Wednesday, January 18, 2012

Other Good Ideas: Proportional Representation

(Note: This post was going to go up sooner, but with Wikipedia going down to protest SOPA, I put it off for a bit.)

Before disappearing for a third of a year, I was asked a simple question in the comments: Would I support an effort in favor of proportional representation (PR), even if it were for single transferable vote (STV), the PR method on which instant runoff voting (IRV) is based. And yes, I would. I often refer to Duverger's Law, the observation that single-winner plurality elections tend toward two-party dominated government, and PR effectively attacks one of the two legs of that problem.

What is Proportional Representation?

The familiar method for electing a legislature is to divide the region into districts, and elect a single representative from each one, the idea being that someone from where you're from, will want what you want. In practice, this isn't always true. There tend to be factions in each district that split along similar lines, and these ideological splits may be completely independent from the geographic split of the districts. It's possible (by accident or by design) for the narrowest of political majorities to win every single seat in a legislature, or for a group that is a minority in the electorate to have a majority of legislative seats. If the legislature is intended to represent the viewpoints of the whole electorate, only with a smaller number of participants, then these single-winner districts often fail.

The goal of PR is to fix this deficiency. Rather than each district electing a single winner, they elect multiples, and in proportion to the number of votes they have received. In this way, the diversity and popularity of viewpoints among the people will more-closely match the diversity and popularity of viewpoints among the representatives.

Variations and Usage

There are many different variations of PR. The most basic is changing the number of winners elected from each district. This number may be as small as 3 or 5 (usually, but not always, the number is odd) or as large as the 120 used for the Israeli Knesset, which is elected from a single, nation-wide "district". The other major point of variation is whether the system is candidate-based, like our current single-winner elections are, or whether it is explicitly party-based, so that voters do not vote for individual candidates but rather for a party or a list of candidates they have supplied; there are also mixed systems, like the one used in Germany, which provide for both simultaneously. There are also different ways of dividing the votes--basically, which way to round fractions--which can favor having fewer, larger parties or more, smaller parties. Depending on the precise method used, a PR ballot may look just like a plurality ballot, like a pair of plurality ballots (one listing candidates and one listing parties), or it may ask you to rank several choices (candidates or parties), or it may even look like an approval voting or score voting ballot

Most of the worlds democracies use proportional representation, which may come as a surprise to those of us in the United States, but that's because some of the largest exceptions--the United Kingdom and Canada--also happen to be some of the countries we Americans are most familiar with.

Criticisms and Difficulties

Since multiple winners are elected in each district, some critics have said that PR severs the connection between a voter and "their" representative. However, since each district elects multiple representatives, it is more-likely that you'll be in political agreement with at least one of "your" representatives. I also have commonly heard complaints about the party-centric aspects of party-based PR, but usually this is from people who don't realize that there are PR methods that are not party-based. Candidate-based PR tends to become somewhat unwieldy as the number of winners increases, as it becomes more difficult for voters to determine and remember which candidates they prefer. The counting procedures, regardless of the specific form of PR used, are also more complex than plurality voting is. Finally, there are people who legitimately believe that it is best to have as few large political parties as possible, in order to best guarantee the existence of a cohesive majority government, although this is less-significant in the US since we have less expectation of party-line voting, as well as extra veto-points in the President and in one chamber of congress acting more-and-more often under super-majority rules via the filibuster.

Also particular to the US, it is currently federal law that the states elect their representatives from single-winner districts (the law was enacted because some states had chosen to elect all their representatives at-large, guaranteeing that 51% of the voters would choose 100% of the representatives, eliminating representation for minorities) and that law would have to modified in order for any form of PR to be usable for House elections. Furthermore, the constitution requires that House seats be apportioned to the states, so no cross-state proportionality would be possible without a constitutional amendment, making the system rather moot for the 7 states who have a single representative.

This hasn't stopped PR from being used in several US cities, even, at one point, New York City; although most of these later returned to single-winner elections, so that, as of today, Cambridge, Massachusetts has the only governing body in the nation that is still elected via proportional representation.


For those who favor a diversity of opinions, proportional representation can be an effective way to achieve it. It has shortcomings and legal hurdles standing in the way of its use, and obviously can't be used for single-seat offices like a governor or the president. I won't stop pressing for approval or score voting, but there's no reason these advances can't be pursued in parallel. Although, anyone who sought to do so should familiarized themselves with how and why PR elections were rolled-back in recent US history, in order to avoid running into the same mistakes again. But that's a blog for another day...

Now, why would I support STV, but not IRV? It's a known property of STV that, the fewer winners it elects per district, the less proportional it is (and the balancing act becomes ballot complexity versus proportionality.) Since IRV is precisely equivalent to STV with one winner, we know that it loses all proportionality (in other words, IRV is not a PR election method.) This, combined with the other deficiencies of IRV make it an unacceptable alternative to my mind.


  1. STV is the most likely method of PR that would be the most well received, and it as you already know, has a history of usage in the United States (a history that's been buried pretty far and which needs to be dug out). I just can't see any party list form of PR ever getting off the ground in the United States. American's just don't want parties to control the process (despite our single winner seat system, they do now), and STV is the only widely practiced form of PR that isn't party list based. (Or am I wrong there?) In fact, evidence suggests it reduces parties abilities to game the system.

  2. That's a complex question. There are non-STV non-party-list systems out there... although they're generally implied party-based (you vote for a candidate, but the party of the candidate determines how many seats each party get), or they're mixed like Germany.

    STV is probably still the most-widely used, but that's because there are so many variations. And it's not THAT widely used either. If you look at the map, there's Ireland, plus half-credit in India and Australia (there's a few others, but they're so small as to be invisible on the map.)

    Party-list PR is clearly the most popular world wide, and I agree it's a not going to be popular here.

    Like I said, I'd support STV. I'd prefer re-weighted range (or re-weighted approval). Or Asset. Or even an implied party list. But I'd support STV.

  3. "And it's not THAT widely used either"

    Good point, I should have said "The only currently practiced and not purely theoretical". (Though I should point out the map isn't entirely accurate, some localities use it in nation's that don't use it federally, such as for example New Zealand ; Auckland uses STV, and all of their local health boards use STV as well) Of course there's good theoretical ones as well, like the PR variants of Approval and Range.

    True, party lists are by far the most popular, but I don't think they'd be very popular in the US, which you agree. American's just have a distaste for that type of politik, whether erroneous or not.

    I'll revise my statement. All non party list forms of PR are the ones that have the best chance of "catching on" in the US, whether STV, cumulative (though that's only Semi-PR), or Re-weighted Approval or Range. Though my guess is some STV variant would be picked, since that has a longer history of use, and it's what organizations like FairVote seem to prefer.

  4. Party-list PR is clearly the most popular world wide, and I agree it's a not going to be popular here.

    I'm not a believer in the creed of American Exceptionalism. Find some people with money to finance a good ad campaign and I think you could get open list PR passed for one house of the state legislature in one or more initiative/referendum states. Bicameralism is popular for some reason in the US, but what is the point of two houses with mirror-image constituencies when other arguably better alternatives exist? Go for broke-make the state as a whole a single constituency in partial homage to Netherlands and Israel.

    Term limits could be included in the package to allay the fears of those who predict a legislature loaded with party hacks who have permanent sinecures.

    If you think of the vast social changes of the last 50 years it seems strange to think that the population would somehow draw the line at PR. It just needs a good sales pitch.

    IIRC, all of the American Empire's postwar colonial reconstructions (Japan, Germany, Iraq) use PR-have Krauthammer explain that to the peasants. Also, #1 ally/friend (which every POTUS candidate must pledge permanent allegiance to) has used an extreme form of PR for its entire existence.

  5. @no name,
    I think there might be a misconception here. No one was arguing that PR period would be unpopular in the US, just that party list based PR wouldn't be very popular. Thus it might be more prudent to explore non party list based PR systems, and see how well they'd work in the US.

    Of course, you have a point about advertising. It does largely shape the viewpoints in whatever society that has a heavy amount of it, but keep in mind (and there would be in anything hotly contested and political) the opposing ads that would be aired. The best selling point against any party list system in the US is, well, the party lists themselves. They'll just air ads with "Do *you* want to choose your representatives, or some party bureaucrats to choose them for you?" I don't think it's unfair to suggest that if one was asked if they'd rather have parties choose their representatives rather than constituents, the most popular answer would be "over my damn dead body!".

    But yeah, PR is definitely worth exploring, we should just keep in mind the United State's population unique culture and mindset in these regards when determining what would be "viable", at least in terms of a sales pitch.

  6. Excellent points, no name.

    The popularity of bicameralism surprises me, too. (Go Nebraska!) For congress, there is at least a difference in the relative sizes of constituencies, and used to be a difference in the electorate, but most states have never had such things. (My own state uses the exact same districts, only it pairs them up for senators, so the senate is precisely half as large.)

    Term limits is something I've written about in the past, but, while researching this post, I'd noticed something about Mexico. In Mexico, a legislator cannot succeed themselves. I think that's brilliant!

    And good points about Japan, Germany, Iraq, and Israel. I don't know about the rest, but I know Americans were very involved in ensuring Iraq used PR.

    Thanks for posting!

  7. thanks for sharing information.. nice article,.

  8. Hi Dale,
    I think it deserves mention that 3-seat LR Hare is the only third party friendly party list form of PR that doesn't require party lists, since they'd be implied. So it's still potentially useful for the US.

    IMO, there needs to be a balance between intra-party discipline and intra-party democracy. A balance that has been disrupted for the Republican primaries by Citizen United enabled Super PACs. Open lists for all elections and term limits can unhealthily subvert that balance. Parties need Edward Kennedy-type visible leaders to act effectively together. Mexico's law forbidding incumbents to run again has been disastrous, since the only accountability an elected official has is to their party, who may give them access to going bigger and better...

    Here's a relevant article on whether or not FL GOP primary will be PR or winner-take-all.

    Also, you didn't mention the use of 3-seat quasi-PR in IL from 1870-1980. That kept one party from dominating their politics, which allowed other states to be more independent politically from IL, who they were economically dependent on.

    I think Duverger was wrong. If you only use single-winner FPTP rules (or multi-winner plurality rules that act just like single-winner FPTP rules) then the consequence is single-party rule, not unlike what existed in MX and Egypt til they started using some PR.

    At issue is not whether we use PR or not, but how to use PR, taking as given that there need be single-winner elections. But I do appeal to Duverger's law and the strength of the US president (and how the two major parties have gamed the system in their favor) for why I take as a given that the US will have 2 major parties. This in turn renders the diffs among alternatives to FPTP to be small, so what matters most is to get the reform asap, not to struggle to figure out what the best election rule in "real life" would be.

  9. FPTP has some merit as a way of electing a single winner in a single member district, but is not a good way of electing a parliament/legislature or government.
    A PR alternative you didn't mention, but might like to consider is Direct Party and Representative Voting (see

  10. Interesting. Rather than use "top up" seats, like MMP, you just re-weight existing seats.

    I wonder though how DPR would handle the case of, for example, the German FDU, which won zero constituency seats, but 93 party seats (15% of the Bundestag!) Who gets to cast those votes?

    For comparison, the CDU had 31.2% of the Bundestag with 173 constituency seats, and the Alliance/Greens got 10.9% with 1 constituency seat, so that one Green would have about 60.4 times the voting power of each CDU member. Okay, fine. But how much voting-power should the non-existent FDP member get?

    I think you need to find an answer to at least that question.

  11. I don't see any good reason that the R part of the equation couldn't be done with a better election rule than FPTP.

    Also, what happens if a party doesn't get a plurality anywhere but gets a significant fraction of the overall vote?

    Why not consider super-district with 4 candidates each and use a 3-seat LR Hare for 3 of them and then something better than FPTP for the remaining seat?

    That way you'd be multi-party and keep the status quo, more or less...


  12. I think DPR is a form of MMP and like MMP, it tends to make the outcomes of the nation-wide election more important than the "more local" election".

    I'd rather have it go the other way around...