Thursday, November 6, 2008

Election wrap-up

Allowed myself a day to recover, now back to 'work' as it were. A piece of advice: don't give blood and then drink heavily.

Two graphs caught my attention in the post-election round-up. First, here's one from the NYT showing areas of the country where McCain did better than Bush in '04. Hmm, what do these areas of the country have in common?

I suppose that the Louisiana redness can be attributed to Katrina -- New Orleans was a Democratic stronghold, and many of them have left the state. But for the Ozarks and southern Appalachians, I guess you either have to posit that they for whatever reason really liked John Kerry in '04 or they wouldn't vote for a black guy in '08. My money is on the latter. For Arkansas, I wonder if there were a lot of Clinton voters there who just refused to vote Obama.

A second graph shows that the states pretty much lined up as they did in '04, but that Obama simply 'moved the line' a few percentage points in his favor, enough to have a huge swing on the electoral college:

This shouldn't be surprising, of course, but it is still neat to see such a nice correlation. The one data point that I find particularly interesting is Massachusetts, as apparently there were people who voted Kerry in '04 but couldn't be moved to vote Obama in '08. If you happen to know anyone who voted as such, I'd be fascinated to meet them.

When looking at the final results, a few things stick out. First, Obama won two Bush states -- Nevada (+12) & New Mexico (+15) -- by larger margins that three Kerry states -- Pennsylvania (+11), Minnesota (+10), and New Hampshire (+10). I would imagine that increased Latino vote share accounts for a good chunk of this. Also, while Obama won more states than he needed to, the state that put him over the top was (as I predicted) Colorado. He won Colorado by 7 points, which game him his 270th to 278th EV. Viginia was next at +5, followed by Ohio (+4), Florida (+2), Indiana (+1) and possibly North Carolina (<1).>

In general, the pollsters did a very good job with this election. I'm sure or will do a thorough run-down. Of course, it is easier to get it right when it isn't all that close, as calling one or two states wrong will still give you the right winner. Also, unlike previous elections, this one was remarkably stable in the polling pretty much from the second debate onward, while both '00 and '04 showed a lot of late movement.