One thing that has bugged me about the coverage of the election was the idea that there was a high probability of Romney winning the popular vote by a decent amount (>1%) and Obama winning the electoral college. Joe Scarborough seems to be one of the worst offenders at this. Much of the polling seemed to suggest that not only is this a possibility but a fairly likely event as well. However, it never seemed intuitively to me to be a all that probable. Yes, there have been a few elections where the popular vote loser won the electoral college, but in each of those cases the popular vote was quite close.
Well today a couple of the better analysts have examined the problem and basically said that one of the polls has to be wrong. Sean Trende says when you aggregate all of the states, the state polling is not consistent with the national polling. Nate Silver suggests that the state polling is better. While Jay Cost seems to be skeptical of the state polling, at least in Ohio. It is good to see some confirmation of my own intuitive thoughts.
One caveat is Hurricane Sandy. If the hurricane really depresses voting in New York and New Jersey, it probably will not have much of an impact on the electoral college as Obama would likely win both states anyway. However, it could impact Obama's popular vote total enough that Romney wins the popular vote by a bigger amount (i.e. 1-2%) and Obama still wins the electoral college. There again this scenario would not be able to explain the current polling discrepancy.