Quality Analysis

If you're looking for more information on the fine details of the Killian memos authenticity debate, The Washington Monthly has masses of data. The consensus is that the memos probably are frauds, but not necessarily. Bush-lovers will cry Aha! Democrat Lies! while Bush-haters will cry Aha! Karl Rove dirty tricks again!

The Monthly also has some interesting analysis of the wild variations in US pre-election polls:

"I'm not sure why this year's polls are so wildly out of sync, but I suspect what we're seeing these days is less a clash of polling and more a clash of polling models. The variance seems to be mostly a product of different algorithms for identifying 'likely' voters, which means the most accurate poll is whichever one has the best algorithm."

Some US polls only analyze pre-registered voters, assuming others are unlikely to vote. It could be a different story this year.