I have added a brief post-election postscript on the sample-weighting controversy to my sample-weighting website:
Thorough discussion and relevant links are available at the website. Here is a brief summary, however. According to the adjusted exit polls (i.e., making the overall Bush-Kerry percentages match the actual popular vote), this year's electorate was comprised of 37% D, 37% R, and 26% I. I realize that the exit polls were messed up this year and that any adjustments may not adequately repair the situation; still, it's all we have to go by. These figures represent a 2% decline from the Democrats' 39% share in 2000 and a corresponding GOP gain of 2%. This year's polls that weighted on 2000 turnout thus were not heavily led astray. Polls such as Gallup's that frequently had substantially greater representation of R's than D's during the final months of the campaign were clearly out of bounds.