It’s difficult to explain how Nate Silver calculates his “tipping point state” formula, so if you are really interested in the details, go read his article. The basic idea, however, is fairly simple. In any scenario where one single state decides the presidential election, how likely is each state to be that state?
Using his simulations, it turns out that Virginia decides the election 28.4% of the time. The only other states that are decisive more than 7% of the time are Ohio (21.8%), Colorado (10.5%), Pennsylvania (8.3%), Nevada (7.8%), and Iowa (7.4%).
Nate brings this up in the context of Obama’s decision to change his immigration policy, but that’s not what interests me. Even if Latinos are underrepresented in these tipping point states, they are still overrepresented in two of them: Nevada and Colorado. So, he has good reason to want to make Latinos happy. And, frankly, votes are votes. Obama wants to maximize his support from Latinos in every state, even New Hampshire and Oregon. After all, he wants to do more than win. He wants some coattails so he can govern in his second term.
One of the glaring things here is the absence of Florida. It appears that however Florida votes it is unlikely to play a decisive role like it did in 2000. Basically, if Romney wins Florida, he would have won the election even without out, and the same is true for Obama. That’s probably not good news for Marco Rubio.
Another interesting tidbit is that Nate currently has Obama winning Pennsylvania with a 81.5% probability, Nevada with a 68.2% probability, Virginia with a 62.8% probability, Iowa with a 61.2% probability, Colorado with a 56.9% probability, and Ohio with a 53.2% probability. So, in every likely tipping point state, Obama is doing better than he did nationally (52.9%) four years ago. In Ohio, he’s well above the 51.5% he carried in the state against McCain. I think it is fair to say that he is in a commanding position.
But he’s not yet in a position to expand the map. Historically, presidents are either reelected by a bigger than original margin or they are defeated. It would be highly unusual for Obama to win reelection and not add any new states to his column. The most likely candidate is Arizona, where some new polling (pdf) shows him within the margin of error. I’m not sure what states come after Arizona, but my best guess is Missouri, Georgia, South Carolina, and Montana are in that group. The two states he won last time but is most likely to lose this time around are Indiana and North Carolina, followed by Florida. But I remain unconvinced that he will win by a narrower margin than he won last time around. It could certainly happen, but it is far more likely that he will lose outright than that he will fail to build on his margins.
