THIRD PARTY OUTCOMES: On the third-party questions below: Yep, there was a break away from 3rd parties after all. In South Dakota, the L, even though he had withdrawn and thrown his support to Thune, exceeded the R-D gap. But I think that was the only Senate seat tipped by one third-party candidate. In Missouri, neither the G nor the L exceeded the Carnahan-Talent gap, though both combined did (meaning that if Carnahan had somehow managed to get both those groups in her camp she could've won). In Minnesota, neither the Indepedence nor Green candidates (though, again, both combined) exceeded the Mondale-Coleman gap, and that's in a state that, for the moment, has both a senator and a governor from the Independence Party. NH Senate: the L is close to but not at the Sununu-Shaheen gap. It's actually striking how many Senate races were blowouts (look here for the tallies). There are R-D gaps of 5 points or less in only MN, MO, SD, NH.
Less of a 3rd-party collapse in gubernatorial races, suggesting that voters understood that their decisive Senate vote might become thedecisive Senate vote. For governor: L solidly bigger than the gap in Alabama; the Republicans can fairly think that the Ls cost them that one assuming even a slight R-over-D preference among southern Ls. Arizona governor: Mahoney twice the size of the gap, costing R Salmon the race-- though Mahoney's listed as an independent and I don't know where his support came from. California governor: Neither G nor L, though both combined, bigger than the gap. That means that California's not as much in play for Republicans as they might briefly think; they were hardly going to get that 5% G vote, and even the 2% L vote is likely out of reach because California Libertarians are a much more pro-choice, pro-gay, pro-drug group than are southern guns-and-taxes Libertarians. Maine: G (9%!) greater than the gap, but it's unlikely the losing R could have picked them up under any circumstances. Mass: it would have taken not only G+L but also the votes of "other" Johnson to make up the D-R gap. MN governor is a special case; Independence candidate and former Congressman Tim Penny exceeds the gap, but 16% isn't very impressive considering his resume and the fact that he's from the incumbent governor's party. NY: even had every last Golisano voter gone to McCall (unlikely!) it wouldn't have tipped the race. Incomplete results make it look likely that an L tipped Oregon governor to the Ds. An "other" tipped OK to the Ds. An "other" tipped Vermont to the Rs. L Thompson, Tommy Thomspon's brother, tipped Wisconsin to the Ds. The L is within a hair's breadth of the gap in Wyoming, but the D would have scraped it out even if every L had gone R.
A quick scan of CNN's list of the competitive House races: L greater than the gap twice, G greater than the gap twice, other greater than the gap twice; but Colorado 7 is all three, so these six thrid-party candidates were in just four districts. But G did well in a district that went D, L did well in a district that went R. Continuing to assume that Ls break more for Rs than for Ds, no L actually tipped a race. The Green tipped Colorado 7 to the Rs (though R+L>D+G).
G alone greater than the gap: Maine gov (but the Ds won). L alone bigger than the gap: SD Senate, Alabama, Oregon, Wisconsin gov (each of which the Ds won, meaning that the Ls at least arguably tipped them). Other or independent greater than the gap: Arizona Vermont, Minnesota, Oklahoma gov. No more than four House races, and probably only two, affected by the presence of third-party candidates.
The Dems seem to have neutralized the Green threat this time out. The Republicans still have a Libertarian problem; but unlike in 2000, it didn't cost them the Senate. It didn't even cost them their gubernatorial majority. The party Perot began and the one Ventura spun off from it are now basically dead.
Note: Most of the above makes the disputed-by-Greens assumption that all Green voters would otherwise be Democrats; (except where noted) the genuinely-false assumption that all Libertarian voters would otherwise be Republicans (I know this is false, because it was certainly false for me for Illinois governor); and the probably-false though I think not-terribly false assumption that all third-party voters would have gone to the polls, or could possibly have been motivated to go to the polls, in the absence of the third-party candidates. I’m well aware of the arguments around these claims. But from the perspective of major-party strategists, I think these are the working assumptions, i.e. Democrats think that every Green voter is someone who would vote for their candidate if there were no Green in the race.
UPDATE: The Alabama figures are in flux, and the R may pull it out.