Is 538 in the Bag for Hillary?

[Karmadillo]

http://www.counterpunch.org/2016/04/04/is-538-in-the-bag-for-hillary/

APRIL 4, 2016
Is 538 in the Bag for Hillary?
by PETER WHITE

<edit>

Voters are also influenced by polls, which have consistently shown Hillary beating Sanders with three key demographics: seniors, women, and minorities. One of the most widely read poll aggregators with its statistical hand on the political pulse of the nation is Nate Silver’s FiveThirtyEight. Perhaps more than any other prognosticator Silver has repeatedly given Clinton the edge in his stat-driven predictions. It turns out Silver’s figures and methodology are biased in favor of Clinton and his predictions have been wrong in a number of states where Sanders did much better than expected.

I’ve based several articles on Silver’s flawed analyses of the primary races and have predicted a Clinton win, as most MSM pundits have, because Sanders doesn’t have and won’t get the delegates he needs to win in the upcoming big primary contests in New York, Pennsylvania, California, and New Jersey. I’ve never been more pleased to admit that I just may be wrong.

Enter Doug Hatlem, once a Jerry Falwell Baptist in Virginia, then a Mennonite street pastor working with the homeless in Toronto, and now a stay-at-home Dad in Chicago, Hatlem’s peregrinations have quixotically led him to stop and consider numbers: specifically, the old adage about lies, damned lies, and statistics.

Hatlem claims that in 18 of 21 states outside the South, Silver’s predictions have a pro-Clinton bias of 12.5%. He bases his analysis from what polls predicted, on what 538 forecasted, and on the results from those primaries. Silver has been averaging polls to predict primary outcomes and he has also been mapping polls and 538’s predictions to track over time how well candidates are gathering the necessary delegates to capture the nomination. Significantly, Hatlem does something Silver doesn’t do: admit his own bias—for Sanders. He writes that he has been wrong in twelve of eighteen states since Super Tuesday because his predictions had a pro-Clinton bias of 7%. In other words, he “undersold Sanders a bit” just to be safe in his voting forecasts.

<edit>

I suppose it’s possible Wisconsin democrats have massively switched their preference from Clinton to Sanders in just a week, but it’s much more likely that Hatlem is correct when he says the polls and pundits, and politicians have gotten a lot wrong about the Democratic race and they may all be eating crow if Sanders pulls off an upset not only in Wisconsin but elsewhere as well.

more...

[Califonz]

1. Does Mickey Mouse work for Disney?

[notadmblnd]

2. Is 538 part of the establishment? If so

It's in the bag for Hillary

[NanceGreggs]

5. And who is "The Establishment"?

Apparently it's any person, group, organization or union that supports/endorses HRC over BS.

I'm still trying to get my head around the idea that Planned Parenthood is "the man".

[delrem]

11. "The Establishment" is probably the wrong designator.

538 is a for profit business esp. concerned with politics, and these are the first years after Citizens United unleashed a flood of cash input to that kind of business.
That's fact.

Nate Silver will have many many clients in a bidding war - because he made a good guess, once. That good guess won him permanent punditry status and after that it's just been a matter of him choosing clients and managing the money flow.

For some it doesn't matter that some of his recent guesses have been howlers.
For some that information doesn't factor in.

What shows that 538 is a paid political propaganda site is that the main commodity it churns out is editorials - to explain the wizardry.
538 bends statistics to its clients' will.

[NanceGreggs]

16. Sure, okay then. What ev. n/t

[delrem]

18. I can see that you're a deep thinker. n/t

[NanceGreggs]

19. I can see that you believe ...

... I'm interested in your assessment of my thinking.


I'm not.

[delrem]

21. So why did you reply? You say nothing.

You don't have a word to say about 538.

So you haven't a word.

[NanceGreggs]

24. To you?

No, I don't.

[delrem]

27. So why did you reply to me? Why?

To sort of ... spit on me?

[bettyellen]

38. Who is paying him? Facts please, thanks!

[mythology]

50. Calling 99 out of 100 states across 2 different elections is a guess and once?

That's just laughably uninformed. It shows that you actually don't have a clue what Silver does. The guys at unskewthepolls.com also thought Silver had no idea what he was doing. They were wrong too.

[Kalidurga]

3. I only looks that way because of their analysis...

If all they did was explain how they got the numbers they got then I would say they are neutral. But, they have become pundits and they look all kinds of biased.

[itsrobert]

4. 538 = Math, Math is in the bag for Hillary

Sorry Bernie.

[giftedgirl77]

12. Math, logic & reason.

sometimes I think these ppl have to be trolling.

[CalvinballPro]

37. Oops, there goes math, right under the bus! nt

[geek tragedy]

6. it's a freaking poll aggregator, if the polls miss the mark, he missses the mark

how hard is this to understand?

[delrem]

13. NO! In fact he's a predictor.

Especially now, when he has nothing to sell that would be worth buying, he's just a predictor.
And he's been fantastically wrong.

[dchill]

28. +1

[paulthompson]

20. He gives two sets of numbers

He gives two sets of numbers. One set is straight poll averages. The other is what he calls "polls plus," where he adds in his own fudge factor. His polls plus numbers have consistently leaned in Clinton's direction.

[NuclearDem]

25. They've leaned Clinton because turnout and endorsements lean Clinton

[dchill]

29. Suppressed turnout and purchased endorsements.

[NuclearDem]

39. Higher turnout actually helps Clinton.

Caucuses, which suppress turnout, favor Sanders.

[dchill]

42. Let's go with that.

[NuclearDem]

44. 1.5 million people voted in Florida, and Clinton won 2-to-1.

A few thousand go to caucus in Kansas, and Sanders wins in a landslide.

Your candidate depends on voter turnout being low.

[dchill]

45. I've ALWAYS been a big believer in Florida elections.

[paulthompson]

31. Yes, but...

Yes but his "polls plus" have been consistently wrong, as the OP points out.

Endorsements in particular don't seem to matter much in this election, with voters being so angry at the status quo. On the Republican side, they're probably a minus, especially with Trump voters. But Silver hasn't updated his new numbers to adjust. Either he leans Clinton's way, or he's just stubborn and not very good at predictions.

[NuclearDem]

40. No, it hasn't.

It's been wrong in Oklahoma and Michigan on the Democratic side. That's it.

The polls-plus forecast has otherwise accurately predicted the winner of all contests it has predicted for.

[paulthompson]

41. Are you kidding me?

I've been following his website and I've seen just the opposite.

Show me the numbers please. And it's not just win or loss. For instance, Silver predicted there was a 99% chance Clinton would win Illinois, and she did, but only by 1%. That's hardly the sure thing Silver predicted.

[NuclearDem]

43. That's not what probability is.

If there's a 99% chance of rain, and it only sprinkles, does that mean the weather forecast was wrong?

Since you've been following 538, you'd know there's a separate part on those pages for predicting the results--you seem more interested in arguing those numbers.

[RandySF]

7. Math is pretty neutral.

[AgerolanAmerican]

9. spelling is, too!

[JackRiddler]

35. Yes, like any tool it can be used for false purposes.

[AgerolanAmerican]

8. if it is, it's already paid the price

It will never live down Michigan.

Ever.

In one state election they went from the industry standard to just another bunch of guys throwing darts and making wild guesses.

[Dem2]

15. You don't understand statistics

What you're saying makes no sense, none.

[delrem]

17. It's because the pollsters that they depend on

know bottom line what kind of poll numbers their clients most want to see.
This is why they're professional pollsters.

Likewise Nate knows who's offering big dollars for results.

It's post Citizens United paradise.

[SCantiGOP]

10. No

Very simply, no. If you knew their history you would never make such a statement.

[WhaTHellsgoingonhere]

14. From the start. And it was very obvious.

[jfern]

22. Yup, 538 has been rather pro establishment

They've been more hilariously wrong with Trump than Bernie, but quite wrong in both cases.

[NuclearDem]

23. Counterpunch!

Speaking of in the bag...

[Renew Deal]

26. Truth has a liberal bias.

[Jester Messiah]

51. So, not pro-Clinton then? n/t

[Renew Deal]

55. Liberal

Not socialist

[TM99]

30. Does a bear shit in the woods?

Of course there is a bias.

People yammer on about math and 'liberal' truth, yet forgetting the realities of human psychology.

Garbage in and garbage out. If Silvers receives bullshit numbers, he will crunch out bullshit results. It is really all quite simple. If the corporation that now owns 538 is decidedly supportive of Clinton, it will be impossible for him to remain good at his job AND being objectively neutral.

But I await the tiresome and usual suspects who will inform us that we are all stupid, naive, hate math, and live in a fairy tale land.

[Yavin4]

32. 538 just gave Bernie a 73% chance of winning Wisconsin tomorrow

So how is he in the bag for the establishment exactly?

[brooklynite]

33. "Influenced voters" aren't spending a minute's time at 538...

[Agnosticsherbet]

34. Is fact in the bag for Clinton?

LOL

[JackRiddler]

36. Dear god, this thread! The positivism, it burns!

[mhatrw]

46. LOL

[qdouble]

47. Another crazy thread...

Polls are never 100% on the money and 538 mostly bases their predictions on other polling....which would mean you'd have to accuse all pollsters of being anti-Bernie (which I'm sure many of his crazed supporters have no problem doing). From my observation, only one prediction was way off base, but everyone got that one wrong, not just 538.

[leftynyc]

48. Counterpunch?

And I'm guessing you consider them neutral. 538 is based on MATH. Perhaps you want to ignore MATH but I wouldn't think that's a good place to be...especially if you're replacing with with counterpunch.

[Nonhlanhla]

49. No, they're in the bag for math.

Math happens to be in the bag for Hillary.

Geez.

[Jester Messiah]

52. I think 538 is at the mercy of outdated polling methodologies.

They've been badly surprised a few times this season. They don't do any polling themselves, but they do aggregate and average polls taken by others, so if those others screw up then 538 is left holding the bag.

[mythology]

53. I'm sorry, I couldn't get past the first sentence

The article starts off praising that god awful uninformed Margot Kiddor screed that had no basis in reality. At that point, there's nothing left worth reading.

It's just another repeat of the unskewthepolls.com from 2012, just updated to say that the polls are wrong and Sanders is really winning somehow.

[artyteacher]

54. math has a clinton bias yeah... eom

[Depaysement]

56. Not that useful

538 on the Wisconsin Dem primary - March 28 Hillary had an 84% chance of winning. On April 5th Bernie had a 73% chance of winning.

[Orsino]

57. The nature of Sanders' campaign defies easy analysis of numbers...

...which is all 538 is supposed to use.

Sanders' story is one of ever-increasing recognition, with support subsequently trending upward, consequences of his having been a relative unknown less than a year ago. These things make last month's polling data a bit less of a guide to his future performance.

That's good. We've needed peaceful revolution, and Sanders' popularity so far seems linked to a nascent one. The revolution will not be televised, but neither will polls document it in real time.