kelticwizard wrote:In a race where all candidates are considered, the latest Zogby poll shows Bush up by a mere three points. The Zogby poll a week before this showed Bush up by four points in an all-candidates race! [..] The election is six weeks away. If Kerry goes up a point a week for the next six weeks, as he is presently doing, he finishes ahead of Bush by three points in the popular vote. And three points advantage in the popular vote will result in the Electoral College agreeing.
Of course those on our side have their own tunnel visions ... <smiles at keltic>
Some things in life, I guess, must simply be accepted. Foxfyre believes only the Rasmussen polls. Kelticwizard believes only the Zogby polls. Such is the world. ;-)
Seriously though, I want to propose a question again. Zogby may have been the closest in 2000 (well, apart from Harris, I believe, actually), but several other polls were within the margin of error as well. If six polls had the result guessed right within the margin of error (see
here), then how does one explain a situation in which the poll that, of those six, had the result second-closest now says one thing, but several of the other six say a very different thing? They were all close last time - are we to reject the several others who had it close then but now say Bush is leading big, just because the one poll that had it one percent closer still says the opposite? Does that make sense, statistically?
No. But there's more. For one, a one percent shift in any poll is indistinguishable from statistical noise. Well within the margin of error, so can't really base any conclusions on it. Example in case: in the two-way race, likely voters, the very same Zogby poll actually had the race move to
Bush by one point (47%/45% to 47%/44%).
Finally, you can't extrapolate purely on the basis of the single shift from one poll to the very next. I mean, shift the reference point by one or two counts, and you get a totally different result. For example: exactly a month ago, Kerry lead by 7 points in the Zogby 2-way poll. Now he trails by 3 points. So going on Zogby, by the same logic, in one and a half month Bush will win the elections by 18 points. You gotta take a little bit of a wider horizon than that ...
I want Kerry to win too, and the divergences between the polls are large enough to suggest there's still a window of opportunity, even if imho it's small. (That's it's small in my HO is because of what I expect from the debates - namely, an advantage for Bush. More about that later.) In the meantime however, I'm trying to sift the sense from the senseless in the analysis mix ... no offence intended.