Favourite blogs for 小雪的后花园

My favourites » blogistic regression

November 10, 2004

More on voting…

Follow-up to Post–election statistical anomaly from blogistic regression

This picture says a lot, I think…
(in case it is not legible, the LH columns show Exit polls (Blue = Kerry) and the RH columns show the "actual" vote count. The states are
Row 1. Illinois, Maine, Wisconsin (all paper ballots)
Row 2. N. Carolina, New Hampshire, New Mexico (all electronic voting)
Row 3. Florida, Ohio, Pennsylvania (all electronic voting)

November 04, 2004

A comment on the US election from the distant past…

This captures my feelings – though hopefuilly my mood will improve as I convince myself of the poisoned chalice theory. Hope there is no prohibition against posting this:


Post–election statistical anomaly

A friend in Southern California has done a statistical analysis on several swing states, and found a perfect and suggestive correlation. Compared to exit polls, every state that implemented e-voting in a way that provided no independent paper trail shows Bush getting around 5% more in the "actual" result.

By contrast, in every state where e-voting has an independent paper audit trail, the exit poll and "actual" results are within the margin of error.

Kinda makes you think…

For stats folks with time on their hands – look for districts with Diebold machines…