Posted on 10/25/2006 11:59:44 AM PDT by Torie
In a new research paper, three political scientists attempt to use the results of generic congressional polls to predict the outcome of the midterm elections.
"Via computer simulation based on statistical analysis of historical data, we show how generic vote polls can be used to forecast the election outcome. We convert the results of generic vote polls into a projection of the actual national vote for Congress and ultimately into the partisan division of seats in the House of Representatives. Our model allows both a point forecast-our expectation of the seat division between Republicans and Democrats-and an estimate of the probability of partisan control. Based on current generic ballot polls, we forecast an expected Democratic gain of 32 seats with Democratic control (a gain of 18 seats or more) a near certainty."
Marking for the eventual gloat.
Did they say anything about tomorrow's weather?
OH good. This must mean that Republicans will win!
Did you look at the ads on the respective campaign sites? I would like to see them.
I don't think that there is any chance that OH-13 (or OH-06) will vote GOP when the Republican candidates for governor and senator are headed for a double-digit defeat statewide. More significantly perhaps, the NRCC is making no effort in OH-13 and Sutton has outspent Foltin by better than 2 to 1. The only independent poll I know of back in mid July had Sutton leading Foltin by 48% to 30% and there's little reason to think the gap has narrowed, especially when the Ohio GOP seems to be in worse shape now than it seemed then.
Yep, that's correct. I recently watched all the ads I could find for all the New York seats as I've been trying to get a better handle on where those contests stand. Here are the Arcuri ads and here are the Meier ads.
This will be fun to watch the surprise on their childish faces
Thanks for the explanation.
I'll ask a few questions on a few of the races, out of curiosity. Once again, it's really an excellent job.
I do think that, out of all the states w/competitive races, NY is the one that raises the most questions in my mind, because of the up-ballot GOP massacre. Yet, at the same time, the polling is just so questionable. It's kind of obvious that C-D's NY-19 and NY-20 polls were dead wrong (based on later internals and Indy polls), which leaves me unsure of the rest of their polls in the state. Yet, I really don't know what to make of NY-26 either and SUSA's polls of the race.
Anyway, enough of my own commentary. Now for the questions:
10. PA-06: What do you make of the fact that Murphy's own internal only puts her 3 in front of Gerlach? Personally, I have the race in toss-up towards the top end.
18. WI-08: This is a race I don't get, though it's obviously close. What's your thinking here?
23. MN-06: How do you think the SUSA poll of a few days ago (showing Bachmann up 6) fits in with the overall contest? I've moved the race down a bit on my own list.
27. PA-08: The NRCC doesn't seem to have this on their list for spending and yet we have a Keystone Poll out today showing Fitz up 9. Your thoughts?
28. FL-13: I'm guessing this one's in toss-up because of the money Buchanan's spending. I personally think he's behind, outside of MOE, and place this one much higher on my list. Your thoughts?
On 38 and 41, I'm curious which one you think is more competitive? I know that Novak has placed NV-03 above NV-02 and yet I'm only seen the national campaigns spending money on NV-02.
40: NH-02: Opinions vary widely on this one, including my own, especially after the questionable uni poll showed Bass down 8. What's your thought?
56: TX-23: The race I always ask about. I've noted that Gilliland got Bolanos endorsement and is basically the only candidate running advertising that I've heard about. Are there any polls on the race (I haven't seen any), or are you just assuming that the multitude of candidates plus one actually advertising and the anti-straight party vote structure, along with the lean of the CD is going to push this one into a runoff?
Both guys' ads seem pretty good, although I have not reviewed them all.
Here is the quote: "Nevada-3: Rep. Jon Porter (R) is now under water in his suburban Las Vegas district, caught in a tough race against Sen. Harry Reid's (D) press secretary Tessa Hafen (D). There is much fear that Mormon Republicans will pull the lever for Hafen, their co-religionist."
It's those darn Mormans again. :)
OK, here goes!
PA-06: The highest that Gerlach has polled all cycle long was 47% and that was in his own internal POS poll. In indy polls Gerlach has polled at 46%, 41%, 44%, and 45%; Murphy has polled at 52%, 43%, 41%, and 50%. We all know that Gerlach barely eked out wins in both 2002 and 2004, and we all know that Murphy got 49% in 2004. It is implausible to me that Murphy will do worse in 2006 than in 2004, and one thought that I've had lately is that everyone seems to be ignoring the incumbent-under-50% rule. To top it off, as of Oct 18 Murphy had $729,717 CoH versus $555,960 CoH for Gerlach, and Murphy has outspent Gerlach by $2,827,042 versus $2,684,384.
WI-08: This one is fairly difficult for me to grasp neatly as well, but Kagen has outspent Gard by $2,140,887 to $2,028,258 and more importantly is a self-funder. Since the 10/18 reports Kagen has given his campaign a further $770,000. That's what tips the scale to Kagen in my ratings. My sense is that the race is essentially tied (Kagen actually has a slight lead within the MoE in the few polls) and so the money advantage combined with a floundering GOP campaign for governor and the national mood tips it to the Dems.
MN-06: The October polling average is 45.4% Bachmann to 45.6% Wetterling. I am rating the seat accordingly with a tip to Bachmann due to the partisan lean of the district. Wetterling continues to have a cash advantage ($783,552 v $563,653 as of 10/18) so that's what also keeps things close and in Toss Up in my estimation. I did however seriously consider rating MN-06 below the quartet of freshmen below it and do not object to the idea myself. I also think the race may be firming toward Bachmann, so if there's more evidence of that I'm quite open to moving it so far down as slight Lean GOP (around current FL-22 range).
PA-08: I posted my ratings before I saw the latest Keystone Poll released late last night. I would likely have notched PA-08 down just a bit if I'd seen that first. However, I would note that the Keystone Poll tends to favor the GOP and that Fitzpatrick polls below 50% at 48%. The two are about even on CoH ($490,090 Fitzpatrick; $448,884 Murphy) and the district leans Dem. That latter is a big part of why I've moved Fitzpatrick up to Toss Up from Lean GOP: The Rendell machine should be in full force in the Philly suburbs and Santorum certainly isn't going to be much help either. The NRCC has not written out PA-08. These are their PA expenditures in the past week: PA-04, $140,607; PA-06, $703,893; PA-07, $840,112; PA-08, $782,450; PA-10, $177,331. If anyone's been written off it's Sherwood; and Fitzpatrick is clearly still thought to be in need of help.
I really have to run now! I won't be able to post on the remaining questions until sometime this evening, but I'll definitely get to them.
I'll explain my reaction further tonight, but I gotta go or I'll be late for a lecture!
I see nothing to really disagree with on your points.
I would remark that the reasons why I think the incumbent under 50% rule has been ignored more of late is because of the great amount of data that's come in over the past few years that shown that since the early 1990s, the trend has been towards undecideds breaking close to even between incumbent and challenger, as opposed to years prior to that period.
Oh, and about the Keystone Poll, it appears to be your standard uni poll. Strangely enough, its results have favored the Republican in comparison to other polls this year, but in 2004, its results skewed Democratic compared to other polls and the final result. The final result was Kerry +5. Just some food for thought. I just usually consider this to be your "crappy uni poll" bias.
Oh, among the news items, I thought you might like to know for your list that in FL-16, the Republicans will be allowed to put signs up at polling places saying "a vote for Negron means Foley". Only the Secretary of State is not allowed to. Apparently, Democrats are not appealing the decision either.
http://blogs.tampabay.com/buzz/2006/10/signs_going_up.html
From there, I applied your ratings, overriding others where it occured (see Notable Differences below).
For reference, here's my scale. I'm using a progressive scale (from the Republican perspective) where each category is twice better than the prior, where "better" is defined as distance from Toss Up. The scale is as follows:
Toss Up = 50%
Tilt = 6.25% better than Toss Up
Lean = 12.5% better than Toss Up
Strong/Likely = 25% better than Toss Up
Safe = 50% better than Toss Up
The scale ranges from 0% (Safe Democrat) -> 50% (Toss Up) -> 100% (Safe Republican). I've been running 100,000 simulated elections.
Notable Differences (FWIW)
OH-15 -- You have it as Lean Democrat, Barone has it as Lean Republican.
PA-06 -- You have it as Lean Democrat, Barone has it as Lean Republican.
NC-11 -- You have it as Toss Up, Barone has it as Sure Democrat.
OH-18 -- You have it as Toss Up, Barone has it as Sure Democrat.
CT-02 -- You have it as Toss up, Barone has it as Sure Republican.
VA-10 -- Who's afraid of Virginia Wolf? sorry, I couldn't resist.
KS-2 -- BUG - you have it as both Likely Republican (Ryun) and Likely Democrat (Moore). Moore is KS-3.
Results
Running 100,000 simulated elections with your predictions and my scale, the expected value of Republican seats is 208.5, a net loss of 23.5 seats. The probability of attaining 218 seats (majority) is 1.6%.
Comments
There are 12 Republican seats with less than 50% probability of retaining, plus another 14 seats that are Toss Up. No Democrat seats are Toss Up or leaning Republican. Does this make sense to you?
If you have other numbers that you want me to factor in, let me know.
-PJ
-PJ
That's very cool. Thanks for posting! Your QC figure of 187 safe D and 162 safe R seats is correct.
Dems take the House.
I dread the next two years.
The percentages below toss up appear too low to me, particularly past lean.
Can you elaborate? I'm not sure I follow what you mean.
It's a bit fuzzy, to be sure. Pundits are using words like Tilt, Lean, Likely/Strong, Safe, but what does that really mean?
I decided against a linear scale, such as 50-60-70-80-90-100 because I didn't want to get into "lean and a half" type distinctions. I wanted something that clearly differentiated the categories. That's why I came up with the doubling scale, where each category is twice better than the prior.
I also played with a scale along the same lines, but had Tilt at 5% (giving a 5-10-20-40 range), which left nothing truly safe -- that is Safe Democrat was 10% and Safe Republican was 90%. This implied that one in every 10 elections could result in a loss (last-minute gaffe)?
Do you have some ideas on how to quantify the verbal scale that people are using?
-PJ
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.