Support NeoGAF

[Journeywalker]

Why is Nate Silver in the spotlight this year? I missed out on that.

Apparently, when some people don't like the results of their spreadsheets, their first thoughts are:

1. Microsoft must have made Excel biased
2. Someone must have royally screwed up on data entry
3. The numbers don't matter.

And certain operatives are all too keen to exploit that thinking.

 

[Dash27]

I thought he did well in 2010

Not for US congressional seats or the UK elections. Unless there's something else I'm unaware of he didnt do well at all compared with other predictions.

 

[Cyan]

I'll agree with you on the marijuana issue but i still cant get over the Nate Silver humping here and on the left in general. And links to dailyKoS to back it up to boot. It's like saying Romney will win because I heard on Hannity this week that...

No, no it's really not. Unless I missed Hannity coming out with a statistical model in 2008 that predicted the election more accurately than anyone else's?

 

[B-Dubs]

If you want Obama to win, he's the guy telling you there's a 70% + chance it will happen.

He did well in 2008, not so much in 2010.

He did well in 2010, stop rewriting history.

 

[Kusagari]

Not really. You get Ohio and something else, who cares really, Colorado and Virginia or something and it's over anyway. Florida doesn't even really have to come into the equation just to reach 270.

I meant that FL gets called. Hopefully Obama easily wins Ohio and it's a wrap anyway.

 

[Amir0x]

I'll agree with you on the marijuana issue but i still cant get over the Nate Silver humping here and on the left in general. And links to dailyKoS to back it up to boot. It's like saying Romney will win because I heard on Hannity this week that...

It's simple statistics. There is no 'yes' or 'no.' No 'right' or 'wrong.' That's a pundits foul game. The reason we love Nate Silver is because he cuts through the bullshit relating to people's "gut feelings" or Republican or Democratic spin, and simply says "this is the state of the race, currently, in all probability."

Nate Silver being an Obama supporter has nothing to do with it. He's not adding that fact into his predictive model. If his predictive model told him Obama was going to lose, he'd post it.

And he does: His model says Obama will lose 16% of the time.

 

[Brinbe]

Such nonsense, comparing Silver to f'n Hannity?

 

[Dash27]

He did. It's Dash. (And it makes sense -- if you think Romney will win because it's going to be 2010 again, you need to get around the fact that 538 showed a large Republican victory in 2010 and shows the opposite today. Easiest way, as always, is lying.)

He did well in 2010, stop rewriting history.

He didnt:

http://www.fivethirtyeight.com/2010/05/uk-forecasting-retrospective.html

Our UK forecasting model, which tried to improve upon the deficiencies inherent in uniform swing, performed underwhelmingly. We went out on something of a limb here, and sometimes when you do that, the limb breaks!

http://blog.lib.umn.edu/cspg/smartpolitics/track-record.php

For example, in the 2010 election cycle, Smart Politics' U.S. House race projections were more accurate and precise than any of the go-to national prognosticators, predicting a net GOP pick-up of 62 seats - just one shy of their final tally and more accurate than Larry Sabato (55 seats), Nate Silver (54 seats), Rasmussen Reports (55 seats), Charlie Cook (50-60 seats), and Stuart Rothenberg (55-65 seats).

 

[Black Republican]

lol redskins

 

[HylianTom]

Such nonsense, comparing Silver to f'n Hannity?

It's just the beginning. Watch.

Bill Clinton said at the convention that a key concept for this election was arithmetic. Hmm.

 

[Stumpokapow]

He did well in 2008, not so much in 2010.

538, 2010:
- Senate Projection: GOP pickup 7. Actual: GOP pickup 6. 34 out of 37 races correctly predicted, of which 2 were the two closest senate races (within margin of error) and 1 was actually dead wrong.
- House Projection: GOP pickup 54/55. Actual: GOP pickup 63.
- State Gubernatorial Projection: 36 out of 37 correctly predicted, the 1 miss was closest gubernatorial race in country

This knowing full well that his model works better the more polling there is, and there's relatively less polling in off-years and house races. What a terrible record.

 

[syllogism]

For house race projections a good argument can be made that pure poll aggregating and modelling is not enough due to there not being enough high quality public polling and people like Cook and Sabato can be more accurate. There are just too many races and they are also more difficult to poll than state or nationwide races. Anyway, the only rational argument against nate is that what he does isn't that complicated; indeed it isn't. Other than that all the attacks are irrational and clueless as what he does is almost entirely pure math.

 

[KHarvey16]

Are you saying prediction models that rely mostly on polling information do poorly in elections with less polls and of which most are of lesser quality? I am shocked.

 

[cousins]

538, 2010:
- Senate Projection: GOP pickup 7. Actual: GOP pickup 6. 34 out of 37 races correctly predicted, of which 2 were the two closest senate races (within margin of error) and 1 was actually dead wrong.
- House Projection: GOP pickup 54/55. Actual: GOP pickup 63.
- State Gubernatorial Projection: 36 out of 37 correctly predicted, the 1 miss was closest gubernatorial race in country

This knowing full well that his model works better the more polling there is, and there's relatively less polling in off-years and house races. What a terrible record.

He was referring to Silver's UK predictions, which played out atrociously.

 

[Amir0x]

Are you saying prediction models that rely mostly on polling information do poorly in elections with less polls and of which most are of lesser quality? I am shocked.

what that sounds like some sort of black magic lifted from the festering bowels of hell

you're a witch. Witch!

 

[The Lamp]

I'm a busy student so I haven't had much time to keep up with Romney and Obama myself in the past 6 months. Are there any good sources I can "crunch" on between now and November 6th to learn more about their positions, promises, etc. so I can make the most informed decision on who to vote for by the 6th?

This is my first election and it's really tough because I really don't like either candidate at all. I think I used that Iallignwith.com or whatever site earlier this year and found that I most align with Ron Paul's views but he's out now. But I have to make a decision. My vote probably doesn't even count anyway, though, because I live in Texas and many people I know are voting for Romney...

EDIT: missed that the blue lines at the top were actually links to that XD

 

[Snaku]

It's simple statistics. There is no 'yes' or 'no.' No 'right' or 'wrong.' That's a pundits foul game. The reason we love Nate Silver is because he cuts through the bullshit relating to people's "gut feelings" or Republican or Democratic spin, and simply says "this is the state of the race, currently, in all probability."

Nate Silver being an Obama supporter has nothing to do with it. He's not adding that fact into his predictive model. If his predictive model told him Obama was going to lose, he'd post it.

And he does: His model says Obama will lose 16% of the time.

He's like a real life Ian Malcolm.

 

[KHarvey16]

what that sounds like some sort of black magic lifted from the festering bowels of hell

you're a witch. Witch!

A pox on you! You'll never get me!

 

[Piggus]

I think the good guys could sweep marriage this year. WA and ME look solid, Obama's endorsement gave it a boost in MD and PPP's poll of MN gives me hope.

This is something of a final frontier for the anti-marriage folks, so we'll see how that works out for them.

Oregon will apparently put it on the ballot in 2014. Washington passing it now would hopefully give it a big boost.

 

[Journeywalker]

For house race projections a good argument can be made that pure poll aggregating and modelling is not enough due to there not being enough high quality public polling and people like Cook and Sabato can be more accurate. There are just too many races and they are also more difficult to poll than state or nationwide races. Anyway, the only rational argument against nate is that what he does isn't that complicated; indeed it isn't. Other than that all the attacks are irrational and clueless as what he does is almost entirely pure math.

There are other reasons that Nate Silver is trusted:

If he's wrong, he'll straight-up tell you. There's no obfuscating with him. In fact, he'll explicitly tell you how wrong he was, and possible reasons for why he was wrong.

In fact, I think he recognized that in 2010, his model was middling in predicting House seats due to the volatility in these races, the difficulty of polling, and the small number of polls for each race. I suspect that's why he's not publishing predictions for the House this year. Critics may try to claim that he is trying to hide a deficiency in his model, but I view it as humbly accepting its limitations.

 

[SouthernDragon]

He was referring to Silver's UK predictions, which played out atrociously.

True enough, but he's done well with all American elections thus far.

 

[gutter_trash]

UK has 3 major parties and a couple of lesser parties. A multi party election in a parliamentary system is harder to predict correctly with a 1st past the post rules then a 2 party Presidential system like the US

 

[Alpha-Bromega]

man i'm so conflicted, i agree wholeheartedly with Stein but don't want a spoiler vote. Obama at his worse is better than Romney at best, triple cliche but true imo

what do people like us do? throw in the symbolic vote? vote for the lesser evil?

 

[Effect]

When i poll-watched in PA the guidance was that everyone in line as of the time if closes is entitled to stay and vote - the place has to stay open until they've all voted.

The law may be different elsewhere.

When those lines are still several hours long in some states that's going to make this very interesting. There aren't enough police to calm people down if the crowds turn angry.

 

[SteveWinwood]

man i'm so conflicted, i agree wholeheartedly with Stein but don't want a spoiler vote. Obama at his worse is better than Romney at best, triple cliche but true imo

what do people like us do? throw in the symbolic vote? vote for the lesser evil?

Vote for who you want on a local or state level where you have more power and vote for who has a chance when you get to the federal level.

Might change in the future, but that's the best you can do right now without fucking yourself over.

 

[Clevinger]

There are other reasons that Nate Silver is trusted:

If he's wrong, he'll straight-up tell you. There's no obfuscating with him. In fact, he'll explicitly tell you how wrong he was, and possible reasons for why he was wrong.

And the pundits who are shitting on him right now, like Joe Scarborough, will never admit to being wrong. They'll just move on to the next bullshit narrative.

man i'm so conflicted, i agree wholeheartedly with Stein but don't want a spoiler vote. Obama at his worse is better than Romney at best, triple cliche but true imo

what do people like us do? throw in the symbolic vote? vote for the lesser evil?

Depends on where you are. If you're in a swing state and you're worried, go Obama. Otherwise, vote with your conscience if that's what you want.

 

[B-Dubs]

There are other reasons that Nate Silver is trusted:

If he's wrong, he'll straight-up tell you. There's no obfuscating with him. In fact, he'll explicitly tell you how wrong he was, and possible reasons for why he was wrong.

In fact, I think he recognized that in 2010, his model was middling in predicting House seats due to the volatility in these races, the difficulty of polling, and the small number of polls for each race. I suspect that's why he's not publishing predictions for the House this year. Critics may try to claim that he is trying to hide a deficiency in his model, but I view it as humbly accepting its limitations.

His models works when there is a lot of polling, when the amount of polling is very small it's not so good. I agree with your assessment, the fact is there is not a lot of polling done on house races which means not as much data which means it's harder to spot the trends. Math is all data data data. The more you've got the better.

 

[Dany]

the potential danger Romney could do to international relations is scaring me the most about tuesday.

 

[Rocket Scientist]

No, no it's really not. Unless I missed Hannity coming out with a statistical model in 2008 that predicted the election more accurately than anyone else's?

I can't believe he's equating Silver to Hannity... I... WHAT?!

 

[HylianTom]

 

[Bel Marduk]

the potential danger Romney could do to international relations is scaring me the most about tuesday.

Don't be scared. Romney has less than a 1/5th chance of winning.

 

[Rocket Scientist]

Wish there was a site which had good quality pictures of voting lines. Need to post on FB.

 

[pigeon]

Matthew Dowd told ABC News that every time "you feel a losing campaign, these three things happen":

"The first thing happens is, don't believe -- the public polls are wrong. That's the first sign of a campaign that's about to lose. The second thing, we're going to change the nature of the electorate, and you're not seeing it reflected in the polls. And the third thing is, the only poll that counts is Election Day. When you hear those things, you know you're about to lose."

http://politicalwire.com/archives/2012/11/04/how_you_know_youre_going_to_lose.html

Mittmentum.

 

[Dany]

Don't be scared. Romney has less than a 1/5th chance of winning.

Hold me! D: You better be right. Then why the hell are these polls saying otherwise?!

 

[Majine]

Don't be scared. Romney has less than a 1/5th chance of winning.

Media says it's close.

I'm still worried.

 

[Ravensmash]

Media says it's close.

I'm still worried.

Isn't the media basing this on popular polls and not the electoral college votes or whatever ?

 

[Hawkian]

true story: you can win the election with 22% of the popular vote

 

[pigeon]

Media says it's close.

I'm still worried.

Would you tune in if they said it was already over?

 

[Takamura-San]

I'm trully scared about tuesday.

I wouldn't put it past the american people that they would vote romney for president.

I which I knew which romney they think they are voting. There are so many of them.

 

[Xeke]

Media says it's close.

I'm still worried.

Media says what will bring in ratings.

 

[SteveWinwood]

Goddammit, I was going to make some money on some intrade bets that looked pretty good (some of those electoral votes ones are easy money) but I have to send a wire and it won't clear before tuesday

Fuck I should have planned ahead.

 

[ViewtifulJC]

Media says what will bring in ratings.

Romney's America...

 

[Brinbe]

Media says it's close.

I'm still worried.

LOL

 

[Bel Marduk]

Hold me! D: You better be right. Then why the hell are these polls saying otherwise?!

Aww.. well, as I'm sure you know, its not a direct popular vote but an electoral college system in the U.S. Obama has enough states wrapped up to surpass the needed 270 electoral votes on Tuesday. The key state there is Ohio, where he has lead in nearly every poll. As long as he has Ohio (which I can assure you he will) he wins

Most of the media will tell you that its a close race because that makes for a more interesting story than President will still be President.

 

[maquiladora]

Patricia Mazzei ‏@PatriciaMazzei

With 15 minutes before closing, the line outside the Miami-Dade elections HQ is about 500 people deep.

.

 

[Brinbe]

 

[Karma Kramer]

Whoever wins , America loses, because its another 4 years of corporate government, shifting more power to a few instead of back to the people.

 

[Puddles]

His models works when there is a lot of polling, when the amount of polling is very small it's not so good. I agree with your assessment, the fact is there is not a lot of polling done on house races which means not as much data which means it's harder to spot the trends. Math is all data data data. The more you've got the better.

This is basically a rule of thumb for statistics. When n is large, the standard error is lower, and the confidence interval is narrower.

 

[GremlinFool]

Whoever wins , America loses, because its another 4 years of corporate government, shifting more power to a few instead of back to the people.

You sound like a broken record.

 

[ivysaur12]

Liberal Xanax.

Whoever wins , America loses, because its another 4 years of corporate government, shifting more power to a few instead of back to the people.

At least try to say something different every once in a while.