Thousands of Mainers have already cast the first ranked-choice ballots ever to be used in a U.S. presidential election, but most of what we know about the system is based on the unsubstantiated claims of proponents and opponents. That’s the backdrop against which Jesse Clark, a political science doctoral candidate at MIT, conducted the most comprehensive study of ranked-choice voting ever attempted. He published a draft in September and concluded without reservations that the new system, lauded as a revolutionary and overdue improvement, will be harder, less satisfying, and likely to produce the same results.

The study, “Rank Deficiency? Analyzing the Costs and Benefits of Single-Winner Ranked-Choice Voting,” included a simulated election with 1,500 participants, observational data collected from the 2018 midterm election, including a text analysis of Facebook ads, and a survey of 500 Maine voters about their experience of the 2018 midterms, which was the first time Maine, or any state, picked federal legislators by RCV.

“Previous reports and studies on this, they can’t show with any sort of causality that ranked-choice voting is doing one thing or another, or at least they can’t very strongly,” Clark told me in a phone conversation last week. “You can only really do that with experiments. And that’s what I [did].”

Clark framed the study around the arguments of proponents and opponents of RCV in Maine. His hypotheses more or less aligned with the criticism put forward by Republicans: voting by ranked-choice would be more confusing, it would take longer, more ballots would be spoiled through errors, voters would be less satisfied with the outcome of the election and more likely to think the system had benefited the opposing party, and many of the effects above would get worse the more candidates appeared on the ballot.

In his lone pro-ranked-choice hypothesis, Clark predicted an increase in “sincere voting,” and therefore more votes for non-major-party candidates. He was right, but the bump was only about 5%. Clark noted that arguments for RCV often gloss over the possibility that a voter could sincerely prefer a major-party candidate.

In the end, nearly all of his projections were right — the one exception, which he was unable to explain: voters botched more traditional ballots than their more complicated RCV counterparts.

Clark puts “ranked-choice” voting in quotes. The term itself, he said, was created to sell voters on what had previously been called instant run-off elections. “Of course, you want to rank your candidates, right?,” he said. “Nobody wants an instant run-off, because what is an instant runoff?”

Here’s what it is: Voters assign each candidate — or those they would accept — a rank. The first-place choices are tallied and if no candidate receives more than 50%, the candidate with the fewest votes is eliminated and the second-place votes from those ballots are redistributed to the remaining candidates. The reallocation process gets somewhat trickier in later rounds, but at some point one candidate accumulates a majority of the ballots that are left and is declared the winner.

Opponents are quick to point out that some ballots are invariably “exhausted” because the voter didn’t rank all the candidates. As a result, the winner of a “majority” after several rounds of ballot reallocation might have fewer than 50% of the ballots cast on Election Day.

A 2019 study by the conservative Maine Heritage Policy Center of 96 ranked-choice voting races from across the country where additional rounds of tabulation were necessary to declare a winner found that this happened 61% of the time.

Clark found that not many of the arguments advanced by RCV supporters hold up to scrutiny. And those that do are often outweighed by other negative factors.

For example, ranked-choice ballots took the study participants a lot longer to complete than traditional ballots. In large part, Clark said this comes down to the fact that it’s easier to pick a favorite from among a group of choices than to order those choices from best to worst, especially when it comes to the middle: “Do people really know how to vote for their third through fourth places when there’s eight candidates that you’re going to rank?” he said. “It’s probably mostly random.”

Supporters of ranked-choice say it’s been used successfully elsewhere, but Clark found that many of the examples they cite, particularly in the U.S., tend to be from cities that lean heavily Democratic. They also tend to conflate the style of ranked-choice voting used in Maine, also known as instant run-off, with a very different type of ranked election called the single transferable vote.

“STV, the type you see in Cambridge, Massachusetts, that gets talked about a lot is really proportional voting that layers ranking on top of it,” Clark said. Multiple candidates are elected from the same ballot, and the results are generally thought to give fair representation. “Instant run-off voting like we see in Maine is just the same first-past-the-post system [used in the rest of the U.S.] with ranked ballots on top of that.”

The argument that RCV leads to more civil campaigns, Clark said, is based on the untested belief that candidates in a ranked-choice race have an incentive to be the second choice of voters who would pick another candidate in a traditional election. Clark said that idea overlooks the possibility that the election will be decided in the first round of counting.

In 2018, while Bruce Poliquin was challenging the constitutionality of ranked-choice voting in a bid to keep his House seat, Maine’s other member of Congress, Chellie Pingree, took more than 58% in the first round, ending the race before ranked-choice was put into play.

In a race with a strong frontrunner, Clark said, trailing candidates actually have an incentive to chop down the leader and force additional rounds of ballot reallocation.

Voters might be just as calculating. RCV proponents say the system eliminates strategic voting, but Clark said there’s always a way to game the system. “That’s one of the base-level knowledges we have of any sort of electoral system; there is a way to game it. [With ranked-choice] we just don’t necessarily know what that is yet.”

Ranked-choice and traditional plurality voting are just two of dozens of possible selection systems that the social scientist Kenneth Arrow considered before coming to the conclusion in 1951 that none of them are immune to outcomes that appear to defy common sense. For example, a candidate who would beat every other candidate in individual head-to-head elections, known as the “Condorcet winner,” might lose an election when facing all of them at once. Similarly, a group of diners forced to choose a single entree might pick salmon over chicken and chicken over beef, but then pick beef over salmon without any of them changing their minds. (Hint: the majority is composed of a different combination of people in each case.)

In the summer of 2010, a group of mathematicians convened in France to elect “the best voting procedure,” and out of some 20 contenders they gave the highest marks to “approval voting,” in which voters mark any of the options they approve of and the one with the most votes wins. Perhaps knowing this at the outset, the group used approval voting to pick the best voting system. Ranked-choice, which they termed “the alternative vote,” came in second place. Plurality voting, the system Americans have used in every presidential election, was one of three systems that received zero votes.

But in politics, math only goes so far. Experts can demonstrate that some voting systems are inherently fairer than others, but they may not produce better results in real elections.

Clark said many of the negative effects of ranked-choice voting will improve as we get used to the system. “The best comparison I think of is automatic vote-by-mail out West where you saw voter confidence go down in that first election, and it sprang right back up in the second election. And I think that’s what’s probably going to happen [with ranked-choice in Maine]. Maybe not after a single election, because I don’t think Republicans are going to fare particularly well this year, and they’re the ones that need to be convinced. But you get a Republican year and I think some of those people start to turn around.”

“Long term, it’s just going to be kind of a harder system. People will work it out, but they’ll never totally climb out of the impacts of ranked-choice voting. And system wide, no, it’s not really going to change who gets elected or how they get elected.”