The incumbent rule

"The incumbent rule" says that in an election with an incumbent, the voters who wait to the very end to decide will usually vote for the challenger. The historical evidence for this rule is overwhelming: "analysis of 155 polls reveals that, in races that include an incumbent, the traditional answers are wrong. Over 80% of the time, most or all of the undecideds voted for the challenger." Therefore, my working assumption is that this Presidential election will follow the incumbent rule, and that Bush's situation in key states is therefore quite precarious.

That being said, over at Slate there's a wonderful "Consumer's Guide to the Polls" which has some data relevant to the incumbent rule's applicability to this presidential election. Some polls press respondents to decide on their vote for President, and for the polls that press and make the relevant data available, the authors tell us what the results of the pressing were, averaged over the last three samples:

Poll	Bush boost	Kerry boost
AP	0.78	1.23
Battleground	1.33	1
Democracy Corps	1.33	1.67
Gallup	0.75	1
ICR	2.63	1.97
Time	0.33	0.67

My quick and dirty unweighted average of the above data gives Bush a boost of 1.19 and Kerry one of 1.42. Now, one shouldn't leap to conclusions. These voters who only decided upon being pressed may be mere leaners, not "true" undecideds; perhaps they've really already decided but just have trouble admitting it to pollsters and/or themselves. Anyhow, it's something to think about.