I've had people say it's a hardening, actually ~ US presidential election 2016 part 9/11 never forget

https://twitter.com/adrian_gray/status/784211583421542401

Clinton winning 94% of Obama approvers. Would be new high.

McCain '08: 89%
Gore '00: 77%
Bush '88: 84%
HHH '68: 64%
Nixon '60: 68%

poor Hubert

They're also all pretty much dismissing the now-cast at this point. I guess they'll do a lot of rethinking after the election. I still like their podcasts, though.

This is good; I stopped reading their polling articles because they kept throwing "but the nowcast says this" analysis into every article. They'd always include their boilerplate explaining the limits of the nowcast, but it still felt like scare-tactic clickbait, not much better than random speculation about who has "momentum."

I don't mind that they're bullish on Trump compared to other predictors, that's fine and makes sense to me. It's the volatility that has caused me to take them less seriously. Clinton will have a few weaker polling days and suddenly her chances drop 10-15%. Then they're back up 20% in the 4 days after the debate.

obviously that debate was major, i'm not denying that, it just feels like 538's numbers have been all over the place to the point that I don't really trust the heights or depths of their bounces.

Nate has actually done the opposite of what you'd expect (well, if he was still at NYT and not at Disney) and interpreted these swings as this race being one of the more "volatile" races in recent memory, while meanwhile Sam at PEC has said the complete opposite, that this has been the most stable campaign in years (it should be noted, his Bayesian forecast even went through a significant reduction in Clinton's odds too, but nowhere near as stark or immediate as Nate's).

Sam has basically said, in very nice veiled words, that maybe Nate should look at his model.

This may have already been linked, but i thought this 538 critique was interesting, even as i admit that i'm not really qualified to evaluate it on its merits. http://predictwise.com/blog/2016/09/poll-aggregation-fight/

http://www.politico.com/story/2016/10/donald-trump-town-hall-debate-prep-229267

“I said forget debate prep. I mean, give me a break,” Trump said at one point. “Do you really think that Hillary Clinton is debate-prepping for three or four days. Hillary Clinton is resting, okay?”

Y'all talking shit like you don't remember the bad old days of the Carville/Huffington/Dowd/Tucker Carlson/etc/etc reality-free punditocracy. We live in a golden age of evidence-based election coverage thanks largely to NS. If you don't like his particular prediction, take an average of his, the daily kos one (actually very good), sam wang, and predictwise. The result will be more stable (good for your delicate constitutions) and is likely more accurate than any one of them alone.

i mean, maybe silver is right and it is uniquely volatile! I think i'm maybe also swayed by the fact that, whatever his talents as a data journalist, his website is just not very good by any measure.

thank god there's no more reality-free punditocracy

Like, as far as I know, his model and those of any other principled predictor are set in stone (code) from the beginning of their coverage of this election cycle. It's not like they go in by hand and tweak shit to make the results more click-baity (I hope!).

Hey, the punditocracy doesn't exist for those of us who Don't Watch TV™

it still shows up in my twitter feed from time to time tbh

The historical data piece of his model is based on basic antagonisms like incumbent party/opposition and republican/democrat that are less meaningful in a year where a weirdo is facing off against a lightning rod

fwiw i don't believe silver is changing his model in real time at all, and my critique wasn't meant to deny him credit for his past accomplishments. he may be right about this election's unique day-to-day volatility, though that seems hard to truly measure given that there will only be one outcome. I've just mentally started to file his forecast as more noise than signal, that's all.

go in by hand and tweak shit to make the results more click-baity

Oh I think that's part of what they have been doing, incrementally anyway

NS and 538 seem to be backsliding towards a more fact-fungible world of political reporting where every day is new exciting horse race fodder, while other peers (mostly hobbyists with other day jobs) are actually pressing forward with the sober analysis that made 538 viable in the first place. Additionally, his OG, more boring methods have been picked up by other wonky sites that generally avoid horse race reporting because they've staked claims to other foci and perspectives as their bread and butter, cf TPM

DailyKos and HuffPo didn't make their first projections until later in the election process than PEC/538 and I think that was to solidify the models.

538 was the only aggregator (major) that ever dipped below 60% for Hillary, and his was also the slowest to recover when polls went back towards Hillary. Enten even wrote as far as to write an article suggesting the evidence for a Hillary rebound was weak (when, at the time, there was barely enough data to say one way or another, making the article fairly pointless since they had to acknowledge it was there the next week). Sam Wang is much too diplomatic to call anybody out (in fact, all of the aggregators are fairly respectful of each other), but I was reading his comments to say that perhaps 538's model design is at fault for the fluctuation, and he strongly believes this has been the most steady campaign (odds-wise) in recent memory. He has consistently given Hillary the best odds.

who cares tho at this point Hillary is obv leading mostly comfortably and 538 acknowledged Johnson and Stein's numbers are dipping as are undecided voters so we might avoid that scruffy idiot yet

Anyway my prediction is Hillary 537 Johnson 1 Trump 0.

that is until Julian Assange gets Hillary arrested.

I might be wrong but I'd bet that, if not now, Clinton has done sessions of at least four hours with breaks with multiple debate/question possibilities weighed. It's not like weird-ass debate club with Cruz where that guy would love to do L-D mock debates all day, but classical debate stuff is definitely in her wheelhouse

The idea that Trump can't even contemplate people being into debate prep or mock debates is, if anything, evidence he can't understand how anyone else lives or works even if he thinks it's not necessary

Silver doesn't change his model at all once they start - which might be the reason why it's off.

Still looking at the Iowa electronic markets a bit

Trump's totally the type of dude that said "you can't study for the SATs, man. you either know it or you don't!"

I guess he thinks maybe Hillary baited those traps for him just by thinking quickly on her feet? that she's part of some awesome Democrat Improv Troupe?

boy, he might wanna skip the second debate after all.

i can't quite decide if i think the town hall will be better for him or worse. Even after the drubbing he got last time i'll probably be sitting on my couch stressed as fuck on sunday.

I suspect whether he wants to or not, he's going to be coached on what he did wrong last time. but he's so hair trigger that I can't see him doing that much better. but with town halls, you sometimes get loaded/leading questions from the civilians asking questions which don't always help.

I kinda feel like after his first meltdown, the strategy of lowered expectations (ie the "St Anger" method) worked and he was deemed Presidential after his awkward visit to Mexico and due to basically not fucking up too badly for a while, but now that he's melted down again, I think those folks who may have come home briefly are pulling back again because they're seeing this wasn't an aberration.

Kinda think Hillary just needs to stay the course - ball control offense, maybe a little more aggressive to run up the score and show the GOP this strategy won't work in 2020.

Hillary screwed him by saying he is a bad speaking man and screwing with his microphone and the moderator was biased.

i bet he brings his own mic to this one

They (Hillary of course) should give him a faulty stool this time.

He has a diaper to mitigate his faulty stool

he sits on the stool and realizes in horror it's a unicycle

he'll definitely make a reference to how it's nice they gave him a working mic this time

"do I get to win this time?"

everything sopan deb posted tonight on trump's NH rally material was amazing btw

https://pbs.twimg.com/media/CuIPPILVIAEkMqg.jpg

i mean how can anyone give an answer this long to anything and still not do the obvious thing of just spinning it around into an excuse to say great things about mike pence and the great vision they have for america and how great his kids are or fucking anything normal

also: "We had a guy inside the room, oscillating my mic."

"I used to say I think someday he'll be an anchor. Guess what, he's still doing the maps!"

That's actually pretty funny. I mean, I bet John King can't believe he's still doing the maps.

Nate Silver has admitted their model is volatile several times. He's also explained why he thinks that's been the right way to describe this election. Iirc it's because 1) There's been a shitton of undecideds/thirdpartyvoters compared to last election, meaning much more room for change 2) There's been so many explosive news stories, from Trump, but also scandals about Clinton, making voters more uncertain and 3) There's simply been more change in the polls. Which is of course the most important. It's also true. Hillary's lead almost disappeared in the polls until the debate, what kind of poll aggregator wouldn't say her chance was shrinking under that circumstance? Now there's been a big swing over this last week, and she's gained almost 30% in the forecasts. Which does seem kinda insane to me.

Their model has definitely behaved very differently than in 2012, and it might seem as if 2012 was simply an election extremely well suited to the data approach. But of all the things that Camaraderie mentioned as should perhaps have been included in the forecast: early voting, better ground game by Democrats, expected performances in the next debates, higher voter registration among minorities, high numbers of late deciders, etc. one of them is included - the late deciders, as in undecideds - and most of the others are guesswork. There isn't even enough demographic data amongst voter registrations to fully include in a model. Economic data, otoh, is objective data that can be put into a model and updated as new data comes in.

All of the aggregators reacted to the polls but 538 had the strongest reaction. Nate tells you not to overreact to individual polls and yet his model seemed to do that.

Trump's totally the type of dude that said "you can't study for the SATs, man. you either know it or you don't!"

this hurts my feelings a little

I'm sure someone would have posted this, but I won't wade through five days of posts to check:

https://www.youtube.com/watch?v=1chlgN3Stxk

Has Silver accounted for this hidden unknown in polls-plus?

Nate's model reacted to the overall consensus. As they pointed out, even for the polls that kept Clinton comfortable ahead, the trendlines were awful. I think they've definitely suffered from overfitting and old ideas, with the most extreme example being them relying on endorsements in the primaries, which turned out to have almost no predictive value, but a lot of the critique of the models is overblown. Except for the now-cast, the now-cast sucks. But for instance, the predictwise guy that intheblanks linked to, he wants them to both be more selective in the polls they include, but then critiques them for 'adjusting' the polls. Even says 'Why not just average the polls?' a paragraph after complaining that they average too many polls.

if you're worried about the Sunday town hall debate just watch what he did in NH yesterday

at the end of the day Trump can't put together remotely coherent answers to questions he knows are coming in advance, and while this format may be a little more friendly to him b/c he doesn't have to engage with Clinton as much, she'll still be around to explain why every answer he gives is idiotic

not to mention all that's happened since Debate One

I wonder if Trump U is gonna get brought up at all in these debates. It's kind of a slam dunk since there's practically nothing Trump can do to make himself look good there, but we all know Trump will spend several minutes defending himself against what is an obvious fraud (compare to Hillary's email answer: "Yep, I did it, I regret it, and I won't do it again" - 10 seconds, done)

Seriously https://twitter.com/justinwolfers/status/784380522214461440

It raises an obvious question: Why has the Freedom Caucus, which built its reputation on promoting conservative principles through internecine conflict, quietly fallen in line behind Trump? According to interviews with a host of Freedom Caucus members — and a National Review analysis of demographic trends and voting patterns in their congressional districts — the explanation owes both to partisan self-interest and political self-preservation.

Read more at: http://www.nationalreview.com/article/438984/donald-trump-house-conservatives-conquest-explained

x-post: Wow... That second debate is going to be something as well.

xxp that... is fantastic

that was one of the more composed answers i've seen trump give recently and it was still total rambling bullshit from start to finish

Beggars belief tbh.

The other thing to remember is the town hall is 90 minutes. The closest he's got to that in real world similar environments is like 30 minutes. He is going to need at least a couple of extra bumps.

Guessing Apple will make their iPods in the US around the same time Donald's Make America Great caps are made in the US

is Hillary allowed to respond to his answers? I can't remember how these town hall debates go.

i feel like i should be entirely inoculated against trump word salad by now but i still find myself trying to follow the logic of his responses and quickly getting a headache and a deep, unsettling sense of existential terror