This diary is a continuation of my diary series first published here which was originally intended to present results from an election model that I had been tinkering with for the US senate races.
I am abandoning my model for a couple of reasons:
1. There are other sites (538, Sam Wang's Princeton Election Consortium, right here on Daily Kos, etc.) that do a great and professional job at aggregating the data. While I am a statistician, I don't think I can compete with those sites.
2. I am somewhat frustrated by the available polling data. I believe that likely voter (LV) results are subject to inherent biases that are not easily identified or quantified. In particular, I see 2 primary sources of bias -- the likely voter selection criteria and the voter composition estimates. Registered voter (RV) results should be published side-by-side along with LV results because they are only subject to the latter type of bias. Unfortunately, few if any polling outfits are providing the RV numbers at this stage. I think that using the LV numbers alone is potentially misleading.
3. I think "get-out-the-vote" (GOTV) efforts do not get fully reflected in polling results. As implied above, a good GOTV effort would bring "unlikely voters" to the polls, again rendering LV poll results misleading.
So what I would like to do in this diary is present my opinion of the Senate races, based partially on current polls, but also on subjective considerations as well.
First, those that I will consider "safe" picks:
Safe D:
DE, HI, IL, MA, MN, NJ, NM, OR, RI, VA
Safe R:
AL, ID, ME, MS, MT, NE, OK1, OK2, SC1, SC2, TN, TX, WV, WY
Now I will present the rest, in my subjective (descending) order of confidence and guesstimated margin of victory:
Very Confident:
1. MI Peters (D)+10. He has built and maintained a strong and consistent lead.
2. NH Shaheen (D)+10. Despite some recent polling showing a close race, I don't think it is really that close. Brown is not the intriguing, fresh candidate he was a few years ago. And not remembering what state you are running in on multiple occasions does not engender confidence with your constituents.
Fairly confident:
3. NC Hagan (D)+5. She has a solid, single-digit lead, confirmed by multiple polls. While it may be close for a while, I don't think Tillis has the favorables that will motivate voters to make this a closer race.
4. KS Orman (I)+5. Even though recent polling has shown this race tightening up, it's one thing to answer a call and give your opinion to a pollster, but another to actually get up off the couch, get in your car and go vote. I think Republicans in Kansas this year will not be very motivated to go to the polls. In addition, some polling has shown Orman (and Davis) doing better among the 50+ year-old, more reliable voters.
5. KY McConnell (R)+3: While I'd love to see ALG beat McConnell in this race, it is more likely that McConnell will pull it out, though I expect it to be too close to call for quite a while. I think McConnell's nonsense has fooled some low-information voters (you can keep KyNect, even if ACA is repealed). It is still the turtle's race to lose as of now.
6. CO Udall (D)+3. The polls that now give Gardner a slight edge should be taken seriously, but so should Colorado polling's clear R-lean between 2006-2012. With Colorado going to a full vote-by-mail approach this year, with all currently registered voters automatically receiving a ballot in the mail, turnout should be excellent. Gardner's lead right now is about where Buck's lead was over Bennett in 2010, and I don't think pollsters have changed their methodology to take into account CO's high turnout rate or Hispanic population.
Not very confident (could go either way):
7. AK Sullivan (R)+2: There are competing forces at play here -- the fact that previous AK polling data has tended to overestimate the blue turnout vs. the ground game that Begich has built with rural staffers and his "True Alaska" campaign. One of these forces will win out, but which one is unpredictable at this time. So far, I think the race is leaning towards Sullivan.
8. IA Braley (D)+2: Polls are indicating a small but persistent lead for Ernst. This cannot be simply put aside. However, the early vote is in Braley's favor, and Iowa has one of the best Democratic ground games, thanks in part to OFA and Obama's early organizing there. Braley's "farmer" comment early in the race has clearly hurt him, as the race shouldn't be this close.
9. AR Cotton (R)+2: This has moved in Cotton's direction based on polling. However, when some RV polls were still being published in September, Pryor had some apparent leads. Therefore, the question is whether GOTV operations in Arkansas will get "unlikely voters" who lean strongly in Pryor's direction to the polls.
10. GA Nunn (D)+2: The sense I am getting is that Perdue's campaign is imploding at just the wrong time for him with his "outsourcing" comments and doubling-down on them. Nunn is capitalizing on it, and may get to 50%+ with a final surge. However, a 3rd party candidate (Swafford) complicates this a little bit, doesn't it? A 50-48-2 results (assuming it is 50% + 1) is certainly possible, but so is a 49-48-3 outcome, pushing this to a run-off. While the conventional wisdom is that Perdue would win a run-off, I think it is too early to tell (polling after Nov. 4 could be affected by whether one party or another already has reached a clear majority, for example).
Upset Specials (More likely that the Republican will win (with small level of confidence), but I will root for the underdog to the end):
11. LA Landrieu (D)+1. While Landrieu is likely to win a plurality of votes in November, it is unlikely that she will get to the 50% threshhold. She has a good record of actually picking up additional votes in run-offs, so if it does go to a run-off, I think it would be closer than expected.
12. SD Weiland (D)+1. Going back 50 years, South Dakota has elected as many Democratic senators as Republican, including high profile ones such as George McGovern and Tom Daschle (in addition to current senator Tim Johnson, who beat John Thune by 500 votes in an off-year (2002)). Polling over the next 1-2 weeks will help indicate whether this is truly a 3-person race between Weiland, Pressler, and Rounds, or whether one will fade down the stretch (or pull ahead). If one does fade/pull ahead, it is rather unpredictable right now who it would be -- an argument could be made for (or against) each candidate. For someone interested in game theory, I have to say that this is easily the most interesting race right now. A 3-person dynamic is very different than a 2-person dynamic because a potential voter is not just thinking of who they want to win (or lose), but also perhaps considering - would I vote for my 2nd favorite guy if I thought they were the most likely one to win? Or "I'm kind of 50/50 on these 2 candidates -- how do I choose?"
Overall Results (fairly to very confident):
D/I: 49
R: 45
Tossups: 6
Result for the Senate (no upsets, no tossups):
D/I: 51
R: 49
Result for the Senate (with upsets):
D/I: 53
R: 47