Reply #10: Hmmmm

[waggawagga]

This probability analysis is nutty. The assumptions the authors are making is that voter behavior is essentially random. For every voter who changes from A to B there has to be someone who goes from B to A, and so on. If that were true you'd only have to take one poll.

The truth is basically the opposite. Public support for a candidate surges and wanes. An odd poll result is one where, over time, changes in support for a candidate appear random (eg. one day at 60%, the next day at 39%, the day after 55%). A good poll result usually shows broad patterns (eg. the accumulation or loss of 1-2 points day by day) punctuated by large changes which are tied to specific events (eg. the bump presidential candidates get around convention time).

The fact that the polls cited were all "off" by the same 10-11 points doesn't suggest fraud or even some huge failure in the methodology of these polls. A random spread (+5 here, -5 there) would be strange because public sentiment, generally, is not random. One party goes up while the other goes down. That's the expected result.

Lastly, anyone who has followed elections during the past 15 years knows that election eve poll numbers are not that reliable. There seems to be this large segment of the public which doesn't make up their minds until the last moment. Decades ago it was often the case that this group would split down the middle (thus cancelling each other out). Lately, though, as a group they seem to break for one candidate.

Clinton got these people in '92. Dole got them in '96. Gore got them in 2000. If you look at statewide races you'll see the same phenomenon (eg. Hillary Clinton got them in NY). In 2002 the Republicans got them.

The above, btw, is one reason why I don't put too much stock in Bush's high poll numbers since 9/11. These people, whatever they tell pollsters today, are still on the fence and will break for whichever candidate looks good at the last minute (understanding who these people are should be job one of the Democrats, as much as I hate to say it I think it was the Wellstone memorial service which turned them off in 2002, they're a fickle bunch, but if a candidate knows how to appeal to them bumps of ten points or more during the last few days of an election cycle are not unusual, expect this in 2004).