Swing state polling and the electoral college map

[Lantern]

States of Play: Battleground Survey on COVID-19 - Change Research

CNBC has the first swing state polls in a while so it’s a good time to examine state polling and the implied electoral college map.

The poll has Biden up 7 points nationally but the swing states aren’t nearly as comfortable.

Biden and Trump are at parity in Arizona (+1 Trump), North Carolina (+1 Biden), and Wisconsin (even), Biden leads in Florida (+3) and Michigan (+2), and Trump leads in Pennsylvania (+4).

If we give the Wisconsin tie to Trump, and allot the other states by the lead in this poll and 2016 results I think that would imply this map with Biden up by 292-246. However it would be a precarious lead based on 1-2 points in Michigan and North Carolina.

Biden is more comfortable based on the full RCP swing-state polling but still not nearly as comfortable as his national lead of 8 points.
Wisconsin Biden +2
Florida Biden +3.4
Michigan Biden +4.2
PA Biden +4
NC Trump +0.6
Arizona Biden +2.5

If we take the states where Biden is up by 3 and the Hillary states that would be 297 but many of these polls are outdated which is why that CNBC poll is a bit worrying.

Overall Biden has the edge but it looks like the electoral college is going to be close. We should get more polls fairly soon and then we will have our first clear picture of the race before the final months.

[Do_Not_Taunt]

I don’t think it’s worth overreacting to a single poll. This is the only poll that’s shown Trump ahead in PA, though it is the most recent. But it’s also a C- pollster by 538’s rating. Wisconsin I would worry much more about - for whatever reason, it seems to be drifting right. But, if the election were held today, I’d feel pretty confident about Biden picking up Michigan and Pennsylvania, and probably making up for Wisconsin with Arizona. NC/Fla/GA also seem within striking distance for him.

[septimus]

The race looks very close. A tinge of optimism may come from the punters at Betfair who have bid up the chance that Ds will win the White House to 53+%, higher than I’ve seen it in earlier months.

[PhillyGuy]

State polling is historically less accurate than national polling. So those swing state polls mean little.

As for worrying, when your President talks like a dictator, there is no such thing as a comfortable lead for the opposition:

Will he go?

[John_Stamos_Left_Ear]

I would be very shocked if Biden loses Pennsylvania. He was born in Scranton and has ties here still and he will spend a lot of time here and launch a comprehensive GOTV campaign in and around Philly and Pittsburgh.

[Lantern]

@PhillyGuy

State polling is historically less accurate than national polling. So those swing state polls mean little.

State polling is less accurate but it measures what actually determines the winner so there is no escaping it.

[John_Stamos_Left_Ear]

The map at Electoral-Vote.com based on current polling is, to say the least, unkind to Donald Trump.

It has Texas and Iowa as tied and even Utah can only manage barely GOP.
Just adding up Strongly Dem and Likely Dem is only seven EVs short of 270.

The website adds the following explanation:

CNN has released its latest poll of the presidential race, and it’s grim for Donald Trump. They have the President down by a staggering 14 points to Joe Biden, 55% to 41%. That is the best head-to-head result for Biden (and the worst for Trump) since way back in October of last year, when ABC News/The Washington Post gave Biden a 17-point lead. There have been 81 polls of the race since that one.

Biden is outperforming Clinton, which is good news for the Democrats. On the other hand, Trump 2020 is generally outperforming Trump 2016, which would seem to be bad news for the Democrats. However, as you can see, even with Trump doing a bit better, the margin between him and Biden is larger than the one between him and Clinton (excepting that spring bump). That can only mean one thing: that Trump has a tougher hill to climb in 2020 than he did in 2016, and he’s got to climb it with fewer undecided voters available to be won over. Notice, for example, that the average June poll has Biden up 7.7% with only 8.8% of voters undecided. This is entirely consistent with other analyses we’ve seen this cycle, suggesting that the number of undecided voters is already extraordinarily low.

At the moment, Trump is being hurt by two concurrent dynamics. The first is that, as everyone knows, he’s running a base-only campaign. Tweets about Antifa, and e-mails about loony leftist Democrats, and videos about those persnickety rioters are great at firing up his devoted supporters. However, his support among Republicans is at 95% right now. It’s reached its ceiling; it really can’t grow any further.

The second problem is that the President is bleeding support among non-college educated white voters; dropping 10% among that demographic since spring. It used to be Trump 65.5% and Biden 34.5%, now it’s Trump 60.5% and Biden 39.5%. If those figures hold, it is unsurvivable for Trump, since Biden leads among virtually every other key demographic—even seniors. That is doubly true in the Rust Belt states, where Trump won by narrow margins, powered by non-college white votes.

In short, it sure looks like the Trump campaign, outside of the occasional Mark Meadows, is carefully constructing the mother of all bubbles around the President. And that will presumably encourage him to double- and triple-down on his current approach, sticking to his gut and his guns, and allowing McLaughlin and others to convince him that the clocks are indeed striking 13. That is not only a losing approach, it is a lose-in-a-landslide approach.

I’m not spiking the ball or anything but this is as hopeful as I have been in some time.

[Lantern]

I believe the biggest high-quality state poll recently has been a Fox poll in Wisconsin (Biden+9) and Arizona(Biden +4). His RCP national lead is now 8 points. Things are looking pretty good for him but it would be nice to have a critical mass of state polling to confirm it. Trump’s handling of the protests have hurt him but I suspect once the story changes, the race will tighten again by a couple of points though still leaving Biden with a decent lead.

[Deeg]

@John_Stamos_Left_Ear

The map at Electoral-Vote.com based on current polling is, to say the least, unkind to Donald Trump.

It has Texas and Iowa as tied and even Utah can only manage barely GOP.
Just adding up Strongly Dem and Likely Dem is only seven EVs short of 270.

The website adds the following explanation:I’m not spiking the ball or anything but this is as hopeful as I have been in some time.

You left out perhaps the best part:

But doesn’t Trump see polls, too? Yes he does, but that brings us to the second part of the current (apparent) strategy. The campaign has just hired “pollster” John McLaughlin, whose work is so shoddy that he makes Scott Rasmussen look like Ann Selzer. McLaughlin has earned a reputation for comically overstating how well Republican candidates are doing (e.g., the time he told Eric Cantor he was up by 34 points, only to see Cantor lose his election by 10 points). McLaughlin is so bad, most Republican candidates and entities (including the National Republican Congressional Committee) won’t use him anymore. For Trump 2020, his job will be to explain how wrong polls like the CNN poll are, and how people shouldn’t pay attention to fake news like that. That’s all well and good, unless the “fake” polls are…actually correct. Campaign manager Brad Parscale probably won’t even look at McLaughlin’s polls. He undoubtedly has hired a reliable pollster to tell him which states to focus on. But he probably won’t show Trump those results for fear of being sued.

I don’t know if that’s a good assessment of McLaughlin but it certainly fits with Trump’s M.O. That likely mean the hits will keep on coming.

[Elendil_s_Heir]

The Hill says Ohio is suddenly a battleground state again: https://thehill.com/homenews/campaign/501427-ohio-is-suddenly-a-2020-battleground

[Velocity]

November 2020 is going to be the biggest Democratic blowout that there’s been in many decades. Bigger than any of Clinton or Obama’s wins.

[Do_Not_Taunt]

I do like electoral-vote.com, but as far as I can tell, its methodology is to mark each state as red or blue (or white for a tie), either in light, dark, or outlined based solely on the result of the most recent poll. There are some polls - partisan or particularly low-quality - that they completely disregard, but nonetheless, that is not a great methodology. Right now, the gap is big enough that it doesn’t matter, but I would pick 538 modeling over electoral-vote.com any day. Particularly if the race does tighten.

[Red_Wiggler]

Fivethirtyeight reported this morning that the swing states that Biden needs are still way too damn close for comfort.

Wisconsin and Michigan are indeed looking good, however, and I think Arizona is going blue in November. But if no other states, I think my scenario adds up 269-269 and then chaos, where Trump is the Master Lord.

[Lance_Turbo]

@Do_Not_Taunt

Right now, the gap is big enough that it doesn’t matter, but I would pick 538 modeling over electoral-vote.com any day.

What does the 538 model say abut the current state of the race?

[JKellyMap]

@Lance_Turbo

What does the 538 model say abut the current state of the race?

AFAICT, they haven’t yet launched the model you probably have in mind.

A summary of their recent output might be this: From “leaning Trump” to “leaning Biden,” the swingiest states at this point seem to be: Arizona, Wisconsin, Florida, North Carolina, Nevada, Michigan, Pennsylvania.

(Some polls have Arizona and Wisconsin more pro-Biden than this ordering suggests; and some have Pennsylvania more pro-Trump).

If this ordering holds, the tipping-point state is North Carolina. If it (and the rest before it, on this list) goes for Trump, it’s a 269-269 tie, and Trump wins. If it (and the rest after it) goes with Biden, he wins, 284-254.

[JKellyMap]

BTW, the least-well-polled state in my swingy list is Nevada (as usual). There is at least SOME evidence to place it as (for now) “very tight, but leaning Biden” — see this aricle.

[Lance_Turbo]

Here’s my own list of PredictIt derived swing states from bluest to reddest with cumulative D electoral votes. This is every state/district that both parties have at least a 10% chance to win.

State/district	Electoral Votes	P(D)	P(R)	Cumulative D EV
Minnesota	10		0.8678	0.1322	228
Michigan	16		0.8532	0.1468	244
New Hampshire	4		0.8501	0.1499	248
Pennsylvania	20		0.7863	0.2137	268
Wisconsin	10		0.7202	0.2798	278
Arizona		11		0.6902	0.3098	289
Nebraska CD-02	1		0.6830	0.3170	290
Florida		29		0.5974	0.4026	319
North Carolina	15		0.5840	0.4160	334
Ohio		18		0.4117	0.5883	352
Georgia		16		0.3969	0.6031	368
Iowa		6		0.3789	0.6211	374
Maine CD-02	1		0.2747	0.7253	375
Texas		38		0.2249	0.7751	413

In this snapshot, Wisconsin is the tipping point. Any of Wisconsin, Arizona, Florida, or North Carolina will be enough for the blue team to win.

[DinoR]

@JKellyMap

If it (and the rest before it, on this list) goes for Trump, it’s a 269-269 tie, and Trump wins.

If the current House selected the winner today that would be accurate for a vote along partisan lines. They do not though. November’s election will select the Reps that would handle the tie.

A darkly humorous side note just came to mind. We are most likely still looking at a Senate GOP majority. If November produces a House with Dems controlling a majority of delegations we could be looking at a Biden-Pence administration.

[Lance_Turbo]

@DinoR

If the current House selected the winner today that would be accurate for a vote along partisan lines. They do not though. November’s election will select the Reps that would handle the tie.

A darkly humorous side note just came to mind. We are most likely still looking at a Senate GOP majority. If November produces a House with Dems controlling a majority of delegations we could be looking at a Biden-Pence administration.

Which states do you think are going to flip from R majority delegation to D?

Do you think this is more likely to happen than D control of the senate?

[JKellyMap]

That would be nutty, all right!
(By the way, the best hopes for the Democrats to pick up Senate seats at this point may be North Carolina, Arizona, and Maine, and just maybe Montana, Iowa, and Colorado. They’ll lose Alabama, though).