America is having a reform moment. According to Pew Research, hundreds of jurisdictions are adopting or considering “alternative” electoral systems. Several began doing so only in the past few years. There are many reasons for this, but two big ideas stick out. One is the belief that we need more parties and that a different electoral system would create them. Another is the recurring push for nonpartisan elections. By “nonpartisan election,” I mean a system in which parties do not control candidates’ use of their labels. An extreme example would remove party labels from ballots entirely.
For those who want more parties, proportional representation (PR) is the holy grail. This would involve two changes if used for congressional elections. First, elections would be held in districts with multiple seats each. (Several states and cities already have multi-seat districts.) Then those seats would be divided among parties in proportion to their vote shares.
Few are advocating PR as just described. Rather, they are looking for an option that satisfies proponents of nonpartisan elections. Enter the single transferable vote (STV), recently rebranded as “proportional ranked-choice voting” (P-RCV), and which became law in Portland, Oregon, last year. Most know ranked-choice voting (RCV) from the “instant runoff” versions passed in Alaska, Maine, and elsewhere. These ask voters to rank candidates in pursuit of majority winners. STV is a bit more complicated, as we’ll see below. For now, what matters is that STV does not give seats to parties as parties.
Many may not know that STV was widespread during the Progressive and New Deal periods. Reformers had formed a belief that “normal” PR could not win. So, they hooked up with the movement for nonpartisan elections, won STV in 22 cities, and then saw it repealed in all but one. That’s because it eventually frustrated powerbrokers in both major parties. STV survives in three countries (five if we include local elections), but their parties have adapted to it—often by introducing elements of “normal” PR.
History suggests that party involvement in elections cannot be escaped. Reforms either work with a party system, or that system spits them out. So, if we want PR for the long haul, we should take a second look at what I’ve called the “normal” version—even if that miffs supporters of nonpartisan elections.
More broadly, the current conversation is too focused on the number of parties. What matters is that majority coalitions can hold themselves together—be they single-party or alliances of parties. PR might help with that and has been adopted for this reason. That PR might strengthen parties is conceptually separate from how many there are. It also shifts the focus to the type of PR.
Proportional Representation 101
What I’ve called “normal” PR is known as a list system. “List PR” here means two things. First, political parties put forward groups of candidates (their lists). Second, these lists earn legislative seats in proportion to their vote shares. So, if a list is entitled to three seats, the first three people on it win office. Every PR system in the world builds on lists. (Some reformers present STV and other systems without lists as forms of PR, but experts would say that they are not truly proportional.)
Who makes the lists? Rules vary. Party leaders might draw them up in a “smoke-filled room.” Local party committees might send delegates to a nominating convention. Multiple parties might negotiate a “joint” list. A single party might even field multiple lists. The point is that these lists exist in advance of a general election.
And what does the voter do? It depends on whether the rules call for “open” or “closed” lists. With open-list PR (OLPR), the voter sees and may choose among candidates. A vote then counts for the chosen candidate and their list as a whole. Variations on this system exist in 40 democracies. Other countries use “closed” lists, which let voters choose among lists, not candidates.
Then how do we determine winners? The fine print matters, but the basic idea is to divide total votes by the number of seats in the district. So, in an election with five seats and 100 votes, the “price” per seat is 20 votes. A list with 60 votes would get three seats. If the list were open, these would go to the most-voted-for candidates. If it were closed, they would go to candidates in party-determined order.
By contrast, in non-list systems, votes do not “pool” at the level of a preelection list. What does this mean? Think back to OLPR above. A party could run as many candidates as it wanted, not fearing that those candidates might “spoil” the overall list. That’s because votes for each candidate also benefit the list as a whole. The quintessential non-list system is “single non-transferable vote” (SNTV). This limits each voter to one candidate vote, which does not also benefit a list. The winners in our running example are just the top five vote-getters. SNTV was used in Japan for years and, at times, has been a remedy in U.S. voting rights cases. Non-list systems (like SNTV) make it possible for “too many” candidates to try to represent a party, since they can split the vote to such an extent that none of the candidates win seats. More precisely, because there is no “pooling,” a party can get fewer seats than it might have gotten in a list system.
STV tries to rescue “pooling” while avoiding controversial lists—controversial because parties typically make them. It does this by borrowing from list systems the idea of a “price” per seat. Then it tries to prevent “spoilers” by asking voters to rank candidates. This works by moving votes among candidates over several “rounds” of counting. Hence it is called the single transferable vote. The basic idea is to transfer any vote in excess of the “price,” then any vote for a candidate who has no chance of winning. The point is that a vote in STV does not benefit the party as a whole—unless voters rank sufficient choices and keep those rankings within a party.
STV Comes to America
STV was used in 22 American cities between 1915 and 1962. (That number rises to 26 if we include two single-tax colonies and two federally planned communities in Tennessee.) Headliners included Cleveland, Cincinnati, Sacramento, and New York. All but one repealed it—several after just a few elections. The outline of that story is as follows:
Reformers settled on STV because some among them disliked the idea of parties.
Reformers then started forming parties to direct voters’ rankings.
STV got repealed anyway because it made it hard to hold those parties together. Or, in New York City, the system made it hard to bind a coalition of parties. (New York was the only city with visibly multiparty politics.)
Much of that history is detailed in my book More Parties or No Parties, published in 2022. Other excellent titles come from political scientist Doug Amy, the late political scientist Kathleen Barber, and historian Daniel Prosterman. I went beyond these in three ways. First, I asked why reformers were promoting STV instead of party-list. Then I sought a data-driven way to explain repeal activity. Finally, I asked why Cambridge, Massachusetts, never repealed STV.
Intellectually, STV was part of the Progressive attack on party politics. Many early PR supporters disliked it for this reason. “The Hare system,” wrote a leading pro-PR historian in 1896, referring to STV, “is advocated by those who, in a too doctrinaire fashion, wish to abolish political parties.” But public opinion was turning against parties for many reasons. One was a view that cities had fallen under the control of party “machines.” Another was frustration with party control of nominations. The country was diversifying, women were gaining voting rights, and nominating primaries were spreading as a way to hold the major parties together. An early form of “instant runoff” was gaining traction at the local level, alongside nonpartisan elections to small city councils. (The pro-reform literature from this period is eerily similar to what we read today.) PR advocates were piggybacking on a reform agenda that targeted parties.
But like many reformers today, they had also formed a belief that party-list could not win. The long-running debate over nonpartisan elections came to a head in March 1913, with a party-list initiative in Los Angeles. OLPR was one of eight reformer-proposed amendments to the city charter. And it was the only one to fail. The Los Angeles Times had equated party-list with socialism and told voters that “it provides for a government by parties.” Reform leaders publicized the episode to build support for STV.
Two years later, in 1915, a wave of STV adoptions began in Ohio. It also touched California, Colorado, Connecticut, and Michigan. Repeal activity, though, was not far behind. A late-1920s report based on stakeholder interviews found that party leaders did not like STV, that voters “resented” its complexity, and that some thought it was not electing “first-class candidates.” One Cleveland mayor told the interviewer: “The councilmen are elected by a so-called proportional representation ballot. Just what proportion and what the proportion is, with regard to any particular councilman, I have never been able to discover.”
Cleveland repealed STV in 1931, bringing total repeals to five, and leaving STV in just three places. (A second adoption wave began in 1935.) One key message from the lessons-learned report might strike some as ironic: STV seemed to provoke less disaffection where its proponents had formed their own parties.
These parties would set policy direction, recruit candidates, and try to get voters to rank them. They went by names like Civic Association, Charter Committee, and Citizens’ Non-partisan Committee. They typically included a city’s smaller major party and the pro-reform faction of the other. So, in a city that had been heavily Democratic, the Republicans might team up with a faction of the Democrats. No party labels would appear on ballots. Rather, the coalition would publicize its slate and run get-out-the-vote operations—sometimes with resources from the major party it subsumed.
The reformer parties worked for a while. Without one, STV lasted 9.8 years on average. With a party, the average was 13.7. Several cities went much longer: three decades each in Cincinnati and Hamilton, Ohio. Only one without such a group kept STV that long: Boulder, Colorado.
The Problem of Vote Leakage
Why did all but one city repeal STV? Something must have undone the majorities that liked it. Conventional wisdom points to the election of Black Americans, Communist Party members, and other “unpopular minorities.” But many cities repealed STV long after such figures were present. Some did so without them present at all.
I collected data on a problem called “vote leakage.” “Leakage” refers to ballots cast for one party’s candidates, but that end up (via transfer rounds) electing politicians from another. Some leakage is normal. For example, votes might “leak” from Working Families to the Democratic Party, or from the Libertarian Party to Republicans. Such leakage can reflect explicit coalition deals or simply that these parties are ideologically similar.
But leakage is a problem when it upsets the main divide in a party system—for example, when it goes from Democrats to Republicans (or vice versa), “good government” to “old machine” (or vice versa), or from the Forward Party to both of the majors. One side might blame leakage for reversing majority control of the legislature. Or leakage might elect legislators who owe their seats to voters on both sides.
I realize that the potential for votes to “leak” across the aisle—and/or elect “centrists” on leaked votes—is exactly what makes STV so popular. The problem seems to be that it also gets the system repealed. I found such patterns in three very different cities: Cincinnati, New York, and Worcester, Massachusetts. Case history also points to leakage problems elsewhere. One complaint from repeal activists was that STV had turned elections into “lotteries.” (STV is unpredictable, and recent experimental work shows that voters are dissatisfied with unexpected outcomes.) Another was that it had destroyed “party responsibility,” which was an old term for party discipline in government.
This is why both major parties agreed that STV was problematic. It came to hurt both of them at once.
What about Cambridge, which never repealed STV? Leakage also cropped up there, and it helps us understand the timing of repeal efforts. The local solution was not to reverse the reform. Rather, the local reformer party began running fewer candidates—just enough to claim majority control of city council. This is what is known as STV “vote management”—limiting the number of candidates, then making sure that voters distribute first-choice rankings roughly evenly among them.
The Case for Party-List
The best book on STV in other countries calls it “an embedded institution.” What this means is that, in the three countries that use the system for national elections, their party systems have adapted STV. Ireland is the key example: multiparty politics with tight control of nominations. Australia lets voters vote “above the line,” which means letting parties rank the candidates for you. (Some might say this sounds like party-list.) And Malta gives a seat majority to the party with the most first-choice votes. All these devices contain leakage by giving parties greater power.
Could STV in the United States also become “embedded”? Maybe it could, and maybe not. I came away from my research thinking that STV is not a good fit for the U.S. party system. Our parties are too big and undisciplined for voters and candidates to make it stick. There is little will to create the sorts of agencies that like running STV elections. Much of the education work—for voters, politicians, and others—falls to the groups that want to see STV enacted.
STV’s persistence in Cambridge shows that there are ways to make it work. But I think it’s easier just to use list PR instead. Studies show that, when STV isn’t facing repeal, it works much like party-list. That goes for historic elections in Cincinnati and New York City (1925-37), as well as recent analyses of Irish and Estonian data. If parties are going to be factors anyway, why not build the electoral system around them?
Then there are the equity concerns that have dogged RCV for years. Voters who use more rankings have more effect on outcomes; voters with less formal education are less likely to use rankings. So are voters whose preferred politicians tell them not to rank. It is hard not to conclude from the growing body of research that reformers should base proposals on a single vote instead. (Such proposals also might improve the existing multiseat systems in ten states, from Arizona to West Virginia.)
In closing, I want to be clear about my own position. PR should not be promoted for its abstract appeal. Changing rules is about reallocating power. That should make us cautious. There is more to learn about PR and the sorts of coalitions it might produce (or not). For now, advocates should stop focusing on nonpartisan elections and instead revisit party lists.