« The Economist Swing State Series (or, My Rant About the Kerry Campaign) | Main | Kerry Up Big in New Ohio Poll »

Wednesday, May 12, 2004

WaPo Op-Ed Revisited

Posted by DavidNYC

The other day, I took issue with an op-ed in the Washington Post which argued that the list of swing states changes a lot more than a lot of people claim.

Well, angry moderate over at DKos took a sledgehammer to the piece. He (or she) points out that the op-ed is even more wrong-headed than I realized. If you look at the states which had a Dem-GOP margin of 10% or less in 2000, 19 of 21 were also under 10% in 1996 and 20 of 21 were in 1992.

What's actually happened is that the total number of swing states has diminished over time, as the nation has seemingly become more polarized. In `92, there were 33 states which were 10% or less, and in `96 there were 27. (The Perot factor might have skewed this a bit, but exit polls show that he took votes pretty equally from Clinton and Bush/Dole.)

Angry moderate's point essentially is this: If you were running the Swing State Project back in 2000, you would have started with a list of 27 states. After the election, you would have missed only Maine and Virginia, and you would have over-included California, Connecticut, Delaware, Georgia, Illinois, Kentucky, Maryland & New Jersey. That's a pretty good track-record, I think.

My aim at the very start was to cast my net as widely as possible. If I miss one or two close states, I'll be alright with that. And if I cover too many states, that's certainly fine with me as well.

Posted at 07:14 PM in General | Technorati


I was vague on what I meant by the Perot effect. I didn't mean that he drew more from the Republicans (though I do think that must be what happened in Montana), since that should have an overall net balanced effect on the number of swing states: some safe Republican would appear as swing, but some swing as safe Democrat. I just meant that he occupied a large percentage of the vote so that the two candidates in 1992 shared only 80.46% of the total vote vs. 96.25% in 2000. Therefore, 10% is a statistically larger number in 1992 than in 2000 so there will be more swing states. This made me curious as to how large an effect this would be. It's about 2%, that is if you take 12% for 2000 that would be about the same as 10% in 1992, which would make Vermont, California and Illinois swing in 2000. So it's not the primary reason why there is such a reduction from 33 to 21 states.

Posted by: angry moderate at May 14, 2004 11:00 PM | Permalink | Edit Comment | Delete Comment