Nate Silver @ 538: Obama has 80% Chance of Winning Nov. 6th

[Lief 🗽]

He tweeted on Friday that he's made the preliminary models for the Senate races and should have them up by next week.

[Middle-aged Europe]

And Mittens' chances of winning North Carolina are down to 58% on that site...

[mileslunn]

Well, it's a good thing Obama's win in NC wasn't just a fluke as many thought it would turn out to be. The fact that NC is competitive while VA leans Obama this year is a good sign that Democrats can remain competitive in this area of the south.

I think Obama has only a 60% chance of winning not 80%.  Now I agree by mid October if the numbers are unchanged or if he gets a strong convention bounce and it lasts for a couple of weeks then the assessment may be correct.  As for Virginia and North Carolina, they are quite different than most southern states.  Northern Virginia is culturally more Northeastern than Southern.  Much like the Philadelphia suburbs and Long Island use to vote Republican, but now vote Democrat, it is the same thing here.  If the Republicans were more moderate like they used to be, Virginia wouldn't be in play.  As for North Carolina it has a large transplant population who are urban, well educated, and not the types who buy the tea party line or the more right wing Republican views.  Lets remember the Republicans in the Northeast are more moderate than in the South, so I suspect when they see the views of the GOP after moving some transplants switch never mind the Democrats in the South are mostly Blue Dog Democrats not liberal Democrats like in the Northeast.  Georgia is a possibility down the road, although not this time around, but South Carolina lacks the transplant population that VA and NC have and is too rural.  Yes it is 30% African-American, but that group already votes over 95% Dem and in 2008 they turned out in large numbers, so unless the Dems can make inroads amongst the white population, they cannot win here and I don't see how they realistically can.  Also NC and VA have a sizeable and growing Asian and Latino population which Georgia has too, but South Carolina lacks. 

[Middle-aged Europe]

And Mittens' chances of winning North Carolina are down to 58% on that site...

Correction, down to 57% now...

[Silent Hunter]

This is likely to be the second election in a decade where the opposition has managed to miss an open goal.

[Oldiesfreak1854]

I'm sorry, President Obama does NOT have an 80% chance of winning.  Nate Silver is obviously biased toward him.  I'd say 50-50 for either one winning.

[Yank2133]

I'm sorry, President Obama does NOT have an 80% chance of winning.  Nate Silver is obviously biased toward him.  I'd say 50-50 for either one winning.

I love when people spout this nonsense, it really shows they aren't familiar at all with Silver's work.

[Craigo]

I'm sorry, President Obama does NOT have an 80% chance of winning.  Nate Silver is obviously biased toward him.  I'd say 50-50 for either one winning.

Why?

One guy has explained his methodology very carefully. One guy has not. Which are you, and which should someone trust?

(I have quibbles with his methodology, certainly, but I have no reason to say that he's off by such a huge margin.)

[TJ in Oregon]

The answer to all of this is in the convention bounces. Nate Silver's model uses the underlying assumption that the convention bounces will look like this:



So when the curves are added together, the total effect should look like this:



But for whatever reason, Mitt Romney didn't really have a convention  bounce, maybe 1-2 points maximum, so the polling right now most closely resembles that blue curve on top rather than the purple curve below the model is expecting it to follow.

Now, this is obviously good news for President Obama since it could include a more heartfelt swing with the convention bounce that will not diminish. It also suggests the possibility that Mitt Romney simply has a ceiling around the 46% he's been polling and there are no more persuadable independents he could win. That's entirely possible.

On a related anecdote, I asked my roommate last night (a disaffected independent) which convention he thought was better and he immediatly responded that the Democratic Convention was better. I then asked him why, what about the Democratic convention was better, to which he answered "Bill Clinton's speech". I asked him why he thought Bill Clinton's speech was so effective and persuasive and he told me "I don't know. I didn't watch it. I just read online and on Facebook it was really good". This serves as a casual reminder how the American indepent voter thinks.

Now perhaps Obama's lead does have some staying power, but if it really is a bounce from a speech that most people didn't watch by a guy who's not running, I have to think the basis is superficial enough that it will be forgotten. If this were baseball, it did give Obama a couple runs on the board, but it's only like the 3rd inning. The debate performances or whatever hits the news cycle next could easily outweigh it.

[Filuwaúrdjan]

Kerry didn't have much of a bounce, IIRC. 2004 in reverse?

[Craigo]

Kerry didn't have much of a bounce, IIRC. 2004 in reverse?

2000, 2004, and 2008 all saw the in-party get the better bounce. I don't know if that's just noise, but it happened again this year.

[後援会]

I take Nate Silver at his word.

That being said, he said that his forecast only holds if Obama continues making gains in the next few days. Because he extrapolated that the convention bounce was 7-9 points from the whole "hey we're Gallup and we do 7-day rolling averages".

I personally don't think that will happen. Just because a nine-point bounce is patently ludicrous. There aren't even that many persuadable voters in America. And if/when it doesn't, we'll see Obama's chances probably tumble down to around the 65%-75% region. Which is about where I think it is at.

[pbrower2a]

80% of an Obama win coming from left-wing NYT? Anybody can take that with a grain of salt. More like 50% is realistic.

This is the most charitable view that I can show toward the prospect of a Romney victory in view of the right-wing groups pulling ads from Michigan and Pennsylvania, which those interests consider lost causes. 

At this stage the best way of looking at the chance of an Obama or Romney win is based upon probability. President Obama has locked up 242 electoral votes. Florida alone has enough electoral votes to put President Obama over the top, and that alone is about a 50-50 proposition. Mitt Romney could win all other states in blue or gray and he would still lose.

If you are at all familiar with probability you know what an 'independent event' is -- one whose result does not influence the likelihood of another. Prime examples include tosses of coins and dice. Except for Iowa and Wisconsin, states that generally move in tandem (that they were won respectively by Dubya and Kerry in 2004, but both by razor-thin margins, doesn't change that reality) all other states are so separated by geography or have differences of political heritage that no single one-size-fits-all appeal works on all of them.  Except for Iowa and Wisconsin, which I will treat as if one large state, all others should be treated as if coin tosses for this purpose.

So here are the 'independent events' in ascending order of electoral votes:

New Hampshire   4
Colorado             9
Virginia              13
North Carolina   15
IA + WI              16
Ohio                   18
FLORIDA             29 

There are 2^7 possible results (128) of random chances if one assumes each state or the one combination is a 50-50 proposition. 64 of those involve Obama winning Florida. That is half of them right there.

OBAMA 64, Romney 0 so far.

With Obama winning Ohio but not Florida (32 chances), winning propositions for Obama include losing all other states or winning only Colorado or New Hampshire but not both. That's three of the remaining 32.

Obama 93, Romney 3.

With Obama losing Florida and Ohio but winning Iowa and Wisconsin (in tandem, of course),  he still wins if he gets any other two states, or at least one of Virginia and North Carolina.       

Obama 106, Romney 3. Sixteen remain.

Losing FL, OH, IA, and WI, President Obama wins if he wins North Carolina and either Virginia or the combination of Colorado and New Hampshire. He has eight chances of winning   Virginia and one chance of winning Colorado and New Hampshire -- 9 wins and  7 losses.

Obama 114, Romney 10.

Four remain, and only one of those is a winner -- Obama wins Virginia, Colorado, and New Hampshire. The others are Obama losses.

Obama 115, Romney 13. 

I see about a 90% chance of an Obama win (89.8%).

Please check the math -- someone!   


Time is running out for chances for Mitt Romney to make major changes in the dynamics of this election. He has to do so if he is to have more than about a 10% chance of winning. States usually solidify around this time barring only a collapse.

[Franzl]

Why would one assume that each result is an independent event?

[Mehmentum]

I would say that North Carolina and Virginia cold be considered one 'state'.  Though that wouldn't change the calculation much.

[H.E. VOLODYMYR ZELENKSYY]

I would say that North Carolina and Virginia cold be considered one 'state'.  Though that wouldn't change the calculation much.

It's very possible, probable even, that Obama could win VA and lose NC.

[Zanas]

Why would one assume that each result is an independent event?

Just because they occur simultaneously. If you toss two coins at the same time, the side you get from one does not determine the side you'll get from the other. On the other hand, if you withdraw a card in a deck of cards, you'll get a reinforced probability of withdrawing any other given card on the next turn, assuming you don't reintroduce the first one.

Magics of probabilities... Reminds me good times

[Franzl]

Why would one assume that each result is an independent event?

Just because they occur simultaneously. If you toss two coins at the same time, the side you get from one does not determine the side you'll get from the other. On the other hand, if you withdraw a card in a deck of cards, you'll get a reinforced probability of withdrawing any other given card on the next turn, assuming you don't reintroduce the first one.

Magics of probabilities... Reminds me good times

As a former statistics major...I understand that concept. It's just that the way states vote isn't independent in the way a coin toss is.

[Adam Griffin]

538 Current Chances:

National - Obama 80.7%
Nowcast - Obama 83.0%

New Hampshire - Obama 86.1%
Nevada - Obama 85.6%
Wisconsin - Obama 84.7%
Virginia - Obama 76.7%
Iowa - Obama 75.7%
Colorado - Obama 75.5%
Ohio - Obama 74.6%
Florida - Obama 66.9%
--------------------------------
North Carolina - Romney 56.5%
Missouri - Romney 82.7%
Montana - Romney 84.9%
Arizona - Romney 85.1%
Indiana - Romney 87.1%

[Zanas]

Why would one assume that each result is an independent event?

Just because they occur simultaneously. If you toss two coins at the same time, the side you get from one does not determine the side you'll get from the other. On the other hand, if you withdraw a card in a deck of cards, you'll get a reinforced probability of withdrawing any other given card on the next turn, assuming you don't reintroduce the first one.

Magics of probabilities... Reminds me good times

As a former statistics major...I understand that concept. It's just that the way states vote isn't independent in the way a coin toss is.

Sorry to patronize you if you are a fellow scientifically educated being

But it kind of is, in the way that no state knows for sure what the score of any other state will be when it vote.

[Franzl]

Why would one assume that each result is an independent event?

Just because they occur simultaneously. If you toss two coins at the same time, the side you get from one does not determine the side you'll get from the other. On the other hand, if you withdraw a card in a deck of cards, you'll get a reinforced probability of withdrawing any other given card on the next turn, assuming you don't reintroduce the first one.

Magics of probabilities... Reminds me good times

As a former statistics major...I understand that concept. It's just that the way states vote isn't independent in the way a coin toss is.

Sorry to patronize you if you are a fellow scientifically educated being

But it kind of is, in the way that no state knows for sure what the score of any other state will be when it vote.

True, but for events to be independent, would also mean that what influences one state cannot be related to any other state either.

For example, just for arguments sake, if opinion polls underestimate white turnout in general....that would have a universal effect on every state. If you then have two lean-D states that show an Obama lead of 1-2%, which way they go is not at all random as a coin toss.

[pbrower2a]

Why would one assume that each result is an independent event?

That's to keep the model simple. Political campaigning is far more complicated than probability and statistics.  

So is the 50-50 assumption, which a recent poll in Ohio made into a travesty. 75%? I would need to set up a computer program, and current computers seem not to support BASIC, which I find more reliable than repeated calculations.

New Hampshire - Obama 86.1%
Nevada - Obama 85.6%
Wisconsin - Obama 84.7%
Virginia - Obama 76.7%
Iowa - Obama 75.7%
Colorado - Obama 75.5%
Ohio - Obama 74.6%
Florida - Obama 66.9%
--------------------------------
North Carolina - Romney 56.5%
Missouri - Romney 82.7%
Montana - Romney 84.9%
Arizona - Romney 85.1%
Indiana - Romney 87.1%

Early in a campaign even a 10% lead looks as if it can be pared away with good campaigning. late in a campaign a 4% lead can be impossible to crack, and the only question about polling is how valid the figure is.   The calendar pages available until November 6 are running low.

...At this stage I see a complete meltdown of the Romney campaign that allows a bunch of seemingly-safe R states like Missouri, Montana, Arizona, Indiana, and perhaps Georgia or South Dakota slipping away from Romney far more possible than a Romney win. There are just too many fires to extinguish.

[mileslunn]

Why would one assume that each result is an independent event?

That's to keep the model simple. Political campaigning is far more complicated than probability and statistics.  

So is the 50-50 assumption, which a recent poll in Ohio made into a travesty. 75%? I would need to set up a computer program, and current computers seem not to support BASIC, which I find more reliable than repeated calculations.

New Hampshire - Obama 86.1%
Nevada - Obama 85.6%
Wisconsin - Obama 84.7%
Virginia - Obama 76.7%
Iowa - Obama 75.7%
Colorado - Obama 75.5%
Ohio - Obama 74.6%
Florida - Obama 66.9%
--------------------------------
North Carolina - Romney 56.5%
Missouri - Romney 82.7%
Montana - Romney 84.9%
Arizona - Romney 85.1%
Indiana - Romney 87.1%

Early in a campaign even a 10% lead looks as if it can be pared away with good campaigning. late in a campaign a 4% lead can be impossible to crack, and the only question about polling is how valid the figure is.   The calendar pages available until November 6 are running low.

...At this stage I see a complete meltdown of the Romney campaign that allows a bunch of seemingly-safe R states like Missouri, Montana, Arizona, Indiana, and perhaps Georgia or South Dakota slipping away from Romney far more possible than a Romney win. There are just too many fires to extinguish.

Those are way too optimistic for Obama.  I would say 65-35 in favour of Obama in Wisconsin, 60-40 in Nevada, 55-45 in Ohio, Iowa, New Hampshire, and Colorado, while 50/50 in Virginia and Florida.  North Carolina, Indiana, and Missouri sound about right, while for Montana, more like 95-5 in favour of Romney.  The state has quite libertarian so I think it is pretty tough to win there if you are someone who believes government can be a force for good.  I would also say Romney has a greater than 15% chance of winning Michigan and Pennsylvania although I still think Obama will win those.  I agree if he messes up other states could come into play, but likewise if Obama messes up, Oregon, New Mexico, and Minnesota could come into play and if Obama's numbers take a large nosedive i.e. another big recession comes on, Washington, Maine, and New Jersey could even come into play.  The point is the dramatic shift to Obama in October 2008 was largely a reaction of the strong economic downturn as the Republicans were in power and blamed for it.  Unless something that dramatic happens I doubt the swing states will change, if anything they will narrow down to fewer.