If you want to really know what the Wisconsin electorate is thinking, you have to wait for Election Day. If you want to have the best possible idea what it’s thinking before that, you look at the Marquette Law School Poll.
Although it was created only eight years ago, the Marquette poll is the longest-running, most frequent and most closely watched survey of Wisconsin voter moods. By the Nov. 3 presidential election, poll director Charles Franklin and his team will have queried state residents about their political leanings 65 times over the years.
The poll’s questions delve beyond outcomes for state and federal elections to lay out the reasons behind those attitudes; this year COVID-19 and racial justice issues have joined the usual questions about the economy and social issues.
This extensive track record has made Franklin, an Alabama native, the state’s preeminent pollster. Here’s a peek behind the curtain of his polling process.
How has the coronavirus pandemic affected polling this year?
We’ve been pretty lucky because it’s a telephone survey – there’s no in-person interviewing. We contract out the phone calls to a “phone house,” and ours in the spring did shut down. [In July] we ended up using three phone houses to complete the survey, but that was the biggest trouble we ran into.
Why use phones instead of mail or email surveys?
Everybody has an equal chance of being selected. We have a list of every area code and exchange [the three digits after the area code] in the state. And then we draw a four-digit random number from 0000 to 9999. We randomly pick an area code and exchange and then we add that four-digit number to it and we dial it. That guarantees that every working number in the state has an equal chance of having been selected into the sample. Now, practical problems do emerge. Not everybody picks up the phone when we call, not everybody agrees to do an interview. So, it’s not perfect.
When you call them and they don’t answer, is that it? Or do they get a second chance?
We call four to six times over the course of a five-day survey. The biggest barrier we face is people who don’t pick up the call. Lots of people say, “I never pick up a call I don’t know,” but we actually get a lot of interviews completed on the third call. Or the fourth call. Calling a second, third, fourth time – at different times on different days – really does result in a lot of people getting reached and included in the survey that weren’t available the first time you called.
How many calls do you make versus how many people pick up and say yes?
We almost always do about 800 completed interviews. That’s a response rate of about 2%. So, we’ve called more than 40,000 times to reach that final completed 800 respondents. When people pick up – before the pandemic – we would get about a 55% response rate [of people agreeing to take the survey], and that’s quite good. Since the pandemic, we’ve been running over 80%. So the social isolation that we’ve been seeing since March really has had an effect on response and cooperation rates. And that might be the only good thing to come out of the pandemic.
Is political polling making an election prediction?
I don’t think so, but that the polls are inevitably seen as prediction is just a fact. It’s very hard to avoid that. I prefer that we look at today’s polls as telling us who supports whom and why. We would all be a little better off if we focused more on understanding what is as of now, rather than trying to gaze into the crystal ball.
Like other pollsters, you had Hillary Clinton winning Wisconsin in the 2016 presidential race. That didn’t happen. Do you count that as being wrong?
It’s really wrong. And I don’t think any of us that were in that position should try to spin it as not being wrong. There’s plenty of room to discuss why it was wrong, but we have to live with our failures as much as our successes. The Marquette poll has been quite accurate in most of the races we’ve looked at with a glaring and very important exception of November 2016. I try to close each final release of the poll before Election Day saying, “Remember, polls don’t vote, people do.” The election is in your hands. Go and vote for who you want to vote for or don’t vote. But for goodness’ sake, don’t do it based on anything in this poll. Do it based on what you believe and what you want to express about government.
You’ve asked about recent events, such as the pandemic and mask mandates as well as support for police and the Black Lives Matter movement. How do you decide what to ask?
It’s ultimately my judgment, informed by what the public discussion is at the time. I’m obviously paying attention to the news and reflecting those issues. We’ve devoted a lot of questions to the COVID epidemic, because how could we not? It’s the single most important issue in the country. And it has ramifications for other really important things, like the economy. We’ve asked a lot about the Black Lives Matter protests and police reform. All of those have been really driven by events this summer.
What does it take for poll results to surprise you?
I get surprised fairly often. [For the poll in early August], we asked, are you comfortable with reopening public schools or not? And more parents of school-aged kids are comfortable than nonparents. I was surprised by that. I thought parents would be more cautious, not less cautious.
How much do you agonize over how a question is worded?
Quite a bit. The rules are clear: Keep the question as simple and direct as possible, and avoid strong adjectives. The problem is, you have to appreciate that a lot of people come to surveys without strong opinions about a number of issues. And therefore, potentially, they were affected by the way you word the questions. You want to make your questions as boring as possible and not think of the survey questions as a creative writing exercise.
How do you think poll results influence voters?
Our goal is to inform them in the same way that a newspaper informs voters, to provide information that they would not have otherwise. I think there is a risk – and I think 2016 is an example of this – of seeing the polls as an infallible prediction of the future and then choosing to act on that in some way.
When was the last time you voted and why don’t you vote now?
I think I voted in the spring election in 2002. But I haven’t voted since. It’s a judgment call, and I’m certainly not holding it up as a moral or ethical position. I just felt it was a way for me to separate my analysis from my personal preferences. Once I retire, I plan to start voting again.
How is Wisconsin voter turnout looking for the presidential election this year?
We’re not seeing much difference between the parties in likelihood of voting this year – Democrats and Republicans look pretty steady. It looks like we’re going to have high turnout this year. I would guess we’re going to be in the 3.2 million turnout range.
Election Night Results? How Quaint
IF YOU’RE PLANNING TO STAY UP until you know who won the presidential election, you’ll need something stronger than coffee to keep you awake. A lot stronger.
Early voting has spiked over COVID-19 concerns; by mid-September more than 1 million Wisconsinites had requested absentee ballots, a number expected to reach as high as 1.8 million by the cutoff date. All those ballots take time to count, something clerks can’t legally start doing until Election Day.
A lawsuit to allow clerks as much as an additional week to receive and count ballots was still being adjudicated as this issue went to press, but even without that extension, nobody was expecting a timeline like that which allowed The Associated Press to call the race for Donald Trump at 1:30 a.m. on Nov. 9, 2016.
“We are not anticipating that we will be done and have results right at 9 p.m. or 10 p.m.,” Milwaukee Clerk Claire Woodall-Vogg said in a Sept. 17 online forum, “but I’m hopeful that by the time the sun comes up on Nov. 4, that we will be finished and have election results.” – Allie Habeck, Chris Drosner