Monday, June 13, 2011

No, Really, Forget the Electoral College For Now

One more time: there's just no reason to pay very much attention to state-by-state stuff if you're trying to project a winner in 2012.

Maybe I'm just grumpy today, but when I read this sort of thing it just sets me off:
To win reelection, President Obama needs to match his 2008 performance in four swing states: Florida, Nevada, Michigan and North Carolina.
That's the normally very sensible Jamelle Bouie, reacting to a WSJ article about unemployment in those states. It's not just Bouie; Dylan Matthews quoted the original article seemingly approvingly in this morning's Wonkbook. Sorry, but no dice.

OK, to begin with: obviously, Obama doesn't need to match his 2008 performance in those states. I mean, among other things, Obama won Nevada by 12.5 points; that means that he could run 12 points worse and still win there.

Continuing...a better way to look at it would be to think about the effects of a nationwide 7 point swing away from Obama, since he won by 7 points. Looking at the results...Obama would lose IN, FL, NC, OH, and VA (and I'll toss in the EV from Nebraska). Without adjusting for reapportionment, that would have been 87 EVs lost, for a 278-260 win. So Obama can actually afford a uniform 7+ point swing.

Suppose that it's worse. Then the question is: where? The next few states (that is, the closest states on the Obama side after a uniform 7 point swing) are CO, IA, NH, MN, PA. So, once again positing a uniform swing, those are the states that Obama needs to do well in. As the swing goes from 7 points to the GOP to 11 points, he gradually loses each of those, and with them the election. To begin with, then, I'd start with those states, none of which are the WSJ four.

Now, here's the thing. Suppose that, for example, Nevada has a larger-than-average swing away from Obama, thanks to its unusually high unemployment rate. But that implies that some other state will have a smaller-than-average shift! And that's true, also, if the question is the demographic factors Bouie looks at as opposed to general economic effects. So maybe the states reshuffle a bit; in fact, looking election-to-election, there's always some reshuffling going on, for a wide variety of short- and long-term reasons. Some of them are easy to see in advance (demographic changes can yield fairly predictable partisan changes), while others are a lot harder from this point, because they depend on how the campaigns go and other things we can't yet predict. Note, too, that the various states mentioned above and the others that were a bit stronger for Obama in 2008 (WI, NM, NJ) really don't have all that much in common, so it's not especially likely that there will be something that pushes all (or none) of these states in the same direction compared to the overall average.

The point is that all of this is going to be basically irrelevant if Obama runs within a few points nationally of where he was in 2008, and it's also going to be irrelevant if he runs a dozen points worse -- in other words, if he wins or loses nationally by a few points or more. It probably doesn't matter even if he only wins or loses by a single point nationwide. And if it's closer than that, it's basically impossible to know right now if the key states turn out to be, say, Iowa or Nevada or Pennsylvania, and those are three very different states.

2 comments:

  1. You wrote, "I mean, among other things, Obama won Nevada by 12.5 points; that means that he could run 12 points worse and still win there."

    If you mean that his margin of victory was 12.5 points, then the loss of only 6.3 points to a single opponent would lose the state for Obaman, wouldn't it, as those points would be added to the challenger's total?

    ReplyDelete
  2. Bill,

    Well, 12 points net worse. Or however you want to word it.

    ReplyDelete

Who links to my website?