Which proportional representation method is best for America?

Party List vs Proportional Ranked Choice Voting vs Mixed Member Proportional -- are they all the same? Or is one method better than the others?

[Warning to readers:  we are going to get a bit wonky in this one. I am going to take you along on a ride as I examine a few different varieties of proportional representation (PR) electoral systems. Recently a discussion has broken out among a group of policy specialists in the US over which PR method would be the best fit for the United States. FairVote and Rob Richie have published their plans for use of Proportional Ranked Choice Voting in all 50 states, as embodied in its Fair Representation Act for Congress; and political scientists like Lee Drutman at FixOurHouse, Jack Santucci, Aziz Huq, Matthew Shugart and Michael Latner have put forward interesting thoughts on what a US-style Party List PR system might look like, and have added immeasurably to our understanding of these matters.

So this article is meant to offer my own thoughts about the most salient pros and cons of different PR methods (some of this material is drawn from my three books related to this topic: Fixing Elections: The Failure of America's Winner Take All Politics, 10 Steps to Repair American Democracy and Whose Vote Counts (with Rob Richie)). This article is not meant to be a comprehensive discussion; entire books and long research papers have been written on this topic, and that is beyond the scope of this article. Rather, it is more of an outline of some of the key details that will need to be addressed as these discussions continue to unfold, and as we take into account the uniquely American culture, traditions and history].

Probing the “operating system” of representative democracy

Electoral systems are to a representative democracy what the operating system is to a computer. Think Windows 10, Apple iOS, Linux. They work invisibly in the background, and we really don’t notice them until they break down or stop working smoothly. Depending on which “operating system” of a democracy you choose, the same votes cast through different electoral systems can result in completely different candidates or parties winning elections.

Broadly speaking, proportional representation (PR), which is used by most of the established democracies in the world, refers to a category of electoral systems in which political parties (or in non-partisan elections, groupings of like-minded voters, i.e. liberals, conservatives, progressives) win representation in proportion to their share of the votes. Elections take place in multi-seat “super districts”, where in a district with 10 seats, if a party or grouping wins 20 percent of the popular vote it wins 20 percent – two – of the seats; if it wins 60 percent it is awarded 60 percent – six – of the seats.

But in the US, we mostly use different variants of a “winner take all” electoral system (also referred to as “first past the post”) where the highest vote-getter wins, so that same 20% of the vote usually wins no representation, and 60 percent wins 100% of the representation. That’s because legislative offices are elected one district seat at a time and result in a single winner instead of multiple winners (there are exceptions, but that is the basic rule, especially in major elections). So proportional representation tends to lead to multi-party legislatures in which a number of parties are able to compete and win legislative seats, including major and minor parties. While winner take all elections tend to result in a two-party system.

But even that is a bit of a misnomer because most winner take all districts, indeed most states, are little more than noncompetitive one-party fiefdoms due to partisan regional demographics, i.e Democrats dominating in urban areas, Republicans in rural areas and many suburbs (important to note: the lack of competition in most districts and states is not due to gerrymandering the district lines but due to these partisan regional demographics, i.e. where people live). The real “choice” for voters is to ratify the candidate of the party that dominates their district and state.

PR systems generally result in more choice for voters, higher voter turnout, broader representation and – with more points of view at the legislative table – it leads to more “policy congruence,” which is the political science way of saying that the policies passed by the legislatures are more in keeping with what the majority of the society generally desires. Most of the established democracies in the world use some type of proportional voting system, with just a handful of democracies – primarily the UK and its former colonies, such as the US, India and Canada – that continue to use the antiquated winner-take-all system.

Different types of PR – which is best for America?

But within the category of PR electoral systems, there are different methods. For this discussion, three subcategories of proportional voting methods are most relevant: 

1) a Party List PR system, in which voters vote for a political party and its pre-selected list of candidates, rather than for individual candidates;

2) a variant of Party List known as Mixed Member Proportional (MMP), in which voters are allowed two votes – the first for a political party and its candidate list, the second for a US-style winner-take-all district representative; and

3) a ranked ballot method, known as Proportional Ranked Choice Voting (P-RCV) (and by political scientists as “single transferable vote”), in which voters rank candidates, 1, 2, 3, and it can be used either in partisan or nonpartisan elections (with a single-winner variant – instant runoff voting – that is used to elect governors, mayors and other executive branch offices).

In my view, all of these PR methods are superior to the current winner-take-all method used in the US, for the reasons outlined above. But it’s also true that, of the various PR methods, one or two of them will likely prove to be more suitable for use in the US.

Specifically, some PR systems would allow the continuation of longstanding American traditions that: 1) enable voters to vote for individual candidates, 2) preserve some degree of a geographic connection between voters and their representatives, 3) don’t require a massive increase in the size of US legislatures (though a modest increase would certainly be beneficial), and 4) waste fewer votes and allow more voters to cast a vote for a winning candidate or party.

Let’s look at each of these PR methods in turn.

Party List PR.  Of all the PR methods, the most widely used is the Party List system. It is simple to use, including for low-information voters, because all the voter has to do is to pick a single party that she/he wants as their representative. The votes are added up, and parties are awarded seats in proportion to their share of the popular vote. Simple. Parties are elected from multi-seat “super districts” which tend to be large -- dozens of seats per district. Some countries, like the Netherlands and Israel, have a single nationwide electoral district with 120 or 150 seats.

The party lists often are used to present a diverse slate of candidates to attract more voters. More parties have started using their lists to nominate a lot more female candidates, even rotating male-female slots, which in various countries has elected a higher percentage of women than any other method. Other represented diversities include younger candidates, ethnic minorities and regional diversity. The head of the list, called the “puller,” is usually the most popular personality who appears in debates and on campaign literature.

However List systems are known for having weak links between elected representatives and constituents.  With “closed lists,” which usually result in party leaders picking the candidates, voters have no opportunity to determine the identity of the persons who will represent them, and no identifiable representative for their town or village. Nor can they easily reject an individual representative if they feel that he or she has performed poorly in office. In fact, the ballot paper (see link) contains only the party names and symbols, and a photograph of the party leader (the puller), but no names of individual candidates.

Open List variant. For this reason, some countries use an “open list” system in which voters are permitted to vote for an individual candidate, and that vote then doubles as a vote for that candidate’s party in figuring out the proportion of legislative seats won by each party. Candidates that receive the most individual votes are pushed higher up the party list. In some open list countries that determines which of the party’s candidates will get elected, but in other countries not necessarily. The Institute for Democracy and Electoral Assistance (IDEA) based in Stockholm warns that Open List “is notable for the large number of variations in the way it is implemented rather than for a set of rules common to all OLPR frameworks” and that it is “more of a concept” than a single system.

Hence, at least in theory, individual voters can influence which candidates ultimately take the seats in the open list legislature, depending on the country. In reality, according to IDEA and the experts at ACE Electoral Knowledge Network, most voters still mark their ballots only for parties rather than candidates, and in many countries the open choice option has limited effect. Canadian political scientist Henry Milner says “Sweden has found the right balance.” Any candidates who receive a number of personal votes equal to five percent or greater of the party's total number of votes in the multi-seat district will automatically be bumped to the top of the list, regardless of where the party leaders ranked them on the party’s list. But, says Milner, “most voters do not do this, which to me means that party supporters tend to have confidence in their party’s leaders.”

So in some countries, even in a number of open-ish list systems, a candidate’s position on the list, and therefore his or her likelihood of election, is dependent on currying favor with party bosses; their relationship with the electorate is of secondary importance. This overbearing influence can extend into the legislative process itself, when legislators sometimes struggle to remain independent from the directives of party leaders. This has sometimes led to entrenchment of power for certain leaders, with no easy way to oust them.

Wasted votes. Another important — though underappreciated — criticism of List systems is that in many (though not all) countries they tend to waste a lot of votes. There is a percentage of votes – called a victory threshold – that a party must pass in order to win any seats. In Germany, for example, that threshold is five percent of the vote. In the 2013 federal elections in Germany, 34 political parties competed but only five parties crossed the 5% threshold and won seats. Votes for the other 29 parties were wasted, resulting in 15.7 percent of the List vote going to losing parties. In Germany’s more recent 2021 elections, approximately 7.5 percent of the List vote went to parties that fell short of five percent. In Italy’s recent election, 11.2 percent of the List vote went to losing parties; in the Czech Republic 19.9 percent, Slovenia 24 percent, Slovakia 28.4 percent, Latvia 28 percent, New Zealand 7.8 percent. Elections in other countries by Party List have resulted in much lower percentages of wasted votes, so there is no hard and fast rule. But when it happens in a close election, it can affect the outcome of which party finishes first and is allowed to try and form a coalition government.

In fact, in the Israeli election of 2022, there were 40 political parties and 10 of them crossed the 3.25% electoral threshold necessary to win seats. With votes for the other 30 parties being wasted, it meant that 8.5 percent of the vote did not count for a successful party. The front runner party Likud, led by former Prime Minister Benjamin Netanyahu, achieved only 23 percent of the popular vote. Yet because nearly 300,000 anti-Netanyahu votes were wasted, his right-wing bloc was able to barely squeak out a majority of seats.

Meanwhile, two Arab-based political parties, Meretz and Balad, barely failed to cross the threshold and win any seats for the first time since the early 1990s. Votes for those two parties totaled over six percent and were wasted, as were votes for other small Arab-based and progressive parties. That drop-off allowed the conservative Netanyahu to barely scrape together a majority of legislative seats for his rightwing, quasi-religious coalition government.

What if the U.S. House was selected by Party List PR?

If that were to happen, this tendency to waste votes going to small parties would be greatly compounded by the fact that 37 states have too few House seats to have a statewide victory threshold under 10%. That includes 21 states with only four seats or fewer, where the victory quota would be at least 20 percent of the popular vote. So any voter casting their ballot for a political party with less than that amount would be wasting her/his vote. Seven states have only a single representative, so no proportionality would be possible there at all.

Political scientist Lee Drutman and Professor Aziz Huq have written in favor of an open-list system with super districts of 5 to 7 seats; Professor Jack Santucci has spoken in favor of a similar system. This would be an interesting experiment to try in the US, yet would struggle in the many US states that have four seats or fewer. A number of smaller nations, such as Denmark, Luxembourg and Slovenia elect seat magnitudes of anywhere from 4 to 11 by open lists. But from what we know about open list systems, few voters actually pick individual candidates, most simply vote for the party. The smaller district magnitude would preserve a sense of regional or geographic representation, but in a five-seat district, any voter casting a ballot for a party with less than 17 percent of the vote would waste their vote; in a seven-seat district about 12.5 percent would be the cutoff.

Currently in the US, only 29 states have five or more representatives, and only 13 states have 10 or more. We have seen how in Germany, with a five percent victory threshold, that has resulted in wasted votes of anywhere from 7.5 to nearly 16 percent. This number would increase exponentially with a threshold of 20 percent or higher; elections would be hemorrhaging wasted votes. That in turn would pretty much ensure that minor parties would never win 20 percent and get elected, especially if there was no ability for several minor parties to co-aggregate their vote, either through transferable cross-ranked ballots or some kind of fusion capability. With the latter, small parties are allowed to group together for an election but are listed separately on the ballot paper, and the votes gained by each are totaled together for the joint list, increasing the chances their grouping will finish above the threshold (a practice called apparentement).

And that in turn would not only undermine multi-party democracy, but it likely would lead to a certain amount of disproportionality. In a four-seat district that leans Republican in a 60-40 split, the GOP list of candidates would win three out of four seats and the Democrat list would win only one seat. Now add in minor party lists, in which voters would be wasting their votes on small parties that have no chance of reaching the 20 percent threshold; in a barely leaning Republican district with four seats, say 52-48, the presence of a vigorous Green Party list could easily result in a lot of wasted votes from progressive voters, splitting the center-left vote and resulting in the Democrat list still winning one seat and the GOP list winning three seats. Multiply this “wasted vote” dynamic across the geography of 435 House districts and its quite conceivable that this ostensibly “proportional” voting system would not result in nationwide proportional representation at all.

This is a real problem that, as far as I know, Party List proponents have yet to address. It is quite conceivable that, in the US context, a successful Party List system would require an enormous increase in the size of the US House of Representatives, which currently has 435 members. While a moderately larger House is certainly long overdue, to make a Party List system meaningful in states like Alaska, Wyoming and the Dakotas, which currently have one seat each, the number of House members would need to increase from one to at least 10 seats, resulting in a 10 percent threshold for a party to win one seat, forty percent to win four seats, etc.

That means the rest of the House, in each state delegation, would have to also increase by a factor of ten. Having a House of Representatives with over 4300 members, or individual states like California with over 500 members, Texas with 380 or Florida with 280 House members, seems like a non-starter. Even half that size seems wildly unrealistic. Indeed, looking at the German Bundestag with its MMP system, currently it has 736 Members of Parliament representing its population of 84 million people, about half of those from winner take all districts and the other half from the party lists. With a population four times that size, the US would proportionally require a House elected by Party List PR of nearly 1500 members. It seems farfetched to believe that the famously “big government”-averse sentiments of most Americans would accept a reform touting a federal legislature of that size. It might set back the drive for PR to have its success contingent on a controversial call for such an enormous increase in the size of the federal House of Representatives.

The best immediate target for Party List reform, therefore, would likely be state legislatures, which often have anywhere from dozens to hundreds of elected legislators. In addition, Party List advocates would do well to propose some method for reducing the number of wasted votes, such as either allowing voters to rank their parties and using a RCV-like counting process to transfer the ballots from those voters who picked unsuccessful parties to their next party choice; or a fusion-type mechanism, as used in countries like Poland and elsewhere, that allows parties to run together on joint lists.

Minor parties are going to try to win their own seats and build their brand, including in the presidential race, as they should, if they are to be full participants in a representative democracy based on proportional representation. It behooves any design to make sure that nearly all voters are able to select a political party that can get elected and represent them. This would be new uncharted territory, since I’m not aware of any PR democracies that have combined a List system with a ranked ballot in a way that avoids wasted votes.

While state legislatures are a good target, at some point Party List advocates will have to put forward their plan for the U.S. House and grapple with the large number of states that only have one, two, three or four representatives. That is not enough seats for a Party List election that has not put into place a mechanism to deal with wasted votes. It might be possible to have the larger states like California, with its 52 House members, elected by a List method with larger multi-seat districts and the smaller states elected by a method that is more friendly to small multi-seat districts, such as proportional ranked choice voting (more on that below). But the small size of the US House is going to be a significant barrier to a true Party List system arriving in the US.

Mixed Member Proportional (MMP). This Party List variant might be an easier sell in the US than either an open or closed list system because it combines US-style "winner take all" districts with a Party List system. So it combines something familiar with something new. The voter actually has two votes, one for their district representative and a second vote for the political party they most prefer. Overall, the number of seats won by parties will be proportional to the List vote, but some of the seats will be filled by representatives from districts (there are versions of this, called non-compensatory, in which the overall results are not exactly proportional).

So MMP has all the advantages of the List system, and combines that with local, geographic representation. Though that can have the undesirable side effect of creating two classes of legislators—one group primarily responsible and beholden to their district constituents, and another from the national Party List without geographical ties and beholden to the party. This may have implications for the cohesiveness of the party’s team of elected representatives. Also, just like in the US, the winner take all districts tend to elect fewer women, and so MMP has not been as beneficial for women’s representation as straight up Party List PR.

Proportional Ranked Choice Voting.  Also known as Single Transferable Vote, P-RCV is a ranked ballot proportional voting method, and the advantages claimed for List PR systems generally – broader representation, more diversity, higher turnout -- apply to P-RCV systems as well. While P-RCV is used in partisan elections in Ireland, Australia and elsewhere, voters are ranking individual candidates, and therefore it is the best method for achieving proportional representation in nonpartisan elections. It is also the best method if the goal is to allow for choice between both parties and candidates.

The multi-seat “super districts” tend to be much smaller than in List systems, anywhere from 3 to 7 seats per district, so it retains a flavor of geographic, i.e. local representation and a link between voters and their representatives. Furthermore, the system provides a better chance for the election of popular independent candidates than List PR, because voters are choosing candidates rather than parties. It also provides incentives for coalition-building through the reciprocal exchange of preferences between parties, which can help reduce partisan polarization.

A voter can decide to rank a candidate from a particular party first, a candidate from another party second, an independent candidate third, and so on, which provides maximum choice for voters (also, a Party List option can be added to a P-RCV election; this is done for the Australian Senate, in which the voter has the option of checking a box for a single political party, which automatically ranks all of that party’s candidates).

Fewer wasted votes

One big advantage of P-RCV over Party List or MMP is fewer wasted votes. The ranked ballots mean that votes for parties or candidates that don’t have enough support to reach the victory threshold are not wasted – voters’ rankings are used to reallocate votes to parties or candidates, using every ballot efficiently. P-RCV maximizes the number of voters who actually cast a vote that helps elect a candidate or party. In the most recent Irish elections, only one percent of votes were cast for losing parties and therefore wasted. The use of transferable ranked ballots is why P-RCV can be used successfully in a district magnitude of 3 to 7 seats (and a victory threshold of 12.5 to 25 percent), a design that is used in the Republic of Ireland and Australia. Few votes are thrown away. In contrast, a List system with low magnitude districts would likely result in larger numbers of wasted votes and consequent instability in election outcomes.

P-RCV also is a strong fit for many other aspects of America’s governing structures and traditions. It can be used for many nonpartisan local elections, and for Top Two/Four/Five primary elections that are nominating multiple candidates. Like a Party List system’s ability to allow parties to diversify candidates on their list, P-RCV incentivizes the major parties to use the endorsed rankings to diversify their pool of candidates and represent more of their “big tent” of supporters. The ability among parties to swap ranked preferences allows major parties to harness minor party enthusiasm while still enabling minor parties to have their own identity and hold the major parties accountable.

In addition, the single winner form of RCV (i.e. instant runoff voting) would be sensible to use for elections for president, governor, Senate and other executive offices, simplifying the role for voters by using ranked ballots across all types of races.

It should be noted that there are other alternative electoral methods used in the US that allow voters to vote for candidates rather than parties, such as cumulative voting and limited voting, which are used in more than 100 US localities, often to settle Voting Rights cases. But the results for cumulative and limited voting can be erratic and less than proportional (in fact, they are referred to by some experts as semi-proportional methods). Proportional ranked choice voting is the only fully proportional system of the candidate-based alternatives to the winner-take-all electoral method used in the US.

So which of these methods will work best in the US?

Since all of these methods are defined as proportional representation, and all benefit from the virtues of PR systems in general, this question really hinges around the differences between these different methods, and which are a better fit for US politics, traditions and culture.

As previously outlined, Party List systems are simple to use and to administer, and ensure broad representation of multiple parties and higher numbers of elected women. But it seems to me that party list systems have other features that would not be popular among Americans.

First, they completely sever the geographic connection between voters and their representatives. The US has such a long history of voting for representatives by district, I fear this might well be too great of a stretch. Voters wouldn’t have an opportunity to vote for candidates at all, instead they would vote for political parties. Yet most Americans appear to loathe at least one, if not both, of the major political parties. In fact, I would say that most Americans don’t vote for a party or candidate they like, they vote against the party or candidate they most fear. Independent voters – which in a number of states is the largest bloc of voters – generally loathe both political parties.

Some critics of that perspective say, “Well, if we had PR the parties would act differently and so they wouldn’t be so unpopular.” I believe there’s some truth to that, but it’s kind of a Catch 22, i.e. you can’t do it because you’ve never done it before. It is going to be hard to convince most Americans to try this “European” system where the dreaded political parties are even more prominent than they are currently.

These unattractive features could be mitigated somewhat by using an open list method rather than a closed list. But as previously mentioned, depending on how it is structured, the open list method does not necessarily change the list selection of candidates all that much, and so party bosses could retain formidable influence. This would not go over well with most Americans, I don’t imagine.

Given this antipathy towards political parties, and the familiarity with voting exclusively for geographic-based candidates, it would seem that trying to move to a political system where voters are voting only for parties and not candidates with little connection to geography flies against deeply-embedded American traditions and culture. Add to that the likelihood that, to enjoy the full benefits of Party List PR it will be necessary to dramatically increase the size of the US House of Representatives – more politicians, in other words – and this doesn’t seem to portend a successful passage for reform.

In addition, I’m troubled by the way that Party List methods waste so many votes. When the election is close between two major center-left and center-right blocs, the high number of wasted votes actually could make a difference in the outcome. Using a Drutman-Santucci style open list method with a district magnitude of 3 to 7 seats – meaning victory thresholds of 12.5 percent to 25 percent – would result in even more wasted votes as voters cast ballots for smaller parties that usually will be incapable of reaching that high threshold. Those voters will not contribute to a winning candidate, party or coalition unless a mechanism is included that allows voters to use transferable ballots and rank their candidates/parties, or allows parties to fuse and run together on joint lists.

For all of these reasons, I think the American culture and tradition matrix is going to be more open to a PR method that allows them to 1) vote for candidates that, 2) to some degree preserves a geographic connection between voters and their representatives, and 3) wastes fewer votes, and 4) doesn’t involve a huge increase in the number of officeholders.

With those four criteria for the best method, proportional RCV would seem to have a leg up on the competition. P-RCV in smaller multi-seat super districts of 3 to 7 seats would preserve some degree of a geographic connection, as well as allow voters to rank their candidates. It is the method that wastes the fewest votes of all because of how its system of fractional transfers of surplus votes allows candidates and parties to benefit from the rankings that facilitate coalition-building and vote-sharing among different constituencies. Some critics say the system is too complicated, and certainly this PR method would require the most voter education. But given how precisely it uses every voter’s ballot, I don’t think of it as complicated as much as I think of it as sophisticated. And the other methods are primitive in comparison.

Mixed Member Proportional also has some great virtues in that voters still get to retain their district representative while also gaining a second vote that allows them to vote for the party they like. So they are benefiting from both geographic and ideological representation. This also retains a core part of the current US political system, that of the district representative.

However, because the second vote is a List system, that vote could easily result in too many wasted votes, and would also likely require a significant increase in the size of US legislatures to result in meaningful representation. Germany uses MMP and in recent years its Bundestag has swelled to 736 Members of Parliament, the largest democratically elected legislature in the world. The current German government has proposed legislation to cap the number of seats at 598, which is still enormous by US standards, given Germany’s much smaller population.

Political scientists have been weighing these factors for decades. Its seems telling that a comparative assessment of the potential impact of 37 structural reforms by 14 political scientists — including PR, redistricting reform, open primaries and other possible reforms — found that proportional ranked choice voting was found to have the greatest positive impact on US democracy. MMP also was rated highly.

One thing is certain: there are no easy answers to some of these fundamental dilemmas of modern representative democracy, and figuring out which electoral system design will work best. Nevertheless, at some point an electoral system designer must propose a real plan. And I can tell you, after nearly three decades of doing just that, the devil is in the details.

It is incumbent upon Party List advocates to do the hard work of drafting their electoral plan(s) with enough concrete specificity that addresses many of these points. It seems to me they cannot avoid arguing for their design within the well-known, path-dependent constraints of US traditions, culture and history. In that direction lies the possibility of overthrowing Winner Take All-ism.

Steven Hill @StevenHill1776