Comparing 4 Voting Methods: Chicago Mayoral Election 2023

(UPDATED April 3, 2023)

Results of an online survey suggest that alternative voting methods could dramatically reshape Chicago’s politics, and that different voting methods might do so in different ways — with implications for campaign funding incentives, the quality of public discourse, and the viability of different types of candidates.

The 2023 Chicago municipal election took place on February 28. Besides mayor, Chicagoans voted on alderpersons (mini-mayors of Chicago’s 50 wards who comprise Chicago’s City Council) and police district council representatives. There were nine mayoral candidates on the ballot (plus several declared write-in candidates).

I created an online survey using Google Forms where people could try out voting on the nine official mayoral candidates using four different methods of voting. I intended it in part as an educational tool to spread awareness about alternative voting methods, and in part as an experiment to see if different voting methods, with the same set of voters, would yield different outcomes and/or paint a different picture of popular opinion.

What follows is an in-depth discussion of the election and the results of the survey. If you don’t have time to read through it all, feel free to scroll to the bottom, where all the results are summarized in a single table. Independent journalist Eric Zorn has also summarized these results succinctly in his online newsletter.

The four voting methods used in this survey included:

• Plurality voting plus top two runoff (the current method used in Chicago): Vote for one candidate only. The candidate with the most votes wins. If no candidate receives a majority, the top two candidates advance to a runoff.
• Approval voting (or “pick-all-you-like voting”): Vote for all the candidates you approve of. The candidate with the most votes wins OR the top two candidates advance to a runoff, depending on which version of approval voting is used. Two cities currently use approval voting: St. Louis, MO and Fargo, ND. The Center for Election Science is an organization that advocates for this method.
• Ranked choice voting (or instant runoff voting): Rank up to 5 candidates.¹ If no candidate receives a majority, the lowest scoring candidate is eliminated and those votes are redistributed based on voters’ preferences; the process is repeated until a candidate achieves a majority. Ranked choice voting is currently used statewide in Maine and Alaska and in several cities around the country, including New York City, Minneapolis, and San Francisco. Evanston, IL plans to start using ranked choice voting in 2025. FairVote is an organization that advocates for this method.
• STAR voting (Score-Then-Automatic-Runoff): Rate candidates on a scale of 0 to 5 stars; candidates left blank receive 0 stars. The two top-scoring candidates are finalists. An automatic runoff phase determines which of the two finalists voters prefer based on relative rankings. STAR voting is not yet used in any public elections, but there is a campaign to adopt the method statewide in Oregon. Equal Vote Coalition is an organization that advocates for this method.

See my previous post for a more in-depth introduction to each of these methods, plus a discussion of how in this particular mayoral election, Chicago’s method of voting artificially disadvantages Black and progressive candidates.

I created the survey on February 1 and closed it on February 28 soon after polls closed. 326 people responded. I discarded 10 responses, either because the respondent didn’t fill out the STAR voting ballot or because they filled out the ranked choice ballot incorrectly, leaving 316 respondents who submitted valid responses to all 4 polls.

If you would like to analyze the data yourself, you can access it here. Please let me know if I made any mistakes.

Limitations:

• This was not a randomized study and it doesn’t indicate the views of the Chicago electorate. Specifically, the respondents skewed much more progressive than the Chicago electorate.
• Since respondents voluntarily took the time to respond to a fairly complex political survey, they were probably more politically engaged than the average voter and might have been more likely to make full use of the alternative voting methods’ expressive potential.
• Respondents who voted strategically did so based on plurality polling of Chicago’s actual electorate, not on polls of this specific group of respondents (which would have indicated different frontrunners), nor on polls that use any of the alternative voting methods.
• The survey can’t account for the possibility that if Chicago actually used any of these alternative voting methods, a different set of candidates might have decided to run, or different candidates might have attracted funding and endorsements.

Despite these limitations, I think these results still might yield useful insights into the properties of the voting methods that may be helpful for deciding which one is a best fit for Chicago.

I’m very grateful to everyone who took the time to submit responses, as well as to the several people who helped share the survey.

Cast of Characters

(If you don’t want to read all this background, skip ahead to the results and refer back here for reference.)

Chicago’s municipal elections are officially nonpartisan. The nine candidates who ran for mayor seemed to fall into three broad ideological groups:

Left Progressives

These candidates support ambitious redistributive economic policies, are closely aligned with social movements, and are skeptical of policing-based solutions to public safety. They include three candidates:

• Ja’Mal Green as CONTRARIAN OUTSIDER — Ja’Mal (pronounced “Jay-Mall”) is a 27-year-old community organizer. His surname is Green, his yard signs are green, and he supports a Green New Deal. His platform is full of ambitious ideas, like expanded cash transfers to people in poverty, a public Bank of Chicago, and universal pre-K education from the age of 3. His political style is combative and he seems to enjoy insulting other candidates, plus random people on Twitter.
• Brandon Johnson as SOCIAL MOVEMENT CANDIDATE — Brandon is a Cook County Commissioner who lives in Chicago’s Austin neighborhood. He’s worked as a public school teacher and as an organizer with the Chicago Teachers Union, which endorsed him and funded his candidacy. He has endorsements from several other unions and progressive groups. He wants to open mental health clinics, provide youth job training, tax expensive real estate transfers to fund homelessness response, and promote some cops to detectives to help solve crimes (half of Chicago’s homicides go unsolved, disproportionately in Black neighborhoods).
• Kam Buckner as URBANIST ALTERNATIVE — Kam is an Illinois House Member who led a successful charge to increase state funding for public transit. He’s a regular rider of the CTA, and his public transportation plan is ambitious: he wants to expand and better integrate CTA and Metra train lines, and establish separate bus lanes to make service faster. He also wants to hire social workers to respond to crises instead of police, and establish new City Charter guaranteeing a powerful, independent City Council. He’s a former college football player who comes from a family of soul singers. He considers himself a “pro-business progressive.”

Moderate Progressives

These candidates also support redistributive economic policies, but are more closely tied to establishment politics. They support a mixed public safety approach that combines policing-based solutions with alternatives to policing. They include four candidates:

• Chuy García as PROGRESSIVE ELDER STATESMAN — Chuy has been a force in Chicago politics since the 80s, when he was an alderman and an ally of Chicago’s first Black mayor, Harold Washington. Since then, he’s been a state senator and a county commissioner, and is currently a member of the U.S. House of Representatives. He ran for mayor in 2015 with the endorsement of the Teachers Union, but was defeated in the runoff by Rahm Emmanuel. His platform calls for a Municipal Restitution Commission to address health disparities and a Chicago Dream Act to provide internship opportunities for immigrant youth. He wants to expand and modernize the police force. He immigrated from Mexico at the age of 9 and lives in Chicago’s Little Village neighborhood.
• Sophia King as PRAGMATIC ALDERPERSON #1 — Sophia is a South Side alderwoman and chair of the Progressive Caucus on City Council. Her campaign calls for “safety and justice.” She wants to expand the size of the police force while also providing alternative 911 responses and criminal justice reform. She wants to attract more tech jobs and make Chicago the “Silicon Valley of the Midwest.” She’s a longtime friend of the Obama family.
• Roderick Sawyer as PRAGMATIC ALDERPERSON #2 — Rod is a South Side alderman, a founding member of City Council’s Progressive Caucus, and chair of City Council’s Black Caucus. He wants to aggressively incentivize development of abandoned properties in blighted and disinvested areas, while reforming Chicago’s TIF (Tax Increment Financing) program so that tax dollars are only used to fund private development in neighborhoods that really need investment. He wants to expand the police force, improve police training, and implement community policing (where officers focus on one neighborhood for an extended period and get to know the community). He is the son of longtime alderman and second Black mayor of Chicago Eugene Sawyer, and practically grew up in City Hall.
• Lori Lightfoot at BELEAGUERED INCUMBENT — a former prosecutor, Lori was elected in 2019, as Chicago’s first Black female mayor, and the nation’s first openly LGBT Black female mayor. She presided over challenging times: the pandemic, the George Floyd protests, the homicide spike. She has an combative demeanor and managed to alienate both the right and the left — for example in public clashes with both the progressive Chicago Teachers Union and the conservative Fraternal Order of Police. As mayor, she increased Chicago’s minimum wage, piloted a guaranteed income program, implemented a program called INVEST South/West to incentivize development in underinvested neighborhoods, and paved the way for a future casino to open in Chicago.

Urban Conservatives

These candidates might not be viewed as particularly conservative within state or national politics, but in a major metropolitan city, they constitute the right wing. They support charter schools and school vouchers, policing-based solutions to crime and public safety, and low taxes to make the city attractive to businesses. They include two candidates:

• Paul Vallas as MANAGERIAL INSIDER — prior to running for lieutenant governor of Illinois on the unsuccessful Democratic ticket in 2014, Paul had a career as a superintendent/CEO of public school systems in Chicago, Philadelphia, New Orleans, and Bridgeport, CT. He expanded charter schools in all those places. He comes from a family of cops, is endorsed by Chicago’s powerful Fraternal Order of Police, and is running a tough-on-crime campaign, insisting “public safety is a human right.” He wants an expanded police force, community policing, and a witness protection program. He also wants a cap on property taxes.
• Willie Wilson as PEOPLE’S BUSINESSMAN — Willie Wilson grew up working in cotton and sugarcane fields, the son of a Louisiana sharecropper. He moved to Chicago in the 60s and became a wealthy businessman, founding a medical supply company. Back for his third self-funded mayoral campaign, Wilson wants to lower taxes and “take the handcuffs off the police and put them on the criminals.” He has said the solution to crime is for police to go after criminals and “hunt them down like a rabbit.” He wants Chicago to stop welcoming asylum seekers and focus on helping the people who already live here. Over the past several years, he has engaged in splashy public philanthropy, giving away millions of dollars in gas, groceries, and cash.

Frontrunners

Polls showed that only four candidates had a realistic chance of advancing to the runoff. Paul Vallas, endorsed by the Fraternal Order of Police, was consistently at the top of the polls. Three candidates vied for second place: Lori Lightfoot, the incumbent mayor, Brandon Johnson, endorsed by the Chicago Teachers Union, and Chuy García, the the progressive runner-up for mayor in 2015.

Demographics

This was an unusual field of candidates demographically. Seven of the candidates were Black, even though Chicago’s Black population has been falling in recent years and is now less than 30% of the total population. Chuy García was the only Latino candidate and Paul Vallas the only White candidate. Journalist Natalie Moore, author of the excellent history The South Side: A Portrait of Chicago and American Segregation, commented recently that the idea that Chicago’s Black community needs to unify around a consensus candidate has faded from the discourse in recent years.

The Actual Election Results

36% of eligible voters turned out to the polls. As expected, no candidate got a majority. The two top candidates were Paul Vallas and Brandon Johnson, who advanced to the runoff election. 26% of voters supported candidates I categorized as left progressives, 32% supported moderate progressives, and 42% supported urban conservatives.

This was an unusual election for Chicago in that the incumbent mayor didn’t even advance to the runoff. Lori Lightfoot became Chicago’s first incumbent mayor in 40 years to lose a re-election bid.

This year, the top two candidates collectively received a majority of the vote (54.5%), which was a pleasant surprise; in the previous mayoral election in 2019, the top two candidates, Lori Lightfoot and Toni Preckwinkle, collectively received only 33.5%.

However, this year there were four aldermanic elections where the top two candidates collectively received less than 50% support (wards 5, 6, 21, and 48), which means those candidates advanced to runoffs despite most voters in their wards having preferred other candidates. Three of those wards each had 10 or more candidates for alderperson, splitting the vote.

This was the first municipal election where voters could vote for police district council representatives. Each of the city’s 25 police districts could elect a three-member council to address community concerns with policing. However, the voting method Chicago chose to use was plurality block voting — vote for 3 candidates, and the 3 highest scoring candidates win seats. This method results in interest groups running slates of candidates, and an entire slate of candidates usually wins.

Plurality block voting (also used to elect local Metropolitan Water Reclamation District Commissioners) is the same voting method Skokie (a nearby suburb) just voted last November to abolish because it had resulted in single-party rule there since 1965. It is interesting that, for police oversight, Chicago would choose a voting method that functions to ensure uniformity of opinion.

A Stark, Binary Choice

The runoff between Vallas and Johnson presented voters with a choice of two radically different visions for the city. For both Vallas’s and Johnson’s supporters, the election must have been exciting but also terrifying. For anyone caught in the middle, it must have been profoundly frustrating.

Vallas and Johnson represented the extremes of Chicago’s political spectrum. Among the other candidates, only Wilson’s platform was more rightwing than Vallas’s, and only Green’s platform was arguably more leftwing than Johnson’s. The five candidates I would consider ideologically in between Vallas and Johnson collectively received 34.2% of the vote, more than either Vallas or Johnson received individually.

This is an example of the center-squeeze effect — a systemic disadvantage against candidates in the ideological middle — and it’s a feature of plurality voting. In this case, it’s possible that Lightfoot, García, or another candidate might have been able to defeat both Vallas and Johnson in head-to-head races, but those candidates were eliminated in the first round.

During the runoff campaign, both sides engaged in negative, fear-based campaigning. Vallas accused Johnson of being a radical who wants to defund the police, and Johnson accused Vallas of secretly being a Republican with ties to the far right. Fear-mongering and accusations largely replaced substantive policy debate. When the ideological contrast is so clear, most voters don’t need to know about actual policy details in order to come to a decision.

This dynamic probably sounds familiar because it’s the norm of how politics functions in the U.S. We are used to politics as a high-stakes football game between two radically opposing worldviews, a zero-sum contest where if your side loses then the other side wins. It’s a dynamic of fear that drives polarization, legislative gridlock, psychological stress, widespread disillusionment with politics, and, at the national scale, the perpetuation of two-party domination.

Our politics doesn’t have to be this way. Discourse could be more civil and more nuanced, and could bring a greater diversity of voices and ideas to the table. The enduring nastiness and shallowness of our politics is rooted in its underlying incentive structures. One of the most tractable ways to improve incentives is to switch to a better method of voting. Doing so won’t solve every problem, but in terms of improving our democracy and political discourse, it would probably be the most impactful single reform within the realm of realistic possibility.

Plurality Voting Survey Results

Just like in the actual election, the plurality voting survey results show Brandon Johnson and Paul Vallas advancing to a runoff. The survey respondents were not a representative sample of the Chicago electorate, so the percentages were very different from the real election. In the survey, Brandon Johnson finished in first place, rather than Paul Vallas. The top two candidates collectively got 62.1% of the vote.

The survey respondents were more progressive and more anti-incumbent than the Chicago electorate as a whole. Compared to the actual election results, Brandon Johnson and Kam Buckner received much more support, and Paul Vallas, Lori Lightfoot, and Willie Wilson received much less. Brandon Johnson received nearly 40% support, instead of less than 22%.

Proportionally, the most dramatic differences were Willie Wilson and Kam Buckner. Wilson got 9% in the actual election, but less than 1% in the survey. Buckner got 2% in the actual election, but 9% in the survey.

51% of survey respondents supported candidates I categorized as left progressives, whereas 26% supported moderate progressives and 23% supported urban conservatives (compared with 26%, 32%, and 42% respectively in the actual election).

Despite the huge differences in opinion between this group of survey respondents and the actual Chicago electorate, the outcome of the election is the same: Paul Vallas and Brandon Johnson face each other in a runoff. Even with this super progressive group of respondents, relative conservative Paul Vallas still makes the runoff! This shows the power of vote splitting and the tendency of plurality voting to favor polarizing candidates.

In this hypothetical runoff, Brandon Johnson’s victory seems almost certain. Johnson would only need to pick up another 11% from voters who supported García, Buckner, or other candidates. Likely, only a major scandal surfacing during the runoff campaign could prevent Johnson’s seemingly inevitable triumph.

In this case, the runoff would nearly be a pointless political exercise and a waste of money for the city. The outcome would be essentially predetermined, and the ritualistic clash between deeply contrasting ideological visions would not add nuance to the city’s public discourse. Moderate voters would feel shut out of the conversation, with no remaining moderate candidate to support — despite the fact that moderate candidates collectively received more support in the first round than conservative candidates.

Approval Voting Results

When the same group of respondents could support multiple candidates, Brandon Johnson still came in first place. However, he received over 61% support, instead of the less than 40% he got with plurality voting. Such strong majority approval might give Johnson a stronger sense of popular mandate. But should he run for reelection next time, he wouldn’t be able to take his victory for granted. Two other candidates also received majority approval: Kam Buckner (58%) and Chuy García (51%). Johnson defeated Buckner by only 10 votes.

Despite Buckner’s and García’s popularity with this group of respondents, demonstrated by the approval voting poll, the plurality voting poll would have excluded both of those candidates from the runoff. Paul Vallas, who would have advanced to the runoff with plurality voting, came in a distant fourth place with approval voting, with only 34% support.

With single-round approval voting, Brandon Johnson would simply win with these results. But if a two-round system were used, like in St. Louis, Johnson and Buckner would advance to a runoff. Johnson and Buckner are ideologically similar candidates, so public debate during the runoff likely would focus more on policy nuances than on broadly contrasting visions, and the campaigns would probably be less focused on stoking fears about the other candidate.

Even though both these finalists would be from the progressive left, the runoff would now force them to compete for moderate and conservative voters. Unlike plurality voting’s runoff between Johnson and Vallas, where Johnson’s victory would be essentially predetermined, approval voting’s runoff between Johnson and Buckner would actually be a competitive election, and moderate and conservative voters could make a decisive difference in the ultimate outcome.

Besides the question of who wins, there’s also the question of framing discourse. Approval voting paints a different picture of the electorate’s preferences among low scoring candidates. For example, the plurality poll suggests that Lori Lightfoot is more popular than Sophia King, whereas the approval poll shows King as more than twice as popular as Lightfoot, with King’s support almost as high as Vallas’s. Perhaps such a result would encourage King to run again in the next election, and improve her ability to attract funding and endorsements if she does. Approval voting makes visible King’s broader appeal, whereas plurality voting obscures it.

However, the most dramatic difference between the results of the plurality and approval polls is the strength of Kam Buckner’s candidacy.

What’s Up with Kam Buckner?

In the plurality voting poll, Buckner came in fourth place with 9%, showing him to be a minor, second-tier candidate. With approval voting, he came in second place with 58% support, a major frontrunner nearly as popular as Johnson. This result is all the more astounding when you consider that Buckner’s campaign only had around $300,000 of funding, while most of the candidates in this race had millions of dollars at their disposal.

In fact, Buckner easily could have won this poll. As people filled out the survey, I could see the plurality and approval results in real time (unlike ranked choice and STAR voting, which require more complex calculations). As the first few hundred responses came in, Buckner was consistently on top. It wasn’t until election day, the very last day the survey was open, that Johnson pulled ahead.

Why did Buckner perform so poorly with plurality voting despite the fact that he seems to have been extremely popular among respondents? Polls showed that Buckner had no chance of winning, so many people who liked Buckner probably felt compelled to vote for a frontrunner instead. Furthermore, respondents who liked winning candidate Brandon Johnson also liked Buckner, and vice versa: 74% of respondents who approved Johnson also approved Buckner, and 78% of respondents who approved Buckner also approved Johnson. Plurality voting only takes into account first-choice votes, so Buckner’s broad appeal does not register in the plurality results.

Once again, Buckner’s popularity with this group of respondents doesn’t mean he was similarly favored among the Chicago electorate. Buckner got only 2% support in the actual election. However, his emphasis on public transportation made him especially popular among policy aficionados who like to hang out on social media talking about urban planning — the exact sort of person who is likely to fill out a survey about alternative voting methods.

Nevertheless, I suspect there’s a chance that if Chicago used approval voting, Buckner might have emerged as a frontrunner. During the election campaign, progressives were mainly divided between two candidates, Brandon Johnson and Chuy García. Johnson’s policy proposals were more leftwing, but García was seen as a less polarizing candidate and an empathic communicator, perhaps more electable than Johnson.

Buckner’s policy proposals were similar to Johnson’s, but his humble communication style was similar to García’s. Throughout the campaign, rival candidates accused Johnson of being a pawn of the Teachers Union, and some progressives criticized García for his establishment ties, but Buckner was seen as an independent figure. Perhaps, if voters were able to support multiple candidates, Buckner might have emerged as a progressive unity candidate, attracting more funding and endorsements along the way.

A common critique of approval voting is that it might allow a bland candidate to win just by being everyone’s second choice. The approval voting poll shows a similar level of support for Johnson and Buckner, but were respondents genuinely enthusiastic about both candidates, or was Buckner simply the backup candidate of almost all of the Johnson supporters (or vice versa)? Ranked choice voting removes that ambiguity.

Ranked Choice Voting Results

When the same group of respondents could rank up to five candidates in order of preference, respondents felt more free to vote their conscience than with plurality voting: 19% of respondents chose a first choice candidate who was different from the candidate they supported in the plurality poll.

Brandon Johnson won again, but only after 7 candidates were eliminated in instant runoffs. Nevertheless, while the approval voting poll showed a close contest between the top candidates, ranked choice creates the impression of Johnson winning by a landslide, with two thirds of respondents preferring him to the second place candidate, Paul Vallas. Such an election result might give Johnson an even stronger sense of popular mandate: not only did he achieve majority support, but he seemed to do so without any close competition.

Kam Buckner, the second place candidate with approval voting, came in third place with ranked choice. Both Johnson and Buckner got lots of first choice votes — neither one was simply everyone’s second choice — but Johnson did get significantly more: 32% ranked him first, compared to 20% for Buckner. Respondents may have found the two candidates similarly acceptable (according to the approval voting poll), but it appears they found Johnson more exciting.

That’s information that ranked choice voting gives that approval voting doesn’t — but is it useful? It shows many respondents strongly liked Johnson, but perhaps they were counterbalanced by many others who disliked him. Perhaps Buckner was less strongly liked, but also less disliked, and maybe that’s one reason Johnson’s and Buckner’s scores were so much closer with approval voting.

Overall, the ranked choice voting results look somewhat in between the plurality voting results and the approval voting results. Like with approval voting, Kam Buckner outperformed Chuy García, and Sophia King outperformed Lori Lightfoot. But like with plurality voting, Paul Vallas came in second place to Brandon Johnson, so that two highest scoring candidates were polarizing opposites.

One critique of ranked choice voting is that, relative to other alternative methods, it favors polarizing candidates. These results seem to support that view, although it’s important to note that the effect isn’t as strong as with plurality voting. In this ranked choice poll, Kam Buckner, a less polarizing alternative to Johnson, at least did much better than with plurality, garnering 27% support instead of 9%. Ranked choice voting might help candidates like him attract funding and endorsements.

An Artificial Landslide Victory

Why does ranked choice voting show such a stronger victory for Johnson than approval voting? With ranked choice voting, Johnson’s lead was inflated by three factors: the discarding of ballots whose rankings were exhausted, the persistence of vote splitting dynamics, and the effect of the particular order in which candidates were eliminated.

1. Ballot exhaustion.

In this ranked choice poll, 7% of ballots (22 of them) were exhausted, meaning none of the candidates ranked made it to the final round. (This is below the average rate of ballot exhaustion in ranked choice elections nationally, which a 2019 analysis reported as over 10%.) Exhausted votes were excluded from the final tally, artificially inflating Johnson’s margin of victory. If, unconventionally, exhausted ballots were included in the final result, Johnson’s level of support would be 62% rather than 66%.

2. Vote splitting.

With ranked choice voting, no matter how many rankings a voter uses, that ballot can only support one candidate at a time. In the approval voting poll, a lot of Kam Buckner’s support came from people who also liked Brandon Johnson; in the ranked choice poll, Brandon Johnson was never eliminated, so candidates such as Buckner or Chuy García never got a chance to benefit from the second choice votes of Johnson supporters.

With ranked choice, like with plurality voting, popular candidates can see low levels of support merely because they are similar to other candidates. Unlike plurality voting, ranked choice is built to withstand the effects of vote splitting in the ultimate selection of the winner (and it successfully does so the vast majority of cases), but it doesn’t prevent vote splitting dynamics from affecting the results of other candidates, which can be important for framing political discourse.

3. Order of elimination.

With ranked choice voting, the order in which candidates are eliminated is the crucial factor that determines the level of support reported for various candidates. In this ranked choice poll, when Kam Buckner was eliminated, he had 82 votes and Vallas had 88. If just 7 fewer Vallas supporters had participated in the survey, Vallas would have been eliminated instead.²

Combining data from the ranked choice and STAR voting polls, I could see that Vallas supporters strongly preferred Buckner to Johnson: 61% of respondents who chose Vallas as their first choice in the ranked choice poll rated Buckner higher than Johnson in the STAR voting poll, whereas only 17% rated Johnson higher. (From their perspective, Buckner may be too progressive, but at least he’s not tied to the Teachers Union.)

Just to see what would happen, I recalculated the final instant runoff round, arbitrarily eliminating Vallas instead of Buckner. Johnson got 54% and Buckner 46%, with 13% of ballots exhausted. (If I include exhausted ballots in the final tally, Johnson would only have 47% support and Buckner 40%.) Thus, Johnson’s landslide win in the ranked choice poll is an artifact of the method itself rather than an accurate reflection of his popularity relative to other candidates.

In another ranked choice voting poll, promoted by the advocacy organization Reform for Illinois, Brandon Johnson won with 58% support, with Chuy García in second place and Kam Buckner in third. 1,860 people participated in that poll. If the same voters had voted using approval ballots, I don’t know what the results would have been, but my guess is that Buckner would have come in second place instead of third. In that ranked choice poll, Buckner didn’t get any vote transfers from Johnson or García supporters, even though he was a natural compromise candidate between those two progressive camps. It looks like he was subjected to a center-squeeze effect.

STAR Voting Results

When the same group of voters could specify their degree of support for candidates by rating them on a scale of 0–5, Brandon Johnson and Kam Buckner were selected as finalists. Brandon Johnson won by a margin of 9 votes in the automatic runoff.

The first round results were even more competitive amongst the top candidates than with approval voting. Not only was the margin between Johnson and Buckner less than 1%, but Chuy García came only a few percentage points behind the top two.

Paul Vallas, who came in 2nd place with plurality and ranked choice voting, and in 4th place with approval voting, came in 6th place with STAR voting, behind Sophia King and Ja’Mal Green.

Polarizing vs. Consensus-Style Candidates

One argument STAR voting advocates make is that STAR voting doesn’t have any ideological bias; ranked choice advantages polarizing candidates, and approval voting advantages moderate candidates, but STAR, they claim, doesn’t unfairly advantage anyone. (This view is backed by evidence from mathematical simulations.) Based on this argument, I expected that moderates would perform better with approval voting than with STAR.

However, at least in this particular STAR poll, the opposite was true: moderate candidates performed better with STAR than with approval. Chuy García came close to being a finalist, and Sophia King jumped from 5th to 4th place. Polarizing candidates performed worse: Vallas fell from 4th to 6th place, and Johnson was the only candidate who received lower support (in terms of percentage) with STAR than with approval voting. (All other candidates, including Vallas, had increased percentage support).

Here is a list of candidates in descending order of how polarizing they were for this group of respondents, using the standard deviation of their STAR voting scores as an indicator.³

A high standard deviation suggests a polarizing candidate who got a lot of high and low scores, but not many scores in between. A lower standard deviation suggests a less polarizing candidate about whom respondents either had a uniform opinion (e.g. Willie Wilson, who was uniformly unpopular) or a wide range of opinions, with many respondents giving medium scores (such as Chuy García).

Here we can see that Johnson and Vallas — the top two candidates in both plurality and ranked choice voting — were by far the most polarizing candidates for this group of respondents. We can also see that, with STAR voting, candidates who were less polarizing outperformed candidates who were more polarizing, relative to approval voting.

Paul Vallas and Sophia King got similar scores with approval voting, but King, the less polarizing candidate, strongly outperformed Vallas with STAR voting. Similarly, Lori Lightfoot and Roderick Sawyer got identical scores with approval voting, but Sawyer, the less polarizing candidate, outperformed her with STAR.

These results suggest that STAR voting favors moderate and consensus-style candidates even more than approval voting. However, it would take more surveys comparing STAR and approval to know if that’s truly the case, or if this was an anomaly.

A voting method that eliminates the center-squeeze effect, making things easier for moderates, is a good thing, even if you’re not a moderate. For example, if you’re on the left, the outcome of Chicago’s election will either be a progressive dream or a progressive nightmare. If Vallas wins, the hard work of progressives to send an exciting progressive candidate to the runoff will have gotten progressives the worst possible outcome; if Johnson hadn’t bothered to run, perhaps one of the moderate frontrunners — Lightfoot or García — would have been able to defeat Vallas instead.

Without plurality voting’s extreme center-squeeze effect, a progressive loss wouldn’t necessarily entail a conservative triumph, and progressives voters wouldn’t have to bite their nails about electability when choosing candidates.

Dollar-Take-All: Changing Funding Incentives

STAR voting also might have interesting implications for role of money in politics. Here is a list of approximately how much funding each campaign had in the first round of the election (according to the Chicago Tribune):

Five candidates were well funded, each with several million dollars at their disposal. Four candidates were poorly funded, with less than a million dollars each. The five well funded candidates were the top five vote-getters in the actual election. All four frontrunners — candidates that polls suggested had a chance of winning — were among them. The only well funded candidate who was not a frontrunner was Willie Wilson, whose campaign was self-funded.

Funding was remarkably well correlated with electoral support. The well funded candidates collectively got 94% of the vote. On average, they had 12.4 times as much funding as poorly funded candidates, and got 12.8 times as much support.

Survey respondents voted differently from the electorate, and the correlation was not as strong, but it was still present. In the plurality poll, well funded candidates on average got 4.5 times as much support as poorly funded candidates. Ranked choice voting reduced this ratio to around 2 times as much support.

However, with the approval and STAR voting polls, the correlation basically disappeared. Well funded and poorly funded candidates got similar support. With approval voting, well-funded candidates faired better, but very slightly.

STAR voting was the only voting method where poorly funded candidates actually did better than well funded candidates! The average candidate with less than a million dollars in funding got 43.5% support, whereas the average candidate with several million dollars got 39.0%.

It seems that the current allocation of political dollars in Chicago is well calibrated both to the Chicago electorate and to plurality voting. The candidates who can do well with plurality voting are most likely to attract funds, and the candidates with funds are most effective at getting votes.

Perhaps the non-zero-sum nature of approval and STAR-voting would help reduce the power of money in politics. I still expect money would play a powerful role, but perhaps it would be allocated more in favor of consensus-style candidates, or more equally among candidates rather than just to the top frontrunners.

This is only one survey, and I don’t know if the same pattern would hold true for others, but it suggests that STAR voting produces results that are the most different from plurality voting of any of these methods, and that presumably STAR voting would lead to the most dramatic reapportionment of funds.

How Well Did Voters Use These Methods?

One criticism that alternative voting methods often face is that they are too complicated and voters wouldn’t use them correctly, or would feel discouraged from participating. An occasional charge is that too many people would engage in bullet voting — voting only for their first choice candidate — and the dynamics of plurality voting would be perpetuated. In this survey, bullet voting was not very common.

With approval voting, only 17 respondents (5%) chose to support a single candidate. With ranked choice, only 3 respondents (1%) used just a single ranking, and likewise with STAR, only 3 gave just a single candidate a score greater than 0. The average respondent approved 3.0 candidates with approval voting, ranked 4.2 candidates with ranked choice voting, and gave 6.2 candidates a score greater than 0 with STAR voting.

In the approval voting poll, 2% of Johnson supporters bullet voted, compared to 8% of Vallas supporters. Johnson supporters approved 3.2 candidates on average, and Vallas supporters approved 2.8. This confirms that, with plurality voting, vote splitting was worse for progressives than for conservatives.

However, the difference is less extreme than I expected. Despite the fact that seven of the candidates identified as “progressive,” the ideological schism between many of them and conservatives is less than I supposed. This is further evidenced by the fact that three of the losing “progressive” candidates (Green, King, and Sawyer) subsequently endorsed Vallas in the runoff, and only two (Buckner and García) endorsed Johnson.

I only had to discard 2 responses because of incorrectly filled out ranked choice ballots (using Maine’s standards) — although the fact that it was a digital ballot allowed me to make it difficult to fill out incorrectly. I discarded 8 responses because respondents didn’t fill out the STAR voting ballot at all. It seems like STAR voting may have seemed like too much work for some people. (However, it was also the last ballot in the survey.)

Of course, this was a self-selected survey, and perhaps in an actual election, more people might bullet vote or fill out ballots incorrectly, though I would be surprised if it were a lot more. I do worry, however, about people’s abstentions from the STAR voting poll. I wonder if the STAR voting ballot seems intimidating and would discourage some people from turning out to the polls (though perhaps, if that happens, the effect would be counterbalanced by others feeling more motivated to turn out on account of the more expressive ballot and the more positive styles of campaigning).

Local Considerations for Reformers

Which method is best for Chicago? That’s for you to say just as much as me. I don’t think there is any strong argument for keeping plurality voting, but here are some arguments for and against each of these three alternatives:

Ranked Choice Voting

Ranked choice is the only alternative that has already achieved momentum in Chicago. Nearby Evanston recently voted to adopt it, alderman Matt Martin advocates for it on City Council, nonprofits such as Reform Illinois and FairVote Illinois are pushing for it, and it has gotten lots of local media coverage. Recently, former mayoral candidate Kam Buckner endorsed ranked choice. (It’s somewhat ironic that he chose that reform, since this survey suggests he would have performed better with approval or STAR voting.) Perhaps, the widespread awareness and support for ranked choice makes it most likely to actually be implemented.

However, this survey illustrates some of ranked choice voting’s disadvantages. Compared with approval and STAR, ranked choice voting benefits polarizing candidates, reduces competitive pressure between similar candidates (allowing frontrunners to dominate more easily), and produces misleading results that don’t paint an accurate picture of candidates’ popularity. It is less of a deviation from the status quo and current campaign funding incentive structure.

Furthermore, ranked choice voting very occasionally can lead to weird outcomes where vote splitting results in the elimination of the most popular candidate. For example, in a 2022 special election for Alaska’s U.S. House seat, a moderate candidate could have defeated either of the more polarizing top candidates, but was eliminated with ranked choice voting.

Approval Voting

Approval voting doesn’t have the same momentum as ranked choice, but it does have precedent in the Midwest. It is used in St. Louis, Missouri, and Fargo, North Dakota. (There may soon be a statewide approval voting initiative in Missouri, led by the cross-partisan advocacy group Show Me Integrity. North Dakota, however, will soon ban approval voting along with ranked choice in reaction to the difficulty far right Republicans have faced with ranked choice voting in Alaska).

Despite having less momentum, approval voting might have even more of a practical advantage because of its sheer simplicity. It might be easier to get alderpersons to agree to support approval voting because it’s easier to implement logistically, requiring no ballot redesign. However, there might be some pushback from ranked choice supporters due to the fact that approval voting doesn’t provide a way for voters who support multiple candidates to distinguish which one is their favorite.

Also, there’s the question of runoffs. Approval voting doesn’t require a runoff, but it arguably works even better when you add one. If Chicago, like St. Louis, decides to adopt the version of approval voting with a top-two runoff, then the runoff elections would entail administrative expenses for the city that otherwise would be eliminated by ranked choice or STAR.

STAR Voting

STAR voting has neither the advantage of momentum nor the advantage of simplicity. However, it does have a key advantage of its own: it addresses many of the criticisms of both ranked choice and approval voting. Unlike ranked choice, STAR voting doesn’t advantage polarizing candidates, produce misleading results, or sometimes eliminate the most popular candidate. Unlike approval voting, STAR allows voters to specify their degree of support for candidates and doesn’t provide any reason to add a separate runoff. In short, STAR is the natural “compromise candidate” between ranked choice and approval.

However, STAR voting was invented in 2014 and has never been used in public elections, and I find it hard to imagine Chicago being the first adopter of a method many would criticize as “untested.” If Oregon’s statewide initiative succeeds in 2024, then maybe things would be different. Also, the ugh, it’s so complicated reaction that a lot of people have with ranked choice would be an issue with STAR as well. Chicago’s election turnout is already low, and it’s an open question what effect STAR’s more complex ballot would have on turnout.

None of these methods is perfect, but their imperfections are vastly outnumbered by plurality voting’s flaws. Any of these alternatives would be better than our current method. If we want voting reform, the first step is to spread awareness that these alternatives exist and encourage people to consider all of them with an open mind.

Better Democracy Starts With Better Discourse

I am writing this the evening before the April 4 runoff election. Polls suggest it will be close. Soon, a winner will be determined, and everyone will draw the wrong conclusions. If Johnson wins, it will be attributed to the unprecedented strength of the progressive movement and the power of the Chicago Teachers Union, not to the fact that Johnson was lucky not to have to face a moderate in the runoff. If Vallas wins, the national media will draw sweeping conclusions about how pro-policing tough-on-crime Democrats are the zeitgeist, rather than about how Vallas was lucky not to have lots of conservative competitors splitting the vote, like he did in 2019 when he only got 5% support.

Rarely will reality be acknowledged: that in Chicago, median public opinion lies somewhere in the middle, and whichever polarizing candidate wins, his victory will be a somewhat arbitrary outcome of the system itself — a system of uneven competition, divide-and-conquer politics, and strategic voting, which tends to disadvantage the very candidates most capable of building consensus across the electorate.

In local media, whenever systemic factors such as vote splitting are brought up, usually they are discussed as if an immutable part of how democracy works, rather than an artifact of an antiquated system our city choses to maintain. In the rare cases that reform is acknowledged as a possibility, ranked choice voting is typically presented as The Solution, rather than simply as one of reforms we can choose from.

In the spirit of the nuanced politics reformers wish to bring about, media coverage should strive to present the issue of voting reform as a open-ended debate among a variety of competing ideas, not as a stark, binary choice between a single reform and the status quo. A better politics is possible, but if we want Chicago to implement the reform that’s the best fit for our city, we need discourse that pushes the discussion forward, with a political analysis that does three essential things:

1. Acknowledges the centrality of the structural factors that shape our politics.
2. Acknowledges that those structures can be changed for the better.
3. Acknowledges that there are a variety of possible reforms we can choose from.

In politics, the discussion of who wins often obscures the discussion of how the system works. If instead we adopt a structural lens and shift our focus to the system itself, there’s hope that democracy can be deepened and remain resilient through our century’s turbulent changes.

Results Summary Table

Footnotes

1. Ranked choice sometimes allows people to rank as many as 10 candidates, as in the Bay Area, or as few as 3, as in Minneapolis. I chose 5 because I didn’t want to overwhelm respondents and because it’s the same number of rankings that New York City permits.
2. In the ranked choice poll, Vallas supporters ironically would have gotten closer to achieving a more favorable outcome if fewer of them had participated in the first place, causing Vallas to be eliminated. If it were the case that Johnson’s lead was less strong, say if 25 fewer Johnson supporters had participated in the poll (with all else equal), then Buckner would have been eliminated despite being able to defeat either Johnson or Vallas in a head to head race. In that particular hypothetical, savvy Vallas supporters could have strategically ranked Buckner higher than Vallas — or simply abstained — and potentially swung the outcome from Johnson to Buckner; it would only have taken 7 abstentions or 4 strategic votes for them to succeed.
3. Standard deviation is the average difference between individual scores and the mean score. The higher the standard deviation, the more polarizing the candidate is for this group of respondents. For example, Paul Vallas’s scores have a standard deviation of 2.08, meaning that the average respondent gave Vallas a score that was over 2 points away from the mean score of 1.80. A lot of people gave him a score of 4 or 5, and a lot of people gave him a 0, but there weren’t a lot of scores in between; he was polarizing.