Latest Marquette-Franklin poll will send GOP spinning, convulsing…Barrett supporters take heed

The last Marquette University poll (May 30 poll toplines) from Charles Franklin before the recall election, when analyzed, will send Scott Walker and the GOP into full spin and attack mode. Even after spending tens of millions of dollars, outspending Tom Barrett 12-1, and plastering the airwaves with the basest propaganda – the race is a virtual tie.

The GOP and conservative pundits will say Walker has a commanding lead. Among likely voters, Walker leads Barrett in this poll 52 – 45%, with a 4.1% margin of error (MoE), larger than the 3.7% MoE in the rest of the poll due to smaller sample size. Based on the MoE, the results are virtually unchanged from the last poll. A look deeper into the poll will explain why this is good news for Barrett supporters.

As Badger Democracy has always maintained, the polling done by Charles Franklin is sound – the only errors arise from media outlets and propagandists misrepresenting what the poll says. Franklin does a much better job than many polls at disclosing all the information, methodology, and crosstabs. Here are the facts – suitable for offsetting right-wing radio talking points. The bottom line…due to the random sample and geographic population distribution, this is a poll which leans significantly conservative – by as much as 5-8%. Here’s how:

First, post-stratification, or “weighting” of the samples to reflect the overall population. The only weighting done in this poll was for age and sex. The rest of the poll results were “sufficiently close to the population values that only age and sex were used for estimation of the weights.” As samples are derived from random phone dialing, here are significant statistical differences from the samples based on the May 30 poll Methodology pdf:

1. “PID3 w/lean” – This is the party people identify with, when pushed into choosing. Unlike the last poll, the sample actually breaks a little more toward Democratic, 48-42.5%. So why the higher percentage for Walker? The poll is under-representative of Independents (by 7% under), a demographic Barrett is doing very well with, according to previous polls. The other issue, is this is simply an overly conservative sample. According to the crosstabs, Walker gets nearly 18% of the “Democratic” vote in this poll, and 45% of Independents – these reflect what is often considered “outlier” samples…leaning considerably further than expectations based on previous samples. In past polls, Walker has gotten no more than 10% of Democratic voters, and under 40% of Independents. The reason this poll leans so conservative can be found in the population distribution.

2. Franklin is sampling his polls to be reflective of population distribution, based on the most recently available Claritas estimate. Here is the population distribution from this poll, very close to the actual population:

City of Milwaukee                                    12.66%
Rest of Milwaukee DMA                       32.68%                                                                                              Madison DMA                                         16.83%                                                                                           Green Bay
Appleton DMA                   18.50%                                                                                          Rest of Wisconsin                                  19.33%                                                                                               Total                                                              100%

While this reflects the POPULATION distribution, the VOTING distribution is very different. Here is the vote distribution by corresponding region from the 2010 General Election, and the corresponding difference in sample (source, GAB election data by county 2010):

City of Milwaukee                                   16% (-4) 
Rest of Milwaukee DMA                     19% (+13)
Madison DMA                                            15% (+1)
Green Bay‐Appleton DMA                 10% (+8)
Rest of Wisconsin                                    40% (-21)
   Total                                                               100%

Significant that the “collar counties” of Milwaukee are over-represented by 13% because of the emphasis on population distribution, while Milwaukee is under-represented by 4%. Meanwhile, in 2010, Dane County only had statewide average turnout of about 43%, instead of its usual 55-60% – thus only accounting for 15% of the vote. A much higher turnout in Dane County will mean a higher impact than the 15% reflected by the voting distribution, potentially as high as the 16% range seen in the population distribution. For purposes of analysis, let’s consider Dane County even, somewhere in the 15-20% range.

Also of significance is the over-representation of the Fox Valley (traditionally more conservative) by 8%, and the outstate vote – accounting for 40%, is under-represented by over 20%. This is the most significant number for Tom Barrett and his supporters. Based on the differences in population and voting distribution; along with the virtual even distribution in party (though slightly under-representative of independents) it is fair to say this poll could be conservative leaning by as much as 8% (based on the MoE), bringing the race to dead even.

Despite the 12-1 financial advantage for Scott Walker; Tom Barrett and his supporters have brought the race to even, and there is a noticeable shift in momentum.  Mobilizing the Democratic and progressive voters in the upstate areas of Wisconsin, Dane and Milwaukee Counties will make the difference in this election. Every poll has been consistent in that fact.

The Marquette – Franklin poll, by being factually accurate about who is sampled, and the composition of the sample, has given us a clear path to victory – mobilizing voters in these key areas of the state will lead to victory on June 5th.

As Mahlon Mitchell and the fire fighters say… “All hands working” for the next 6 days. Solidarity.


11 thoughts on “Latest Marquette-Franklin poll will send GOP spinning, convulsing…Barrett supporters take heed

  1. Very good analysis of this poll. When I’ve said before how I don’t like the polls run by Charles Franklin, it’s not that he doesn’t know how to poll — it’s that he consistently oversamples heavy R areas, which makes the state look much more R than it is. He’s done this for years, though some polls are better than others; to his credit, in an interview on WTMJ-AM Radio 620 in Milwaukee, he admitted the poll results are “within the margin of error,” but he highlighted a curious result that said that many voters _believe Scott Walker’s claims_ regarding job creation in Wisconsin. Do you have any insight into this? (Was Franklin correct there?)

    • I understand your analysis, but if the poll was so over-weighted, then why was Obama up by 8 points? According to the logic in the blog, this means Obama could be up by 16. Not likely.
      Also, if you say the pollster is a good one, then why would he over-weight this poll? He has the same figures you do and would know if it was overweighted. Thus he would know that his polling methodology would be proven false in a week and not ant to look so bad.

      • Thanks Elliot – first, the poll is not intentionally overweighted, it is a function of who is reached in random dialing – the issue is WHERE the polling is done as a function of vote representation…as for the Obama/Romney race, even the current MoE could put him up by as much as 12 (in the likely voting sample), or up by only 4. This could be a reason the DNC has gotten into Wisconsin so late – they know Romney is not very electable in Wisconsin.

    • I believe that Walker cannot be trusted in anything other than a deep driven desire to slash costs no matter what. He did that in Milwaukee County and it showed. He seems driven to do what it takes, end justifies the means. The job data is not going to be fleshed out till after the election. He seemed so ready to respond, his ad came out hours after the announcement, eerily so. There are powerful interests that are orchestrating public opinion. I think it will backfire in the end. As for Charles Franklin, this analysis proved what some have always said. The polls are what you make them. Why claim the poll is representative when it is clearly not? The goal here, even by a respected University, is to tell people that it is hopeless, stay home, Walker is going to win. It has worked in the past.
      Wisconsin is a progressive state and it contains people generally supportive of the common good. It is hoped that those will see the outside money, the heavy handed tactics, the orchestration of a weak governor, controlled by power brokers, should sway enough to say Enough Of This and throw the rascal out of office.

  2. Another good analysis.But I beg to differ with you about the methodology. To build on what Ms Caffrey posted, a reputable pollster doesn’t oversample populations known to skew one way or the other. This poll is executed by the Universal Survey Center in NYC, specialists in market research surveys. The outfit is not a member of the American Assn for Public Opinion Research (AAPOR), so it is not bound to uphold their high ethical standards. As USC’s own website says describing their ‘Public Insights:Vote’ polling:
    “A dedicated team of Universal Survey professionals works directly with you to understand and execute the requirements of your project. This team approach provides you with undivided and personalized attention. Our teams are championed by Account Directors who are your research partners. Their role is to understand your needs and provide you with the right guidance and support. Dedicated Client Service Managers are responsible for the implementation of your project. They ensure you have the critical information required to help manage your project.”

    IOW, if Franklin wanted to get a respresentative sampling of the electorate, this outfit would deliver it. But he doesn’t, He wants a GOP propaganda tool – and that is exactly what they gave him.

    • I would like to believe Badger’s analysis, but: 1) if the polling was over-weighted in certain parts of the state, the author would know that and thus would correct it (otherwise your depositive description of him makes no sense) and 2) the past polling by PPP appeared to confirm the Marquette poll. Now if PPP is scewed also, why would they ignore that issue of what part of the state do people live in?

      • It’s how samples are generated (random dialing), and how they determine distribution of the sample. Noteworthy that there is no precedent for a recall election like this, so to Franklin’s own admission, their models aren’t entirely accurate.

  3. The Milwaukee DMA is Dodge, Jefferson, Kenosha, Milwaukee, Ozaukee, Racine, Sheboygan, Walworth, Washington and Waukesha counties, not just the 4 referenced here.

    Madison DMA comprises Columbia, Dane, Grant, Green, Iowa, Juneau, Lafayette, Marquette, Richland, Rock and Sauk counties, not just Dane.

    Green Bay-Appleton DMA comprises Brown, Calumet, Door, Fond du Lac, Green Lake, Kewaunee, Manitowoc, Marinette, Menominee, Oconto, Outagamie, Shawano, Waupaca, Waushara and Winnebago counties.

    So I’m afraid your calculations are a bit off.

    • Geoff – thanks for that…fixed. That still results in a significant bias toward the Conservative counties, and under-represents Milwaukee and upstate vote. I appreciate the correction.

  4. FWIW, I’m okay with Walker supporters believing that they are 7 points up in the polls in the final week before the election. Voter turnout is easier to motivate for the perceived underdog, at least when it’s this close.

    • This election has always been about turnout. If we get presidential level turnout, Barrett wins. 2010 turnout and Walker wins. That’s why it would have been better for the recall election to have been in November. Bad tactical decision, which lowers student and minority turnout.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s