Clinton hanging on to an electoral lead by a thread in 538's polls+forecast

[Reck]

http://fivethirtyeight.com/features/...d-poor-voters/

How ty of a candidate do you have to be to be hoping that this is the stuff you wish to happen so the other person doesn't win?

[ducks]

Who is wishing it?

[baseline bum]

Who is wishing it?

You're not?

[Reck]

So Nate put her chances of winning at barely 60% yesterday and today she's almost at 70%. Why the quick turnaround?

[Reck]

BTW baseline bum Nate has flipped Nevada, North Carolina and Florida all back leaning blue in all 3 forecasts.

[florige]

So Nate put her chances of winning at barely 60% yesterday and today she's almost at 70%. Why the quick turnaround?

S bag Comney announcement Hillary emails nothing there maybe?

[baseline bum]

BTW baseline bum Nate has flipped Nevada, North Carolina and Florida all back leaning blue in all 3 forecasts.

Good to see Clinton rebounding there in his model, but , now he's got New Hampshire going for Ayotte. What's Silver's record like in handicapping state races? Betting markets seem to have Hassan as a slight favorite while Silver has Ayotte as one.

[Reck]

Good to see Clinton rebounding there in his model, but , now he's got New Hampshire going for Ayotte. What's Silver's record like in handicapping state races? Betting markets seem to have Hassan as a slight favorite while Silver has Ayotte as one.

Man, I dont know. Nate has been panicky with his forecast lately. All over the place.

But most predictions from different election forecasters give Hassan the slight edge.



http://www.nytimes.com/interactive/2...-forecast.html

[MannyIsGod]

Those states haven't really "flipped". Silver's model may give you a percentage down to a tenth of a point, but it doesn't actually have that precision. A state race that is 51-49% is basically no different than 50-50, or 1:1. A tossup in the model. Especially with the sigma bars around the percentages. The model is essentially saying the same thing it has all weekend with the presidential race.

The senate races are much more volatile. Sadly the Hispanic vote can't help in many of those cases except Nevada. In Florida, Marco Rubio does much better with Latino voters than Trump, so its unlikely to sink him there. The senate races and the state prop votes will be the real things to watch tomorrow. I will be exceptionally dissapointed if the GOP holds onto the senate.

[Axl Rose]

BTW baseline bum Nate has flipped Nevada, North Carolina and Florida all back leaning blue in all 3 forecasts.

Polls no longer matter, trump outperforming Romney in those states by 100s of thousands of votes

[florige]

Polls no longer matter, trump outperforming Romney in those states by 100s of thousands of votes

Gotcha

[Axl Rose]

You think polls matter more than the actual votes coming in?

[HarlemHeat37]

I dropped $2500 on Hillary @-325 on Friday morning, tbh..the number went up dramatically on Saturday night to about -575..massive jump..

I expect it'll be -650 by tomorrow IMO..

[Warlord23]

Polls no longer matter, trump outperforming Romney in those states by 100s of thousands of votes

Do stick around on the forum for the next couple of days, old sport. And keep guns and sharp objects locked up before they start announcing the results.

[florige]

You think polls matter more than the actual votes coming in?

Idk maybe Trump will pull the impossible and shock the world and prove all polling is not right. The polls the night before the election are usually the only ones that I've ever seen hold true.

[FuzzyLumpkins]

Fuzzy, what physics is Silver's model built on? Silvers trend line and state by state adjustments are built on pure statistical correlations between those variables. That isn't what has driven those relationships. For instance, PA and OH are linked historically because of their similar demographic makeups. However, if this changes, then the correlation changes as well. You can't just say that they stay the same and still use them in your model. If the model were built upon the demographics as opposed to the historical relationship then it would be more robust to changes. Second of all, Climate Models are publilc. They're available for all to view and understand the underlying physics that go into them. Its a valid critique of someone touting how great their PRIVATE FOR PROFIT model is to point out that we don't know whats in it.

Your comparison to climate deniers is extremely superficial and lacks merit if you dig even a tiny bit into what my points were.

Your final point hits the nail on the head. But if Silver's model is based on simple statistical relationships of a changing environment then its increasingly less valid. Especially when the polls themselves are getting worse. Its still the best prediciton method, but Silver should understand these caveats better than most which is why his childish rant yesterday is ridiculous. His ego has gotten the better of him and I suspect that this election will see his model perform worse than either of the last two.

To be clear, I don't have a problem with Silver's methods. I don't think they are arbitrary and I think he has statistical reasoning for it. But I also think they're wrong in many instance; none more so than Nevada. This is because we have further information that shows that. Prediction markets who have factored this in show Clinton at about an 8.5:1 favorite while Silver essentially shows a 1:1 race in Nevada. He also rarely tallks about how poorly Latinos are being sampled by polls this year. And unlike the GOP unskewing talk of 2012, there is a lot of obvious evidence that this is the case.

Then you have no problem saying that his approach is empirical.

I think the word you are looking for is mechanics and I agree with you that his approach doesn't appear to be comprehensive but what he does have is clearly empirical. Overall I don't think we are not in accord. We both have presented plausible and testable hypothesis as to what he is not considering. I have been crowing that the electorate has been at a tipping point for some time with the end of the boomer hegemony after all.

That being said, don't feign certainty where none exists. It could be that he is right and we are wrong.

[Splits]

Good to see Clinton rebounding there in his model, but , now he's got New Hampshire going for Ayotte. What's Silver's record like in handicapping state races? Betting markets seem to have Hassan as a slight favorite while Silver has Ayotte as one.

In 2008 he predicted 50/51 Pres states and every Senate race correctly.

In 2012 he predicted 51/51 Pres states correctly, but don't recall the Senate. I think he had 2 misses (ND and some other ty state out West).

[CosmicCowboy]

margin of error is margin of error.

[Reck]

margin of error is margin of error.

The race is not within margin of error. The are you talking about?

[CosmicCowboy]

The race is not within margin of error. The are you talking about?

From 538

Even at the end of the campaign, the polls probably won’t perfectly predict the results. Polls get more accurate as Election Day approaches, but even on the eve of voting they’re not perfect. Everyone talks about the margin of sampling error — the error introduced by not surveying every voter. But polls are also subject to all kinds of other errors, none of which disappear as we get closer to the election. The average presidential poll within the final 21 days of the election has been off by 3.6 percentage points since 2000.

[FuzzyLumpkins]

From 538

That is him describing why there is a margin of error and not that it is within the margin of error.

[baseline bum]

Nice, the transit strike in Philly is over.

[UNT Eagles 2016]

still thinking HRC 279-259 stay tuned

[UNT Eagles 2016]

I dropped $2500 on Hillary @-325 on Friday morning, tbh..the number went up dramatically on Saturday night to about -575..massive jump..

I expect it'll be -650 by tomorrow IMO..

your ROI is what, like a few hundred bucks, if that? Maybe you should have bet $50k with the Koch brothers.

[Reck]

Nice, the transit strike in Philly is over.

ducks