From “How to Solve the Swing-State Puzzle,” Nate Silver, NYT:
Using my FiveThirtyEight model, I’ve determined — through about 25,000 simulations that I run each day — which states could put either candidate over the top. Crucially, the model takes into account not only how states poll relative to national trends but also to one another. Demographically similar states can rise and fall together. If Romney makes gains in Wisconsin, for example, he will probably also do so in neighboring Minnesota.
Perhaps more important, the program evaluates the order in which the states might line up. Obama could win North Carolina, where the polls show a competitive race, but he’s unlikely to do so without already having won Ohio, Florida and Virginia, where the demographics are slightly more favorable to him. Some combination of those states would probably get him to 270 electoral votes anyway. By that point, North Carolina would be redundant.
Which states, then, are the most strategically important? The answer exists along a spectrum rather than in absolutes. Ten states could play an important role in the electoral calculus. I have listed them below, in four groups, along with the chance that each state will be the one that determines the next president.
The Big Two: Ohio (32 percent chance of determining the Electoral College winner) and Florida (20 percent).
The auto industry’s recovery has helped drop Ohio’s unemployment rate from 8.6 percent when he took office to 7.2 percent now, making it one state where voters really are better off than they were four years ago.
Florida’s economic recovery has not been as robust, but Obama may be buoyed by long-term demographic factors there. The G.O.P. has long been buffered by Cuban-Americans, a historically right-leaning group, who made up a majority of the state’s Hispanic electorate. Now not only are non-Cuban Hispanics growing in the electorate, but the Cuban population is increasingly divided along generational lines, with younger voters leaning heavily left.
The New Breed: Virginia (9 percent), Colorado (9 percent) and Nevada (5 percent).
In these states, which Obama carried in 2008 but Kerry and Gore lost, swift demographic changes have become manifest. Obama won Nevada — which now resembles a West Coast state to some degree — by an unexpectedly large margin, 12 percentage points, in 2008. And despite a wretched economy there, he has led in every state poll conducted this year.
The polling has been more inconsistent in Colorado and Virginia.
In the end, Obama might simply conclude that Florida or Ohio — and not Colorado and Virginia — represents his path of least resistance. If the president can win either Florida’s 29 electoral votes, or Ohio’s 18 plus Nevada’s 6, then Romney’s shot at 270 will become vanishingly thin, and it won’t matter how Virginia and Colorado turn out.
Primary Purple: Iowa (6 percent) and New Hampshire (3 percent).
The voters in Iowa and New Hampshire are almost entirely white and mostly rural — factors that ordinarily favor Republicans. But they are also highly educated, which gives Democrats a chance. … [I]t’s worth remembering that if Al Gore had won New Hampshire in 2000, he wouldn’t have even needed Florida.
The Blue Wall: Wisconsin (9 percent), Pennsylvania (5 percent), and Michigan (1 percent).
In Michigan, Romney’s opposition to the auto bailout may be too much of an albatross. In Pennsylvania, though, the issue may be that while the polls are close, they are also hard to move; each party has its respective constituencies, and there may be few true undecided voters left.
My calculations suggest that, despite Romney’s deficit, the upside of his winning Pennsylvania is so great that he might want to take a chance. It’s Obama’s closest equivalent to a must-win state, and the combination of losing Pennsylvania and Ohio would essentially ensure his defeat. Unfortunately for Romney, it may be too late to adopt that strategy, as Obama has come close to clinching a majority of the state’s electorate in recent surveys.
Wisconsin, however, is the state that Romney must contest. If Romney can’t overtake him in Wisconsin, considering his problems in Ohio and Florida, he’ll leave Obama with too many paths to 270, and himself with too few.
The most plausible range of outcomes runs from Obama losing the election by about two percentage points, slightly better than John Kerry did, to his winning it by perhaps six or seven, slightly worse than his margin from four years ago. Given where the election is being contested, however, the most likely outcome is that Obama wins enough tipping-point states to eke out a victory.
Heading emphasis in original; text emphasis added.