2012 US election a 'Moneyball' win for geeks

Nov 07, 2012 by Rob Lever
Attendees react to election results at a reception sponsored by the Republican Party of Florida in Tampa. It was not just a victory for President Barack Obama, it was validation for the number-crunchers and statistical model geeks, including a New York Times blogger who became a target for conservatives.

It was not just a victory for President Barack Obama, it was validation for the number-crunchers and statistical model geeks, including a New York Times blogger who became a target for conservatives.

Tuesday's made a star out of Nate Silver, whose FiveThirtyEight blog for the US daily tracked the president's statistical odds and on election day offered a 90.9 percent probability of an Obama win.

Silver's model correctly predicted the presidential outcome in 49 states, and will be correct in all 50 if Obama's lead holds in Florida.

"Here is the absolute, undoubted winner of this election: Nate Silver and big data," said Chris Taylor in an opinion column on the website Mashable.

"What does this victory mean? That mathematical models can no longer be derided by 'gut-feeling' pundits. That Silver's contention—TV pundits are generally no more accurate than a coin toss—must now be given wider credence."

It was a similar result for three other models, including from Princeton University neuroscientist Sam Wang, Stanford's Simon Jackman and Emory University's Drew Linzer.

The results evoked the popular book and film starring Brad Pitt "Moneyball," which showed how statistical models can help win in baseball.

Obama's victory "is also a victory for the Moneyball approach to politics," said John Sides, a political scientist at George Washington University.

"It shows us that we can use systematic data—economic data, polling data—to separate momentum from no-mentum, to dispense with the gaseous emanations of pundits' 'guts,' and ultimately to forecast the winner," he said in a blog post.

In the weeks leading up to the election, Silver—whose FiveThirtyEight blog is named geekily after the number of electoral college votes up for grabs—and others had been pilloried by conservative commentators.

"The -commissioned polls have been far more skewed this election season than most in the past," said conservative analyst Dean Chambers on September 26, as he predicted a win for Republican Mitt Romney.

Chambers, seen as the head of the so-called "unskewed movement," on Monday predicted a narrow Romney win in electoral votes.

Conservative consultant Dick Morris had argued that the polls "understate" the number of people who favored Romney, and ahead of the election predicted a "landslide" for the Republican.

"All of the polling out there uses some variant of the 2008 election turnout as its model for weighting respondents and this overstates the Democratic vote by a huge margin," Morris said in September.

On Wednesday, Morris acknowledged he was wrong: "I've got egg on my face," he said, arguing that the impact of superstorm Sandy "stopped Romney's post-debate momentum."

The predictions from Silver and other analysts—who for the technically literate use techniques known as quantitative forecasting or Bayesian analysis, named for mathematician Thomas Bayes—drew wide audiences during the campaign, and prompted many to place bets on accuracy.

Former Republican congressman called Silver an "ideologue" on MSNBC and bet Silver $1,000 on the race.

Princeton's Wang said he would "eat a bug" if Romney won Ohio, and expressed relief on Wednesday after the results.

Wang had accurately predicted 49 states and called Florida "a tossup." He had on Tuesday given a 99.2 percent probability of Obama's re-election.

The popular vote total giving Obama a slight edge "exactly matches my prediction," Wang said. "Bottom line: I will not have to eat a bug."

Yet some said the four predictions were simply aggregations of individual polls, and cautioned against giving too much credit to the aggregators.

George Washington University Henry Farrell said the popularity of these systems might lead to "less incentive to produce these individual polls" and make the predictions less accurate.

"The models might, over the longer term, drive the individual polls out of the market, cannibalizing the conditions of their own existence unless someone figures out a new business model," Farrell said.

Explore further: Ant colonies help evacuees in disaster zones

add to favorites email to friend print save as pdf

Related Stories

Election most tweeted event in US political history

Nov 07, 2012

Election day in the United States became the most tweeted about event in US political history Tuesday with enthusiastic netizens firing off 20 million poll-related tweets, the social network said.

Recommended for you

Freight train industry to miss safety deadline

13 minutes ago

The U.S. freight railroad industry says only one-fifth of its track will be equipped with mandatory safety technology to prevent most collisions and derailments by the deadline set by Congress.

IBM posts lower 1Q earnings amid hardware slump

1 hour ago

IBM's first-quarter earnings fell and revenue came in below Wall Street's expectations amid an ongoing decline in its hardware business, one that was exasperated by weaker demand in China and emerging markets.

Microsoft CEO is driving data-culture mindset

2 hours ago

(Phys.org) —Microsoft's future strategy: is all about leveraging data, from different sources, coming together using one cohesive Microsoft architecture. Microsoft CEO Satya Nadella on Tuesday, both in ...

User comments : 12

Adjust slider to filter visible comments by rank

Display comments: newest first

Claudius
1.6 / 5 (7) Nov 07, 2012
There is really no way to know who got more votes. With all the "glitches" and probable hacking of voting machines on both sides, what really seems to matter is which side hacked the most votes.

It would be interesting to have an election in which every vote cast was accurately counted and used to determine a winner. ATM machines can do it, why can't elections? It is just a matter of reforming the whole process. As long as people continue to rely on these dicey results, that will never happen.
LariAnn
2.3 / 5 (4) Nov 07, 2012
It would be interesting to have an election in which every vote cast was accurately counted and used to determine a winner. ATM machines can do it, why can't elections? It is just a matter of reforming the whole process. As long as people continue to rely on these dicey results, that will never happen.


The persistent problem with any voting scheme is that somewhere in the process, people get involved with the results. Someone has to program and maintain the machines, and process the results inputted by voters. It is virtually impossible to find anyone without a preference or bias for or against a particular candidate so the possibility of fudging or hacking is always going to be present if people have anything to do with the process. Sadly, there is no way to eliminate people from the processing of votes, so one can never be fully confident of untainted vote totals. Give anyone access to vote tallies and the temptation is too great.
Claudius
2.3 / 5 (3) Nov 07, 2012
@LariAnn

What you say is a problem that has to be overcome. Is it possible?

If we do not reform the voting system (I like your choice of the word "scheme") there is no hope of ever solving our political problems.

If paper ballots were used, and on the precinct level concerned citizens were on hand to count all the votes and come to an agreement that the count was accurate, and then those same citizens went to the next higher level to report their count, and ensure that the count was added to the other precincts reporting (with their own groups of citizens), with all concerned certifying that an accurate count was made, and with this process continuing all the way to the final count, with a chain of custody supervised by citizens, I believe we could have honest elections.

(cont.)
Claudius
2.3 / 5 (3) Nov 07, 2012
As it stands now, your vote disappears into the machine, where it may be added to the totals for a different candidate. Or, when it goes to the higher level computer, that computer may be programmed to adjust vote totals to provided a desired result. National votes are sent to a company in Spain run by former Goldman-Sachs employees where the final tallies are made and then reported back to us in this country. Citizens are not allowed to participate at any of the critical steps of this "scheme."

Obviously, things not only can be improved, they must be improved. It is only a persistent problem because we are doing nothing about it.
kochevnik
3.7 / 5 (3) Nov 07, 2012
Give anyone access to vote tallies and the temptation is too great.
List the votes on a website with pgp signatures. This is done everywhere except in elections, because the machine prefers fraud. The powers have a vested interest in blocking simple cryptographic solutions.
Claudius
3 / 5 (4) Nov 07, 2012
For those who doubt our votes are counted in Spain, see this:
http://www.wnd.co...e-fears/

"In January, SCYTL, based in Barcelona, acquired 100 percent of SOE Software, the leading software provider of election management solutions in the United States."
ValeriaT
1 / 5 (1) Nov 07, 2012
It seems, we could replace the election with statistical models soon...
VendicarD
3.7 / 5 (3) Nov 07, 2012
Mindless Conservative mouthpiece George Will predicted a Ronmey landslide, claiming that he would receive over 340 electoral college votes.

That shows you how far out of touch with reality these ConservaTard pundants are.
ForFreeMinds
1.5 / 5 (2) Nov 09, 2012
While Nate Silver's model was accurate this time, other models were not. And the article fails to examine the models that failed this time. I recall one model from a university in Colorado predicted a Romney win. The article fails to examine why some models worked, while others didn't.

I have more faith in the people who actually put their own money on the line, i.e. the gamblers.
antialias_physorg
5 / 5 (1) Nov 09, 2012
The article fails to examine why some models worked, while others didn't.

Statistics have a confidence interval. Even the best statistic will fail occasionally. That doesn't mean that it doesn't 'work'. It's a STATISTIC (i.e. a measure of PROBABILITY) - not an algorithm (i.e. something that will always give a CERTAIN output).

So with statistics you don't have the works/doesn't work dichotomy. You have the works more often/works less often dichotomy.

I have more faith in the people who actually put their own money on the line, i.e. the gamblers.

I'd pick a statistician over a gambler any day. In the long run the statistician will win. Always. That's why casinos are run by statisticians and not gamblers.
FrankHerbert
1 / 5 (1) Nov 09, 2012
I have more faith in the people who actually put their own money on the line, i.e. the gamblers.


Nate Silver started in sports odds. He is successful precisely because he doesn't have a partisan bend.

The great portion of his model is simply weighting the polls based on their past performance and basic electoral math.
ReallyWannaKnow
1 / 5 (2) Nov 11, 2012
Obviously, things not only can be improved, they must be improved. It is only a persistent problem because we are doing nothing about it.

Given that the count can't be trusted, nor the counters, then what to do?

More news stories

IBM posts lower 1Q earnings amid hardware slump

IBM's first-quarter earnings fell and revenue came in below Wall Street's expectations amid an ongoing decline in its hardware business, one that was exasperated by weaker demand in China and emerging markets.

Microsoft CEO is driving data-culture mindset

(Phys.org) —Microsoft's future strategy: is all about leveraging data, from different sources, coming together using one cohesive Microsoft architecture. Microsoft CEO Satya Nadella on Tuesday, both in ...

Quantenna promises 10-gigabit Wi-Fi by next year

(Phys.org) —Quantenna Communications has announced that it has plans for releasing a chipset that will be capable of delivering 10Gbps WiFi to/from routers, bridges and computers by sometime next year. ...

Down's chromosome cause genome-wide disruption

The extra copy of Chromosome 21 that causes Down's syndrome throws a spanner into the workings of all the other chromosomes as well, said a study published Wednesday that surprised its authors.

Researchers see hospitalization records as additional tool

Comparing hospitalization records with data reported to local boards of health presents a more accurate way to monitor how well communities track disease outbreaks, according to a paper published April 16 in the journal PLOS ON ...

Ebola virus in Africa outbreak is a new strain

The Ebola virus that has killed scores of people in Guinea this year is a new strain—evidence that the disease did not spread there from outbreaks in some other African nations, scientists report.