The Presidential Campaign is now only happening in four states. The Romney campaign isn't spending any time in North Carolina, though they are certainly spending money on ads there. Instead, almost all of the Republican nominee's campaign events (as opposed to fundraisers) are focused on Florida, Virginia, Ohio, Iowa, and New Hampshire. Barack Obama has even further narrowed his appearances to those four states.
The polls in New Hampshire have shown Obama with a modest but consistent lead, so it's unlikely that Romney has a path to victory through the Granite state. Instead, he will have to pull of a clean sweep of the other four true swing states. Naively multiplying Romney's 538 win probabilities in those states gives him a 5% chance of victory, though that probably understates his chances a bit.
"Florida, Virginia, Ohio, Iowa, and New Hampshire" are five states.
There ought to be a Holy Hand Grenade joke in there somewhere.
And really, it's going to come down to the biggest three of that group: Florida, Virginia, and Ohio. If Romney can sweep them, he's the overwhelming favorite. But if he loses just one, Obama's almost certain to win.
It definitely illustrates the stupidity of our Electoral College system. The rest of us might as well be in some other country. Shifting 100,000 votes in California or Texas ought to matter just as much as shifting 100,000 votes in Florida, but only the latter counts.
"Naively multiplying Romney's 538 win probabilities in those states gives him a 5% chance of victory, though that probably understates his chances a bit."
Respectfully, that calculation is extremely naive, and understates Romney's chances a lot. The probabilities in question aren't anything close to independent. Multiplying the numbers in question is a pretty worthless exercise.
Let's try another take: although on the whole, Romney's fortunes in the several states being contested should rise or fall together (i.e. assuming a fair degree of dependence), let's suppose a group of states, in the final polls, are in a dead heat between the two candidates.
At that point, which way each state swings is more or less random, because we've already factored in the group move to equality; now we're at the point where a few thousand votes either way in each state can make the difference.
And that part is random, coin-flip stuff.
So if Romney needs all three of the states I mentioned, his chances are 1/2^3 = 1/8, or about 12.5%.
At present, this *overstates* his chances a good bit, since he's not particularly close to even in any of the three key states. But we can use that as an upper bound of his chances of winning, assuming that he never gains the lead outright in the polls of any of the three states.
No argument from me that multiplying the probabilities is extremely naive. Especially since I trust the 538 probabilities about as far as I can throw them. But I'm not sure what else there is to do.
LTC's point about Romney needing to get to the point where 3 of the states are coinflips (and one is locked in for Romney) is a good one. We're really looking at an 8-10% chance for Romney unless something drastic happens to change the shape of the election.
I guess in the best case, something changes such that FL & IA, which are a bit closer than VA and OH, flip to Romney, and VA and OH are coin flips. Then Romney has a 25% chance of victory.
LTC: I think I mentioned that New Hampshire isn't actually in play even if candidates are visiting there.
I don't really know what to make of the Obama campaign shifting resources to WI. Right now it's actually even in the polls, but Obama's not advertising, and I think the difference between "zero ads" and "some ads" is about 4-5% based on the MN county results from the Fargo media market in '08. So my guess is that it will return to being a small but persistent Obama lead.
Post a Comment