It is Silver's decision to apply these unreliable tools anyway, and to sell it as predictive science, and to even make arrogant predictions like "99% certainty," and to aggressively market this. And to do so in an environment in which his word is treated as gospel, and in which his statements are used in attempts to affect the results (as on this board). At best this is unscrupulous, self-aggrandizing, commercial behavior. Whether he is furthermore in the tank and actively skewing for the establishment is another possibility.

[mountain grammy]

54. Fantastic!

Huge turnout=Bernie wins.. turnout, turnout, turnout!

[TexasBushwhacker]

67. And huge turnout is critical for the Dems to win the GE

Other than delegates for the convention, and that is important, Clinton has absolutely ZERO chance of winning Alabama, Georgia, Mississippi, Louisiana, Arkansas and Texas in the GE.

[mountain grammy]

69. Yes, low turnout has been killing us.

[Feeling the Bern]

56. Translation: I am in the HRC tank and I need to save face because I fucked up big time

Maybe his expiration date has come and gone, just like Missouri's pick as the "bellweather" state.

[rury]

57. MSNBC just called it for Sanders

Can you say "big upset??"

WOW!!!

[tomm2thumbs]

60. Yay!

[tomm2thumbs]

58. Nate is so full of it, he has lost ALL credibility -- this is a repeat fail on his part

when you're in the tank, it's hard to see it when you hit the glass <THUNK>

[SoapBox]

61. Ditto!

...what you said!

[reformist2]

64. ^^^ SO TRUE.

[Joe the Revelator]

86. I still like his sports matrices.

But I'm not taking his word for gospel in politics anymore.

[pat_k]

66. THRILLED He Was Wrong!!!

[bbgrunt]

71. polls = propaganda

[silenttigersong]

80. Poor Nate

[Tierra_y_Libertad]

81. So much for the "scientific" and "reality based" polls.

[whatchamacallit]

83. They took a

[Bluenorthwest]

88. Nate trades in a highly curated form of bullshit and he sells it by using very certain language

about his 'math'.

[BernieforPres2016]

89. STFU Nate you fraud

"I had an intuition Sanders could beat his polling"

I think the odds on that are 50.000000000% Nate. What are the odds that your models are Garbage In, Garbage Out? I'd put it at 99.7683452%.

[JackRiddler]

93. Nate Silver knew he was pushing a flawed approach.

Or he's not as smart as advertised.

One or the other.

[Enthusiast]

95. Just my take but....

I think the polling inaccuracies are created to demoralize Bernie supporters.

[riderinthestorm]

99. No shit. He's really damaged his credibility. Not sure how he comes back nt