PEC (Princeton) predicts Biden 360 EV.

[triron]

[Dr. Jack]

1. Didn't they give Clinton a 99% chance of winning in 2016?

I'm not saying Biden is going to lose because he's not but is Princeton's election site trustworthy? I feel like they were one of those sites giving Clinton a 99% chance of winning the 2016.

[Dem2]

2. Almost everybody was wrong in '16

There was an underlying current that wasn't seen it the polls, add in Comey and Russia and it was a difficult one to call...

One election does not a trend make.

[Dr. Jack]

3. I disagree

There certainly where websites like Huffington Post, Daily Kos, and Princeton that were saying Clinton was as close to 100%chance of winning as possible. But that isn't what polls were saying and that isn't what other experts, like 538, were saying. There was plenty of evidence in 2016 that it was going to be very close, so I personally wouldn't trust any site that got 2016 so wrong but claims now they know what they are doing.

[Dem2]

4. The polls were still moving at the time the voting took place

Let's not forget that, it was a difficult to capture phenomenon.

[Tom Rinaldo]

6. A comment on the bigger picture

In 2016 polls that seemed to indicate that Hillary was comfortably ahead fed into a sense of complacency among some potential Democratic voters causing some to stay home, and others to vote third party.

In 2020 polls that seem to indicate that Biden's support is surging are energizing Democratic Party coalition voters. If anything people are moved to pile on further to Trump's defeat, to give more and more money to Democrats up and down the ticket, and to drive every last nail possible into both Trump's and the Republican Party's coffins. People are eager to humiliate Trump and virtually NO ONE is looking for reasons to not bother voting. Trump was not a fully known quantity in 2016. Now he is. Hillary had negative approval ratings in 2016, Joe has positive approval ratings in 2020, There is no real comparison with 2016.

Meanwhile the building sense of certainty that Trump is going to lose is drying up donations to the Trump campaign, and more and more people, including some politicians but, more importantly, actual voters no longer want to be closely associated with a loser.

The perception that Trump is going down has become a psychological weapon on behalf of Democrats. It should not be tamped down!

[Yavin4]

5. There's a disconnect between polling and turnout.

Polls are a snap shot in time. They show the mood of the electorate, but they should not be used to make predictions because they do take turnout into account. People can tell a pollster whom they will vote for but the actual process of voting is something else entirely.

In general, the 2016 polls were right, but the turnout was down. That's why Hillary lost.

[Awsi Dooger]

7. Polls were not correct in the swing states

They underestimated white working class turnout and percentage for Trump. That's why analysts like Sam Wang of this Princeton Consortium were more wrong than anyone else, since Wang stupidly relies on state polls alone. He actually was moronic enough to actually brag about that through 2016, even as Nate Silver repeatedly chastised him for the ignorant approach. Nate's model gave Trump a higher chance because Nate recognized if there was a polling error it would likely translate to all the similar states. Wang's numbskull approach treats each variable as isolated from others.

The problem in 2016 was preference, not turnout. Comey's letter impacted preference. Late deciders also tilted sharply to Trump, whether impacted by Comey or not.

Turnout is always nothing but a sloppy lazy rationalization after the fact. It is impossible not to laugh uncontrollably when someone counts the difference in some state and announces that all Hillary needed was 14,000 extra votes. Oh sure. No problem. We're going to assign some fantasy world in which exactly 14,001 additional Hillary voters show up, and zero extra voters turn out for Trump or anyone else. I have no idea how the commentators say it with a straight face, or any type of belief they are making a fleck of sense. All the dozens of variables that combined for that 50/50 race are somehow ignored and now we're going to bake a cake of 14,0001 to 0.

That crap would be laughed out of the room in the Las Vegas environment I am accustomed to. That's why it is so maddening.

In 2020 some state poll models have adjusted for education. Others have not.

[triron]

8. No. Well, maybe a little bit. Mostly Russian manipulation. Effected turnout, election vote totals.

You probably will not accept this but I claim it's what happened.

[Yavin4]

9. Look up the turnout for Milwaukee, WI in 2016.

It was the lowest since 1996. Trump's core voters were more enthusiastic about Trump than Hillary's core voters were.

Turnout is not some "sloppy lazy rationalization after the fact". See the 2018 midterm elections.