PureSpectrum/COVID-19 Consortium: Biden +10% and 50 states+DC (wins with an odd 413 map)

[TiltsAreUnderrated]

http://www.kateto.net/covid19/COVID19%20CONSORTIUM%20REPORT%2016%20MAIL%20OCT%202020.pdf

August 7 - September 27
26838 likely voters
Changes with June 12-28

Biden 50% (+3)
Trump 40% (+1)

Why such a large sample size? Well, they've posted margins for all 50 states including breakdowns for VBM/non-VBM in each state (they do claim to have weighted properly for all states in their respective results).

The report tries to analyse 3 issues:

• How many people report that they will vote by mail in each state?
• What will be the difference in candidate choice between those who vote
on the day of the election versus those who vote by mail?
• What is the potential shift in the margin between Trump and Biden from the day of voting to a count that incorporates votes received by mail?

All sample sizes and results posted below are for likely voters. I assume Biden keeps ME-01 but loses ME-02 with the statewide margin; I also assume he takes NE-02 but loses the other two districts with NE's margin. I leave MT as a tossup.

AK - 161 - Biden +11%
AL - 590 - Trump +21%
AR - 383 - Trump +9%
AZ - 628 - Biden +3%
CA - 1147 - Biden +27%
CO - 551 - Biden +10%
CT - 518 - Biden +19%
DC - 365 - Biden +53%
DE - 398 - Biden +18%
FL - 869 - Biden +13%
GA - 726 - Biden +5%
HI - 280 - Biden +3%
IA - 402 - Trump +1%
ID - 355 - Trump +23%
IL - 716 - Biden +18%
IN - 542 - Trump +4%
KS - 446 - Trump +14%
KY - 511 - Trump +16%
LA - 448 - Trump +1%
MA - 706 - Biden +31%
MD - 747 - Biden +32%
ME - 380 - Trump +2%
MI - 668 - Biden +12%
MN - 604 - Biden +15%
MO - 667 - Trump +15%
MS - 448 - Trump +8%
MT - 304 - TIE
NC - 683 - Biden +3%
ND - 214 - Biden +4%
NE - 375 - Trump +9%
NH - 316 - Biden +22%
NJ - 652 - Biden +26%
NM - 407 - Biden +25%
NV - 487 - Biden +15%
NY - 847 - Biden +22%
OH - 677 - Biden +6%
OK - 537 - Trump +21%
OR - 595 - Biden +11%
PA - 679 - Biden +7%
RI - 347 - Biden +31%
SC - 508 - Trump +3%
SD - 193 - Biden +2%
TN - 649 - Trump +15%
TX - 982 - Biden +7%
UT - 451 - Trump +21%
VA - 701 - Biden +13%
VT - 161 - Biden +28%
WA - 657 - Biden +24%
WI - 592 - Biden +15%
WV - 439 - Trump +29%
WY - 130 - Trump +35%

In map form:

[TiltsAreUnderrated]

Very possible. To Biden friendly in the midwest tho.

Winning the Dakotas while losing Iowa? Only winning Hawaii by 3 points? I think not.

It's probably better data than SurveyMonkey but not by much at the state level, given the small sample sizes.

[TiltsAreUnderrated]

The topline number looks fine, but it's never a good idea to assume you can break a national poll down into states. Also its a month and a half old.

It's not normally a good idea, I'll agree, but they claim to have weighted for individual states.

This is now my favourite take on what "the 413 map" really means. When Biden said "return to normalcy," we just didn't appreciate how far he intended to reverse trends.

[TiltsAreUnderrated]

Very possible. To Biden friendly in the midwest tho.

Winning the Dakotas while losing Iowa? Only winning Hawaii by 3 points? I think not.

It's probably better data than SurveyMonkey but not by much at the state level, given the small sample sizes.

The Survey Monkey numbers actually seem far more plausible to me. If someone told me their numbers were the final results, or that they were from a higher quality pollster, I would not be surprised at all.

They're plausible in bulk but individual results can be very odd. Their LV results consistently have Biden down by only about 6% in Nebraska. That is shocking and I'd have expected more statewide polling if the sorts of shifts necessary for that to happen were really going on.

[TiltsAreUnderrated]

Very possible. To Biden friendly in the midwest tho.

Winning the Dakotas while losing Iowa? Only winning Hawaii by 3 points? I think not.

It's probably better data than SurveyMonkey but not by much at the state level, given the small sample sizes.

The Survey Monkey numbers actually seem far more plausible to me. If someone told me their numbers were the final results, or that they were from a higher quality pollster, I would not be surprised at all.

They're plausible in bulk but individual results can be very odd. Their LV results consistently have Biden down by only about 6% in Nebraska. That is shocking and I'd have expected more statewide polling if the sorts of shifts necessary for that to happen were really going on.

Well they’re pretty much the only ones polling Nebraska so hard to say. For what it’s worth, Morning Consult tracker consistently found Trump with a pretty medicore approval rating in Nebraska, something I found interesting. Plus, it’s not really that different from Kansas, and everyone seems to agree that’s swung left.

Democrats haven't really had the same kind of success in NE they've had in KS recently. I think the lack of other polls here speaks for itself - there could well be a Democratic trend, but I doubt it's anywhere near that strong or else we would probably have seen serious recruitment in the Senate election as opposed to the primary that produced Janicek and an abandoned write-in bid from one of the biggest names in the Nebraska Democratic Party. Sasse seems like the sort of incumbent who could be vulnerable were his state not titanium R (which it is).

There was one NE-01 poll from Kate Bolz's campaign in July and there could be an upset there, but with a +6% margin statewide, I'd have expected her campaign to be publishing more surveys too.