Wisconsin's purple state tradition doesn't match demographics

Why is Wisconsin a battleground state?

It’s a question I’ve been hearing (and pondering) for decades because Wisconsin has been a battleground for a very long time.

The explanations for that are not quite the same today as they were 20 or 30 years ago. The state’s political map has changed. This is a battleground for different reasons in 2024 than in 2000.

But at the heart of Wisconsin’s remarkably enduring competitiveness is a bit of a puzzle.

Demographically speaking, you would expect this to be a red state, not an almost perfectly purple one.

Why?

Because it’s overwhelmingly white, heavily blue-collar and disproportionately rural.

White non-college voters and white rural voters make up a larger share of the electorate in Wisconsin than they do in any of the six other 2024 battlegrounds: Michigan, Pennsylvania, Nevada, Arizona, Georgia and North Carolina.

In most states, these are very Republican-leaning groups.

In Wisconsin, too, they have shifted in a more Republican direction.

But crucially, they are still considerably less Republican than they are nationally; otherwise, this would not be a state famous for its cliffhanger elections.

Put another way, Wisconsin’s demographic makeup pulls it in a “red” direction, but its political character and culture pull it back in a “blue” direction. Presto. We’re purple.

This makes Wisconsin a great illustration of both the power and the limits of demographics — data that describe the characteristics of a population — in understanding how people vote.

Demographics tell us more about this state’s political map than it used to. They do a better job of predicting how a place is going to vote than they did even a decade ago. They help explain why some parts of Wisconsin are getting redder and some are getting bluer.

But demographics aren’t everything. They fall dramatically short of explaining the politics of many communities. There are places in Wisconsin that vote completely differently, but closely resemble each other with respect to race, education, population density and other basic demographic measures that pollsters, professors and political analysts use to track the trends and fault lines in the American electorate.

To help illustrate this, my Marquette Law School colleague John Johnson created a basic statistical model to see how well the demographic makeup of counties across the state and the nation explain the way they vote for president.

He looked at the makeup of every county with respect to race, education, income, age, religion and population density. Nationally, these variables explain about 75% of the variation in how counties vote.

Johnson calculated how these counties would have voted in 2020 based on their demographics alone. Then he compared that to how they actually voted. The difference between the two gives us a rough sense of whether a place is redder or bluer than their demographics suggest.

The findings in Wisconsin are pretty striking.

One is that the vast majority of Wisconsin’s 72 counties — even the red ones — are more Democratic, or less Republican, than this demographic data alone would have predicted.

In fact, the only counties that are “redder” than you’d expect based on their demographics are the three “WOW” counties outside Milwaukee (Waukesha, Ozaukee, and Washington) and tiny Florence County on the Michigan border.

The other 68 Wisconsin counties are all bluer than “predicted” by the model, including almost all the state’s dozens of smaller, heavily rural counties.

This reflects the fact, as noted above, that rural and blue-collar whites are not as lopsidedly Republican in Wisconsin as they are nationally.

In polling by the Marquette Law School this fall, for example, “non-college” whites favor Republican Donald Trump over Democrat Kamala Harris by 35 points nationally, but by a much smaller margin – 9 points – in Wisconsin.

In polling earlier this fall by the New York Times, rural voters favored Trump by just under 20 points in Wisconsin and Michigan, but by 26 in Pennsylvania, 31 in Georgia, and 35 nationally.

Just as the relationship between demographics and voting is different in Wisconsin than it is nationally, the relationship also differs widely from one Wisconsin county to another. In other words, some counties that appear pretty similar demographically look nothing like each other politically.

Take two small rural counties in Wisconsin: Vernon in the southwest and Rusk in the northwest.

They are both about 95% white. They are both sparsely populated.  In both, less than 15% of adults have four-year college degrees.  Vernon has a younger population and a higher median income, but the demographic model predicts that Republicans should have won the 2020 presidential race in both Vernon and Rusk by about 50 points (had these two counties voted like the average of their demographic counterparts nationally).

Instead, Trump won Rusk by 35 points and Vernon by only 5, a massive difference.

Similarly, Waukesha in the southeast, Brown in the northeast and La Crosse in the west are three larger, more densely populated counties than Vernon and Rusk.  They have higher rates of college education. All three are about 90% white. Waukesha has proportionately more college grads and a higher median income than Brown and La Crosse. La Crosse has a much smaller share of Catholics than Brown and Waukesha.

But the demographic model predicts fairly similar election outcomes, ranging from a 9-point GOP edge in Waukesha to an 11-point GOP edge in La Crosse to a 15-point GOP edge in Brown (home to Green Bay).

Instead, the actual voting differences are much larger. Trump in 2020 won Waukesha by 21, won Brown by 7 and lost La Crosse by 14.

More: Once a GOP juggernaut, Waukesha County's shifting voters makes for Wisconsin battleground

One conclusion from all this is that our demographic model isn’t all-encompassing (and isn’t intended to be). It doesn’t contain every possible data point about the population, such as the nature of the employment base, or whether the region was settled in a past century by Germans or Norwegians, or whether a county's politics are influenced by its proximity to Madison or Milwaukee or Minnesota's Twin Cities.

But another conclusion is that demographic traits and other metrics can’t account for all the differences we see in how states or the communities within them vote, because they have different political cultures and different political histories.

At the same time, demographics do have real value in understanding political change, including the two parties’ shifting coalitions.

When I started covering Wisconsin elections in the late 1980s, two things really jumped out about the state’s political map.

One involved southwest Wisconsin, where overwhelmingly rural white counties were consistently voting Democratic for president. Nowhere in America was there a greater concentration of blue-leaning rural white counties than the upper Mississippi River where western Wisconsin, southeastern Minnesota and northeastern Iowa come together in what’s known as the Driftless Area. It helped explain why the first thing Democrat Al Gore did after the 2000 Democratic convention was fly to La Crosse and campaign along the Mississippi in a houseboat.

The other outstanding “quirk” of the Wisconsin political map involved the WOW Counties outside Milwaukee, full of ultra-Republican, hyper-mobilized communities that were collectively far “redder” than the suburbs of virtually any other northern metropolis.

These two regions were glaring examples of Wisconsin’s distinctive political character and how it defied powerful national voting patterns.

And to a significant degree, both areas still defy national patterns. Those rural southwestern counties like Vernon are still way bluer than demographics alone would suggest. And the WOW counties such as Waukesha are practically the only counties in Wisconsin that are redder than predicted by Johnson’s demographic model.

But neither region is as idiosyncratic as it used to be, because their demographic makeup is making its power felt. Across the country, college-educated suburban whites have been trending Democratic. Rural and blue-collar whites have been trending Republican.

In the Trump era, the state’s rural southwest has gotten less blue. Vernon voted Democratic for president by 8 points in the jump-ball presidential election of 2004 but Republican by 5 points in the jump-ball election of 2020. And the suburban ring around Milwaukee has gotten a lot less red. Waukesha voted Republican by 35 points in 2004 but 21 in 2020.

In fact, the demographic model fits much better with the 2016 election in Wisconsin than the 2012 election, and even better with the 2020 election than the 2016 election.

Wisconsin still has a political character of its own. The same demographic segments of the electorate still vote differently from one state to another and one community to another.

But our politics is more nationalized. Rural Wisconsin is following the national rural trend. Suburban Wisconsin is following the national suburban trend.

And when we wake up after Election Day and see how Wisconsin’s political map has changed, demographics will be a big part of the explanation.

(This story was updated because an earlier version included an inaccuracy.)

Craig Gilbert provides Wisconsin political analysis as a fellow with Marquette University Law School's Lubar Center for Public Policy Research and Civic Education. Prior to the fellowship, Gilbert reported on politics for 35 years at the Journal Sentinel, the last 25 in its Washington Bureau. His column continues that independent reporting tradition and goes through the established Journal Sentinel editing process.

Follow him on Twitter: @Wisvoter.

This article originally appeared on Milwaukee Journal Sentinel: Wisconsin's purple state tradition doesn't match demographics