The US Election Polling / Data Thread

[direwolf]

Originally Posted by Drak

Dixie Strategies looks suspect by name alone.

Yea the more I look at both of those polls, they do look a little suspect. Also, 538 rates them pretty low at C+, as compared to the A and B rated polls that mostly show Clinton being up.

[Senator Clay Davis]

Hate to do another calm down post but...I posted the Bloomberg article in the other thread, but here's the Trump campaign admitting they are well behind.

Several things jump out. Despite Trump’s claim that he doesn’t believe the polls, his San Antonio research team spends $100,000 a week on surveys (apart from polls commissioned out of Trump Tower) and has sophisticated models that run daily simulations of the election. The results mirror those of the more reliable public forecasters—in other words, Trump’s staff knows he’s losing. Badly. “Nate Silver’s results have been similar to ours,” says Parscale, referring to the polling analyst and his predictions at FiveThirtyEight, “except they lag by a week or two because he’s relying on public polls.” The campaign knows who it must reach and is still executing its strategy despite the public turmoil: It’s identified 13.5 million voters in 16 battleground states whom it considers persuadable, although the number of voters shrinks daily as they make up their minds.

Trump’s team also knows where its fate will be decided. It’s built a model, the “Battleground Optimizer Path to Victory,” to weight and rank the states that the data team believes are most critical to amassing the 270 electoral votes Trump needs to win the White House. On Oct. 18 they rank as follows: Florida (“If we don’t win, we’re cooked,” says an official), Ohio, Pennsylvania, North Carolina, and Georgia.

So two things stand out, one significantly more than the other: One, they say models like 538 are mostly right, except for timing differences. They are effectively conceding their internals show much of the same external polls have showed. But the significant thing that stands out is they put Georgia on the list of states they have to win. Georgia should not even be close, so that they conceded that it is close pretty much tells you all you need to know. In the same article they also point out driving down Hillary's turnout is probably their best chance to win.

[opendoor]

Originally Posted by Drak

What did 538 look like (snap shot) of this time 4 years ago?

On this date in 2012 they had Obama with a 73.6% chance of winning.

[photon]

The snake will keep me safe, the snake will protect me.

Even if she doesn't take Ohio or Arizona, that's a lot of states Trump has to take including Pennsylvania.

[Senator Clay Davis]

Originally Posted by Drak

Dixie Strategies looks suspect by name alone.

Forget the name, Trump has a 53% approval rating in that poll. Pretty hilarious.

[mrkajz44]

Originally Posted by direwolf

I've seen quite a few negative polls for Trump this morning, and yet he continues to tick upwards in the 538 forecast, which is odd. He's now at 19% in polls plus, up 2 percentage points from yesterday, possibly due to Florida and Nevada being fairly tight at the moment.

http://projects.fivethirtyeight.com/...d=rrpromo#plus

Photon posted this link a few pages back, but it's a really good summary about why 538 gives Trump much better chances than other forecasters.

http://fivethirtyeight.com/features/...hers-on-trump/

Basically, their model considers that all the states are much more interconnected than the other models. So if it looks like certain states are close races, then other similar states would likely also be close as well. The model subscribes to the theory that a surprise win in one state would likely mean multiple other states would fall differently than predicted as well.

[photon]

Some interesting early voter poll numbers:

Clinton leads 48 – 42 percent among Georgia early voters

Clinton leads 61 – 27 percent among Iowa likely voters who have cast ballots.

North Carolina likely voters who cast early ballots go 62 – 34 percent for Clinton

https://poll.qu.edu/images/polling/p..._S397fgbw.pdf/

[GGG]

Likely a sign of ground game rather than of that level of support. My theory is that it should mean that Clinton will outperform polling by a few points.

[photon]

Yeah that wouldn't be surprising. Trump's also running out of money (http://www.nbcnews.com/politics/2016...roblem-n674341) with only $16 million on hand and $2 million in debt as of Oct 19, which would hurt his get-out-the-vote efforts.. if he had any...

[Street Pharmacist]

Originally Posted by photon

Yeah that wouldn't be surprising. Trump's also running out of money (http://www.nbcnews.com/politics/2016...roblem-n674341) with only $16 million on hand and $2 million in debt as of Oct 19, which would hurt his get-out-the-vote efforts.. if he had any...

He's said he's going to start self funding again. Any day now. Any day

[Jordan!]

Fiance and I got our ballots in and confirmed they're counted. 2 for Hill dawg. Turn AZ blue!!

[Iowa_Flames_Fan]

You obviously don't have to disclose if you don't want, but I am curious about your senate vote... Do you think many Hillary supporters will be voting for McCain?

[Iowa_Flames_Fan]

On another note, it is 12 days before the election, and over 16 million votes have already been cast, which is likely to be a little more than 10% of the total.
http://www.electproject.org/early_2016

Importantly, it's not evenly distributed, and there are some key states where the early vote is closing in on 25% or even higher:

(numbers are approximate because I'm lazy)

Arizona: 882,000 votes cast, a bit less than 40% of the 2012 total;
Colorado:575,000 votes cast, or about 20% of the 2012 total;
Florida: 2.5 million votes cast, or a little more than 30% of the 2012 total;
Georgia: about 1 million votes are in, or a bit more than 25% of the 2012 total;
Iowa: almost 350,000 votes in, or about 20% of the 2012 total;
Nevada: 280,000, or 28% of the 2012 vote; and
North Carolina: 880,000, or almost 20% of the 2012 total.

There are some exceptions, but it does appear to me that the "early vote" is disproportionately concentrated in the most competitive swing states. One notable exception s Pennsylvania, which doesn't have early voting.

One thing to keep in minds as polls "move" closer to Election Day is that the cake is increasingly more "baked" as that day gets closer. I'm not one who thinks we can learn much about who is "winning" based on early voting numbers--but I do think it's safe to conclude that as Election Day gets closer it is harder to move the needle when a significant part of the vote is already set in stone.

[Acey]

Originally Posted by photon

The snake will keep me safe, the snake will protect me.

The snake is my favourite graphic. All the others can go home.

[Senator Clay Davis]

So a week ago the ABC News tracking poll started and had Hillary up 12, Trump supporters dismissed it and said trust the IBD/TIPP poll because it is the most accurate pollster from 2012 (it also had Trump up 2). A week later the ABC tracking poll now has Hillary up only 2.....but the IBD/TIPP poll has her up 4. So maybe we are heading for a big polling error. It just might not be easy to tell which way though.

[Iowa_Flames_Fan]

Originally Posted by Senator Clay Davis

So a week ago the ABC News tracking poll started and had Hillary up 12, Trump supporters dismissed it and said trust the IBD/TIPP poll because it is the most accurate pollster from 2012 (it also had Trump up 2). A week later the ABC tracking poll now has Hillary up only 2.....but the IBD/TIPP poll has her up 4. So maybe we are heading for a big polling error. It just might not be easy to tell which way though.

It is odd, and this cycle there has been way more volatility than usual I think. I kind of chalk that up to two candidates that (fairly or not) nobody really likes, but it could make for an interesting finish.

As for ABC and IBD, one explanation for what you are seeing could be that both polls are sort of "reverting to the mean", and the race was always at that mid-single-digits level. But who knows?

[Iowa_Flames_Fan]

On the early voting front, more than 20.5 Americans have now already voted.

[photon]

Could also be outliers for both polls. 538 also mentioned herding as a late cycle phenomenon where polls that are viewed as outliers are sometimes not even released to the public, artificially selecting polls that narrow the race.

The email news could shave 1 or 2 % off of Clinton's lead. Trump's been going up as all the news falls away and people revert to their partisan defaults. Johnson's been going down but will he go much lower than the 5% he's at now. How much of a difference will get-out-the-vote efforts make (of which Trump is far behind), could that basically offset any loss due to the email news? Trump's running out of money could also impact his end game. Is there one more big negative story out there against Trump?

Ugh.

[octothorp]

I can't find the original post but someone previously posted this on IBD, which explains the factors that cause it to always be right of others, and then move left at the last minute. It could be that they're still using essentially the same approach, but 'smoothing' it so that it it happens over the last couple weeks rather than the last couple days.

http://www.pollster.com/blogs/hittin..._a_bullet.html

[Bunk]

Originally Posted by Iowa_Flames_Fan

On the early voting front, more than 20.5 Americans have now already voted.

That's not very many people.