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."
I don't really, have good definitions for the terms, but to me safe means a seat isn't in play, and absent a huge wave, the challenger has less than a 5% chance to win. Likely is like an 85% chance. In other words, the formula is more exponential than yours seems to be.
Safe Democrat = 0% (50% from Toss Up)
Strong Democrat = 25% (25% from Toss Up)
Lean Democrat = 37.5% (12.5% from Toss Up)
Tilt Democrat = 43.75% (6.25% from Toss Up)
Toss Up = 50%
Tilt Republican = 56.25 (6.25% from Toss Up)
Lean Republican = 62.5% (12.5% from Toss Up)
Strong Republican = 75% (25% from Toss Up)
Safe Republican = 100% (50% from Toss Up)
How does this look?
-PJ
Better. I should have divined it without your help, but I have my senior moments, and I hate when that happens (the wraith of the conservator becomes ever nearer at hand). I would put strong at 85%, but other than that, I agree that your percentages resonant in my pea brain. Great job on your part. Kudos, for what it is worth from this end!
-PJ
FL-13: This one's a bit of a dilemma for me. It's one of my lower confidence ratings (OH-02 also pops to mind in that regard). The buzz has definitely grown on this one but I'm somewhat perplexed as to why. The only independent poll that I'm aware of was the Constituent Dynamics that put Jennings at a slight 47% to 44% lead. But quite significantly, the C-D poll was conducted Oct 8-10, at the peak of the Foley clamor in the neighboring FL-16 district. Otherwise, there are a trio of polls showing a daunting lead by Jennings, but they are all internal polls and I must all but disregard them.
Then, not only was this a 55.8% GWB district in '04, but there's the major financial advantage for the self-funding Buchanan, as you noted. That's hardly chump change: Buchanan gave his campaign $975,000 today 10/27 on top of $800,000 on 10/20; he'd already spent a somewhat astounding $5,926,098 as of 10/18. Meanwhile, Jennings had just $176,901 CoH as of 10/18, so she'll be outspent on the order of about 7 to 1 in the closing two weeks.
In short, it's unclear to me why you'd think Buchanan is behind and probably outside the MoE. Is there something important I've missed? It does happen! :)
NV-03 & NV-02: The DCCC isn't playing in NV-03 but the NRCC has spent $395,575 there in the past week. The NRCC spent $245,553 in NV-02 in the same timeframe (10/20-10/27). BTW, I'll follow up with a post linking to where you can quickly look up such figures. Anyhow, there's also the obvious difference in partisan lean: NV-02 was 57.2% Bush in '04 while NV-03 was 49.9% Bush in '04. Another big factor in my ratings was that Hafen led Porter in the 9/30 CoH figure which I used for that round of ratings. As of 10/18 Porter led $225,171 versus $181,573 for Hafen, but it's noteworthy that between 9/30 and 10/18 Hafen nonetheless outspent Porter by $407,339 to $60,958, so the $44,000 10/18 CoH differential will hardly make up the difference. Porter has also been hit by allegations of ethical lapses (Google him) and by ads claiming that he voted to cut military/veterans benefits.
Meanwhile, Heller leads Derby by $251,991 CoH versus $123,192 CoH as of 10/18 reports. In October Heller outspent Derby by $255,058 versus $219,391 and has outspent her in total by $1,237,515 versus $1,123,631. A late September Mason-Dixon poll had Heller leading 45%-42% while Research 2000 had Heller leading 45%-37%. Getting outspent in a district where she's at a 10% partisan disadvantage is not the way for Derby to overcome Heller's polling lead, even if that lead was weak. In a political environment where Dem crossover voting seems to have all but dissipated, I think the district partisanship outweighs the advantage of incumbency so that NV-02 is a bit firmer than NV-03. That being said, I am still rating both with a decisive GOP lean.
NH-02: I can't think of any good reason why Bass would've suffered such a spectacular collapse unless there were a tsunami sweeping across the GOP nationwide and if that were the case we'd be seeing it in polling from all over, and we're not. So, my assumption is that the Becker Institute poll is whacked. To begin with, I would note that I've largely broken free from the polls in this revision to a more 'holistic' analysis of the various districts. Regardless of that, my assessment is that Bass probably had the hefty 20% or so lead that was registering as of late summer, with his support around 50% and Hodes about 30%, and the undecideds generally inclined to vote Dem. Since then, I think UNH probably revealed an genuine autumn tightening to a 10% margin, with Bass at 46% and Hodes at 36% in late Sept. Now I'd say it's probably inside of that but nowhere near the Becker figure. In short, I think Bass likely still has a firm edge and that it's somewhere around 5%-8%. So, I'm basically rejecting the Becker poll, rightly or wrongly.
TX-23: I am effectively assuming that a 51.5% or so GOP district is very unlikely to give Bonilla an outright majority in a 'jungle primary' this year, and that the $700,000 that Gilliland has spent on his campaign should easily get him into second place (and the presumptive runoff). As for polls, the only one I know of is the Aug 25 Dem poll by Greenberg Quinlan Rosner that had the combined Dem total at 47% and Bonilla at 44%. Gilliland was not included in that poll, and the mildly ridiculous Ciro Rodriguez led the Dem pack at 24%.
OK, I think I've covered everything you asked about. Sorry it took longer than I thought to get back to this!
With regard to the Keystone Poll, when I noticed that it clearly favored the GOP this year by comparison to other polls I went back to look at 2004 and also noticed that in that cycle it slightly favored Dems. I don't really know what to make of that, except to speculate that perhaps they overcompensated somehow in their methodology. Who knows? In general, I've not much worried about it and simply averaged in the Keystone Poll with all the rest when assessing the various contests, but with a relative dearth of polling in PA-08, and with the Keystone Poll at some variance to the latest CW on that one, I have to make an executive decision to minimize its influence on my rating. :)
With regard to FL-16 I've already been rating that one below 'consensus' and with the presumption that most voters who want to vote for Negron will know that a vote for Foley is a vote for Negron. The problem, as I've stated previously, is that I think FL-16 would be a Toss Up even if it were just a 'normal' open seat contest. Another way to look at it is that FL-16 was 1.8% less GOP than FL-13 in 2004, and 1.4% less GOP in 2000. If the open Harris seat can be a Toss Up then certainly an open Foley seat could've been so. But again, I have not leaped to the conclusion that FL-16 is a sure loss which is why I have it rated last in the Lean Dem seats.
With regard to the 50% rule, that's a controversy I'd just as soon avoid! I'll simply add that if there's any election where it should be applicable it is this year's prospective 'wave' election, so if there's any merit to it I think we'll get a clear sense of that once the election results are analyzed. That takes us back to the $1,000,000 question of whether this year is a 'normal' election or a 'wave' election. We'll certainly find out soon enough!
The problem with the Meier ads is that they revolve around what is 'wrong' and what needs to be 'fixed' so to speak. In short, if voters think that the country is on the wrong track, and Meier's ads appear to implicitly concede as much, then it's difficult to see why they'd vote for a Republican over a Democrat (if they're a 'swing' voter) or why they'd be inspired to turn out (if they're a 'weak' GOP voter). I also think the donut ad is silly. When candidates have to resort to sneaking donuts from the wife or to how their opponent would kick their dog if they had one (Harold Ford Jr) you generally know they have serious problems (and are trying to avoid directly addressing the issues at hand).
But, to be sure, my reaction to the ads is ultimately subjective. The idea though is to get a sense of the 'issue terrain' where the campaign is being fought and whether that inherently favors one candidate over the other, regardless of the quality of the ads. Since both Arcuri and Meier are focusing on 'change' - with the added undercurrent of countering negative ads on the part of Meier (i.e., "bad habits") - I think that favors Arcuri since he's more the 'agent of change' so far as that district, if for no other reason than that he's challenging the incumbent party.
I think you may have overanalyzed the ads. I don't see them as changing many votes myself. Sure a marginal seat in NY that is open = Dem victory this year, but one has to know more about the pull and reputation of the candidates on the ground, how they interact with votes, and what the local rags are saying, and whom they endorse, and whether they are Dem or GOP rags in general in the past. Absent that, we go with the polls, the partisan history, the wave, and the money tree. That is usually enough, except when it isn't.
Those links will give you the itemized expenditures listed by state/district from October 1 to October 27. Also included with each district is a sub-total of all spending by the committee for that time frame.
If you look at each link, you will see a string as follows:
&Count=20061001&sYR=2006&sEndDate=20061027
the underlined part is the date range. So, if you want the last week instead of the full month of October you would simply change 1001 to 1020 and that will make your start date October 20th instead of October 1st, with the same end date of October 27th.
Hopefully I've explained that well enough! There's probably an easier way to get the data, but that's how I've been doing it.
OK, you're probably right!
On FL-13: In the link to one of Jennings' internal, I found this remark, which fuels my suspicions.
http://www.bradenton.com/mld/bradenton/news/local/15848966.htm
"Buchanan dismissed the results, saying his internal polling shows him trailing "within the margin of error" of Jennings. He did not divulge details, and has not publicly released any of his poll results since the September primary."
I buy your rationales on NV-02, NV-03 and NH-02 and thanks for the link on the monthly DCCC, NRCC party stats. I often look at the day-by-day stuff, but I never could figure out how to get the monthlies. (my computer expertise sometimes leaves a lot to be desired) Also, the Becker Institute was a terrible NH polling company in 2004, so I can understand that.
On TX-23, I really think you should be counting Bonilla's number @ around Bush's 2004 number, not the base Republican number, simply because he's perhaps the one Republican candidate in Texas who can pull similar numbers at this time and place. (even though I know his Latino figures have been declining over the years) I also think the Republican breakdown in this area is going to be less than it is nationwide (just a gut feeling).
The real question I have in this "jungle primary" continues to be (and I can't answer it): Even if you vote a straight-party ticket, you still have to vote in the special election. How is this going to affect the numbers?
In a normal election, roughly 30%-35% of the voters in Bexar County vote straight party. These voters tend to fairly evenly divided (considering the usual low turnout among Democrats in south-side Bexar). The rural counties exhibit much less straight-party voting patterns on the whole.
For that reason, I think polling of this race is next to impossible and the poll we've both seen is stale anyway.
This also reminds me of a big Gilliland problem. Those South Siders that were added to the CD are simply not going to vote for a white guy at the local level, regardless of party. He's going to have to pick up votes from the border areas or Bonilla strongholds in order to make a runoff, imho.
I still think he makes it past a runoff, but that's just me. I doubt we really know for sure until Election Day.
That's interesting about Buchanan and certainly does suggest that FL-13 is slipping away. Hmm. I think I'm gonna switch FL-13 and MN-06 in my ratings above.
http://tray.com/cgi-win/x_IndepExp_SQL_By_State_Date.exe?DoFn=&Count=20060822&sYR=2006&sEndDate=20060906&sID=C00075820
As should be clear, you simply need to change the dates in your browser to whatever you want to see. Here's the segment with the date string again:
=&Count=20060822&sYR=2006&sEndDate=20060906&sID
I've underlined the part I changed.
Thanks for making sure on the communication, but yes I did get how to manipulate the links to get the specific date periods. It is very helpful, to put it mildly.
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.