Biden still has the advantage after #Election2020
but it's very clear that pollsters & modelers like Nate Silver badly messed up likely voter models.
I tried pitching publications on this topic but editors love group think and were too afraid. Here's some of what happened:

I tried pitching publications on this topic but editors love group think and were too afraid. Here's some of what happened:
Firstly, a huge number of mailed ballots are never returned by voters. Estimates put it between 10 to 20 percent. Likely voter models should have removed some of these people, the plurality of whom were Democrats.
The Trump campaign began boasting about this in mid October but not one MSM or liberal outlet ever responded to this boasting. https://www.breitbart.com/politics/2020/10/22/data-shows-democrats-failing-early-voting14790167/
Democrats who were at the state level, especially in Florida, SAW THIS COMING. They tried to sound the alarm but were ignored by national Dems.
First time mail voters are much less likely to be good at doing the process successfully so Trump's warning to his voters probably saved a lot of them from messing up what can be a difficult thing for elderly, disabled, or people who don't read English well
Florida Democratic organizers tried to warn the Biden campaign in July that they were wasting too much money on TV ads and not enough on connecting to citizens. They were ignored. https://www.miamiherald.com/news/politics-government/article244480437.html
As a result of the Biden team failure, instead of getting a lopsided number of absentee votes, he ended up with 50.2% versus 48.9% for Trump. https://www.nytimes.com/interactive/2020/11/03/us/elections/results-florida.html
This is why Biden lost Florida, pure and simple.
Another flaw of the likely voter models used this year was that they didn't seem to reflect the several states that force people who request a mail ballot but change to in-person to cast a provisional ballot.
This means it's not counted on the night of #Election2020
This means it's not counted on the night of #Election2020

Then of course there is the fact that mailed ballots have a higher rejection rate. This further lowered the Biden vote total, although not to a huge degree. Was it reflected in likely voter models? Beats the hell out of me. No one discussed it as far as I can tell.
It is tough to say nationally how much all of these factors contributed to lowering Biden's totals but my private estimate was that between 2 and 5 percent of Biden support in polls would not survive the voting process.
What's unfortunate about political data and journalism though is that because most editors don't understand math well, they fear publishing others who do.
Center left media in the US is also very afraid of criticizing Dem leaders.
Center left media in the US is also very afraid of criticizing Dem leaders.
And so that's is why America is in such a strange and precarious situation right now.
I'm tired of this and will be starting a new outlet very soon to tackle the questions that aren't asked. Please follow if you're interested.
I'm tired of this and will be starting a new outlet very soon to tackle the questions that aren't asked. Please follow if you're interested.
Just to continue on the thread, here is a look at the very low return rates of mailed ballots as currently tallied by @ElectProject.
They're still being counted but Biden got absolutely killed by a very low 56% return rate by independents and fairly low 71.6% among Dems
They're still being counted but Biden got absolutely killed by a very low 56% return rate by independents and fairly low 71.6% among Dems