FiveThirtyEight: Nate Silver sets us worrywarts straight
Nate Silver of FiveThirtyEight has been getting a bit testy lately, which I suspect is reflective of the types of e-mails he's been receiving recently.
Yesterday, he took the time to ease Obama supporters' fears surrounding Pennsylvania:
Pennsylvania Sanity Check
Suppose that Barack Obama were to concede Pennsylvania's electoral votes. Literally, concede them. Throw 'em back, like a Chase Utley home run at a Cubs game. How often would he still win the election?
... 89.0% of the time, according to our most recent run of simulations, along with another 2.4% of outcomes that ended in ties. This is because in the vast majority of our simulations, Obama either:
a) was winning at least 291 electoral votes, meaning that he could drop Pennsylvania's 21 and still be over 270, and/or
b) was winning at least 270 electoral votes, while already being projected to lose Pennsylvania in the first place.
(a) was much, much more common than (b), obviously.
On Pennsylvania Being "In Play", also posted yesterday, offered Nate's extended explanation on this subject.
He's also been addressing concerns about the electoral map as a whole.
Yesterday, he explained what Obama needs to do to win:
Today's Polls, 11/1
This is beginning to look like a five-state election. Those states are Virginia, Colorado, Pennsylvania, Ohio and Nevada. Essentially all relevant electoral scenarios involve some combination of these five states.
I should caution that by far the most likely scenario is that Obama wins some relatively decisive victory of anywhere from 3-12 points in the popular vote. If Obama wins the popular vote by anything in this range, he will find plenty of blue territory, accumulating somewhere between 300-400 electoral votes. The electoral math will matter very little.
We can probably assume, however, that IF the national polls tighten significantly (and to reiterate, the likelihood is that they will NOT), McCain will edge out a victory in North Carolina, Florida, Indiana, North Dakota, Montana, Georgia, and Missouri; put those states in the McCain column for the time being. Likewise, New Mexico, New Hampshire, Minnesota, Michigan, Wisconsin and Iowa all appear safe for Obama, even in the case of significant tightening. Put those in the Obama column.
That leaves our five states in play. The victory conditions for Obama involving these five states proceed something as follows:
1. Win Pennsylvania and ANY ONE of Colorado, Virginia, Ohio, or Nevada*
2. Win Ohio and EITHER Colorado OR Virginia.
3. Win Colorado AND Virginia AND Nevada.
(* Nevada produces a 269-269 tie, which would probably be resolved for Obama in the House of Representatives.)
Now, suppose you think that Colorado is already in the bag for Obama because of his large edge in early voting there. We can then simplify the victory conditions as follows:
1. Win Pennsylvania
2. Win Ohio
3. Win Virginia AND Nevada
That's basically what it comes down to, although I'm sure each campaign would claim that there are a larger number of states in play.
Today, he explained what McCain needs to do to win:
What A McCain Win Looks Like ...
[... T]here are some states that truly do appear to be "must-wins" for McCain. In each and every one of the 624 victory scenarios that the simulation found for him this afternoon, McCain won Florida, Georgia, Missouri, Indiana and Montana. He also picked up Ohio in 621 out of the 624 simulations, and North Carolina in 622 out of 624. If McCain drops any of those states, it's pretty much over.
Just to clarify, there are 624 victory scenarios for McCain out of the 10,000 simulations that Nate ran this afternoon.
So, keep these states, and their math, in mind, as you watch the election coverage on Tuesday.
Yesterday, he took the time to ease Obama supporters' fears surrounding Pennsylvania:
Pennsylvania Sanity Check
Suppose that Barack Obama were to concede Pennsylvania's electoral votes. Literally, concede them. Throw 'em back, like a Chase Utley home run at a Cubs game. How often would he still win the election?
... 89.0% of the time, according to our most recent run of simulations, along with another 2.4% of outcomes that ended in ties. This is because in the vast majority of our simulations, Obama either:
a) was winning at least 291 electoral votes, meaning that he could drop Pennsylvania's 21 and still be over 270, and/or
b) was winning at least 270 electoral votes, while already being projected to lose Pennsylvania in the first place.
(a) was much, much more common than (b), obviously.
On Pennsylvania Being "In Play", also posted yesterday, offered Nate's extended explanation on this subject.
He's also been addressing concerns about the electoral map as a whole.
Yesterday, he explained what Obama needs to do to win:
Today's Polls, 11/1
This is beginning to look like a five-state election. Those states are Virginia, Colorado, Pennsylvania, Ohio and Nevada. Essentially all relevant electoral scenarios involve some combination of these five states.
I should caution that by far the most likely scenario is that Obama wins some relatively decisive victory of anywhere from 3-12 points in the popular vote. If Obama wins the popular vote by anything in this range, he will find plenty of blue territory, accumulating somewhere between 300-400 electoral votes. The electoral math will matter very little.
We can probably assume, however, that IF the national polls tighten significantly (and to reiterate, the likelihood is that they will NOT), McCain will edge out a victory in North Carolina, Florida, Indiana, North Dakota, Montana, Georgia, and Missouri; put those states in the McCain column for the time being. Likewise, New Mexico, New Hampshire, Minnesota, Michigan, Wisconsin and Iowa all appear safe for Obama, even in the case of significant tightening. Put those in the Obama column.
That leaves our five states in play. The victory conditions for Obama involving these five states proceed something as follows:
1. Win Pennsylvania and ANY ONE of Colorado, Virginia, Ohio, or Nevada*
2. Win Ohio and EITHER Colorado OR Virginia.
3. Win Colorado AND Virginia AND Nevada.
(* Nevada produces a 269-269 tie, which would probably be resolved for Obama in the House of Representatives.)
Now, suppose you think that Colorado is already in the bag for Obama because of his large edge in early voting there. We can then simplify the victory conditions as follows:
1. Win Pennsylvania
2. Win Ohio
3. Win Virginia AND Nevada
That's basically what it comes down to, although I'm sure each campaign would claim that there are a larger number of states in play.
Today, he explained what McCain needs to do to win:
What A McCain Win Looks Like ...
[... T]here are some states that truly do appear to be "must-wins" for McCain. In each and every one of the 624 victory scenarios that the simulation found for him this afternoon, McCain won Florida, Georgia, Missouri, Indiana and Montana. He also picked up Ohio in 621 out of the 624 simulations, and North Carolina in 622 out of 624. If McCain drops any of those states, it's pretty much over.
Just to clarify, there are 624 victory scenarios for McCain out of the 10,000 simulations that Nate ran this afternoon.
So, keep these states, and their math, in mind, as you watch the election coverage on Tuesday.