These Battleground State Polls are Not Adding Up

[ElectionsGuy]

Taking the RCP average of polls this morning, I realized the battleground state averages are not lining up with national polls, at all. The national polling average is around Biden +5-6 right now, with a few outliers.

National: Biden +5.3

AZ: Biden +4.4% (implied trend: D+4.7%)
FL: Biden +3.2% (implied trend: D+1.4%)
MI: Biden +5.5% (implied trend: D+2.5%)
NC: Trump +0.3% (implied trend: D+0.1%) (includes an outlier Trump +7 poll)
PA: Biden +6.5% (implied trend: D+4.0%)
TX: Trump +1.4% (implied trend: D+4.4%)
WI: Biden +2.7% (implied trend: D+0.3%)

So every single swing state trends D... makes sense. Here are the polling averages in 2016 right before the election:

National: Clinton +3.2

AZ: Trump +4.0% (implied trend: D+5.7%)
FL: Trump +0.2% (implied trend: R+0.4%)
MI: Clinton +3.4% (implied trend: R+5.3%)
NC: Trump +1.0% (implied trend: D+1.7%)
OH: Trump +3.5% (implied trend: R+5.8%)
PA: Clinton +1.9% (implied trend: R+2.8%)
TX: Trump +12.0% (implied trend: D+5.5%)
WI: Clinton +6.5% (implied trend: D+0.3%)

At least in 2016 it makes sense, some states trend R and some trend D. Ultimately, they underestimated the trends significantly, but it seemed realistic. Right now, it doesn't even make sense.

And then these specific polls...

AZ: Biden +9 (implied trend: D+9) - OH Predictive Insights
IA: Trump +2 (implied trend: D+4) - PPP
MI: Biden +8 (implied trend: D+6) - Fox News
MT: Trump +5 (implied trend: D+12) - MSU
PA: Biden +8 (implied trend: D+6) - Fox News
TX: Biden +1 (implied trend: D+7) - PPP

Almost every poll that is coming out of these states is implying around a D+10 environment or the extreme trends above, they are way too Dem friendly for these national polls. Either the national polls are off or the swing states are off.  Given the history of national polls being only slightly off in 2016 and 2018, I think it's best we don't put too much emphasis into these swing state polls yet (despite all the 'national polls are useless' talk, they actually seem to be more useful right now at gauging the overall race).

[Arizona Iced Tea]

You have to wait till before the election, in October we get a lot more polls so the aggregate doesn't get skewed so easily. Also this is still anyone's game, and this will be a 50/50 election at some point between now and November 3rd. You aren't supposed to compare the 2020 early polls with the 2016 final polls, because it won't work.

[Amenhotep Bakari-Sellers]

Everything isnt about 2016

[Non Swing Voter]

That doesn't necessarily add up to D+10 if states like New York and California are just staying flat because they are maxed out.  That said I think it could be a D+8 or 9 environment like it was in 2018.

[lfromnj]

That doesn't necessarily add up to D+10 if states like New York and California are just staying flat because they are maxed out.  That said I think it could be a D+8 or 9 environment like it was in 2018.

Its really convenient for Democrats that literally all the good trends are in the swing states? Why would California and New York be maxed out but not Texas or PA? Theres certain areas in all 4 states which are maxed out but in the suburbs of all 4 states swings can go in a lot of ways.

Trends in a general form at-least work geographically or demographically .

[Xing]

First of all, it’s early, and the situation with COVID-19 and the economy probably make this race somewhat volatile right now. Also, while it’s possible that the state polls are off and the national numbers aren’t (similar to 2016), it’s also possible that the state polls are somewhat close to accurate, while the national numbers are off (similar to 2012.)

[Amenhotep Bakari-Sellers]

Rs and conservatives want to associate everything to 2016. Hillary was a wife of Bill Clinton whom had ran for NY Senator with no competetive races, and Biden is a 40 yr vet that have had tough races. That's why Biden crushed the Rs in 2008 and 2012

[TrendsareUsuallyReal]

There’s like two polls that show the national race tied which might be skewing the disparity a little.

[Alben Barkley]

“The polls don’t fit my priors, which apparently are that it’s not possible for all the battleground states to trend D or for this year to be a much stronger year for Dems nationally than 2016, so they don’t ‘add up.’ Also, I’m comparing polls taken 6 months before an election to polls taken days before an election. Also, I’m comparing polls from two entirely different elections in two entirely different political climates with two entirely different Democratic candidates.”

And no, it wouldn’t necessarily have to be D +10 nationally for the reasons people have said.

That doesn't necessarily add up to D+10 if states like New York and California are just staying flat because they are maxed out.  That said I think it could be a D+8 or 9 environment like it was in 2018.

Its really convenient for Democrats that literally all the good trends are in the swing states? Why would California and New York be maxed out but not Texas or PA? Theres certain areas in all 4 states which are maxed out but in the suburbs of all 4 states swings can go in a lot of ways.

Trends in a general form at-least work geographically or demographically .

Just because it sounds “convenient” doesn’t mean it can’t be true. It certainly was “convenient” for Trump in 2016 that all the polls were off in the same direction, favoring him, and that multiple key states all trended towards him even though nationally the trend was only about 2 points more R than 2012 and he got a lower share of the vote than Romney. Doesn’t mean it didn’t happen. And part of the reason is that states like California ran up the score while there were huge swings in, well, swing states. That’s one reason to believe California is already pretty much maxed out. But these swing states could all trend D, and be narrowly won by the D, without it necessarily having to even change the national popular vote margin that much.

[ProgressiveModerate]

Like in 2016, most swing sattes shifted R while Ds racked up huge amrgin in CA. Could be the same in 2020 for the Rs, as Trump has energized his base but hasn't made any real outreach effort to new voters

[Non Swing Voter]

That doesn't necessarily add up to D+10 if states like New York and California are just staying flat because they are maxed out.  That said I think it could be a D+8 or 9 environment like it was in 2018.

Its really convenient for Democrats that literally all the good trends are in the swing states? Why would California and New York be maxed out but not Texas or PA? Theres certain areas in all 4 states which are maxed out but in the suburbs of all 4 states swings can go in a lot of ways.

Trends in a general form at-least work geographically or demographically .

Perhaps you missed the word "IF" in my post.  He asked why they weren't adding up, I gave a reason.  Pretty histrionic response.

[Fmr. Gov. NickG]

What’s more likely is that the RCP national average includes some trashy polls showing a very close race, and that is bringing down the average unrealistically.  I’d guess the real national margin right now is about Biden +7 or +8, which would much more even out the trend you are seeing in the swing states.

Note that even polling in the noncompetitive states is suggesting a larger national margin than Biden +5 (like today’s polls showing Trump +5 in Montana or Biden +18 in Colorado).

[ElectionsGuy]

You have to wait till before the election, in October we get a lot more polls so the aggregate doesn't get skewed so easily. Also this is still anyone's game, and this will be a 50/50 election at some point between now and November 3rd. You aren't supposed to compare the 2020 early polls with the 2016 final polls, because it won't work.

I get that, but there are large amounts of people who are already settled into the idea that this election is going to be a huge Democratic wave and it's 2008/Jimmy Carter/Herbert Hoover all over again. My point is, we can't tell yet, and that these polls right now are not making sense. That's the point of this, to say not so fast. In October, if some of the swing state polls are the same as they are now with larger national leads, then we can probably fairly say it will be a very bad year for the Republicans.

There’s like two polls that show the national race tied which might be skewing the disparity a little.

There's one poll with a tie, one with Biden +2, and one with a 10 point lead for Biden. There are 4 out of 8 polls in the average with Biden +6. So let's add a point to Biden's margin for the sake of the argument, it still does not make sense in the context of the battleground polls. And certainly, some of the recent polls still do not make sense (like Trump down 18 in CO)

First of all, it’s early, and the situation with COVID-19 and the economy probably make this race somewhat volatile right now. Also, while it’s possible that the state polls are off and the national numbers aren’t (similar to 2016), it’s also possible that the state polls are somewhat close to accurate, while the national numbers are off (similar to 2012.)

While that's definitely a possibility of that, the reason I wouldn't trust the state polls over the national ones at the moment is because of the better consistency and larger quantity of them over state polls.

And I see some people get triggered when I compare or contrast anything to past elections (especially 2016) to make any sort of sense to a current situation. And I thought I was being generous to Democrats by comparing to the end of the 2016 cycle (where Republican pollsters like Trafalger skewed the averages R towards the end). It's like they have PTSD and still can't explain or process how that election turned out, without dismissing it as a giant fluke, and they are terrified of something like that happening again. Is it possible something like that could happen again? I think it can, or it could not, and the polls could possibly have their most accurate year, or they can even underestimate Democrats. But too many people are closed off to the idea of polls underestimating Republican support because 'that happened in 2016 and 2020 isn't 2016', or whatever other reason they want to say it can't happen ('Trump is an incumbent now' (the polls underestimated Obama's margin by 3 points in 2012)).

[pbrower2a]

We may be seeing a collapse during which polls go haywire.

[Gass3268]

Add up now?

https://twitter.com/gelliottmorris/status/1263607470112624641?s=20

https://twitter.com/ForecasterEnten/status/1262021936038719490?s=20

[Non Swing Voter]

Add up now?

https://twitter.com/gelliottmorris/status/1263607470112624641?s=20

https://twitter.com/ForecasterEnten/status/1262021936038719490?s=20

I think Biden +8 is exactly where things are, which seems consistent with the general mood of the country which is around where it was in 2018 when Democrats won by the same margin. 

[MidwestPollster]

I’m sorry to beat a dead horse, so to speak, but I just don’t see where trump has enough offense to beat Biden at this point in the game. Trump has already hit his ceiling in 2016.

[Figueira]

That doesn't necessarily add up to D+10 if states like New York and California are just staying flat because they are maxed out.  That said I think it could be a D+8 or 9 environment like it was in 2018.

How the hell is New York "maxed out"? Clinton did worse there than Obama.

[wbrocks67]

D+8 also exactly matches where the GCB average is right now as well. Should also be noted that the GCB has been incredibly consistent all year long

[Stranger in a strange land]

That doesn't necessarily add up to D+10 if states like New York and California are just staying flat because they are maxed out.  That said I think it could be a D+8 or 9 environment like it was in 2018.

How the hell is New York "maxed out"? Clinton did worse there than Obama.

A lot of Brooklyn DSA & DSA-Adjacent types are making a big show of not voting for Biden. There aren't a ton of them, but concievably there are enough to shift the margin by 1-2 points.

[Figueira]

That doesn't necessarily add up to D+10 if states like New York and California are just staying flat because they are maxed out.  That said I think it could be a D+8 or 9 environment like it was in 2018.

How the hell is New York "maxed out"? Clinton did worse there than Obama.

A lot of Brooklyn DSA & DSA-Adjacent types are making a big show of not voting for Biden. There aren't a ton of them, but concievably there are enough to shift the margin by 1-2 points.

Yeah, I don't see those people materializing any more than they did in 2016. But even if Trump does do better in New York this time, that doesn't mean Democrats have hit some magical ceiling.

[Roronoa D. Law]

That doesn't necessarily add up to D+10 if states like New York and California are just staying flat because they are maxed out.  That said I think it could be a D+8 or 9 environment like it was in 2018.

How the hell is New York "maxed out"? Clinton did worse there than Obama.

A lot of Brooklyn DSA & DSA-Adjacent types are making a big show of not voting for Biden. There aren't a ton of them, but concievably there are enough to shift the margin by 1-2 points.

Please tell me you are not going by Twitter because every DSA on twitter is either a Bot, a Green party member trying to disrupt the Democrat Party, a con artist trying to sell something, or a sheep following the heard cause they can't make up an idea on there own. 

[Ferguson97]

That doesn't necessarily add up to D+10 if states like New York and California are just staying flat because they are maxed out.  That said I think it could be a D+8 or 9 environment like it was in 2018.

How the hell is New York "maxed out"? Clinton did worse there than Obama.

A lot of Brooklyn DSA & DSA-Adjacent types are making a big show of not voting for Biden. There aren't a ton of them, but concievably there are enough to shift the margin by 1-2 points.

Every single one oft hese people live in Brooklyn, Portland, or SoCal. They will not shift the PV by any meaningful margin. They're just loud on Twitter.

[Pollster]

We saw two polls of safe R states the other day (KY and TN) showing Trump greatly underperforming his 2016 numbers. If these polls are accurate (they may not be) and are pointing to a broader national trend (they may not be), this would have repercussions in the NPV that extrapolations based on battleground state polls wouldn't pick up.

[Amenhotep Bakari-Sellers]

We wont know definitely how the battlegrounds will stack up until October,  but we are living in a 20% unemployment rate not a 3.5 percent unemployment,  I have been saying this all along. It's not gonna be a typical 2016 election