Yesterday’s release of the most recent Badger Poll by the Madison Capital Times made big news. It should not have. It was a bad poll. I understand why the Green campaign must put out a press release putting a positive spin on the numbers; I wil instead say straight up that this is a bad poll.
For the US Senate race, the poll asks: “Herb Kohl will be the Democratic candidate for US Senator this year, but it is unclear who will be on the ballot for the Republicans. Suppose the final choice in November came down to Herb Kohl for the Democrats and Dave Redick for the Republicans. As of now, would you prefer Kohl or Redick?”
DAVE REDICK LEFT THE REPUBLICAN PARTY AND JOINED THE LIBERTERIANS JUNE 9th!!! The poll was in the field starting June 23rd! It should be totally tossed out for this fact alone – the polling team failed to notice for two weeks that Redick had switched parties, even though it was on Wispolitics and The Wheeler Report, and printed in the Madison Capital Times. Utterly inexcusable.
The question wording above was also used in the gubernatorial poll question. “Jim Doyle will be the Democratic candidate for Governor this year, but it is unclear who will be on the ballot for the Republicans. Suppose the final choice in November came down to Jim Doyle for the Democrats and Mark Green for the Republicans. As of now, would you prefer Jim Doyle or Mark Green?”
“[B]ut it is unclear who will be on the ballot for the Republicans”!!!! Who are these people?!??! Mark Green has no challenger on the GOP side – Scott Walker dropped out months ago. No wonder the poll has weak support among Republicans for Green – the way the question is worded he doesn’t even sound like the party’s nominee, even though he is!!! Again, utterly inexcusable.
Then there’s the brief note on polling methodology. “Results shown here are based on 508 persons who were randomly chosen within households with working telephone numbers…” So, there was no screening for likely voters, or even screening for registered voters. They just talked to random adults. Which, for a poll asking if the people of Wisconsin approve of the job their governor is doing, is fine. Just recognize that you might get a few illegal aliens in your sample. But for a question about who you would vote for, screening for registered or likely voters just makes sense.
Then there’s the time span the poll was in the field: June 23 through July 2. 11 days- that is a huge time frame. Look at other pollsters taking political polls in Wisconsin during the 05-06 cycle: Rasmussen and SurveyUSA conduct their polls on one day. Strategic Vision takes 3 days. Diversified Research for WPRI took two days. Tarrance Group for Walker took 3 days. Public Opinion Strategies for WHA/WMS took three days. Wood Communications’ Checkpoint is five days. Democratic firm The Mellman Group took five days in the 8th. Fallon Research for the Builders took two days. Capitol Opinion for Enough! took two days. Democratic firm Garin-Hart-Yang took four days. Democratic firm Fairbank, Maslin, Maullin & Associates took four days. — That averages out to 2.9 days per poll. And UWSC took 11!
College sponsored polls in Wisconsin are notoriously unreliable. The St. Norbert’s College poll, UWM poll, UW Survey Center, and Edgewood College poll all suffer similar flaws. They have extrordinarily time spans (SNC – 10 days, UWM – 15 days, UW – 11 days, Edgewood – 10 days). I think/assume they use college students as their interviewers, who are apparently less reliable than professional interviewers or touch-tone calls. They are often run by professors who do this as a side job rather than by professional pollsters. They are just historically less reliable. Now, the media reports them like any other poll, but they shouldn’t. As Democratic strategist Bill Christofferson said in a blog post, “The Wisconsin media, however, usually treat all polls equally — one taken by a college class is as good as one taken by one of the country’s top political pollsters. So you can be up one day in a professional poll and have a story saying you are falling the next day, based on another poll that isn’t half as reliable.”
Then there’s the fact that the poll results were weighted, which raises red flags. According to the National Council on Public Polls resource page for journalists, “You should always ask if the poll results have been ‘weighted.’ This process is usually used to account for unequal probabilities of selection and to adjust slightly the demographics in the sample. You should be aware that a poll could be manipulated unduly by weighting the numbers to produce a desired result. While some weighting may be appropriate, other weighting is not. Weighting a scientific poll is only appropriate to reflect unequal probabilities or to adjust to independent values that are mostly constant.” We have no idea from the release what kind of weighting was done on this poll, and what standards (exit polls?) were used to determine relative weights.
Finally, the date of release raises major questions. The Badger Poll claims it “adhere[s] to the highest standards of … rigerous independence.” But that doesn’t mean that the survey sponsor doesn’t own the survey, and get to determine when it is released. UW Poli Sci prof Charles Franklin said today “A new UW-Madison/Badger Poll was conducted 6/23-7/2 though curiously not published until today, 7/11.”
THERE IS NOTHING CURIOUS ABOUT IT PROFESSOR! I would be willing to bet anything this poll’s release by the Madison Capital Times, which proudly proclaims itself as “Wisconsin Progressive Newspaper,” was timed to have the maximum negative impact on Mark Green’s campaign for Governor. One might not unreasonably think that this poll’s release was delayed by about a week and released the day that President Bush visited so all the stories (ex. the Washington Post) about President Bush’s visit would also include the news that the most recent poll shows Green down by double digits. No mention is made that the poll is methodolically flawed and was very poorly written. But alas.
P.S. There are those Terri McCormick apologists whose first line of attack is to say that John Gard is a career politician. But another name for ”career politician” is “long-time community leader” and “experienced legislator.” And the experts at the non-partisan National Journal say that this year, Gard is among the top recruits on either side of the aisle because voters want experienced leadership.
P.P.S. Speaking of Terri McCormick, apparently she broke the law earlier this week by failing to file a statement of non-candidacy as required with the State Elections Board.
P.P.P.S. Nick Reid must be giving Dave Obey a little bit of a scare, because Obey Out reports that he’s been engaging in some push polling of Republican senior citizens in the 7th. Reid’s camp responds.
P.P.P.P.S. For the Wisconsin political junkies, please also be sure to check out the second half of another post of mine today, lampooning DPW/Doyle’s “special interest cash” line of attack.
Last 5 posts by Daniel- Tom Barrett's Savings Plan - June 8th, 2010
- C.J. Abrahamson on J. Stevens - April 11th, 2010
- Why always Bill Bablitch? - March 24th, 2010
- Suhr on Nixon on Health Care - February 23rd, 2010
- Did the USCCB Foresee Dead People? - February 8th, 2010
Oops, maybe I didn’t make myself clear.
John Gard is a career politician, a political opportunist, and the essence of a RINO. I’m not supporting Terri McCormick because she’s promised me anything, unlike some of the Gard Sycophants here. She brings real world experience to the job, unlike Gard, who’s only career has been under the protective Capitol dome.
Gard’s record in politics is dismal, although he takes credit for everything the Assemblu has done while he’s been Speaker. It takes an independent mind to come up with real solutions, not just taking credit for the work of others. Didn’t they used to call that plagerism in school?
Well, if he’s going to take the credit, perhaps we should assign him the blame for the out of control state budgets his Assembly has okayed. or the out of control Department of Natural Resources. Or the failure of the Assembly to act on the UW nut this week. The list goes on.
I too read the entire poll, with particular interest in the Attorney General race. Where on earth did these numbers come from? If the polling was truly ramdom, there should have been a very different distribution. Paul Bucher has no campaign activity west or north of Madison, and Van Hollen has a huge volunteer network that Bucher could only dream of buying, yet somehow Bucher ends up decisively on top. Additionally, I would estimate that nearly 40% of Democrats with whom I’ve spoken have a VERY unfavorable opinion of Lautenschlager. How she is in a dead heat with JB is unexplainable.
The methodology paragraph (which in and of itself is horrifically insufficient for a scientific study) fails to discuss where the “508 randomly sampled” indivuals were from. Based on the obviously skewed distribution, I would opine that they were randomly sampled in key areas only, such as Waukesha and Milwaukie, thus nullifying the randomness of the sample. There was no detail given to clarify whether the samples were statewide or localized.
My undergrad (one of them) is in Economics, with an emphasis on Statistics. This poll fails even the most basic of scientific methods. It is quite clear that this poll was designed to solicit a particular, programmed result. If anybody else out there, like me, understands advanced math and statistics, you’re probably sick to your stomach in reading through this poll.
I’d really like to see the original regressions and verify the statistical significances of the deductions, because they all seem to be outliers to me, well outside an acceptable number of standard deviations.
So Dan, how does Terri McCormick relate to your badger poll? So quick to add that little snippet just to stay critical. John Gard must be perfect. Looking at the election sheets released by the State Election Board… I don’t think Mark Green declared Non-Candidacy either? MMMmmm… is that necessarily breaking the law?
In regards with the Badger poll, take it for a grain of salt. Remember, they put Bush ahead of Kerry by 14 points in the summer of 2004. I remember that clearly… I’m waiting for a better poll like Strategic vision.
Uh, Terri violated a state law, not a Federal Law. Mark Green currently holds a federal office, and I am not aware of a federal declaration of non-candidacy (but I could be wrong). Besides, the fact that he transferred his congressional funds to his governor campaign account, and he has the party nomination, and he endorsed a Republican in the race, and so forth, kinda let the cat out of the bag that he’s not running.
Whereas Terri could very well have filed 200 signatures if she wanted to drop out of the Republican Primary and run again for the Assembly.
SurveyUSA election polls are conducted over three days and nights.
[...] P.P.P.P.S. I still think the methodology for the most recent Badger poll was trash. But just think – if the poll shows the Wisconsin Marriage Amendment winning 53 to 44, imagine what the margin of victory for marriage must be like in a poll with a sample that’s fair to Republicans! [...]
[...] The University of Wisconsin Survey Center poll, on the other hand, is simply not credible (as with the last one). The eight day poll surveyed 508 Wisconsin adults without screening for registered voters, little less likely voters. There is no way Green trails Doyle by 14 points, when every professional poll since mid-September shows the race within 0 to 6 points. The reality is that Green has pulled within a few points of Doyle, that undecideds are going to break for the challenger, and that the Republicans will make ground on turnout. [...]
[...] the fact that it polls residents, not voters; the 14-day sample window (professional polls average three days); the underrepresentation of Republicans; the small sample [...]