On Tuesday, voters will head to the polls in Wisconsin to choose a new state Supreme Court justice between the Republican-aligned former Supreme Court justice Dan Kelly and the Democratic-aligned judge Janet Protasiewicz for a seat on the state’s Supreme Court. With control of the court on the line and issues such as redistricting, abortion access, and possibly even the 2024 presidential election at stake, it is fair to say that the partisan stakes are as high as they could possibly be in this battleground state.
Because of the nonpartisan election system and a dearth of polling, the bands of uncertainty in this race are significantly higher than normal, limiting our ability to make confident predictions. However, there is some data we can use to get a better understanding of what’s going on.
In the absence of polling, one of the first data points to use is spending and media airtime. Here, Democrats have considerably outraised the Republicans in candidate fundraising, leading the GOP to rely on their commanding edge in outside spending (and a massive late influx of cash) to narrow the gap. Per AdImpact, Protasiewicz has outraised Kelly $11.3M to $0.9M, while Republican outside groups have outspent their Democratic counterparts $11M to $3M.
On the face of it, this might indicate that Democrats have a narrow edge in spending. However, it is worth noting that candidate dollars usually go much further towards booking airtime than independent expenditures (IEs) do because candidates get cheaper advertising rates. Medium Buying suggests that until March 23, Protasiewicz maintained an incredibly commanding lead on the airwaves, booking 66% of all media airtime.
Since then, a late influx of outside GOP cash has narrowed the airtime disparity significantly, and over the course of the last week, Republicans have nearly matched Democrats on the airwaves, with pro-Democratic groups seeing their share reduced to 52% of media impressions.
Part of the problem here for Kelly, however, is that this intervention still has not dragged him to parity, and there are questions about whether this may have come a little bit too late, especially given that early voting has already been going on for the last few weeks, and because he has had to withstand a sustained and massive on-air disadvantage for quite some time.
As we mentioned in our primary recap, in choosing Dan Kelly over Waukesha County Judge Jennifer Dorow, the GOP nominated the more populist, controversial, and Trump-adjacent candidate. Kelly has been linked to the failed false electors scheme and has made a number of controversial comments on hot-button issues, likening Social Security and affirmative action to slavery. He has also made past remarks opposing abortion in a state that is extremely pro-choice.
In our view, all of these factors make Kelly a greater general election liability than Dorow would have been, especially among left-trending, college-educated voters in the Milwaukee suburbs. The concerning point for Republicans is that this demographic was a large part of Dorow’s base, and Kelly’s weaknesses here appear especially glaring, especially after a brutal primary and an even harsher set of attacks in the general election. As our modeling will show, unless the GOP receives a historic election-day turnout surge, Kelly can almost certainly not afford to lose any portion of Dorow voters if he is to win this election.
In February’s nonpartisan jungle primary, Democrats enjoyed a fairly good turnout environment en route to a combined 54–46 lead in the first round, especially compared to their underwhelming turnout in the November general election. This time around, the turnout differential is not as pro-Democratic as it was in the D+8 February primary.
This is evident in a map made by Ryan Brune, where we can see that the Democratic bulwarks of Madison and Milwaukee currently combine for a lower share of the statewide early electorate than they did in the primary runoff. And although the rurals are turning out at a lower rate as well, the ruby-red Milwaukee suburbs (Waukesha, Ozaukee, and Washington counties) are all making up a much higher share of the electorate.
However, this electorate is still more Democratic than the one seen in the November 2022 general election, where Tony Evers still won by 3.4 and Mandela Barnes only lost by 1. Madison constitutes a higher share of the early vote overall than it did in November, and Milwaukee is close to unchanged in statewide turnout share. Meanwhile, the more Republican rurals have seen a precipitous plunge in turnout without a proportionately high spike from the red Milwaukee suburbs to compensate.
This makes the current overall election picture more Democratic-leaning than not. But what, exactly, is the picture?
Unlike Georgia, Wisconsin only shows the total number of ballots cast by vote type (mail, early in-person, and election day), and in most cases, it does not break out the result splits by ballot type or by precinct — only by municipality. All of this makes it impossible for us to construct a model that is as precise or as sophisticated as our Georgia runoffs model. However, we can still do some geographic-based modeling to gain a better understanding of just what is happening and to gauge the turnout environment that we find ourselves in.
When modeling this, we needed to figure out which margins to use on a geographic level, because we cannot accurately model persuasion well. Given the results of the primary, the spending and advertising disparity, and the fact that this is a statewide and not federal race, we believe that the best thing to do is to use a blend of Tony Evers’ margins and the Democratic margins in the Supreme Court primary. Under this scenario, our municipality-based model gets something where Protasiewicz is clearly favored.
How much so? If Dan Kelly got Ron Johnson’s margins, this race would only be a pure coin-flip; however, if he got margins closer to those that Tim Michels pulled against Evers (which we believe to be a better baseline), it would be very difficult to envision a conservative win here. And if he lost even 3% of Dorow’s voters, this race would become an even more difficult climb for him.
In past Supreme Court elections, the differential between primary and general election composites has averaged roughly 10 percent, which means that our bands of uncertainty with regards to the final margin should be high. But even then, with the available data, Protasiewicz emerges the winner in the majority of probable outcomes. In essence, assuming no massive coalitional or persuasion-based shifts, we believe Janet Protasiewicz is a clear favorite. Under these circumstances, we think that she is on track to win by roughly 6 points and is a 6:1 favorite.
We should note that due to the dearth of polling, our model cannot track persuasion well, and thus has its limitations. But it is quite robust when coalitions are roughly static, as they have been in Wisconsin over the last few elections, and it does remarkably well when backtested. This modeling approach suggested that Evers was on track to win by 2.8 in 2022 (he won by 3.4), and that liberals were on track to win the nonpartisan primary by 6 (they won by 8). Given that post-2016 elections in Wisconsin are increasingly polarized, we expect coalitions to be quite similar to those seen in the primary and the gubernatorial election, and so we are comfortable with using this approach here.
Dan Kelly still has a very decent chance of winning. In fact, his odds of victory (~15%) are only slightly lower than what the New York Times gave Donald Trump in 2016. But between the early vote and the advertising discrepancy, there is not a single datapoint that suggests he is favored in this election. A Kelly win is possible, but it would be a clear upset. Our rating for this race is Likely Protasiewicz.
SD08 Special Election
We favor Republican Dan Knodl to win the SD-08 special election against Democrat Jodi Habush Sinykin to replace outgoing Senator Alberta Darling. This Trump +5, Michels +3.6 seat in the northern Milwaukee suburbs is democratic trending, but Republicans benefit from the effects of down-ballot lag.
Because the concentration of high-propensity, persuadable Dorow voters in the WOW counties is higher than elsewhere in the state, we expect Knodl to outrun Kelly by a few points, especially given how in 2020, Darling outperformed the top of the ticket by three points. Sinykin has outspent Knodl, but we do not think this is enough to mark her as favored in a district that is so Republican downballot. If Protasiewicz wins by more than expected, this race could get close, but we believe it is a race in which the GOP still holds the clear advantage going in. Our rating for the special is Leans Republican — an upset is clearly possible, but the Republicans enter as favorites.
Editor note: A previous iteration of this article incorrectly stated that Knodl outspent Sinykin.
I’m a software engineer and a computer scientist (UC Berkeley class of 2019 BA, class of 2020 MS) who has an interest in machine learning, politics, and electoral data. I’m a partner at Split Ticket, handle our Senate races, and make many kinds of electoral models.
My name is Harrison Lavelle and I am a political analyst studying political science and international studies at the College of New Jersey. As a co-founder and partner at Split Ticket, I coordinate our House coverage. I write about a variety of electoral topics and produce political maps. Besides elections, my hobbies include music, history, language, aviation, and fitness.
Contact me at @HWLavelleMaps or firstname.lastname@example.org
Leave a Reply