One of the "fun" things about presidential elections is that every four years there's a new dumb thing about the process for political junkies to yell at each other about despite the disinterest or genuine disgust of normals, and this year it's polling! Did Nate Silver's devil-math suck all the fun out of democracy, forever? Were the polls skewed because they didn't reflect Republican understanding of reality? Were Gallup and Rasmussen "in the tank" for Republicans? Well, Gallup Editor-in-Chief Frank Newport has decided to weigh in on this controversy, and would like you to know that (a) Gallup was
good point frank. no one in the history of ever has ever used 'information' to do 'analyses' and 'draw conclusions' and 'make predictions'.
man can you IMAGINE how bad THAT would be? why, such laxity might even produce, i don't know, a class of political pundits.
First of all, Gallup was off by quite a bit. They suspended polling for about a week and took that opportunity to get back on track. By Nov. 6th they were quite close.
Second, Nate Silver is no Sludge. He doesn't just aggregate. He has a very complicated model that runs endless (well, not quite - just some 25,000 per day!) iterations of what the data really mean. All the polls are just the first step in his analysis.
What about Real Clear Politics? They just post the polls and average them. TMC? I thought they were smoking some pretty good shit, but they came pretty close. Yahoo News averaged the various groups like Intrade.
As others have pointed out, why do they dog Nate Silver? Because he is so good?
What's so great about this "argument" is that Nate Silver would be the first one to say, "Yes! Totally true!" If there are no polls, he can't do his aggregation model. No polls, no Nate. Of course, it's true only in that it is a tautology and doesn't get into the actual issue, but, eh, whatever.
"[T]he Romney campaign was falling apart, people were not optimistic, nobody thought there was a chance of victory and I felt that it was my duty at that point to go out and say what I said."
Not hard to see where he thought his &quot;duty&quot; <strike>lay</strike> lied.
This &quot;hate Nate&quot; shit is amazing. The day after the election, some yoyo on Slate or Salon or something (I&#039;m not gonna bother to re-find it) wrote a quite long piece, the gist of which was &quot;Nate Silver is not a genius. It&#039;s the polling companies that do all the work, and they&#039;re pretty much right anyhow&quot;. (It was much longer than that.) Now this Gallup guy.
They completely ignore what it is that Nate actually does. The body of work that underlies his his forecasts is this: He has detected evidence of statistical bias in the the published results of various polling organizations, and he has invested a lot of work in quantifying these biases, based on historical data. This is the secret sauce in his models. His model makes adjustments and/or assigns weights to the published polling results, in an attempt to create a less-biased composite result for each election.
It&#039;s worth remarking that &quot;statistical bias&quot; is a mathematical concept. Nate does not presume that any polling organization <i>intentionally</i> skews its results right or left. The numbers speak for themselves -- if bias exists, it exists, and it can be measured (historically) and corrected for.
If Gallup, or any other pollster, wanted to put Nate Silver out of the elections business, all they would have to do is analyze their own results for statistical bias, and then take steps to either eliminate the source of bias or correct for it. In the short run, this would screw Nate up, because he&#039;d be applying history-based corrections that were no longer appropriate. In the long run, it would remove his added value, because anybody can meta-average unbiased polls to get an unbiased composite.
Based on Gallup-guy&#039;s remarks, it doesn&#039;t look like they&#039;re going to be trying introspection any time soon, so Nate&#039;s business model seems secure.
good point frank. no one in the history of ever has ever used &#039;information&#039; to do &#039;analyses&#039; and &#039;draw conclusions&#039; and &#039;make predictions&#039;.
man can you IMAGINE how bad THAT would be? why, such laxity might even produce, i don&#039;t know, a class of political pundits.
that would suck.
also frank, no one understands &#039;commons&#039; and &#039;cows&#039; anymore.
you should talk about, say, military men sharing women from florida.
First of all, Gallup was off by quite a bit. They suspended polling for about a week and took that opportunity to get back on track. By Nov. 6th they were quite close.
Second, Nate Silver is no Sludge. He doesn&#039;t just aggregate. He has a very complicated model that runs endless (well, not quite - just some 25,000 per day!) iterations of what the data really mean. All the polls are just the first step in his analysis.
What about Real Clear Politics? They just post the polls and average them. TMC? I thought they were smoking some pretty good shit, but they came pretty close. Yahoo News averaged the various groups like Intrade.
As others have pointed out, why do they dog Nate Silver? Because he is so good?
ff:
I&quot;m beginning to get the idea they are more common than the Kardashians.
What&#039;s so great about this &quot;argument&quot; is that Nate Silver would be the first one to say, &quot;Yes! Totally true!&quot; If there are no polls, he can&#039;t do his aggregation model. No polls, no Nate. Of course, it&#039;s true only in that it is a tautology and doesn&#039;t get into the actual issue, but, eh, whatever.
So Nate Silver is the Ariana Huffington of the polling industry? I can hardly wait for him to start bussing us to rallies in DC!
&quot;[T]he Romney campaign was falling apart, people were not optimistic, nobody thought there was a chance of victory and I felt that it was my duty at that point to go out and say what I said.&quot;
Not hard to see where he thought his &quot;duty&quot; <strike>lay</strike> lied.
I want to see other pollsters polled on this question. And Silver should parse the results.
Now don&#039;t go making an issue of Mitt&#039;s religion. It&#039;s not right to... wait, what?
Within 24 hours of Nate first putting Obama in the lead.
<i>&quot;... I felt that it was my duty at that point to go out and say what I said...&quot;</i>
Spelled &quot;doody&quot; wrong.
Best comment: Karl Rove says it&#039;s too early to call the guy dead.
Should be a h/t to Defeatably_Joe for posting the Morris non-apology on Wonkville.
So does this mean that deer isn&#039;t dead after all?
This &quot;hate Nate&quot; shit is amazing. The day after the election, some yoyo on Slate or Salon or something (I&#039;m not gonna bother to re-find it) wrote a quite long piece, the gist of which was &quot;Nate Silver is not a genius. It&#039;s the polling companies that do all the work, and they&#039;re pretty much right anyhow&quot;. (It was much longer than that.) Now this Gallup guy.
They completely ignore what it is that Nate actually does. The body of work that underlies his his forecasts is this: He has detected evidence of statistical bias in the the published results of various polling organizations, and he has invested a lot of work in quantifying these biases, based on historical data. This is the secret sauce in his models. His model makes adjustments and/or assigns weights to the published polling results, in an attempt to create a less-biased composite result for each election.
It&#039;s worth remarking that &quot;statistical bias&quot; is a mathematical concept. Nate does not presume that any polling organization <i>intentionally</i> skews its results right or left. The numbers speak for themselves -- if bias exists, it exists, and it can be measured (historically) and corrected for.
If Gallup, or any other pollster, wanted to put Nate Silver out of the elections business, all they would have to do is analyze their own results for statistical bias, and then take steps to either eliminate the source of bias or correct for it. In the short run, this would screw Nate up, because he&#039;d be applying history-based corrections that were no longer appropriate. In the long run, it would remove his added value, because anybody can meta-average unbiased polls to get an unbiased composite.
Based on Gallup-guy&#039;s remarks, it doesn&#039;t look like they&#039;re going to be trying introspection any time soon, so Nate&#039;s business model seems secure.
It&#039;s so cute when boring people fight.