Thin and effeminate statistics

Written By: - Date published: 4:05 pm, November 8th, 2012 - 25 comments
Categories: science, uk politics - Tags: , ,

Here’s another entry in the Very Important Columnist vs Parasitic Bloggers file. While the professional mainstream media were all chanting in unison that the American presidential election was “too close to call”, one (famous!) blogger correctly predicted the result – for every state. The Guardian reports:

 So who is he? … A statistical analyst of elections and a former statistical analyst of baseball. His blog, FiveThirtyEight, now hosted by the New York Times, is the go-to site for cold, hard election stats.

…  Silver didn’t just predict Obama was going to win it. He correctly projected the results for every single one of America’s 50 states. Something 50 tossed coins would have less than a one in a trillion chance of getting right.

Ah, well. That’s a little more impressive. How’d he do it? With his “election simulator” statistical model, which aggregates hundreds of polling figures and economic data, weights it all for accuracy, factors in a load of fancy maths about past elections and boils the whole thing down to a percentage chance of victory for each candidate. On Tuesday morning, while many pundits were insisting it was too close to call, Silver’s model put Obama’s chances at a precise and healthy 90.9%.

How did pundits feel about that? Pretty uncomfortable. He was derided in the media as “a joke”, who was “getting into silly land”. One rightwing blogger even dismissed his predictions on the grounds that Silver – an openly gay Democrat – was “thin and effeminate”.

(Today’s xkcd is also on the case.)

In 2008 Silver correctly predicted 49 / 50 states. So, thin and effeminate statistics 2, mainstream media 0. Political punditry, at least in America, is going to have to lift its game…

Just for the record, Nate Silver’s predictions:

The result:

25 comments on “Thin and effeminate statistics ”

  • Comments are now closed

  1. Idiot/Savant 1
    8 November 2012 at 4:12 pm

    There can be only one explanation for such accurate predictions: Science! Witchcraft.

  2. kea 2
    8 November 2012 at 4:13 pm

    Yep, I’ve been following Nate during this election with great interest – if anyone’s interested, follow his rss feed.

    Cheers ..

    • mickysavage 2.1
      8 November 2012 at 7:37 pm

      Nate is the anti Fox news.

      Those guys were still trying to say that at least Romney had won the popular vote when only a quarter of California’s votes has been reported in …

      Obama is currently winning the popular vote by 60.6 million to 57.8 million …

  3. Zorr 3
    8 November 2012 at 4:16 pm

    I think we’re going to have to rethink the old addage “Lies, damned lies and statistics” because this guy seems to be proving the worth of the field…

    • Matthew Whitehead 3.1
      9 November 2012 at 7:06 am

      That’s not what the adage refers to.

      http://en.wikipedia.org/wiki/Lies,_damned_lies,_and_statistics

      It’s referring to the fact that lies that pretend to be statistics are the worst and hardest to debunk of all three types. 😉 Nate Silver’s statistics were honest.

    • RedLogix 3.2
      9 November 2012 at 8:19 am

      Yes. Matthew nails it correctly.

      It’s very easy to dress up a lie as data and statistics. Sadly far too many people have no real idea how these things work, and how very powerfully statistics can extract real and valuable meaning from difficult, noisy and incomplete information.

      In the real world data is very rarely clean, complete and noise-free. There are almost always other variables in play that taint or skew the raw data. Good statistics when done well (as this guy does ) can yield remarkable and powerful insights.

      Unfortunately it’s also very easy for people to hijack this power to lend credibility to their lies by dressing them up as statistics. And because most people quite understandably have little to no skill in the subject, makes them very hard to debunk effectively. Most people just shrug their shoulders and mutter the ‘damn lies’ line to themselves.

  4. tracey 4
    8 November 2012 at 4:31 pm

    Perhaps because he isnt manipulating stats he is collecting them and genuinely analysing them for numerical not ploitical trends. As long as people like farrar, hooten and mccarten are relied upon for political analysis we can hardly expect a scientific outcome.

  5. Rich 5
    8 November 2012 at 4:54 pm

    Only three or four states actually ever change hands, right? And they aren’t independent variables – if many Nevadans swing left, then so might Coloradans and Floridese. So it’s maybe more like a 1:5 chance of getting a prediction right.

    With 57 million blogs on WordPress alone, I’d think that a very large number of random bloggers would get the election result right in every state, if they bothered to make a call.

    • kea 5.1
      8 November 2012 at 5:12 pm

      One of the things Nate does is to repeatedly re-run many models, each tweaked to cover various possibilities, and his ‘predictions’ are the aggregates of the results. The published photo I saw was of him sitting in what looked like a suburban hut with a laptop – although it may have been connected to a server via fibreoptic, who knows. I don’t know what program he was using, but it is probably a combination of baseball number crunching experience and contemporary computing power ..

  6. kea 6
    8 November 2012 at 5:32 pm

    I have to give credit to someone on this blog who put me onto it. Can’t remember who ..

  8. gobsmacked 8
    8 November 2012 at 5:50 pm

    “Hello, welcome to Close-Up Sunday Breakfast on TVNZ.

    Today we’re going to discuss facts and figures. Do they matter?

    Here to tell us are some Experts, all of whom you know well, because they’re appearing in all media every week. There’s 1) a washed-up former MP who lost his seat, 2) a grumpy old git who copys his column from Wikipedia, and 3) a token woman who likes to say “Look”.

    So panel, how about Nate Silver? Worth a whole lot more than anything you ever have to say, wouldn’t you agree?

    1) “Well, I’ve got a feeling about this election, a real sense that something is happening, and it’s come fresh from my arse. So don’t bother me with all these ivory tower numbers – get back in the real world!”

    2) “At the end of the day, who can you believe? Statistics, at the end of the day, are just something spouted by bloggers. I remember Muldoon once said to me …

    3) “Look …

    (audience chucks coffee cup at TV, goes back to bed)

    • Anne 8.1
      8 November 2012 at 6:49 pm

      (audience chucks coffee cup at TV, goes back to bed)

      It’s true. I can vouch for it.

    • Colonial Viper 8.2
      8 November 2012 at 8:00 pm

      That show worsens my hangover. Which is saying something.

    • Tracey 8.3
      9 November 2012 at 8:17 am

      for Gobsmacked NOT the show

    • QoT 8.4
      9 November 2012 at 6:32 pm

      Holy shit, you totally called Firstline’s coverage this morning. Some dude excusing the blatantly biased coverage/spinning of polls in the US as being about Nate Silver being some kind of magically advanced wizard, and “oh people just aren’t used to treating statistics like this” … as opposed to, you know, corporations with obvious agendas telling their anchors how to talk down polls they don’t like.

  9. kea 9
    8 November 2012 at 6:40 pm

    That’s about right ..

  10. mike e 10
    8 November 2012 at 7:52 pm

    Imagine what key would have said if he had been wearing red

  11. Brett Dale 11
    8 November 2012 at 10:49 pm

    Is there a Nate Silver type person In New Zealand???

    He is brilliant.

  12. SukieDamson 12
    9 November 2012 at 7:56 am

    Once upon a time. His name was Marty G.

  14. kea 14
    9 November 2012 at 3:47 pm

    For the psephologists and election junkies

    An extended video interview with Nate Silver.

    http://www.thedailyshow.com/watch/wed-october-17-2012/exclusive—nate-silver-extended-interview-pt–1

The server will be getting hardware changes this evening starting at 10pm NZDT.
The site will be off line for some hours.