Nate Silver: Trump 57.5% Chance of Winning

[deacon blues]

Nate Silver, a statistician that predicted all 50 states in the 2012 election has Trump winning the election as of this week 57.5% to Hillary's 42.5%.

https://www.washingtonpost.com/news/...trump-can-win/

[n david]

"""Our polls-only model has Trump with a 19 percent chance of beating Clinton as of early Wednesday afternoon. (The forecasts will continually update as new polls are added.) Our polls-plus model, which considers economic conditions along with the polls, is more optimistic about Trump, giving him a 26 percent chance."""

http://fivethirtyeight.com/features/...ing-president/

[deacon blues]

Originally Posted by n david

"""Our polls-only model has Trump with a 19 percent chance of beating Clinton as of early Wednesday afternoon. (The forecasts will continually update as new polls are added.) Our polls-plus model, which considers economic conditions along with the polls, is more optimistic about Trump, giving him a 26 percent chance."""

http://fivethirtyeight.com/features/...ing-president/

"Polls-only model" being the operative words here. Using his statistical model in 2012 he nailed all 50 states. The 57.5% probability figure comes from his statistical model.

[n david]

Originally Posted by deacon blues

"Polls-only model" being the operative words here. Using his statistical model in 2012 he nailed all 50 states. The 57.5% probability figure comes from his statistical model.

You missed the last part, where he stated the polls-plus model gave 26%.

Not sure what Nate is seeing, but it's highly improbable that Trump went from 26% a month ago to 55% today, especially with a new Gallup poll showing 51% are less likely to vote for Trump post-convention, with 36% more likely.

[deacon blues]

Yet DT in the polls was beating Hillary this week.

[deacon blues]

The 57.5% probability was in his "now cast" earlier this week. That's is not based on national polls but based on the electoral college if the election were held today.

[deacon blues]

From the Washington Post article that talked about Silver's predictions:

How can this be, many Democrats — and even some Republicans — ask? How, given all of the mistakes, gaffes and other issues that have dogged Trump's campaign, can he be tied or maybe even ahead?

First, see above. All of the rules are off. Assuming they aren't ignores recent history.

Second, Clinton is a uniquely flawed candidate. She has been in the national eye for a very long time and people have largely made up their minds about her. It is very hard for her to change those perceptions. What that means is she has a hard and relatively low vote ceiling; no matter what she says or does (or what Trump says or does) there is a rock-hard group of people who will not vote for her.

In a binary choice election — which is what this is — Trump benefits from the fact that he is simply not Clinton. What's happened to date in the race — pre-GOP convention — is that support has peeled off Clinton but not gone to Trump. Rather, it's moved to "undecided." The GOP convention, as Philip Bump documented here, led to some movement to Trump. Hence the tightening in polls.

What this all means is that the race is close today and there's plenty of reason to believe it will stay like that all the way until the election. For those who say Trump can't win, you probably said he would have never made it this far. All assumptions need to go out the window in an election like this one. Trump seems to understand that.

[n david]

Today it shows 55.3%.

http://projects.fivethirtyeight.com/...-forecast/#now

[n david]

Originally Posted by deacon blues

"Polls-only model" being the operative words here. Using his statistical model in 2012 he nailed all 50 states. The 57.5% probability figure comes from his statistical model.

From Silvers "User's Guide":

◾The now-cast is basically the polls-only model, except that we lie to our computer and tell it the election is today.
◾As a result, the now-cast is very aggressive. It’s much more confident than polls-plus or polls-only; it weights recent polls more heavily and is more aggressive in calculating a trend line.
◾There could be some big differences around the conventions. The polls-only and polls-plus models discount polls taken just after the conventions, whereas the now-cast will work to quickly capture the convention bounce.

So the 55.3% is polls-only (not statistical), but with him telling the computer the election is today.

I'm not a statistician, but again, it's highly improbable that Trump would have a 19% chance one month ago in Silver's polls-only model, and suddenly now leap to 55.3% -- again, polls-only model.

[jfrog]

Why is that unlikely?

Originally Posted by n david

From Silvers "User's Guide":

◾The now-cast is basically the polls-only model, except that we lie to our computer and tell it the election is today.
◾As a result, the now-cast is very aggressive. It’s much more confident than polls-plus or polls-only; it weights recent polls more heavily and is more aggressive in calculating a trend line.
◾There could be some big differences around the conventions. The polls-only and polls-plus models discount polls taken just after the conventions, whereas the now-cast will work to quickly capture the convention bounce.

So the 55.3% is polls-only (not statistical), but with him telling the computer the election is today.

I'm not a statistician, but again, it's highly improbable that Trump would have a 19% chance one month ago in Silver's polls-only model, and suddenly now leap to 55.3% -- again, polls-only model.