While polling elections is a challenge, analyzing polls can be even more dangerous. I considered scrapping a couple days of reporting work on a survey memo on an upcoming Democratic primary, but decided to share what happened to provide a small window into how I digest polling.
Illinois Rep. Daniel Lipinski fended off a competitive challenge in the 2018 Democratic primary and recently released a poll that showed him with a 23-point lead in a 2020 rematch with Marie Newman. That felt like a stretch considering their 2-point race a couple years ago.
The Jan. 7-9 survey of 600 likely Democratic voters by Expedition Strategies showed Lipinski with a 50 percent to 27 percent lead over Newman in the March 17 primary, when leaners were included.
The memo presented the results as the initial ballot — where voters are asked to pick their candidate of choice in a hypothetical primary match-up before any messages are tested. A spokeswoman for the Lipinski campaign told Greg Hinz of Crain’s Chicago Business that the horse-race question “was asked before any introduction of positives or negatives on the candidates, or any push question.” This is proper methodology for a campaign poll and would be in line for Expedition Strategies, which is headed by one of Democrats’ top pollsters, Pete Brodnitz.
A subsequent exchange, however, between the Lipinski campaign and one of my CQ Roll Call colleagues raised a red flag for me about whether the numbers released were indeed an initial ballot question.
“The poll tested candidate favorability and asked dozens of questions about the pertinent issues in the race on a wide range of topics,” according to an email from the campaign.
The “dozens of questions about pertinent issues in the race” potentially being asked before the ballot test was concerning. By asking that many questions, there is a risk of influencing the ballot test question by giving voters information they may not have when they vote in a few weeks.
So after a couple of unanswered emails to the campaign, I reached out to Brodnitz. He confirmed that the ballot test came before any questions that would have influenced responses. It was preceded by “screeners” (to understand who was taking the survey), a question about the respondent’s most important issue, and favorability of all of the candidates on the ballot. Brodnitz added that there was “no message testing or any other content before the ballot and the ballot was a normal ballot with no message or added information included in it.”
With clarity on the methodology, I still didn’t think the results were stellar for Lipinski, even in the face of his large lead. While reaching 50 percent might be considered a position of strength for a vulnerable incumbent, I thought it demonstrated a potentially low ceiling for an eight-term incumbent among voters who have been supporting him or his father (former Rep. Bill Lipinski, who preceded him in office) for more than 30 years. Even in his own survey, Daniel Lipinski was slightly down from this 51.1 percent showing in the 2018 primary.
Brodnitz had another view. In a multi-candidate race, Lipinski is well-positioned to win, even if all of the undecided voters support Newman. Unlike 2018, when there were just two candidates, the congressman doesn’t need to win a majority because there are four candidates on the 2020 ballot. And after spending nearly $1.5 million in the 2018 race, Newman starts with higher name identification compared to a typical upstart challenger. That means her current lack of support isn’t necessarily because voters are not familiar with her candidacy.
Up to this point, the Newman campaign has chosen to criticize the Lipinski polling without releasing its own survey numbers.
I still believe Newman is better positioned than Lipinski to win a disproportionate share of the undecided voters and that this will be another close race. But I’m convinced his overall position is stronger than I previously thought. This exercise has also been a lesson in clarifying methodology by reaching out to the key principal and being willing to see another perspective.