You're gonna cite Dailykos diarists as a source?

[triggercut1]

...well, yes.

All primary season, A Dkos blogger under the handle of "Poblano" has been kicking the virtual butts of every pollster in the country. On the eve of elections he's been more accurate than any other pundit in the country. James Wolcott at Vanity Fair has given him a shoutout. The National Journal has remarked on his methods. The Field has praised his accuracy up and down, and added Poblano's political blog, Five-Thirty-Eight.com to his blogroll. Poblano uses a standard regression statistical model with math that sweeps right over my head, but basically takes a bunch of known statistical facts and then plugs them in to get a prediction.

Sound familiar?

It should. Today "Poblano" stepped out from behind the curtain. He's Nate Silver, the creator of PECOTA and of course a writer/Managing Editor at Baseball Prospectus. What he did for projecting Democratic Primary elections was in a sense very little different from what he does at BP--he plugs in data and uses statistical analysis to make a projection.

I just thought it was exceedingly cool that a seamhead was showing the soundness of statistical regression analysis in a milieu other than baseball.

[007]

...well, yes.

All primary season, A Dkos blogger under the handle of "Poblano" has been kicking the virtual butts of every pollster in the country. On the eve of elections he's been more accurate than any other pundit in the country. James Wolcott at Vanity Fair has given him a shoutout. The National Journal has remarked on his methods. The Field has praised his accuracy up and down, and added Poblano's political blog, Five-Thirty-Eight.com to his blogroll. Poblano uses a standard regression statistical model with math that sweeps right over my head, but basically takes a bunch of known statistical facts and then plugs them in to get a prediction.

Sound familiar?

It should. Today "Poblano" stepped out from behind the curtain. He's Nate Silver, the creator of PECOTA and of course a writer/editor at Baseball Prospectus. What he did for projecting Democratic Primary elections was in a sense very little different from what he does at BP--he plugs in data and uses statistical analysis to make a projection.

I just thought it was exceedingly cool that a seamhead was showing the soundness of statistical regression analysis in a milieu other than baseball.

Wow...that's interesting....

[triggercut1]

He did a thing predicting North Carolina with frightening accuracy a few weeks ago where he used a regression model based on "15 Most Similar Districts" and the first thing I thought of was "this sounds almost like VORP or PECOTA.

In hindsight, there were a ton of clues based on his statistical breakdowns. They were clearly not the work of some lucky amateur.

[triggercut1]

Poblano/Nate Silver's blog, btw. Expect a news network or three and perhaps even one of the two candidates to get him on payroll. He's basically Moneyballing the election.

[backstop]

a regression model based on "15 Most Similar Districts" and the first thing I thought of was "this sounds almost like VORP or PECOTA.

That's pretty much exactly what pecota does.

[El Establishment 1]

That is really cool.

[Moon Man]

Pollsters have been using regression techniques for quite some time. Though, I do find it amusing that a baseball dude was outperforming some of the experts.

Moon

[backstop]

And yeah, that's wild. First Mos Schrute is Ken Tremendous, now this.

[hansolo]

bonus points for the title of the thread.

see if it'll reel him in.

[triggercut1]

bonus points for the title of the thread.

see if it'll reel him in.

"A daily kos analysis? That's what you are relying on? Is this where you get all of your "insight"? This is just some dude sitting in his basement crunching some numbers. There's no indication at all that his analysis is even statistically relevant or applicable."

[WinstonSmith]

Who said that?

[triggercut1]

Who said that?

Nah. Just a (very mildly) heated exchange in another thread I had with someone. It wasn't the disagreement that peeved me, it was the "I'm not even going to accept the numbers and explanation of analysis of same due to where they first showed up" part of it. Understandable, I suppose--if someone posted a statistical analysis from Powerline, I'd tend the same way, although I'll note that I had to agree that the work LittleGreenFootballs and Pline did to expose the fraudulence of the Sixty Minutes letter about the President's military service was excellent and spot-on.

[hansolo]

Who said that?

think far far far 52 times.