• Welcome to OGBoards 10.0, keep in mind that we will be making LOTS of changes to smooth out the experience here and make it as close as possible functionally to the old software, but feel free to drop suggestions or requests in the Tech Support subforum!

SayHeyDeac's Thread For Serious Political Discourse Only--Trolls Need Not Apply

Clinton campaign is still confident in PA (Philly suburbs), CO (Denver suburbs, Latinos), VA (NOVA), and NH because their numbers are much better than expected among college grads. More wary about NC now because of the shooting, but like their chances because of CLT/RDU suburbs. Think they started their daughter ads too early and telegraphed a debate line of attack. Conway & Pence freaked out and overcompensated about Flowers. They still have zero control over Trump on tweeting and executing strategy. Trump doesn't need to win suburbs and cities, but he can't get crushed in either since can't make it up solely in rural areas.

Conway had to have explained to Trump why he was going to Mexico, visiting Black churches, and trying to calm down wary suburbanites, but he still fucks up. Buffett, Bezos, and Bloomberg are all way richer than Trump and Cuban combined, but HRC campaign knew Cuban would provoke a bigger Trump meltdown than the others.
 
FuckmouthedRube said:
Deplorables certainly didn't help, but do think Hillary's collapse and not disclosing her illness hurt her. Plouffe certainly could be bullshitting, but find it hard to believe many people in America know swing state polling and GOTV better than he does.

Question for birdman: 538's model looks like it's stabilized, but for weeks the EV probability distribution peaks were between 340 & 350, but the model was claiming it expected ~ 320 EVs. Much cleaner now since the peaks correlate almost exactly with the expected EVs, but was weird that model expected EVs and the probability distribution were out of whack for several weeks. Thoughts?
Click to expand...

Hmmm, I didn't follow that progression. When you say peaks, do you mean the high end of a 95% confidence interval, or do you mean the average value?
 
RJKarl said:
All indications are the Latino vote is going to be big. Another key is people with college degrees. Here's a study from 2009 that states :Well-educated citizens vote more frequently than the poorly educated in some countries, including the USA. "

http://www.sciencedirect.com/science/article/pii/S0261379409001176


This also shows that some college and post grads vote more than non-college people http://www.electproject.org/home/voter-turnout/demographics

If this is true, then the dramatic swing from Romney to Hillary would have a great impact.
Click to expand...

It might be the flailing insecurity that draws you in, but it's the clumsy false bravado that keeps you hanging around to the exciting conclusion. The RJ Karl Experience.
 
RJKarl said:
All indications are the Latino vote is going to be big. Another key is people with college degrees. Here's a study from 2009 that states :Well-educated citizens vote more frequently than the poorly educated in some countries, including the USA. "

http://www.sciencedirect.com/science/article/pii/S0261379409001176


This also shows that some college and post grads vote more than non-college people http://www.electproject.org/home/voter-turnout/demographics

If this is true, then the dramatic swing from Romney to Hillary would have a great impact.
Click to expand...

You missed the point of my post RJ by focusing too much of the example. Switch the word Latino/Hispanic in my post with any demographic you want. The point is that a candidate can get a larger percentage of any given Demo than a candidate did in 2012, but still get few actual votes if fewer people vote. The reverse is also true...a candidate can get a smaller percentage of an given Demo but still get more votes if lots of a specific demo is motivated to vote. Comparing poll numbers to previous elections is highly influenced by the "likely voter" component.
 
birdman said:
Hmmm, I didn't follow that progression. When you say peaks, do you mean the high end of a 95% confidence interval, or do you mean the average value?
Click to expand...

Just talking about the raw distribution of outcomes when they run the model. Current distribution has the two highest peaks (most probable outcomes) between 270 and 280, but the third highest peak is around 360. For a long time, the tallest peaks were between 340 & 360, even though the published expected outcome was around 320 EVs.
 
I fully understand that, but there are more people in each of the Clinton leaning demographics than in 2012. Thus, unless there is a dramatic attack of apathy, the larger group from which to draw can easily overcome the lack of interest.

If you look at the 15% swing in college educated voters, even if this number goes down from 2012, it's a huge swing.

http://www.pewresearch.org/fact-tan...rate-will-be-the-most-diverse-in-u-s-history/

As you can see, there are 18-20% more eligible Hispanic voters than in 2012. Even at a 10% lower turnout rate (which is highly unlikely), this would be a larger number of raw votes.
 
FuckmouthedRube said:
Just talking about the raw distribution of outcomes when they run the model. Current distribution has the two highest peaks (most probable outcomes) between 270 and 280, but the third highest peak is around 360. For a long time, the tallest peaks were between 340 & 360, even though the published expected outcome was around 320 EVs.
Click to expand...

I see. So that is probably a function of the electoral college weirdness. With states like FL and OH having a lot of EC votes and being near an even probability of either candidate winning the simulation out puts (e.g.the sum of EC votes for each candidate) would jump around a lot. In that case you can end up with a bimodal or Trimodal distribution of out comes. When ever you have odd distributions of data like that the average is going to end up a pretty meaningless summary statistic. I don't know what their "published expected outcome" is (e.g, average EC votes for each, median) but using the average would be misleading, median, or mode with percentiles would be a better metric.
 
RJKarl said:
I fully understand that, but there are more people in each of the Clinton leaning demographics than in 2012. Thus, unless there is a dramatic attack of apathy, the larger group from which to draw can easily overcome the lack of interest.

If you look at the 15% swing in college educated voters, even if this number goes down from 2012, it's a huge swing.

http://www.pewresearch.org/fact-tan...rate-will-be-the-most-diverse-in-u-s-history/

As you can see, there are 18-20% more eligible Hispanic voters than in 2012. Even at a 10% lower turnout rate (which is highly unlikely), this would be a larger number of raw votes.
Click to expand...

Ok then...the math in all the polls must be wrong.
 
birdman said:
Ok then...the math in all the polls must be wrong.
Click to expand...

As I asked earlier, which set is wrong? The most critical one could be whites with a college education. That would be the most dramatic as the base of that demographic is growing and it appears to the largest change in any group.

Could they be under-accounted for in the national numbers? Or is the swing grossly over-estimated?

I'm not sure which numbers are wrong, but some of them have to be.
 
RJKarl said:
As I asked earlier, which set is wrong? The most critical one could be whites with a college education. That would be the most dramatic as the base of that demographic is growing and it appears to the largest change in any group.

Could they be under-accounted for in the national numbers? Or is the swing grossly over-estimated?

I'm not sure which numbers are wrong, but some of them have to be.
Click to expand...

It's pretty simple, RJ. You can talk about how Hillary is tracking Obama's percentages within groups, and you may be correct, but it won't matter, because you are ignoring the turnout within groups and the enthusiasm gap. Trump is turning out hordes of the poor white voters you look down on, and Hillary isn't turning out enough of the black voters you look down on.
 
jhmd2000 said:
It's pretty simple, RJ. You can talk about how Hillary is tracking Obama's percentages within groups, and you may be correct, but it won't matter, because you are ignoring the turnout within groups and the enthusiasm gap. Trump is turning out hordes of the poor white voters you look down on, and Hillary isn't turning out enough of the black voters you look down on.
Click to expand...

wow
 
jhmd2000 said:
It's pretty simple, RJ. You can talk about how Hillary is tracking Obama's percentages within groups, and you may be correct, but it won't matter, because you are ignoring the turnout within groups and the enthusiasm gap. Trump is turning out hordes of the poor white voters you look down on, and Hillary isn't turning out enough of the black voters you look down on.
Click to expand...

This was my attempted explanation of the weird numbers, but without the hostility. You're are free to disagree RJ, but I am not going to make an effort to find some other explanation for you.
 
bird, that's why jhmd hasn't existed to me for years.

It's weird how there is a disconnect somewhere. It appears to be fairly large.
 
marquee moon said:
wow
Click to expand...

I was trying to be clear, but perhaps I was overly broad. RJ looks down on just about everybody, so that's not much of a qualifier. I should have added "but don't chide for not voting like they're supposed to." That was what I meant to say. Sorry for the confusion.

My apologies to my father's brother Tom Thomas.
 
Last edited:
FuckmouthedRube said:
Just talking about the raw distribution of outcomes when they run the model. Current distribution has the two highest peaks (most probable outcomes) between 270 and 280, but the third highest peak is around 360. For a long time, the tallest peaks were between 340 & 360, even though the published expected outcome was around 320 EVs.
Click to expand...

I've had a chance to look at the distributions now and they are really really flat for both candidates (http://projects.fivethirtyeight.com/2016-election-forecast/#electoral-vote). Clinton has basically an equal probability of getting any where from 220 to 350. Trump is a little more normally shaped but still has almost an equal probability of getting anywhere between 180 and 310. The x-axis shows that the "peaks" are taking you from a 0.05 probability to a 0.07 probability, that means that if they ran the simulation 1000 times, 50 replicates had clinton with 252 EC votes and about 70 or 80 had her with 272; it is really hard for me to see a lot of difference in terms of probability. All this really tells us is that they each have a pretty solid floor; the big blue wall and the big red wall are alive and well and these two candidates/parties and fighting over a very small percentage of the popular and EC votes.
 
birdman said:
I see. So that is probably a function of the electoral college weirdness. With states like FL and OH having a lot of EC votes and being near an even probability of either candidate winning the simulation out puts (e.g.the sum of EC votes for each candidate) would jump around a lot. In that case you can end up with a bimodal or Trimodal distribution of out comes. When ever you have odd distributions of data like that the average is going to end up a pretty meaningless summary statistic. I don't know what their "published expected outcome" is (e.g, average EC votes for each, median) but using the average would be misleading, median, or mode with percentiles would be a better metric.
Click to expand...

Thanks, makes sense.

Plouffe did say an 8 point lead was unnatural. Find it wild that there's a reasonable chance (~33%) that a 1.5 point race nationally could still yield 350 EVs.

Paradoxically, more states in play isn't all good news for Trump. Clinton's campaign manager Robby Mook is a turnout specialist; Conway was a pollster for Todd Akin in 2012. Clinton campaign won't pull out of Iowa or Ohio, but they're comfortable going ala Carte in FL, PA, CO, NV, and NH with NC and OH as back ups. Obama won the former group twice and they like the Latino and Mormon blocs in CO & NV. Trump knows he's still hurting among Mormons, that's why his latest judicial list includes Mike Lee.

Trump would be better off with a straight final four of FL, PA, OH, and NC. IA gives him some slack, but major city suburbs aren't friendly to him. It's why KAC shit over Flowers. They already have non-cucks hyperventilating about the hole in HRC's tongue and fantasing about Taylor Swift becoming a Nazi Princess; they need moms and dads who drive their tween daughters to Swift shows.
 
FuckmouthedRube said:
Thanks, makes sense.

Plouffe did say an 8 point lead was unnatural. Find it wild that there's a reasonable chance (~33%) that a 1.5 point race nationally could still yield 350 EVs.

Paradoxically, more states in play isn't all good news for Trump. Clinton's campaign manager Robby Mook is a turnout specialist; Conway was a pollster for Todd Akin in 2012. Clinton campaign won't pull out of Iowa or Ohio, but they're comfortable going ala Carte in FL, PA, CO, NV, and NH with NC and OH as back ups. Obama won the former group twice and they like the Latino and Mormon blocs in CO & NV. Trump knows he's still hurting among Mormons, that's why his latest judicial list includes Mike Lee.

Trump would be better off with a straight final four of FL, PA, OH, and NC. IA gives him some slack, but major city suburbs aren't friendly to him. It's why KAC shit over Flowers. They already have non-cucks hyperventilating about the hole in HRC's tongue and fantasizing about Taylor Swift becoming a Nazi Princess; they need moms and dads who drive their tween daughters to Swift shows.
Click to expand...

lol
 
For the record, did fuck up the probability of getting to 350. An earlier distribution had three huge spikes: 270, 280, and 350. Everything else was significantly shorter.
 
FMR and birdman are really showing why they're allowed to post on this thread. Nice work, guys.
 
Meanwhile, no one's talking about the VP debate, sponsored by Wonderbread and NyQuil.
 
You must log in or register to reply here.
Back
Top