Choosing a voting method for British Columbia: the case for a Mixed Member Proportional system

jq finalIn the second of two posts on the voting reform referendum in British Columbia, Jameson Quinn argues that the province’s electorate should support a move to some form of proportional representation. Specifically, he argues that the version known as Mixed Member Proportional is best for both the province itself and representative democracy as a whole. 

This is the second of two posts on the upcoming British Columbia (BC) referendum on proportional representation (which I’ll call ‘pro-rep’, because ‘PR’ has too many other meanings). In the first post, I discussed the context and rules of the referendum itself. In this one, I’m going to discuss the options available, in the context of theories of voting and democracy. I’ll also look at some of the arguments being used in this campaign.

Before I start, I should lay my cards on the table. I am unabashedly in favour of voting reform. For over 20 years, I’ve felt that choose-one voting, as used in most English speaking countries, is a badly-flawed form of democracy. Also known as FPTP, for ‘first past the post’, choose-one almost inevitably leads to spoiled elections, dishonest strategic voting, or both. My support of voting reform is what led me to join the board of the Center for Election Science, a non-profit that advocates for better voting systems.

As my previous post mentioned, I was one of the primary organisers of the BC Symposium on Proportional Representation. Though many of the experts and activists attending that symposium were, like me, advocates for reform, the symposium itself remained neutral on whether changing to pro-rep was a good idea. Our role was to impartially lay out the evidence regarding the relative advantages and disadvantages of various options.

The functions of representative democracy

In order to argue that pro-rep is a better option for BC, I’m going to be explicit about what I mean by ‘better’; that is, what representative democracy is for. I believe that representative democracy serves three basic functions:

  1. Provides a format for regular, orderly, non-violent transitions of power. If you’re dissatisfied with the current government, you don’t have to pick up a gun; if you’re currently in power, you avoid excessive corruption, because you know the next government would investigate. Any voting method can fulfill this function, as long as it’s seen as legitimate; so I won’t discuss this further.
  2. Helps make relatively good decisions. Diverse groups in society get to have input, and all in all the wisdom of these diverse points of view can, at least sometimes, add up to more than the sum of its parts. We all know that democratic decisions can still sometimes be terrible, but as far as I can tell, any other government structure is more often worse.
  3. Is relatively efficient at making decisions. Unlike direct democracy, where everybody has to weigh in on every argument, representative democracy keeps that task to a relatively small group of professionals. Ideally, the voting method should make voters’ task easy, while choosing representatives who are both well-qualified for their jobs and reflective of the community from which they are elected.

Any well-designed pro-rep method—which all three of the options on the BC referendum are or could be—is clearly superior to the current choose-one method on the latter two points: each would give outcomes that are more representative without being more divisive (point 2), all the while remaining comparably simple for voters (point 3).

Proportional representation versus ‘First Past The Post’

It’s almost tautological to say that a pro-rep method is more representative than a choose-one, winner-take-all one. But tautology or not, it bears repeating. So to underline this point, I’ll  examine some of the ways choose-one can lead to horrible results.

The most obvious one is spoiled/‘wrong-winner’ elections, where one party wins more seats despite getting fewer votes than another party. That’s been the case for the BC government after the 1954 and 1996 elections, but looking around Canada, there are much more recent examples. In New Brunswick recently, the Progressive Conservatives got the most seats, 22, with 32% of the vote, while the Liberals got just 21 with 37% of the vote. In Canada as a whole, it is commonplace for the New Democrats to get more votes but less seats than the regionally-concentrated Bloc Québécois. You can also look at ‘false majorities’, where one party wins a majority of seats with far less than a majority of votes; for instance, in Ontario, Doug Ford’s Progressive Conservatives have 61% of seats on just over 40% of votes.

A more hidden problem is strategic voting and major-party bias. Since choose-one votes for minor parties are predictably wasted, many voters choose to strategically vote for a larger party even though they honestly prefer a smaller one. This distorts the democratic message of the election, and can make it possible for interest groups that ‘capture’ the power structure of one or more parties to impose their will even if a majority of voters oppose them. It’s tough to measure the extent of the impact of this voting strategy, but the wild last-minute polling swings in the recent Ontario election suggest it is no minor matter.

Underlying both these points is the problem of fruitless votes. In choose-one voting, on average, almost half of all votes (and sometimes more) are ‘fruitless’ — they do not help elect a winning representative. In proportional representation systems, only a tiny minority of votes are fruitless. For instance, in an at-large proportional election for 99 seats, an ideal proportional method can ensure that under 1% of votes are fruitless, and almost any pro-rep method can easily keep that under 5%. Given the requirements for regional representation in BC, fruitless votes might be as high as 10% there; but that’s still far less than the 50% average under choose-one.

Of course, there are still losers. Just because your vote fruitfully helps elect a representative you prefer, doesn’t mean that your representative becomes part of the governing majority. But even if your representative ends up in the minority, casting losing votes in the legislature, I’d argue you’re probably better off for having one than somebody whose vote was fruitless before the legislature was even seated.

The three methods of proportional representation on offer in British Columbia

So pro-rep, in general, is better than choose-one. Let’s take a closer look at the relative advantages and disadvantages of the three methods on the BC referendum ballot, all of which were among those evaluated and recommended by the BC Symposium.

First, there’s Dual Member Proportional (DMP). Voters are divided into two-seat ridings (the Canadian word for constituencies or districts) plus a few single-seat ridings in the very lowest-density rural areas. They get to choose between the two-candidate slates each party has on the ballot in their riding, or to vote for an independent. The first seat in each riding is given to the plurality winner, as with choose-one. Second seats are first given to any independents who finished in second place, and then the remainder are distributed in a way that ensures party proportionality while favoring the best-performing candidates within each party. Candidates who got a lower proportion than a set threshold (suggested to be 5%) locally would not be eligible for election; thus tiny parties could end up with no seats, even if they have enough votes to merit one or two on a purely proportional basis.

There’s two ways that a vote could be fruitful in this system: directly, by helping elect a local representative, or indirectly, by increasing a party’s proportion and thus helping elect a candidate somewhere else. Chances are, almost half of votes would be directly fruitful, and most of the remainder would be indirectly so, leaving under 5% of votes as fruitless. This system has the simplest ballot format of the three options; but also the lowest breadth of choice for voters, who would have only one choice per party in the general election, plus independents.

The second option on the ballot is Mixed Member Proportional (MMP). In this option, the province would be divided into a few general regions, with each region split into ridings. For instance, one region might have 25 seats and 15 ridings. The ridings would get one seat each; the other 10 seats would be allocated in such a way as to ensure proportionality.

Unfortunately, the referendum materials do not specify any further details about how MMP would work. That leaves it up in the air whether it would be a closed-list system (good examples are Scotland or Wales), an open-list or semi-open one (the best example for this being the one used in Bavaria, Germany). Such decisions, along with the specific region and riding boundaries, would be made by an all-party commission after the referendum.

In Welsh-model MMP, voters choose one local candidate and one party. Candidates win locally on a plurality basis, as BC does currently; and list seats are allocated to balance proportions, targeting the proportions from the party votes only. The problem with this system is that it creates a strong incentive for strategic voting. By voting for different parties on the local and party portions of the ballot, voters can in some cases increase their impact. Such a strategy is evident in Welsh elections. While it’s not as bad, in my opinion, as the problems with choose-one, it’s not as good as other pro-rep systems.

That’s why the BC Symposium recommended that, if MMP was to be an option, it should follow the Bavarian model. In this system, voters would choose one local candidate, and one other candidate from the combined list of candidates in the region but outside the riding. As above, the local candidates win on a choose-one basis. But the regional candidates are assigned to parties in order to balance seats to the proportion of all votes, counting local and regional votes equally. Within parties, the candidates with the most votes get any party seats. This has two big advantages over the Welsh model. First, voters still get to help choose individual candidates, not just parties, even if their local candidate does not win. And second, there is less of an incentive for strategic cross-party voting. In Bavaria, this has let voters have real say in the direction their party moves, without evidence of dishonest voting strategies.

If BC bases their system on a Bavarian model, they could make three changes that would improve the strategic resistance even further. First, if exactly one of a voter’s two votes (local and regional) goes to a party that ends up getting no seats, it should be transferred to the other party they voted for. That way, it’s safe to honestly vote for a smaller party with half of your ballot, as long as you choose a clearly-viable party with the other half. Second, if a voter’s chosen candidate wins locally, their regional vote should be counted for the same party as their local vote. This reduces yet further the incentive for strategic cross-party voting. And third, candidates below a certain threshold locally (which could be as high as 20%) can be eliminated. That prevents fringe candidates from winning with regional votes only, unless they can make a strong showing locally.

A Welsh-like system would be, in my opinion, the worst of the three options, though still better than choose-one. A Bavarian-like system, especially with the modifications above, would be the best. It gives voters excellent breadth of choice; preserves local and regional representation; and minimises fruitless votes. With a high threshold rule, it would also tend to lead to parliaments without too many tiny parties (a problem in the Israeli Knesset), while still allowing small parties to grow over time into larger ones, because voters could still choose them without risking a fruitless ballot.

The third system offered by the referendum is Rural Urban Proportional (RUP). This would have ridings consisting of between three and seven seats, using Single Transferable Vote (with ranked ballots) in urban areas, and MMP (perhaps with two-mark local/regional ballots) in lower-density rural areas. It offers the advantages of STV for city dwellers — reasonably broad and fine-grained options and good proportionality — while avoiding its primary disadvantage for rural voters — excessively large ridings.

To me, any of the three options would be a clear step up. Though it is unrealistic to hope that all BC voters would become experts in the ins and outs of all these systems, the media, government, and ‘yes’ campaign have all worked to produce educational materials on the options, which are now widely available and appropriate. In my previous post I mentioned the regrettable lack of a ‘citizens’ jury’ to help advise the voters, but other than that, I believe that the choice between these three options is being made as transparently as is possible in practice.

In terms of details that would be set after the referendum, the only one I see as truly key is the choice of model (Welsh, Bavarian, or other) for MMP. On this matter, I am optimistic that a desire to see pro-rep succeed would push parties to prefer a more open system, rather than attempting to retain more power by using a purely closed method. Note that there are still some reasons that including some closed list aspects could be valid, such as ensuring gender balance.

The opposing view

Of course, the ‘no’ campaign disagrees with me. But Dennis Pilon has responded more than adequately to the weakness of the anti-pro-rep arguments in the academic sphere. And the public arguments seem to me just as weak:

  • Pro-rep would not lead to excessive instability or discourage investment. In fact, there would likely be more overlap between successive governments, and thus less risk of policy whiplash.
  • Pro-rep would not mean that the Vancouver area, which has a majority of the province’s population, would steamroll rural interests. In fact, quite the opposite; it would make it far more likely that the governing majority would include both urban and rural MLAs.
  • Pro-rep would not increase the risk of empowering radical or racist minority parties. Not only do all three options include threshold mechanisms to ensure tiny parties win no seats; pro-rep in general ensures that the governing majority represents a majority of voters, avoiding empowering parties in which racist minorities have an outsized voice. BC need only look to the US or Ontario for examples of how choose-one can work out, examples I think most British Columbians wouldn’t want to emulate.
  • Pro-rep is not too complicated for the voters. It’s used in most democracies without appreciable confusion, and no country has ever switched from pro-rep back to choose-one. Presumably this is why the BC government feels safe in offering a second referendum.
  • It is not scary that the riding boundaries for a pro-rep system would be drawn after the referendum. Under pro-rep, unlike under choose-one, riding boundaries are relatively low-stakes issues; and in all three options under consideration, the largest rural ridings would be roughly the same size as today.


In sum, I think the referendum is an opportunity for BC to put itself at the forefront of democracy. All three options on the ballot give at least an opportunity for the province to use one of the most thoughtfully-designed pro-rep systems in the world. I eagerly look forward to seeing what British Columbia decides.

This is the second of two posts on the upcoming voting reform refeendum in British Columbia. The first post, which relates the political-historical background and discusses the format and rules of the referendum, can be found here.

About the author

Jameson Quinn lives in Cambridge, Massachusetts, and serves as a board member of the Center for Election Science, a US voting reform non-profit.