I will try to produce a quick memo on this tonight, but here are some things I’m seeing in the pattern of “roll off” (a possible proxy for “residual votes”) in the Franken/Coleman race in Minnesota:
- Roll-off appears to be greater in precincts that gave Obama a larger vote
- Roll-off appears to be greater in larger precincts
- Roll-off appears to be variable across counties
In addition, the small literature comparing optically scanned ballots with a recount finds about a 0.5% difference between the initial and final count. Back-of-the envelope calculations suggest that we might expect about 15,000 more votes to show up in a Minnesota recount, if this literature (by Herron and Wand, Ansolabehere and Reaves, etc.) is applicable here.