FiveThirtyEight now showing a presidential forecast

[Angel of Death]

http://fivethirtyeight.blogs.nytimes.com/2012/06/07/election-forecast-obama-begins-with-tenuous-advantage/

[Senator Snowstalker]

The map seems about right.

[Yelnoc]

I would interpret their "291.3 EVs" as this map.

[Old Europe]

I would interpret their "291.3 EVs" as this map.

FiveThirtyEight places Obama's chances of winning Nebraska 2 at only 20%.

His chance of winning Virginia: 57.5%
Ohio: 55.9%
Colorado: 54.9%

[Angel of Death]

Current map:



which would work out as 303-235 for Obama. This number being higher for him than the global projection implies that he's more on the right side of the tossups right now.

[OC]

I think that Obama at the end win either VA or OH. But he doesn't have to win either one as long as he secures Democratic friendly states of CO,NV,PA and NH

[Political Junkie]

Nate Silver did a great job on the 2012 update. I expect Obama to win Ohio and Virginia thanks to the fairly low unemployment in both states.

[Lief]

This is way too complicated. A simple average of the last three or five polls in each state is probably as accurate as this gargantuan model at the end of the day.

[TheDeadFlagBlues]

This is way too complicated. A simple average of the last three or five polls in each state is probably as accurate as this gargantuan model at the end of the day.

Assigning a "state fundamentals" aspect to their model might be a mistake.

[The Mikado]

Nate Silver did a great job on the 2012 update. I expect Obama to win Ohio and Virginia thanks to the fairly low unemployment in both states.

Nate Silver's projections the day before election day were almost dead on (he missed IN and NE-02).  His predictions from June 2008...

My point is that the race's dynamics are far from set yet and that Nate Silver's oracular powers don't change that.  I'm reasonably confident that in November 2012, his day-before-the-election prediction will be dead on.  Now, not so much.

[retromike22]

I think he has Romney's chances of winning NC as way too high.

[Lief]

This is way too complicated. A simple average of the last three or five polls in each state is probably as accurate as this gargantuan model at the end of the day.

Assigning a "state fundamentals" aspect to their model might be a mistake.

It makes sense for a few states, where the polls are historically off in one direction or the other (Nevada, Alaska, Hawaii), but not otherwise.

[Fuzzy]

lol Oregon is more likely to go to Romney than Wisconsin. 

[Nichlemn]

This is way too complicated. A simple average of the last three or five polls in each state is probably as accurate as this gargantuan model at the end of the day.

Assigning a "state fundamentals" aspect to their model might be a mistake.

Why? If there's only poll of a state and it's from "Sketchy Pollsters Ltd" showing a result way out of kilter from what we would expect, it'd be stupid to use it as the sole basis for your forecast. Even if there are a number of polls from pollsters with reasonable track records, they still be unintentionally biased.

[You kip if you want to...]

Didn't 538 get the right result last time? (With the exception of MO being blue and IN being red).

[BaldEagle1991]

I think the fact that Obama is not doing as strongly in Oregon compared to other solid blue states is a danger sign for him.

[Nichlemn]

Also, PA being more likely to be a tipping point state than FL - reasonable, but no-one here is saying "Obama shouldn't bother with FL because he's already won then".

[Mr. Morden]

I wouldn't worry about the fact that Silver's forecast isn't going to be "accurate" this early in the cycle.  That isn't the right way to think about it, IMHO.

Silver's model is probabilistic, like Intrade.  He'll say "The probability of Obama winning Arizona is X; the probability of Romney winning Colorado is Y", etc.  The right way to assess whether the model is "accurate" is to ask "Do events that it predicts are 30% likely happen 30% of the time?  Do events that it predicts are 95% likely happen 95% of the time?", etc.  I haven't run the numbers on how he did in past cycles, but it's of course possible for his model to be accurate in that probabilistic sense even at an early stage, when things are very uncertain.

[MilesC56]

I think the fact that Obama is not doing as strongly in Oregon compared to other solid blue states is a danger sign for him.

We've only had one pollster do work in Oregon over the past year (SurveyUSA); I wouldn't read too much into Oregon specifically due to the dearth of polling there.

[Californian Tony]

Nate is always the best. My prediction map is based on his forecasts (tossup=within 67%, lean=67-90%, strong=over 90%).



Oregon and Minnesota are a tad surprising to me (I'd switch them off), but that's probably a fluke.

[TJ in Wisco]

Oregon being less solid for Obama in his predictions than most people think is mostly from the SurveyUSA poll that showed Obama up by 4. Nate Silver's model is really good, but in this case it's a garbage in, garbage out scenario. Oregon will correct itself once we get more polls there to dilute and bury the outlier.

[○∙◄☻¥tπ[╪AV┼cVê└]

Obama wins at least one state he failed to carry in 2008 10.0%
Obama wins Montana 17%

OK, Nate.

[True Federalist]

Oregon being less solid for Obama in his predictions than most people think is mostly from the SurveyUSA poll that showed Obama up by 4. Nate Silver's model is really good, but in this case it's a garbage in, garbage out scenario. Oregon will correct itself once we get more polls there to dilute and bury the outlier.

Compared to the U.S. as a whole, Oregon was more severely impacted by the Great Recession.  Even if there were no polling data from Oregon, I'd expect Oregon to be only +7 or +8 Obama.  That May poll is a bit of an outlier, but it's no more so than SUSA's March poll that had Obama at +11.  Obama is an unlikely Romney pickup, but not an impossible one, especially if the economy takes a turn for the worse between now and November.

[Holmes]

Obama wins at least one state he failed to carry in 2008 10.0%
Obama wins Montana 17%

OK, Nate.

Yeah, but Montana wasn't the only state he lost in 2008, was it? I'm guessing that 10% takes into account the probability of winning a state that he lost other than Montana, like Missouri or Arizona... or any of the rest for the matter. And it says "at least" as well, so it's also probably taking into consideration the chances of winning multiple states that he lost in 2008 as well. Maybe. I dunno.

[Grad Students are the Worst]

Obama wins at least one state he failed to carry in 2008 10.0%
Obama wins Montana 17%

OK, Nate.

Yeah, but Montana wasn't the only state he lost in 2008, was it? I'm guessing that 10% takes into account the probability of winning a state that he lost other than Montana, like Missouri or Arizona... or any of the rest for the matter. And it says "at least" as well, so it's also probably taking into consideration the chances of winning multiple states that he lost in 2008 as well. Maybe. I dunno.

Right...which is why 10% would make no sense.  Except the site now shows it as 28%, indicating that 10% was probably an error.