Ahead of the second presidential debate tonight, it's worth taking stock of the task facing each candidate in the last three weeks of the campaign. It is clear that Mitt Romney has received a sizeable bounce since the first debate, closing the gap to Barack Obama by probably around 4 points nationally. Nevertheless, it looks like he still remains about one point adrift of the President, and Nate Silver's Fivethirtyeight forecast makes Obama the clear favourite, with the odds against his victory at about 1/2.
Before the debates, talk of swing states and the electoral college seemed superfluous. Obama looked likely to win the popular vote by around four points, a margin which would guarantee him the 270 electoral votes he needs. A split between the popular vote and the electoral college is unlikely — it has occurred just three times in US history: in 1876, 1888 and 2000. But when the popular vote margin is two points or less (as it looks like being this time), it is certainly worthy of consideration.
According to Silver's model, there's about a 7 per cent chance of it happening this year — and if there is a split, it's most likely to favour Obama. It finds a 4.3 per cent chance of Romney getting more votes but losing the electoral college, compared to just a 2.7 per cent chance for Obama. In other words, if Obama won the popular vote, Romney would only have about a 4 per cent chance of securing the White House, whereas if Romney won the popular vote, Obama would still have about a 12 per cent chance of re-election.
Why the split? Well a lot of it may come down to Obama's strength in Ohio. The Buckeye State has tended to be slightly more Republican than the country as a whole (in 2008, for example, Obama beat McCain 53-46 nationally, but only 51-47 in Ohio), but this year Obama seems to have turned that around. Before the debate, Obama was actually been doing about one point better in Ohio than he was nationally. And the polls suggest Romney's post-debate bounce has been smaller there — around two points, compared to about four points nationally. So whereas Silver's model gives Obama a 64.4 per cent chance of getting more votes than Romney nationally, it gives him a slightly better 69.1 per cent chance of getting more votes in Ohio and securing the state's 18 electoral votes.
Ohio was the decisive state in 2004, and is undoubtedly the most important state again this year — if the election comes down to one state, the most likely one by far is Ohio. And Obama's strength there is certainly a big problem for Romney. If Ohio were still two points more Republican-leaning than the country as a whole, as it has been in the past, Romney's path to victory would be much easier. But it'd be wrong to say, as others have, that it's near impossible for Romney to win without it. If you take all the states where Romney is doing better than he is in Ohio, you get 269 electoral votes — enough for a tie (as shown in the map below, which I put together at 270.com). That'd mean the House of Representatives would decide the winner, with each state's group of Representatives getting one vote. That'd very likely amount to a Romney victory, as the Republicans currently hold the majority in 33 House state delegations to the Democrats' 16, and are very likely to maintain an advantage after these elections.
Of course, a tie is always improbable (Silver's model rates the chances of one at just 1.3 per cent), but it is a demonstration of how Romney could take the White House without Ohio. And he could substitute two other competitive states — New Hampshire and Wisconsin — for Virginia or Colorado, or for Nevada and Iowa. The point is that when we talk of Ohio as a 'must win' state, it is in a way more true for Obama than Romney. That is, while it's unlikely either candidate wins the election without Ohio (indeed, no candidate has since John F Kennedy did so in 1960), it's more likely that Romney manages it than Obama does.