A PR Primer for Ontario. Drafted by Réal Lavergne Oct. 8, 2026
(I have made some change to this copy of the Primer)
Executive Summary
If you are reading this, you are probably well aware of the general case for proportional representation (PR). However, you might be concerned about some of the details of how such a system would work, some of the risks involved, and how to bring about a sufficient level of consensus for change.
This paper is for you. The issue addressed here is how to build a better electoral system with a broad level of support from voters and political actors alike across party lines.
The paper starts by making the basic case for PR and briefly considers ranked ballots as an alternative, before turning to the political challenges of securing electoral reform and some of the conditions for overcoming those challenges.
It provides the basics of what we need to know and do to achieve the sort of multi-party consensus on electoral reform that we are going to need to make progress on this issue between now and the next Ontario election. The paper discusses what a basic PR model might look like for Ontario on an illustrative basis.
Electoral systems in the rest of the world
From our vantage point in Canada, it’s easy to think that our first-past-the-post system (FPTP) is the natural and most common way of holding elections. However, FPTP is still used in just three OECD countries: the UK, Canada and the US. To these can be added France, which uses a two-round majoritarian system and Australia, which uses a majoritarian system with ranked ballots (AV) for parliamentary elections and PR for senate elections.
The norm elsewhere is some form of PR, which is currently used in 80% of OECD countries, most of which transitioned to PR over 100 years ago. PR has had plenty of time to prove its worth. Comparative research has shown PR countries to outperform FPTP countries on pretty much any criterion one might want to study, as summarized in Fair Vote Canada’s A Look at the Evidence.
The Case for Proportional Representation
The basic case for PR is conventionally made in two principal ways. The first approach focuses on voters’ rights and voter equality. The second shines a spotlight on the more egregious mismatches that one observes between votes cast and electoral results under FPTP.
The first approach teaches us that voters are not equally treated under FPTP. Every citizen has a vote and every vote is counted, to be sure, but not every vote helps to elect someone of one’s choice. Those who live in safe seats legitimately feel disempowered. Voters who live in swing ridings have a greater chance to make a difference, but even they may need to vote strategically to make their votes count.
Under FPTP, if your vote fails to elect someone of your choice in your riding there is no place for those votes to go. Hence the expression “wasted votes.” With PR—using multi-member districts of one sort or another— almost all votes do have a place to go. Consider the example of a five-seat district used later in this paper. A district like this can elect a combination of candidates from different parties, and because of this almost every vote—usually 95% of votes or more— typically helps to elect a candidate from the party of one’s choice in a proportional way. Under FPTP, the comparable share of voters who can make their voice count is usually less than 50%.
The other way to make the case is to point to the distortions that appear under FPTP. Such distortions are legion in Ontario. We are currently in our third consecutive Progressive Conservative Party (PCP) majority government in Ontario, in which the governing party has won a large majority of the seats without winning a majority of the popular vote.
What explains this is that FPTP typically produces a “winner’s bonus” favouring the largest party. While FPTP typically overrepresents the winning party, smaller parties are generally underrepresented. Here in Ontario, the Ontario NDP (ONDP) and the Green Party of Ontario (GPO) have typically won a smaller proportion of seats than their share of the votes. However, in the last three elections the Liberals have been the biggest loser. In 2025, the Ontario Liberal Party (OLP) took 30% of the popular vote but won only 11% of the seats.
This same phenomenon also leads to whole regions that are represented by only one party. In Ontario each party has its geographic strongholds, with little or no representation of voters who chose other parties in those strongholds.
Finally, there are distortions over time, due to vote-splitting by the OLP, the ONDP and the GPO. In Ontario, this has been going on for over a century, most notably since 1943 when vote splitting [became a significant problem]. As a result, since 1943, the PCP has formed government 71% of the time, most often with the sort of unbridled power that comes with a majority of the seats. Overall, we have had false majority Conservative governments 58% of the time since 1943.
PR would eliminate or significantly moderate all of these problems. False majorities would be a thing of the past, all parties would get a share of the seats matching how voters cast their ballots, and different parties would have representation in every region. We would no longer have one party winning at the expense of other parties who are splitting the vote.
What About Ranked Ballots?
Ranked ballots can be used in PR systems or as a simple tweak to FPTP.
In the latter case, it’s formally called Alternative Vote (AV). While this approach is regularly and successfully used in single-winner contests such as leadership races, the only OECD country that uses AV is Australia for its parliamentary elections.
AV might help to address the problems of vote-splitting and strategic voting. However, it does not give proportional results and could make seats-to-votes distortions even worse. It’s possible that it might further concentrate power in two parties. This is what has happened in Australia, which has very much remained a two-party system in which an outside party like the Greens can get 11% of the popular vote but end up with no seats at all.
That said, let us remember that ranked ballots could be used to good advantage as part of just about any proportional system, allowing voters to more fully express their preferences. The best-known use of ranked ballots as part of a PR system is called Single Transferable Vote (STV). This model uses multi-member districts, and allows voters to rank their preferences across party lines if they wish.
The Politics of Electoral Reform
In light of all that we know about the problems of FPTP, one would have expected countries such as Canada and the UK to also have adopted some form of PR by now. However, tradition weighs heavily in these countries, and politics gets in the way of change.
A major source of resistance to change has to do with our elected politicians. It’s intuitively easy to understand that politicians once elected might be loath to change the system that brought them to power. Beyond that, when FPTP involves a predominantly two-party system as it currently does in most Canadian provinces, both of the major parties may decide that they are “OK” with FPTP, provided they have a fair chance of forming government from time to time.
Voters for these parties may feel the same way. Canada has had numerous referendums on PR and all have failed in one way or another, due to obstacles put in place by the government itself, but also a certain hesitation on voters’ part. A typical pattern is for up to 60% of voters to opt for the status quo, and around 40% for change. We might call those 60% of voters for the status quo the “comfortable majority.” This comfortable majority may understand that FPTP is failing them in important ways, but they have never known anything else. In the hockey-loving country that we are, many of us are “OK” with a win-some, lose-some electoral system.
This does not make it right that a referendum should allow this comfortable majority to impose its will on an aggrieved minority. Nor does it mean FPTP is a better system even for this comfortable majority. Most Canadians do not have the time nor the interest to investigate electoral systems in depth.
A more reliable source of legitimacy would be a multi-party agreement based on a recommendation from a citizens’ assembly or a consensus of non-partisan experts.
Those holding majority power inevitably feel that their majority enables them to get things done, ignoring the added value of minority voices and the dangers of policy lurch when the other main party takes over (on climate change or municipal reform for example).
Yet change requires a consensus across political lines. Indeed, it’s this sort of consensus that allowed most European countries to bring in PR in the first place. It was never one party winning power and deciding to just vote in PR. Indeed, there is probably no case in history of a party forming a majority government and bringing in PR. That’s not how it works in the real world.
The basics
To understand how our FPTP system gives us the distorted results that it does requires a closer look at our use of single-member districts as the fundamental source of these problems. The problem with having single-member districts is that this breaks the link between the popular vote and the way seats are allocated. To immediately understand this, imagine if party X won 40% of the vote in every riding and other parties won 30%, 20% and 10% respectively. The result would be an election in which party X won 100% of the seats with 40% of the vote. Of course, this never happens quite like that, but we do get false majorities, and we have had provincial elections in which one party won 97% of the seats or even 100% of the seats. Single-member districts are what creates the winner’s bonus.
It doesn’t have to be that way. Imagine that you were to merge a number of single-member ridings into one multi-member district (MMD), with the aim of electing a slate of district representatives in proportion to each party’s vote share. You could now vote for a candidate from the party of your choice and help elect someone from that party in a proportional way. In a 5 member district, a party with 40% of the vote would win 2 seats. A party with 20% of the vote would win one seat.
Multi-member districts are what makes PR systems proportional. Indeed, one of the best indicators of proportionality is the district magnitude of a country’s voting system, expressed as the average number of seats in each of a country’s voting districts, District magnitude (DM) can be as low as 1 in FPTP countries and as high as the total number of seats in the country in “pure PR” systems such as the one in the Netherlands.
Not all PR systems are equally proportional. The larger the district, the more proportional the result, because the share of votes for a party to win a seat would be smaller. However, anything larger than single-member districts would considerably enhance the proportionality of the voting system.
Some basic criteria
This raises questions about what sort of PR systems might be best in Ontario.
A first step might be to enquire which features of FPTP voters and politicians might like to keep. It does not matter here whether those features are equally loved by everyone. If we are looking for a broad consensus for change, there is a need to respect traditions and habits that are engrained in our existing system.
If we are looking to accommodate a wide swath of the electorate and of the political elite, we might consider three such criteria, in addition to a higher level of proportionality:
some form of local representation;
direct accountability of elected representatives to the voter; and
a simple system that is easy to use and easy to implement.
Of these three criteria, the notion of local representation needs some elaboration. Most people would associate this criterion with the status quo involving single-member districts. Such districts are certainly “local” in the most basic way, but can single-member districts truly provide meaningful representation for all or even most voters? Keeping in mind that the main function of our representatives is the crafting and adoption of legislation in parliament, it is not possible for a member to simultaneously represent all voters in a single-member riding.
So, what would it take for local representatives to represent all voters in a riding? This takes us back to the notion of multi-member districts. If we want “meaningful local representation,” in which representatives of a local district should represent as large a share of constituents as possible, the only meaningful option is some form of multi-member representation in which representatives of different political leanings are elected. representing the vast majority of voters. These multi-member districts may be more or less “local” depending on how large they are and how the boundary lines are set. The challenge is to find the right balance when agreeing on the size of multi-member ridings we are looking for.
An illustrative model
The most basic PR model that might be considered with due regard for the above criteria would be to replace single-member districts with relatively small multi-member districts (MMDs).
One can get quite large improvements in proportionality even with small MMDs. Already, with a two-member district, you would most often elect representatives from two different parties. With a three-member district you could elect two members from one party and one from another party, or one member each from three different parties.
Ireland uses multi-member districts with district magnitudes of 3 to 5 and ranked ballots and the Single Transferable Vote system.
However, most countries use somewhat larger districts than this and often add what are called “adjustment seats” as top-ups to compensate parties earning less than their proportional share of seats at the district level. Sweden and Denmark are examples of this approach.
Below is an illustrative example of how a 5-member district might work, using voting numbers from the 2025 Ontario provincial election in Etobicoke and York.
Let’s begin with what the ballot might look like, as illustrated below. The only difference compared to FPTP, in this basic example, is that there are more candidates to choose from. Voters would still get a single vote and put an X next to the candidate of their choice. The X would count for the candidate they prefer from the party of their choice.
When tabulating the results, the first step would be to determine how many seats each party deserves. The candidates elected to fill those seats would be the ones with the most votes.
Below are pie charts showing respectively
how citizens in this district voted in 2025,
what they got under FPTP and
what the result would have been in a 5-seat MMD.
A particularity of this example is that the Conservatives and Liberals won approximately the same share of the vote, but the Liberals won just one seat to the Conservatives’ three seats under FPTP as shown in the second pie chart.
The third pie chart shows what the result would have been had seats been awarded district-wide instead. As one can see by comparing charts 1 and 3, the outcome is a significant improvement over FPTP, and we now have the PCP and the OLP each winning two seats. Readers who wish to see how this result was calculated in detail can consult Appendix A, below.
A welcome feature of this sort of model is how easy it is to make the transition from FPTP to PR, since the only redistricting that is needed is to group existing ridings together, using existing boundaries to form a multi-member district. Under this approach, all incumbents could still run again and would still have a chance of winning.
Further improvements on a basic model of this sort could be achieved in four different ways:
better ballots - instead of voting with just an X, voters might be invited to rank their preferences;
larger districts (DMs) that would give more proportional results and could make the transition easier for some incumbents; and
the addition of top-up seats at a regional level to compensate parties that did not win a proportional share of seats due to limits on district size;
different ways of counting the results; the above simulation uses what is called the largest remainder method, which is easily understood and very easy to simulate; however there are other ways of doing this that systems experts might wish to consider.
Features such as these could be built in from the start, or added at a latter stage. Whatever works best. What matters most is to get this train out of the station.
Better than FPTP in every respect
A system like this, using relatively small multi-member districts, would be better than FPTP in every respect, so should not be particularly controversial.
Small multi-member districts would provide a much improved form of local representation, because voters would have a range of representatives to speak for them.
Similarly, politicians would be even more accountable to voters, because voters would now have more than one candidate from each party to speak for them.
Most voters could now successfully help elect a representative of their choice.
Knowing that their votes would count, the vast majority of voters would no longer feel the need to vote strategically.
Geographical misrepresentation and false majorities would largely disappear as features of our voting system.
Simulated results for Ontario
It is possible to do simulations like the above for Etobicoke—York for other MMDs across the whole of Ontario, the major challenge being how best to group existing ridings that share a geographical sense of identity. Doing my best in this regard, and seeking advice from others, I have come up with 28 MMDs averaging 4.4 seats each, to see what the 2025 election results would have been under such a system.
As MMDs go, an average size of 4.4 is quite small compared to what is used in other countries. For example, this is less than half the size of similar districts in Sweden and Denmark, and comparable to what they have in Ireland. The largest MMD in the spreadsheet is an 8-seater in densely-populated Toronto Centre.
The summary results for the whole of the province are shown at the end of the sheet. As we can see in row 296, the big winner in 2025 would have been the OLP, which would have tripled its seat count from 14 seats to 43. This makes sense, since the OLP was the main party to suffer from the distortions of FPTP in 2025.
To this basic model could be added a small number of “top-up” seats as a way of further enhancing the proportionality of the system. To test this idea and get a sense of its impact, we have grouped the 28 MMDs into 10 regions adding one top-up seat to each. That top-up goes to the largest remaining fraction of a seat that could not otherwise be accommodated as shown in yellow highlighting in the bottom row for each region. Taking Eastern Ontario as an example, one can see that the NDP would have ended up with the top-up in this case.
The summary table at the end of the sheet allows us to compare actual vs. perfectly proportional results. Rows 295-297 show the number of seats each party won under FPTP, the number corresponding to perfect proportionality and the number of seats they would have won with MMDs but no top-ups. Comparing rows 296 and 295, it’s clear that the use of MMDs leads to a considerable improvement over FPTP, while overrepresenting the OLP somewhat, at the expense of the NDP and the Greens.
Adding 10 top-ups to this helps to correct for residual disproportionalities, with the end result that all four of the parties currently represented in the Ontario legislature would enjoy seat shares approximately equal to their vote shares. Those that are underrepresented are New Blue, the Ontario Party and others who collectively win only 1 seat for an independent candidate. Small parties such as these would have to increase their vote shares to win seats.
As noted, this simulation is intended to be illustrative of what might be politically acceptable for most voters and most parties while yielding more proportional results. The overall result, achieved by combining quite small MMDs with a modest number of top-ups, are highly proportional.
An idea worth exploring as a next step would be the use of ranked ballots rather than a single X ballot as part of the MMD model. The best known version of ranked ballots in multi-member districts is the Single Transferable Vote (STV) recommended by the BC citizens’ assembly in 2004.
The central point here is to emphasize the central importance of multi-member districts as a central concept of proportional representation.
If there is agreement on that, the discussion can go in any number of directions to enhance the basic model.
Appendix A - Actual and Simulated Results for the 5-seat district of Etobicoke and York - 2025
This table calculates the number of seats that would have been awarded to each party in the 5-seat district of Etobicoke and York. The candidates elected to fill those seats for each party would be the ones who received the most votes. Each vote counts as a vote both for the party and for a particular candidate from that party.
Share of the Vote | Share of Seats with FPTP | Proportionate # of seats deserved | Seats awarded with rounding | Share of seats with MMD-PR | |
PCP | 43.4% | 60.0% | 2.17 | 2 | 40% |
NDP | 13.7% | 20.0% | 0.69 | 1 | 20% |
LPC | 38.6% | 20.0% | 1.93 | 2 | 40% |
Green | 2.2% | 0.0% | 0.11 | 0 | 0% |
Other | 2.1% | 0.0% | 0.10 | 0 | 0% |
Total | 100.0% | 100.0% | 5 | 5 | 100% |
Comments