Two weeks ago, I published my first diary, showing results from election model that I have been tinkering with for the US senate races.
This is an update, incorporating new polling data, to look at predicted outcomes for the senate. A common theme that you will see below is that this is going to be a get-out-the-vote (GOTV) election. There are no waves present (Roberts-KS would not be losing in and R-wave, Braley would not be under-performing in a D-wave).
First, the easy ones -- these are well outside of the margin of error, so I will call them safe:
Safe D (alphabetical, no numbers provided):
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
The rest (in order of reddest to bluest):
1. SD Rounds+7: No change in last 2 weeks-- no new polls.
2. KY McConnell+4: McConnell seems to have a stable single-digit lead. I expect a busy October on behalf of ALG -- she'll need to do something to shake up the race.
3. AK Sullivan +3: This has moved in Sullivan's direction based on polling, but there are 2 unknowns: (1) is the current polling really accurate in a notoriously difficult-to-poll state, and (2) how are the GOTV operations on both sides?
4. AR Cotton +3: This has moved in Cotton's direction based on polling. However, this is where it gets tricky. Likely voter polls are generally published now, and I tend to consider these biased. An early September CNN/ORC poll had Cotton+2 among likely voters, but Pryor+9 among registered voters. The question is whether GOTV operations in Arkansas will get those "unlikely voters" who strongly favor Pryor to the polls.
5. GA Perdue+2: Nunn is hanging in there. This is likely to go to a run-off, but I have no good prediction on what would happen during the run-off. I will offer that the Democrats seem to be well-organized in GA, and it bodes well for ensuring voters come back for the run-off(s).
6. IA Ernst+2: This one is painful. There is no way Ernst should be leading, but Braley is underperforming at this time (at least poll-wise over the last 2 weeks). The good news for Braley is that there are a number of events that could turn the tide. The first of 3 debates took place yesterday, though I don't know if it garnered any buzz. And a decent campaign over the next month should allow Braley to pull out the victory, as Iowa's fundamentals should give him an edge once voters tune in to the race.
7. LA Landrieu+0. Actually, I have Landrieu+4 over Cassidy in November, but not at 50%. So this will probably go to a run-off. Landrieu has pulled elections out in run-offs before. I list this at Landrieu+0 because I'm not making a run-off prediction, other than I think it would be very close.
8. NC Hagan+5. She has a solid, single-digit lead, confirmed by multiple polls.
9. CO Udall+6. The polls show this to be neck-in-neck, some even giving Gardner a lead. It should be noted that Colorado polls between 2006-2012 have had 3%-10% biases in the direction of the Republican in statewide elections. My model includes voter turnout assumptions (consider it state "fundamentals") that benefit Udall, despite the recent polling. He also benefits from ballots being mailed to all registered voters in October as well as an already top-5 turnout rate for CO voters.
10. MI Peters+8. Peters has a strong and consistent lead. He may start to pull away in upcoming polls.
11. KS Orman+8. It may seem surprising that I am predicting that Orman will win by this much over Roberts, but available head-to-head polling implies this (and more -- see note below). Even if Republicans come home to Roberts, I think Orman will win this race.
12. NH Shaheen+10 (this is just outside my 95% confidence interval, roughly implying that Shaheen has a 95% chance of winning this race.
Current Result for the Senate:
D/I: 49 (9/14: 52)
R: 50 (9/14: 47 -- swings from D to R are IA, AK, AR)
Tied: 1 (9/14: 1 -- LA still a pure toss-up)
Notes:
1. The data indicate that Greg Orman is way ahead in Kansas. I don't see how he loses, at this point (subject to any new data, of course). Why am I so confident? Well, among older voters, he has a commanding lead. A Fox News poll from Sept 14-16 showed him with a 52-38 lead among those 55 or older. The only age group where he was losing was under 35, where he was losing 39-49. Are the younger voters more conservative in Kansas, or are they just responding to a poll based on name recognition of Roberts? Hard to say, but 2012 exit polls do not indicate more conservative youth vote (no exit polls exist from 2010). My point is an important one -- Roberts losing by 14 points among older, more reliable voters, which should be his base. This is a big deal! A PPP poll from around the same time had similar results (Orman +16 over 46 year-olds or older, Orman +10 overall). Will Bob Dole turn the tide? Bob Dole may think so, but I'm surprised Bob Dole even cared to show up for Roberts. Roberts certainly didn't show up for him.
2. Apologies to a few of you who, after viewing my first diary, wanted more statistical modeling details in my next diary. I have pressures from my day job (including a re-structuring and potential takeover) that have prevented me from spending more time refining this diary. So I've just provided my own brief commentary instead.
3. I still hope to provide updates every couple of weeks until the weekend before the election.