A quote from Nate at 538.....

[getagrip_already]

Still, Trump has a meaningful chance per our forecast — a little worse than the chances of rolling a 1 on a six-sided die

Think about that. Then get people to the polls.

Just sayin......

[Cirque du So-What]

1. I don't understand that analogy

Voting is not random like the toss of a die. Using Nate’s analogy, it seems that the die is restrained in its movement. Perhaps I’m missing something...

[vollehosen]

3. It's a probability

Trump's chances are basically tied to polls numbers being wrong.

[brooklynite]

6. Or an unanticipated outside event affecting the result.

[Claustrum]

5. What he means is if there is a large enough polling error, there is still a chance Trump can win

For example, WI, MI, and PA are about 5-6 points Biden lead right now. If the polling error is 5+, then Trump might still take those states, thus winning EC.

But historically, the polling error is between 0-3 points (2016 has 2-3 points error) other than 1980 that has a 7 point error. So it's very unlikely for Trump to win but it's not impossible.

[Silent3]

7. There's random noise in the data upon which the prediction models are based...

...uncertainties in the weighting of factors in the models, and there are always unforeseeable factors that aren't/can't be captured by the models. All that leads to randomness in how well a model will match a predicted outcome, and yes, that's a lot like throwing dice.

[getagrip_already]

8. It's the way his analysis works....

He runs thousands of model simulations with different parameters all based on polling what if cases. Then he looks at the results and graphs them as chances in a hundred.

For example, what if white non college educated men turn out at a rate 5% more than expected.....

Of all the possible outcomes to his simulations, trump wins 12 times out of 100. A 1 in 6 chance is about a 16% chance that any one side will come up.

Statistically, an 8 sided die would be more accurate, but people know what a 6 sided die looks like.

[Klaralven]

10. Voting is not random, but polling is

Polls are based on a sample of from a few hundred to a couple thousand people that were contacted and responded to the pollster.

Consequently, they are subject to both random errors due to the small sample size and to systematic errors due to some people being unreachable, unresponsive, etc.

Note also that if the gap between two candidates is 55 to 45% in a poll, a margin of 10%, a switch by 1% of the respondents would shift the result to 54 to 46%, a margin of 8%. That is, a switch by 1% of voters changes the margin by 2%, twice as much.

[PTWB]

2. If you knew you had a 10% chance of getting struck by lightning the next time you went out...

You’d be pretty freaking worried. It is amazing to me the folks who are acting like Trump doesn’t have a chance. Even Joe’s campaign manager said that in critical swing states their internal polling shows the race is “neck and neck” and also “functionally tied.”

GOTV! Everyone here is politically aware. Get your kids to vote, get your spouse, your ex-spouse, your parents and grandparents and brothers and sisters to vote. Get your co-workers and friends to vote. Volunteer to make phone calls on behalf of national or local candidates (or both!) - just DO SOMETHING!

[liberalmuse]

9. I don't know a single Democrat who isn't scared shitless.

I think most of us know the stakes. Any information that even slightly favors Trump makes my blood run cold.

[Claustrum]

12. The confidence for the democrats in 2016 has nothing to do with polling.

I was one of them. It was just a general feeling that no one in the world would elect a disaster like Trump. It wasn't because of the polls. Even after Comey and the polls tightened, I still thought there is no way Trump can win. That bubble has been burst and I, and many other, know that some people are that stupid. So I really don't think there is any complacency or overconfidence this year.

[Ferrets are Cool]

11. This

[NoMoreRepugs]

4. Perhaps the problem is that far too many people think that there's a possibility that a meaningful

number of the Deplorables will not support RUMP, to which I say, you are grossly underestimating how fucking stupid that portion of our population is.

[Warren_Pointe]

13. He referred to it as "Russian Roulette" on NPR this weekend.