Wisconsin’s political world will turn its attention to the Marquette Law School Poll on Wednesday as the state’s longtime polling benchmark releases its first look at the race for governor since the August 11th primary.
The poll comes amid growing scrutiny of public polling after several high-profile misses this election cycle, including primary races in Wisconsin and Michigan. In Wisconsin, that skepticism intensified after David Crowley’s stunning Democratic primary victory over Francesca Hong.
Most public polling showed Hong comfortably ahead in the closing weeks, with some surveys giving her leads of more than 20 points. Instead, Crowley surged late and defeated Hong by fewer than 4,000 votes. Even Marquette — long considered the gold standard of Wisconsin polling — was well off on the final result.
While its final pre-primary poll showed Hong at 38 percent, almost exactly matching the 39.3 percent she ultimately received, it also showed Crowley in third place at 7% of the vote. The survey was conducted before Mandela Barnes dropped out, but Marquette’s polling suggested Barnes’ supporters were more likely to migrate toward Hong than Crowley. That ultimately didn’t happen.
The recent misses provide an important backdrop for Wednesday’s poll. But there is another reason to exercise caution: Marquette’s August polling has historically been a shaky predictor of where Wisconsin races ultimately finish in November.
August is a snapshot, not a prediction
Candidates often receive a post-primary bump as their parties consolidate and more general-election voters begin paying attention. Meanwhile, the barrage of fall advertising and campaign developments that will ultimately shape the race has barely begun.
Marquette’s August polling over the past decade shows just how dramatically races can move.
In August 2016, Marquette had Hillary Clinton leading Donald Trump by 10 points and Russ Feingold leading Ron Johnson by six. Three months later, Trump carried Wisconsin by 0.8 points, while Johnson won by 3.3.
In 2018, Marquette showed Tony Evers and Scott Walker tied, close to Evers’ eventual one-point victory. But polling in the Senate race showed Republican Leah Vukmir trailing Tammy Baldwin by only three points in August. Baldwin would ultimately win by 11 points in November.
In August 2020, Marquette showed Joe Biden leading Trump by six points. Biden ultimately carried Wisconsin by less than a percent.
Perhaps the most striking example came in 2022, when Marquette had Mandela Barnes leading Johnson by an astonishing seven points and Evers leading Tim Michels by two. Evers eventually won by roughly three percentage points, while Johnson erased Barnes’ seven-point advantage and won by approximately one point.
Then came 2024. Marquette’s August poll had Trump ahead by one point and Baldwin leading Eric Hovde by seven. Trump’s margin proved almost exactly right, but Baldwin’s seven-point advantage dwindled to roughly one point by Election Day.
The history isn’t evidence that Marquette’s polling is inherently unreliable. In some cases, its August numbers proved remarkably close to the final result. But in others, the final margin was well off from the August polling. Candidates who appeared to be cruising in August found themselves in nail-biters by November, while candidates who appeared to be losing comfortably went on to win.
That’s because an August poll isn’t a November prediction. It’s a snapshot of an electorate still emerging from the primary, before many general-election voters have fully tuned in and before millions of dollars in campaign advertising begin shaping the race.
What to watch Wednesday
Wednesday’s poll will offer the first major indication of whether Crowley received a post-primary bounce and where his race against Republican Tom Tiffany stands as Wisconsin turns toward the general election.
But if the last decade is any indication, Wednesday’s numbers will tell us where the race begins — not necessarily where it ends.
