Quebec: the electoral distortion of 2026

Essi Parent · October 15, 2026

The calculations in this post can be reproduced on tangent/note.

CAQ: 0. Not 0% of the vote, but 0 candidates elected. No conspiracy here, nothing but the expected result of the governments are elected in Quebec: first past the post, one member per riding. Here is how elections work in Quebec.

  1. you vote in your riding, with an x beside the name of the person you choose
  2. that’s it.

Your vote doesn’t go beyond your riding (well, almost; we will come back to that). Most candidates are tied to a party, and the party with the most members forms the government. A party is a private organization authorized by the Chief Electoral Officer of Quebec. You form a party, there are rules on how you raise money and how you spend it. Once the results are official, the lieutenant governor, the King’s representative, invites the person who can win the confidence of the National Assembly to form the government. That is normally the leader of the party with the most members elected (the winning caucus). The party’s members chose that person, not the voters. They stay premier as long as the Assembly keeps its confidence in them (with a minority government, they need the support of other parties), and pick their ministers among the members (in practice, among their gang). Normally, we have a clear idea of who will be chosen. But no, you do not choose who will be premier. You only choose the person who will represent your riding: your vote stops there.

A party can therefore get 13% of the valid votes without electing a single member: CAQ 0 with 13%. It can also squeak through just about everywhere, get 28% of the valid vote, and form the government with 59 members out of the 127 available positions (the seats): that is the PQ in 2026. It can also come close everywhere and form a large majority of 72% of the seats with 41% of the vote: that was the CAQ in 2022.

Not everyone has the right to vote, not everyone who has it votes, and (a result of the electoral system) those who vote are not necessarily well represented.

Chart sources: population on 1 January 2026, Institut de la statistique du Québec; registered electors and votes, Élections Québec.

Oblivion voting

In each riding, only the votes for the winner elect someone. The other votes have no effect on the makeup of the National Assembly. In 2026, that is the case for 58% of the votes. None of the CAQ’s 572,000 votes elected a member.

A party loses all the more votes as it comes second or third in many ridings. Three quarters of Québec solidaire’s votes elected no one, against a little over a third of the PQ’s.

But compared to…

Political scientists like to have indicators of vote distortion. The Gallagher index (no, not named after Brendan), , measures the gap between vote shares and seat shares. The higher the index, the larger the gap. Wikipedia lists Gallagher indices for much of the world. Where does Quebec stand?

Countries that vote by proportional representation have low indices, but a majoritarian system is not necessarily very distorted: France (two rounds) and Canada in 2025 stay under 10. In Germany, a party gets seats in the Bundestag only if it collects at least 5% of the votes. In 2025, about 14% of the votes went to parties under that threshold, hence a higher index. For the United States House of Representatives, most states elect one person per district, as in Quebec. Almost all votes go to two parties, and seats roughly follow votes. Quebec lives in the zone of the most distorted majoritarian systems. The CAQ: 0 effect is not an exceptional case, though. In 2022, the CPQ also got about 13% of the votes, and no seat. And in the 2022 election, the CAQ benefited from the distortion even more than the PQ did in 2026.

Fiction: the 2026 election under proportional representation

Under strict proportional representation, a party gets a share of the seats equal to its share of the vote (rounded, of course).1 The proportional distribution gives the 127 seats to the five main parties according to their 2026 votes, with no threshold (Sainte-Laguë method). The other parties, about 1% of the votes, get no seat. A 5% threshold would change nothing in 2026: the five main parties all exceed 12%. That is not, however, what the Mouvement démocratie nouvelle (MDN) proposes, a Quebec organization that has been calling for electoral reform since 1999, and whose current proposal is a mixed-member proportional system with regional lists, the Pro-Mix model (no, not the potting soil brand). In short, the MDN proposes electing members in a smaller number of ridings, with proportionality brought in by lists of candidates in the regions.

Under pure proportional representation, the PQ would have 36 seats rather than 59, and 48 under Pro-Mix (which would have 129 seats, hence a majority threshold of 65 seats). A majority coalition government, 48 (PQ) + 23 (CPQ) = 71 (with the QLP in opposition), would have been possible. Today, the PQ could join forces with the CPQ or QS (or count on 5 floor-crossers) to reach a majority.

Does your vote change anything?

The electoral system also fundamentally changes the weight of a vote. Today, a vote changes a riding’s result only if it breaks a tie. Since 1970, the closest race was Hull in 1976, won by 2 votes. In 2026, Vachon was won by 10 votes.

Over nearly 2,000 riding races since 1970, 25 were won by fewer than 100 votes. A perfect tie in a riding happens about once in 7,800 riding races.2 From 1970 to 2026, 25 ridings out of 1,946 were won by fewer than 100 votes. If these margins are spread evenly between 0 and 100 votes, a perfect tie happens about once in 7,800.

The Votovision forecast for 2030 allows the same calculation for each riding. The chance that one vote decides the winner depends on the forecast margin between the top two and on the number of electors.3 The chance that one vote decides the winner in 2030 is computed from the forecast margin between the top two and the uncertainty on that margin, derived from the probability of winning. When the forecast is nearly certain, the typical uncertainty of the other ridings is used.

In the riding of Ungava, the closest and one of the least populous, that chance is about 1 in 4,100. In the median riding, it is about 1 in 26,000. In 7 ridings, it is under 1 in 10 million, the same order of magnitude as the Lotto 6/49 jackpot (about 1 in 14 million).

Under proportional representation, every vote counts, but a vote changes the distribution of seats about once in 34,000,4 Under proportional representation, 4,282,551 valid votes for 127 seats give one seat per about 34,000 votes. One vote therefore changes one seat about once in 34,000. a little less than today’s median. That chance would be the same everywhere, though. Today, a vote weighs much more in a close riding than in a safe one.

Why vote anyway

For the same reason that danger is not risk. A shark attack is dangerous. But it is so unlikely (especially if you are on a bus right now) that the risk is practically nil. Risk is the danger × the probability that it happens. What matters is the size of the consequence relative to its probability. A small probability that something enormous happens: that is no longer a shark attack on a bus, and it is serious.

Whether it is one chance in 26,000 or in 34,000 of making the difference, it is negligible. But the stakes affect all of Quebec (budgets, laws, emergency measures), all spread over four years of government (in the case of a majority government). The more electors there are, the smaller the probability of being decisive… but the greater the stakes. The two effects combined make voting a rational democratic act for anyone who cares about others and about posterity.

Your vote also has a definite effect, even if your choice loses. Every year, the State pays the parties about $2 per registered elector, nearly $13 million, shared according to their share of the votes at the last election. The envelope is fixed: your vote creates no money, it moves about $3 a year to your party ($12 over four years), at the expense of the others. A person can give at most $100 a year to a party.

The CAQ has no member, but its votes bring it about $1,700,000 a year. A party’s score also counts beyond money: it affects its media coverage and its ability to recruit candidates.

Strategic voting means voting to block a party you want to see lose. It changes the result only if your riding is close and your vote breaks a tie, the same tiny bet as an ordinary vote. On the other hand, your preferred party loses about $3 a year for sure, and a bit of its score. Under proportional representation, strategic voting loses much of its point, since almost no vote is wasted. It does not disappear entirely: with a threshold, some lend their vote to an ally at risk of falling under the bar, as in Germany.

Why the electoral system does not change

According to its supporters, bringing in more proportionality would allow better representation of political currents, more territorial fairness and, since the parties would form more of a continuous spectrum than sharply drawn clans (in theory at least), a better spirit of cooperation among elected members. Why, then, does the electoral system seem set in concrete?

Changing the electoral system requires a law passed by the National Assembly. But the governing party is almost always the one the electoral system favoured. Proportional representation was in the Parti québécois’s platform from its beginnings, and René Lévesque never managed to get it adopted. Fifty years later, in 2018, the CAQ committed, with the PQ, Québec solidaire and the Green Party, to introducing a share of proportionality. Once in power, it tabled Bill 39 in 2019, which was dropped before the 2022 election.

Defenders of the current system put forward several arguments, to which supporters of reform respond. Today, each riding has its member, whom the voters judge directly, and sparsely populated regions keep their members: that would also be the case with the MDN’s proposal, even if ridings would be larger. It is said that the current system produces stable majority governments, but Quebec had minority governments in 2007, 2012 and 2026. A mixed system would be too complicated, but it is in use in Germany, New Zealand and Scotland.

Finally, it is said that proportional representation would open the door to extremist parties. The MDN’s model sets a 4% threshold overall, and since compensation is done by region, in practice it takes much more to win a seat outside the more populous regions. But, and this matters here, a threshold protects against fragmentation, not against extremism. The AfD, a German far-right party, passed 5% and now sits in the Bundestag. In Quebec, the current system did not prevent the election of people who had made anti-vaccine claims or posted photos of supposed chemtrails, a theory with no scientific basis.

At the Radio-Canada leaders’ debate, the leader of the Parti québécois, now premier, committed to reforming the electoral system, without giving details.