Full analysis of the most recent date
Full analysis of the most recent date
On this site, I take the various state-by-state polls seriously; probably a lot more seriously than I ought to. As an observational astrophysicist, I know that weak data are usually beyond the aid of any analysis methods, no matter how sophisticated the analysis may be. But I am mathematically inclined with intense interest in the present election. I find the Electoral College to be complicated enough to be fascinating yet simple enough to model with statistics, which I have done here. Elsewhere, there's further discussion of the model.
I do not attempt to second-guess the polls or to remove the many known systematic errors (e.g. the no-cellphone problem, the identifying-likely-voters problem, the adjusting-for-party-affiliation problem, or the undecideds-break-for-the-challenger effect). I do not know enough about these issues to make adjustments for them. As a partisan Democrat, I take heart in thinking that (apart from the likely voters problem) all of these effects would appear to work in Kerry's favor.
Let me say it again: I think that the data presented here underestimate Senator Kerry's chances of winning the election.