-
Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.
PoliGAF 2016 |OT11| Well this is exciting
- Thread starter Crayons
- Start date
- Status
Officerrob
Banned
Joe and Mika went to Trump to kiss his ring? Yikes!
whyamihere
Banned
WELL HMM
@jhagner
This Elon poll Likely Voter screen is bananas. A screen this tight doesn't help you understand the electorate. Can you name your precinct?
ItWasMeantToBe19
Banned
Trump is very mad that the U.S. bombed Russia aligned Syrian and Hezbollah troops for some reason:
balladofwindfishes
Member
this makes their polls entirely useless...
Jerry_Seinfeld.jpg
D
Deleted member 231381
Unconfirmed Member
Is that such a terrible screen? I mean, 644 of 799 met those conditions, which means "turnout" is 80.6%. That's... pretty generous.
Lots of this is veering into poll "unskewing" now. It's probably just a truth Clinton is doing quite poorly at the moment. It's also a truth that's probably enough to beat Trump. Even if we did want to point and laugh at pollsters for bad LV screens, that screen would definitely not be first on the list.
Lots of this is veering into poll "unskewing" now. It's probably just a truth Clinton is doing quite poorly at the moment. It's also a truth that's probably enough to beat Trump. Even if we did want to point and laugh at pollsters for bad LV screens, that screen would definitely not be first on the list.
That would be Gravis.
Aaron Strife
Banned
Yeah that's bogus.
Iowa (and Ohio) is trending red because Trump and the GOP is making gains with white voters without college education and there aren't enough minorities to offset those gains. That's more true in Iowa than it is in Ohio.
Obama being an amazing candidate probably staved off the inevitable in Iowa, also the state party's operation there is top notch with GOTV and early voting in particular.
Gang bangers! Black people are simultaneously terrifying precise super criminals, and inept bunglers who can't shoot straight.Trump is very mad that the U.S. bombed Russia aligned Syrian and Hezbollah troops for some reason:
balladofwindfishes
Member
That LV screen would cut me out since I have no idea what my precinct is. I don't imagine the "average" voter commits their precinct to memory
ItWasMeantToBe19
Banned
Hillary vs. Trump wasn't polled and the poll was fine for the Dems as it showed them tied for Senate and Governor.
There's nothing to unskew, it's just some pollsters think that it was done badly.
Aaron Strife
Banned
The fact that so many people met the criteria suggests to me that some of them are lying.
What if you're voting absentee by mail, for example? Your precinct wouldn't even matter in that instance.
whyamihere
Banned
Yes, asking if you know where your precinct is a very dumb screen.
There isn't even a Clinton/Trump question (yet?) in that poll, I'm literally saying it is a dumb LV screen. There are many polling methods and LV screening methods that I do not love!
Exactly, though I guess there must be an ulterior motive?
Aaron Strife
Banned
I will say showing Ross ahead but Cooper behind is a confusing result. Ross vs. Burr is a toss up at best whereas even Republicans seem to have written off McCrory.
Nazgul_Hunter
Member
James Woolsey on CNN doesn't realize he's contradicting Trump?
User 463088
Banned
I have voted in every election since I turned 18. I couldn't tell you what my precinct is. I think it's at a nursing home? Maybe....I've never voted there in person.
Trump and Hillary's Must Win States.
http://fivethirtyeight.com/features...ates-for-trump-and-clinton/?ex_cid=538twitter
http://fivethirtyeight.com/features...ates-for-trump-and-clinton/?ex_cid=538twitter
D
Deleted member 231381
Unconfirmed Member
Again, 644 of 799 did. So the "average" voter very much does remember this - or at least, the average poll respondent. Otherwise it would have been 399 of 799 or less!
And yes, some people who don't know their precinct will still vote. Some people who do know will vote. That's just the nature of these things. But it doesn't seem at all improbable to me that filtering on knowing your precinct or not will produce a more accurate result than not filtering at all. It's not as good as say a self-reported likelihood to vote weighted for socioeconomic demographic or whatever, but there are genuinely much worse LV screens.
I'm kind of tempted to do a big write-up on how pollsters actually do LV screening and why it's so hard to do!
Fragamemnon
Member
There are plenty of urban/suburban core voters in NC that early vote and haven't stepped into their actual voting precinct since moving here. The only reason I know mine is that I volunteer there on election day because it occupies the time.
Bam Bam Baklava
Member
When the hell did one random idiot unskewing polls become a dome of impervious defense for bad pollsters?
It's like scientists calling everyone who criticizes a finding in their study a flat-earther. Feels like a cheap cop out.
It's like scientists calling everyone who criticizes a finding in their study a flat-earther. Feels like a cheap cop out.
whyamihere
Banned
It's sort of annoying. It happened when we complained about landline-only pollsters before.
Maybe be a good pollster.
AntraxSuicide
Banned
Eh, there's nothing wrong with examining a poll. Nate Silver literally skews every poll in his model for bias, and he's doing it based on the same things most people here are.
If your method is garbage, your results will be garbage, and not every poll should count. If I run 40 polls real quick with bad methodology, then these should be excluded. I think even you can see why that USC thing and the Ipsos thing are poorly done. As Adam pointed out a few times, the Ipsos one even gets unskewed twice before getting counted, and that's pretty strange.
As someone from the UK, you should be more comfortable with polling misses!
exmachina64
Banned
Yes, but I'm probably an exception.
User 463088
Banned
D
Deleted member 231381
Unconfirmed Member
I mean, there are genuinely bad polls and bad likely voter models. Like, anything landline only is trash tier. Gravis' likely voter model is trash tier. But my first real job was in working on likely voter models, and I can assure you it's hard as all fuck. If you find even a half-reasonable instrumental variable you thank the gods for your luck and never touch it again in case it breaks.
shinra-bansho
Member
Does Trash Nate tell us how he skews and weights polls in his trash model for trash bedwetting.
Based on "NATIONAL TRENDZ"
whyamihere
Banned
um.
D
Deleted member 231381
Unconfirmed Member
I know this, I worked in the industry. My first proper job was with YouGOV. I'm not saying "never criticize a poll". All I'm saying is "as likely voter screens go, that one is not actually that bad". Getting likely voters right is the hardest bit of polling, by far. Getting a good sample is easy-ish if you have a fuckton of money, but even money doesn't solve the likely voter problem.
USC is fine, although I don't know why Nate treats it the way he does. Because it follows the same group of people, it will always have the same bias, which should probably be accounted for over time. On the flipside, because it follows the same group of people, the direction of travel should always be correct (to within the margin of error etc.). I'd actually be pretty happy if the majority of polls before an election were panel-based, because when examining an election season we're usually more interested in the accuracy of the direction of travel than the accuracy of what the election would be if it was held tomorrow.
IPSOS is similarly fine if you remember that on a state-by-state basis it has an absolutely enormous margin of error. More useful for establishing the national context than anything else.
Hobbestetrician
Member
This whole family needs to be shot into the sun. Can't wait for them to lose.
ItWasMeantToBe19
Banned
um.
Counter proposal: All Holocaust deniers get put on a watchlist.
This may be 80% of Trump Jr.'s Twitter follower base, but we have to be safe!
um.
This is sick.
BrokenFiction
Member
ShadowSwordmaster
Banned
um.
I hate this man.
ItWasMeantToBe19
Banned
Donald Trump's father was one of the bad ones. Member of a terror organization, very bad guy. And look at his son!
But I'm okaying letting in German immigrants to the United States still.
But I'm okaying letting in German immigrants to the United States still.
Nazgul_Hunter
Member
um.
Set aside how gross the pic is; two qs:
1) Is this an official campaign image?
2) If not, why do these idiots RT stuff with their logo like that?
Aaron Strife
Banned
Pointing out flaws in polls isn't unskewing, whether that's wonky LV screenings or undersampling certain demographics.
Unskewing is when you take that next leap into manually "correcting" those perceived flaws and rerunning the numbers for a more favorable result. You can't take a poll that shows Trump up 3 and then say "oh well Clinton is actually up 2". You might as well just be making up your own numbers at that point.
Also smh that skittles picture. I can't believe this is a mainstream candidate and not just some fringe third party asshole who makes the ballot in like one state.
Unskewing is when you take that next leap into manually "correcting" those perceived flaws and rerunning the numbers for a more favorable result. You can't take a poll that shows Trump up 3 and then say "oh well Clinton is actually up 2". You might as well just be making up your own numbers at that point.
Also smh that skittles picture. I can't believe this is a mainstream candidate and not just some fringe third party asshole who makes the ballot in like one state.
jack_package_200
Banned
um.
Is that something his son just retweeted or does that come straight from the campaign?
bangbang i'm a timpani
Banned
does nate silver disappear when this election is over? the pandering this website is doing now is ridiculous.
Wolfgunblood Garopa
Member
They wouldn't make that kind of grammar mistake. I hope not, anyway.
Aaron Strife
Banned
Depends on how well his model holds up, I suppose. And unfortunately this time he has 3 so he can just pretend he's on the mark if any of them are close.
um.
Dude...
Okay this is legitimately the first tweet that makes me really, really mad in weeks from either of these goons. Fuck them.um.
D
Deleted member 231381
Unconfirmed Member
He will have 2 by the time of the election - Nowcast and Polls Plus will converge over time.
EDIT: And Polls Only being wrong is sort of a rejection of him anyway, given from what I understand it's just a basic UNS model. He'll live and die by Polls Plus.
i think the political side of my brain is just a raging swirl of internal expletives.
D
Deleted member 231381
Unconfirmed Member
It doesn't even make a good point. I'd totally risk it on a handful of Skittles if only three were poisonous. I love Skittles.
um.
You know, at first this made me angry. Then I thought about it and realized it's not much compared to the Snake story Trump loves to tell, which I heard for the first time today. It ends with: "oh shut up silly woman ... you knew damn well I was a snake before you took me in."
Both are vile, but the tweet is no worse than what Trump's saying to packed arenas and thunderous applause.
- Status
Similar threads
- Locked
- Locked