Wednesday, January 25, 2012

Marquette Law poll says Walker leads with Baggers, old suburbanites

Sometimes you see a headline and your immediate reaction is "Bullshit!" And that's exactly what I did today when I saw the a Marquette Law School poll that Scott Walker led all listed Dem challengers in a recent poll, and that Walker had jumped to a 51% approval rating. I understand that sometimes living in Madison can be a bit of a bubble, but I've done plenty of traveling around the state the last 2 months and seen little to zero outward support of standing behind Walker in a large amount of communities, so I tend to be suspicious when I see numbers that show Walker to have better approval ratings than President Obama.

And it turns out with good reason, because the poll is garbage, with fixed numbers and bad samples. The Marquette people even admit they fiddling with the figures on their "methodology" page.
...In telephone surveys it is common for potential respondents who are younger and have fewer years of formal education to exhibit higher rates of non-response resulting in these groups being under-represented in the sample. To compensate for these non-response effects the sample is weighted to bring sample characteristsics into line with the population values.
In other words, the pollsters had to change their actual numbers of the responses to make up for the fact that they didn't ask a representative group of people.

And I'm sure this'll stun you, but if you look on Page 2 of the methodology, you'll find the people they asked were disproportionately conservative, from the Walker-supporting 262, old, and on landlines. For example, 76.4% of those who answered the phone were 45 or older, and over 40% were past 60, so Marquette had to change its numbers by changing the raw totals for both these groups, and the ones under 45. In addition, 77% of the respondees answered on their landlines, and the Marquette people only deflated that down to 73.2% weight. Now do you really believe that 3/4 of the people in Wisconsin answer their land line first, and talk to complete strangers? Well, maybe if they're shut-ins, elderly, or live way out in the sticks, but not most people with jobs, lives and interests in metro areas with size.

Geography is another major fail. It's no secret that suburban Milwaukee is the area most likely to support Walker, and the Madison area and City of Milwaukee is most likely to vote him out. But again, the suburbs/ Milwaukee DMA (which brings in Racine, Kenosha among other places) got more respondees than they warrant (31% vs. 24% of the Wisconsin votes in 2010), while Madison and outstate were a good 2-3% less than their vote share in 2010.

Even the people they could get in the City of Milwaukee were a shaky group, because the poll's crosstabs of Milwaukee's vote in a Barrett-Walker rematch have Barrett winning 62-31, which sounds OK until you realize that Barrett beat Walker 74-25 in Milwaukee in 2010. So are you telling me that Walker would GAIN 18 points on Barrett after Scotty has tried to dienfranchise Milwaukee minorities with voter ID, and his collective bargaining stances would probably cost him the endorsements and votes of police and firefighters that backed him in 2010? Think about it for 2 seconds and you know those numbers are seriously flawed, and you feel like an ESPN analyst and say "C'MON MAN!"

And the Milwaukee poll results go to the biggest failure of this poll - oversampling of Tea Baggers and other right-wingers. As a starting point, let's use CBS's 2010 exit poll of the Feingold-Johnson Senate race, and figure the voters would be the same for the Walker-Barrett race that was also voted on at that time. At the height of TeaBaggery in an election filled with motivated Republicans who saw a chance to win their first big statewide elections in the 21st Century, we had the following ideological breakdowns, and support for the Tea Party:

Ideology, 2010 election
Conservative 36%
Moderate 43%
Liberal 21%

Tea Party, 2010 election
Support 36%
Oppose 32%
Neutral 29%

Now compare those figures with this poll.

Ideology of poll responders (those who answered question)
Conservative/ Very Conservative 44%
Moderate 34%
Liberal/ Very Liberal 22%

Tea Party, poll responders
Favorable 33%
Unfavorable 41%
No Opinion/ Neutral 26%

So they've basically flip-flopped conservative and moderate numbers in Wisconsin's voting population. Well, no kidding that Walker leads if you overstate conservatives by 8% at the expense of moderates, especially when he loses moderates by 6-10% in this poll (that's probably light, I've seen other polls that have Walker disapproval around 60-70% with moderates). Get the same numbers but an accurate moderate vs. conservative number, and this thing is a toss-up if not Walker trailing.

And the Tea Party stat is especially BS, because most national polls have Bagger support around 26-30%, but non-support over 60%. But in this poll the favorable/unfavorable bi-normial is more like 45% aprove, 55% disapprove. But that's who picks up and answers the phone when strange people call, so instead of the pollsters using an ounce of common sense and realize that they've overloaded their poll with Baggers that will make Walker look better than he really is, they just shrugged and went with the numbers. This is either negligent, or it's intentionally trying to make news by producing a story line of "people still support Walker" that doesn't exist.

This is the reason I worry over polls like this, because when outlets like Madison.com just run with the numbers without doing any analysis of where the numbers came from, it favors Walker, and leads to flawed coverage and opinions. The 2010 polls in the Feingold-Johnson race consistently overstated Johnson's lead, and both that race and the Governor's race ended up being 4-5 points closer than most polls on Election Night. You think the casual citizen who likes to "be on the side of the winner" might have been more likely to vote for either Feingold or Barrett if they thought the race was a virtual toss-up instead of a 8-10 point GOP lead? You bet.

That's why you have to hammer these polls, because you know the media won't have the guts or time to, and good news from fake polls gives the Walker folks enough of an alibi that can allow them to try to steal a close election with more Kathy Nickolaus-style shenanigans. We need to be vigilant at all turns, and make sure fake stories and faulty polls don't get made into a media reality.

P.S.: And We Are Wisconsin illustrates how you respond, by going right to the JS and pointing out why it's garbage. In addition to my point on oversampling conservatives, they point out that union households are underrepresented vs. the 2010 exit poll (if anything, they'll vote more), and they call out the collective bargaining question (the poll says "limits" on collective bargaining- Walker ABOLISHED it. Big difference).

We need this type of rapid response to propaganda like this. Well done, We Are Wisconsin.

5 comments:

  1. Thank you for this great analysis, Jake. Unfortunately, even left-leaning people are falling for the surface spin (e.g., Eric Kleefeld at Talking Points Memo; DDay at Firedoglake News; a Daily Kos diarist who seems to take every bit of news as a reason that we need to put a gun to Russ Feingold's head and force him to run for governor). It's so depressing how quickly the lies and half truths get out there.

    ReplyDelete
  2. Great analysis. This poll really bothered me, because it just didn't seem to reflect reality. That 2010 exit poll seems to be an objective basis for comparing sampling techniques, and by this standard the Marquette poll looks like an epic fail. Thanks for posting.

    ReplyDelete
  3. This comment has been removed by the author.

    ReplyDelete
  4. Contacted Badger Democracy & they said they looking into republican bias in these polls

    @bdgrdemocracy: ...working on it with a UIC epidemiologist/statistician...looking at crosstabs/methodology for Repub bias in samples-stay tuned

    In addition the questioning was privatized by LHK Partners-basically a marketing company.

    ReplyDelete
  5. Thanks for reading and for following along everybody, and a good tidbit about LHK Partners- we know the Walker folks would pay some business off in order to spread their propaganda.

    We need to stay vigilant on crap like this, because the media will let it slide and sink into people's heads. Keep doing your good deeds.

    ReplyDelete