Okay, Doug, I've got more detail here for you...
I'm not quite as well-versed with the polling process in a personal way as Iron is, but I do know my numbers, and I've been crunching some tonight (what a Friday night!), so I have some more thoughts.
First, let me start with this. There are 64 seats not up for election. 30 GOP, 32 Dem, and 2 Indies who always caucus as Dems. So 34-30, essentially.
Of the 36 seats up for election, I've got 26 of them as clear no-contests. Although I came to that conclusion differently, they are the same 26 races that Nate Silver has below the 90% threshold. They include Minnesota, West Virginia, and every other state below them on the list sorted by contestability. Those states shake out to 15 GOP, 11 Dem. So now we are at 45 GOP and 45 Dem, with 10 races to decide the whole thing: Alaska, Arkansas, Colorado, Georgia, Iowa, Kansas, Kentucky, Louisiana, North Carolina and New Hampshire. GOP needs 6 of 10 to win the Senate.
I base my analysis of these states below only on poll results released since October 1, weighted by date. I'll update with November polls between now and Tuesday. I ran three sets of numbers. One in which the undecideds/third party candidates don't break for either major party, one in which they break 2-1 Dem, and one in which they break 2-1 GOP*. This is a pretty simplistic look at the numbers compared to the complicated simulations run by Silver, et al., but, hey, I'm still running Windows 7. Also, I firmly believe that sometimes it's helpful to step back and just take a simple glance at things instead of getting bogged down in minutiae that ultimately don't amount to anything. Anyway:
Alaska - Impossible to predict because of lower number of polls and high variability. Raw contest virtually tied (GOP + 0.05), state breaks with wherever the national party trend goes.
Arkansas - Not all that competitive. Even with a Dem break, over 50% go to GOP.
Colorado - Although a Dem break keeps it close - and both candidates under 50%, this is a GOP seat in each scenario.
Georgia - Slight GOP lead, but goes with the national trend. In no scenario does either candidate get 50%. So this is going to a runoff.
Iowa - Goes GOP all three ways.
Kansas - Needs a GOP break to swing their way. Orman looking better than I thought.
Kentucky - GOP no doubt.
Louisiana - Not only GOP all three ways, but assuming at least some break away from undecideds to major party candidates, probably avoid a runoff.
North Carolina - Repubs need to cool their jets. Hagan isn't done by a long shot. Raw numbers have her up by a little less than a point.
New Hampshire - Sorry, Brown isn't winning this. Even with a GOP break, this stays Democratic.
Long story short? Best case scenario for Dems at this point is 46 seats + all 3 indies, + a chance to compete for that Georgia seat in a runoff. That's not ideal.
I'm sticking with 52 outright seats for the Repubs at the moment (everything above except for Kansas, North Carolina and New Hampshire), with Orman as #53 if he decides to caucus with them.
goat
* Actually, for the sake of Georgia and Louisiana, they actually break 2-2-1, with the other 40% staying with the third party candidate.
This post was edited on 10/31 10:28 PM by TheOriginalHappyGoat