Nate Silver Say... Everything Nate Silver goes here

[deaddogseye]

Nate Silver isn't quite perfect. He miscalled the Tester/Rehberg race in Montana. He gave Rehberg a 65.5% chance of winning, but Tester won by almost 4 points. To cut him some slack, there wasn't a whole lot of polling done there.

he aslo blew berg in ND. That one shows an unavoidable weakness in his methodology. All the recent polls showed him ahead so there was no way (i think n)n for the methodology to show anyone winning but him -- there was no data indicating anything else and of course he lost.

on tester -- you are right and there wasnt a lot of polling but the last pre election poll showed tester winnig (see RCP) so there was data out there to show a Tester win. I has Tester winning (although I was wrong on ND)

 

[deaddogseye]

well -- yes he had it as you say tuesday morning but then later that morning before the polls closed he pulled it back and put it 50/50 and called it a tossup -- and thats how it still is now - check it out so i think he didnt call it right. I think im right on this.

Provide a link. I'm looking at his predictions here, and he currently has Obama with a 50.3% chance of winning Florida. He doesn't do things like 50/50 that I recall.

Election Forecasts - FiveThirtyEight Blog - NYTimes.com

ok -- this is weird. I was going to send you a link but i clicked on your link and it shows (me at least) exactly what i said. Scroll down the left side to the tables of state and look at florida? doesnt it say exactly what I say?

Im sorry - i mean scroll down the RIGHT side of the page

 

[Dick Tuck]

well -- yes he had it as you say tuesday morning but then later that morning before the polls closed he pulled it back and put it 50/50 and called it a tossup -- and thats how it still is now - check it out so i think he didnt call it right. I think im right on this.

Provide a link. I'm looking at his predictions here, and he currently has Obama with a 50.3% chance of winning Florida. He doesn't do things like 50/50 that I recall.

Election Forecasts - FiveThirtyEight Blog - NYTimes.com

ok -- this is weird. I was going to send you a link but i clicked on your link and it shows (me at least) exactly what i said. Scroll down the left side to the tables of state and look at florida? doesnt it say exactly what I say?

I see what you're saying, and looking there, he's rounding all those final state predictions. But if you look at his data points, he still has a .2% advantage to Obama, in the Adjusted polling average, and that tiny advantage is reflected in his map.

 

[deaddogseye]

Provide a link. I'm looking at his predictions here, and he currently has Obama with a 50.3% chance of winning Florida. He doesn't do things like 50/50 that I recall.

Election Forecasts - FiveThirtyEight Blog - NYTimes.com

ok -- this is weird. I was going to send you a link but i clicked on your link and it shows (me at least) exactly what i said. Scroll down the left side to the tables of state and look at florida? doesnt it say exactly what I say?

I see what you're saying, and looking there, he's rounding all those final state predictions. But if you look at his data points, he still has a .2% advantage to Obama, in the Adjusted polling average, and that tiny advantage is reflected in his map.

Thanks

 

[Dick Tuck]

Nate Silver isn't quite perfect. He miscalled the Tester/Rehberg race in Montana. He gave Rehberg a 65.5% chance of winning, but Tester won by almost 4 points. To cut him some slack, there wasn't a whole lot of polling done there.

he aslo blew berg in ND. That one shows an unavoidable weakness in his methodology. All the recent polls showed him ahead so there was no way (i think n)n for the methodology to show anyone winning but him -- there was no data indicating anything else and of course he lost.

on tester -- you are right and there wasnt a lot of polling but the last pre election poll showed tester winnig (see RCP) so there was data out there to show a Tester win. I has Tester winning (although I was wrong on ND)

The thing is, he's not a pollster or a pundit. He puts the data he gets in, weighs it for the past accuracy of the specific pollster, and what comes out comes out. When working with a small number of polls, he'll be less accurate, and with more data, his accuracy improves.

The bottom line remains, his method is very accurate, and his objective predictions are better than anyone else out there doing similar things. It's certainly better than those who tried to weigh their results on a 50-50 split in voter party id, like unskewedpolls and Rassmussen, or Gallup who used 2010 numbers to weigh their race demographics for their likely voter algorithm.

 

[deaddogseye]

Nate Silver isn't quite perfect. He miscalled the Tester/Rehberg race in Montana. He gave Rehberg a 65.5% chance of winning, but Tester won by almost 4 points. To cut him some slack, there wasn't a whole lot of polling done there.

he aslo blew berg in ND. That one shows an unavoidable weakness in his methodology. All the recent polls showed him ahead so there was no way (i think n)n for the methodology to show anyone winning but him -- there was no data indicating anything else and of course he lost.

on tester -- you are right and there wasnt a lot of polling but the last pre election poll showed tester winnig (see RCP) so there was data out there to show a Tester win. I has Tester winning (although I was wrong on ND)

The thing is, he's not a pollster or a pundit. He puts the data he gets in, weighs it for the past accuracy of the specific pollster, and what comes out comes out. When working with a small number of polls, he'll be less accurate, and with more data, his accuracy improves.

The bottom line remains, his method is very accurate, and his objective predictions are better than anyone else out there doing similar things. It's certainly better than those who tried to weigh their results on a 50-50 split in voter party id, like unskewedpolls and Rassmussen, or Gallup who used 2010 numbers to weigh their race demographics for their likely voter algorithm.

I agree with you on this -- althogh i do maintain that what he does isnt that hard to do and most people who just eyeballed the state poll tables on RCP on a fairly regular basis would come up with the same conclusions on their own.

He deserves credit for making a little career out of this but it isnt hard to do -picking the winners of the states anyway -- at least IMO.

 

[JimH52]

Republicans at a real disadvantage with the Electoral College system

Election Forecasts - FiveThirtyEight Blog - NYTimes.com


OOOOUUUUCCHHHHH That's got to hurt!

 

[JoeB131]

will someonme explain something to me -- what is the basis for saying silver got 50 states right?

On the morning of election day he had florida at 52 / 48 for Obama (earlier he had it for romney i believe) BUT by noon before the polls closed he had it 50/50 and labled it "Tossup" (and that is still on the site now). So he didnt make a call on it as of the end and really punted it. So he wasnt wrong but he also wanst right as far as I could tell.

So if you count DC he did get 50 elctoral jurisdictions right (or even more counting the seprate distircts in Maine in Nebraska) but he did not get all of them right as far as I can as far as I can tell.

Am i missing something and if i am where is his call on Florida on the site because what I see doesnt indicate obama. Thanks

I must have missed where he had it 52/48, and he called it a tossup.

 

[Dot Com]

It's fun watching a dying snake writhe on the ground.

It is...

Please be here wednesday...

Its almost Friday now

 

[Synthaholic]

Link?

I call bullshit.

Good for you! Silver was on a TV show explaining what he does a while back. It was interesting. He's a really nice guy that likes to gamble. I don't care if your believe me or not.

Here is what he says about how he weights polls at his NY Times blog:

"a combination of freshness, sample size, past accuracy, and methodological standards".

I see plenty of room for human factors in that explanation.

Bottom line- He's a gambler. If he's right- he'll make a bloody FORTUNE. If he's wrong, he'll make a tiny little less. He can't walk away from the pot odds...he wins either way.

You're awfully quiet tonight? What happened?

He's been busy converting all his money into gold, and buying ammo.

 

[Synthaholic]

Whoa. Ohio just turned red. Won't change the election, but does change the numbers.

Nope:

Ohio Election Results

 

[Zander]

Nate made a good bet....

 

[EriktheRed]

Nate made a good bet....

Good on you, sir.


Now where's the other shit-talker whose bet featured prominently in this thread...

 

[elvis]

Whoa. Ohio just turned red. Won't change the election, but does change the numbers.

Nope:

Ohio Election Results

No one has a crystal ball, synth.

 

[Procrustes Stretched]

We all know polls are skewed Democratic. The liberal media again helped elected Obama. Life is unfair

boo hoo hoo

 

[konradv]

#drunknatesilver has taken over the Twitterverse. Funny article. Here's a sample.

Drunk Nate Silver knows what you did next summer. #DrunkNateSilver

Twitter commandeered by 'Drunk Nate Silver' - CNN.com

 

[candycorn]

It is going to be hilarious watching you morons howl in pain.

he he he

 

[JoeB131]

Nate made a good bet....

No, Nate conducted solid analysis based on data.

 

[Political Junky]

Nate made a good bet....

No, Nate conducted solid analysis based on data.

Yes, as did the Obama campaign. Romney's campaign should have done that too, and Willard wouldn't have been shell-shocked.

 

[Cowman]

I wonder, will conservatives mock Nate Silver next election? Or will they actually trust his data?