Thread by @NoTeamsIndy

Final Biden v. Trump UPDATE

Nat’l: Biden +9.0/Clinton +5.2
EC: Biden 279-121 (+178)/Clinton 223-179 (+43)

Biden 50%+ w/lead greater than all undecideds + 3rd party in ELEVEN battlegrounds, ZERO for Clinton

THREAD on polling, early vote, exit polls, and my GUESS at the outcome

NOTES (con’t):

1. "Battleground": ≤10% in 2016 results or 2020 averages

2. Likely Voter (LV), otherwise Registered Voter (RV)

3. Polls w/3rd party used when given

4. Turnout-hypothesizing polls not used

5. No partisan polls

5. Outliers (10%+ off of average) excluded

ETERNAL REMINDERS:

1. 2016 nat'l + battlegrounds polling AVERAGES were accurate (≥5%), except WI/IA:

aapor.org/Education-Reso…

2. The maps above are NOT A FORECAST; it is a REPORT on the current data. My GUESS is given at the end of the thread.

3. And FYI, forecasts aren't "yes/no" *predictions* They're probability reports, and 538's largely matched Team Trump's ~30%. Think of them like the hurricane path charts that show all possibilities, and indicating which ones are more likely.

CONTEXT:

10/28/16 (10 days before election): Comey letter (more below)

10/14/20 (20 days before election): NY Post publishes Hunter Biden story from Giuliani/ Bannon w/o verification, based on materials held by Russian Security Forces in 2019:

https://twitter.com/NoTeamsIndy/status/1296561193277165569

Comey letter: Clinton’s lead slips 3-4% in battlegrounds as undecideds moved to Trump, making several tossups. Also, enthusiasm levels rose for him, fell for her, as reflected in turnout. This was seen in polling for several key states (except WI):

fivethirtyeight.com/features/the-c…

A2. On 11/6/16, ALL battlegrounds (margin ≤10%) had double-digit undec/3rd, ZERO double-digit margins on 11/1/20

In 2016, Trump needed 26-48% to flip PA, WI, MI

In 2020, Trump needs 129-147% of them to take leads in those three states.

A2a. Without 3rd party supporters, by 11/6/16 Trump needed only 40% and 60% of undecideds to flip MI and PA. In WI, where polls missed the post-Comey slip, he needed 92% to flip it.

On 11/1/20, Trump needs 179-231% to flip these three states.

A2c. ZERO 2016 battlegrounds lean Trump in 2020 (including GA and TX!); ALL 2020 tossups went red in 2016

Also, Biden is doing better than Clinton in ALL BUT ONE battleground (IA, thanks to last night's Selzer poll), leading by as much as 10.8% in MT, 9.7% in TX, 8.4% in CO.

A3. In 2020, several pollsters started "weighting for education" by including more non-college voters, based on 2016 turnout. We’ll see if that turnout repeats in 2020, but it benefits Trump ~3%. So by 2016 practices, Biden would be as much as +12.0% nat'l, with 324 EC votes

A4. In fact, Biden hasn't trailed in a SINGLE nat'l poll by an A-B rated pollster in 2020. Clinton trailed in TWENTY by 11/1/16. Only 4 ties for Biden (8 for Clinton), and the last one was on 4/9, one day after Sanders dropped out of Dem primary.

A5. In 2020, we've seen more "alternate reality" polls than usual, all favoring Trump. Not quality pollsters whose model favors Trump (e.g., IBD/TIPP, Emerson), but those who appear to be purposefully skewing with unsupported (or even outlandish) methods.

A5a. E.g., Trafalgar’s "social desirability" skews ~R+5%, w/no firm data to support it, w/results 5-10% to the right of averages. Susquehanna’s FL poll excludes new voters, despite polls showing record-level turnout, thanks to new voters (data below).

A5b. Some Trump supporters will point out that Trafalgar got MI right in 2016, but their "accuracy" is largely restricted to races that break late for GOP candidates. They are often way off when a race breaks for the Dem. That's not predictive, that's partisan.

A5c. Using only A+ to B/C rated pollsters removes most of these from my data, though I'm still using B-rated pollsters who appear to be engaging in similar practices (InsiderAdvantage, Atlas, Spry). But given in/exclusion practices of RCP, Biden likely overperforms their averages

A6. Biden’s lead is also the STEADIEST in modern polling. Why? Because both candidates have 95-100% name recognition, and both have served in the Executive. Add to that extreme political polarization, and you have a historically steady race. But also...

cnn.com/2020/09/06/pol…

A6a. Voters just LIKE Biden more than Trump:

He’s currently +8 in un/favorables, 22% BETTER than Trump, whereas Clinton’s rating was only slightly better than Trump's in 2016. Trump's "very/highly" unfavorable is also 14% higher than Biden's!

fivethirtyeight.com/features/ameri…

So why all these comparisons to the last election? First, there are some valuable contrasts to be made, especially since Biden’s running against the same opponent.

But mostly, it’s cuz if ya don’t, people who just don’t like the data will lazily say "BUT 2016!!!"

B1. How's he doing it? Per A/B-rated nat'l polls, 10/1-10/31 vs 2016 exits, he's outperforming Clinton with:

White: +15
White/college: +20
White/no coll: +18
Men: +9
50-64 years: +10
Over 65: +13
College grads: +14
Indies +15
Moderates: +26
Suburban +20

B1a. The most noteworthy of these is that Biden is just 5%(!!!) behind Trump with white voters! No Dem has won this group since LBJ in '64. Only others to get within 5% were Carter 76, and Clinton 92/96, two of the last three Dems to win the presidency. All others lost by 10-20%+

B1b. Reflective of this, he's doing 20% better with white college grads, a group that’s been a key swing demographic since 1988. More importantly, he’s doing 18% better with white voters who didn’t go to college, a group that turned out BIG for Trump in 2016.

B1b. The next most notable data point is that he's essentially TIED with men. The last three to do it were Carter 76, Clinton 92/96, and Obama 08. All others lost by 10-20%+.

B1c. He’s winning two other traditional GOP blocs, namely voters 50-64, and over 65. Only Carter and Gore have carried the latter group since 1972. He’s also outperforming Clinton among voters with a college degree by 14%, a key swing group in the last few cycles.

B1b. Biden's been doing historically well in the suburbs, currently outpacing Clinton by 20%! This is notable, considering Trump’s less-than-subtle appeals to suburbanites, who were a key swing demographic in 2016.

cnn.com/2020/07/19/pol…

B1c. Biden is absolutely OWNING the political middle, doing 15% better with independents, and 26%(!) better with moderates.

B1d. A popular myth these numbers explode is that Biden's struggling with young voters. While pollsters divide age groups differently (e.g., 18-29 vs. 18-34, 30-44 vs. 35-50), he’s running 9% stronger than Clinton among the youngest group, 18-29.

fivethirtyeight.com/features/biden…

B1e. One other myth to put to bed is that Biden struggles with liberals. In fact, he’s outpacing Clinton by 15%, while running basically even with her among conservatives.

B1f. After a stronger September, Biden’s weakness w/Hispanics appears to have returned. In fairness, there is an unusual level of variation among the individual crosstab results, so the margin is debatable, but it’s clear that he is not matching Clinton’s margin.

B1g. Similar story with Black voters, and an unusually high level of variation between individual crosstab results. That said, the overall variation from 2016 is smaller, just 9% short of Clinton’s support level.

C. Finally, let’s talk about some key factors that will impact how the final results come out. Also, at the end, I will finally make a... GUESS at the final result 😲

C1. Enthusiasm: polls show it’s equally high for both candidates, unlike 2016, when the Comey letter led to a sharp dip in enthusiasm among Clinton supporters, and a rise in Trump’s as polls got close enough that supporters began to believe he could actually win.

C1a. Desire to beat Trump ("negative partisanship") turned out record numbers in primaries (). Given the sharp difference between the two candidates’ un/favorable numbers, that’s advantage Biden.

.C2. Early voting: first, anyone telling you they can predict the result based on early voting is lying, misinformed, or deluded. We don’t know who and how many will turn out on election day, so unless 100% of eligible voters vote early, nothing is "over."

C2a. Polls show a record-breaking 70-80%+ of eligible adults plan to vote. 73% of Biden supporters planning to vote early, while 55% of Trump supporters plan to vote on Election Day. Polls show a 59-34 Biden advantage among those who’ve already voted.

C2b. But % planning to vote on election day has decreased ~10% in more recent polls (top of chart), as "already voted" numbers go up, so maybe some Trump supporters aren’t waiting for election day after all. This is also suggested by tightening partisan margins in early votes.

C2c. Even if Dems are ahead by election day, remember they were in 2016 too. For example, Clinton had a 90,000 early voting lead in FL on Election Day, but Trump won the state by 107,000 votes.

But here are a few key factors to watch as results roll in:

politico.com/states/florida…

C3. Turnout level: Monmouth releases high/low-turnout versions of battleground polls, and high favors Biden EVERY time, between 2-5%. Early voting turnout suggests we will see high overall rates.

Over 70% of eligible voters participating = good for Biden.

C3a. High turnout fueled in part by 2016 non-voters, who favor Biden 58-32. In 2016, new voters favored Clinton by similar margin, but were only 10% of the electorate. They're ~20% of early vote so far, but we'll see where the numbers are in exit polls.

nbcnews.com/politics/donal…

C3b. Biden also leads with those who voted 3rd party in 2016, 51-29. This group was larger than usual in 2016 at ~5%, so this could give Biden another point or so in tight races.

C3c. Young voters turning out big, but we won’t know how much of the *overall* electorate they'll be until election day. In 2016, 18-29 were 19%, but largest group was 45-64, who still favor Trump. 18-29 taking 5%+ from any group helps, but if its 45-64, BIG plus for Biden.

C3d. Black/Hispanic turnout varies WIDELY between states in early vote, but in FL/PA Dem officials worry % not as high as past elections. But if Biden maintains his polling margins w/white voters, he may not need esp high turnout from these two groups.

vox.com/2020/10/31/215…

C3e. Finally, district polling can catch warnings statewide polls miss, like they did in 2016. Basically ZERO for Biden.

Finally, my GUESS!

1. ME/CO/NM/VA not competitive, NH 10%+

2. MN/WI/MI averages have never slipped under +5%, currently +7-9% (remember: A+ to B/C rated pollsters, so no Trafalgar, Susquehanna, etc). ALL THREE have moved 1-2% toward Biden in averages since 10/1.

2a. Slight tightenening in PA, thanks entirely to the poll from Atlas (whose 9/1 nat’l poll gave Trump 28% Black support). But even w/o that, margins have always been a little closer than MN/WI/MI, and Trump’s going full bore with rallies there. Biden wins, but closer.

3. NV/NE-2 strong for Biden. Smaller lead in ME-2, where anti-Collins sentiment helps him win.

4. NC/GA/FL tight, Biden slightly ahead. Latest high-quality polls have trendlines moving 2-3% toward Biden. I expect Biden to win one/ all, but if I had a farm, I wouldn't bet it...

5. AZ widely inconsistent from one pollster to another, w/Biden slightly ahead in averages. He wins, with help from Kelly voters.

6. TX competitive thanks to Biden's cross-demographic appeal. An outright win, tho, will be thanks to Beto's organization, but I don't think he will

7. Competitive in OH/IA/MT, but like TX, I think he wins only w/3-4% overperformance, as part of a massive wave victory.

Final GUESS: Biden wins, with a margin between 280-258 and 351-187.

But REMEMBER: whatever polls show, they can CHANGE, and candidates can overperform. Clinton's polling slide continued until Election Day, leaving her w/5%+ leads in states worth only 203 EV:

So make your voting plan at IWillVote.com today, and VOTE!!!