Nov 06 2006
Where’s The Senate II
Updates At The End
About a week ago I did my first detailed forecast on the Senate using a bias correction for the polls that has been regular and systemic for years (before that I was intuitively bullish on a Rep pickup of 1-2 seats) . It is a step required on early polls since they underestimate GOP turnout and thus can miss the mark between 3-5%. Just look at this RCP comparison of Gallup and Mason Dixon polls and results in 2004 and you can see a 5% correction is not unreasonable on Gallup.
For consistency and simplicity I am going to include all the latest polls, average them and apply the bias correction to deweight polls like Gallup. As before I am going to place the races, after the bias correction, into Probable Rep, Toss Up and Probable Dem categories. I apply the bias in all races at the state and national level (as opposed to clear republican House districts) since the electorate is larger and much more diverse than in House races where gerrymandering controls voter pool make up.
I am using the RCP Senate lists and rankings. If the rankings have changed since the previous post I will note that with italic-bold text. Note, I list the races in the order of RCP ranking within a column. So the top race is the riskiest race and the bottom one is the most solid in any category
Probable Rep
3. RI: (Chafee – R) – Chaffey barely moved into the Toss Up category last time because the bias correction was barely making it a real race. Since then it seems that was the right call since Chafee has seen the most movement in all the key Senate races. If you look at the last three races you can see the outliar (pun intended) immediately. If I include this Zogby poll the average is a 5% lead for the Dem putting this in a tie. But the momentum alone would cause me to push this into the Rep column. If I do not use the Zogby poll and use the correction we see a Chafee lead of 3+. I think it is safe to say this one is trending to the Reps.
4. VA: (Allen – R) – RCP had this as #7, and have been unreasonably negative about Allen. As a Virginian I have found the RCP view of this race a bit biased. Anyway, the fact is Allen is not in much trouble. The three latest polls (+3, tie, -1) show a stastical tie. Applying the correction would give Allen the classic VA 52.5-47.5 win. But if we look at trends, Rasmussen went from -5 to tie in less than a week! At that rate Allen will be 4+ by tomorrow. Rep Hold.
5. MO: (Talent – R) – If you look at the last three polls you have an average of just over 1% lead for McCaskill. If you apply the correction we get a Rep win. But again, if you look at the trend in Rasmussen we see a -1 deficit going to a +1 lead in the span of a few days. I would like a sharper trend line, but I think this will go Rep. I would have this race right behind Chafee as one tottering between a toss-up and a Rep pick up.
6. MT: (Burns – R) – Was #4 in the RCP ranking last time. This race has lots of momentum towards Burns. Given the rate of change in the polls it is clear this one is heading right. Applying the bias correction gives Burns a chance to win. But if you look are Rasmussen, Zogby and Mason Dixon all three show a shift. So I see this race as a Rep Hold.
9. TN: (Open – R) – Was #8 in last time I did this. And since that time Corker has been moving out smartly. Applying even a modest correction (which should be done) makes this another embarrassing loss in TN for the Dems (remember Al Gore?).
Toss Up
2. OH: (DeWine – R) – Last time I said if Dewine started closing I would have to put OH in the Toss Up category. Well he started closing. RCP has one of those wild outliars (pun intended) with a 24% blow-out for Brown even RCP cannot bring themselves to use. But if you look at the last two polls you get a 6.5% lead for Brown. Add in the bias correction and it drops to 1.5% or a statistical ties. Also, if you look at trends in a single poll you see the lines heading to DeWine, so their is momentum in one direction. There are strong headwinds against DeWine, but he could pull this out of the GOTV effort is as good or better than 2004. And supposedly it is.
8. NJ: (Menendez – D) – Was RCP #6 last time. While RCP moves away from this race I see it closing. There are a wealth of polls out and the RCP average has it going for Menendez by 6.5%. If you toss the Gallup outlier we get 5%. If there is a bias correction of 5% (and that is a big if in NJ) then that would put it into a tie. The reason I think it may really be in a tie is the Manmouth/Gannet poll showing the Dem lead down to 3% (which mustr be in the margin of error). I think the Dems history of corruption will get them in the end as the GOP GOTV mobilizes and the Dems sit back in disgust.
Probable Dem
1. PA: (Santorum – R) – Sadly, not much has changed in this race. Santorum has been at risk for a long time. If you average the last three polls he is trailing by more than 10. Even applying the bias correction only get him within 5%. As of now I think Santorum will lose slightly.
The other races so clearly staying with the incumbents it is not worth the effort to even discuss them. And that includes Lieberman’s cruise to re-election. So where does that leave us? 1 Dem Pick Up, 2 Toss Ups and 1 Rep Pick up. If the Dems and Reps split the Toss Ups along Party lines (or trade them) that leaves a 0 Seat loss in the Senate. That would crush the Democrats because this was the cycle where most seats up for election were Rep seats. If the tides switch and the Dems pick up all the Toss Ups then make that a 1 seat pick up for the Dems. If the tide goes towards the reps there a 1 seat pick up for the Reps.
The cone of possible outcomes could be broader. One more of those races I put in the Leans Rep column after the bias correction could go Dem. That would give the dems a measely two seat pick up. Well below the average 6th year itch were the President’s party loses 6 Senate seats. Now are these predictions accurate? I cannot say. I have been struggling to avoid a pollyannish approach like those from Cook and Rothenburg. There is no way the Dems pick up the Senate – I am sure of that. And if the Dem wave never existed because pollsters were not measuring the full electorate, and there is a Rep wave out there they missed because they never measure the true Rep intensity, then there is a possibility for a stunning day tomorrow. It all depends on who goes out and votes.
Update: Clarice Feldman has noticed an interesting last minute PA poll (with some questions surrounding it) which shows Santorum within 4% points. I will have to say Santorum is the longest long shot here – but nothing is for certain this election.
Very true. I’ve always felt Santorum was the only “goner” among the GOP Senate.
Can we trade Chafee and Santorum? Please?
A McCulloch Research poll conducted over the weekend suggests the senate race in PA is tightening. It reports State Treasurer Bob Casey, Jr. leads Senator Rick Santorum 48.1-44.1% among “very likely voters.” There’s an interesting breakdown of how the election is shaping up regionally, and while McCulloch maintains the race is still Mr. Casey’s to lose, he believes that Senator Santorum is now within striking distance. He even uses the word “upset.”
Given the dismal flatness of the PA polls over the last eight weeks, the only factor that explains this shift is the John Kerry Factor. After Senator Kerry (D-EU) made his unfortunate “gaffe” last week, Mr. Casey, secure in his lead, played the party loyalist and praised Mr. Kerry’s leadership and patriotism. Mr. Santorum pounced, released a powerful new ad (view it below), and may well be getting some traction at very long last. As the Senator is fond of saying, “My opponents win polls. I win elections.” He speaks with the wisdom of experience and while this is still a long shot folks, if PA is back in play tomorrow’s results remain anyone’s guess despite what you might read in the Washington Post or the New York Times.
http://www.redstate.com/stories/archived/is_santorum_getting_a_kerry_bounce
Well–Not to get your hopes up too much–Red State has this comment:
Early this morning, Keystone Politics editors received and released a poll by McCulloch Research and Polling showing that Rick Santorum was within 4 points of retaining his Senate seat. Further research into McCulloch Research and Polling shows that Rod McCulloch, principal at the firm, has been indicted in voter fraud and forgery in Illinois.
Interesting! I think I have a crush on Clarice and I have no idea what she looks like.
Oh, and your anaysis is interesting too.
Well, Keystone has pulled the McCulloch poll because the background of the pollster is fishy. Hope that doesn’t change your affection for me, but truth it truth.
Update on Santorum..NRO Corner reports that Rendell’s internals show Sanotrum clsing.
In his interview with HH, Rove iindicates surprising strength in Michigan. (Rick Ballard, Flares into the Darkness spotted this a couple days ago.)