Sunday, November 11, 2012

By The Way, Nate Silver @ FiveThirtyEight Got it Right (Again)

Nate Silver of FiveThirtyEight.com
Look, all of you "skewed poll" conspiracy theorists out there can hate him all you want but Nate Silver has proved for the third consecutive election season that he knows what he's talking about.  During the 2008 Election, he called the Presidential race nearly 100% correct (he missed 1 state (Indiana) and 1 district within Nebraska).  During the 2010 Midterm Election, he accurately predicted 34 out of the 36 US Senate races, 36 out of the 37 Gubernatorial races, and 426 out of the 435 US House of Representatives races.  Despite all of that, there were many critics (predominately on the Right) who refused to believe that Nate Silver's analytic model was correct this time in 2012.



Per NY Times:


It was not on any ballot, but one of the biggest election contests this week pitted pundits against pollsters. It was a pitched battle between two self-assured rivals: those who relied on an unscientific mixture of experience, anecdotal details and “Spidey sense,” and those who stuck to cold, hard numbers. 
When the results were tabulated, it became clear that data had bested divination.
The election results that delivered a second term to President Obama on Tuesday left some well-known pundits, many of whom have a partisan bent, eating crow on Wednesday morning — including analysts like Karl Rove, Dick Morris and Michael Barone, all of whom had confidently predicted a victory by Mitt Romney.
The results were much kinder to pollsters and the data devotees who aggregate and average polls, or who use mathematical models to make projections.
“While the election’s biggest winner was President Barack Obama, the other victory on Tuesday night went to the careful application of reason, data and, yes, to the science of modern survey research,” Mark Blumenthal, the senior polling editor of The Huffington Post and the founding editor of Pollster.com, wrote on Wednesday. “The losers were the amateur poll mavens who sought to ‘unskew’ the polls and the pundits who saw what they wanted to see.”
Quite a few of the numbers-crunchers could claim bragging rights on the morning after.
The models used by Sam Wang at the Princeton Election Consortium, Drew Linzer at Votamatic and Nate Silver, who writes the FiveThirtyEight blog for The New York Times, all came remarkably close to forecasting the final result, depending on which candidate ultimately wins Florida.
On the night before the 2012 Election, Nate Silver had the following predictions:


Battleground States:
Colorado 81.8% chance for Obama
Florida 53.1 % chance for Obama (was 53.1% for Romney on Nov. 5, 2012)
Indiana 99.6% chance for Romney
Iowa 86.0% chance for Obama
Michigan 99.5% chance for Obama
Nevada 94.4% chance for Obama
New Hampshire 86.2% chance for Obama
North Carolina 72.5% for Romney
Ohio 91.8% chance for Obama
Pennsylvania 98.8% chance for Obama
Virginia 81.4% chance for Obama
Wisconsin 97.2% chance for Obama


Which projected the following Electoral College Vote:
Obama: 332
Romney: 206
How many of these projections did he get correct?

Answer: ALL OF THEM!!!

Let me say that again in case you weren't paying attention; Nate Silver accurately predicted all (as in 100%) of the Battleground States for the 2012 Presidential Election.  But he didn't just stop there. He also accurately predicted all 50 out of the 50 states as well. He projected that the final electoral college tally would be 332 Obama to 206 Romney and that is, in fact, what it ended up being.

Silver also correctly predicted 30 out of the 32 US Senate races. His first error: Montana.  Silver's model projected that Republican candidate Denny Rehberg would defeat Democratic candidate Jon Tester 49.9% to 48.4%, respectively.  It was actually Tester who won the race with 48.7% of the vote (0.3% higher than Silver's prediction) to Rehberg's 44.8% (5.1% lower than Silver's prediction; Independent candidate Dan Cox took 6.5% of the vote). His other error occurred in North Dakota where Silver's model showed Republican Rick Berg with 52.8% of the vote defeating Democrat Heidi Heitkamp with 47.2% of the vote.  The final result: Heidi Heitkamp won 50.5% of the vote to Rick Berg's 49.5%.  Silver was off by exactly 3.3% for each candidate. [EDITOR'S NOTE:  we'd be wrong not to mention that Silver called one that none of us here thought was going to happen - Democrat Tim Kaine's defeat over Republican George Allen for the US Senate Seat in Virginia.  Silver called it Kaine 51.0%, Allen 47.6%. Final result: Kaine 52.5%, Allen 47.5%.  Scary.]

To recap, Silver batted 100 for the Presidential race, and he got 94% of the Senate races correct.  And for all the cries of "liberal bias" that came from the Right during the election, Silver's only two errors occurred where he actually gave the benefit of the doubt to Republican candidates, not to Democratic candidates.  Maybe next time the folks living the bubble will learn to trust math.  Maybe.

 



blog comments powered by Disqus