Press release No. 2 – Report of the Chief Electoral Officer on the voting system

December 21, 2007

Québec, December 21, 2007 – In his report on the compensatory mixed system, the Chief Electoral Officer of Québec illustrates, with the help of simulations and analyses, the various results that may ensue from the different ways of designing and applying this voting system. What characteristics make it possible to attain the greatest proportionality? How is it possible to promote the representation of small parties? Does the choice of a voting system have an impact on the representation of regions or on the number of women likely to be elected?  These are the types of questions that the Chief Electoral Officer and the associated specialists endeavoured to answer.

"The application of a theoretical model constitutes undoubtedly the most "specialized" and certainly the most "arid" part of our report, emphasized Mr. Marcel Blanchet, Chief Electoral Officer, but it is also the part that allows us to obtain neutral and impartial insight on the impacts of the various paths that the National Assembly could take, if it were to decide to change the voting system."

Compensatory mixed system: compensating for the lack of proportionality of the plurality system

Québec's debate on the voting system stems from the finding that in the past, some general elections have given rise to results that did not necessarily correspond to the votes cast across Québec by electors. This was notably the case when the party elected to govern Québec was not the one that obtained the greatest number of votes at the national level or when a party having obtained a significant percentage of the votes cast wound up with very few elected members. A way of correcting this type of situation would be to adopt a compensatory mixed system, where a portion of the members are elected by the plurality system (as is currently the case in Québec) and where the remaining seats are filled according to the rules of proportional representation "using the list system". These latter seats compensate for the lack of proportionality of the majority system.

For the purposes of his report, and in accordance with the mandate conferred by the government, the Chief Electoral Officer examined three types of compensation:

• compensation at the national level: compensatory seats are attributed according to the total percentage of votes obtained at the national level by each party, by drawing from a national list of candidates, without the seats being distributed regionally;

• national compensation with regional redistribution: it is the votes obtained by a party at the national level which make it possible to determine the number of compensation seats that this party will receive, with these compensation seats then being distributed regionally based on the percentage of votes obtained by the party in each of the regions;

• regional compensation: in this case, the territory is divided into a certain number of regions and it is at this level that the number of compensation seats to which each of the parties will be entitled is calculated, based on the results obtained at the regional level (and not all across Québec).

It should be recalled that of the 27 countries that switched electoral system in recent past, seven switched from a majority system comparable to that of Québec to a mixed-type system between 1993 and 2004.

The "theoretical" Québec of statistical simulations

To evaluate the impact of a compensatory mixed system, the Chief Electoral Officer and his partners made simulations on the basis of theoretical situations, but which realistically could exist here in Québec. The tested models present the following "theoretical" Québec:

• There is a population of 8 million people with 6 million electors;

• There are 127 elected members, with 77 division members elected according to the traditional majority system and 50 "list members" elected according to a compensatory mixed system;

• The compensation seats and the division seats are distributed according to four possible delimitations, namely in 9 regions having an equal population, 9 regions having a population that is not equal, 17 regions or 26 regions;

• There were two election scenarios, during which 10 political parties sought to win the favour of the electorate: 600 candidates ran in these elections, 360 as "division candidates" and 240 as "list candidates".

• These two electoral scenarios produced two "election nights" during which the political parties obtained various results;

• For each of both elections, the compensation lists were drawn up by alternating male and female candidacies, and then without resorting to such an alternation;

• The two scenarios considered two alternate situations, one where electors voted once, and the other where they voted twice. In this latter case, the aim of the first vote was to elect a division candidate and that of the second vote to elect a list candidate; 

• Different representation thresholds were also tested during the two elections. A party had to obtain 2%, 3% or 5% of the votes to have access to the compensation seats and to elect one or more list members;

• Three calculation methods were also applied during the two fictional polls for the allocation of compensation seats;

• Both electoral scenarios considered three types of compensation.

Once the parameters of the model had been set, the Institut de la statistique du Québec proceeded with the various simulations.

A few results…

The objective of the statistical simulations was to see what type of effects could be produced by a compensatory mixed poll, depending on the choices made when this poll had been configured. Indeed, the simulations show that the various options produce different situations:

• National compensation and national compensation with regional redistribution produce more proportional results (results that better correspond to the votes cast) than does regional compensation. This observation is in line with the fact that a small number of compensation regions produces more proportional results. National compensation, owing to the fact that it has only one electoral region, is therefore the one which offers the most proportional results.

• As the number of regions for compensation purposes increases, small parties become ever less favoured. The 26-region delimitation scenario is the one that is the least favourable to small parties. Conversely, the "nine-region" scenarios are more advantageous for small political parties, while the map having only one region is the one that most favours their representation although in these two cases, we note few differences.

• The way of calculating the attribution of compensation seats has an impact on the proportionality of the results. Of the three mathematical scenarios tested by statisticians, the Hare method (described on page 18 of the report) is the one that produces the most proportional results and that most favours the representation of small parties.

• As the representation threshold goes up, the results become less proportional. The distortion of the voting results is three times greater with a threshold of 5% than it is with a threshold of 2%. Moreover, a lower threshold makes it easier for small parties to have access to Parliament, since "the bar is lower" for having a first candidate elected.

• A two-vote poll favours small parties. Indeed, it is not unusual for electors to first choose a division candidate belonging to a political party, with their second vote going to a list candidate of a different party. Moreover, the election of list candidates from small parties, within the context of a two-vote system, makes the general result of the poll more proportional.

• The alternation of male and female candidacies on the lists of parties has a direct effect on the representation of women, an effect that is accentuated if parties are required to put a female candidate at the top of one out of every two lists.

The report of the Chief Electoral Officer on the characteristics of a compensatory mixed member voting system is available on the website of the Chief Electoral Officer at the following address: www.electionsquebec.qc.ca.

Categories : Provincial, Election system, DGE