Friday, October 24, 2008

FiveThirtyEight Watch

As I might have mentioned I'm a little wary of Nate Silver's analysis; his statistical model is overloaded with bells and whistles, he tweaks it too often, and I don't entirely trust him to keep partisanship out of his model. Nevertheless he's generally sharp and well-informed, and I was surprised to see him say this in his new TNR column:
Bill Clinton--running as an outsider in 1992--won Montana, and came within single digits of George Bush in states like Wyoming and Alaska. By 1996, however, when his incumbency had transformed him into an insider by default, Clinton lost Montana, and was crushed in Wyoming and Alaska by 13 and 18 points, respectively.

Here's some data from the site he links to.

1992
Montana: Clinton 38 Bush 35 Perot 26
Wyoming: Bush 40 Clinton 34 Perot 26
Alaska: Bush 40 Clinton 30 Perot 28

1996
Montana: Dole 44 Clinton 41 Perot 13
Wyoming: Dole 50 Clinton 37 Perot 12
Alaska: Dole 50 Clinton 33 Perot 11 Nader 3

2000
Montana: Bush 58 Gore 33 Nader 6
Wyoming: Bush 68 Gore 28
Alaska: Bush 58 Gore 28 Nader 10

2004
Montana: Bush 59 Kerry 39
Wyoming: Bush 69 Kerry 29
Alaska: Bush 61 Kerry 35

If you aggregate the Perot/R vote and the Nader/D vote, these states were stable over the period, with Montana going roughly 60/40, Wyoming between 65-35 and 70-30, and Alaska steadily bluing from 70/30 to more like 60/40. Clinton in fact did anomalously well in 1996.

What really happened in 1992, of course, was that Perot siphoned off a lot of Bush votes. In 1996 he was no longer a particularly interesting third-party candidate, so he siphoned off fewer votes and Dole won even though Clinton's 1996 showing was better in absolute terms. (Clinton generally dominated that election.) Kerry did about as well as Clinton '92, and better than Gore, which can largely be explained by the Nader vote.

No comments: