however, the row regarding the bradley effect is only updated manually.

simulating what would happen in my simulaitons if every state experienced a 2% or 3% "bradley effect", that is if polls overestimated obama's share by 2 or 3%. David Stromberg estimates here the bradley effect to be between 2 to 3% in a set of recent elections.

wanted to do that for a long time :-)

a worst-case scenario for obama, in which he loses 7 pivotal states where McCain is concentrating campaigning efforts. Karl Rove wrote an article about that, he's right that this is a possibility for McCain, even if quite remote.

from Josep Colomer's blog

to the conditional probs... somehow it didn't make the first cut

including obama losing more pivotal states. I am now reporting only if they differ from the main statistics by more than 15%. Suggestions for more welcome!

much more readable now. Unfortunately the google chart api only lets me interpolate between 3 colors, so I can't do much better than this.