Friday, November 9, 2012

Nate Silver, Polls and Arithmatic


Well, Nate Silver was right. Exactly right. He couldn't have been more right. His Monte Carlo models for 2012 accurately predicted the number of electoral votes each candidate would receive, as well as which states would go blue for Obama. Well-done, Sir. You have now become part of election history, and I'm glad you also stuck to your guns when others were taunting you and with Politico calling you a "one-term celebrity." I guess they should have said "Won Term Celebrity."

What's important is that you get the last laugh. :) I think you'll be around in the next election. I'm not sure about the hacks at Breitbart or those guys from Politico. What were their names again?

Excerpt of Silver's Book "The Signal and the Noise" via Daily Beast
Political news, and especially the important news that really affects the campaign, proceeds at an irregular pace. But news coverage is produced every day. Most of it is filler, packaged in the form of stories that are designed to obscure its unimportance. Not only does political coverage often lose the signal—it frequently accentuates the noise.

Snark Amendment: Pundits Wagering War on Nate Silver
Previous Posts on Nate Silver and the Polls


The Romney camp was "shellshocked" over the fact that the Republican-leaning polls were the ones that were skewed, since the party-line had been the opposite for months.

Rush, Dick and Rove Skewed into Frenzy by Polls

In the last days of the campaign, people were coming out of the woodwork to predict a landslide, which had nothing to do with the reality of the polls, most of which showed Obama ahead at that time. I gathered up their quotes and tweets here:

Snark Amendment: Republicans Predict Romney Landslide

Yeah, the GOP got a landslide, all right. They are going to be digging their way out of it for years to come!


I think one reason ordinary voters drank that Kool-Aid in addition to the Romney Camp, was that some spineless pundits in the mainstream were hedging bets on the outside chance that all the Democratic pollsters were A. Delusional and B. Liars. For instance, CNN kept talking about the "razor thin margin" which somehow gave the advantage to Mitt instead of Obama, when their own internal polling showed otherwise. Chris Cillizza at Washington Post moved Ohio into the "toss-up" column just because Romney needed it so much to win.

The Dems who took a "poll of polls" approach and checked in with statisticians and oddsmakers were not nearly as confused. Yes, we were also nervous and scared about the unknown outcome, yes, but we were not AS confused about the data. And in the end, this election was all about the math.

Bookmark these websites for future elections - math works! - Nate Silver
Princeton - Sam Wang

Below is a list compiled by professor Costas Panagopoulos, Ph.D. of Fordham University. It shows which pollsters did a good job of accurately predicting the election outcome. It's a good idea to save this list for the next election so you can avoid the polls from CNN down to AP because those were the most biased and least accurate.

Fordham University List of Pollsters Ranked from Most Accurate to Least Accurate

Most (22) polls overestimated Romney support, while six (6)
overestimated Obama strength (indicated with a * below), but none of the 28 national preelection polls I examined had a significant partisan bias.
The following list ranks the 28 organizations by the predictive accuracy of their final, national pre-election estimates (as reported on

1. Ipsos/Reuters
2. YouGov
3. PPP (D)
3. Daily Kos/SEIU/PPP
4. Angus-Reid*
5. ABC/WP*
6. Pew Research*
6. Hartford Courant/UConn*
7. Purple Strategies
8. YouGov/Economist
11. Democracy Corps (D)*

12. Monmouth/SurveyUSA
12. Politico/GWU/Battleground
12. FOX News
12. Washington Times/JZ Analytics
12. Newsmax/JZ Analytics
12. American Research Group
12. Gravis Marketing
13. National Journal*
14. Rasmussen
14. Gallup
15. NPR
16. AP/GfK

No comments:

Post a Comment