A note on measurement

To control for the see-saw nature of politics, most of the statistics I present are the “relative margin,” which is the partisan vote minus the statewide lean. First, I calculate the area’s vote margin (% Democrat minus % Republican); then I subtract the statewide margin. For example, the City of Waukesha voted 42% for Evers and 56% for Walker, so its absolute margin is -14%. The state as a whole had a margin of +1%. So Waukesha’s relative margin is -15% because it voted 15% more Republican than the state as a whole.

Overview

Since 1990, the Greater Milwaukee Area combined has voted Democratic in 5 out of 7 Presidential elections and Republican in 7 out of 9 Gubernatorial races. In his first three elections, Scott Walker won the Milwaukee Area by 47,200; 44,900; and 47,900 votes, respectively. In 2018 he lost by 17,600 votes. Since Tony Evers only won statewide by 29,227 votes, you could say Walker’s deteriorating support in the Milwaukee area cost him the election.

So what changed? Let’s first examine the vote in three broad swathes of the Milwaukee Area which collectively hold 85% of the region’s population.

1. The City of Milwaukee, home to 595,000 people or about 44% of the population.
2. Milwaukee County’s 18 suburbs, home to 357,000 people–about 23% of the population.
3. Waukesha County’s 38 suburbs, home to 401,000 residents, or about 30% of the total.

Each of these areas has followed a different trajectory over the last few decades.

Consider the pre-Trump era. From 1990 to 2014, the City of Milwaukee grew steadily more Democratic. By the 2010s, Democrats in Milwaukee were beating their statewide performance by more than 50 points. On the other extreme, Waukesha County grew steadily redder. In the early 1990s, it usually voted around 26 points more Republican than the rest of the state. By the 2012/2014 elections, this had increased to 40. The Milwaukee County suburbs began the ’90s around 7% more Republican than the state, but gradually moved closer to the state average.

The two elections of the Trump era suggest things have changed. In Milwaukee City support for Clinton was even higher relative to the state average than it was for Obama in 2012. Conversely, there was a slight decline in the Democrat’s relative performance between the governor’s race in 2014 and 2018.

The big changes happened in the suburbs. In 2012, the Milwaukee County suburbs voted 4% more Republican than the rest of the state. In 2016, they voted 11% more Democratic. The shift between gubernatorial races was smaller (1% more Republican in 2014 compared with 6% more Democratic in 2018) but similarly abrupt and consequential.

Waukesha County experienced a strikingly similar trend. In 2012 it voted 41 points more Republican than the state overall. In 2016 this fell to 26 points. Likewise, Waukesha voted 40% more Republican than the state in Walker’s 2014 reelection, but this fell to 35% in his 2018 defeat.

Pre-Trump vs Trump-era

To get a better sense of the overall shifts from the pre-Trump to the current Trump era of partisan politics, I generated two statistics. First, I calculated the average adjusted margin of the 2012 and 2014 elections for president and governor (left map). Then I found the same statistic for the following elections in 2016 and 2018 (middle map). The difference between these two numbers shows a durable shift toward the Democratic party across nearly all Milwaukee Area suburbs, but the size of this shift varies considerably (right map).

Whitefish Bay, Fox Point, Bayside, Wauwatosa, and Elm Grove all shifted toward the Democrats by over 20% relative to the rest of the state. A handful of smaller, more far-flung communities actually shifted toward the Republicans. These include Newburg, Kewaskum, Big Bend, and Wayne. Barton, Hartford, Cudahy, South Milwaukee, Addison, and Farmington all experienced less than a 1% shift.

So what causes some suburbs to be more Democratic than others, and why have some changed greatly in the Trump era while others haven’t? To explore those questions, I created two regression models. The first model takes the margin in 2018 governor’s race as its dependent variable. The second model takes the pre-Trump to Trump era shift described above. In each case I test the influence of the following independent variables: % Black or African-American, % Latinx (of any race), % with a graduate degree, median income (divided by 10,000), median age, and the percent of workers commuting to the City of Milwaukee. In both models the unit of analysis is suburbs in Milwaukee, Waukesha, Ozaukee, and Washington counties. Each suburb is weighted to its total population, so large communities count more than small ones.

All of these variables have a significant, independent effect on the 2018 vote margin. Collectively, the model explains 93% of the variance–a remarkably high figure for this sort of analysis. Perhaps the most surprising result is the strong correlation between the portion of workers commuting to Milwaukee and improved Democratic performance. For every additional 1% of workers commuting to the central city, Tony Evers improved his margin of victory by about 0.93%.

Fewer variables correlate strongly with the partisan shift of a given suburb from the pre-Trump elections of 2012-14 to the Trump-era elections of 2016-18. Income and age don’t have a significant relationship one way or the other. The share of workers commuting to Milwaukee actually has a small negative coefficient. Conversely, larger Black and Latinx shares of the population do positively correlate with a larger shift toward the Democrats. The biggest predictor of a pro-Democratic shift is, by far, high levels of education. For every additional 1% of the population with a graduate degree, the average suburb shifted by 0.8% toward the Democrats from 2012/14 to 2016/18.

Statewide consequences

Tony Evers’ 2018 victory is only the second time since (at least) 1990 that a Democratic candidate for governor carried the 4-county Milwaukee Area. The last time it happened was 2006 when Jim Doyle won reelection. In that year, Doyle won the state handily by 7.4%, but he only eked out a victory in Greater Milwaukee by 1,578 votes.

Things were much different twelve years later. Evers’ performance in the Milwaukee Area (a 2.4% lead) exceeded his statewide victory of 1.1%. Continued Democratic strength in the central city combined with eroding Republican support in the suburbs was more than enough to overcome growing GOP margins in the more rural regions of the state.