I will now accept my accolades megathread

[Crumpets]

The election obviously hasn't been called yet, but let's start a thread for people to post their accurate predictions that we know for a fact have come true.

[VAR]

See sig.

[lfromnj]

Well I called VA05 as Likely R

[Penn_Quaker_Girl]

(Ugh I hate doing this but...)

I insisted throughout this cycle that Biden would not win over 300 EC votes.  My final prediction was 278/279-290.   

[Orwell]

I have no accolades I'm just extremely sad

[Farmlands]

I am glad to have stuck to my gut and given Florida to Trump at the last minute. Looking at the early vote thread, you'd think there was no way Biden was losing the state, but it was the exact same tale in 2016.

[ملكة كرينجيتوك]

I knew the “shy Trump vote”/“non response bias” was real, just underestimated how big it was. AAPIs probably ended up voting more D than Latinos after all, but I figured both would swing D not R...

I feel vindicated in not supporting Biden or Sanders in the primaries, since it isn’t obvious Sanders would’ve done much better in the general than Biden (bless his campaign managers’ hearts).

[Okay, maybe Mike Johnson is a competent parliamentarian.]

GA and NC left of FL.

[Hope For A New Era]

Remember those posts I made freaking out about North Carolina (particularly the senate race, but the presidential too) this past week? Talking about how Tillis seemed to have suddenly broken through his ceiling and how Sunbelt states seemed to be all converging on Biden +1 at the last moment?

I think I get some accolades for that.

[RaphaelDLG]

(Ugh I hate doing this but...)

I insisted throughout this cycle that Biden would not win over 300 EC votes.  My final prediction was 278/279-290.   

I'd bet he still breaks 300 even now

[𝕭𝖆𝖕𝖙𝖎𝖘𝖙𝖆 𝕸𝖎𝖓𝖔𝖑𝖆]

I spat out this hot take four months out.

I think it should count for the accolades.

I don't even believe in my own take, but:

Trump gains a surprising number of votes among non-college-graduate Hispanics.

[Middle-aged Europe]

Two days ago I issued the hot take that Biden will lose PA but win the election and if that really comes to pass I will of course be willing to accept my accolades.

https://talkelections.org/FORUM/index.php?topic=409802.msg7719464#msg7719464

[OSR stands with Israel]

Texas was never a tossup despite all the gaslighting TrendsAreReal did

[Neptunium]

I think I would only lost ME02 and NE02 in my first 2020 prediction.

and Ohio and Iowa are not tossup.

[Okay, maybe Mike Johnson is a competent parliamentarian.]

Texas was never a tossup despite all the gaslighting TrendsAreReal did

Forecasting something that turned out to be wrong is not gaslighting, OSR.

[OSR stands with Israel]

Texas was never a tossup despite all the gaslighting TrendsAreReal did

Forecasting something that turned out to be wrong is not gaslighting, OSR.

But mocking people non stop who thought otherwise is gaslighting. He would constantly say i was living in fantasy land for saying it was not a tossup and was just in denial .

[Ferguson97]

My prediction of AZ voting to the left of WI seems to be accurate.

[Xing]

Biden is still going to win NV (although it'll be much closer than I expected), even though the night didn't go as well for Biden as expected.

[Senator Incitatus]

I don't really do confident predictions and think you're insane if you do (see my accurate signature), but I had a lot of correct hunches. You can basically get a lot of credit by just having been slightly Biden skeptic, but here are a few pre-election takes that are looking good:


On polling being wrong:

Yes, I can. There are two main reasons I am skeptical of polls (and a third category that confirms my skepticism).

1.) Non-response bias, redux. In 2016, this was the primary reason polls were off in places like Wisconsin and Michigan. In short, 2016 was the first election in recorded history where there was a significant gap in "social trust" between supporters of the two candidates; Clinton supporters were dramatically more like to poll highly on "social trust" as measured in an apolitical sense. People with lower "social trust" are less likely to answer opinion polls and less likely to answer truthfully when they do.

Pollsters have spent much of the time since 2016 trying to correct for this error with varying success. But I think with the rise of things like QAnon (on the further fringe) and the death of Jeffrey Epstein (which has dramatically affected even mainstream social trust), this gap has grown larger at a rate pollsters haven't accounted for. This has been shown in a number of special elections where Republicans outperformed expectations and Republican primary elections in which "fringier" candidates like Lauren Boebert and Marjorie Green Taylor have outperformed expectations.

n.b. This polling problem wasn't as large in 2018, where Trump wasn't on the ballot, but polls still did slightly underestimate Republicans in some states with relatively high white populations without degrees (the group least likely to have "social trust"), like Florida, Missouri, Iowa, and Ohio. In 2020, the impact of this error should be unevenly distributed. That's why I take both Georgia and Wisconsin seriously as competitive states; polling in the former is less likely to suffer from this error.

2.) The pandemic. This doesn't affect polls in a specific direction, but the pandemic has dramatically undermined a number of assumptions that are key to polling models. Turnout should be dramatically different with a shift to VBM and population distributions have shifted massively.

Exodus from cities and across state lines will shuffle voting populations in a way that no model currently accounts for (or could). New Yorkers who have moved out or back to their parents' should boost Democratic votes in otherwise Republican suburbs; Las Vegans leaving Nevada should leave the population whiter and wealthier, meaning that polling based on traditional populations underestimates Trump. Unemployment, as well, should materially influence voting patterns -- but I'm not certain how.

*3.) Hard numbers. See PM's "fundamentals" post above (though not as extreme). 2020 primary turnout, voter registration, and VBM returns in key states all break even, slightly favor Trump, or favor Biden by much less than polling would indicate. This is subordinate to both points above; it doesn't create skepticism, but rather reinforces it. If the hard numbers were in line with a big Biden win, I'd tend to believe the polls despite my problems with them.

n.b. None of this is to say Trump will or even can win.

Be more specific, DTT.  What mechanisms point to a Trump victory?

The growing gap in social trust between Trump voters and Biden voters, which led to non-response bias in 2016. Attempts at correction by pollsters have been underwhelming, even relative to the 2016 gap, and the rate of divergence has accelerated. There has been a complete collapse in social trust by Trump voters.

They don't point to a Trump victory, but they point to a race significantly closer than polling suggests, especially in places where the Trump campaign has done lots of direct voter outreach like Pennsylvania, Wisconsin, and Michigan.

(Additionally, the unpredictable nature of the present crisis makes modeling turnout difficult or impossible. This favors neither candidate, but it suggests a greater error bar than pollsters may be accounting for.)

This is starting to become a bit like reading chicken entrails. All the polls are wrong, we must find out what people really think! Trump will win because gun ownership has gone up!!

Ask yourself which is more abstract from voting decisions: a thousand-dollar, highly politically-charged purchase or a brief phone conversation with a pollster. (Or party registration.)

Campaigns use all sorts of data (gun ownership, car ownership, pets, number of children) to model voter behavior, not just polling responses. I get that this is a forum mostly dedicated to discussing polls, but be careful about giving them undue weight. They're also a bit like reading chicken entrails, even knowing the scientific rigor of their composition.

I wonder if part of the reason CD polls show such a major drop for Trump (besides being mostly taken in swing districts, where he's lost more ground than in rurals) is that knowing who is running for House is an informal proxy for education.

The CD polls were uniquely terrible. We don't know if that's the reason for certain, but I stand by that hunch.


On trends:

...A big part of the reason Trump won 2016 and has even an outside shot this time is the drive to register new voters and turnout inactive voters from among the disaffected lower and middle class...

This appears to have held up as the defining characteristic of the race.

I've said this before, but the Cuban-American tendency to single-issue voting against improved relations with Cuba never ceases to amaze me. I can't think of any other diasporas, even other diasporas of political refugees, that are that reflexively hostile to the "old country".

Venezuelans, in the same state. I don't know if anyone here follows Spanish-language media but there have been a ton of think-pieces PLEADING with Venezuelans, especially in Florida, not to vote Trump.

Trump's gains were in Spanish-language communities across the board.

Gender will be the defining divide of 21st century politics.

State- or race-specific takes:
I made a whole thread on VBM rejection rates that you lot derailed: https://talkelections.org/FORUM/index.php?topic=402834.msg7639416#msg7639416

I also said multiple times, though not ever in one long, self-contained post, that I thought Nevada was the state with the most potential to surprise, due to unemployment and COVID displacement. That seems to have held up.

Not sure if this has already been said, but outside chance Harrington becomes the top-performing Libertarian candidate ever, both in raw votes (369,807) and percentage (29.16%), though those were notably in three- or four-way races.

He did both.

Deve Daines, Owner of the Dainesdale Dimmadaines +6

Bredesen redux.

Currently +7.2 and probably going to tighten slightly. Didn't see anyone go higher.

James (40%), Harrison (30%), Bollier (10%), Gross (refusetopredictAlaska%)

Rating the likelihood of candidates. This seems to have held up, although I did elsewhere bump James's chances.

Some things I was dead wrong about:

• Florida to the left of Georgia (no clue what I was thinking there, could've been a typo)
• I thought Sara Gideon would win relatively comfortably, though I did counter the argument that it would be a Blanching.
• I thought Kevin O'Connor could get near 45% of the vote. (In fairness, this was before he decided not to run a campaign.)
• I posited two contradictory possibilities: that the electoral college advantage for Republicans would expand to the highest since 1948, but also that state-level depolarization could occur due to COVID. It has to be one or the other, though we don't know which yet. Leaning towards the latter.
• I said Wyoming wouldn't be Trump's best state. (In my head, I was thinking of North Dakota.) With almost all of the vote in, that seems clearly wrong.
• I thought Trump would lose Texas suburban whites faster than in any other state. We can't say for sure yet, but he seems to have held more ground there than in, say, Ohio. It's actually possible he broke even with suburban Texas whites, but that the suburbs are becoming dramatically less white.

[True Federalist (진정한 연방 주의자)]

While not a presidential accolade, my prediction that Graham would easily defeat Harrison because of people voting straight ticket rather than for him directly appears to have been borne out.

[McGarnagle]

My map wasn't fully accurate - but if MI, PA and NV get called for Biden as it appears they will be, I only missed FL, NC and GA. Definitely not as bad as my 2016 prediction, but not as accurate as I had hoped to be. Maybe one day, I'll get 'em all right.

Yes, Trump still leads PA - but a lot of the ballots that haven't yet come in appear to be from Philadelphia and Pittsburgh, where Biden leads handily. He may be able to flip it yet.

[Meclazine for Israel]

I predicted COVID-19 infection would cost Trump the top job.

I am predicting Trump will be removed by Secret Service agents in the middle of the night.

[Hammy]

I'll accept mine for:

Sticking to my Arizona prediction based on underpolling of Dems
Sticking to my GA/TX/NC predictions--following my gut (which is usually right with predictions) than Atlas enthusiasm
Keeping Florida R on my non tossup map
Never buying the Iowa/Ohio bandwagon
WI being the first state to go to Biden of the Rust Belt trio, and Pennsylvania being Biden's weakest

By the looks of things, my only two missed calls were ME-02 and Pennsylvania

[Beefalow and the Consumer]

Re: what margin in Hamilton county (IN) will end Trump quickly?

If Biden is within 10 points, Trump is in trouble.

If Biden is within 5 points, it's probably an electoral landslide.

With 95% of the vote in, Trump is ahead in Hamilton County, 54.2% - 43.4%. Trump won by just barely over 10 points, and Biden just barely won the election. I nailed that one. It's good to be right about something occasionally.

[Not Me, Us]

My map wasn't fully accurate - but if MI, PA and NV get called for Biden as it appears they will be, I only missed FL, NC and GA. Definitely not as bad as my 2016 prediction, but not as accurate as I had hoped to be. Maybe one day, I'll get 'em all right.

Yes, Trump still leads PA - but a lot of the ballots that haven't yet come in appear to be from Philadelphia and Pittsburgh, where Biden leads handily. He may be able to flip it yet.

Outstanding votes are also heavily mail-in, which skew extremely Democratic, and almost all are from Allegheny, Philly, and Philly suburbs.