politicalbetting.com » Blog Archive » As WH2016 moves into the final straight: The PB/Polling Matter

System · November 2016

politicalbetting.com » Blog Archive » As WH2016 moves into the final straight: The PB/Polling Matters Podcast team looks in detail at the polling & what might happen

In this week’s episode of the PB/Polling Matters podcast Keiran Pedley (@keiranpedley), Leo Barasi (@leobarasi) and Rob Vance (@robvance) look at the latest US election polling and discuss whether the they are tightening.

Read the full story here

foxinsoxuk · November 2016

First like Hillary!

IanB2 · November 2016

Second like Zac

IanB2 · November 2016

I wish Rob could overcome his verbal tic of saying 'you know' so very often; it spoils what is otherwise interesting insight.

Sandpit · November 2016

Grr, too slow. 4th like the Lid Dems.

DecrepitJohnL · November 2016

No, there could not be a big surprise. President Trump is priced at 5/2 which is about 29, call it 30 per cent, so only a small surprise.

ydoethur · November 2016

DecrepitJohnL said:
No, there could not be a big surprise. President Trump is priced at 5/2 which is about 29, call it 30 per cent, so only a small surprise.

6th like Boris Johnson in the Tory leadership contest, or Donald Trump in a British election with 5 candidates.

JennyFreeman · November 2016

President Trump is starting to look increasingly plausible. He has the Big Mo and what a time to have it.
http://fivethirtyeight.com/features/election-update-yes-donald-trump-has-a-path-to-victory/
http://news.sky.com/story/trump-claims-momentum-as-he-takes-lead-in-presidential-poll-10641469

This from Nate: "Whenever the race tightens, we get people protesting that the popular vote doesn’t matter because it’s all about the Electoral College, and that Trump has no path to 270 electoral votes. But this presumes that the states behave independently from national trends, when in fact they tend to move in tandem."

nunu · November 2016

"Note - Large split ticket vote favours Clinton. 28% of Florida GOP voters went Clinton. Follows Marist early polling analysis of Florida."
.
Jack no way this can be right?? 10% maybe but over a quarter? They must have their modelling wrong.

IanB2 · November 2016

Surely this is a big misjudgement by Suzanne Evans; an appalling reason for anyone to go into teaching?:

"Ukip activists should sign up to train as teachers so that they can influence what children are taught from an early age, one of the frontrunners for the party’s leadership has said."

Chris · November 2016

JennyFreeman said:
President Trump is starting to look increasingly plausible. He has the Big Mo and what a time to have it.
http://fivethirtyeight.com/features/election-update-yes-donald-trump-has-a-path-to-victory/
http://news.sky.com/story/trump-claims-momentum-as-he-takes-lead-in-presidential-poll-10641469

This from Nate: "Whenever the race tightens, we get people protesting that the popular vote doesn’t matter because it’s all about the Electoral College, and that Trump has no path to 270 electoral votes. But this presumes that the states behave independently from national trends, when in fact they tend to move in tandem."

Just for a bit of balance, here's the paragraph immediately before that:
Our model takes all this data in stride, along with all the other polls that nobody pays much attention to. And it thinks the results are most consistent with a 3- or 4-percentage point national lead for Clinton, down from a lead of about 7 points in mid-October. Trump remains an underdog, but no longer really a longshot: His Electoral College chances are 29 percent in our polls-only model — his highest probability since Oct. 2 — and 30 percent in polls-plus.

Innocent_Abroad · November 2016

Chris said:

JennyFreeman said:
President Trump is starting to look increasingly plausible. He has the Big Mo and what a time to have it.
http://fivethirtyeight.com/features/election-update-yes-donald-trump-has-a-path-to-victory/
http://news.sky.com/story/trump-claims-momentum-as-he-takes-lead-in-presidential-poll-10641469

This from Nate: "Whenever the race tightens, we get people protesting that the popular vote doesn’t matter because it’s all about the Electoral College, and that Trump has no path to 270 electoral votes. But this presumes that the states behave independently from national trends, when in fact they tend to move in tandem."
Just for a bit of balance, here's the paragraph immediately before that:
Our model takes all this data in stride, along with all the other polls that nobody pays much attention to. And it thinks the results are most consistent with a 3- or 4-percentage point national lead for Clinton, down from a lead of about 7 points in mid-October. Trump remains an underdog, but no longer really a longshot: His Electoral College chances are 29 percent in our polls-only model — his highest probability since Oct. 2 — and 30 percent in polls-plus.

A more interesting question is - will Trump ever shake HRC's hand and wish her well? And if not, how long before the first assassination attempt?

JennyFreeman · November 2016

Chris:
Yep but that's also based on a polling average which includes polls taken before most of the Trump momentum. I've never found averaging a very reliable method. There's no doubt that Donald Trump is in the ascendancy with the Big Mo.

SouthamObserver · November 2016

Innocent_Abroad said:

Chris said:

JennyFreeman said:
President Trump is starting to look increasingly plausible. He has the Big Mo and what a time to have it.
http://fivethirtyeight.com/features/election-update-yes-donald-trump-has-a-path-to-victory/
http://news.sky.com/story/trump-claims-momentum-as-he-takes-lead-in-presidential-poll-10641469

This from Nate: "Whenever the race tightens, we get people protesting that the popular vote doesn’t matter because it’s all about the Electoral College, and that Trump has no path to 270 electoral votes. But this presumes that the states behave independently from national trends, when in fact they tend to move in tandem."
Just for a bit of balance, here's the paragraph immediately before that:
Our model takes all this data in stride, along with all the other polls that nobody pays much attention to. And it thinks the results are most consistent with a 3- or 4-percentage point national lead for Clinton, down from a lead of about 7 points in mid-October. Trump remains an underdog, but no longer really a longshot: His Electoral College chances are 29 percent in our polls-only model — his highest probability since Oct. 2 — and 30 percent in polls-plus.
A more interesting question is - will Trump ever shake HRC's hand and wish her well? And if not, how long before the first assassination attempt?

Much more interesting will be the Obama/Trump chemistry at the latter's inauguration. And will the 45th President really seek the imprisonment of his defeated rival?

With Congress on his side and the Supreme Court falling into his lap Trump will be among the most powerful US presidents there has ever been.

ydoethur · November 2016

IanB2 said:
Surely this is a big misjudgement by Suzanne Evans; an appalling reason for anyone to go into teaching?:

"Ukip activists should sign up to train as teachers so that they can influence what children are taught from an early age, one of the frontrunners for the party’s leadership has said."

The thought is worrying but the stupidity is worse. As a teacher I have comparatively little influence over what I can and can't teach. If she wants to influence the curriculum, she should aim at the DfES, OFQUAL and OFSTED, only one of which actually employs teachers.

If she is so dense as that, she should fit right in at the DfES...

Sandpit · November 2016

IanB2 said:
Surely this is a big misjudgement by Suzanne Evans; an appalling reason for anyone to go into teaching?:

"Ukip activists should sign up to train as teachers so that they can influence what children are taught from an early age, one of the frontrunners for the party’s leadership has said."

She's only railing against the influence of the 'Common Purpose' types in education. To UKIP's membership this probably resonates more than it does to the general public, it was certainly something they were talking about in the run up to the referendum.

CD13 · November 2016

Mr B2,

"Ukip activists should sign up to train as teachers so that they can influence what children are taught from an early age."

Although many teachers don't often bring their politics into the classroom, I could have guessed the voting patterns of several of my teachers at grammar school in the 1960s. In that era, they were generally for Queen, country and the Conservatives.

As I switched to science as quickly as possible, it faded quickly, but in the Micky Mouse sciences (sociology et al), it can become a matter of personal interpretation. The ones I bumped into at university from 1967 onwards were often obvious left-wingers, but that's less important. We weren't too bothered about safe spaces.

I meet young family members and their friends and some do echo the prevailing ethos as if it were the holy grail. I assume many will grow out of it anyway.

foxinsoxuk · November 2016

nunu said:
"Note - Large split ticket vote favours Clinton. 28% of Florida GOP voters went Clinton. Follows Marist early polling analysis of Florida."
.
Jack no way this can be right?? 10% maybe but over a quarter? They must have their modelling wrong.

I think there must be a lot of Hispanic GOP voters in Florida.

logical_song · November 2016

JennyFreeman said:
Chris:
Yep but that's also based on a polling average which includes polls taken before most of the Trump momentum. I've never found averaging a very reliable method. There's no doubt that Donald Trump is in the ascendancy with the Big Mo.

Nate has Trump now winning Florida, Ohio and North Carolina, where Hillary used to be ahead - but he still loses by 60 electoral college votes.

IanB2 · November 2016

CD13 said:
Mr B2,

"Ukip activists should sign up to train as teachers so that they can influence what children are taught from an early age."

Although many teachers don't often bring their politics into the classroom, I could have guessed the voting patterns of several of my teachers at grammar school in the 1960s. In that era, they were generally for Queen, country and the Conservatives.

As I switched to science as quickly as possible, it faded quickly, but in the Micky Mouse sciences (sociology et al), it can become a matter of personal interpretation. The ones I bumped into at university from 1967 onwards were often obvious left-wingers, but that's less important. We weren't too bothered about safe spaces.

I meet young family members and their friends and some do echo the prevailing ethos as if it were the holy grail. I assume many will grow out of it anyway.

The point however is that encouraging people to go into teaching so they can influence young kids in a UKIP-direction is an appalling thing for a politician to be suggesting.

Theuniondivvie · November 2016

IanB2 said:
Surely this is a big misjudgement by Suzanne Evans; an appalling reason for anyone to go into teaching?:

"Ukip activists should sign up to train as teachers so that they can influence what children are taught from an early age, one of the frontrunners for the party’s leadership has said."

Jeezo, and she's supposed to be the credible one.

foxinsoxuk · November 2016

IanB2 said:

CD13 said:
Mr B2,

"Ukip activists should sign up to train as teachers so that they can influence what children are taught from an early age."

Although many teachers don't often bring their politics into the classroom, I could have guessed the voting patterns of several of my teachers at grammar school in the 1960s. In that era, they were generally for Queen, country and the Conservatives.

As I switched to science as quickly as possible, it faded quickly, but in the Micky Mouse sciences (sociology et al), it can become a matter of personal interpretation. The ones I bumped into at university from 1967 onwards were often obvious left-wingers, but that's less important. We weren't too bothered about safe spaces.

I meet young family members and their friends and some do echo the prevailing ethos as if it were the holy grail. I assume many will grow out of it anyway.
The point however is that encouraging people to go into teaching so they can influence young kids in a UKIP-direction is an appalling thing for a politician to be suggesting.

It is, but at the mement she just needs support from kipper members. This is not an election by the general public.

I have a fiver laying Nuttall, as kipper elections are full of incident that affects frontrunners!

SquareRoot · November 2016

Where are all those posters maintaining that the pounds crash will not affect the UK or inflation...

http://www.bbc.co.uk/news/business-37838087

Morris_Dancer · November 2016

Good morning, everyone.

Agree that Evans teaching comment is disturbing but it's worth remembering that foster children were taken from parents because they supported UKIP (the parents, not the kids...), and that BNP members have been banned as teachers in the past.

Sadly, political views have had an influence when it comes to children and their upbringing and who ought to do it long before Evans made her ill-advised comments.

DavidL · November 2016

JennyFreeman said:
President Trump is starting to look increasingly plausible. He has the Big Mo and what a time to have it.
http://fivethirtyeight.com/features/election-update-yes-donald-trump-has-a-path-to-victory/
http://news.sky.com/story/trump-claims-momentum-as-he-takes-lead-in-presidential-poll-10641469

This from Nate: "Whenever the race tightens, we get people protesting that the popular vote doesn’t matter because it’s all about the Electoral College, and that Trump has no path to 270 electoral votes. But this presumes that the states behave independently from national trends, when in fact they tend to move in tandem."

This is what I have been saying. There is a whole crop of states, Fl, Az, NC, OH, Col, that are very close and follow the national trend. If Trump gains another 2% or so they all go into his column. That makes it very close with smaller states like Nev, NH and Iowa then becoming important.

If that movement doesn't come Clinton still wins big. If it does its going to be a long night.

SquareRoot · November 2016

IanB2 said:

CD13 said:
Mr B2,

"Ukip activists should sign up to train as teachers so that they can influence what children are taught from an early age."

Although many teachers don't often bring their politics into the classroom, I could have guessed the voting patterns of several of my teachers at grammar school in the 1960s. In that era, they were generally for Queen, country and the Conservatives.

As I switched to science as quickly as possible, it faded quickly, but in the Micky Mouse sciences (sociology et al), it can become a matter of personal interpretation. The ones I bumped into at university from 1967 onwards were often obvious left-wingers, but that's less important. We weren't too bothered about safe spaces.

I meet young family members and their friends and some do echo the prevailing ethos as if it were the holy grail. I assume many will grow out of it anyway.
The point however is that encouraging people to go into teaching so they can influence young kids in a UKIP-direction is an appalling thing for a politician to be suggesting.

The teaching profession(state) is stuffed full of lefties .. I doubt the espouse Tory policies in their thinking, as do teachers in public school

JackW · November 2016

nunu said:
"Note - Large split ticket vote favours Clinton. 28% of Florida GOP voters went Clinton. Follows Marist early polling analysis of Florida."
.
Jack no way this can be right?? 10% maybe but over a quarter? They must have their modelling wrong.

I accept that 28% is certainly too high but a significant split vote will be important in Florida. One other point to note the poll shows Rubio +6 so this is not an outlier in the normal sense.

Another point is the Hispanic spike. Previously they have underperformed their demographic. Not so this time. If crosstabs are provided look to see the likely voter screen.

We certainly require more evidence.

PlatoSaid · November 2016

This is insane - Podesta's lawyer is apptd to investigate emails by DOJ.

foxinsoxuk · November 2016

SouthamObserver said:

Innocent_Abroad said:

Chris said:

JennyFreeman said:
President Trump is starting to look increasingly plausible. He has the Big Mo and what a time to have it.
http://fivethirtyeight.com/features/election-update-yes-donald-trump-has-a-path-to-victory/
http://news.sky.com/story/trump-claims-momentum-as-he-takes-lead-in-presidential-poll-10641469

This from Nate: "Whenever the race tightens, we get people protesting that the popular vote doesn’t matter because it’s all about the Electoral College, and that Trump has no path to 270 electoral votes. But this presumes that the states behave independently from national trends, when in fact they tend to move in tandem."
Just for a bit of balance, here's the paragraph immediately before that:
Our model takes all this data in stride, along with all the other polls that nobody pays much attention to. And it thinks the results are most consistent with a 3- or 4-percentage point national lead for Clinton, down from a lead of about 7 points in mid-October. Trump remains an underdog, but no longer really a longshot: His Electoral College chances are 29 percent in our polls-only model — his highest probability since Oct. 2 — and 30 percent in polls-plus.
A more interesting question is - will Trump ever shake HRC's hand and wish her well? And if not, how long before the first assassination attempt?

Much more interesting will be the Obama/Trump chemistry at the latter's inauguration. And will the 45th President really seek the imprisonment of his defeated rival?

With Congress on his side and the Supreme Court falling into his lap Trump will be among the most powerful US presidents there has ever been.

I agree with your second point. POTUS Clinton would be significantly restrained by a Republican Congress, POTUS Trump would have control of Congress and SC too, with very little restraint. Trade war and nuking Raqqa, here we come!

Pulpstar · November 2016

SquareRoot said:
Where are all those posters maintaining that the pounds crash will not affect the UK or inflation...

http://www.bbc.co.uk/news/business-37838087

Pulpstar said:

taffys said:
You can't eat sovereignty.

Oooh a forecaster. Like the ones who got the immediate aftermath of Brexit totally, utterly and completely wrong. Massive recession, wasn;t it? higher interest rates? need I go on?

Good things come to those who wait.

matt · November 2016

SquareRoot said:

IanB2 said:

CD13 said:
Mr B2,

"Ukip activists should sign up to train as teachers so that they can influence what children are taught from an early age."

Although many teachers don't often bring their politics into the classroom, I could have guessed the voting patterns of several of my teachers at grammar school in the 1960s. In that era, they were generally for Queen, country and the Conservatives.

As I switched to science as quickly as possible, it faded quickly, but in the Micky Mouse sciences (sociology et al), it can become a matter of personal interpretation. The ones I bumped into at university from 1967 onwards were often obvious left-wingers, but that's less important. We weren't too bothered about safe spaces.

I meet young family members and their friends and some do echo the prevailing ethos as if it were the holy grail. I assume many will grow out of it anyway.
The point however is that encouraging people to go into teaching so they can influence young kids in a UKIP-direction is an appalling thing for a politician to be suggesting.
The teaching profession(state) is stuffed full of lefties .. I doubt the espouse Tory policies in their thinking, as do teachers in public school

There may be a confusion here between the public face, through the NUT, and teaching staff generally.

Chris · November 2016

JennyFreeman said:
Chris:
Yep but that's also based on a polling average which includes polls taken before most of the Trump momentum. I've never found averaging a very reliable method. There's no doubt that Donald Trump is in the ascendancy with the Big Mo.

As I said, I quoted that for balance, as Nate Silver's view wasn't evident from the bit of his article you chose to quote.

rottenborough · November 2016

foxinsoxuk said:

SouthamObserver said:

Innocent_Abroad said:

Chris said:

JennyFreeman said:
President Trump is starting to look increasingly plausible. He has the Big Mo and what a time to have it.
http://fivethirtyeight.com/features/election-update-yes-donald-trump-has-a-path-to-victory/
http://news.sky.com/story/trump-claims-momentum-as-he-takes-lead-in-presidential-poll-10641469

This from Nate: "Whenever the race tightens, we get people protesting that the popular vote doesn’t matter because it’s all about the Electoral College, and that Trump has no path to 270 electoral votes. But this presumes that the states behave independently from national trends, when in fact they tend to move in tandem."
Just for a bit of balance, here's the paragraph immediately before that:
Our model takes all this data in stride, along with all the other polls that nobody pays much attention to. And it thinks the results are most consistent with a 3- or 4-percentage point national lead for Clinton, down from a lead of about 7 points in mid-October. Trump remains an underdog, but no longer really a longshot: His Electoral College chances are 29 percent in our polls-only model — his highest probability since Oct. 2 — and 30 percent in polls-plus.
A more interesting question is - will Trump ever shake HRC's hand and wish her well? And if not, how long before the first assassination attempt?

Much more interesting will be the Obama/Trump chemistry at the latter's inauguration. And will the 45th President really seek the imprisonment of his defeated rival?

With Congress on his side and the Supreme Court falling into his lap Trump will be among the most powerful US presidents there has ever been.

I agree with your second point. POTUS Clinton would be significantly restrained by a Republican Congress, POTUS Trump would have control of Congress and SC too, with very little restraint. Trade war and nuking Raqqa, here we come!

Morning all,

What a bleak prospect before us. Let's hope the 30% chance remains at that level.

Chris · November 2016

And I'm very glad that JennyFreeman raised the question of polling averages, because I've just prepared a little round-up of selected polls released in the last 24 hours - all concluded at least 2 days after the FBI intervention. The reason I've done this is specifically to illustrate the importance of using averages rather than selected polls!

National OCT. 20-NOV. 1 RAND (American Life Panel) (2,269) C: 44% T: 35% Clinton +9

Arizona OCT. 25-31 Google Consumer Surveys (656) C: 38% T: 33% Clinton +5
Colorado OCT. 25-31 Google Consumer Surveys (637) C: 43% T: 28% Clinton +15
Florida OCT. 25-30 TargetSmart/William & Mary (718) C: 48% T: 40% Clinton +8
Iowa OCT. 25-31 Google Consumer Surveys (429) C: 38% T: 30% Clinton +8
Maine’s 2nd District OCT. 28-30 Emerson College (375) C: 44% T: 42% Clinton +2
Michigan OCT. 31 Mitchell Research & Communications (737) C: 50% T: 43% Clinton +7
Minnesota OCT. 25-31 SurveyMonkey (773) C: 46% T: 37% Clinton +9
Nevada OCT. 27-31 Auto Alliance / ESA / Pulse Opinion Research (525) C: 44% T: 40% Clinton +4
New Hampshire OCT. 25-31 SurveyMonkey (659) C: 48% T: 35% Clinton +13
North Carolina OCT. 25-31 SurveyMonkey (1,574) C: 47% T: 44% Clinton +3
Ohio OCT. 27-31 Auto Alliance / ESA / Pulse Opinion Research (525) C: 47% T:41% Clinton +6
Pennsylvania OCT. 26-30 Franklin & Marshall College (652) C: 49% T: 38% Clinton +11
Virginia OCT. 25-31 SurveyMonkey (2,089) C: 48% T: 38% Clinton +10
Wisconsin OCT. 25-31 Google Consumer Surveys (816) C: 41% T: 29% Clinton +12

SquareRoot · November 2016

matt said:

SquareRoot said:

IanB2 said:

CD13 said:
Mr B2,

"Ukip activists should sign up to train as teachers so that they can influence what children are taught from an early age."

Although many teachers don't often bring their politics into the classroom, I could have guessed the voting patterns of several of my teachers at grammar school in the 1960s. In that era, they were generally for Queen, country and the Conservatives.

As I switched to science as quickly as possible, it faded quickly, but in the Micky Mouse sciences (sociology et al), it can become a matter of personal interpretation. The ones I bumped into at university from 1967 onwards were often obvious left-wingers, but that's less important. We weren't too bothered about safe spaces.

I meet young family members and their friends and some do echo the prevailing ethos as if it were the holy grail. I assume many will grow out of it anyway.
The point however is that encouraging people to go into teaching so they can influence young kids in a UKIP-direction is an appalling thing for a politician to be suggesting.
The teaching profession(state) is stuffed full of lefties .. I doubt the espouse Tory policies in their thinking, as do teachers in public school
There may be a confusion here between the public face, through the NUT, and teaching staff generally.

On reflection I guess you are right, in the same way the BMA has been taken over by a bunch of lefties.

IanB2 · November 2016

Arron Banks saying he doesn't like any of the UKIP leadership candidates nor their MP. Surely this indicates he is thinking seriously of starting afresh to try and re-model the Italian Cinque Stelle?

Only problem being the need to a big personality (come Grillo) to build it around; hard to see Farage being up for this?

dugarbandier · November 2016

PlatoSaid said:
This is insane - Podesta's lawyer is apptd to investigate emails by DOJ.

"Kadzik, as head of the Office of Legislative Affairs, does not have a direct role in ‘chain of command’ for the Clinton investigation
Read more at http://thefreethoughtproject.com/podesta-clinton-doj-investigation/#wKkdpu4bGH19bbKx.99"

Nigelb · November 2016

Much more interesting will be the Obama/Trump chemistry at the latter's inauguration. And will the 45th President really seek the imprisonment of his defeated rival?

With Congress on his side and the Supreme Court falling into his lap Trump will be among the most powerful US presidents there has ever been.

That's an interesting question. Just how much damage could the election of an obvious scofflaw who bears grudges do to the rule of law in the US ?
The system has a large number of checks and balances, but a Trump win would test them severely.

As for Clinton, in law, even assuming the worst email conspiracy theories are true, it's not at all clear that a criminal case would succeed:
https://www.justsecurity.org/34012/
A common refrain in discussions of the Clinton case has been that “anyone else” would have been charged in similar circumstances. Yet if you look closely at the supposedly parallel cases where lower-ranking individuals have been prosecuted for mishandling, you find that invariably the circumstances aren’t similar. You’ll find prosecutions involving classified material knowingly and intentionally provided to uncleared persons (as in the case of Gen. David Petraeus), or where large quantities of documents were literally removed from secure facilities, but I haven’t turned up any cases where conversations about classified subjects on an insecure line have been treated as a criminal matter. In a country where literally millions of individuals hold top secret clearances, we may safely assume this is not because such indiscretions never occur, but because they had not been understood to be criminal acts, and were dealt with as matters of administrative discipline…

Morris_Dancer · November 2016

Miss Plato, Podesta being a Clinton campaign chap?

Chris · November 2016

DavidL said:

JennyFreeman said:
President Trump is starting to look increasingly plausible. He has the Big Mo and what a time to have it.
http://fivethirtyeight.com/features/election-update-yes-donald-trump-has-a-path-to-victory/
http://news.sky.com/story/trump-claims-momentum-as-he-takes-lead-in-presidential-poll-10641469

This from Nate: "Whenever the race tightens, we get people protesting that the popular vote doesn’t matter because it’s all about the Electoral College, and that Trump has no path to 270 electoral votes. But this presumes that the states behave independently from national trends, when in fact they tend to move in tandem."
This is what I have been saying. There is a whole crop of states, Fl, Az, NC, OH, Col, that are very close and follow the national trend. If Trump gains another 2% or so they all go into his column. That makes it very close with smaller states like Nev, NH and Iowa then becoming important.

If that movement doesn't come Clinton still wins big. If it does its going to be a long night.

Yes. For all the emphasis on states, the overwhelming likelihood is that whoever wins the popular vote will also win the presidency (Nate Silver's model puts that likelihood at 89%).

rottenborough · November 2016

IanB2 said:
Arron Banks saying he doesn't like any of the UKIP leadership candidates nor their MP. Surely this indicates he is thinking seriously of starting afresh to try and re-model the Italian Cinque Stelle?

Only problem being the need to a big personality (come Grillo) to build it around; hard to see Farage being up for this?

Farage is lined up for Trump TV and big bucks, reportedly. I suppose he could do both?

DavidL · November 2016

I don't know enough about US federal criminal procedure to ascertain if this is just someone's wet dream but if this is right then the application for a warrant here was a much more significant step than I appreciated at the time: http://www.americanthinker.com/articles/2016/11/dont_be_fooled_hillarygate_probe_is_now_a_formal_federal_criminal_investigation.html

It says that the granting of the warrant means, amongst other things:

"1) A federal judge supervising a grand jury has now made a finding, based on FBI affidavits which present evidence gathered during the preliminary Hillary inquiry (the one which the FBI director stated had been closed back in July), that there's probable cause to believe that a federal crime was committed in connection with Mrs. Clinton's use of a private email server.

We still, however, don't know what crime(s) are suspected to have been committed. Or by whom.

2) The FBI can use this new grant of grand jury authority to investigate Mrs. Clinton's use of a private email server for the first time to issues subpoenaes to obtain testimony from witnesses and compel the production of documents and things. The Bureau and DOJ can, furthermore, use the judge's probable cause finding to support further warrant applications."

As I say it may be overstated but it is definitely worth a read.

Pulpstar · November 2016

@Chris Good point with all those polls,

Clinton is still ahead where she needs to be

HYUFD · November 2016

Chris said:
And I'm very glad that JennyFreeman raised the question of polling averages, because I've just prepared a little round-up of selected polls released in the last 24 hours - all concluded at least 2 days after the FBI intervention. The reason I've done this is specifically to illustrate the importance of using averages rather than selected polls!

National OCT. 20-NOV. 1 RAND (American Life Panel) (2,269) C: 44% T: 35% Clinton +9

Arizona OCT. 25-31 Google Consumer Surveys (656) C: 38% T: 33% Clinton +5
Colorado OCT. 25-31 Google Consumer Surveys (637) C: 43% T: 28% Clinton +15
Florida OCT. 25-30 TargetSmart/William & Mary (718) C: 48% T: 40% Clinton +8
Iowa OCT. 25-31 Google Consumer Surveys (429) C: 38% T: 30% Clinton +8
Maine’s 2nd District OCT. 28-30 Emerson College (375) C: 44% T: 42% Clinton +2
Michigan OCT. 31 Mitchell Research & Communications (737) C: 50% T: 43% Clinton +7
Minnesota OCT. 25-31 SurveyMonkey (773) C: 46% T: 37% Clinton +9
Nevada OCT. 27-31 Auto Alliance / ESA / Pulse Opinion Research (525) C: 44% T: 40% Clinton +4
New Hampshire OCT. 25-31 SurveyMonkey (659) C: 48% T: 35% Clinton +13
North Carolina OCT. 25-31 SurveyMonkey (1,574) C: 47% T: 44% Clinton +3
Ohio OCT. 27-31 Auto Alliance / ESA / Pulse Opinion Research (525) C: 47% T:41% Clinton +6
Pennsylvania OCT. 26-30 Franklin & Marshall College (652) C: 49% T: 38% Clinton +11
Virginia OCT. 25-31 SurveyMonkey (2,089) C: 48% T: 38% Clinton +10
Wisconsin OCT. 25-31 Google Consumer Surveys (816) C: 41% T: 29% Clinton +12

If you had looked at average polling in the EU referendum then Remain would have been ahead, it was the minority of polls which had Leave ahead which were more accurate given Leave won by 4%

rottenborough · November 2016

Chris said:

DavidL said:

JennyFreeman said:
President Trump is starting to look increasingly plausible. He has the Big Mo and what a time to have it.
http://fivethirtyeight.com/features/election-update-yes-donald-trump-has-a-path-to-victory/
http://news.sky.com/story/trump-claims-momentum-as-he-takes-lead-in-presidential-poll-10641469

This from Nate: "Whenever the race tightens, we get people protesting that the popular vote doesn’t matter because it’s all about the Electoral College, and that Trump has no path to 270 electoral votes. But this presumes that the states behave independently from national trends, when in fact they tend to move in tandem."
This is what I have been saying. There is a whole crop of states, Fl, Az, NC, OH, Col, that are very close and follow the national trend. If Trump gains another 2% or so they all go into his column. That makes it very close with smaller states like Nev, NH and Iowa then becoming important.

If that movement doesn't come Clinton still wins big. If it does its going to be a long night.
Yes. For all the emphasis on states, the overwhelming likelihood is that whoever wins the popular vote will also win the presidency (Nate Silver's model puts that likelihood at 89%).

If Jenny doesn't like aggregating polls then I suggest keep off Nate's site.

IanB2 · November 2016

SquareRoot said:

matt said:

SquareRoot said:

IanB2 said:

CD13 said:
Mr B2,

"Ukip activists should sign up to train as teachers so that they can influence what children are taught from an early age."

Although many teachers don't often bring their politics into the classroom, I could have guessed the voting patterns of several of my teachers at grammar school in the 1960s. In that era, they were generally for Queen, country and the Conservatives.

As I switched to science as quickly as possible, it faded quickly, but in the Micky Mouse sciences (sociology et al), it can become a matter of personal interpretation. The ones I bumped into at university from 1967 onwards were often obvious left-wingers, but that's less important. We weren't too bothered about safe spaces.

I meet young family members and their friends and some do echo the prevailing ethos as if it were the holy grail. I assume many will grow out of it anyway.
The point however is that encouraging people to go into teaching so they can influence young kids in a UKIP-direction is an appalling thing for a politician to be suggesting.
The teaching profession(state) is stuffed full of lefties .. I doubt the espouse Tory policies in their thinking, as do teachers in public school
There may be a confusion here between the public face, through the NUT, and teaching staff generally.
On reflection I guess you are right, in the same way the BMA has been taken over by a bunch of lefties.

Teaching (and maybe medicine) might be professions that tend to attract more socially-minded and hence left wing people, but I doubt many of them joined their profession because they wanted to promulgate their political outlook amongst school children. Which is the point here.

rottenborough · November 2016

Ok, maybe time to panic a little:

"...national polls are tightening with Hillary Clinton’s missteps while a key Wall Street indicator that has correctly predicted past elections is lining up in the billionaire’s favor.

The S&P 500 Index is down 3.6 percent since Aug. 8 — a trend that could signify a Trump victory next week, according to an analysis cited in a Bloomberg report. That’s because when there’s a gain in the index during the three months prior to the vote, the incumbent party — in this case, Democrats — have won 86 percent of the time since 1928. The pattern has rung true 100 percent of the time since 1984."

http://www.bostonherald.com/news/us_politics/2016/11/signs_pointing_up_for_donald_trump

PlatoSaid · November 2016

Morris_Dancer said:
Miss Plato, Podesta being a Clinton campaign chap?

He's running Hillary's campaign and all the Wikileaks are from his email accounts.

HYUFD · November 2016

logical_song said:

JennyFreeman said:
Chris:
Yep but that's also based on a polling average which includes polls taken before most of the Trump momentum. I've never found averaging a very reliable method. There's no doubt that Donald Trump is in the ascendancy with the Big Mo.
Nate has Trump now winning Florida, Ohio and North Carolina, where Hillary used to be ahead - but he still loses by 60 electoral college votes.

That can't be right, if Trump wins those three states, the other Romney states, Iowa and Nevada then Hillary only leads 273 to 265. Either Colorado or Pennsylvania would win Trump the Electoral College

Philip_Thompson · November 2016

rottenborough said:
Ok, maybe time to panic a little:

"...national polls are tightening with Hillary Clinton’s missteps while a key Wall Street indicator that has correctly predicted past elections is lining up in the billionaire’s favor.

The S&P 500 Index is down 3.6 percent since Aug. 8 — a trend that could signify a Trump victory next week, according to an analysis cited in a Bloomberg report. That’s because when there’s a gain in the index during the three months prior to the vote, the incumbent party — in this case, Democrats — have won 86 percent of the time since 1928. The pattern has rung true 100 percent of the time since 1984."

http://www.bostonherald.com/news/us_politics/2016/11/signs_pointing_up_for_donald_trump

100% of the time since 84 is a pretty pathetic claimed precedent.

https://xkcd.com/1122/

PlatoSaid · November 2016

For the alternative Trump view

https://youtu.be/uw3T1V9DQZc

rottenborough · November 2016

HYUFD said:

logical_song said:

JennyFreeman said:
Chris:
Yep but that's also based on a polling average which includes polls taken before most of the Trump momentum. I've never found averaging a very reliable method. There's no doubt that Donald Trump is in the ascendancy with the Big Mo.
Nate has Trump now winning Florida, Ohio and North Carolina, where Hillary used to be ahead - but he still loses by 60 electoral college votes.
That can't be right, if Trump wins those three states, the other Romney states, Iowa and Nevada then Hillary only leads 273 to 265. Either Colorado or Pennsylvania would win Trump the Electoral College

where is nate saying FL goes trump? it is pale blue on my copy of his map.

Philip_Thompson · November 2016

HYUFD said:

Chris said:
And I'm very glad that JennyFreeman raised the question of polling averages, because I've just prepared a little round-up of selected polls released in the last 24 hours - all concluded at least 2 days after the FBI intervention. The reason I've done this is specifically to illustrate the importance of using averages rather than selected polls!

National OCT. 20-NOV. 1 RAND (American Life Panel) (2,269) C: 44% T: 35% Clinton +9

Arizona OCT. 25-31 Google Consumer Surveys (656) C: 38% T: 33% Clinton +5
Colorado OCT. 25-31 Google Consumer Surveys (637) C: 43% T: 28% Clinton +15
Florida OCT. 25-30 TargetSmart/William & Mary (718) C: 48% T: 40% Clinton +8
Iowa OCT. 25-31 Google Consumer Surveys (429) C: 38% T: 30% Clinton +8
Maine’s 2nd District OCT. 28-30 Emerson College (375) C: 44% T: 42% Clinton +2
Michigan OCT. 31 Mitchell Research & Communications (737) C: 50% T: 43% Clinton +7
Minnesota OCT. 25-31 SurveyMonkey (773) C: 46% T: 37% Clinton +9
Nevada OCT. 27-31 Auto Alliance / ESA / Pulse Opinion Research (525) C: 44% T: 40% Clinton +4
New Hampshire OCT. 25-31 SurveyMonkey (659) C: 48% T: 35% Clinton +13
North Carolina OCT. 25-31 SurveyMonkey (1,574) C: 47% T: 44% Clinton +3
Ohio OCT. 27-31 Auto Alliance / ESA / Pulse Opinion Research (525) C: 47% T:41% Clinton +6
Pennsylvania OCT. 26-30 Franklin & Marshall College (652) C: 49% T: 38% Clinton +11
Virginia OCT. 25-31 SurveyMonkey (2,089) C: 48% T: 38% Clinton +10
Wisconsin OCT. 25-31 Google Consumer Surveys (816) C: 41% T: 29% Clinton +12
If you had looked at average polling in the EU referendum then Remain would have been ahead, it was the minority of polls which had Leave ahead which were more accurate given Leave won by 4%

Didn't Leave win more polls than Remain in the weeks before the referendum? That was my recollection, it just wasn't taken seriously.

Chris · November 2016

rottenborough said:

HYUFD said:

logical_song said:

JennyFreeman said:
Chris:
Yep but that's also based on a polling average which includes polls taken before most of the Trump momentum. I've never found averaging a very reliable method. There's no doubt that Donald Trump is in the ascendancy with the Big Mo.
Nate has Trump now winning Florida, Ohio and North Carolina, where Hillary used to be ahead - but he still loses by 60 electoral college votes.
That can't be right, if Trump wins those three states, the other Romney states, Iowa and Nevada then Hillary only leads 273 to 265. Either Colorado or Pennsylvania would win Trump the Electoral College
where is nate saying FL goes trump? it is pale blue on my copy of his map.

There are three flavours of his model. Florida is pale blue in the "polls only" model, but pale red in the "polls plus" and "nowcast".

PlatoSaid · November 2016

I see LA Times is pulling further ahead for Trump 5.4%

dugarbandier · November 2016

rottenborough said:
Ok, maybe time to panic a little:

"...national polls are tightening with Hillary Clinton’s missteps while a key Wall Street indicator that has correctly predicted past elections is lining up in the billionaire’s favor.

The S&P 500 Index is down 3.6 percent since Aug. 8 — a trend that could signify a Trump victory next week, according to an analysis cited in a Bloomberg report. That’s because when there’s a gain in the index during the three months prior to the vote, the incumbent party — in this case, Democrats — have won 86 percent of the time since 1928. The pattern has rung true 100 percent of the time since 1984."

http://www.bostonherald.com/news/us_politics/2016/11/signs_pointing_up_for_donald_trump

I'm not going to bother to plot it but suspect the S&P 500 goes up when HRC's lead goes up and goes down when Trump is in the ascendant

foxinsoxuk · November 2016

HYUFD said:

Chris said:
And I'm very glad that JennyFreeman raised the question of polling averages, because I've just prepared a little round-up of selected polls released in the last 24 hours - all concluded at least 2 days after the FBI intervention. The reason I've done this is specifically to illustrate the importance of using averages rather than selected polls!

National OCT. 20-NOV. 1 RAND (American Life Panel) (2,269) C: 44% T: 35% Clinton +9

Arizona OCT. 25-31 Google Consumer Surveys (656) C: 38% T: 33% Clinton +5
Colorado OCT. 25-31 Google Consumer Surveys (637) C: 43% T: 28% Clinton +15
Florida OCT. 25-30 TargetSmart/William & Mary (718) C: 48% T: 40% Clinton +8
Iowa OCT. 25-31 Google Consumer Surveys (429) C: 38% T: 30% Clinton +8
Maine’s 2nd District OCT. 28-30 Emerson College (375) C: 44% T: 42% Clinton +2
Michigan OCT. 31 Mitchell Research & Communications (737) C: 50% T: 43% Clinton +7
Minnesota OCT. 25-31 SurveyMonkey (773) C: 46% T: 37% Clinton +9
Nevada OCT. 27-31 Auto Alliance / ESA / Pulse Opinion Research (525) C: 44% T: 40% Clinton +4
New Hampshire OCT. 25-31 SurveyMonkey (659) C: 48% T: 35% Clinton +13
North Carolina OCT. 25-31 SurveyMonkey (1,574) C: 47% T: 44% Clinton +3
Ohio OCT. 27-31 Auto Alliance / ESA / Pulse Opinion Research (525) C: 47% T:41% Clinton +6
Pennsylvania OCT. 26-30 Franklin & Marshall College (652) C: 49% T: 38% Clinton +11
Virginia OCT. 25-31 SurveyMonkey (2,089) C: 48% T: 38% Clinton +10
Wisconsin OCT. 25-31 Google Consumer Surveys (816) C: 41% T: 29% Clinton +12
If you had looked at average polling in the EU referendum then Remain would have been ahead, it was the minority of polls which had Leave ahead which were more accurate given Leave won by 4%

I think that you are mis-remembering a bit. This was the polling as reported on PB in June, and while the phone polls were for Remain, the average was not decisively for Remain, and one week prior was for Leave. I think the "late swing" to Remain was ficticious.

http://www2.politicalbetting.com/index.php/archives/2016/06/22/two-massive-poll-boosts-for-remain-with-voting-starting-in-less-than-nine-hours/

RobD · November 2016

rottenborough said:

HYUFD said:

logical_song said:

JennyFreeman said:
Chris:
Yep but that's also based on a polling average which includes polls taken before most of the Trump momentum. I've never found averaging a very reliable method. There's no doubt that Donald Trump is in the ascendancy with the Big Mo.
Nate has Trump now winning Florida, Ohio and North Carolina, where Hillary used to be ahead - but he still loses by 60 electoral college votes.
That can't be right, if Trump wins those three states, the other Romney states, Iowa and Nevada then Hillary only leads 273 to 265. Either Colorado or Pennsylvania would win Trump the Electoral College
where is nate saying FL goes trump? it is pale blue on my copy of his map.

If you are on polls+forecast its light red (51% trump)

Bromptonaut · November 2016

Back in the EU
Back in the EU
Back in the EUSSR

http://www.independent.co.uk/news/uk/politics/theresa-may-has-decided-to-opt-into-eu-crime-fighting-measures-risking-anger-of-brexit-tory-mps-a7391666.html

Chris · November 2016

HYUFD said:

logical_song said:

JennyFreeman said:
Chris:
Yep but that's also based on a polling average which includes polls taken before most of the Trump momentum. I've never found averaging a very reliable method. There's no doubt that Donald Trump is in the ascendancy with the Big Mo.
Nate has Trump now winning Florida, Ohio and North Carolina, where Hillary used to be ahead - but he still loses by 60 electoral college votes.
That can't be right, if Trump wins those three states, the other Romney states, Iowa and Nevada then Hillary only leads 273 to 265. Either Colorado or Pennsylvania would win Trump the Electoral College

The point is that it's a probabilistic model. The "polls plus" model gives Trump a 51% probability of winning Florida, and the same for North Carolina. It doesn't assume he wins both. In fact, the number of those states that he's expected to win in the model is 1.02.

Pulpstar · November 2016

Philip_Thompson said:

HYUFD said:

Chris said:
And I'm very glad that JennyFreeman raised the question of polling averages, because I've just prepared a little round-up of selected polls released in the last 24 hours - all concluded at least 2 days after the FBI intervention. The reason I've done this is specifically to illustrate the importance of using averages rather than selected polls!

National OCT. 20-NOV. 1 RAND (American Life Panel) (2,269) C: 44% T: 35% Clinton +9

Arizona OCT. 25-31 Google Consumer Surveys (656) C: 38% T: 33% Clinton +5
Colorado OCT. 25-31 Google Consumer Surveys (637) C: 43% T: 28% Clinton +15
Florida OCT. 25-30 TargetSmart/William & Mary (718) C: 48% T: 40% Clinton +8
Iowa OCT. 25-31 Google Consumer Surveys (429) C: 38% T: 30% Clinton +8
Maine’s 2nd District OCT. 28-30 Emerson College (375) C: 44% T: 42% Clinton +2
Michigan OCT. 31 Mitchell Research & Communications (737) C: 50% T: 43% Clinton +7
Minnesota OCT. 25-31 SurveyMonkey (773) C: 46% T: 37% Clinton +9
Nevada OCT. 27-31 Auto Alliance / ESA / Pulse Opinion Research (525) C: 44% T: 40% Clinton +4
New Hampshire OCT. 25-31 SurveyMonkey (659) C: 48% T: 35% Clinton +13
North Carolina OCT. 25-31 SurveyMonkey (1,574) C: 47% T: 44% Clinton +3
Ohio OCT. 27-31 Auto Alliance / ESA / Pulse Opinion Research (525) C: 47% T:41% Clinton +6
Pennsylvania OCT. 26-30 Franklin & Marshall College (652) C: 49% T: 38% Clinton +11
Virginia OCT. 25-31 SurveyMonkey (2,089) C: 48% T: 38% Clinton +10
Wisconsin OCT. 25-31 Google Consumer Surveys (816) C: 41% T: 29% Clinton +12
If you had looked at average polling in the EU referendum then Remain would have been ahead, it was the minority of polls which had Leave ahead which were more accurate given Leave won by 4%
Didn't Leave win more polls than Remain in the weeks before the referendum? That was my recollection, it just wasn't taken seriously.

Leave's position was stronger purely on polling than Trump's is.

SquareRoot · November 2016

IanB2 said:

SquareRoot said:

matt said:

SquareRoot said:

IanB2 said:

CD13 said:
Mr B2,

"Ukip activists should sign up to train as teachers so that they can influence what children are taught from an early age."

Although many teachers don't often bring their politics into the classroom, I could have guessed the voting patterns of several of my teachers at grammar school in the 1960s. In that era, they were generally for Queen, country and the Conservatives.

As I switched to science as quickly as possible, it faded quickly, but in the Micky Mouse sciences (sociology et al), it can become a matter of personal interpretation. The ones I bumped into at university from 1967 onwards were often obvious left-wingers, but that's less important. We weren't too bothered about safe spaces.

I meet young family members and their friends and some do echo the prevailing ethos as if it were the holy grail. I assume many will grow out of it anyway.
The point however is that encouraging people to go into teaching so they can influence young kids in a UKIP-direction is an appalling thing for a politician to be suggesting.
The teaching profession(state) is stuffed full of lefties .. I doubt the espouse Tory policies in their thinking, as do teachers in public school
There may be a confusion here between the public face, through the NUT, and teaching staff generally.
On reflection I guess you are right, in the same way the BMA has been taken over by a bunch of lefties.
Teaching (and maybe medicine) might be professions that tend to attract more socially-minded and hence left wing people, but I doubt many of them joined their profession because they wanted to promulgate their political outlook amongst school children. Which is the point here.

NO the point being made by Matt was that the leaders of unions such as teachers are not necessarily representative of the political thinking of the membership... and he is right, but it does affect the membership when the leadership speaks..

Chris · November 2016

Pulpstar said:
@Chris Good point with all those polls,

Clinton is still ahead where she needs to be

The point really wasn't that Clinton is still ahead, but that you can prove anything by taking selected polls.

JackW · November 2016

National Tracker - LA Times - Sample 3,004 - 2 Nov

Clinton 42.4 .. Trump 47.8

Note - Trump winning 18-34 by 2.6% .. Gender divide by 11.3% and 30% more Hispanic than Romney ..

http://graphics.latimes.com/usc-presidential-poll-dashboard/

Mortimer · November 2016

IanB2 said:

SquareRoot said:

matt said:

SquareRoot said:

IanB2 said:

CD13 said:
Mr B2,

"Ukip activists should sign up to train as teachers so that they can influence what children are taught from an early age."

Although many teachers don't often bring their politics into the classroom, I could have guessed the voting patterns of several of my teachers at grammar school in the 1960s. In that era, they were generally for Queen, country and the Conservatives.

As I switched to science as quickly as possible, it faded quickly, but in the Micky Mouse sciences (sociology et al), it can become a matter of personal interpretation. The ones I bumped into at university from 1967 onwards were often obvious left-wingers, but that's less important. We weren't too bothered about safe spaces.

I meet young family members and their friends and some do echo the prevailing ethos as if it were the holy grail. I assume many will grow out of it anyway.
The point however is that encouraging people to go into teaching so they can influence young kids in a UKIP-direction is an appalling thing for a politician to be suggesting.
The teaching profession(state) is stuffed full of lefties .. I doubt the espouse Tory policies in their thinking, as do teachers in public school
There may be a confusion here between the public face, through the NUT, and teaching staff generally.
On reflection I guess you are right, in the same way the BMA has been taken over by a bunch of lefties.
Teaching (and maybe medicine) might be professions that tend to attract more socially-minded and hence left wing people, but I doubt many of them joined their profession because they wanted to promulgate their political outlook amongst school children. Which is the point here.

Nope - not going to accept the premise of that supposition. Being socially minded is nothing to do with left and right. Being left wing involves seeing class as the major cleavage in our country and wanting to correct it through state intervention.

HYUFD · November 2016

foxinsoxuk said:

HYUFD said:

Chris said:
And I'm very glad that JennyFreeman raised the question of polling averages, because I've just prepared a little round-up of selected polls released in the last 24 hours - all concluded at least 2 days after the FBI intervention. The reason I've done this is specifically to illustrate the importance of using averages rather than selected polls!

National OCT. 20-NOV. 1 RAND (American Life Panel) (2,269) C: 44% T: 35% Clinton +9

Arizona OCT. 25-31 Google Consumer Surveys (656) C: 38% T: 33% Clinton +5
Colorado OCT. 25-31 Google Consumer Surveys (637) C: 43% T: 28% Clinton +15
Florida OCT. 25-30 TargetSmart/William & Mary (718) C: 48% T: 40% Clinton +8
Iowa OCT. 25-31 Google Consumer Surveys (429) C: 38% T: 30% Clinton +8
Maine’s 2nd District OCT. 28-30 Emerson College (375) C: 44% T: 42% Clinton +2
Michigan OCT. 31 Mitchell Research & Communications (737) C: 50% T: 43% Clinton +7
Minnesota OCT. 25-31 SurveyMonkey (773) C: 46% T: 37% Clinton +9
Nevada OCT. 27-31 Auto Alliance / ESA / Pulse Opinion Research (525) C: 44% T: 40% Clinton +4
New Hampshire OCT. 25-31 SurveyMonkey (659) C: 48% T: 35% Clinton +13
North Carolina OCT. 25-31 SurveyMonkey (1,574) C: 47% T: 44% Clinton +3
Ohio OCT. 27-31 Auto Alliance / ESA / Pulse Opinion Research (525) C: 47% T:41% Clinton +6
Pennsylvania OCT. 26-30 Franklin & Marshall College (652) C: 49% T: 38% Clinton +11
Virginia OCT. 25-31 SurveyMonkey (2,089) C: 48% T: 38% Clinton +10
Wisconsin OCT. 25-31 Google Consumer Surveys (816) C: 41% T: 29% Clinton +12
If you had looked at average polling in the EU referendum then Remain would have been ahead, it was the minority of polls which had Leave ahead which were more accurate given Leave won by 4%
I think that you are mis-remembering a bit. This was the polling as reported on PB in June, and while the phone polls were for Remain, the average was not decisively for Remain, and one week prior was for Leave. I think the "late swing" to Remain was ficticious.

http://www2.politicalbetting.com/index.php/archives/2016/06/22/two-massive-poll-boosts-for-remain-with-voting-starting-in-less-than-nine-hours/

Yet the average was still for Remain and Leave won by 4%, if the average in the US is out by the same margin Trump is heading for the White House

Philip_Thompson · November 2016

Chris said:

HYUFD said:

logical_song said:

JennyFreeman said:
Chris:
Yep but that's also based on a polling average which includes polls taken before most of the Trump momentum. I've never found averaging a very reliable method. There's no doubt that Donald Trump is in the ascendancy with the Big Mo.
Nate has Trump now winning Florida, Ohio and North Carolina, where Hillary used to be ahead - but he still loses by 60 electoral college votes.
That can't be right, if Trump wins those three states, the other Romney states, Iowa and Nevada then Hillary only leads 273 to 265. Either Colorado or Pennsylvania would win Trump the Electoral College
The point is that it's a probabilistic model. The "polls plus" model gives Trump a 51% probability of winning Florida, and the same for North Carolina. It doesn't assume he wins both. In fact, the number of those states that he's expected to win in the model is 1.02.

The problem with using probability is that they're not independent events. In a purely independent 50-50 shot you'd expect him to win 1.

However they events aren't independent. He's quite likely to win either 2 or 0 depending upon if the polls shift one way or another, or if there's a mistake in the polling one way or another.

Philip_Thompson · November 2016

HYUFD said:

foxinsoxuk said:

HYUFD said:

Chris said:
And I'm very glad that JennyFreeman raised the question of polling averages, because I've just prepared a little round-up of selected polls released in the last 24 hours - all concluded at least 2 days after the FBI intervention. The reason I've done this is specifically to illustrate the importance of using averages rather than selected polls!

National OCT. 20-NOV. 1 RAND (American Life Panel) (2,269) C: 44% T: 35% Clinton +9

Arizona OCT. 25-31 Google Consumer Surveys (656) C: 38% T: 33% Clinton +5
Colorado OCT. 25-31 Google Consumer Surveys (637) C: 43% T: 28% Clinton +15
Florida OCT. 25-30 TargetSmart/William & Mary (718) C: 48% T: 40% Clinton +8
Iowa OCT. 25-31 Google Consumer Surveys (429) C: 38% T: 30% Clinton +8
Maine’s 2nd District OCT. 28-30 Emerson College (375) C: 44% T: 42% Clinton +2
Michigan OCT. 31 Mitchell Research & Communications (737) C: 50% T: 43% Clinton +7
Minnesota OCT. 25-31 SurveyMonkey (773) C: 46% T: 37% Clinton +9
Nevada OCT. 27-31 Auto Alliance / ESA / Pulse Opinion Research (525) C: 44% T: 40% Clinton +4
New Hampshire OCT. 25-31 SurveyMonkey (659) C: 48% T: 35% Clinton +13
North Carolina OCT. 25-31 SurveyMonkey (1,574) C: 47% T: 44% Clinton +3
Ohio OCT. 27-31 Auto Alliance / ESA / Pulse Opinion Research (525) C: 47% T:41% Clinton +6
Pennsylvania OCT. 26-30 Franklin & Marshall College (652) C: 49% T: 38% Clinton +11
Virginia OCT. 25-31 SurveyMonkey (2,089) C: 48% T: 38% Clinton +10
Wisconsin OCT. 25-31 Google Consumer Surveys (816) C: 41% T: 29% Clinton +12
If you had looked at average polling in the EU referendum then Remain would have been ahead, it was the minority of polls which had Leave ahead which were more accurate given Leave won by 4%
I think that you are mis-remembering a bit. This was the polling as reported on PB in June, and while the phone polls were for Remain, the average was not decisively for Remain, and one week prior was for Leave. I think the "late swing" to Remain was ficticious.

http://www2.politicalbetting.com/index.php/archives/2016/06/22/two-massive-poll-boosts-for-remain-with-voting-starting-in-less-than-nine-hours/
Yet the average was still for Remain and Leave won by 4%, if the average in the US is out by the same argon Trump is heading for the White House

But the average was correct when early voting was happening. If the same happens in the US then HRC is heading for the White House.

HYUFD · November 2016

rottenborough said:

HYUFD said:

logical_song said:

JennyFreeman said:
Chris:
Yep but that's also based on a polling average which includes polls taken before most of the Trump momentum. I've never found averaging a very reliable method. There's no doubt that Donald Trump is in the ascendancy with the Big Mo.
Nate has Trump now winning Florida, Ohio and North Carolina, where Hillary used to be ahead - but he still loses by 60 electoral college votes.
That can't be right, if Trump wins those three states, the other Romney states, Iowa and Nevada then Hillary only leads 273 to 265. Either Colorado or Pennsylvania would win Trump the Electoral College
where is nate saying FL goes trump? it is pale blue on my copy of his map.

It is red now with RCP

HYUFD · November 2016

Philip_Thompson said:

HYUFD said:

Chris said:
And I'm very glad that JennyFreeman raised the question of polling averages, because I've just prepared a little round-up of selected polls released in the last 24 hours - all concluded at least 2 days after the FBI intervention. The reason I've done this is specifically to illustrate the importance of using averages rather than selected polls!

National OCT. 20-NOV. 1 RAND (American Life Panel) (2,269) C: 44% T: 35% Clinton +9

Arizona OCT. 25-31 Google Consumer Surveys (656) C: 38% T: 33% Clinton +5
Colorado OCT. 25-31 Google Consumer Surveys (637) C: 43% T: 28% Clinton +15
Florida OCT. 25-30 TargetSmart/William & Mary (718) C: 48% T: 40% Clinton +8
Iowa OCT. 25-31 Google Consumer Surveys (429) C: 38% T: 30% Clinton +8
Maine’s 2nd District OCT. 28-30 Emerson College (375) C: 44% T: 42% Clinton +2
Michigan OCT. 31 Mitchell Research & Communications (737) C: 50% T: 43% Clinton +7
Minnesota OCT. 25-31 SurveyMonkey (773) C: 46% T: 37% Clinton +9
Nevada OCT. 27-31 Auto Alliance / ESA / Pulse Opinion Research (525) C: 44% T: 40% Clinton +4
New Hampshire OCT. 25-31 SurveyMonkey (659) C: 48% T: 35% Clinton +13
North Carolina OCT. 25-31 SurveyMonkey (1,574) C: 47% T: 44% Clinton +3
Ohio OCT. 27-31 Auto Alliance / ESA / Pulse Opinion Research (525) C: 47% T:41% Clinton +6
Pennsylvania OCT. 26-30 Franklin & Marshall College (652) C: 49% T: 38% Clinton +11
Virginia OCT. 25-31 SurveyMonkey (2,089) C: 48% T: 38% Clinton +10
Wisconsin OCT. 25-31 Google Consumer Surveys (816) C: 41% T: 29% Clinton +12
If you had looked at average polling in the EU referendum then Remain would have been ahead, it was the minority of polls which had Leave ahead which were more accurate given Leave won by 4%
Didn't Leave win more polls than Remain in the weeks before the referendum? That was my recollection, it just wasn't taken seriously.

Remain led in more final polls than Leave did and in the poll average done by the FT

DavidL · November 2016

JackW said:
National Tracker - LA Times - Sample 3,004 - 2 Nov

Clinton 42.4 .. Trump 47.8

Note - Trump winning 18-34 by 2.6% .. Gender divide by 11.3% and 30% more Hispanic than Romney ..

http://graphics.latimes.com/usc-presidential-poll-dashboard/

My understanding is that the open data of the LA Times has indicated that if you remove their rather odd but consistent distortions then you basically get something in line with RCP average.

Unless that has changed this increase in the lead should reflect the movement in the RCP average towards Trump and may even anticipate it a little with newer data. I think the LA Times have generally had a bias of about +4 for Trump. Again, unless that has changed, this indicates that it is getting extremely close in the popular vote.

Chris · November 2016

HYUFD said:

Chris said:
And I'm very glad that JennyFreeman raised the question of polling averages, because I've just prepared a little round-up of selected polls released in the last 24 hours - all concluded at least 2 days after the FBI intervention. The reason I've done this is specifically to illustrate the importance of using averages rather than selected polls!

National OCT. 20-NOV. 1 RAND (American Life Panel) (2,269) C: 44% T: 35% Clinton +9

Arizona OCT. 25-31 Google Consumer Surveys (656) C: 38% T: 33% Clinton +5
Colorado OCT. 25-31 Google Consumer Surveys (637) C: 43% T: 28% Clinton +15
Florida OCT. 25-30 TargetSmart/William & Mary (718) C: 48% T: 40% Clinton +8
Iowa OCT. 25-31 Google Consumer Surveys (429) C: 38% T: 30% Clinton +8
Maine’s 2nd District OCT. 28-30 Emerson College (375) C: 44% T: 42% Clinton +2
Michigan OCT. 31 Mitchell Research & Communications (737) C: 50% T: 43% Clinton +7
Minnesota OCT. 25-31 SurveyMonkey (773) C: 46% T: 37% Clinton +9
Nevada OCT. 27-31 Auto Alliance / ESA / Pulse Opinion Research (525) C: 44% T: 40% Clinton +4
New Hampshire OCT. 25-31 SurveyMonkey (659) C: 48% T: 35% Clinton +13
North Carolina OCT. 25-31 SurveyMonkey (1,574) C: 47% T: 44% Clinton +3
Ohio OCT. 27-31 Auto Alliance / ESA / Pulse Opinion Research (525) C: 47% T:41% Clinton +6
Pennsylvania OCT. 26-30 Franklin & Marshall College (652) C: 49% T: 38% Clinton +11
Virginia OCT. 25-31 SurveyMonkey (2,089) C: 48% T: 38% Clinton +10
Wisconsin OCT. 25-31 Google Consumer Surveys (816) C: 41% T: 29% Clinton +12
If you had looked at average polling in the EU referendum then Remain would have been ahead, it was the minority of polls which had Leave ahead which were more accurate given Leave won by 4%

Of course the average won't be exactly right, and almost certainly there will be a minority of polls that happen to be closer to the right result. But that doesn't help you, unless you have a magic way of picking out the polls that are right!

PlatoSaid · November 2016

Hmm

YouGov
Those big poll swings in the US presidential election? We think they're phantoms, and that this race has been stable https://t.co/XHI7dSnndC https://t.co/yILKwM8GfC

HYUFD · November 2016

Philip_Thompson said:

HYUFD said:

foxinsoxuk said:

HYUFD said:

Chris said:
And I'm very glad that JennyFreeman raised the question of polling averages, because I've just prepared a little round-up of selected polls released in the last 24 hours - all concluded at least 2 days after the FBI intervention. The reason I've done this is specifically to illustrate the importance of using averages rather than selected polls!

National OCT. 20-NOV. 1 RAND (American Life Panel) (2,269) C: 44% T: 35% Clinton +9

Arizona OCT. 25-31 Google Consumer Surveys (656) C: 38% T: 33% Clinton +5
Colorado OCT. 25-31 Google Consumer Surveys (637) C: 43% T: 28% Clinton +15
Florida OCT. 25-30 TargetSmart/William & Mary (718) C: 48% T: 40% Clinton +8
Iowa OCT. 25-31 Google Consumer Surveys (429) C: 38% T: 30% Clinton +8
Maine’s 2nd District OCT. 28-30 Emerson College (375) C: 44% T: 42% Clinton +2
Michigan OCT. 31 Mitchell Research & Communications (737) C: 50% T: 43% Clinton +7
Minnesota OCT. 25-31 SurveyMonkey (773) C: 46% T: 37% Clinton +9
Nevada OCT. 27-31 Auto Alliance / ESA / Pulse Opinion Research (525) C: 44% T: 40% Clinton +4
New Hampshire OCT. 25-31 SurveyMonkey (659) C: 48% T: 35% Clinton +13
North Carolina OCT. 25-31 SurveyMonkey (1,574) C: 47% T: 44% Clinton +3
Ohio OCT. 27-31 Auto Alliance / ESA / Pulse Opinion Research (525) C: 47% T:41% Clinton +6
Pennsylvania OCT. 26-30 Franklin & Marshall College (652) C: 49% T: 38% Clinton +11
Virginia OCT. 25-31 SurveyMonkey (2,089) C: 48% T: 38% Clinton +10
Wisconsin OCT. 25-31 Google Consumer Surveys (816) C: 41% T: 29% Clinton +12
If you had looked at average polling in the EU referendum then Remain would have been ahead, it was the minority of polls which had Leave ahead which were more accurate given Leave won by 4%
I think that you are mis-remembering a bit. This was the polling as reported on PB in June, and while the phone polls were for Remain, the average was not decisively for Remain, and one week prior was for Leave. I think the "late swing" to Remain was ficticious.

http://www2.politicalbetting.com/index.php/archives/2016/06/22/two-massive-poll-boosts-for-remain-with-voting-starting-in-less-than-nine-hours/
Yet the average was still for Remain and Leave won by 4%, if the average in the US is out by the same argon Trump is heading for the White House
But the average was correct when early voting was happening. If the same happens in the US then HRC is heading for the White House.

Early voting is included in some polls and even then Leave did not lead by 4% in the poll average when postal voting started

ThreeQuidder · November 2016

rottenborough said:

Chris said:

DavidL said:

JennyFreeman said:
President Trump is starting to look increasingly plausible. He has the Big Mo and what a time to have it.
http://fivethirtyeight.com/features/election-update-yes-donald-trump-has-a-path-to-victory/
http://news.sky.com/story/trump-claims-momentum-as-he-takes-lead-in-presidential-poll-10641469

This from Nate: "Whenever the race tightens, we get people protesting that the popular vote doesn’t matter because it’s all about the Electoral College, and that Trump has no path to 270 electoral votes. But this presumes that the states behave independently from national trends, when in fact they tend to move in tandem."
This is what I have been saying. There is a whole crop of states, Fl, Az, NC, OH, Col, that are very close and follow the national trend. If Trump gains another 2% or so they all go into his column. That makes it very close with smaller states like Nev, NH and Iowa then becoming important.

If that movement doesn't come Clinton still wins big. If it does its going to be a long night.
Yes. For all the emphasis on states, the overwhelming likelihood is that whoever wins the popular vote will also win the presidency (Nate Silver's model puts that likelihood at 89%).
If Jenny doesn't like aggregating polls then I suggest keep off Nate's site.

It's axiomatic that the best polls are better than the polling average.

The trouble is in determining which are the best polls. And confirmation bias doesn't help.

Richard_Tyndall · November 2016

IanB2 said:
Arron Banks saying he doesn't like any of the UKIP leadership candidates nor their MP. Surely this indicates he is thinking seriously of starting afresh to try and re-model the Italian Cinque Stelle?

Only problem being the need to a big personality (come Grillo) to build it around; hard to see Farage being up for this?

Not sure how realistic that is. the Cinque Stella movement is actually very close to Carswell's position on many key issues and I am pretty sure from looking at his history that Banks is a long way away from them on most things.

MarqueeMark · November 2016

PlatoSaid said:
Hmm

YouGov
Those big poll swings in the US presidential election? We think they're phantoms, and that this race has been stable https://t.co/XHI7dSnndC https://t.co/yILKwM8GfC

Ah yes, YouGov's "large panel of respondents". Like that has never caused problems with a self-selecting over-informed overly politically-committed bunch before....

IanB2 · November 2016

Mortimer said:

IanB2 said:

SquareRoot said:

matt said:

SquareRoot said:

IanB2 said:

CD13 said:
Mr B2,

"Ukip activists should sign up to train as teachers so that they can influence what children are taught from an early age."

Although many teachers don't often bring their politics into the classroom, I could have guessed the voting patterns of several of my teachers at grammar school in the 1960s. In that era, they were generally for Queen, country and the Conservatives.

As I switched to science as quickly as possible, it faded quickly, but in the Micky Mouse sciences (sociology et al), it can become a matter of personal interpretation. The ones I bumped into at university from 1967 onwards were often obvious left-wingers, but that's less important. We weren't too bothered about safe spaces.

I meet young family members and their friends and some do echo the prevailing ethos as if it were the holy grail. I assume many will grow out of it anyway.
The point however is that encouraging people to go into teaching so they can influence young kids in a UKIP-direction is an appalling thing for a politician to be suggesting.
The teaching profession(state) is stuffed full of lefties .. I doubt the espouse Tory policies in their thinking, as do teachers in public school
There may be a confusion here between the public face, through the NUT, and teaching staff generally.
On reflection I guess you are right, in the same way the BMA has been taken over by a bunch of lefties.
Teaching (and maybe medicine) might be professions that tend to attract more socially-minded and hence left wing people, but I doubt many of them joined their profession because they wanted to promulgate their political outlook amongst school children. Which is the point here.
Nope - not going to accept the premise of that supposition. Being socially minded is nothing to do with left and right. Being left wing involves seeing class as the major cleavage in our country and wanting to correct it through state intervention.

For socialists, maybe, but not for liberals or greens.

Despite your assertion, my experience has been that those occupations that involve working with people in a supportive context, whether teaching, nursing or social work, tend to attract people with a more left wing mindset.

JackW · November 2016

NBC on total state early voting numbers as of last night :

https://twitter.com/NBCNightlyNews/status/793564692287483904/photo/1?ref_src=twsrc^tfw

Chris · November 2016

Philip_Thompson said:

Chris said:

HYUFD said:

logical_song said:

JennyFreeman said:
Chris:
Yep but that's also based on a polling average which includes polls taken before most of the Trump momentum. I've never found averaging a very reliable method. There's no doubt that Donald Trump is in the ascendancy with the Big Mo.
Nate has Trump now winning Florida, Ohio and North Carolina, where Hillary used to be ahead - but he still loses by 60 electoral college votes.
That can't be right, if Trump wins those three states, the other Romney states, Iowa and Nevada then Hillary only leads 273 to 265. Either Colorado or Pennsylvania would win Trump the Electoral College
The point is that it's a probabilistic model. The "polls plus" model gives Trump a 51% probability of winning Florida, and the same for North Carolina. It doesn't assume he wins both. In fact, the number of those states that he's expected to win in the model is 1.02.
The problem with using probability is that they're not independent events. In a purely independent 50-50 shot you'd expect him to win 1.

However they events aren't independent. He's quite likely to win either 2 or 0 depending upon if the polls shift one way or another, or if there's a mistake in the polling one way or another.

Yes - I shouldn't have said it was exactly 1.02 in the model, because the model does take into account that the outcomes in different states are closely correlated. But the expected number of Trump wins in those states will still be very close to 1, given that both probabilities are 51%. The correlation means that the probability of it being 1 is less than 50% and the probability of it being 0 or 2 is greater than 25%..

JosiasJessop · November 2016

Nigelb said:
Much more interesting will be the Obama/Trump chemistry at the latter's inauguration. And will the 45th President really seek the imprisonment of his defeated rival?

With Congress on his side and the Supreme Court falling into his lap Trump will be among the most powerful US presidents there has ever been.

That's an interesting question. Just how much damage could the election of an obvious scofflaw who bears grudges do to the rule of law in the US ?
The system has a large number of checks and balances, but a Trump win would test them severely.

As for Clinton, in law, even assuming the worst email conspiracy theories are true, it's not at all clear that a criminal case would succeed:
https://www.justsecurity.org/34012/
A common refrain in discussions of the Clinton case has been that “anyone else” would have been charged in similar circumstances. Yet if you look closely at the supposedly parallel cases where lower-ranking individuals have been prosecuted for mishandling, you find that invariably the circumstances aren’t similar. You’ll find prosecutions involving classified material knowingly and intentionally provided to uncleared persons (as in the case of Gen. David Petraeus), or where large quantities of documents were literally removed from secure facilities, but I haven’t turned up any cases where conversations about classified subjects on an insecure line have been treated as a criminal matter. In a country where literally millions of individuals hold top secret clearances, we may safely assume this is not because such indiscretions never occur, but because they had not been understood to be criminal acts, and were dealt with as matters of administrative discipline…

As we have seen countless times before, including Erdogan in Turkey and Putin in Russia, a populist democratically-elected leader can start off moderate and increasingly grab power over the state, judiciary and populace. The checks and balances - such as a free press - are amongst the first to be subsumed to the will of the leader.

If Trump wins, it will be interesting to see if he starts to make baby steps towards such moves.

Jobabob · November 2016

Chris said:

Pulpstar said:
@Chris Good point with all those polls,

Clinton is still ahead where she needs to be
The point really wasn't that Clinton is still ahead, but that you can prove anything by taking selected polls.

The cherry picking on here has reached insane levels. It's a betting site for god's sake. I suspect the Trump rampers never, ever bet.

Philip_Thompson · November 2016

HYUFD said:

Philip_Thompson said:

HYUFD said:

Chris said:
And I'm very glad that JennyFreeman raised the question of polling averages, because I've just prepared a little round-up of selected polls released in the last 24 hours - all concluded at least 2 days after the FBI intervention. The reason I've done this is specifically to illustrate the importance of using averages rather than selected polls!

National OCT. 20-NOV. 1 RAND (American Life Panel) (2,269) C: 44% T: 35% Clinton +9

Arizona OCT. 25-31 Google Consumer Surveys (656) C: 38% T: 33% Clinton +5
Colorado OCT. 25-31 Google Consumer Surveys (637) C: 43% T: 28% Clinton +15
Florida OCT. 25-30 TargetSmart/William & Mary (718) C: 48% T: 40% Clinton +8
Iowa OCT. 25-31 Google Consumer Surveys (429) C: 38% T: 30% Clinton +8
Maine’s 2nd District OCT. 28-30 Emerson College (375) C: 44% T: 42% Clinton +2
Michigan OCT. 31 Mitchell Research & Communications (737) C: 50% T: 43% Clinton +7
Minnesota OCT. 25-31 SurveyMonkey (773) C: 46% T: 37% Clinton +9
Nevada OCT. 27-31 Auto Alliance / ESA / Pulse Opinion Research (525) C: 44% T: 40% Clinton +4
New Hampshire OCT. 25-31 SurveyMonkey (659) C: 48% T: 35% Clinton +13
North Carolina OCT. 25-31 SurveyMonkey (1,574) C: 47% T: 44% Clinton +3
Ohio OCT. 27-31 Auto Alliance / ESA / Pulse Opinion Research (525) C: 47% T:41% Clinton +6
Pennsylvania OCT. 26-30 Franklin & Marshall College (652) C: 49% T: 38% Clinton +11
Virginia OCT. 25-31 SurveyMonkey (2,089) C: 48% T: 38% Clinton +10
Wisconsin OCT. 25-31 Google Consumer Surveys (816) C: 41% T: 29% Clinton +12
If you had looked at average polling in the EU referendum then Remain would have been ahead, it was the minority of polls which had Leave ahead which were more accurate given Leave won by 4%
Didn't Leave win more polls than Remain in the weeks before the referendum? That was my recollection, it just wasn't taken seriously.
Remain led in more final polls than Leave did and in the poll average done by the FT

But the final polls were not the only polls. They were wrong, but the polls during postal voting were correct. There was also the twin issues of herding and probable "shy leavers" after the murder of Jo Cox that possibly made the final polls wrong while the earlier polls were correct.

If the same occurs in the USA Hillary has won.

Sandpit · November 2016

JackW said:
NBC on total state early voting numbers as of last night :

https://twitter.com/NBCNightlyNews/status/793564692287483904/photo/1?ref_src=twsrc^tfw

Any explanation for the massive increase in Michigan - a change in the rules on early voting?

Alistair · November 2016

Philip_Thompson said:

HYUFD said:

Chris said:
And I'm very glad that JennyFreeman raised the question of polling averages, because I've just prepared a little round-up of selected polls released in the last 24 hours - all concluded at least 2 days after the FBI intervention. The reason I've done this is specifically to illustrate the importance of using averages rather than selected polls!

National OCT. 20-NOV. 1 RAND (American Life Panel) (2,269) C: 44% T: 35% Clinton +9

Arizona OCT. 25-31 Google Consumer Surveys (656) C: 38% T: 33% Clinton +5
Colorado OCT. 25-31 Google Consumer Surveys (637) C: 43% T: 28% Clinton +15
Florida OCT. 25-30 TargetSmart/William & Mary (718) C: 48% T: 40% Clinton +8
Iowa OCT. 25-31 Google Consumer Surveys (429) C: 38% T: 30% Clinton +8
Maine’s 2nd District OCT. 28-30 Emerson College (375) C: 44% T: 42% Clinton +2
Michigan OCT. 31 Mitchell Research & Communications (737) C: 50% T: 43% Clinton +7
Minnesota OCT. 25-31 SurveyMonkey (773) C: 46% T: 37% Clinton +9
Nevada OCT. 27-31 Auto Alliance / ESA / Pulse Opinion Research (525) C: 44% T: 40% Clinton +4
New Hampshire OCT. 25-31 SurveyMonkey (659) C: 48% T: 35% Clinton +13
North Carolina OCT. 25-31 SurveyMonkey (1,574) C: 47% T: 44% Clinton +3
Ohio OCT. 27-31 Auto Alliance / ESA / Pulse Opinion Research (525) C: 47% T:41% Clinton +6
Pennsylvania OCT. 26-30 Franklin & Marshall College (652) C: 49% T: 38% Clinton +11
Virginia OCT. 25-31 SurveyMonkey (2,089) C: 48% T: 38% Clinton +10
Wisconsin OCT. 25-31 Google Consumer Surveys (816) C: 41% T: 29% Clinton +12
If you had looked at average polling in the EU referendum then Remain would have been ahead, it was the minority of polls which had Leave ahead which were more accurate given Leave won by 4%
Didn't Leave win more polls than Remain in the weeks before the referendum? That was my recollection, it just wasn't taken seriously.

The issue with the Brexit polling was the massive disconnect between phone and online where phone polls have remain big leads while online gave narrow Leave leads with big don't know figures.

On the basis of 2015 GE people opted to believe the phone rather than online polls

ThreeQuidder · November 2016

HYUFD said:

foxinsoxuk said:

HYUFD said:

Chris said:
And I'm very glad that JennyFreeman raised the question of polling averages, because I've just prepared a little round-up of selected polls released in the last 24 hours - all concluded at least 2 days after the FBI intervention. The reason I've done this is specifically to illustrate the importance of using averages rather than selected polls!

National OCT. 20-NOV. 1 RAND (American Life Panel) (2,269) C: 44% T: 35% Clinton +9

Arizona OCT. 25-31 Google Consumer Surveys (656) C: 38% T: 33% Clinton +5
Colorado OCT. 25-31 Google Consumer Surveys (637) C: 43% T: 28% Clinton +15
Florida OCT. 25-30 TargetSmart/William & Mary (718) C: 48% T: 40% Clinton +8
Iowa OCT. 25-31 Google Consumer Surveys (429) C: 38% T: 30% Clinton +8
Maine’s 2nd District OCT. 28-30 Emerson College (375) C: 44% T: 42% Clinton +2
Michigan OCT. 31 Mitchell Research & Communications (737) C: 50% T: 43% Clinton +7
Minnesota OCT. 25-31 SurveyMonkey (773) C: 46% T: 37% Clinton +9
Nevada OCT. 27-31 Auto Alliance / ESA / Pulse Opinion Research (525) C: 44% T: 40% Clinton +4
New Hampshire OCT. 25-31 SurveyMonkey (659) C: 48% T: 35% Clinton +13
North Carolina OCT. 25-31 SurveyMonkey (1,574) C: 47% T: 44% Clinton +3
Ohio OCT. 27-31 Auto Alliance / ESA / Pulse Opinion Research (525) C: 47% T:41% Clinton +6
Pennsylvania OCT. 26-30 Franklin & Marshall College (652) C: 49% T: 38% Clinton +11
Virginia OCT. 25-31 SurveyMonkey (2,089) C: 48% T: 38% Clinton +10
Wisconsin OCT. 25-31 Google Consumer Surveys (816) C: 41% T: 29% Clinton +12
If you had looked at average polling in the EU referendum then Remain would have been ahead, it was the minority of polls which had Leave ahead which were more accurate given Leave won by 4%
I think that you are mis-remembering a bit. This was the polling as reported on PB in June, and while the phone polls were for Remain, the average was not decisively for Remain, and one week prior was for Leave. I think the "late swing" to Remain was ficticious.

http://www2.politicalbetting.com/index.php/archives/2016/06/22/two-massive-poll-boosts-for-remain-with-voting-starting-in-less-than-nine-hours/
Yet the average was still for Remain and Leave won by 4%, if the average in the US is out by the same margin Trump is heading for the White House

Well, that rather depends on which way it is out...!

Philip_Thompson · November 2016

Jobabob said:

Chris said:

Pulpstar said:
@Chris Good point with all those polls,

Clinton is still ahead where she needs to be
The point really wasn't that Clinton is still ahead, but that you can prove anything by taking selected polls.
The cherry picking on here has reached insane levels. It's a betting site for god's sake. I suspect the Trump rampers never, ever bet.

If you're betting you should be glad for people who disagree with you, all the moreso if you're correct.

If you're right but everyone else agrees with you then you won't make money betting as the odds will be terrible.

Monksfield · November 2016

foxinsoxuk said:

nunu said:
"Note - Large split ticket vote favours Clinton. 28% of Florida GOP voters went Clinton. Follows Marist early polling analysis of Florida."
.
Jack no way this can be right?? 10% maybe but over a quarter? They must have their modelling wrong.
I think there must be a lot of Hispanic GOP voters in Florida.

Cubans. Usually reliable republicans but perhaps less reliable when the candidate has been as dimwitted in his courting of hispanics as Trump.

It's one reason why the registration stats for early voting in Florida could be unreliable this time.

Philip_Thompson · November 2016

Chris said:

Philip_Thompson said:

Chris said:

HYUFD said:

logical_song said:

JennyFreeman said:
Chris:
Yep but that's also based on a polling average which includes polls taken before most of the Trump momentum. I've never found averaging a very reliable method. There's no doubt that Donald Trump is in the ascendancy with the Big Mo.
Nate has Trump now winning Florida, Ohio and North Carolina, where Hillary used to be ahead - but he still loses by 60 electoral college votes.
That can't be right, if Trump wins those three states, the other Romney states, Iowa and Nevada then Hillary only leads 273 to 265. Either Colorado or Pennsylvania would win Trump the Electoral College
The point is that it's a probabilistic model. The "polls plus" model gives Trump a 51% probability of winning Florida, and the same for North Carolina. It doesn't assume he wins both. In fact, the number of those states that he's expected to win in the model is 1.02.
The problem with using probability is that they're not independent events. In a purely independent 50-50 shot you'd expect him to win 1.

However they events aren't independent. He's quite likely to win either 2 or 0 depending upon if the polls shift one way or another, or if there's a mistake in the polling one way or another.
Yes - I shouldn't have said it was exactly 1.02 in the model, because the model does take into account that the outcomes in different states are closely correlated. But the expected number of Trump wins in those states will still be very close to 1, given that both probabilities are 51%. The correlation means that the probability of it being 1 is less than 50% and the probability of it being 0 or 2 is greater than 25%..

Indeed the expected number doesn't vary but if the standard deviation varies then that's the difference between a locked on Clinton victory and a possible Trump victory (and equally possible Clinton landslide).

MikeK · November 2016

Jobabob said:

Chris said:

Pulpstar said:
@Chris Good point with all those polls,

Clinton is still ahead where she needs to be
The point really wasn't that Clinton is still ahead, but that you can prove anything by taking selected polls.
The cherry picking on here has reached insane levels. It's a betting site for god's sake. I suspect the Trump rampers never, ever bet.

I'm not a Trump ramper but I've bet on Trump to win, if even by a scrape, and I think it will be a bit larger than that. The Times gives Trump the lead this morning, if that means anything at all. It will be hearts in mouths for both sides until polling day.

Theuniondivvie · November 2016

foxinsoxuk said:

HYUFD said:

Chris said:
And I'm very glad that JennyFreeman raised the question of polling averages, because I've just prepared a little round-up of selected polls released in the last 24 hours - all concluded at least 2 days after the FBI intervention. The reason I've done this is specifically to illustrate the importance of using averages rather than selected polls!

National OCT. 20-NOV. 1 RAND (American Life Panel) (2,269) C: 44% T: 35% Clinton +9

Arizona OCT. 25-31 Google Consumer Surveys (656) C: 38% T: 33% Clinton +5
Colorado OCT. 25-31 Google Consumer Surveys (637) C: 43% T: 28% Clinton +15
Florida OCT. 25-30 TargetSmart/William & Mary (718) C: 48% T: 40% Clinton +8
Iowa OCT. 25-31 Google Consumer Surveys (429) C: 38% T: 30% Clinton +8
Maine’s 2nd District OCT. 28-30 Emerson College (375) C: 44% T: 42% Clinton +2
Michigan OCT. 31 Mitchell Research & Communications (737) C: 50% T: 43% Clinton +7
Minnesota OCT. 25-31 SurveyMonkey (773) C: 46% T: 37% Clinton +9
Nevada OCT. 27-31 Auto Alliance / ESA / Pulse Opinion Research (525) C: 44% T: 40% Clinton +4
New Hampshire OCT. 25-31 SurveyMonkey (659) C: 48% T: 35% Clinton +13
North Carolina OCT. 25-31 SurveyMonkey (1,574) C: 47% T: 44% Clinton +3
Ohio OCT. 27-31 Auto Alliance / ESA / Pulse Opinion Research (525) C: 47% T:41% Clinton +6
Pennsylvania OCT. 26-30 Franklin & Marshall College (652) C: 49% T: 38% Clinton +11
Virginia OCT. 25-31 SurveyMonkey (2,089) C: 48% T: 38% Clinton +10
Wisconsin OCT. 25-31 Google Consumer Surveys (816) C: 41% T: 29% Clinton +12
If you had looked at average polling in the EU referendum then Remain would have been ahead, it was the minority of polls which had Leave ahead which were more accurate given Leave won by 4%
I think that you are mis-remembering a bit. This was the polling as reported on PB in June, and while the phone polls were for Remain, the average was not decisively for Remain, and one week prior was for Leave. I think the "late swing" to Remain was ficticious.

http://www2.politicalbetting.com/index.php/archives/2016/06/22/two-massive-poll-boosts-for-remain-with-voting-starting-in-less-than-nine-hours/

'Mis-remembering' is...generous.

weejonnie · November 2016

Looking at the LA Times http://graphics.latimes.com/usc-presidential-poll-dashboard/

Hillary's big collapse seems to be with the over $75K voters.

A bit of the change seems to be likelihood to vote with Rs heading upwards.

PlatoSaid · November 2016

Philip_Thompson said:

Jobabob said:

Chris said:

Pulpstar said:
@Chris Good point with all those polls,

Clinton is still ahead where she needs to be
The point really wasn't that Clinton is still ahead, but that you can prove anything by taking selected polls.
The cherry picking on here has reached insane levels. It's a betting site for god's sake. I suspect the Trump rampers never, ever bet.
If you're betting you should be glad for people who disagree with you, all the moreso if you're correct.

If you're right but everyone else agrees with you then you won't make money betting as the odds will be terrible.

IIRC Paddy Power said 90% of their small bets were going on Trump yesterday. Believe or not!

brokenwheel · November 2016

JackW said:
NBC on total state early voting numbers as of last night :

https://twitter.com/NBCNightlyNews/status/793564692287483904/photo/1?ref_src=twsrc^tfw

Only because the number of early voting days was restricted in Florida in 2012.

rottenborough · November 2016

brokenwheel said:

JackW said:
NBC on total state early voting numbers as of last night :

https://twitter.com/NBCNightlyNews/status/793564692287483904/photo/1?ref_src=twsrc^tfw
Only because the number of early voting days was restricted in Florida in 2012.

What about Michigan? 660K more

rottenborough · November 2016

PlatoSaid said:

Philip_Thompson said:

Jobabob said:

Chris said:

Pulpstar said:
@Chris Good point with all those polls,

Clinton is still ahead where she needs to be
The point really wasn't that Clinton is still ahead, but that you can prove anything by taking selected polls.
The cherry picking on here has reached insane levels. It's a betting site for god's sake. I suspect the Trump rampers never, ever bet.
If you're betting you should be glad for people who disagree with you, all the moreso if you're correct.

If you're right but everyone else agrees with you then you won't make money betting as the odds will be terrible.

Philip_Thompson said:

Jobabob said:

Chris said:

Pulpstar said:
@Chris Good point with all those polls,

Clinton is still ahead where she needs to be
The point really wasn't that Clinton is still ahead, but that you can prove anything by taking selected polls.
The cherry picking on here has reached insane levels. It's a betting site for god's sake. I suspect the Trump rampers never, ever bet.
If you're betting you should be glad for people who disagree with you, all the moreso if you're correct.

If you're right but everyone else agrees with you then you won't make money betting as the odds will be terrible.
IIRC Paddy Power said 90% of their small bets were going on Trump yesterday. Believe or not!

When are they going to pay out on Trump and get a free advertisement in daily newspapers?

Sandpit · November 2016

Philip_Thompson said:

Chris said:

Philip_Thompson said:

Chris said:

HYUFD said:

logical_song said:

JennyFreeman said:
Chris:
Yep but that's also based on a polling average which includes polls taken before most of the Trump momentum. I've never found averaging a very reliable method. There's no doubt that Donald Trump is in the ascendancy with the Big Mo.
Nate has Trump now winning Florida, Ohio and North Carolina, where Hillary used to be ahead - but he still loses by 60 electoral college votes.
That can't be right, if Trump wins those three states, the other Romney states, Iowa and Nevada then Hillary only leads 273 to 265. Either Colorado or Pennsylvania would win Trump the Electoral College
The point is that it's a probabilistic model. The "polls plus" model gives Trump a 51% probability of winning Florida, and the same for North Carolina. It doesn't assume he wins both. In fact, the number of those states that he's expected to win in the model is 1.02.
The problem with using probability is that they're not independent events. In a purely independent 50-50 shot you'd expect him to win 1.

However they events aren't independent. He's quite likely to win either 2 or 0 depending upon if the polls shift one way or another, or if there's a mistake in the polling one way or another.
Yes - I shouldn't have said it was exactly 1.02 in the model, because the model does take into account that the outcomes in different states are closely correlated. But the expected number of Trump wins in those states will still be very close to 1, given that both probabilities are 51%. The correlation means that the probability of it being 1 is less than 50% and the probability of it being 0 or 2 is greater than 25%..
Indeed the expected number doesn't vary but if the standard deviation varies then that's the difference between a locked on Clinton victory and a possible Trump victory (and equally possible Clinton landslide).

Yes, there's only a couple of points each way in the popular vote, between a Trump win and a Clinton landslide - as the spread bettors hopefully understand.

brokenwheel · November 2016

weejonnie said:
Looking at the LA Times http://graphics.latimes.com/usc-presidential-poll-dashboard/

Hillary's big collapse seems to be with the over $75K voters.

A bit of the change seems to be likelihood to vote with Rs heading upwards.

Shame many of that demo will have already voted...

Philip_Thompson · November 2016

HYUFD said:

Philip_Thompson said:

HYUFD said:

foxinsoxuk said:

HYUFD said:

Chris said:
And I'm very glad that JennyFreeman raised the question of polling averages, because I've just prepared a little round-up of selected polls released in the last 24 hours - all concluded at least 2 days after the FBI intervention. The reason I've done this is specifically to illustrate the importance of using averages rather than selected polls!

National OCT. 20-NOV. 1 RAND (American Life Panel) (2,269) C: 44% T: 35% Clinton +9

Arizona OCT. 25-31 Google Consumer Surveys (656) C: 38% T: 33% Clinton +5
Colorado OCT. 25-31 Google Consumer Surveys (637) C: 43% T: 28% Clinton +15
Florida OCT. 25-30 TargetSmart/William & Mary (718) C: 48% T: 40% Clinton +8
Iowa OCT. 25-31 Google Consumer Surveys (429) C: 38% T: 30% Clinton +8
Maine’s 2nd District OCT. 28-30 Emerson College (375) C: 44% T: 42% Clinton +2
Michigan OCT. 31 Mitchell Research & Communications (737) C: 50% T: 43% Clinton +7
Minnesota OCT. 25-31 SurveyMonkey (773) C: 46% T: 37% Clinton +9
Nevada OCT. 27-31 Auto Alliance / ESA / Pulse Opinion Research (525) C: 44% T: 40% Clinton +4
New Hampshire OCT. 25-31 SurveyMonkey (659) C: 48% T: 35% Clinton +13
North Carolina OCT. 25-31 SurveyMonkey (1,574) C: 47% T: 44% Clinton +3
Ohio OCT. 27-31 Auto Alliance / ESA / Pulse Opinion Research (525) C: 47% T:41% Clinton +6
Pennsylvania OCT. 26-30 Franklin & Marshall College (652) C: 49% T: 38% Clinton +11
Virginia OCT. 25-31 SurveyMonkey (2,089) C: 48% T: 38% Clinton +10
Wisconsin OCT. 25-31 Google Consumer Surveys (816) C: 41% T: 29% Clinton +12
If you had looked at average polling in the EU referendum then Remain would have been ahead, it was the minority of polls which had Leave ahead which were more accurate given Leave won by 4%
I think that you are mis-remembering a bit. This was the polling as reported on PB in June, and while the phone polls were for Remain, the average was not decisively for Remain, and one week prior was for Leave. I think the "late swing" to Remain was ficticious.

http://www2.politicalbetting.com/index.php/archives/2016/06/22/two-massive-poll-boosts-for-remain-with-voting-starting-in-less-than-nine-hours/
Yet the average was still for Remain and Leave won by 4%, if the average in the US is out by the same argon Trump is heading for the White House
But the average was correct when early voting was happening. If the same happens in the US then HRC is heading for the White House.
Early voting is included in some polls and even then Leave did not lead by 4% in the poll average when postal voting started

Actually it was quite close to that when you strip out don't knows etc in the weeks before the murder of Jo Cox.

Chris · November 2016

Philip_Thompson said:

Chris said:

Philip_Thompson said:

Chris said:

HYUFD said:

logical_song said:

JennyFreeman said:
Chris:
Yep but that's also based on a polling average which includes polls taken before most of the Trump momentum. I've never found averaging a very reliable method. There's no doubt that Donald Trump is in the ascendancy with the Big Mo.
Nate has Trump now winning Florida, Ohio and North Carolina, where Hillary used to be ahead - but he still loses by 60 electoral college votes.
That can't be right, if Trump wins those three states, the other Romney states, Iowa and Nevada then Hillary only leads 273 to 265. Either Colorado or Pennsylvania would win Trump the Electoral College
The point is that it's a probabilistic model. The "polls plus" model gives Trump a 51% probability of winning Florida, and the same for North Carolina. It doesn't assume he wins both. In fact, the number of those states that he's expected to win in the model is 1.02.
The problem with using probability is that they're not independent events. In a purely independent 50-50 shot you'd expect him to win 1.

However they events aren't independent. He's quite likely to win either 2 or 0 depending upon if the polls shift one way or another, or if there's a mistake in the polling one way or another.
Yes - I shouldn't have said it was exactly 1.02 in the model, because the model does take into account that the outcomes in different states are closely correlated. But the expected number of Trump wins in those states will still be very close to 1, given that both probabilities are 51%. The correlation means that the probability of it being 1 is less than 50% and the probability of it being 0 or 2 is greater than 25%..
Indeed the expected number doesn't vary but if the standard deviation varies then that's the difference between a locked on Clinton victory and a possible Trump victory (and equally possible Clinton landslide).

Yes, I agree. I believe the emphasis on correlation was one reason why Nate Silver's model was showing a larger probability of a Trump win than some others, during the period of Clinton's ascendancy in the polls.

But going back to the original comment I replied to, the key thing for the estimate of electoral college votes is that the model doesn't assume a state is "in the bag" for Trump just because it turns light red.

Sandpit · November 2016

PlatoSaid said:

Philip_Thompson said:

Jobabob said:

Chris said:

Pulpstar said:
@Chris Good point with all those polls,

Clinton is still ahead where she needs to be
The point really wasn't that Clinton is still ahead, but that you can prove anything by taking selected polls.
The cherry picking on here has reached insane levels. It's a betting site for god's sake. I suspect the Trump rampers never, ever bet.
If you're betting you should be glad for people who disagree with you, all the moreso if you're correct.

If you're right but everyone else agrees with you then you won't make money betting as the odds will be terrible.
IIRC Paddy Power said 90% of their small bets were going on Trump yesterday. Believe or not!

And they paid out for Clinton a couple of weeks ago. One assumes for relatively small money at the time, otherwise they wouldn't have done it.

Howdy, Stranger!

Quick Links

Categories

politicalbetting.com » Blog Archive » As WH2016 moves into the final straight: The PB/Polling Matter

Comments