The US Election Polling / Data Thread

[Iowa_Flames_Fan]

Originally Posted by Senator Clay Davis

One day later, another poll has Trump up 2 in Florida....

http://www.realclearpolitics.com/doc...ida_August.pdf

I saw that this morning. Frankly, a single digit lead for Clinton in Florida is more plausible than 14, but I continue to think that demographics are a major challenge for Trump, given the high number of Latino voters.

[Senator Clay Davis]

That's what you'd think, but then you remember Florida is the most ####ed in the head state in America, and you realize Hillary winning there isn't a given or anything close. In fact I can totally see a scenario where she wins every swing state and steals Georgia and Arizona, but loses Florida because it's Florida.

[Parallex]

Originally Posted by Senator Clay Davis

One day later, another poll has Trump up 2 in Florida....

http://www.realclearpolitics.com/doc...ida_August.pdf

Yeah, that's your outlier the other way. Wish that firm put out polls a bit more frequently, their prior poll was in January so it's less valuable for trendline watching.

[Itse]

Originally Posted by Senator Clay Davis

The next few weeks of polling should give the clearest sign of whether we are headed to a slaughter or whether this will be close. Trump is now finally taking some risk of alienating his supporters by shifting/softening/lying about his positions towards illegals, and I guess we can call it "wooing", but his wooing of African American voters has also happened. So if his numbers are still in the toilet in two weeks, his attempt didn't work and he's basically stuck with his base and not much else. And his hope then becomes hoping for a Hillary meltdown or Wikileaks/Russia miracle.

While he obviously has to try do something, I suspect it's more likely he'll end up alienating more than he stands to gain. His whole thing is being "not like those other politicians". If he loses that, he's got nothing.

[Senator Clay Davis]

Originally Posted by Itse

While he obviously has to try do something, I suspect it's more likely he'll end up alienating more than he stands to gain. His whole thing is being "not like those other politicians". If he loses that, he's got nothing.

At the same time where does his core have to go to? When the Bernie or Busters busted, they at least had Jill Stein from a progressive perspective, and (sadly) Trump from an anti-establishment perspective. Trump supporters aren't going to Stein or Johnson, and we know Hillary is out. So what do they have after that? Staying home I guess, but I suspect Trump has correctly calculated he won't lose much core support so long as he remains opaque about his intentions. They can try and delude themselves that he's wink-winking to them and that he's lying to minorities.

[CorsiHockeyLeague]

They could go to Johnson couldn't they? I don't see why not. Though I think the answer is almost entirely "they'll stay home".

[ernie]

But with what is likely a weak ground game can Trump afford apathy in his current supporters? And apathy is what could set in if he tries to moderate.

Though I'm not convinced he'll moderate at all. His new handlers clearly want him to try but the guy is a loose cannon. The muzzle will fall off at some point and/or he'll be forced to actually articulate something of substance in front of the broader audience and won't be able to.

[Iowa_Flames_Fan]

A huge data dump of state-level polling from Ipsos today, and it's a bit of a mixed bag. Some good results for Clinton (+7 in both Florida and Ohio) and other results that are pretty good for Trump (down only 1 in Michigan, Wisconsin, and Pennsylvania).

These are a bit odd in that the field time is really long (all of them were in the field either July 29-August 19 or August 12-19), and the sample size is pretty small for most (~500). I'm more concerned about the long field time, as it can mask intervening events/variables, and means that the poll includes responses that are now over three weeks old.

Nevertheless, they are all listed on 538's forecast site if anyone is interested. Net effect seems to have been a very slight uptick for Trump in the "polls only" forecast.

[Parallex]

Originally Posted by Iowa_Flames_Fan

A huge data dump of state-level polling from Ipsos today, and it's a bit of a mixed bag. Some good results for Clinton (+7 in both Florida and Ohio) and other results that are pretty good for Trump (down only 1 in Michigan, Wisconsin, and Pennsylvania).

These are a bit odd in that the field time is really long (all of them were in the field either July 29-August 19 or August 12-19), and the sample size is pretty small for most (~500). I'm more concerned about the long field time, as it can mask intervening events/variables, and means that the poll includes responses that are now over three weeks old.

Nevertheless, they are all listed on 538's forecast site if anyone is interested. Net effect seems to have been a very slight uptick for Trump in the "polls only" forecast.

Yeah it was really just a big sample national poll that they parsed out the state data from.

It's interesting to see the state level data from states that aren't frequently polled... but there are some problems with many of them. I mean less then 200 people were polled in New Hampshire. Some real headscratchers... I will eat my hat if Clinton is actually up 6 in Missouri.

[Iowa_Flames_Fan]

Yeah that doesn't seem like a particularly sound methodology. Presumably that also means no state-by-state demographic adjustments to the data? I don't know where to find that out.

Meanwhile, a new Florida Poll from a GOP pollster shows Trump ahead 44-41:
http://www.flchamber.com/politicians...tatewide-poll/
And national poll numbers:

Rasmussen has Clinton ahead 42-38:
http://m.rasmussenreports.com/public...te_house_watch

USC Dornsife now has the race tied:
http://graphics.latimes.com/usc-pres...oll-dashboard/

And some pollster I've never heard of has Clinton way ahead:
http://www.prri.org/research/lgbt-20...ion/?test=true

[Parallex]

Originally Posted by Iowa_Flames_Fan

And some pollster I've never heard of has Clinton way ahead:
http://www.prri.org/research/lgbt-20...ion/?test=true

... we're a month past the conventions. Why are these people still polling registered voters? Should be screening for likely voters at this point.

[Iowa_Flames_Fan]

Originally Posted by Parallex

... we're a month past the conventions. Why are these people still polling registered voters? Should be screening for likely voters at this point.

It does make the apples-to-apples comparison a lot more difficult. My preference is for pollsters to report both numbers, and to be transparent about the "secret sauce" in their likely voter screen, but that's another story.

[Iowa_Flames_Fan]

Some new national numbers today, and you can basically take your pick according to the result you prefer.

CVOTER has Trump leading Clinton by less than a percentage point:
http://www.upi.com/Top_News/US/2016/...?spt=sec&or=tn

Quinnipiac has Clinton ahead by double digits, 51-41:
https://www.qu.edu/images/polling/us...16_U88mxwn.pdf

Meanwhile, Suffolk has Clinton ahead by 7 in Michigan:
http://www.suffolk.edu/documents/SUP..._marginals.pdf

[Flash Walken]

Could it be possible for Trump to win both Ohio and Florida and lose the electoral college?

He would be the first Republican nominee to win Ohio and Florida and not become president in that circumstance I believe.

[Bunk]

Originally Posted by Flash Walken

Could it be possible for Trump to win both Ohio and Florida and lose the electoral college?

He would be the first Republican nominee to win Ohio and Florida and not become president in that circumstance I believe.

Yeah, lose Virginia, NC, Iowa, Pennsylvania, Michigan, Nevada, New Hampshire (all pretty likely) and yes he still loses by a wide margin (300-237).

[CorsiHockeyLeague]

Originally Posted by Flash Walken

Could it be possible for Trump to win both Ohio and Florida and lose the electoral college?

He would be the first Republican nominee to win Ohio and Florida and not become president in that circumstance I believe.

Yep. And I think this result is actually really generous to Trump even leaving aside OH and FLA - I gave him NC, IA and NV and he still loses.

[GGG]

Originally Posted by Iowa_Flames_Fan

It does make the apples-to-apples comparison a lot more difficult. My preference is for pollsters to report both numbers, and to be transparent about the "secret sauce" in their likely voter screen, but that's another story.

I've been really enjoying Silvers discussion around avoiding poll herding so I'm not sure releasing the secret sauce would be a good thing. If all polls used the same methodology then there would be no gain in using averages as all error should just be the statistical sample errors and any bias in the polls would be undetected.

Instead by having a bunch of polls who use "reasonable methods" to poll with you get a more accurate picture when you roll up the average.

[Senator Clay Davis]

^^How much of a nightmare would losing with that map be for the GOP? They'd obviously convince themselves there's no need to shift or moderate, just have a slightly cleaner candidate than Trump and they win (ignoring Hillary isn't running every election). They need to get decimated to change and improve their chances of winning in the future.

[GGG]

NH plus Maines 2nd district becomes the tipping point state in that scenario.

The problem is that map plus the above is Trumps only real path to victory.

[troutman]

The problem for Trump is the only state he is winning that has a large number of electoral votes is Texas (not sure how recent this picture is):