2016 Postmortem
Related: About this forumNate Silver's Take on the Polling Noise at 538 & a Plea
Forgive me if this has been posted here before - http://fivethirtyeight.blogs.nytimes.com/2012/10/15/oct-15-distracted-by-polling-noise/
But I urge all of you to read it. I also make a plea that today everyone focus on what an excellent job our president will do tonight. Be in support of him. Our country is in danger if the conservatives have their way and the next four years (and beyond) could be very bleak. So, start the positive thinking/posting/talk now and please don't focus on polls today.
longship
(40,416 posts)Nate has some important rationale on why DUers shouldn't be so focussed on polls right now, especially individual polls.
All polls are biased, people. That's why Nate takes them all. That's why I trust him, so far. We'll see how accurate he is soon enough.
But we need to chill out about polls because they are all flawed.
GeorgeGist
(25,323 posts)pushing bullshit.
longship
(40,416 posts)I watch all the poll aggregators, including Nate. I don't care who gets rich or not, I only care who gets it right.
I kind of aggregate the aggregators.
The thing is that the polls ARE all over the place. So, I do not trust them the last couple weeks. Either that, or the US voters are an extremely fickle lot. I would rather have the former than the latter.
We will see in three weeks. In the meantime I am paying attention and am not panicking... Yet.
SunSeeker
(51,697 posts)And, if anything, is reversing.
ShadowLiberal
(2,237 posts)Obama has been moving around from 63% to 67% favored to win the last few days. The worst it got him was like 60.1%.
But it's not getting better again, it had creeped up a bit more to 66%, but then it went back down to 63% today, probably from statistical noise at this point.
progressivebydesign
(19,458 posts)Baitball Blogger
(46,757 posts)And you all don't wonder, that with Gravis being in Florida, that Florida has one of the largest fluctuations in favor of Romney?
longship
(40,416 posts)Polling isn't easy so there is always noise. That's why you shouldn't throw out outliers. You take the whole data set in the hopes that the errors cancel out. If you cherry pick polls, or throw out outliers, that is bad.
Baitball Blogger
(46,757 posts)How about autocalls in English only? And how do you know they're random numbers?
longship
(40,416 posts)Stat 101.
Read about Gaussian statistics.
Baitball Blogger
(46,757 posts)Why Gaussian statistics are mostly wrong for strategic organization
http://soq.sagepub.com/content/3/2/219.citation
Drunken Irishman
(34,857 posts)This is unusual. I posted in another thread that between Oct. 7th and Oct. 16th four years ago, 22 new polls were released. Between a similar stretch this year ... only seven new polls have been released and almost all of 'em are daily tracking polls.
We are seeing an usual amount of under-polling this election and because of it, only a handful of polls are driving the narrative.
It's bizarre.
justiceischeap
(14,040 posts)he talked about how the post-debate spin actually effected the polls. He stated he'd love a half hour media blackout for polling and then let the pundits speak.
jkrichter
(46 posts)I don't like the lobbyist money going to Romney. We need to support our President!