Why KSTP Primary Polls Were Off, But General Election Polls Should Not Be Ignored

KSTP/SurveyUSA conducted three separate polls over the summer months, and each one showed Mike Lindell as the front runner against Lisa Demuth and Kendall Qualls. The last poll showed Lindell with a 6 point lead over Demuth, but on election night, Demuth beat him by 11 points. What happened???

Here’s why I think KSTP’s polling was off…

The cross tabs of the last poll showed that they had a survey size of 1,757 registered voters who were willing to answer their questions. They first asked the voters if they already casted a vote in the primary, were definitely going to vote in the primary or will “probably vote in the primary.” The latter, unreliable category was 25% of respondents.

More importantly, 69% of the people who said they supported Demuth, responded that they will “definitely vote in the primary” vs only 56% of Lindell supporters. Lindell respondents beat Demuth people as probable primary voters (35% to 25%), but that’s an unreliable category.

Lindell probably did well with the low propensity, Republican voter that Trump infamously energized to go to the polls and vote for him. I’m guessing this block of voters liked Lindell and would loyally follow Trump’s endorsements, but didn’t actually plan on making the effort to vote in the primary, which is all that really matters in the end.

KSTP/SurveyUSA’s poll was off because they overcounted the amount of people who were going to actually vote in the primary. When they displayed the overall numbers and had Lindell with a 6 point margin, their numbers also showed that 458 of the 1,757 registered voters would vote in the Republican primary. That’s a whopping 26%!

Historically, only about 9-10% of registered voters in Minnesota vote in the Republican primary. In the 2026 Primary, there were 3,889,4442 registered voters as polls opened and 413,605 cast a vote in the Republican gubernatorial primary, or 10.6%. KSTP/SurveyUSA’s poll overcounted primary voters by a factor of 2.45.

If you are wondering why they were off in the Democrat Senate primary, the same thing happened on that side as well. Their last poll had Angie Craig up over Peggy Flanagan, 42% to 40%. However, they assumed 33% of registered voters were going to vote in the DFL primary, when in actuality, on August 11th, only 18% of registered voters voted in the DFL primary.

But don’t give up on KSTP/SurveyUSA polls just yet. They took a risk and tried doing polling pre-primary, which was not a commonly practiced thing in Minnesota primary races of the past. Hopefully they learn the lessons and improve their modeling for the next one.

When it comes to general elections, they’re polling has shown to be pretty accurate. In 2022, they accurately predicted Walz beating Jensen by a margin of 8. Impressive! Their polling was also accurate in the 2024 election cycle.