Change Research: Biden +8 (National), +7 (AZ), +5 (FL), +5 (MI), +7 (NC), +6 (PA), +8 (WI)

[Granite City]

All conducted June 26-28, +/- from June 12-14

https://www.changeresearch.com/post/states-of-play-battleground-states-of-play-battleground-wave-7

National, 1,663 LV

Biden: 49% (-2)
Trump: 41% (-)

Arizona

Biden: 51% (+6)
Trump: 44% (-)

Florida

Biden: 50% (-)
Trump: 45% (+2)

Michigan

Biden: 48% (+1)
Trump: 43% (-2)

North Carolina

Biden: 51% (+4)
Trump: 44% (-1)

Pennsylvania

Biden: 50% (+1)
Trump: 44% (-2)

Wisconsin

Biden: 51% (+3)
Trump: 43% (-1)

[wbrocks67]

So his margins increased in nearly all of these states, but he fell nationally? Ok Change...

[Amenhotep Bakari-Sellers]

Proves WI isnt a tossup

[JRP1994]

Proves WI isnt a tossup

It's a good practice not to take one poll as definitively proving anything. Just a couple days ago you said the Trafalgar poll of WI showing Trump +1 proved that WI is still a tossup, now this poll proves that it isn't a tossup.

Not trying to tell you how to interpret polls or how you should view any particular race. And I agree with you here, WI isn't a tossup. Biden's definitely favored here. But that was true earlier this week after the Trafalgar poll too.

[TiltsAreUnderrated]

June 26-28, 2020

National sample size: 1663 likely voters
National MoE: 2.4%

Biden 48%
Trump 41%
Libertarian 5%
Green 2%
Would not vote 1%
Undecided 3%

Estimated sample sizes for swing states based on percentage of the full battleground sample they constitute. The MoE for the full battleground sample of 3739 likely voters is 1.6%.

AZ: 298 likely voters
Biden 51%
Trump 44%

FL: 1119 likely voters
Biden 50%
Trump 45%

MI: 559 likely voters
Biden 48%
Trump 43%

NC: 597 likely voters
Biden 51%
Trump 44%

PA: 783 likely voters
Biden 50%
Trump 44%

WI: 373 likely voters
Biden 51%
Trump 43%

[wbrocks67]

North Carolina is interesting. Granted it's Change, but we've now had multiple polls now that are either tied, or like +6-9.

[TiltsAreUnderrated]

North Carolina is interesting. Granted it's Change, but we've now had multiple polls now that are either tied, or like +6-9.

Is it an especially hard state to poll?

[wbrocks67]

North Carolina is interesting. Granted it's Change, but we've now had multiple polls now that are either tied, or like +6-9.

Is it an especially hard state to poll?

Looks like the average was marginally off in 2016 (Trump won by 4%, polls averaged 1%), but nothing crazy

https://www.realclearpolitics.com/epolls/2016/president/nc/north_carolina_trump_vs_clinton-5538.html

[Annatar]

North Carolina is interesting. Granted it's Change, but we've now had multiple polls now that are either tied, or like +6-9.

Is it an especially hard state to poll?

Looks like the average was marginally off in 2016 (Trump won by 4%, polls averaged 1%), but nothing crazy

https://www.realclearpolitics.com/epolls/2016/president/nc/north_carolina_trump_vs_clinton-5538.html

538 had Clinton up 0.7%, so they were off by around 4.5%, so the polls weren't that good in NC if 538's average is looked at.

[Mr. Illini]

How many polls need to be released showing Trump getting obliterated before people stop saying "ok I don't believe this one"?

[GeorgiaModerate]

How many polls need to be released showing Trump getting obliterated before people stop saying "ok I don't believe this one"?

Approximately the same number of such polls that will be released between now and Nov. 3.

[Crumpets]

Arizona being just one point to the right of the nation is my favorite part of this poll. Imagine Biden doing worse than Hillary in popular vote margin and still pulling off a win in Arizona.

[tagimaucia]

North Carolina is interesting. Granted it's Change, but we've now had multiple polls now that are either tied, or like +6-9.

Is it an especially hard state to poll?

Looks like the average was marginally off in 2016 (Trump won by 4%, polls averaged 1%), but nothing crazy

https://www.realclearpolitics.com/epolls/2016/president/nc/north_carolina_trump_vs_clinton-5538.html

538 had Clinton up 0.7%, so they were off by around 4.5%, so the polls weren't that good in NC if 538's average is looked at.

I think its wrong to call the predicted state margins in 538's model a "polling average" though, my understanding is that it's a highly modeled prediction that's also based on national polling, polling in other states, and some demographic data / economic fundamentals as well.

It's possible that with the different weights and adjustments that 538 uses, their simple polling average would be equally off base, but there's definitely additional secret sauce they're putting in there.

[TiltsAreUnderrated]

North Carolina is interesting. Granted it's Change, but we've now had multiple polls now that are either tied, or like +6-9.

Is it an especially hard state to poll?

Looks like the average was marginally off in 2016 (Trump won by 4%, polls averaged 1%), but nothing crazy

https://www.realclearpolitics.com/epolls/2016/president/nc/north_carolina_trump_vs_clinton-5538.html

538 had Clinton up 0.7%, so they were off by around 4.5%, so the polls weren't that good in NC if 538's average is looked at.

I think its wrong to call the predicted state margins in 538's model a "polling average" though, my understanding is that it's a highly modeled prediction that's also based on national polling, polling in other states, and some demographic data / economic fundamentals as well.

It's possible that with the different weights and adjustments that 538 uses, their simple polling average would be equally off base, but there's definitely additional secret sauce they're putting in there.

They do weight with similar state/national polls etc., but it's a lot simpler than their full model (IIRC there aren't many, if any, fundamentals involved in the polling average alone). It's not a bad barometer of the state but JHK etc. are probably still putting out better projections than the 538 polling average right now.

[tagimaucia]

North Carolina is interesting. Granted it's Change, but we've now had multiple polls now that are either tied, or like +6-9.

Is it an especially hard state to poll?

Looks like the average was marginally off in 2016 (Trump won by 4%, polls averaged 1%), but nothing crazy

https://www.realclearpolitics.com/epolls/2016/president/nc/north_carolina_trump_vs_clinton-5538.html

538 had Clinton up 0.7%, so they were off by around 4.5%, so the polls weren't that good in NC if 538's average is looked at.

I think its wrong to call the predicted state margins in 538's model a "polling average" though, my understanding is that it's a highly modeled prediction that's also based on national polling, polling in other states, and some demographic data / economic fundamentals as well.

It's possible that with the different weights and adjustments that 538 uses, their simple polling average would be equally off base, but there's definitely additional secret sauce they're putting in there.

They do weight with similar state/national polls etc., but it's a lot simpler than their full model (IIRC there aren't many, if any, fundamentals involved in the polling average alone). It's not a bad barometer of the state but JHK etc. are probably still putting out better projections than the 538 polling average right now.

Annatar is talking about 538's NC prediction from their model in 2016, not what they're doing now though.

[tagimaucia]

On that note, I'm interested to see how 538's model does this year because my understanding is that Nate refuses to take out polls that don't weight by education, and he argues that simply applying their typical house affect adjustments and weights should solve the problem.  

But because the gaps between voters with different levels of educational attainment have grown so much in just the past few years, it doesn't seem to me that this will work.  I.e. pollsters that have an overall good track record historically because their lack of educational weighting didn't matter as much in the past are now much more likely to have very poor results, and the 538 model will be still giving them a lot of weight.

[Amenhotep Bakari-Sellers]

These polls solidifies, the race for Biden and not only that Rs cannot win the Senate without NC, along with AZ, Co and ME, and a runoff in GA, and MT, KS, and IA still up for grabs, the Trifecta is Ds

[Gass3268]

Regardless of what is says, Change Research is cheap trash.

[Sir Mohamed]

Not bad considering Change Research always belonged to Biden's worst pollsters, especially during the primaries. What strikes me in most national and state polls of 2020 is Biden at 50% or close to this important mark. Back in 2016, HRC was mostly up, but in the low to mid 40s range. In the end, over 80% of voters who disliked either candidate went for Trump. Strongly appears this will completely turn around in 2020. Good news for Biden. We only need to keep it up to early November.

[Dr Oz Lost Party!]

Change is still trash, but Biden is hitting the magic 50% in most of these states.

[Holy Unifying Centrist]

The AZ and WI polls are junk because their sample size is way too low.

[Progressive Pessimist]

Majestic results! But the Michigan result confounds me. In what world is it voting to the right of of all those other battlegrounds, and evenly with Florida?

[TiltsAreUnderrated]

Precise sample sizes obtained by 538 for likely voters:

AZ: 311
FL: 951
MI: 699
NC: 468
PA: 760
WI: 502

[Amenhotep Bakari-Sellers]

It's not the voter screen, its 3rd party candidates, we dont know without them being included how much they are taking away from Biden.

These polls dont include 3rd party candidates

[wbrocks67]

Can we pls not bump threads from 2 weeks ago?