Selzer/DMR/Bloomberg FINAL IOWA POLL: Trump +5, Clinton +3

[Landslide Lyndon]

LOL, according to the poll Rubio's support DROPPED the last two days of the survey, after the debate he "won".

[Mr. Morden]

Nate Cohn has a good story on Selzer's methodology:

http://www.nytimes.com/2016/01/31/upshot/why-this-is-the-iowa-poll-that-everyones-waiting-for.html?_r=0

This table is interesting:

[Adam Griffin]

^^^ Interesting. There are two schools of thought in regards to which is more accurate, but I tend to believe that voter file sampling is the superior method in all but the most unorthodox and/or high-turnout scenarios (like 2008):

[RBH]

there's also not an infinite number of first time voters. It'd be a nice split to see how many 2016 voters are first-timers from 2008.

I'd guess virtually everybody 25 or younger (born in 1991 or after) is going to be first time for the Dems, unless somebody caucused for the uncontested Obama candidacy as their first time in 2012. That universe seems to be kinda lower than you'd think.

But then again, an age breakdown of the 60% first timers in 2008 seems helpful here and idk if that got split by the surveys.

Eh, there might be some Paul 2012 first timers switching to Sanders.  But yeah, you make a good point that many of the 2008 first timers are going to be voting again.

yeah, I wasn't counting 2012 Republican caucusers. I'd think that number is gonna be pretty low. Those voters are probably gonna have to get used to the differences between D and R caucuses.

comparing the 2 Selzer polls, it moved from 42/40 Clinton to 45/42 Clinton. Kinda suggesting that there might be a ceiling on the PV support for Sanders. Considering the IA Dem Caucuses aren't a PV contest, there's a potential that Bernie gets 40%+ in a bunch of places that he loses. Topping out in the mid-50s or such in Johnson Co (where the non-IC vote is gonna cancel out parts of the IC vote)

[IceSpear]

LOL, according to the poll Rubio's support DROPPED the last two days of the survey, after the debate he "won".

but muh surge

If he does finish strongly, we'll see first hand how much influence untrue media spin can have.

[Shadows]

5% Unsure, 4% Uncommitted, 3 % O Malley - That is 12% up for grabs.

Plus this only considers registered voters. There will be a pretty decent number of registrations on caucus day among which demographic Sanders has a good lead. O'Malley's supporters prefer Bernie 2-1.

So I think this could be anywhere between a Clinton, 5-6% point victory, to Sanders 5-6% victory, with likely result as a neck & neck.

With such huge (12%) open voters & new registrations coming in for Bernie, this is TOO close to call

[Alcon]

LOL, according to the poll Rubio's support DROPPED the last two days of the survey, after the debate he "won".

but muh surge

If he does finish strongly, we'll see first hand how much influence untrue media spin can have.

Those individual days have huge MoEs.  It's not that that unlikely that a candidate could be improving (although not dominating) over these days and yet the poll would show a decline.  Selzer is a great pollster, but statistics are statistics.

[○∙◄☻¥tπ[╪AV┼cVê└]

Nate Cohn has a good story on Selzer's methodology:

http://www.nytimes.com/2016/01/31/upshot/why-this-is-the-iowa-poll-that-everyones-waiting-for.html?_r=0

This table is interesting:

Wow, that is interesting. And not too much variation within the 2 groups. Except go home, Loras, you're drunk.

[Badger]

i'm not as inclined with these numbers to view Trump winning compared to earlier this week. evangical and tea party turnout in IA is a lot more reliable than Trump's bastion of first-timers and generally unengaged voters. for once, weather could actually have an impact here.

gun to my head i very reluctantly say Trump.

[Sorenroy]

I can't find the crosstabs for the polls, however I found this in the officially released numbers:

You must be logged in to read this quote.

As such, as long as the numbers for this year match up with elections in the past, this should be pretty accurate.
http://media.bloomberg.com/bb/avfile/r1OvZ1NeDjnY

I also have a fairly significant question regarding the meaning of late Jan-16 and early Jan-16. All of the polls have that weird wording to them when they are compared to the polls conducted earlier in the campaign cycle (all of those have specific dates). I do not question that these polls were conducted over the period of the 26th to the 29th, but what does that wording even mean?

[psychprofessor]

I can't find the crosstabs for the polls, however I found this in the officially released numbers:

You must be logged in to read this quote.

As such, as long as the numbers for this year match up with elections in the past, this should be pretty accurate.
http://media.bloomberg.com/bb/avfile/r1OvZ1NeDjnY

I also have a fairly significant question regarding the meaning of late Jan-16 and early Jan-16. All of the polls have that weird wording to them when they are compared to the polls conducted earlier in the campaign cycle (all of those have specific dates). I do not question that these polls were conducted over the period of the 26th to the 29th, but what does that wording even mean?

She conducted two polls in January, 2016. To differentiate them, she termed the first one "early Jan 16" and the second one "late Jan 16."

[Sorenroy]

I can't find the crosstabs for the polls, however I found this in the officially released numbers:

You must be logged in to read this quote.

As such, as long as the numbers for this year match up with elections in the past, this should be pretty accurate.
http://media.bloomberg.com/bb/avfile/r1OvZ1NeDjnY

I also have a fairly significant question regarding the meaning of late Jan-16 and early Jan-16. All of the polls have that weird wording to them when they are compared to the polls conducted earlier in the campaign cycle (all of those have specific dates). I do not question that these polls were conducted over the period of the 26th to the 29th, but what does that wording even mean?

She conducted two polls in January, 2016. To differentiate them, she termed the first one "early Jan 16" and the second one "late Jan 16."

Thank you. I guess I'm more tired than I thought (1:45 AM here); I thought it was referring to January 16th 2016...

Anyway, to add something new here, they polled more than just the three people running for president of the Democratic side for favorabilities. Here's a full list:

Person — Favorable-Unfavorable (Net)

Barack Obama — 90-9 (+81)
Bill Clinton — 86-11 (+75)
Bernie Sanders — 82-12 (+70)
Joe Biden — 81-11 (+70)
Hillary Clinton — 81-17 (+64)
John Kerry — 65-18 (+47)
Elizabeth Warren — 47-7 (+40)
Martin O'Malley — 46-13 (+33)
Michael Bloomberg — 17-26 (-9)

Obama is so widely liked, I wonder if his endorsement would be able to shift voters away from Sanders.

Edit: I emphasized the word "would" to make it clear that although he has not endorsed anyone (yet), I think that that endorsement would have an impact.

[This account no longer in use.]

I can't find the crosstabs for the polls, however I found this in the officially released numbers:

You must be logged in to read this quote.

As such, as long as the numbers for this year match up with elections in the past, this should be pretty accurate.
http://media.bloomberg.com/bb/avfile/r1OvZ1NeDjnY

I also have a fairly significant question regarding the meaning of late Jan-16 and early Jan-16. All of the polls have that weird wording to them when they are compared to the polls conducted earlier in the campaign cycle (all of those have specific dates). I do not question that these polls were conducted over the period of the 26th to the 29th, but what does that wording even mean?

She conducted two polls in January, 2016. To differentiate them, she termed the first one "early Jan 16" and the second one "late Jan 16."

Thank you. I guess I'm more tired than I thought (1:45 AM here); I thought it was referring to January 16th 2016...

Anyway, to add something new here, they polled more than just the three people running for president of the Democratic side for favorabilities. Here's a full list:

Person — Favorable-Unfavorable (Net)

Barack Obama — 90-9 (+81)
Bill Clinton — 86-11 (+75)
Bernie Sanders — 82-12 (+70)
Joe Biden — 81-11 (+70)
Hillary Clinton — 81-17 (+64)
John Kerry — 65-18 (+47)
Elizabeth Warren — 47-7 (+40)
Martin O'Malley — 46-13 (+33)
Michael Bloomberg — 17-26 (-9)

Obama is so widely liked, I wonder if his endorsement would be able to shift voters away from Sanders.

Obama never endorsed anybody yet. I would give my personal opinion on the Obama endorsement situation, but I'd probably be ripped apart by Hillary and Bernie hacks together.

[Fmr President & Senator Polnut]

I can't find the crosstabs for the polls, however I found this in the officially released numbers:

You must be logged in to read this quote.

As such, as long as the numbers for this year match up with elections in the past, this should be pretty accurate.
http://media.bloomberg.com/bb/avfile/r1OvZ1NeDjnY

I also have a fairly significant question regarding the meaning of late Jan-16 and early Jan-16. All of the polls have that weird wording to them when they are compared to the polls conducted earlier in the campaign cycle (all of those have specific dates). I do not question that these polls were conducted over the period of the 26th to the 29th, but what does that wording even mean?

She conducted two polls in January, 2016. To differentiate them, she termed the first one "early Jan 16" and the second one "late Jan 16."

Thank you. I guess I'm more tired than I thought (1:45 AM here); I thought it was referring to January 16th 2016...

Anyway, to add something new here, they polled more than just the three people running for president of the Democratic side for favorabilities. Here's a full list:

Person — Favorable-Unfavorable (Net)

Barack Obama — 90-9 (+81)
Bill Clinton — 86-11 (+75)
Bernie Sanders — 82-12 (+70)
Joe Biden — 81-11 (+70)
Hillary Clinton — 81-17 (+64)
John Kerry — 65-18 (+47)
Elizabeth Warren — 47-7 (+40)
Martin O'Malley — 46-13 (+33)
Michael Bloomberg — 17-26 (-9)

Obama is so widely liked, I wonder if his endorsement would be able to shift voters away from Sanders.

Edit: I emphasized the word "would" to make it clear that although he has not endorsed anyone (yet), I think that that endorsement would have an impact.

Obama has done everything short of endorsing Clinton, because it's pretty poor form for a sitting president to endorse in the middle of the primary of their own party.

[Mr. Morden]

Harry Enten looks at Selzer’s history in Iowa going back to 1988, finding that she’s only incorrectly called the winner once (Santorum beating Romney in 2012), and that the average error per candidate in the poll’s predictions across all years is just 3.3%.  However, with so many candidates running each year, you can be really close on most of them, but way off on one or two, and still manage a 3.3% average:

http://fivethirtyeight.com/features/the-final-des-moines-register-iowa-poll-is-out-how-accurate-will-it-be/

You must be logged in to read this quote.

[○∙◄☻¥tπ[╪AV┼cVê└]

Harry Enten looks at Selzer’s history in Iowa going back to 1988, finding that she’s only incorrectly called the winner once (Santorum beating Romney in 2012), and that the average error per candidate in the poll’s predictions across all years is just 3.3%.  However, with so many candidates running each year, you can be really close on most of them, but way off on one or two, and still manage a 3.3% average:

http://fivethirtyeight.com/features/the-final-des-moines-register-iowa-poll-is-out-how-accurate-will-it-be/

You must be logged in to read this quote.

Of course if she's off 3.3 points on both Hillary's and Bernie's support (those would tend to be very anti-correlated) in Hillary's favor, that means that Bernie leads by 3-4 points.

[HagridOfTheDeep]

Harry Enten looks at Selzer’s history in Iowa going back to 1988, finding that she’s only incorrectly called the winner once (Santorum beating Romney in 2012), and that the average error per candidate in the poll’s predictions across all years is just 3.3%.  However, with so many candidates running each year, you can be really close on most of them, but way off on one or two, and still manage a 3.3% average:

http://fivethirtyeight.com/features/the-final-des-moines-register-iowa-poll-is-out-how-accurate-will-it-be/

You must be logged in to read this quote.

Of course if she's off 3.3 points on both Hillary's and Bernie's support (those would tend to be very anti-correlated) in Hillary's favor, that means that Bernie leads by 3-4 points.

Or Hillary leads by 9 if it's not in her favour.

Either way, it's better to see these results if you're Hillary than if you're Bernie Sanders.

[Ebsy]

I think we just witnessed the death of the Sanders campaign. jfern actually turning a poll that shows a 3 point lead for Clinton into a 4 point win for Sanders. :whew:

[○∙◄☻¥tπ[╪AV┼cVê└]

I think we just witnessed the death of the Sanders campaign. jfern actually turning a poll that shows a 3 point lead for Clinton into a 4 point win for Sanders. :whew:

Come on, I'm just saying Bernie could be leading, even the gold standard of polling can't avoid random statistical noise.

[BlueSwan]

My guess:

Dems: Slim win to Clinton, probably in the 1-4% range.
One might believe that Sanders will underperform due to him having lots of young voters, but I think his ground operation will be fairly strong, keeping the results close, but probably not enough to overtake Clinton.

GOP: Very slim win to Cruz, probably in the 0-3% range.
Unlike Sanders, I don't think Trump has got the ground operation to pull off the win in a caucus state. His lead in the polls in sustantial and it might be enough to carry him over the finish line, but I believe that Cruz' ground operation is far better and likely enough too squeeze out a win. Will be very interesting to see how Trump reacts if that happens.

[Shadows]

I think we just witnessed the death of the Sanders campaign. jfern actually turning a poll that shows a 3 point lead for Clinton into a 4 point win for Sanders. :whew:
[/quote

If Clinton is leading +3 vs Sanders among the REGISTERED voters with a 12% Undecided votebank (including O Malley whose supporters prefer Sanders 2-1), this does not look VERY GOOD for Clinton.

This does not capture the new voters who will register & Sanders camp have clearly said they don't want new voters to double work & register early & are focused on bringing them on the day as Reg can be done anytime.

+3 among registered guys ONLY with 12% undecided is in noways a MAJOR positive for Clinton

[Fmr President & Senator Polnut]

I think we just witnessed the death of the Sanders campaign. jfern actually turning a poll that shows a 3 point lead for Clinton into a 4 point win for Sanders. :whew:

If Clinton is leading +3 vs Sanders among the REGISTERED voters with a 12% Undecided votebank (including O Malley whose supporters prefer Sanders 2-1), this does not look VERY GOOD for Clinton.

This does not capture the new voters who will register & Sanders camp have clearly said they don't want new voters to double work & register early & are focused on bringing them on the day as Reg can be done anytime.

+3 among registered guys ONLY with 12% undecided is in noways a MAJOR positive for Clinton

I agree that there is some elements that could be problematic for Clinton but this is a MASSIVE test for the machines and how they get their people out in the right places and in the right numbers. This isn't about pure raw voters, but where they are.

[Mr. Morden]

Scanning through the thread, I don't think this has been mentioned yet, but they also polled a hypothetical 2-man race on the GOP side.  If it was just Cruz and Trump running, then you get:

Cruz 53%
Trump 35%

[Fmr President & Senator Polnut]

Scanning through the thread, I don't think this has been mentioned yet, but they also polled a hypothetical 2-man race on the GOP side.  If it was just Cruz and Trump running, then you get:

Cruz 53%
Trump 35%

Well it just illustrates that this is, in so many ways, a mess of the GOPs own making.

[Mr. Morden]

First and second choices, and candidate favorability, over time: