The highly regarded (and usually accurate) Marquette Law School poll swung wildly in Walker’s favor this week. Should the electorate be freaked out that this bloc of “likely voters” will be the ones who vote Walker in to a second term? Not necessarily.
My favorite statistician, Nate Silver, wrote about the challenge that “likely voters” can present to pollsters in a column on his FiveThirtyEight blog on October 22, 2008. That week there were just two weeks to go until the nation would elect Barack Obama the 44th President of the United States. Pollsters had scrambled to predict whether – or by how much—Barack Obama would be elected. One of the measures they cited in their poll results was “likely voters”, and if you drilled down into the data, you’d see that their definitions varied wildly on this term. What exactly is a likely voter?
Silver dissected the polls individually, finding two types of “likely voters.” The first “…considers both a voter’s stated intention and his past voting behavior. The second cluster coincides with their ‘expanded’ likely voter model, which considers solely the voter’s stated intentions. Note the philosophical difference between the two: in the ‘traditional’ model, a voter can tell you that he’s registered, tell you that he’s certain to vote, tell you that he’s very engaged by the election, tell you that he knows where his polling place is, etc., and still be excluded from the model if he hasn’t voted in the past. The pollster, in other words, is making a determination as to how the voter will behave. In the ‘expanded’ model, the pollster lets the voter speak for himself.”
The final Marquette Law School poll asked voters if they were registered to vote, asked them to rate how likely they were to vote, and whether or not they had voted already (early voting was under way when the most recent poll was done). This is not as accurate or as in-depth as the description of the first cluster in the paragraph above. Add to this the in-depth analysis that Harry Enten did of the Marquette and Wisconsin pollsters who’ve tracked state voter data (www.fivethirtyeight.com, August 28, 2014): “The poll’s registered voter results are more in line with the long-term averages of the Marquette poll and Wisconsin polls overall, and there isn’t evidence from past campaigns that Marquette’s likely voter screen produces more accurate results.” Enten goes on to explain, “It’s important to remember there is nothing magical about a likely voter screen. Marquette chooses a simple screen: Those who say they are absolutely certain to vote in November. Marquette could just as easily choose to include participants who say they are very likely to vote. Both methods have been employed in the past by other polls, and studies have shown that some people don’t accurately gauge their likeliness to vote.
Nor is there much of a sign that using a likely voter screen on Marquette’s surveys improves their accuracy, even within a month of the election. An average of Marquette polls during the final month of the 2012 recall had Walker ahead by 6.4 percentage points among registered voters and 6.5 points among likely voters. Walker won by 6.8 points. In the final month of the 2012 presidential campaign, President Obama led Mitt Romney by an average of 4.2 percentage points among likely voters and 6.9 points among registered voters. Obama won the state by 6.9 points.”
What I find most interesting is the responses in the Marquette Law School poll that could only lead the reader to believe that right-leaning voters are overrepresented in the sample. More people are in favor of Walker’s handling of the state economy, fewer people approve of the job President Obama is doing, 37.6% of those polled approve of Mary Burke and 46% disapprove… You see where I am going with this. It’s important to note that some of these responses will get weighted in the end to make up a more representative sample that accurately represents statewide political affiliations and preferences, but I would challenge anyone to read the raw data and not come away feeling like right-leaning individuals were overrepresented. Also, when you look at how they weighted the poll results, though they don’t explain why, they chose to over-represent both “lean Republican” and “Independent” voters, further making polling results lean to the right. I simply can’t come away from the sampling data on this poll and say that it’s entirely balanced. Read more here: https://law.marquette.edu/poll/wp-content/uploads/2014/10/MLSP27ToplinesRV.pdf
Americans are growing less willing – seemingly by the week—to affiliate with either party. I can’t say I blame them. I'd like to do a pretty major overhaul on the DPW, myself. All complaints aside, let's make one last push, everyone. There are still lots of opportunities to volunteer to drive voters to the polls, to get those last-minute reminder e-mails out to your left-leaning neighbors, or reach out to your local party and see what else is left to do that day. Let’s restore and repair this great state’s reputation. Let’s end Walker’s fascist, fiscally irresponsible stranglehold on Wisconsin on Tuesday, November 4. It’s well within our reach, Wisconsin. It is absolutely NOT Walker’s race, even according to the numbers, if you look closely enough. VOTE!