NYTimes/Siena - Biden+9 in MN, +3 in NH, +4 in NV, +5 in WI

[Umengus]

Their last poll had Biden +11 in WI (june), with RV. So not a good poll for trump but better than in june.

[DaleCooper]

In Wisconsin, Mr. Biden received 48 percent support compared with 43 percent for Mr. Trump. That’s a significant drop-off from June, when a Times/Siena poll showed Mr. Biden ahead by 11 points.

LOL maybe its because last time there was a Trump +1 sample and now it's a Trump+6 sample??

No. That's the self-reported sample, but NYT doesn't weight by self-reported vote

https://twitter.com/Nate_Cohn/status/1304740486062759937

https://twitter.com/Nate_Cohn/status/1304746758082265088

Right, that's what I'm saying. There is no weighting, so this is a straight Trump +6 self reported sample. Their June poll was a self-reported Trump +1 sample. So them acting surprised that Biden dropped from the previous poll is rich considering the sample is obviously more predisposed to Trump.

The only self-reported vote ID of any of these states that was accurate was the NH poll.

Do they really not weight for that at all? That might explain the relatively high fixation on law & order in these polls, but it's funny that Trump is even losing on that issue.

[n1240]

The Milwaukee suburb and city numbers are almost certainly flipped here but I'm sure right wing circles will pick up the Milwaukee city numbers anyways and run it as truth whilst ignoring everything else about the poll.

[Interlocutor is just not there yet]

In Wisconsin, Mr. Biden received 48 percent support compared with 43 percent for Mr. Trump. That’s a significant drop-off from June, when a Times/Siena poll showed Mr. Biden ahead by 11 points.

LOL maybe its because last time there was a Trump +1 sample and now it's a Trump+6 sample??

No. That's the self-reported sample, but NYT doesn't weight by self-reported vote

https://twitter.com/Nate_Cohn/status/1304740486062759937
https://twitter.com/Nate_Cohn/status/1304746758082265088

Right, that's what I'm saying. There is no weighting, so this is a straight Trump +6 self reported sample. Their June poll was a self-reported Trump +1 sample. So them acting surprised that Biden dropped from the previous poll is rich considering the sample is obviously more predisposed to Trump.

The only self-reported vote ID of any of these states that was accurate was the NH poll.

Do they really not weight for that at all? That might explain the relatively high fixation on law & order in these polls, but it's funny that Trump is even losing on that issue.

NY Times can't even push-poll their way to a horserace narrative. There's just no budging with this race & it could perhaps widen further if the first debate is a total disaster for Trump

[TiltsAreUnderrated]

MN
MoE: 3.9%

Jorgensen 2
Hawkins 1
Someone else 0
Not voting 0
Don't know/refused 5

NH
MoE: 5.5%

Jorgensen 4
Someone else 1
Not voting 1
Don't know/refused 7

NV
MoE: 5.3%

Jorgensen 3
Someone else 1
Hawkins 0
Not voting 0
Don't know/refused 7

WI
MoE: 4.7%
Changes with October 13-16 poll (their last Likely Voters poll)

Biden 48 (+2)
Trump 43 (-1)
Jorgensen 2
Someone else 1
Not voting 1
Hawkins 0
Don't know/refused 6

[DaleCooper]

I can understand not taking people's recalled 2016 vote at face value, but it seems to me that a +6 self-identified Trump-voter sample would skew the results pretty heavily when the state in question was practically a statistical tie in 2016.

[Ferguson97]

So Biden’s likely leading Wisconsin by double digits if the sample is Trump+6, right?

[wbrocks67]

MINNESOTA:
Males: Biden 44, Trump 43
Females: Biden 56, Trump 38
18-29: Biden 67, Trump 22
30-44: Biden 57, Trump 34
45-64: Trump 52, Biden 40
65+: Biden 52, Trump 40
Whites: Biden 49, Trump 43
Nonwhites: Biden 60, Trump 27
Indies: Biden 52, Trump 33
Trump job approval: 45/52 (-7)
Trump fav: 45/53 (-8)
Biden fav: 52/46 (+6)

NEW HAMPSHIRE:
Males: Trump 51, Biden 36
Females: Biden 54, Trump 35
18-44: Biden 47, Trump 37
45-64: Trump 51, Biden 41
65+: Biden 48, Trump 41
Indies: Biden 41, Trump 41
Trump job approval: 45/53 (-8)
Trump fav: 45/53 (-8)
Biden fav: 51/46 (+5)

NEVADA
Males: Trump 48, Biden 41
Females: Biden 53, Trump 37
18-29: Biden 53, Trump 26
30-44: Biden 42, Trump 40
45-64: Trump 54, Biden 40
65+: Biden 57, Trump 37
Whites: Trump 52, Biden 41
Nonwhites: Biden 58, Trump 24
Indies: Biden 41, Trump 36
Trump job approval: 47/48 (-1)
Trump fav: 45/51 (-6)
Biden fav: 52/43 (+9)

WISCONSIN
Males: Trump 46, Biden 44
Females: Biden 52, Trump 39
18-29: Biden 73, Trump 15
30-44: Trump 52, Biden 36
45-64: Trump 50, Biden 42
65+: Biden 53, Trump 42
Whites: Biden 47, Trump 45
Nonwhites: Biden 71, Trump 21
Indies: Biden 49, Trump 35
Trump job approval: 45/52 (-7)
Trump fav: 45/53 (-8)
Biden fav: 51/45 (+6)

[Fmr. Gov. NickG]

I would tend to agree that you shouldn’t weight by self-reported vote. You generally only want to weight on verifiable demographics if possible.  This means party registration where available.  Party ID is better than reported vote, but not ideal.

However, Nate Cohn seems completely obsessed with education weights.  This strikes me as an overcorrection based on a single recent data point.  Not that you shouldn’t do it, but it seems like Nate sometimes uses this as the sole indicator of poll quality.

[Fmr. Gov. NickG]

Good numbers for Wisconsin in Minnesota.

Not that great for Nevada and New Hampshire.

The Nevada number is just fine for Biden.  You have to remember to add 4 points to the Dem to any poll in NV.

[Brittain33]

I look at the high undecideds and narrow Biden margins for everything except Minnesota and it takes me back to October 2016 polls giving me confidence in Clinton. I know 2020 is different, we have several midterm elections to show that undecideds tend to favor the Dem, but it’s a little cause for anxiety.

[wbrocks67]

I would tend to agree that you shouldn’t weight by self-reported vote. You generally only want to weight on verifiable demographics if possible.  This means party registration where available.  Party ID is better than reported vote, but not ideal.

However, Nate Cohn seems completely obsessed with education weights.  This strikes me as an overcorrection based on a single recent data point.  Not that you shouldn’t do it, but it seems like Nate sometimes uses this as the sole indicator of poll quality.

They have the party ID weight for Wisconsin as R+2. Is that correct?

[wbrocks67]

I look at the high undecideds and narrow Biden margins for everything except Minnesota and it takes me back to October 2016 polls giving me confidence in Clinton. I know 2020 is different, we have several midterm elections to show that undecideds tend to favor the Dem, but it’s a little cause for anxiety.

Yep. Seeing results like 45-42 and 46-42 is a big trigger. Why they couldn't of pushed these people or at least asked which way they were leaning is beyond me...

[GeorgiaModerate]

I look at the high undecideds and narrow Biden margins for everything except Minnesota and it takes me back to October 2016 polls giving me confidence in Clinton. I know 2020 is different, we have several midterm elections to show that undecideds tend to favor the Dem, but it’s a little cause for anxiety.

Yep. Seeing results like 45-42 and 46-42 is a big trigger. Why they couldn't of pushed these people or at least asked which way they were leaning is beyond me...

Somebody asked Nate about that:

https://twitter.com/Nate_Cohn/status/1304762438747926528

Whatever quibbles we may have with the methodology, Nate's transparency on all this is laudable.

[Annatar]

I would tend to agree that you shouldn’t weight by self-reported vote. You generally only want to weight on verifiable demographics if possible.  This means party registration where available.  Party ID is better than reported vote, but not ideal.

However, Nate Cohn seems completely obsessed with education weights.  This strikes me as an overcorrection based on a single recent data point.  Not that you shouldn’t do it, but it seems like Nate sometimes uses this as the sole indicator of poll quality.

They have the party ID weight for Wisconsin as R+2. Is that correct?

This is what Charles Franklin of Marquette has said on party ID in WI

https://mobile.twitter.com/PollsAndVotes/status/1116781482595487750

[Fmr. Gov. NickG]

I would tend to agree that you shouldn’t weight by self-reported vote. You generally only want to weight on verifiable demographics if possible.  This means party registration where available.  Party ID is better than reported vote, but not ideal.

However, Nate Cohn seems completely obsessed with education weights.  This strikes me as an overcorrection based on a single recent data point.  Not that you shouldn’t do it, but it seems like Nate sometimes uses this as the sole indicator of poll quality.

They have the party ID weight for Wisconsin as R+2. Is that correct?

Looks like Wisconsin sample has too many independents, which is unusual.  Independents are much likely to vote than partisans.  It is possible that many Republicans have switched to Ind in the past couple years, in which case the sample would be too Republican relative to the Dems.  Which is why it’s better to weight on registration, although not every state has it.

[GeorgiaModerate]

https://twitter.com/Nate_Cohn/status/1304767299149352960

[Gass3268]

https://twitter.com/Nate_Cohn/status/1304767299149352960

Here’s the map:

https://twitter.com/nate_cohn/status/1304769793317376006?s=21

[politicallefty]

I was about to ask what exactly they mean by Milwaukee Suburbs. That seems very weak for a Republican and some serious erosion in support in the WOW counties, although I don't necessarily think the erosion is even (it seems like Washington County is the most stubbornly Republican of the three). It would certainly mean Waukesha is swinging against Trump, and I'm assuming Biden over 40%? If anything, it likes Biden is underpeforming in Dane County.


The NV numbers don't really concern me. The state is notoriously hard to poll and Democrats seem to always overperform their polling. I'm still waiting to hear from Jon Ralston, although I imagine he won't have much to say until early voting really ramps up.

[woodley park]

So does this represent a stake in the heart of the '2020 Red Minnesota' vampire?

[Gass3268]

I was about to ask what exactly they mean by Milwaukee Suburbs. That seems very weak for a Republican and some serious erosion in support in the WOW counties, although I don't necessarily think the erosion is even (it seems like Washington County is the most stubbornly Republican of the three). It would certainly mean Waukesha is swinging against Trump, and I'm assuming Biden over 40%? If anything, it likes Biden is underpeforming in Dane County.


The NV numbers don't really concern me. The state is notoriously hard to poll and Democrats seem to always overperform their polling. I'm still waiting to hear from Jon Ralston, although I imagine he won't have much to say until early voting really ramps up.

Thats a 4 point swing in Dane County for a Biden from 2016.

[neostassenite31]

So does this represent a stake in the heart of the '2020 Red Minnesota' vampire?

Probably not. Trafalgar (R) is probably going to return in a few weeks with a Trump+2 poll in MN, and there could be more polls showing a low to mid single digit lead just because of variation in polling   

[Tender Branson]

Those results roughly indicate a 4-6% national race in favour of Biden.

NV could actually vote to the Right of the US again this year, just like in 2000.

NV Latinos are not really enthusiastic about Biden and might sit out the election.

All the other results look reasonable, NH maybe too close.

[GeorgiaModerate]

Those results roughly indicate a 4-6% national race in favour of Biden.

NV could actually vote to the Right of the US again this year, just like in 2000.

NV Latinos are not really enthusiastic about Biden and might sit out the election.

All the other results look reasonable, NH maybe too close.

Nate Cohn disagrees with your estimate:

https://twitter.com/Nate_Cohn/status/1304755484788961280

[politicallefty]

Thats a 4 point swing in Dane County for a Biden from 2016.

Good point. I guess I was getting used to the monstrous margins in Dane from the state Supreme Court races. That's probably not realistic, but I don't trust the rural areas or even the Milwaukee suburbs. I think Biden probably does need to win the county by at least 50 points to win the state, but I'd really prefer to see him get closer to Tammy Baldwin's 2018 margin there.