Nate Silver (10/23/2018): "GOP chances of retaining control of the House are down to 14 percent"

[Bootney Lee Farnsworth]

Nate Silver on Hillary Clinton in 2016:

Final Election Update: There’s A Wide Range Of Outcomes, And Most Of Them Come Up Clinton

 

[easyt65]

The main reason some of the 2016 prediction models failed so badly (the ones that gave Clinton a 95+% chance of winning; the 538 model gave her ~70% IIRC) is that they failed to account for the fact that polling errors tend to be correlated. That is, they had terms to account for polling error, but they assumed that the errors in various states would be random, so that if the overall polling average in (for example) Pennsylvania was 2% too favorable for Clinton, that implied nothing about the accuracy of polls in Wisconsin. In reality, polling errors across states are likely to be correlated, which is what we saw in the rust belt. Failing to account for that made them overestimate her chances of success. Note that this isn't a problem with individual polls, or even with aggregates of polls. It was a failure of the specific prediction models. Polls themselves were about as accurate in the aggregate as they usually are, i.e. within ~2% of the national popular vote. But in a close election where the electoral college and popular vote are split a 2% error in favor of Clinton, along with the correlation in errors across states, is enough to give the impression that the entire enterprise was flawed. But mostly it's just that people are bad at interpreting statistical models, and some of the models (the 95+% Clinton victory ones) were flawed.

The main reason some of the 2016 prediction models failed so badly was because the Democrats attempted to effect voter turn-out by declaring early that there was no reason for Republicans to even go to the polls to vote because Hillary was already the President of the United States according to their false models / reporting - she was just waiting for it to become official.

If you remember, network TV no longer is allowed to / no longer provides predictions about what candidate won what states when only something like 15% of the districts have been polled / reported in like they used to do. It used to effect voter turnout....USED TO. 2016 Showed Americans are on to the Democrats and the Liberal media and just ignore that crap.

 

[BlackFlag]

Wow. The Republicans' chances to retain the House majority continue to sink.
You guys are in an extremely ugly position.

I am a person who has a lot of empathy, Therefore I feel bad for you.

Silver explains that Trump's marginally improved approval ratings are very inconsequential.
Trump’s Approval Rating Is Up. Republican House Chances Are Down. Does That Make Any Sense?

Nothing is more important for this nation than Democrats taking back at least the House.

 

[Ted Frazier]

Wow. The Republicans' chances to retain the House majority continue to sink.
You guys are in an extremely ugly position.

I am a person who has a lot of empathy, Therefore I feel bad for you.

Silver explains that Trump's marginally improved approval ratings are very inconsequential.
Trump’s Approval Rating Is Up. Republican House Chances Are Down. Does That Make Any Sense?

Well, Trump had a 1.6% chance of winning, but there you go:

That's not a Nate Silver forecast.

 

[well named]

The main reason some of the 2016 prediction models failed so badly was because the Democrats attempted to effect voter turn-out by declaring early that there was no reason for Republicans to even go to the polls to vote because Hillary was already the President of the United States according to their false models / reporting - she was just waiting for it to become official.

If you remember, network TV no longer is allowed to / no longer provides predictions about what candidate won what states when only something like 15% of the districts have been polled / reported in like they used to do. It used to effect voter turnout....USED TO. 2016 Showed Americans are on to the Democrats and the Liberal media and just ignore that crap.

I think that it's at least prima facie plausible that voter turnout among Democrats was negatively affected at the margins by the expectation that she was a lock to win, driven in part by the way media presented some of these models, but it's probably impossible to measure, and the data are not inconsistent with more banal explanations. Like I said, the polls were not significantly more wrong than they have been in past cycles. But the reason why some models suggested Clinton was 99% and others only 70% has nothing to do with these considerations.

 

[jc456]

The main reason some of the 2016 prediction models failed so badly was because the Democrats attempted to effect voter turn-out by declaring early that there was no reason for Republicans to even go to the polls to vote because Hillary was already the President of the United States according to their false models / reporting - she was just waiting for it to become official.

If you remember, network TV no longer is allowed to / no longer provides predictions about what candidate won what states when only something like 15% of the districts have been polled / reported in like they used to do. It used to effect voter turnout....USED TO. 2016 Showed Americans are on to the Democrats and the Liberal media and just ignore that crap.

I think that it's at least prima facie plausible that voter turnout among Democrats was negatively affected at the margins by the expectation that she was a lock to win, driven in part by the way media presented some of these models, but it's probably impossible to measure, and the data are not inconsistent with more banal explanations. Like I said, the polls were not significantly more wrong than they have been in past cycles. But the reason why some models suggested Clinton was 99% and others only 70% has nothing to do with these considerations.

i'll disagree with you here.

 

[well named]

i'll disagree with you here.

It's not at all clear what you are disagreeing with, or what you think that chart demonstrates.

 

[jc456]

i'll disagree with you here.

It's not at all clear what you are disagreeing with, or what you think that chart demonstrates.

fk some people are stupid.

You posted this: I think that it's at least prima facie plausible that voter turnout among Democrats was negatively affected at the margins by the expectation that she was a lock to win,

I posted the votes. you're not correct that democrats didn't come out, perhaps the difference in count is the rise in count to trump. You fking know absolutely nothing. Trump was up +2 million, and hitlery down -71K. and the other candidates were up +5 milliion.

In all a net 7 million more people voted. how the fk you can tell what party a person voted for from previous elections is beyond me, but the mere fact that there 7 million additional votes tallied in 2016 tells me many more people voted. So your entire negative perception is garbage.

 

[well named]

You posted this: I think that it's at least prima facie plausible that voter turnout among Democrats was negatively affected at the margins by the expectation that she was a lock to win,

I posted the votes. you're not correct that democrats didn't come out, perhaps the difference in count is the rise in count to trump.

You're confused on a couple points

1) "Prima facie plausible" doesn't mean true. In fact, the post you quoted was arguing against the conclusion that Democratic turnout was negatively impacted. I said I think there are better explanations for the data. So you entirely misunderstood my point to begin with.

2) Your chart doesn't actually show what you are claiming it does. The assumption that Clinton's vote totals being about the same as Obama's means that there was no negative impact on Democratic turnout is a non-sequitur. The voting population was larger in 2016 than 2012, as you noted yourself. If 7 million more people voted than in 2012, but Clinton did not get any more votes than Obama, that certainly could be consistent with a turnout problem for Democrats, although as I said I believe there are better explanations.

But one of the obvious problems with the "Democrats didn't turn out because they were sure Clinton would win" argument is that it's impossible to measure, because it depends on knowing how many people would have voted in 2016 absent the hypothesized turnout problem. That counterfactual is not settled by comparing to 2012, although I agree it provides some upper limit to the answer.

All of the above is why I wrote that this argument was possible "but it's probably impossible to measure, and the data are not inconsistent with more banal explanations." You seem to have overlooked that sentence (and the following) when you read my post.

 

[jc456]

You posted this: I think that it's at least prima facie plausible that voter turnout among Democrats was negatively affected at the margins by the expectation that she was a lock to win,

I posted the votes. you're not correct that democrats didn't come out, perhaps the difference in count is the rise in count to trump.

You're confused on a couple points

1) "Prima facie plausible" doesn't mean true. In fact, the post you quoted was arguing against the conclusion that Democratic turnout was negatively impacted. I said I think there are better explanations for the data. So you entirely misunderstood my point to begin with.

2) Your chart doesn't actually show what you are claiming it does. The assumption that Clinton's vote totals being about the same as Obama's means that there was no negative impact on Democratic turnout is a non-sequitur. The voting population was larger in 2016 than 2012, as you noted yourself. If 7 million more people voted than in 2012, but Clinton did not get any more votes than Obama, that certainly could be consistent with a turnout problem for Democrats, although as I said I believe there are better explanations.

But one of the obvious problems with the "Democrats didn't turn out because they were sure Clinton would win" argument is that it's impossible to measure, because it depends on knowing how many people would have voted in 2016 absent the hypothesized turnout problem. That counterfactual is not settled by comparing to 2012, although I agree it provides some upper limit to the answer.

All of the above is why I wrote that this argument was possible "but it's probably impossible to measure, and the data are not inconsistent with more banal explanations." You seem to have overlooked that sentence (and the following) when you read my post.

everyone of the 71k that she was short could have voted for either trump or the other two candidates. It's what I said you couldn't prove. you can't, you can't make any statement that the democrats didn't come out and vote. Now you could say they didn't vote for her, but not that they didn't come out. And the negative number may actually tell us, she wasn't liked by her own party. just maybe? And trump got 2 million more than romney, so more republicans and independents came out as well. She had everyone worried about her shit. That more proves the opposite of your argument that many more people didn't actually want her as president and she was a bad candidate.

 

[well named]

That more proves the opposite of your argument that many more people didn't actually want her as president and she was a bad candidate.

You're still confused about what I'm arguing for. You're mostly agreeing with me. Please try reading my entire post. I'll say it again: I was arguing against the conclusion you think I was arguing for.

In any case, it's not exactly true that "more people didn't want her as president" (she won the popular vote), but I agree that she was a bad candidate in several ways, and I agree a better candidate may have done better than her. If we qualify your statement to say that she was more unpopular than Trump in a handful of very important states, or among certain voter demographics that were pivotal in the election, then that is clearly true.

 

[jc456]

That more proves the opposite of your argument that many more people didn't actually want her as president and she was a bad candidate.

You're still confused about what I'm arguing for. You're mostly agreeing with me. Please try reading my entire post. I'll say it again: I was arguing against the conclusion you think I was arguing for.

In any case, it's not exactly true that "more people didn't want her as president" (she won the popular vote), but I agree that she was a bad candidate in several ways, and I agree a better candidate may have done better than her. If we qualify your statement to say that she was more unpopular than Trump in a handful of very important states, or among certain voter demographics that were pivotal in the election, then that is clearly true.

I'm not confused. The democratic party was not negatively impacted by its base not showing up to vote at all. not at all.

The rest I will give you.

 

[Mike473]

I think that the GOP is going to hold the house and have a big night in the Senate. I wonder how Dems will react to losing twice in a row after being told victory was all but certain?

 

[ptbw forever]

The 7,000-criminal invasion force marching on our southern border, along with the criminal, unethical. despicable attacks on Kavanaugh and Hillary's public plea for MORE liberal violent intolerance / violence are almost as good as running a proven felon who compromised national security as your Presidential nominee and is quickly shifting the tide, polls show.

The Democrats may even LOSE seats in the Senate.

They are going to lose seats in the Senate literally everyone knows this.

 

[hjmick]

Wow. The Republicans' chances to retain the House majority continue to sink.
You guys are in an extremely ugly position.

I am a person who has a lot of empathy, Therefore I feel bad for you.

Silver explains that Trump's marginally improved approval ratings are very inconsequential.
Trump’s Approval Rating Is Up. Republican House Chances Are Down. Does That Make Any Sense?

Nate Silver: 'Calm down,' Donald Trump won't win the GOP nomination

 

[Bruce_T_Laney]

Sounds about right.

so did trump had no path to the white house, how'd that work out for you?

What's your point?

Make it.

Point it the polls were wrong in 2016 and Clinton lost the Electoral College by a wide margin and had it not been for California she would have lost the Popular Vote also.

 

[Mike473]

Based on my calculations, the GOP has a 95% chance of keeping the house.

 

[Missourian]

Wow. The Republicans' chances to retain the House majority continue to sink.
You guys are in an extremely ugly position.

I am a person who has a lot of empathy, Therefore I feel bad for you.

Silver explains that Trump's marginally improved approval ratings are very inconsequential.
Trump’s Approval Rating Is Up. Republican House Chances Are Down. Does That Make Any Sense?

Sounds pretty close to the odds he gave Trump.

https://projects.fivethirtyeight.com/2016-election-forecast/

How'd that work out?

 

[Markle]

Yes, and we know how reliable those forecasts were in the past.

 

[Frankeneinstein]

He must have used his crystal ball and had a séance.

Silver is actually quite good at what he does, I have never suspected him of playing to the media.