RastaMentality
Banned
We're probably gonna need a new thread by tomorrow morning right?
Hell lets start our own weekly gaf poll!
No, I've understood you perfectly. Let me quote the precise part I mean:
This is wrong. You do not average polls because they use different methodologies. Some methodologies will be more accurate than others because the underlying assumptions are better. Some will be worse because the underlying assumptions are worse. All averaging a good set of assumptions with a bad set of assumptions does is produce a third result which is less accurate than the good set of assumptions. You would have been better off *not averaging* and simply using the good set of assumptions to begin with.
You average samples because it is effectively the same as creating a bigger sample - you know that on the balance of probabilities, you will be more accurate by doing so. Note that this isn't guaranteed; but expected in the strict mathematical sense of the word. There is a statistical proof of this. However, you shouldn't average methodologies - you should apply a single methodology, the one which has the most credibility. This is because there is no reason to suppose that on the balance of probabilities, averaging methodology will improve either the accuracy or the precision of your results compared to any given single methodology.
You can see why we say its best to average polls, and to stop fretting so much about single polls.
Morse code only.Landlines only please.
But we have no way of knowing which methodologies are more accurate
You obviously know more about statistics than I do, so if you want a more fruitful discussion on why different LV methodologies are, in fact, a compelling reason to rely on polling averages take it up with Cohn on twitter.
I wonder how much weight a poll consisting only of Youtube comments would get.
Stein had planned to challenge her exclusion from tonights event by hosting a rally outside the secured perimeter of the debate hall beginning around 5 p.m. ET. She said she will be live on Twitters Periscope app answering the same questions as Democratic nominee Hillary Clinton and Republican nominee Donald Trump on the debate stage.
Steins campaign said she will not risk arrest this time, because there is an outstanding warrant for her arrest over her involvement in a recent protest against a controversial pipeline project in North Dakota. Still, her campaign spokeswoman Meleiza Figueroa said they will attempt to get the spirited demonstration as close to the gates as possible."
Oh god noooooooooo
But this isn't true! We have all sorts of reasons. For example, we have strong reason to suppose landline polls won't produce an accurate sample because there are many eligible voters who don't have a landline.
I might, actually. I think that's a pretty terrible conclusion he's made there. It's the statistician's equivalent of "eh, fuck it, just fudge it a bit and it should turn out okay".
A good test of whether a model is too conservative, too aggressive or “just right” is whether it does a good job of matching new polls as they come out in a state. So far in this election, the FiveThirtyEight and Daily Kos Elections model — which also uses a trend-line adjustment — have done a good job of this, while other models sometimes lag behind the trend.
FiveThirtyEight’s models also generally account for more uncertainty than other models — or at least they do in this election because the presence of a large number of undecided and third-party voters, who contribute to polling volatility. That helps Trump’s odds, since he’s (narrowly) the underdog in our forecast.
And if Clinton gains following the debates, FiveThirtyEight’s models will probably be among the quicker ones to detect it.
I find it's the same people with the "burn it all down" mentality. They don't want to vote the big parties and know that they are throwing their vote away.
But this isn't true! We have all sorts of reasons. For example, we have strong reason to suppose landline polls won't produce an accurate sample because there are many eligible voters who don't have a landline.
I might, actually. I think that's a pretty terrible conclusion he's made there. It's the statistician's equivalent of "eh, fuck it, just fudge it a bit and it should turn out okay".
I love how we ended up getting a good day of polling for Hillary after that Selzer disaster.
I love how every time Silver screams about a dead heat, the polling slaps him in the face.
I love how every time Silver screams about a dead heat, the polling slaps him in the face.
I heard Jack Kingston on NewsHour claiming Hillary had 13 private email servers and it just went unchallenged by Judy Woodruff. Gah!
Subdued trump only shows up when teleprompters are there.I love that Hillary campaign has practiced in case they get insane Trump or subdued Trump. It makes my heart flutter
To quote Wang: Reversion to the mean!
If a poll came out ten minutes into the future, we still probably have posted it and discussed it.
Chuck Todd says some of the polls today look like they were designed to be headline generators
But this isn't true! We have all sorts of reasons. For example, we have strong reason to suppose landline polls won't produce an accurate sample because there are many eligible voters who don't have a landline.
Thor +2538 just loaded poll results from the next two weeks.
Trump +4
And yet, they've prepared for it.Subdued trump only shows up when teleprompters are there.
There is none
Chuck Todd says some of the polls today look like they were designed to be headline generators
This may have been posted.
But before anyone listens to the post-debate coverage, read this about the media this year:
http://pressthink.org/2016/09/asymm...s-fries-the-circuits-of-the-mainstream-press/
Best article about it so far.
I feel pretty good about tonight. Hillary is on the right! Let's go khalesi!
You're right, FWIW. But...excluding the obvious, I think his point is that we don't know which voter turnout model is right and so we just have to average those (though I believe things like house effect should be included too).
You know I just realized that usually even if you have democratic and republican pundits on, one side will usually at the very least admit their candidate didn't do as well of job as they liked on a certain issue or two. There is no way any of the Trump people will will be willing to move an inch even if Trump bombs catastrophically.
She's going to attack Trump from the right? Is that a good idea?
Don't worry we'll make sure this thread goes waaay too fast as well. Sorry MambaWell, I'll be posting here instead of the Debate thread. That thread will just be going way too fast for me.
Between twitter and my fantasy football scoring updates, gonna be too hard to manage two threads and so I'll stay here!
See y'all just before the debate!