Shaky political “science” in Minnesota

This anti-RCV “study” by a well-known political scientist is so rife with mistakes, misquotes and cherrypicked info that it is an embarrassment to the profession

Minnesota state capitol building

[Dear readers -- DemocracySOS has launched our “Shaky Political Science” series in which we identify and analyze sloppy research focused on ranked choice voting. In our 38 page report, “Shaky political ‘science’ misses mark on ranked choice voting,” released in early December, we examined over 40 studies focused on ranked choice voting and found that many suffered from puzzling research methodologies, poorly constructed surveys and simulated elections, cherry picking of data, and faulty analyses that often were contradicted by results from real world elections (a link to our full paper is here). Previously in December, DSOS published an initial article summarizing the results of our report (that link is here). Below is the second article in our “Shaky Political Science” series.]

Professor Larry Jacobs is a well-established political scientist at the University of Minnesota. One of us (Steven Hill) has cited Jacobs’s research in his own work, specifically Jacobs’s co-authored book Politicians Don’t Pander: Political Manipulation and the Loss of Democratic Responsiveness, including in this previous DemocracySOS article. But Jacobs’s “study” of ranked choice voting from 2023 is one of the most error-prone of all the research we reviewed in our report. And nobody caught it — until now.

Why aren’t other political scientists calling out the flaws in this study? Did no other academics actually read it? Is there really that little quality control within political science? The study was published by the venerable Hubert H. Humphrey School of Public Affairs, which dents the credibility of HHH as an institution. Studies like this (and dozens of others analyzed in our report) give political science a bad name. And Jacobs has been quoted, and his paper cited, by a number of national media organizations, including National Public Radio. It was cited in a number of media articles by the opposition during the 2024 RCV ballot measures in Nevada, Colorado and Oregon.

Here is a link to Jacobs’s study, co-authored with Penny Thomas:

Where’s the evidence supporting Ranked Choice Voting Claims? by Lawrence Jacobs and Penny Thomas. Hubert H. Humphrey School of Public Affairs, 2023.

This “study” did little more than cherry-pick a number of other academic studies – including several flawed studies criticized in our report, and other studies that actually didn’t conclude what Jacobs claimed -- to attack ranked choice voting (RCV). Jacobs’s study said RCV “fails to support four of the advocates’ promises for improvements over today’s system.” First, it cited “#1 RCV claim: Reduce today’s polarization of the political parties” and then tried to rebut it. For the rebuttal, it relied on two studies that our previous paper analyzed and found deficient. One of these studies, Study 19 in our paper (Cerrone et al., 2021), was not actually about polarization at all, it was about voter satisfaction. It also did not use data from real-world elections, instead it used an opinion survey with little explanation provided to respondents about the different election methods being studied that would help respondents to understand RCV, how it works, or how to use ranked ballots effectively.

The second study cited by Jacobs et al, Study 2 in our paper (Fischer et al., 2021), used an unconventional methodology based on game theory, not on real-world elections, which tried to provoke measurable behaviors from the study’s human participants who were divided up in teams to judge their interparty animosity. This study was too divorced from the reality of real elections to take seriously, as discussed under Study 2 in our paper. Jacobs is located in Minneapolis, which has one of the longest track records of using RCV in the United States, and next-door St. Paul now uses it too, as does St. Louis Park, Bloomington and Minnetonka. There have been dozens of RCV elections in Minnesota to actually study for real-world impacts on polarization. Instead, Jacobs’s conclusions relied in part on two studies of dubious credibility with no data from actual elections.

Next, Jacobs’s study examined the impact of RCV on the diversity of elected officials, increased voter turnout, engagement of voters of color and decreased negative campaigning. Regarding an increase in the diversity of elected officials, the Jacobs et al study concluded “there is little support for this claim.” However, the three studies cited in fact took the opposite viewpoint, presenting compelling evidence for an increase in the diversity of elected officials associated with RCV elections.

Jacobs et al cited the study “Election Reform and Women’s Representation: Ranked Choice Voting in the U.S.” (Terrell et al., 2021), which concluded “As of July 2020, women’s average representation is eleven points higher in the California city councils elected using ranked choice voting compared to the average of comparison California cities.” A second study cited, “The alternative vote: Do changes in single-member voting systems affect descriptive representation of women and minorities?” (John et al., 2018), concluded that RCV led to not only a 9 point increase in the percentage of minority candidates but also a 16 percent increase in the predicted probability that a female candidate will get elected and a 14 percent increase in the predicted probability that a female minority candidate will get elected. Another study done by the same three researchers, The Impact of Ranked Choice Voting on Representation (John et al., 2016), found that “RCV increases descriptive representation for women, people of color, and women of color.”

Bizarrely, the conclusions of these three studies are completely misrepresented by Jacobs and his co-author.

Jacobs also cited Study 34 in our paper (Kimball and Anthony, 2016), “Voter Participation with Ranked Choice Voting in the United States,” which actually didn’t analyze diversity at all. It analyzed turnout. Kimball and Anthony concluded that “RCV helps reduce the substantial drop in voter participation that commonly occurs between primary and runoff elections” by an astounding 32.7 points, primarily due to finishing elections in a single November contest which avoids huge voter drop-offs between the first (primary) and second (general) elections. And in a case study of Minneapolis elections – Jacobs’ own hometown – Kimball and Anthony concluded “we find similar levels of socioeconomic and racial disparities in voter participation in plurality and RCV elections.”

So perplexingly, several of the studies cited by Jacobs et al directly refuted their own conclusions. To support their conclusions on voter turnout and engagement of voters of color, their study cited the thoroughly discredited Study 29 that was analyzed in our paper (McDaniel, 2016), yet bizarrely ignored the Kimball and Anthony study on turnout that their own study had just finished citing – erroneously -- regarding diversity of elected officials. Regarding RCV impacts on negative campaigning, the Jacobs study emphatically declared “most research found little to no impact or even increased negativity.” To support this claim, it cites Study 8 analyzed in our paper (Kropf, 2021), writing “Twitter traffic was more negative in RCV cities.” But that is the exact opposite of what Kropf’s paper concluded. Kropf’s Twitter analysis found “evidence of bargaining and accommodation” in RCV cities, including in Jacobs’ hometown Minneapolis, where “many of the candidates referenced each other,” with some candidates actually retweeting posts from their supporters about which other candidates they planned to rank. Jacobs also ignored the other five studies, grouped in our paper with Kropf’s study, that showed consistent evidence of less negative campaigning in real world elections.

Jacobs has been an esteemed political scientist for decades, so this poor quality study steeped in misinformation and cherry-picking is puzzling. A number of the studies he cited actually refute his conclusions. Despite its clear flaws and poor reliability, Jacobs’s study has been quoted and linked to in various media articles. Unfortunately media outlets these days do not have substantial research or deep reporting budgets, and have tight deadlines too, so journalists often cite the conclusions of academic studies without reviewing the actual studies themselves for accuracy or even common sense.

History of cherry-picking and anti-RCV information

It is additionally illuminating to note that this was not the first time Jacobs chose to portray RCV in such a negative light, with little evidence to support his claims. In 2013, following the first RCV elections in Minneapolis in 2009, Jacobs cited from an exit survey following that election to make unsubstantiated claims in an op-ed in the local Minneapolis Star Tribune titled, “New Minneapolis rules could diminish voting equality.” Jacobs claimed that RCV “may widen disparities” because “Minneapolis voters indicated that they understood RCV better than those who did not vote.” Which seems like an unremarkable conclusion, since non-voters in general are more confused by elections than actual voters, which is one reason they don’t vote, even in non-RCV elections. Most actual voters take at least some time to study the candidates, while non-voters generally do not.

Stretching a thin reed of logic, Jacobs then went on to suggest that “RCV kept people from voting who didn’t feel confident about their understanding of the complicated system,” even though he sideways acknowledged that low turnout in 2009’s mayoral election reflected the fact that the incumbent held a large lead in opinion polls and easily won with 74% of the vote.

Moreover, in his survey, he found that non-voters had lower levels of interest, confidence and belief in the fairness of the election than voters, which again seems unremarkable as that is one reason nonvoters don’t vote, including in non-RCV elections. But Jacobs then tried to suggest that, in this first Minneapolis election using RCV in 2009, the reason nonvoters felt this way was because they were confused by RCV. But it could just as likely have been that the same nonvoters also found non-RCV elections unfair and unrepresentative, and that’s why they didn’t vote. After all, patterns of non-voting are typically developed over many years. Jacobs presented no data from his survey that teased out the reasons why nonvoters don’t vote, whether in RCV or non-RCV elections, and without that his viewpoint was mere speculation.

Jacobs also did not accurately describe how RCV works, and revealed that in fact he apparently does not really understand how it works, when he wrote that RCV may give some voters “three votes” and “is eroding…‘one person one vote’.” In actuality, RCV is a runoff system that only gives each voter a single vote, which is transferable to lower choices in each round of the “instant runoff” if a higher choice is eliminated. Each Minneapolis voter had three rankings, not votes, to indicate which single candidate a voter supported in each runoff round.

Also, Jacobs apparently was unaware that four years earlier, right there in his home state of Minnesota, the state Supreme Court had upheld the constitutionality of RCV, including as it pertains to the “one person, one vote” standard, in Minnesota Voters Alliance v. City of Minneapolis (2009). Lack of awareness of a major Supreme Court ruling is quite an enormous oversight for someone weighing in as a political science “expert.” Without presenting any factual data, Jacobs worried that RCV would hurt elected diversity, yet failed to mention that seven out of 13 (53.8%) of the RCV-elected city councilors in that 2009 election were women.

Ten years after Jacobs’s 2013 op-ed, as he was publishing the study that our paper analyzed about the impacts of RCV, with his warning that it may hurt diversity, Minneapolis voters in the 2023 local elections used RCV to elect candidates of color to nine out of 13 (69%) city council seats, including Blacks, Latino, Muslim and Somali elected officials, and women to eight out of 13 (62%) seats. In the 2021 elections, women candidates won a majority of seats, candidates of color won 62%, and Minneapolis enjoyed its highest voter turnout in local elections in 45 years. An exit poll found that 88% of voters found RCV easy to use, and 76% liked and wanted to continue using RCV. In the nearby Twin City of St. Paul, following the introduction of RCV, voters elected their first Black mayor in 2017, and in the 2023 elections, St. Paul elected an all-women city council, with six of the seven councilors being women of color. That is quite a sea level change from before RCV implementation, when the city council included only one woman and one person of color. And Jacobs never noticed.

So Larry Jacobs’s warnings about RCV not only were unsupported by the very studies he cited, but also were not supported by results from actual real world elections. His non-peer-reviewed “study” is full of errors and misinformation, and it is a discredit to political science itself that no political scientists caught that and called him out, even as this bogus study was cited by a number of media outlets.

Over the next few weeks, DemocracySOS will present more examples of poor and flawed research about ranked choice voting, afflicted by defective methodology and distorted conclusions, that have been published by various outlets.

Steven Hill @StevenHill1776 @StevenHill1776 bsky.social