From time to time I’ve come across references to odds-makers parsing the probabilities of election outcomes, or at least the POTUS one. Do they ever break down the key U.S. States?
I’d like to know what, if anything, has been said about such Electoral College elections as Ohio, Virginia, Florida, Pennsylvania, and North Carolina. (A recent The Economist article pointed NC out as one that could be considered neither red nor blue, but harder for Obama to win this time around. I don’t know how many votes it carries offhand.) If there are any other Battleground States I has missed, please let me know.
I ask because IMHO this election can no longer be presumed not to be close. I do want Obama to win if anyone is wondering.
538 does a state by state breakdown. It’s run by Nate Silver and based upon his forecasting model.
Current odds:
63.3% Obama
I’m not sure why you want the state-by-state Presidential odds, but here they are:
Ohio 66.5%
Virginia 48%
Florida 33.1%
Pennsylvania 90.3%
North Carolina 14.2%. (Much more winnable a month ago IIRC.)
Wisconsin - 77% Obama
Colorado - 49.7% Obama
Website: http://fivethirtyeight.blogs.nytimes.com/
I think the New York Times limits non-subscribers to 10 free views a month, though there are workarounds for that.
Be careful: It looks like you might be making the statistical mistake of assuming that the states are independent events. They’re not: A gain in one state will usually be accompanied by gains in other states, and vice-versa. Nate Silver, in his models, already takes this into account, and thus, if you’re going to be trusting his state-by-state figures, you should also trust his overall figure, which is basically the same as the figure for Ohio by itself.
No, it does not. If you look closer, you’ll see that his prediction for who will win the popular vote is very close to the prediction for the electoral vote. And he’s basing that on the same polls that RCP is using; he just has more sophisticated math for analyzing them.
But he does assume in the end that Obama wins, 50% to 49%. If the actual result is Romney 50%, Obama 49%, then the electoral vote also changes.
I’m just saying that those who reassure themselves with the swing state polls need to unreassure themselves unless Silver’s analysis actually predicts a Romney popular vote win and an electoral vote loss. The odds of either man winning the popular vote and losing the electoral vote, according to Silver’s latest projections, is only 7.3%.
No. That is false. His model predicts that Obama will win 50 to 49%, based on a complicated weighted averaging process of current surveys and economic indicators. Drop the economic indicators and the results tend to be stronger for Obama. More like 68%.
To be clear though, 63.3% odds aren’t all that overwhelming. Nor are 80% odds in my view. Romney has had a serious shot at the Presidency since he smacked down Gingrich, though I’d prefer to have Obama’s odds than his. 63 > 37. 63 !>> 37.
Assume, predicts, the fact remains that Silver’s state projections and national projections go hand-in-hand. If Silver’s model predicts that Romney wins the popular vote, Ohio will not be 63.3% anymore for Obama.
My point is simply that the national polls matter, a lot. The winner of the popular vote is an overwhelming favorite to win the electoral vote. In Silver’s model a 93.2% favorite, to be exact. Posters who believe that Obama can lose the popular vote and still win are deluding themselves.
Agreed, subject to that 7.3% quibble. (Indeed, the state odds by Nate reflect both state and national polling data for that reason, since national trends will flip swing states, and sample sizes are superior at a national level.)
Well, yes, you’re not going to lose the popular vote by 5 points, for example, and still win the election. I’ve said in other threads that I don’t care about the popular vote, just tell me the state-by-state numbers, and I’m referring to numbers from the same polling firm in this example. I simply mean that I don’t need to know anything else but those state numbers, because I obviously can’t do a calculation in my head given the popular vote and translate that to electoral college results, and I have no idea what popular vote margin guarantees someone the election, so in the end the spread is effectively irrelevant.
The problem with the state polls is less accuracy. Sure, in a tied race, the state polls tell you a lot, especially if they all say the same thing and show a big lead for one candidate or the other. With the race now tied, I’m very concerned about Ohio. But if Romney’s up by 1 or more, I’m less worried. If he’s up by 2 and all the polls agree he’s up by 2 or so, then I’m not worried at all.