Stat geek forecasts are Election 2012 winners

Statistical analysts were right; most media pundits were wrong.

And it wasn't even close.

This year's presidential election proved to be the comfortable Obama Electoral College victory predicted by Nate Silver and other stat geeks, not the "too close to call" that many in the mainstream media had declared.

With Florida yet to be called, Obama had 303 electoral votes to Governor Romney's 203. That means it'll either end up 332-203 or 303-232.

Here's what statisticians were predicting one day before the Election:

Nate Silver at FiveThirtyEight: Obama 315.3 to 222. Update: In his state-by-state predictions, Silver went 50 for 50 and is 99 out of 100 in the last two US presidential elections. 

Sam Wang at Princeton Election Consortium: Obama 309-229. "In the races called thus far, pre-election polling medians were correct in 50 out of 50," he notes.

Election Projection: Obama 303-235

Votamatic: Obama 326-212 (that was changed to 332-206 on Tuesday morning).

Those numbers were pretty constant from the prior week. Yet at the same time, Politico was reporting: "Media stumped by 2012 outcome."

More specifically:

"Anyone who claims to know who is going to win is blowing smoke." -- pundit Joe Klein

"I'm completely confused. I have no idea who's going to win." ABC News Jon Karl

"Anybody that thinks that this race is anything but a toss-up right now is such an ideologue, they should be kept away from typewriters, computers, laptops and microphones for the next 10 days, because they're jokes." -- Joe Scarborough, MSNBC

"People are saying it's too close to call because it is." -- Bill Kristol, the Weekly Standard

Except it wasn't.

Nobel prize-winning economist Paul Krugman was correct: Reporting that claimed the race was tied "makes you stupider."

So here are a couple of takeaways on how to be better election-information consumer:

* One poll showing that Candidate X has moved ahead of Candidate Y when Candidate X had been behind does not necessarily mean that the race is shifting. It could mean the new poll is an outlier. As Silver tweeted recently, "The answer to 90% of poll-related questions is 'it's probably random variance and you shouldn't make too much of it.''' Anyone who reports on the latest poll as if it's the best piece of information on an election only because it's new should not be doing stories that involve data. 

* Polls may not be predictive six months before an election; but six days before an election, pay attention to statisticians who are aggregating and analyzing public opinion polls.

* TV talking heads may provide interesting insights into strategies and moods within a campaign; they may indeed also be entertaining. But if the pundits say one thing and the stat geeks say another, do not ignore the geeks.

Computerworld's IT Salary Survey 2017 results
Shop Tech Products at Amazon