That's conviction for you; They are literally DYING to have trump re-elected.
I'm totally fine with his people dying for him. That's a choice they made. Sucks for the non-Trump-supporting friends and family of these people, but hey -- that's one more reason to shun MAGAts.
See, when the Trump campaing sends its people, they're not sending their best. They're drug addicts, sexual predators, they're bringing disease, they're bringing crime...and some, I assume, are nice p--actually, fuck that. They're not nice people.
Anyways, it's Monday so it's time to crunch some numbers. And in this edition, I'm going to use two different models. The first is the standard +/-7% lead and then narrowing down the "lean" and "swing" states. The second one, I added in average "undecided/third party" and adjusted the lead by that with some surprising results. More on that inside the spoiler.
Ok, this is the usual one, and I'm only using the poll aggregates from 538 at this point. RCP is just too janky and easily skewed.
First, the safe states (>=7% lead)
Biden:
California (55)
Colorado (9)
Connecticut (7)
Delaware (3)
D.C. (3)
Hawaii (4)
Illinois (20)
Maine + CD1 (3)
Maryland (10)
Massachusetts (11)
Michigan (16)
Minnesota (10)
New Hampshire (4)
New Jersey (14)
New Mexico (5)
New York (29)
Oregon (7)
Rhode Island (4)
Vermont (3)
Virginia (13)
Washington (12)
Wisconsin (10)
TOTAL: 252
Trump gets:
Alabama (9)
Alaska (3)
Arkansas (6)
Idaho (4)
Indiana (11)
Kansas (6)
Kentucky (8 )
Louisiana (8 )
Mississippi (6)
Missouri (10)
Nebraska + CD3 (3)
North Dakota (3)
Oklahoma (7)
South Carolina (9)
South Dakota (3)
Tennessee (11)
Utah (6)
West Virginia (5)
Wyoming (3)
TOTAL: 121
So Biden needs 18 EVs to win, Trump needs 149.
Remaining on the table are:
Arizona (D+3.1, up 0.2 from last week; 11 EVs)
Florida (D+2.4, up 0.1 from last week; 29 EVs)
Georgia (D+1.1, up 0.7 from last week; 16 EVs)
Iowa (R+1.6, down 2.8 from last week and a flip; 6 EVs)
Maine CD 2 (D+3.2, up 1 from last week; 1 EV)
Montana (R+4.9, down 1.8 from last week and down 3.3 over two weeks; 3 EVs)
Nebraska CD1 (R+2; 1 EV)
Nebraska CD2 (D+4.5; 1 EV)
Nevada (D+4.9, down 1.3 from last week; 6 EVs)
North Carolina (D+1.9, down 0.5 from last week; 15 EVs)
Ohio (R+0.4, down 1.2 from last week; 18 EVs)
Pennsylvania (D+5.3, down 0.3 from last week; 20 EVs)
Texas (R+1, up 0.6 from last week; 38 EVs)
Biden's path to victory has a lot of options. TX, PA, OH, or FL win it for him outright.
As a result, Trump's path is conversely narrow. He has to win TX, PA, OH and FL, as well as almost all the remainder.
Prediction: Biden takes PA, NV, FL and ME2. for a minimum of 307.
Trump takes Montana and Iowa, for a minimum of 165.
The remainder are just too close to call.
But let's talk "undecided" voters. And for sake of argument, this include potential third party voters as well, even though polls show they will have minimal returns this year, no more than 3% other than maybe in Alaska. But it's important to try and account for them, as it was undecided voters which largely flipped the polls n their head four years ago. A number of states in the upper Midwest had 8-10% undecided just days before the election, and roughly 60-65% of them went to Trump, which was enough to swing close races.
So, could that happen again? Unlikely. The average undecided is around 3-5% this time, with some interesting exceptions.
Now, I'm calculating a synthetic undecided percentage, because i'm using poll aggregates. So if the average Biden support is 52% and the average Trump support is 48%, that's a synthetic undecided count of 4 percent, even though the undecided count in the individual polls may be (and currently is) lower than that (about 1-2% lower in comparing most recent polls to the synthetic undecided count). But I'm fine using the higher number because more uncertainty is good for making a conservative forecast. What I then did was subtract the undecided count from the aggregate poll lead, to reflect a "worst-case scenario" where 100% of the undecideds break for the underdog. At that point, the only way a positive result could be off is through systemic poll error. This is also possible (and also occurred in 2016), but the amount it could be off by is almost certainly less than 7%. If the polls are off by more than 7%, then either polling science needs to be completely rethought or there's something rotten in the state of <enter state name here>.
Also, a negative result means it's a true swing state in that the lead is essentially less than the remaining voters. The larger the negative result, the "swingier" a state is. Mind you, that doesn't mean the poll result is wrong -- the undecided voters may split evenly. They may choose not to vote. They may choose to vote third party, which in a mathematical sense is equal to not voting as it does not change the basic electoral math. The only real concern would be if they split very unevenly, which would skew the final result away from the poll. Bear in mind that late deciders typically skew female, skew towards the challenger, and skew towards the perceived front-runner. (2016 being a case where the challenger pressure--insomuch as Hillary was seen as an 'heir apparent' to Obama--was stronger than the front-runner pressure). In this election, all three categories would help Biden. Biden is winning the female vote by about 20 percent, has been the perceived front-runner for months now, and is the challenger to an unpopular incumbent.
Despite that, I'm not going to project any kind of undecided split and instead I'm just looking at how great the uncertainty affects each candidate's lead.
I think 5% would be a high but plausible polling error, so we'll award each candidate any state where their lead minus the undecided percentage is still above 5%.
Biden:
California (55)
Colorado (9)
Connecticut (7) -- Connecticut had the highest undecided average of any state, at 10.1%. However, Biden's lead is 25% so it's still a safe bet.
Delaware (3)
District of Columbia (3)
Hawaii (4)
Illinois (20)
Maine + ME2 (3)
Maryland (10)
Massachusetts (11)
New Hampshire (4)
New Jersey (14)
New Mexico (5)
New York (29)
Oregon (7)
Rhode Island (4)
Vermont (3)
Virginia (13)
Washington (12)
TOTAL: 216
Biden's dropped 36 electoral votes from the first model by shedding Michigan, Minnesota and Wisconsin. But, as we'll see, they're still considerably in the lean category.
Trump gets:
Alabama (9)
Arkansas (6)
Idaho (4)
Kentucky (8 )
Louisiana (8 )
Mississippi (6)
Nebraska CD3 (1)
North Dakota (3)
Oklahoma (7)
South Dakota (3)
Tennessee (11)
West Virginia (5)
Wyoming (3)
TOTAL: 74
That's...impressive. Trump has lost 47 EVs, in the form of Alaska, Indiana, Kansas, Missouri, Nebraska (statewide), South Carolina and Utah.
Does that mean he's not going to win those states? Not at all, it's likely he still does. But it means the certainty level is lower than the poll lead suggests. For instance, Trump has a 10.5 percent lead in Utah, but with 6% undecided, which adjusts the lead to 4.6. In Utah in particular, the Libertarian party will probably outperform their national average and account for most of that. And it's highly unlikely that all Utah polling is off by an average 4.6 percent in favor of the Republicans. But it definitely could mean that Trump only wins Utah by say, 5% instead of 10.5%.
The main pattern is that there's 2-3% higher undecideds in a lot of the Midwest (averaging 6-7% as opposed to the 4-5% in most states). If enough of those undecideds break for Biden it likely won't be enough to swing the state without also a polling error in excess of the adjusted margin. But it could make life interesting.
Here's the new "swing" states, in decreasing order of certainty:
Nebraska (R+5)
Utah (R+4.6)
Indiana (R+3.8 )
Wisconsin (D+3.8 )
Kansas (R+3.6)
Minnesota (D+3.6)
South Carolina (R+2.4)
Alaska (R+2.4)
Michigan (D+2.3)
Missouri (R+2.2)
Pennsylvania (D+0.5)
Montana (R+0.4)
Nebraska CD2 (D+0.4)
Nevada (-1.2)
Florida (-2.2)
Arizona (-2.4)
North Carolina (-2.5)
Texas (-3)
Georgia (-3.6)
Nebraska CD1 (-4)
Maine CD1 (-4.6)
Ohio (-5)
Iowa (-5.2)
Another way of looking at this is that Montana is less of a sure win for Trump than Pennsylvania is for Biden. And amazingly Michigan and Minnesota are more of a sure thing for Biden than *Missouri* is for Trump. 2.2% is certainly a feasible polling error, especially given that there's hasn't been a lot of high-quality polling in Missouri. (The more polls, the smaller the average poll error should be, in theory).
On the other end of things, this tells us that Ohio and Iowa are the "swingiest" of swing states, which seems very legit given how many times they've flipped back and forth between parties in the poll results. Versus a state like Arizona or Nevada that has been close but consistently on the blue side of the line.
The standouts for me are Texas and Montana. Texas is most definitely up for grabs, and when you include the eye-popping early turnout numbers, and the example of the attempted (and failed) voter disenfranchisement mentioned a few posts back....I'm starting to get a gut feeling that this might be the year Texas finally goes blue. Which would not only seal the election, but do so with a slam dunk and make Trump the first Republican to lose Texas since Gerald Ford. And Ford was running against a Southerner.
I'm a little less sanguine on Montana. 0.4 is a completely plausible polling error, even if it had a bunch of high-quality polls, which it doesn't. And Trump's lead has cratered in Montana over the last two weeks. But, I think a sizeable portion of Montana's 4.5% undecided will actually go Libertarian, and it's largely a vote-by-mail state which means those votes have already been cast and won't benefit from late momentum in Biden's direction. But at the same time, I won't do a spit-take if Montana does go blue. I'll do more of a spit-take if Trump were to match his 20-point margin from 2016. I think it's likely he only wins Montana by a couple percent this time around.
South Carolina is also worth looking at. Jaime Harrison's campaign has had shittons of money to finance GOTV efforts, which may possibly pay dividends for Biden. My gut still says Trump wins, but I think it's going to be far closer than his current 7.5% lead suggests.
Because the whole point of this model is uncertainty, I'm not going to award any of these states and make a projection from it. But it does show that Trump may have even more of an uphill battle than the simple lead-based model shows for him. The Midwest, especially the "red" Midwest, seems to be slightly more reluctant to make a choice or possibly unenthused about the prospect of voting for Trump again.
Also, I think it's important to understand these dynamics, because let's take a hypothetical situation: the polls were off nationally by 2.5% in favor of Trump, and undecideds break 70% in favor of Biden (which again, remember that the three historical trends with late-deciders all favor Biden).
Biden then wins Texas, Georgia, Iowa, Ohio, Montana, two out of three of Nebraska's congressional districts, Missouri, Alaska, and South Carolina. There will be a LOT of Trump supporters screaming fraud if that happens. Hell, I'd have a hard time believing it myself if I hadn't looked at the underlying data. The notion of red-meat, "manly" states like Texas, Alaska and Montana going for a "socialist pedophile" just will not compute to them, even though it's well within the realm of possibility. Texas could become the new Florida in terms of legal challenges and the GOP seeking to have ballots thrown out for all sorts of reasons. Granted, if Biden won all the states I mentioned, he could just be like "Fine, have Texas. You still lose, loser." But there will likely be legal challenges across the map, regardless of the result.
The sad reality is that Biden needs to win enough states by a large enough margin that he can afford to have the courts intervene and claw back a few for Trump and still win. It's kind of like in football (and sports in general) when a team loses a game because of a single bad penalty call, "your job is to score enough points that the referee can't lose the game for you". Biden needs to score enough votes that the courts (Supreme, state and Federal district courts) can't steal the election for Trump.
An additional item of note? There's no bad weather in the forecast tomorrow to potentially depress turnout. Other than a little bit of rain in the Pacific Northwest and some snow showers in upper New England (both of which are literally the default weather for those regions), it's sunny and mild across the United States. And historical research has indicated that Democrats benefit from better weather and increased voter turnout, or conversely that Republicans benefit from poor weather and decreased turnout. My gut instinct says this is because Republican voters have tended to be older, retired with more available time and more determined to hobble through Hell itself, while Democrat voters are a bit more flighty and/or have shit to do.
If turnout is light tomorrow, it's because almost 70% of the 2016 electorate has already voted. However, there's good reason to believe that total 2020 turnout will exceed 2016 by about 20 million people. The main states where a signficant majority of the population has not yet voted are Idaho, Oklahoma, West Virginia, Alabama and Mississippi (all below 30%), none of which are going for Biden unless their Republicans stay home and boycott the election.