|
djw175 posted:Honestly my biggest problem with Nate isn't his maps and stuff. Sure, it's conservative. Whatever. It's the think pieces or whatever about how Trump is totally gonna win you guys! that he seems to constantly loving put out or at least tweet. That's usually not what he says though, he says that if circumstances in the race shift, or if Trump is lucky and polling was error-ridden, he could win. His point is always that the race is not a certainty and is becoming tighter, never that Trump is favored to win. People then retweet it and mischaracterize it.
|
#
?
Nov 5, 2016 23:09
|
|
|
|
| # ? Jun 11, 2024 06:30 |
|
|
Cingulate posted:Correct - he talks about the model here: http://fivethirtyeight.com/features/a-users-guide-to-fivethirtyeights-2016-general-election-forecast/ cant cook creole bream posted:
Owen Ellickson has now included Nate in his twitter drama: https://twitter.com/onlxn/status/795019547724025856 quote:SILVER: DICK! You have NO idea how to account for the way polling errors tend to correlate between states!
|
|
#
?
Nov 5, 2016 23:09
|
|
|
RVProfootballer posted:I get the reasoning why it might be that high (and it's actually 12.6% of a popular/electoral split overall), I just don't believe it can actually be that high. Just look up the numbers instead of positing some vague, meaningless perception of reality. There have been 43 presidents. There have been 4 cases where the winner lost the popular vote. It's historically 10% of the time. http://www.factcheck.org/2008/03/presidents-winning-without-popular-vote/
|
|
#
?
Nov 5, 2016 23:09
|
|
|
RVProfootballer posted:I get the reasoning why it might be that high (and it's actually 12.6% of a popular/electoral split overall), I just don't believe it can actually be that high. Look at it this way, Trump has a really small chance to win the popular vote. But he manages to use every single way your country and your election system is broken, so he manages to raise his chances. Using flaws in the system and making them worse is sort of Trumps and the GOPs MO.
|
|
#
?
Nov 5, 2016 23:11
|
|
|
In any case, I do wish we could open up the 538 model and see the guts. Modeling correlations between states seems like an absolutely perfect thing to do, but without seeing exactly how they do it, it is tough to gauge whether there's any potential error there, ya know? Maybe demographics have changed too much across the data set they're using to estimate the correlations for those estimates to be useful. Maybe the weighting of a state's polling vs it's correlation with other states' polls can be tweaked differently. Maybe weighting of polls is too aggressive or permissive. I trust the 538 team is doing things in ways that make sense, but all of those possible sources of bias or error seem like they could interact in very non-obvious ways.
|
|
#
?
Nov 5, 2016 23:11
|
|
|
Clinton Campaign Manager Robby Mook explains her visits to MI: https://twitter.com/rubycramer/status/795025808934969352 I understand this but it makes me annoyed that PA and Michigan don't have early voting available. I feel a little better that she has more surrogates to GOTV in other critical states, but how can states so populous limit voting opportunities this way? So I guess at the end of the day, gently caress John Roberts.
|
|
#
?
Nov 5, 2016 23:14
|
|
|
CottonWolf posted:What he's regularly said is that it's not unlikely that Trump will win, but that it's more likely that Clinton will win, which seems completely fair. RVProfootballer posted:In any case, I do wish we could open up the 538 model and see the guts. Modeling correlations between states seems like an absolutely perfect thing to do, but without seeing exactly how they do it, it is tough to gauge whether there's any potential error there, ya know? Maybe demographics have changed too much across the data set they're using to estimate the correlations for those estimates to be useful. Maybe the weighting of a state's polling vs it's correlation with other states' polls can be tweaked differently. Maybe weighting of polls is too aggressive or permissive. I trust the 538 team is doing things in ways that make sense, but all of those possible sources of bias or error seem like they could interact in very non-obvious ways. There is zero reason for Nate not to do it. If anyone else were to somehow copy his model, people would still go to Nate cause he has better graphics. And he could even put it under some license that disallows it? djw175 posted:I may have been exaggerating a bit, but the amount of times he does it and how often he did it even when Clinton was +8, he's harping on it a lot for something even his models say isn't particularly likely. Not accusing you here, but some people cannot deal with uncertainty. I have a lot of understanding for this because, hey, humans simply are bad with uncertainty, but still, it's not that Nate is doing anything wrong, it's that non-nerds read his nerd stuff and deal with it in the wrong way.
|
|
#
?
Nov 5, 2016 23:15
|
|
|
Convergence posted:Just look up the numbers instead of positing some vague, meaningless perception of reality. There have been 43 presidents. There have been 4 cases where the winner lost the popular vote. It's historically 10% of the time. Nah, I made it very clear that I was saying I personally found it hard to believe, not that I was sure it was wrong. Anyway, there have been 57 presidential elections, so that's 4/57, or 7%. One of those had neither candidate reaching the 50%+ of electoral votes needed, which is not what 538's 12.6% chance reflects, so it's actually 3/57 that have had an electoral college winner that did not win the popular vote, or 5%. sourdough fucked around with this message at 23:20 on Nov 5, 2016 |
|
#
?
Nov 5, 2016 23:17
|
|
|
Periodiko posted:That's usually not what he says though, he says that if circumstances in the race shift, or if Trump is lucky and polling was error-ridden, he could win. His point is always that the race is not a certainty and is becoming tighter, never that Trump is favored to win. People then retweet it and mischaracterize it. Probably the more applicable criticism is that 538 puts out a never ending wall of articles using almost the same language to make the argument that one of two candidates will win the election.
|
|
#
?
Nov 5, 2016 23:17
|
|
|
Pastrymancy posted:Clinton Campaign Manager Robby Mook explains her visits to MI: I don't understand the current reason why states that don't have early voting aren't moving to change that. I also don't understand why early voting ends instead of just running to election day.
|
|
#
?
Nov 5, 2016 23:17
|
|
|
Nate is like a weather forecaster saying it might rain five days from now if that low pressure front doesn't change course. Or it might not rain. Don't blame me either way.
|
|
#
?
Nov 5, 2016 23:21
|
|
|
Pastrymancy posted:Clinton Campaign Manager Robby Mook explains her visits to MI: John Roberts has nothing to do with MI and PA's voting system
|
|
#
?
Nov 5, 2016 23:21
|
|
|
Wyld Thang posted:The Democratic party literally had to schedule the Nevada caucus to be on a Saturday or risk the apple cart being overturned. Nevada dumping caucus is my fever dream scenario. gently caress this loving cheapskate state and gently caress how inept it's political parties often are. The Democrats were the clowns this cycle, but the GOP had issues in 08 and 12. We literally only have a caucus because in 2004ish the state got too cheap to continue primaries and by moving to caucus we could shove the cost onto the parties and ask the Secretary of State and the county election departments to stay home.
|
|
#
?
Nov 5, 2016 23:22
|
|
|
Cingulate posted:There is zero reason for Nate not to do it. If anyone else were to somehow copy his model, people would still go to Nate cause he has better graphics. Come a long way since 2008, eh? 
|
|
#
?
Nov 5, 2016 23:23
|
|
|
I'm out of the country right now, but what's the read everybody is getting from Republican friends/idiot coworkers? Are they in denial like in 2012? I have like one friend who's now making a list of "Trump haters" so he can boycott them after the election. The thing that I remember from 2012 was how certain Nate was. He and Sam were both like, yeah this is 100% Obama.
|
|
#
?
Nov 5, 2016 23:23
|
|
|
RVProfootballer posted:Nah, I made it very clear that I was saying I personally found it hard to believe, not that I was sure it was wrong. hmmmmmm fair enough
|
|
#
?
Nov 5, 2016 23:23
|
|
|
Bloops Crusts posted:Come a long way since 2008, eh? bollig posted:I'm out of the country right now, but what's the read everybody is getting from Republican friends/idiot coworkers? Are they in denial like in 2012? I have like one friend who's now making a list of "Trump haters" so he can boycott them after the election.
|
|
#
?
Nov 5, 2016 23:26
|
|
|
Cingulate posted:Correct - he talks about the model here: http://fivethirtyeight.com/features/a-users-guide-to-fivethirtyeights-2016-general-election-forecast/ Oh, absolutely and without question he shouldn't change now. Heck, I got twitchy when he updated his model to include Mcmullin as a viable third party candidate in Utah, and that was pretty unquestionably needed! But for a little more talk about early voting and why he should change it post-election, let's try a thought experiment. Say there's the state of "Adaven." Adaven's a swing state, and the polling average has Hillary up by 1.5 points right now. We have a lotta polls that average out to that, once you account for house effects and so on (we'll leave thinking too hard about those out right now- it actually doesn't overall change too much from a purely unweighted average). Once we do that, Nate's model goes through and does a bunch of adjustments and he comes out with the prediction that, well, they're pretty much going to be tied on election day. Fine. What if, in Adaven, 100% of the electorate voted early? We still have polls, and they still have some error. But when we feed those polls that have "accounted for early voting", as Nate's put it, into the 538 algorithm, some oddities crop up. We apply a likely voter adjustment to data from voters that have all already voted. We allocate the undecided in an electorate that has already finished voting. Most worryingly, we apply a trend-line adjustment to voters that have already voted. We've taken real data and adjusted it as though it were still in doubt, and artificially shifted our estimate of the mean. This is pretty blatantly and obviously a bad idea. The problem is, what if only 20% of Adaven's electorate voted early? We're still applying those adjustments to everyone. Which means we're still applying adjustments to the voters who have already voted, since those are a subset of the Likely Voters in our database of polls! We're skewing our sample. So, for means-estimation purposes, it would be more apt to soften the effect of some of those adjustments the 538 model is based on in polls that are taken after early voting has started. This will end up being a minor change in the scheme of things, honestly. What's more important is that there's less uncertainty when you have early voting. Go back to Adaven. If you have a stack of four polls taken from an electorate that's completely voted, how sure are you that the poll average is in error? You've stripped out a lot of (but not all) sources of systematic error. The data is, then, the data. You could use voter registration data to add information, but we're going to stick to just the polls. In that case, the uncertainty approaches the uncertainty of a straight combination of polls that arises just from sample size, which is far, far lower than the uncertainty in the 538 model. In fact, it's pretty much the same uncertainty that's in the PEC model! So, even if you assume the 538 model is correct about how much uncertainty is now, you can say with equivalent confidence that Wang's model is correct about the uncertainty if the electorate had wholly voted. If some fraction of the electorate's already voted, does it not then follow that the uncertainty should be somewhere between what 538 has now and what PEC has now? The important part of what I'm saying is that you're not introducing non-pollish demographic data or anything like Ralston's doing in Nevada. Just the fact that X% of the electorate (and yes, that number has its own small uncertainty based on us not knowing a priori the turnout on election day) has voted will both change your estimated mean and reduce your estimated standard deviation. Something like that is what I expect Nate Silver will do after the election. Less adjustment of the strict polling average on polls taken post-early voting, and less uncertainty with the more polls that occur post-early voting.
|
|
#
?
Nov 5, 2016 23:27
|
|
|
bollig posted:I'm out of the country right now, but what's the read everybody is getting from Republican friends/idiot coworkers? Are they in denial like in 2012? I have like one friend who's now making a list of "Trump haters" so he can boycott them after the election. The Trump supporters I know think he is absolutely going to win. About a quarter of the Hillary supporters I know also think Trump is going to win.
|
|
#
?
Nov 5, 2016 23:28
|
|
|
Nate Silver's avoiding making his model public because it's not a very good model, and it'd get ripped to shreds. It's already getting some heavy poo poo when dumb seismic shifts happen in response to two new Google Survey polls and an unexpectedly red Skinny Girl Analytics from Missouri. Also, it's pretty clear from reading his articles that he's keeping a close eye on site metrics. Which is probably bad policy.. the writers should be kept firewalled from the money guys as best as possible, lest the readers smell the bullshit. MrBuddyLee fucked around with this message at 23:31 on Nov 5, 2016 |
|
#
?
Nov 5, 2016 23:29
|
|
|
A second faithless elector has hit Washington's electoral votes for Hillary.
|
|
#
?
Nov 5, 2016 23:30
|
|
|
ComradeCosmobot posted:A second faithless elector has hit Washington's electoral votes for Hillary. I for one am shocked that it's berniebros who are the ones in favor of overturning the will of the electorate.
|
|
#
?
Nov 5, 2016 23:31
|
|
|
Cingulate posted:I'm not American, but I actually just found out I have a lone Trump supporter amongst my Facebook friends. Posted something about immigrants and sweets. No, not the poison one, a different one. Also have a little video about the election, because Rachel Bloom is an awesome person.
|
|
#
?
Nov 5, 2016 23:32
|
|
|
A Winner is Jew posted:I for one am shocked that it's berniebros who are the ones in favor of overturning the will of the electorate.
|
|
#
?
Nov 5, 2016 23:41
|
|
|
A Winner is Jew posted:I for one am shocked that it's berniebros who are the ones in favor of overturning the will of the electorate. There may be even more hiding in the woodwork than just these two. quote:He said he’s heard from a few other national Democratic electors who are considering joining him.  if Trump wins because of leftists splitting the electoral vote for Hillary while Republican electoral vote in lockstep for Trump and it goes to the House.
|
|
#
?
Nov 5, 2016 23:41
|
|
|
ComradeCosmobot posted:
This is so incredibly insane. I made this post earlier today. cant cook creole bream posted:Unless you do some really wonky stuff, it's kinda hard to get exactly 270. But somehow I love the idea that Clinton wins the popular vote and about 275 electoral, but loses 6 this way. Then it is a tie which obviously goes to Trump. I'm pretty sure that's the dream scenario for Russia. If that would happen, I legitimately think there would be a civil war. Just imagine the same outcome with reversed parties. There would be blood on the streets within minutes.
|
|
#
?
Nov 5, 2016 23:44
|
|
|
Periodiko posted:That's such a good point I hadn't even considered. Trump has no former Republican Presidents, all of his primary rivals hate his guts and are only working with him reluctantly except Chris Christie, who can't appear in public because of Bridgegate. Swing state congresscreatures are trying to walk a tight rope of not appearing too close to him. He has no surrogates. What a mess. From 5 pages ago. Does being a TV proxy count? Black Token #3 and Huckelberry have had talking head appearances. Reminder : The only good thing Hucckleberry has ever done is cockblocking Zodiac Killer from the stage/the photo op when they went to defend that incest pedo. Also doing volunteer work with local D-offices and experiencing a real life in-person echo box is pretty good anti-arzying.
|
|
#
?
Nov 5, 2016 23:44
|
|
|
Cingulate posted:There is zero reason for Nate not to do it. If anyone else were to somehow copy his model, people would still go to Nate cause he has better graphics. And he could even put it under some license that disallows it? Who even knows if his contract with Disney would let him do that? And I mean, I'm not sure what he'd gain. Sure it would please people like you and me, but would the general public care?
|
|
#
?
Nov 5, 2016 23:45
|
|
|
MrBuddyLee posted:Nate Silver's avoiding making his model public because it's not a very good model, and it'd get ripped to shreds. cant cook creole bream posted:Similarly an acquaintance absolutely despises Clinton because he thinks that she will let just as many Muslims in, as Merkel does here because women in politics can't think straight. I kinda lost interest in talking to him since then. But he's the boyfriend of a friend, so I get to see him sometimes. Mukaikubo posted:Go back to Adaven. If you have a stack of four polls taken from an electorate that's completely voted, how sure are you that the poll average is in error? CottonWolf posted:Who even knows if his contract with Disney would let him do that? And I mean, I'm not sure what he'd gain. Sure it would please people like you and me, but would the general public care?
|
|
#
?
Nov 5, 2016 23:47
|
|
|
I wonder if these WA dudes are just doing it to see if Bernie will call them. And if they want to use that phone call to try to convince Bernie that he REALLY CAN WIN THIS THING
|
|
#
?
Nov 5, 2016 23:49
|
|
|
Cingulate posted:As with all open science/software: people could tell him about errors. True. But whatever he would end up releasing, I couldn't imagine a private company (read: specifically Disney) letting him do a real code dump. Perhaps I'm overly cynical. CottonWolf fucked around with this message at 23:52 on Nov 5, 2016 |
|
#
?
Nov 5, 2016 23:50
|
|
|
cant cook creole bream posted:This is so incredibly insane. I made this post earlier today. If it the electoral college is that close it isn't hard to imagine intimidation or violence against voters who are openly stating they won't support the popular will, especially when they're enabling Trump.
|
|
#
?
Nov 5, 2016 23:51
|
|
|
Also, one might imagine the reason Hillary's visiting Michigan is at least partially because of the wildly unpredictable result in the primaries compared to the primary polling. Somethin' weird goin on there.
|
|
#
?
Nov 5, 2016 23:53
|
|
|
Cingulate posted:Sorry to answer an effortpost with a one-line, but I don't really understand the second half - what does this line mean? Because I forgot to type the "n't" part of "isn't".  What I was getting at is how likely is the actual vote totals to be near the unmodified average of polls, if everyone's already voted.
|
|
#
?
Nov 5, 2016 23:52
|
|
|
CottonWolf posted:True. But whatever he would end up releasing, I couldn't imagine a private company (read: specifically Disney) letting him do a real code dump. Perhaps I'm overly cynical. Mukaikubo posted:Because I forgot to type the "n't" part of "isn't". I bet there's a stats thesis in that one.
|
|
#
?
Nov 5, 2016 23:57
|
|
|
Mukaikubo posted:Something like that is what I expect Nate Silver will do after the election. Less adjustment of the strict polling average on polls taken post-early voting, and less uncertainty with the more polls that occur post-early voting. OK, for a non-American non-statistician like me, how safe are EV numbers going to be after the election? How does it differ from state to state in collection, how many elections do you need to make safer guesses in your model, do we have that kind of data going back already (I doubt it or we wouldn't be having this discussion)? I find the baying for EV stats concerning given that it seems a very recent phenomenon. And of course if it appears to have had an important effect on the election, it will get hosed with in states where that is not "desireable". It seems to me to be an extra dollop of complexity on top of an already complex problem.
|
|
#
?
Nov 5, 2016 23:57
|
|
|
Bloops Crusts posted:Thanks. Find your angle and tell your truth. I hate the script. I hate how dry it is. So I tell them how a candidate or an issue affects me, my life, my family, my theoretical kids. With Hillary, I talk about the student loan debt relief, since it's a concrete plan that's going to jump start my whole generation. If I hadn't paid my college loans off last year, I'd have jumped at the chance for her program. Ask them simple questions. Do they have opinions? Let them vent if they do. Listen to them. That's really what's missing in today's rhetoric, valid or not. Nobody thinks anyone is listening. So make your two minutes count for THEM. Maybe you can move them. Maybe you can't. Maybe they just won't understand or care. But they'll have told you something. And you can use it. Let your neighborhood tell you who they are, and they'll let you know how to approach them. They're just humans. And they're awesome if you give them time and space. I like humans.
|
|
#
?
Nov 5, 2016 23:58
|
|
|
A Winner is Jew posted:I for one am shocked that it's berniebros who are the ones in favor of overturning the will of the electorate. Ugh and it reads like he's convinced himself that as long as he casts his vote for a third party, it's not his fault even if his vote would be the decider. "If I can't have my way, I want to watch the world burn" makes for odd and terrifying bedfellows. bollig posted:I'm out of the country right now, but what's the read everybody is getting from Republican friends/idiot coworkers? Are they in denial like in 2012? I have like one friend who's now making a list of "Trump haters" so he can boycott them after the election. I have a very devout Christian colleague from a previous job who is absolutely beside himself over the the Podesta/Marina Abramovich email. He thinks Satan worship will occur in the White House if Hillary wins. I'm pretty sure he's for real with his fear of her "connection" to black magic and the occult. I'm having a number of "why am I still Facebook friends with this person" moments over this election but this guy and the hysterical "Hillary for prison" Bernie or bust types frustrate me way more than the guy who just shares a Milo meme with "HAHAHAHAHAHA"
|
|
#
?
Nov 6, 2016 00:06
|
|
|
When I was a kid, elections seemed like a big deal but weren't this debased, country-rending event that we have going on now. I remember the Hanging-Chad Debacle of the 2000 election, but that seemed like mostly a big deal because we just wanted to know who won; for all that it was to decide the presidential election, it seemed like there was less at stake. Am I wrong about this? Was I just out of touch back then because I was in 5th grade at the time, and it's hard to get a grasp on what's important when you can't imagine a world where you don't have to worry about final exams next month?
|
|
#
?
Nov 6, 2016 00:07
|
|
|
|
| # ? Jun 11, 2024 06:30 |
|
|
Your Boy Fancy posted:Find your angle and tell your truth.
|
|
#
?
Nov 6, 2016 00:09
|
|