The new poll aggregation site airs out ideas for Election Night forecasting, using regression and a lot of Bayesian statistics.
Details, as well as some graphs from the latest fundraising data below.
http://stochasticdemocracy.blogspot.com/
*********Cross Posted at StochasticDemocracy**********
No poll update today, there haven't been enough polls to justify one.
On the other hand, we've made a ton of progress toward our election-night plans. I'll write this up more later, but the basic idea is to use math(regression and a lot of Bayesian statistics...) to extrapolate results from early reporting counties to the rest of the county. In effect, we'll be able to predict outcomes in Alaska using data from Maine.
We'll also be able to find patterns in real time. If there turns out to actually be a Bradly effect, our site will be the first to know.
We've tested a trial run on the 2004 elections, and it seems to work very well. It might be useful for people who don't want to stay up until 3 am watching the Begich Senate race in Alaska.
To make up for the lack of a poll update, more fundraising Graphs of the Day:
The number of self described CEO's who have contributed to each Campaign in the last month where data is easily available.
The number of self described Homemakers who have contributed to each Campaign in the last month where data is easily available.
Requests for any other profession graphs? (Best requests will be posted tomorrow)
*********Cross Posted at StochasticDemocracy**********