I have published an update of the paper referenced below, now doing the analysis on the residual vote rate at the precinct level, rather than drop-off. The substantive results are the same. The link to the new paper is here.
I’ve written a short memo, available here, which examines the “drop-off” rate in the Minnesota senatorial election. The drop-off rate, which is a proxy for the residual vote rate (currently uncalculatable), is a way of measuring how many ballots were not counted toward one of the senatorial candidates, and therefore are the most likely to augment the candidates’ votes in a recount. In this particular race, the roll-off rate was higher in areas that gave Obama a greater margin. While this proves nothing about what the count would be after a recount, it is suggestive that Coleman’s margin would narrow further under a recount (and perhaps flip), making Franken’s chances of prevailing after the recount (assuming it happens) greater than a dice roll.
Please excuse the errors induced by sleep deprivation.