Georgia: Nunn not going down without a fight

[SPC]

Right now, based on the assumption that the Libertarian usually takes ~3%, I would peg the odds of a runoff at ~55% (with an outright Nunn win at ~5%), although that could change if Perdue's momentum keeps up or if Nunn turns this race around.

Assuming that there is a runoff, I am not sure that Perdue would be the overwhelming favorite that everyone assumes. If this were a normal winner-take-all election, sans runoff, Nunn would probably have ~25% chance of getting a plurarity on Election Night. Our sample of post-November runoffs consists only of Louisiana 2002 (where incumbent Landrieu gained a net ~7 points over her Republican opponents) and Georgia 2008 (where Chambliss gained a net ~15 points over his Democratic opponnent, although about half of this improvement can probably be attributed to historic black turnout in 2008.) Thus, I think the evidence is too spotty to say a priori that a Republican will do better in a runoff than in a midterm election, especially when it is an open race.

The closest thing to incumbency will be control of the Senate, which will probably give a modest boost to whichever party has it after the Louisiana runoff. Assume that analysts are correct that Republicans have a ~60% chance at control (which is an underestimate in this case, since a 50-49 split would almost certainly favor Perdue). Now, using Bayesian reasoning, if there is a runoff, there is ~20/55=~35% chance Nunn won a plurarilty on Election Night. If having control of the upper chamber boosts the majority party's chances by ~15%, then there is ~30% chance that Nunn wins the runoff, which cumulatively becomes a ~20% chance of Senator Nunn. Thus, while Nunn is certainly in a weaker position than she would be in a winner-take-all state, I do not believe the chances of her winning are significantly diminished by the runoff system.

EDIT: The more interesting case is going to be that in Louisiana, where likely residual uncertainties in Kansas and Georgia will make control of the Senate even more ambigious, which I suspect will place a larger albatross over Landrieu's neck than Nunn's. Assuming that Louisiana and Georgia both end up in runoffs, Republicans would have to pick up fewer than four Senate seats on Election Night (a remote possibility, given the likelihood that at least one of Pryor, Begich, Braley, or Udall will tumble) for Landrieu's seniority to work to her advantage. I suppose theoretically Landrieu could replicate her 2002 victory by Republicans picking up the six requisite seats on Election Night, but if at least two of her colleagues fell to challengers, I find it unlikely that Landrieu would avoid the same fate.