The Marquette Poll was more accurate in '24 than '16 or '20. Why? Better success finding Trump supporters

 

MILWAUKEE (CBS 58) -- While the Marquette Law School Poll struggled to accurately capture the feelings of Wisconsin votes in the previous two elections where President-elect Donald Trump was on the ballot, it was much closer to hitting the mark in 2024.

The final MU poll, released Oct. 30, had Vice President Kamala Harris one point ahead of Mr. Trump, 50% to 49%. As counties complete their canvass of the election results, President-elect Trump appears to have won the state by nearly 30,000 votes and a margin of 0.9%.

While the poll was off by 1.9 percentage points, it was a marked improvement from the final polls ahead of the 2016 and 2020 elections.

2016 marked the poll's worst performance; it missed the final Wisconsin result by 6.8 percentage points, skewing heavily toward Hillary Clinton. The poll fared better in 2020, leaning toward President Joe Biden by 4.4 points compared to the final result.

Speaking at a Milwaukee Press Club panel event Wednesday, Franklin said Mr. Trump's victory should not have been a surprise and cannot necessarily be attributed to campaigning on either side. He pointed to polls consistently showing voters were most concerned about the economy and trusted Mr. Trump more on the issue, and they gave low marks to President Biden throughout his term.

"We had 38% saying the economy is the most important issue, and Trump had a 10-point advantage on that," Franklin said. "Joe Biden's job approval here in Wisconsin has been between 41 and 43 percent for three-and-a-half years. Presidents and presidents' parties rarely get re-elected when their job approval is that low."

Franklin said past polls were less accurate because they struggled to get a representative sample of Trump supporters. He said getting that aspect of polling right was a major emphasis in 2024.

"They certainly don't trust the media, or God forbid, pollsters," Franklin said. "And so, as a result, these are exactly the kind of people we want to make sure we're reaching. But they're the absolute hardest category of people to get to do a survey."

To get a more representative sample, Franklin said the poll moved away from contacting people by dialing random numbers. Instead, the poll began contacting people in 2023 using a list of registered voters. Franklin said the poll broke voters up into 90 different sections of the state based on past voting patterns in those areas.

"We sort it based on the local geography of the Republican to Democratic vote," he said. "So, we sort it from the most Republican wards in the state to the most Democratic wards in the state."

Franklin added the poll now sends emails and text messages with prompts for people to take the survey online. He estimated 80% of the poll's responses came online while only 20% were over the phone, another necessary adjustment as fewer people are willing to take calls from unfamiliar numbers.

Franklin said another segment of the electorate that is typically hard to accurately reflect is the working class. He acknowledged that added another challenge in this particular race and a Marquette analysis of last week's results by voting ward found the poorest 20% of wards favored Harris by 12 points. It continued a slide for Democrats with the working class, as Clinton won in the poorest 20% of wards by 19 points in 2016.