Today, I interviewed guest Stephen Leuchtman, who has been involved in political campaigns for a lifetime. He has great stories and some insights about why polling is not our friend. There are good reasons for this change, and a lot has to do with how we’ve evolved technologically. Gather round and find out why we should not place any best based on polls this year.

Jennifer Carole: Hey everybody, welcome to the Jen Carole podcast. Let’s get started. Stephen, I’m relying on your reputation so this all sticks to you!

Stephen Leuchtman: That’s fine. No worries. So, I’m Steve Leuchtman. I’ve been involved in elections since 1988, starting when I was 18 and worked for Al Gore in the Democratic Primaries. My first paid job in politics was for Harry Reid in his ’92 US Senate race, and I also worked for Dick Stallings in Idaho. That one didn’t turn out as well, but Dick was a fantastic guy—one of the nicest men in politics. Then, I moved to Omaha in the mid-’90s and went to work for Gallup for about three years.

Jennifer Carole: Gallup in the late ’90s? That was the place to be, right?

Stephen Leuchtman: Yeah, absolutely. I was doing debate polls, but I like to say my first casualty of working at Gallup was a generally positive impression of the average voter. People have such idiosyncratic ideas about elections, and it’s almost impossible to capture them in a simple poll. After Gallup, I did field work for Democratic candidates in the Dakotas, Minnesota, Nebraska, and Iowa. Then I started a polling operation, and I’ve generally worked for Democratic causes and 501(C)(3)s, super PACs, and the like. Mostly, when I’m polling, it’s for allocating resources for high-level donors.

Jennifer Carole: That’s important. I don’t think people understand that there’s a whole group of people sitting somewhere deciding where to spend money. It’s a big game, right?

Stephen Leuchtman: Yeah, it is. Although I’m rarely in the room where those decisions happen. I’m the help—I provide them with information so they can decide whether to invest in a race and how to allocate resources for the most impact. When I was at Gallup, polling was based on a few assumptions. One was that everyone had universal access—you could reach anyone because everyone had a landline. The second was that errors fell equally across the distribution curve—Republicans and Democrats were just as likely to lie. You had universal access and random samples. The models accounted for a certain level of error, but it was manageable.

Jennifer Carole: Right, because they built error into the sample.

Stephen Leuchtman: Exactly. For example, when I worked at Gallup, the response rate was around 33%. So, you had a solid chance of getting a truly random sample that reflected America. You also knew what neighborhood you were calling based on the area codes and exchanges, so you could accurately target geographic areas.

Jennifer Carole: Oh yeah, old phone numbers were specific to neighborhoods. It was so precise.

Stephen Leuchtman: Exactly. But then the Telephone Number Portability Act came into play, which allowed people to keep their phone numbers when they moved, and suddenly, you couldn’t be sure if someone in the 415 area code was still in San Francisco or had moved to Alabama. That started to make things much more difficult, especially as landlines started disappearing and cell phones took over.

Jennifer Carole: Cell phones really complicated things, right? People had both landlines and cell phones for a while.

Stephen Leuchtman: Yep, I got my first cell phone in 2006 and got rid of my landline in 2013. But the response rates kept dropping. By 2012, we were seeing about 15-18% response rates, and by 2020, it was down to the single digits—around 7-9%. Some polls now, like the Times-Siena poll, had response rates as low as 1.2%. And with that, the cost of conducting a poll skyrockets.

Jennifer Carole: So, people understand, it’s just endless calls and touches to get one response.

Stephen Leuchtman: Exactly. And the lower response rates create a whole set of new problems, like sample bias and response bias, where the people answering the polls have something in common that skews the results. For example, I had my spam blockers off when I got a polling call last week—otherwise, I never would’ve received it. A lot of people are blocking potential spam calls, so the people who are answering are often those with bad data hygiene. You have to wonder, are these people representative of the general population?

Jennifer Carole: Right! Most of us block spam calls.

Stephen Leuchtman: Exactly. My 18-year-old daughter won’t even answer texts from people who aren’t in her contacts, so they’re never going to reach her or her friends. There’s this whole group of people they just can’t reach.

Jennifer Carole: It’s the same with my daughter. She never picks up the phone unless she knows who’s calling.

Stephen Leuchtman: Exactly. Then, you’ve got people like Nate Silver, who’s made a career out of polling averages, but polling averages are garbage. If you take 10 bad polls and average them, you’re just getting an average of bad data.

Jennifer Carole: Right, garbage in, garbage out.

Stephen Leuchtman: Exactly. And this cycle, we’re seeing Republican-affiliated pollsters flood the zone with low-quality polls to generate narratives that favor their candidates. It’s designed to set a narrative because they believe that whoever wins the narrative wins the race. You’ve got to be very cautious about it.

Jennifer Carole: So what can people do? How do we navigate this?

Stephen Leuchtman: Take polling with a grain of salt. There are good pollsters doing good work, but most polls right now are content for the content mill. The really accurate polls will come out closer to the election, in the last week of October. That’s when it matters. In the meantime, focus on what’s happening on the ground and trust your gut.

Jennifer Carole: That’s great advice, Stephen. Thank you so much for this insightful conversation. I look forward to having you back in October to talk about the 2020 elections and what made them so different.

Stephen Leuchtman: Absolutely. I’d love to come back. Thanks for having me, Jennifer.

Jennifer Carole: If you like the podcast, be sure to subscribe and rate it. And I’ll see you all next time!