Latest Post | Last 10 Posts | Archives
Previous Post: Nieukirk goes national
Next Post: Daily Herald/ABC7 poll: 48-32-12 *** Updated x1 ***
Posted in:
For the second time in a row, Constituent Dynamics has released a poll on the 10th CD that doesn’t jibe with any other numbers.
Two weeks ago, they had Congressman Mark Kirk ahead of Dan Seals 46-44. Nobody believed it. They also had Melissa Bean in a tough race and nobody believed that, either, until the Daily Herald poll suggested she was in trouble.
This time, they’ve got Dan Seals ahead of Congressman Mark Kirk 48-46. Gitcher crosstabs here. Click below for a larger image.
[Hat tip: ArchPundit]
*** UPDATE *** I posted this in comments, but thought it should be elevated to the main page. Here is the methodology this outfit uses:
Majority Watch, a project of RT Strategies and Constituent Dynamics, comprises telephone surveys conducted using interactive voice recognition (IVR) technology.
In each of 30 contested congressional districts, a representative sample of 5,000 likely voters based on previous vote history was selected from state voter registration files. 1,200-1,300 interviews were completed in each district […]
Completed interviews were compared to the voter registration file to validate results. Interview were discarded when the surveyed demographic responses did not match the voter file information of the sampled likely voter. In this manner, approximately 1,000 interviews were validated for each district.
Final results were weighted to represent the likely electorate by demographic factors such as age, sex, race and CD geographical subarea. No more than one time in 20 should chance variations in the sample cause the results to vary by more than 3 percentage points from the answers that would be obtained if all voters in a CD were polled.
posted by Rich Miller
Monday, Oct 30, 06 @ 5:32 pm
Sorry, comments are closed at this time.
Previous Post: Nieukirk goes national
Next Post: Daily Herald/ABC7 poll: 48-32-12 *** Updated x1 ***
WordPress Mobile Edition available at alexking.org.
powered by WordPress.
Whoever paid for the poll should stop payment on the check. Not sure of this organization so maybe they are paying for it themselves. Either way they are fooling themselves if they think these numbers are correct.
Comment by Get Your Money Back Monday, Oct 30, 06 @ 5:39 pm
Isn’t this an online poll? Can it really even be considered scientific?
If you look at the national results, they have Democrats polling far better than expected or how DCCC poll show in many races, yet they have Chris Shays in CT suddenly pushing out to a comfortable lead at the same time.
It just sounds like a bogus method of polling.
Comment by Establishment Republican Monday, Oct 30, 06 @ 5:45 pm
It’s not an online poll, but the methodology has been questioned by some.
Comment by Rich Miller Monday, Oct 30, 06 @ 5:47 pm
Of course this poll is incorrect. If you support Kirk relax, this election is in the bag. There is no way this Seals guy can win. Take it easy, maybe even consider taking a holiday. Your vote really is not needed in a blow out like this election. And if you see other polls begin to show a tightening race, don’t believe them for a minute-This is a done deal.
Comment by Garp Monday, Oct 30, 06 @ 5:49 pm
Methodology:
===Majority Watch, a project of RT Strategies and Constituent Dynamics, comprises telephone surveys conducted using interactive voice recognition (IVR) technology
In each of 30 contested congressional districts, a representative sample of 5,000 likely voters based on previous vote history was selected from state voter registration files. 1,200-1,300 interviews were completed in each district […]
Completed interviews were compared to the voter registration file to validate results. Interview were discarded when the surveyed demographic responses did not match the voter file information of the sampled likely voter. In this manner, approximately 1,000 interviews were validated for each district.
Final results were weighted to represent the likely electorate by demographic factors such as age, sex, race and CD geographical subarea. No more than one time in 20 should chance variations in the sample cause the results to vary by more than 3 percentage points from the answers that would be obtained if all voters in a CD were polled. ====
Comment by Rich Miller Monday, Oct 30, 06 @ 5:49 pm
I’d like to know why it can’t be true? Please be more specific. I know a few of district residents who ARE voting for Seals–and they’re not wacky. There are many people who wish the Democrats would take back the congress and are willing to take the chance of throwing out even an effective current representative to do it…because in the big picture it seems possible to achieve that end. We’ll see next week, I guess.
Comment by NoGiftsPlease Monday, Oct 30, 06 @ 5:58 pm
If you drive or walk around the 10th district you would believe these results. This race is tight. I have no idea about the polling methodology but it is clearly NOT an online poll i.e. the sample is not self selected. I think Dan Seals could pull this off.
Comment by Way Northsider Monday, Oct 30, 06 @ 6:06 pm
http://www.dailykos.com/storyonly/2006/10/30/173950/09
The DCCC is getting involved in IL-10 so that seems to validate the poll.
Comment by Way Northsider Monday, Oct 30, 06 @ 6:08 pm
I’d like to know too. Apparently it’s a telephone poll conducted by robot computers. I’m a professional statistician. My large concern is that if there might be some systematic bias in the polling method, like those who are affluent being more likely to hangup on the poll. That’s really the only way these results can be justified, right?
Comment by Confused?? Monday, Oct 30, 06 @ 6:12 pm
rich, if you can keep track of the final polls for the various important elections, and put in one discussion track, it would be nice to see who did well with predictions. i know it’s a bit of work, but a good review for discussion after. thanks for considering….
Comment by anne Monday, Oct 30, 06 @ 6:13 pm
I don’t think anyone has yet convincingly made the case that IVR polls are worthless or even suspect, despite what traditional pollsters may claim. Their results usually track with other polls. That’s not to say that this particular outfit is reputable. But their stated methodology looks OK. I’d like to see what their unweighted totals are, though.
Comment by Rich Miller Monday, Oct 30, 06 @ 6:17 pm
If you have doubts about the accuracy of this poll, go check what the poll says about the 8th or 6th and compare with the Daily Herald polls that were released today. You can also compare peripheral info in the 10th poll with earlier polls in the 10th (e.g. approve/disapprove of bush). What you’ll find is they line up just fine. Actually, this methodology will somewhat underrepresent dems because of 1) lower sampling of ages under 22, 2) selection bias toward those with touch tone phones in their home, and 3) towards english speakers. Kirk is in serious trouble.
Comment by Betterwatchout! Monday, Oct 30, 06 @ 6:45 pm
Rich is correct there. We don’t know fully understand IVR to other systems. In asking simple horse race questions they may even be better, but that’s an empirical question that hasn’t been answered yet. Charles Franklin who is one of the folks running Pollster doesn’t dismiss IRV and he’s very, very good at what he does. He also points out potential weaknesses.
The methodological transparancy is far greater with these then in other polls–even SurveyUSA which is very good about sharing it’s methodology isn’t this complete–and as Rich says, the unweighted numbers would be nice because I can’t figure out how they are weighting though their numbers aren’t far off in terms of other than horse race numbers. The demographics certainly look reasonable.
Also, this is a far bigger sample than other polls with about 1000 people in each District.
The outfit itself is reputable as they include a marketing company and RT Strategies that works with National Journal and Cook Political Report. They are competent professionals. That doesn’t mean they are right because other competent professionals have different numbers out there that conflict with the CD/RT project.
Given this particular partnership doesn’t have the record of Rasmussen or SurveyUSA I’m not willing to fully buy they’ve got a system down, but I wouldn’t dismiss them either. There’s a potential step ahead in verifying people are actually registered as well. There are some other potential challenges there, but it’s great to see it tried to improve the technology.
Comment by Archpundit Monday, Oct 30, 06 @ 6:57 pm
Wow
First GOPs blow the Hyde seat, then they cannot retake Bean. Now Kirk is going down
Wonder how Shimkkus is doing? He is running radio ads about the moon landing on St. Louis radio.
Looks like the pervert cover up is taking a toll.
What a shame.
Comment by Reddbyrd Monday, Oct 30, 06 @ 7:23 pm
Its not the pervert coverup–it iss two wars going on now and waiting to see what happens with Iran. All paid for off-budget and totaling probably 1 trillion dollars when all’s said and done. Economy’s going bust, declining real wages for the middle class, federal budget unsustainable. Refinancing has carried the economy for the last 5 years and is now a popping heard round the world — and nothing but “happy talk” from the whitehouse. Perverts are the least of our problems.
Comment by NoGiftsPlease Monday, Oct 30, 06 @ 7:31 pm
I might not be confident about any race in play this election… except this race. I’ll bet anyone that Kirk wins. And if I lose, I will pick up a democratic ballot in ‘08 and vote for Hillary. That is how confident I am.
Comment by Lovie's Leather Monday, Oct 30, 06 @ 7:35 pm
I don’t know much about polling science, but as far as this poll being any kind of indicator of Seals pulling ahead, I doubt it. If you look at the leanings of those polled from the “second” round of this poll (Oct 15-16) with 1004 LVs, those who ID’d themselves as Dems were 31%, with 36% undecided and 33% republican. Kirk had 46% with Seals 44%. The Seals camp crowed mightily because their guy had pulled to a tie within the margin of error with the incumbent. However, the “third” round (Oct 24-26) had Seals at 48%, Kirk steady at 46%, but in this round of polling (1008 LVs), 34% of those polled ID’d themselves as Dems, 34% as undecided, and 31% Republicans. Note that the third round doesn’t add up to 100%, so we have a missing 1% somewhere. If you credit the missing 1% to Dems, we have 4% more Dems polled in the third round, which (low and behold) is the same percentage increase of Seals from one round to the next. Moreover, Republicans were 2% undersampled in round three compared to round two, so one might say that Kirk must be gaining ground among independants, which is really what this race will come down to, anyway.
The real touchstone is that no other poll conducted confirms these results, including those done by both candidates, which showed Kirk in double digits leads. If Seals’s more recent polling (which, assuming he’s done some) corroborates rather than debunks these results from RT, then why hasn’t he released the results?
For those who think the DCCC is going to ride in on a white horse, after reading the Herald article about how close Roskan-Duckworth and McSweeney-Bean are, I betcha Rahm is not seriously considering the Seals race as a worthy investment in light of his other battles that he has far from won at this point.
TA
Comment by Team America, World Police Monday, Oct 30, 06 @ 8:49 pm
I may vote for Hillary, regardless of what happens in Il-10. It will all depend on what the choices are.
Comment by steve schnorf Monday, Oct 30, 06 @ 9:04 pm
Team America, I believe the missing 1% is from rounding off to the nearest full percentage point. Suppose, for instance, that Dems are sampled at 34.3%, GOP at 31.3%, and Independents 34.4%. If you round these, then you’ll get the numbers in the poll.
Comment by GlencoeNative Monday, Oct 30, 06 @ 9:09 pm
Team America, does anyone know how the other polls were conducted for this race. I’m referring to the McLaughlin and Melman polls. Thanks!
Comment by GlencoeNative Monday, Oct 30, 06 @ 9:13 pm
I think the point that no other poll confirms these results is the main point for IL-10. That said, the changes in party id isn’t all that strange given the environment. First, there is a bit of natural flucation and error. Second, with a wave election you see changing partisan ID as the election gets nearer. In fact, the most recent Democracy Corps polling shows similar movement on a similar sample with less time in between.
Comment by Archpundit Monday, Oct 30, 06 @ 9:28 pm
I’ve been seriously studying the back and forth on Rich’s Blog and have concluded that we all enjoy a lively and controversial campaign and election. I would like everyone to set aside politics and give serious thought to voting for corruption. It seems to be what catches our interest and by golly it IS the Illinois way. It is our birthright!
Comment by Citizen A Monday, Oct 30, 06 @ 9:34 pm
did anyone see the new kirk mailer? it’s kind of hilarious. talking about how dan seals wants amnesty for all illegal immigrants.
this is not even twisting the truth or taking quotes out of context, it’s inventing statements from whole cloth. do you suppose the people who run the kirk campaign even know the internet exists? that most people know how to use it? that people can just go to the seals web site, or do a google search and find out in 5 seconds that kirk is l-y-i-n-g? that most people in his district are not idiot sheep and WILL bother to check?
i guess desperate times call for desperate measures.
Comment by zena Monday, Oct 30, 06 @ 11:14 pm
I guess voters in the district have learned that Seals’ dad played for the Chicago Bears. If you don’t think that means anything, may I remind you who our state’s Chief Justice is and what he was most noted for before winning election to the bench.
Comment by fedup dem Monday, Oct 30, 06 @ 11:34 pm