Nate Silver moved Florida (and North Carolina) to Trump...

[Joe941]

Last edited Tue Nov 1, 2016, 04:56 PM - Edit history (4)

in the now cast and polls plus. Polls only remains blue. Remember folks the only poll that matters is the election day poll. Also Why does 538 polls change so fast?

Update: Polls only has also now turned red.
Update2: Polls only has now turned bad to blue! Why is it changing so much?
Update3: NC goes red in polls-plus forecast.

[Dem2]

1. Apparently Nate was so shocked that DJT won the primary

Even though his sense told him that he couldn't possibly win, he's now erring on the side of extreme caution. I hope he's way too conservative in his analysis.

[Jason1961]

77. Not happening

I live in Miaimi and travel the state regularly, Florida is about to be a Blue State that is as dependable as New York or California

[Dem2]

82. :)

Sounds good to me. Hoping for the same here in NH.

[AgadorSparticus]

107. Are you sure? Why is Nate so spooked then? It seems to me like FL will be blue too

With the latino vote and all. I know nate can be once bit and twice shy, but....this seems kind of drastic....

[getagrip_already]

2. the polls plus is very "predictive" heavy.....

It takes trending into account, as well as news cycle trends.

But the polls have been tightening, and newer polls count heavier than older ones. You can go to his site and see a detailed explanation.

[geek tragedy]

3. that's his model, not him. He programmed the model and those are the numbers

it is spitting out given state and national polls.

Florida leans about 3 points to the right of the country as a whole, and Clinton is up by about 3 points nationally.

And, early voting results have been pretty disappointing. So, there's a very good case to be made that Trump is slightly favored to win Florida.

[Joe941]

5. I'm pretty sure Clinton wins Florida if...

we GOTV.

[geek tragedy]

7. that's the problem, they're not doing very well at GOTV in Florida, especially

with regard to absentee voters (tons of Democrats requested absentee ballots but never returned) and African-Americans (early vote WAY down for African-Americans from 2012).

I know a lot of people there who are busting their asses and lord knows they're doing everything they can.

But the Clinton management team for Florida is not as good at their job as Obama's team was.

[CherokeeDem]

22. One thing to remember about Florida absentee voting...

Anyone can request a ballot, no reason needed. Those voters who have not mailed them back yet, may find the time to early vote or vote on election day.

Well.. one can hope....

[Joe941]

50. I hope you are right.

[CherokeeDem]

95. Me, too!!!!!

[Native]

61. That's what I did last election. Ended up not trusting the process, so voted in person.

[DonCoquixote]

54. ^this

"But the Clinton management team for Florida is not as good at their job as Obama's team was. "

This is what happens when you leave Debbuie Wassermann Schultz in power for too long, and let Human Abedin be stupid enough to point emails where her cyberporn addict husband can try to use her computer.

[geek tragedy]

59. Can't blame DWS for Clinton campaign strategy in FL.

It's mind-boggling that they just kind of assumed black turnout would happen by itself.

[DonCoquixote]

94. and just who the heck was supposed to help Hill will florida

Oh I dunno, a Florida Democrat? And 2000 did not give hints of shenanigans, or Rick Scott? All of this happened when DWS had her iron hands on the wheel.

[Native]

66. How do you know they are not doing as good a job as Obama's team?

You're throwing down a major supposition and not backing it up. As for AA votes being down from 2012, the article on 538 was only able to compare early votes thru this morning to the TOTAL early votes for 2012. There is still time to reach the same thresholds from 2012.

Additionally, the Tampa Bay Times wrote an article critical of Hill's team with regard to registrations (why aren't they at the ball park registering soccer moms?), and lo and behold when it was all over and done with, they exceeded numbers from 2012 and 2008. Fortunately, the TB Times wrote about that too.

Think positively!

[Lefthacker]

4. Hillary

Is still up in Florida according to 538. I haven't seen Nate change it yet.

[Foggyhill]

6. Hes a hack, 60% unsubstantiated bs

With a veneer of stats

When héll lose big in his predictions I don't want to hear about him ever again

Very bad method and s crapload of opinions

Way too much variance in his predictive model

[Capt. Obvious]

27. I was wondering who would be the first to call Nate a hack

Congratulations, you won nothing.

[okieinpain]

46. +1 n/t.

[jcgoldie]

35. Nice bunch of nonsense

Aside from your inability to type coherent sentences, I also got a kick out of the irony of pulling a percentage completely out of your ass in order to criticize a statistical model.

[Foggyhill]

37. My % are as close to truth as his and they were

Clearly opinion unlike his pseudo science

So your comment is just a load of hot air

Thanks for the laugh though

[ScienceIsGood]

45. Great, link to your history of past elections, would love to see your blog! N/t

[ScienceIsGood]

44. Yes, he knows nothing about math and stuff, a complete idiot! N/t

[helpisontheway]

8. I can't believe he would move to Trump instead

of toss up. I listened to the podcast last night and they were playing down Hillary's national lead. They said she does not have a sturdy lead compared to others that have had leads going into presidential campaign. They said that Hillary's national lead is because of strong support in states like California and Texas. Ugh..I hope Hillary is able to hold Florida.

I have been reading Steve Shales posts and he says Florida is tight but Hillary still has a chance. He is basing his info on early votes.

[Cosmocat]

30. That actually has been my feel for a while

and nate explains what you noted in his analysis.

Hill's national lead is getting watered down with larger proportions being in states she has big leads in or some republican states that she can't win, like Texas, that normally the D will lose by 15, but she is only down like 6 percent.

Ohio has been the one swing state she never has been able to get any real traction, and the way things are breaking right now is a bell weather for the "rust belt" surge Trump has been playing all along, and we are seeing Mich and Wisky tightening more than we would be comforable with. Pa is buffered by the eastern part of the state.

[BobbyDrake]

9. Nate Silver has a website competing for traffic in the waning days of the election.

Stunts are to be expected.

[Cosmocat]

31. Dems tried to hide from the numbers in 2010, Rs in 2012 ...

the numbers are the numbers ...

[Demsrule86]

11. I checked...still blue.

Electoral votes by the way...

Hillary Clinton

307.6

Donald Trump

229.1

[apnu]

14. So long as Hillary holds PA, she's got this

She can lose FL and OH and still win. CO will stay blue.

[AngryAmish]

20. Financial professionals only look at 538 and nothing elss

It appeals to their belief in models. Nonsense, but comforting nonsense.

[Foggyhill]

28. Traders are mostly foin glam artists

Relying on suspect models, data, emotions to move markets so not a surprise

[OkSustainAg]

21. I don't see that many people undecided

at this point in the election. I mean Honestly undecided.

[Joe941]

24. this.

[andym]

26. It's not so much the undecided, as it is who is actually going to vote.

Partisans tend to vote-- for example, I assume almost everyone on DU will vote or already has voted. It is the lukewarm supporters that are always the problem.

[Foggyhill]

29. His model gives no clue about

There is no science behind him

[andym]

36. His models do give a clue about likely voters

because most of the polls that he uses for his models ask that specific question about whether you will vote-- thus the difference between registered and likely voters in all these polls. As for science-- statistics itself is scientific, but its use is an art form.

[Capt. Obvious]

58. You should probably stop offering your eloquent opinions on this thread

[Foggyhill]

62. My background is computer engineering and physics

I went a bit more into his use of
Sampling elsewhere and how it relates to the analysis part.

The stat part and proper sampling are two distinct areas

Btw, You are the one making it personal for who knows what reason

[Capt. Obvious]

74. I don't believe you

[Cosmocat]

32. It isn't just "undecideds"

Johnson and Stein votes are peeling off.

Collectively, it is most certainly enough to chip at a candidate who isn't at 50%.

[regnaD kciN]

80. You might think that to be the case…

…but numerous analysts have noted an unusually-high percentage of true undecideds (not Johnson/Stein backers) for this point in an election. It adds a pretty unsettling "crapshoot" aspect to predictions.

[andym]

25. He is using statistical methodology-- I think he is standing by his models

Florida is actually close according to his models. If you can help with the GOTV there, please do.

[Foggyhill]

33. Again, the same type if comments

Doesn't matter how much stats you use if you data is flawed.
The stat parts only works on a true random poll of the underlying pop you attempt to measure
If you don't have that because most polls have massive methodological failings, you got nothing to say in this pop and stats don't apply
Thats my beef with him


Also, adding up a lot of those crap flawed polls is not supported by math at all

This simply bad math and yes got a load of advanced math in my background

Florida is always close but that's ALL you can say and he keeps changing his percentages around

[andym]

38. Statistics used correctly will include a confidence interval-- for example 95%

I'm sure his models do have defined confidence intervals-- not sure where he reports them though-- would be interesting to read more about his methods.

[Foggyhill]

42. It doesn't matter if the sample is not

A random sample of the likely voters

The confidence intervals relies on a certain assumptions about the distribution of the polled data. If the polled data is heavily biased because say you only called land lines, then you don't have a sample of the general population

You get info on these land line people and a margin of error on them. Extrapolating that to gen pop would be wrong because the characteristics of that pop is very different from the general pop (in the pre cell days this wasn't the case)

If you sample 10000 trump voters out of 10000 because you had the worse polling methods, yoûre not getting a 0.1% error margin out of it

Polling methods and stat analysis are not same though they are conflated

[andym]

109. If the polling data is biased then things become tricky

Determining whether you have a fair sample is difficult. That's why Silver uses historical accuracy as one of his criteria for weighting polls. History in this case includes accuracy of polling during the primaries if available. But 538 and other meta sites try to account for bias-- things like under polling if Hispanic speakers etc. In the past including the recent past, meaning the primaries, which were severely under polled compared to the general election, 538 had an excellent record, despite some very notable failures like the Demoratic Michigan primary, which would be a good example to support your view-- however it was just the exception, and not the rule.

Bottom line it is very difficult to attempt to unskew polls based on presumed bias or under polling as Mitt Romney's team found out, and more importantly the meta polling analysis by sites like 538 have worked well in the past, including the recent past, which suggests that enough good data is being gathered to draw accurate conclusions.

[Foggyhill]

110. My view is supported by math and stat

I am not unskewing data; you can't unskewing something when both the bias is unknown and the actual error is unknown

I'm saying people are abusing the polls actual meaning. ThAt this abusers seems to be done deliberatively by the media and pollers.

In fact since methodology is so often so bad, who knows really how skewed it actually is, they certainly don't

If the actual sample is not a random sample of the gen pop of likely voters the margin of error you get is for what you actually sampled, not the general pop

That's the misrepresentation. So, the actual margin of error would be much bigger and who knows what the actual tesult would be.

Getting a good sample is hard but most don't even try

[AgadorSparticus]

108. He's not including data from the outlier polls with the crap methodology is he?

That would explain the false slants and his skewed numbers. I personally think that they threw out shit polls just to muck up the Blue tide we will see nov 8.

[Yo_Mama_Been_Loggin]

39. Both electoral vote.com and Real Clear Politcs have Hillary ahead

Trump will have to sweep all battleground states to win>

http://www.electoral-vote.com/

http://www.realclearpolitics.com/epolls/2016/president/2016_elections_electoral_college_map.html

[Motley13]

40. I was not pleased with what I read in the paper this morn

votes cast
dem 1.5 mil 30.77% of registered dems
rep 1.5 mil 33.18% of registered reps

[0rganism]

81. iirc, Republicans usually *lead* in FL early voting, and not by a small margin

for HRC to be running even in FL early votes is actually good news

[Native]

98. Souls to the polls is this Sunday!

[Ligyron]

41. Hills will win FL.

There's no way enough people here will vote for Trump. He's too vulgar and crass to be accepted as our President

[Joe941]

43. Update: Polls only has also now turned red.

[Farmgirl1961]

47. yeah, just saw that FL is in the light red column for all three predictors

Suppose that might change after more early voting -- especially this weekend's drive?

It doesn't make sense with the huge Latino population there polling for Hillary. I know Hillary doesn't need Florida, but man -- it'd be nice to not have it go Trump.

[Joe941]

49. Back voter turnout is down according to 538...

I hope that turns around.

[Farmgirl1961]

51. I had heard that as well...but heard that it might be due to limited polling locations

And this weekend was the big "Souls to the Polls" turn out. Does that start Friday or is it Sat/Sun? What's the weather forecast looking like for the upcoming weekend in Florida (I'm sure it depends on location)?

[blm]

92. That was GOP cutting early vote days and locations.

We have to support swing states in their GOTV efforts. NC race, for example, is crucial to who controls Senate.

[BadDog40]

75. I don't think Nates model takes early voting into its analysis

[Farmgirl1961]

53. I want to ask a serious question with seemingly coming across as being overly "concerned" or a troll

We can see that Nate Cohen's 538 site is way more volatile than the majority of other polling sites. The trend during the last week+ has been that Hillary is "losing" a bit of ground and Trump is "gaining" a bit of ground.

Over the course of the next days, as more people engage in early voting and we continue to get more information and of course lots and lots and lots of polling numbers (good, bad, ugly, whatever), what do any of you who follow the statistics and polls and know a heck of a lot more than me and lot of us at DU predict to see on 538?

I ask more so I can be more prepared -- i.e. if the numbers continue on their current trajectory. On the other hand, I'd be thrilled to hear that there will be a "bottoming out" somewhere along the line as all the "undecideds" figure out where they stand.

[Foggyhill]

56. The variations have mostly been eithn margins of error

Shes never been more than 4% up in Florida

Most times 1-3%

Polls done have 3-4.5% margins of error

If you had 10 polls showing him with a higher scores you can probably assume hés ahead. But that would not be certain

[regnaD kciN]

103. A minor clarification…

…the head of 538 is Nate SILVER. Nate COHN runs the New York Times "Upshot" projection service.

[Beartracks]

55. "Don't worry!" say the confidence trolls.

The idea that any expression of concern is unwarranted can skew fellow Dems into thinking, "Hey, it's no biggie if I can't make it to the polls next week, Hillary's got it in the bag anyway."

Run like hell, everyone. Run like hell to the polling places on Election Day. We need Every Vote to send Trump, and a bunch of Repub blowhards, home. Leave no room for spin, or doubt, or statistical whining. Let's win BIG.

============

[Farmgirl1961]

60. ABSOLUTELY!!! Vote like your life depends on it

Because it does. This is not an election to sit out (none of them are...ever...I've never not voted).

[DonCoquixote]

57. I better not hear

Talks of "the blacks did not get out to vote!" When some of us Floridians were warning everyone all along that Rick Scott and his barbie bribe taking attorney general were going to try every dirty trick in the book to make sure blacks and other minorities could not vote.

Florida should have been dealt with 16 years ago, but of course, we had a Florida person running the campaign, the same person who kept pushing Charlie Crist on us.

We Warned YOU

and we warn you now, if you try to scapegoat us again, or do the inevitable "why can't we get rid of Florida" posts, than you can enjoy running with low wallets.

[blm]

65. SEND GOTV DONATIONS TO NC - that is where HRC wins and Senate goes Democratic.

Game over for Trump and McConnell's leadership.

[Farmgirl1961]

68. To whom or where in NC? I'd be happy to send some $$ their way

[blm]

69. Deborah Ross is senate candidate now tied with Burr.

https://www.deborahross.com

GOTV for Mecklenburg County Dem party has the most targeted GOTV machine in the Charlotte area where most of the Dem votes are in NC.

http://www.meckdem.org

[Farmgirl1961]

76. Done!!!

Let's make this a separate link so other DU folks know where to find this thread and can contribute to Deborah Ross.

[blm]

78. Just posted - thanks for the prompt.

.

[chimpymustgo]

90. Just donated. How about a thread with links to Senate races we need to contribute to.

[blm]

91. Good idea. Go for it, kiddo - post a thread.

I think it's NC, Florida, Pennsylvania, Missouri and Nevada that are close.

What a difference it will make in those states to help them with this final push to GOTV. Those blue ballots and gas cards don't pay for themselves.

[LAS14]

70. It's 3 hours later and it's blue. (Lightest blue, yes, but blue).

[LAS14]

71. Oh. It's the difference between polls only and nowcast and polls plus. Damn.

[Joe941]

72. How come it changes so much?

[nolabels]

73. It has been said three days before the election people should ignore polling results

Well this time it looks more than week should be adhered to.

Those people who think they pull the strings need to keep it close and have pulled out all stops quite early.

Just think how they are, some might even get replaced at their jobs because it. I always try to look at the other side's point of view before going to the reaction button

[0ccy01]

79. Not a troll but Clinton will not win FL

I live here and there is no enthusiasm for Hillary. Lots of love for Trump especially along the I-4 corridor that won Florida for Obama, a terrible sign for H. I hate living here

[moonscape]

85. I'm confused. Why are you smiling when you say that, and why are you not

to be confused with a concern troll?

[Joe941]

83. Update: NC turned red for the polls plus forecast.

polls only and now cast remain in the blue column. What data is being used for these changes?

[MadBadger]

93. Survey USA has it +7 Trump today in NC

[obamanut2012]

101. hahaha never

[Donald Ian Rankin]

84. It's moved from very pale blue to very pale pink. The 0 point isn't actually that significant.

The difference between "Trump will definitely win 49.9% of the vote in Florida" and "Trump will definitely win 50.1% of the vote in Florida" would be immensely significant.

But what Silver's models are saying is that we've gone from "Trump has a 49.9% chance of winning Florida" to "Trump has a 50.1 % chance of winning Florida" (or whatever the numbers are). And the difference there isn't terribly significant.

[Joe941]

86. Good point.

[LAS14]

87. At 5:17, all measures are in Hillary's favor.

[budkin]

89. NC was because of that bullshit Survey USA poll that has Trump +7

And they are highly weighted by 538

[StevieM]

96. I hope you are right but it seems clear that the race has tightened greatly since Comey and Chaffetz

Last edited Tue Nov 1, 2016, 10:48 PM - Edit history (1)

committed their illegal acts.

Hillary was on pace to win by 10 points and with 375 electoral votes. She was likely to win Arizona IMO, and possibly even Georgia. We were going to have a substantial margin in the Senate and possibly even challenge for control of the House.

Then the abuses of power happened. And now the race is much closer then we could have imagined. And I do fear that North Carolina may go to Trump.

I pray that Survey USA poll is as flawed as you say it is.

For the record, I am not just handwringing. I canvassed both days last weekend. I am doing my part to beat Trump.

[obamanut2012]

102. Trump is not winning NC by 7

[triron]

97. That's because of the Survey USA poll today of NC

[obamanut2012]

99. Poor OP, they are blue again

At least a nice Carolina blue.

[okieinpain]

99. im not a conspiracy believer but i dont

think that trump won. i believe there is something going on and I don't think even if the dnc knew and had proof that they would expose what happen for fear of creating a panic.

but there is something wrong, how can trump win NC, but the Dem candidate win the Gov Chair. didn't the polls pick that race correctly but it was wrong with clinton / trump.

[link:http://www.realclearpolitics.com/epolls/2016/governor/nc/north_carolina_governor_mccrory_vs_cooper-4096.html|

[DarthDem]

104. Oh Noes!!!

This particular concern has been washed away already. Time to move on to another set of pearls!

[ham_actor]

106. Volatility?

Wasn't only a week ago that 538 had Clinton winning Ohio and Iowa as well as being comfortably ahead in Florida and North Carolina? Now all four of those states have gone to Trump? Whatever the validity of the statistical models 538 uses elections in the real world are not subject to the kind of volatile shifts that 538 portrays. Are there really 5 to 10% of the electorate who change their minds and their votes that quickly and easily? I might add that 538 is the only one that has been subject to such volatility. PEC, The New York Times-Upshot, Huff Post and others have been much more consistent. Though Clinton's chances of winning have decreased from 93 to 88% in the last week in the Upshot estimate for instance, this is a far cry from more than 14 point drop in her chances from 538. Frankly I don't know which is correct but I think that 538 is immersed in a statistical universe and isolated from political reality.