[RD] US 2016 election: Poll watching thread

[Sommerswerd]

Am I misremembering, or did you say that if polls showed the race tightening the actual result would probably be more lopsided as the appearance of a tight race would drive turnout (which generally speaking helps Hillary)?

I don't remember whether I posted that or not, but it certainly sounds like something I would say... and therefore, sounds correct.

However, I do remember saying that it could actually work the opposite way... If the polls are showing a Hillary blowout, it could serve to discourage Trump voters, by causing them to stay home in despair... my theory was that blowout predictions depress turnout overall, but probably slightly more among the side that is expected to lose, because its fun to go out and triumphantly vote for the winner (ie., bandwagon effect), whereas standing in line for an hour to vote for a lost cause is sadtimes.

Also, increased turnout overall might help Hillary, but increased Republican turnout certainly helps the Republican Senate candidates... another thing to keep in mind.

 

[metalhead]

This should read 'non-college grads' right?

Yeah, edit is forthcoming.

The "fun" one... As in the model that reinforces the ratings-raising "horse-race" narrative.

I will repeat... its still early for "likely voter" polls anyway, because there is too much tendency to be squishy with how you define "likely voters" They should still be using registered voters as the sample at this point.

That's the thing though, the demographic split I reference was within the Likely Voter pool. So they had already passed whatever screen CNN was applying to identify LVs, but then got lost in whatever model CNN used to get their final result, while Trump's college grad voters made it through unscathed.

 

[Lexicus]

I don't remember whether I posted that or not, but it certainly sounds like something I would say... and therefore, sounds correct.

However, I do remember saying that it could actually work the opposite way... If the polls are showing a Hillary blowout, it could serve to discourage Trump voters, by causing them to stay home in despair... my theory was that blowout predictions depress turnout overall, but probably slightly more among the side that is expected to lose, because its fun to go out and triumphantly vote for the winner (ie., bandwagon effect), whereas standing in line for an hour to vote for a lost cause is sadtimes.

Also, increased turnout overall might help Hillary, but increased Republican turnout certainly helps the Republican Senate candidates... another thing to keep in mind.

I think you said it, or something like it. I do definitely remember you saying the stuff about the blowout predictions depressing turnout.

I don't know, I guess I'm looking for reassurance because I keep seeing all this dire prognosticating (like eg from Robert Reich) that Trump winning is a serious possibility, and it would be nice if I could think to myself heh, this is just artifice to make sure that Hillary supporters don't get complacent.

 

[Sommerswerd]

That's the thing though, the demographic split I reference was within the Likely Voter pool. So they had already passed whatever screen CNN was applying to identify LVs, but then got lost in whatever model CNN used to get their final result, while Trump's college grad voters made it through unscathed.

That's what happens in a funhouse... people get lost.

it would be nice if I could think to myself heh, this is just artifice to make sure that Hillary supporters don't get complacent.

But if you could do that, it wouldn't be a proper artifice, would it?

 

[metalhead]

I think there are Democrats legitimately wetting their pants thinking Hillary can lose, but the only realistic scenario I see where she could is if she collapses at one of the debates. There's no way in hell WikiLeaks has any smoking gun up their sleeve, so I don't see that October surprise as having much cache. The email and foundation stuff is already baked into the cake.

Democrats want the polls projecting a Hillary blowout, because it will absolutely depress Trump turnout if his non-fanatical voters think he can't win.

 

[Sommerswerd]

I think there are Democrats legitimately wetting their pants thinking Hillary can lose, but the only realistic scenario I see where she could is if she collapses at one of the debates. There's no way in hell WikiLeaks has any smoking gun up their sleeve, so I don't see that October surprise as having much cache. The email and foundation stuff is already baked into the cake.

Democrats want the polls projecting a Hillary blowout, because it will absolutely depress Trump turnout if his non-fanatical voters think he can't win.

Point of fact, as it relates to the Presidential election, I don't particularly want depressed turnout. What I would most like is for Hillary to win a blowout victory with record turnout, because that way there can be no excuse that Hillary only won only due to most voters not participating. I am confident that Hillary would win whether there was 49% turnout or 79% turnout.

 

[Lexicus]

But if you could do that, it wouldn't be a proper artifice, would it?

Shh, you're messing with my self-image. I'm much too smart to be affected by artifice

 

[onejayhawk]

I think there are Democrats legitimately wetting their pants thinking Hillary can lose, but the only realistic scenario I see where she could is if she collapses at one of the debates. There's no way in hell WikiLeaks has any smoking gun up their sleeve, so I don't see that October surprise as having much cache. The email and foundation stuff is already baked into the cake.

Democrats want the polls projecting a Hillary blowout, because it will absolutely depress Trump turnout if his non-fanatical voters think he can't win.

I agree that Democrats fear the debate. While many things will be said about policy and priorities, this is all about appearances. If Hillary looks sick and tired and if Trump looks Presidential, what is said may mean little. I do not expect that to happen.

The other side is that Hillary has a reputation as a great debater. Trump has a reputation for presentation. One will probably suffer, but it is not clear which. I would call Clinton more at risk because this is her venue, so to speak. Trump clearly feared debating Cruz but that may not be relevant.

As an aside, I would love to see a Cruz vs Sanders debate. That would be epic.

J

 

[Zkribbler]

I predict Hillary will win both the debates and the Presidency.
The Democrats will take control of the Senate.
Hillary will fill the vacancy on the Supreme Court by appointing Obama.
Republican heads will explode.

 

[Timsup2nothin]

I predict Hillary will win both the debates and the Presidency.
The Democrats will take control of the Senate.
Hillary will fill the vacancy on the Supreme Court by appointing Obama.
Republican heads will explode.

Their heads will probably just inflate...it's when the next opening comes up and she puts in Elizabeth Warren that they will explode.

 

[onejayhawk]

It's a significant day in poll watching. The 538 forecast on control of the Senate flipped colors. Up to 25 September, the Democrats were favored to win control (pickup at least 5 seats). That is no longer true. Democrats are still favored to make significant gains, including a possible tie.

This coincides with the Presidential debate. On the Presidential side, polls seem to be holding their breath, awaiting the outcome of tonight's debate. For example, the highly regarded Quinnipiac poll favors Clinton by 1%. This is largely true. While Clinton's lead is paper thin, it's still there. Nate Silver calls it a the equivalent of a one score lead in American football. Whether that means +1 or +7 is unclear.

Largely unnoticed in all the key state polling, Trump is unusually weak in Republican core states, particularly Texas' 38 EV. Some wonder why Clinton can be leading nationally when key states like Ohio are leaning Trump. Essentially, her national total is getting a boost from deep red states that are merely hot pink.

J

 

[AdrienIer]

Yes Clinton is weaker than expected in Ohio, Iowa Maine and Nevada, but she's also stronger in Florida (slightly) Virginia North Carolina, Georgia Arizona and severall deep red states. I suspect the main factor is that Trump is doing terribly with non-whites.

 

[metalhead]

Not only is he doing terribly with non-whites, he is doing absolutely putridly with white people with college degrees in blue and purple states, when compared to how Romney did. That's why Hillary is nearly bulletproof in VA, CO, and PA; and more vulnerable in Iowa and Ohio. College whites in culturally conservative states are still heavily into the Republican nominee, which makes Texas and Georgia well out of reach despite relatively close polling. Texas could flip, but it would require about as monumental an electorate shift there as would be requred for Trump to win the election. I.e., it's possible but extraordinarily unlikely.

 

[onejayhawk]

Yes Clinton is weaker than expected in Ohio, Iowa Maine and Nevada, but she's also stronger in Florida (slightly) Virginia North Carolina, Georgia Arizona and severall deep red states. I suspect the main factor is that Trump is doing terribly with non-whites.

According to MH's toy, Trump is doing better with non-whites than Romney did. Specifically, he is doing better with both Hispanics and blacks. I have not seen a breakdown on Asian or other minorities.

J

 

[Thedrin]

According to MH's toy, Trump is doing better with non-whites than Romney did. Specifically, he is doing better with both Hispanics and blacks. I have not seen a breakdown on Asian or other minorities.

J

The Bloomberg poll decoder? If I'm reading it right it has their relative scores as:

Demographic| Obama Clinton Change PopProportion VoteChange
-----------| ---------------------------------------------
Blacks     | 93.5  90      -3.5   12            -0.42
Hispanics  | 72    71.8    -0.2   12            -0.024
Whites     | 40    43.05   3.05   70            2.135

Those population proportions are from the 2010 census though so, as per my understanding of modern immigration to the US and voting rights accorded to all immigrants, that 0.2% percentage point swing is absolutely devastating to Hillary's chances.

These are the 3 largest groups in the US with Asians making up just 3.6%. A swing in their support would have to be enormous to be particularly significant.

Edit: Updated table to use Metalhead's percentage's below.

 

[onejayhawk]

That will do. Much as Trump is supposed to be anathema, the results are slightly better than history.

Where on the Bloomberg do you find percentages? What I see are spreads.

J

 

[AdrienIer]

Uhm, this board seems to say that Clinton will get almost 1.5% more than Obama, who won by 4%... Also it doesn't matter if Clinton does worse than Obama as long as Trump does worse than Romney. Which is likely given the high numbers going to Johnson (coming from both parties).

 

[Thedrin]

I interpret a +80 to Clinton to mean that she is polling 90% against Trumps 10% with a difference of 80%, but I didn't take a closer look at the site to see if that's a correct interpretation.

Workings:

q){$[x>0;x;0]+%[;2]100-abs x}each 87 80 44 43.6 -20 -13.9
93.5 90 72 71.8 40 43.05

 

[metalhead]

That will do. Much as Trump is supposed to be anathema, the results are slightly better than history.

Where on the Bloomberg do you find percentages? What I see are spreads.

J

Not quite. Both the black and Hispanic votes are shaping up very similar to 2012, and if you recall 2012 was a historic wipeout among Hispanic/Latinx voters. Hillary stands to do better with whites than Obama, perhaps significantly so.

@Thedrin - the actual electorate projects to be about 70% white, 12% Black, 12% Hispanic, with the remaining 6% Asian, Native, and various other groups (who are likely to go 60%-ish for Hillary).

 

[AdrienIer]

Not quite. Both the black and Hispanic votes are shaping up very similar to 2012, and if you recall 2012 was a historic wipeout among Hispanic/Latinx voters. Hillary stands to do better with whites than Obama, perhaps significantly so.

@Thedrin - the actual electorate projects to be about 70% white, 12% Black, 12% Hispanic, with the remaining 6% Asian, Native, and various other groups (who are likely to go 60%-ish for Hillary).

So if we take Thedrin's numbers (Blacks 90%, Hispanics 71.8% and Whites 43.5%) and add the vote for the other ethnicities (let's make it 50% Clinton) that makes... 52.5% Clinton. Congratulations you probably just got a landslide.