Here is a long essay on PR and electoral reform. A highlight is a list of PR systems that would easily fit in with the political system used in Canada, with specific advantages of each. ====== First off, I want to say that I agree with the statement - The essential point of any reform of FPTP should be to go to multi-seat districts.
Changing to Instant-Runoff Voting (AKA Alternative Voting) does not result in proportional representation, does not address the lack of fairness and dis-proportionality produced by First Past The Post. Unfortunately, many people use the term "ranked voting" to mean IRV, although ranked votes are used in proportional STV-PR and IRV does not give the fairness that STV-PR does. People may do this to play on the good feeling of STV-PR - although unfortunately relatively few know of STV-PR. It is strange politics that makes Prime Minister Trudeau push for what he calls "ranked voting." He means IRV, ranked voting in single-seat district. IRV elects mostly the same as would be elected under FPTP. And I doubt there are many voters who want the work of deciding how to rank candidates and the work of marking back-up preferences - only to see just one elected in the district anyway. so IRV means more work with (likely) the same result as FPTP. Without multi-seat districts, the result of IRV cannot be proportional (within the district). IRV results will likely, except only in a few cases of individual members, be different from FPTP result. That was how it worked when Alberta and Manitoba used IRV to fill most of their seats from 1920s to 1950s - in all but one or two or just a handful of seats, the same person was elected as who would have been elected under FPTP. That was how it worked when London Ontario used IRV to fill its city council seats in 2018 - in all seats the same person was elected who would have been elected under FPTP. and many other historic cases as well. I expect ranking (marking first preference and back-up preferences) is different under PR-STV than single-winner IRV. In STV-PR you can give all your preferences to members of one party (if your preferred party runs multiple candidates). In single-winner IRV, you must judge candidates of different parties, if you mark multiple preferences. And under PR-STV even if no voters mark back-up preferences, the result will (in all but a few cases in all of history) be different than the result under single-winner FPTP or IRV in single-member districts covering the same territory, That is to say, across a city using FPTP, it is highly unlikely that the same members might be elected as under STV-PR in a city -wide district. And even if it did happen, under such a FPTP or IRV system, there would not be a direct relationship between each voter in the city and each of the elected members as there would be in STV with a city-wide MM district. Under STV-PR, just the act of single voting in multi-member district creates different (more-proportional) result than FPTP. even with no back-up preferences consulted and no votes transferred. The term Instant-runoff voting is nice and clear IMHO. I think it is clear it means single-winner. I think "runoff" indicates narrowing of field to one member. There are various types of runoff elections. Exhaustive ballot is one type. Wikipedia article "Exhaustive runoff" says "Exhaustive runoff is a voting system used to elect a single winner. Under the exhaustive ballot the elector casts a single vote for their chosen candidate. However, if no candidate is supported by an overall majority of votes then the candidate with the fewest votes is eliminated and a further round of voting occurs. This process is repeated for as many rounds as necessary until one candidate has a majority." Instant-Runoff Voting (AKA ranked voting) is a version of this which has just one round of voting. In cases where no candidate has majority of the votes on the first round, elimination(s) take place and votes are transferred as per back-up preferences pre-set by voters. It is called "instant" because there is only one round of voting but it is not at all instant -- sometimes taking hours and the transfer of hundreds or thousands (or more) votes from candidate to candidate before someone takes the seat, with the leader in the first round usually being elected in the end anyway. But sometimes IRV contests involve no transfers at all - the winner is declared on the first round) - The terms "ranked voting" or preferential voting or even STV are less clear as both IRV and STV-PR use ranked votes, preferential votes and single transferable votes, which all usually mean the same (technically it is possible to cast multiple transferable votes in a system that uses ranked votes). But I think it is correct to say that Single Transferable Votes (STV) with capital letters can in most cases be understood to mean STV-PR (or PR-STV if you prefer). And PR is commonly understood to mean PR-STV (or MMP). But it is common, as is sometimes pointed out, that the term PR is used to mean IRV, just as the term "STV" is sometimes used to mean IRV, in some cases. Where people are accustomed to using ranked votes in STV-PR and thnk of STV-0PR as the normal form of PR - that is people in Ireland, Malta or in Canada in the old days - they naturally use the term PR or STV to mean any election where ranked votes are used. I am good with using STV to mean PR-STV, and IRV or Alternative Voting to mean single-winner ranked voting. The real question then becomes which type should Canada adopt. The range of options would be something like the following. In any Canadian federal system, constitutionally it seems no MM districts would cross provincial borders. (But I see the 1979 Task Force on Canadian Unity does not indicate this. (https://primarydocuments.ca/task-force-on-canadian-unity-a-future-together/) Electoral systems can be organized into two groups - those with adjustment seats (different forms of MMP) and those without. The type of districts used (single-seat or multi-seat or a mixture), and the type of votes used (X votes or ranked votes) are other descriptors. Another distinguishing feature is the break-down of which, or how many, of the districts will use the particular combination of type of district or type of votes used, selected. Thirteen different electoral systems are listed here. They are identified by letters, A to M. A (MMP) is much discussed today. J (STV-PR), L (STV/FPTP) and M (STV/AV) are the only systems on the list that were ever used in past elections in Canada. Mixed Member Proportional (A is the most common variant), plus two other systems invented recently - E (Rural-Urban PR) and H (Dual-Member Proportional) - are the only systems that have names. The rest are known by abbreviations derived from the electoral system(s) used. PR systems with adjustment seats (adjustment seats usually filled as per party lists according to party proportions of the votes. Adjustment seats may or may not preserve local representation. Proportionality produced by adjustment seats may be more exact, at wider level of measurement, than STV- or SNTV-district elections.
A Mixed Member (MMP) with FPTP elections in single-seat districts, plus top-up adjustment seats (what Canadians generally think of when they say MMP) (New Zealand uses MMP where adjustment seats allocated at national level. Scotland uses MMP where adjustment seats are allocated at regional level, where it is called the Additional Member system.) Constitutionally Canada would have to allocate adjustment seats at provincial or sub-provincial regional level. Adjustment seats would be based on province-wide stats in some cases. Proportionality would be more exact, at wider level of measurement, than STV- or SNTV-district elections.) Generally dis-proportional results of FPTP elections means large number of top-up seats are needed to achieve certain level of proportionality and thus districts have to be large to cover the area. Question would be whether each voter casts one vote that is used for both district and to set adjustment seats, or if each voter casts separate votes for each. X voting in district elections Advantages include using the same type of voting in all districts. Single-member districts allow smaller districts than multi-seat districts, other things being equal.) B Mixed Member (MMP) with mixture of multi-member districts using SNTV and single-member districts where FPTP is used, plus top-up adjustment seats X voting in all districts. Advantages include using the same type of voting in all districts. The single-member districts, where used, allow smaller districts than multi-seat districts, other things being equal. Mixed representation would be likely elected where SNTV is used, so lesser number of top-up seats needed to achieve a certain level of proportionality. MM districts larger than single-seat districts, other things being equal. Flexibility of district size - single-seat districts in rural areas and multi-seat districts in cities is natural choice.) C Mixed Member (MMP) with mixture of multi-member districts using STV and single-member districts where FPTP is used, plus top-up adjustment seats. (Advantages include mixed representation elected where STV is used, so lesser number of top-up seats needed to achieve a certain level of proportionality. Very high proportion of votes used to elect members in MM districts. The single-member districts, where used, allow smaller districts than multi-seat districts, other things being equal. Flexibility of district size - single-seat districts in rural areas and multi-seat districts in cities is natural choice. Does not force ranked voting in single-member districts.) D MMP with mixture of multi-member districts using STV and single-member districts where IRV is used, plus adjustment seats (Ranked voting in all districts. Advantages include using the same type of voting in all districts. STV producing fairness at city/local level so not as many adjustment seats would be required to achieve certain level of proportionality. Very high proportion of votes used to elect members in MM districts. Majority of votes in all districts used to elect member(s). The single-member districts, where used, allow smaller districts than multi-seat districts, other things being equal. Flexibility of district size - single-seat districts in rural areas and multi-seat districts in cities is natural choice. E (Rural-Urban PR) MMP with mixture of multi-member districts using STV and single-member districts where FPTP is used, plus adjustment seats added just in area(s) where FPTP is used (FVC uses term Rural-Urban PR for this system, where STV is used in cities, and FPTP and adjustment seats in rural districts.) Somewhat similar to MMP used in Scotland, where adjustment seats are allocated at regional level, where it is called the Additional Member system.) Federal level: Constitutionally Canada would have to allocate adjustment seats under a basis that does not cross provincial boundaries.
Generally dis-proportional results of FPTP elections means large number of top-up seats are needed to achieve certain level of proportionality and thus districts have to be large to cover the area. But having STV in some districts means fewer adjustment seats required than under all-single-seat-district system. Question would be whether each voter casts one vote that is used for both district and to set adjustment seats, or if each voter casts separate votes for each. Very high proportion of votes used to elect members in MM districts. Single-member districts allow smaller districts than multi-seat districts, other things being equal. Does not force ranked voting in single-member districts. Flexibility of district size - single-seat districts in rural areas and multi-seat districts in cities is natural choice.) F MMP with multi-member districts using SNTV, plus adjustment seats (X voting in all districts. Advantages include using the same type of voting in all districts. SNTV likely producing mixed representation at city/local level so so likely not as many adjustment seats would be required to achieve certain level of proportionality. (Disadvantage: MM districts larger than single-seat districts, other things being equal.) G MMP with multi-member districts using STV, plus adjustment seats (Ranked voting in all districts. Advantages include using the same type of voting in all districts. STV producing fairness at city/local level so only a few adjustment seats would achieve high level of proportionality. Very high proportion of votes used to elect members in MM districts. (Disadvantage: MM districts larger than single-seat districts, other things being equal.) H (Dual-Member Proportional) Two-seat districts where one seat is filled through FPTP, and the second seat in each district filled as an adjustment seat based on multi-district-regional party vote tallies. X voting. Advantages include using the same type of voting in all districts. System said to have only a single tier of members (with each member having only a single district to represent) although second-seat members could be elected based on support outside the district where they ran. (Could be related to SMDPR system put forward by Denis Falvey <denisfalvey@eastlink.ca>)
Systems with no adjustment seats I Multi-member district elections using SNTV, with no adjustment seats (X voting in each district. Mixed representation likely elected in district elections, so lesser number of top-up seats needed to achieve a certain level of proportionality than if FPTP used.) (Disadvantage: MM districts larger than single-seat districts, other things being equal. Without transferable votes and adjustment seats, potential for dis-proportionality.) J Multi-member district elections using STV, with no adjustment seats (Ranked voting in each district(s). Advantages include using the same type of voting in all districts. Mixed roughly-proportional representation produced in each MM district where more than three seats. Very high proportion of votes used to elect members in MM districts.) (Disadvantage: MM districts larger than single-seat districts, other things being equal.) K Mixture of Multi-member district elections using SNTV and FPTP single-seat district elections, with no adjustment seats. (X voting in each district. Does not force ranked voting. Advantages include using the same type of voting in all districts. Mixed representation produced in each MM district where more than three seats. Without transferable votes, potential for accidental dis-proportionality. Flexibility of district size - single-seat districts in rural areas and multi-seat districts in cities is natural choice.) L Mixture of Multi-member district elections using STV and FPTP single-seat district elections, with no adjustment seats. Mixed representation produced in each MM district where more than three seats. Very high proportion of votes used to elect members in MM districts. Flexibility of district size - single-seat districts in rural areas and multi-seat districts in cities is natural choice. Does not force ranked voting in single-member districts.) M Mixture of Multi-member district elections using STV and IRV single-seat district elections, with no adjustment seats. (ranked voting in each district. Mixed representation produced in each MM district where more than three seats. Majority of votes used to elect winner(s) in each district. Very high proportion of votes used to elect members in MM districts. Flexibility of district size - single-seat districts in rural areas and multi-seat districts in cities is natural choice.)
=================== A is PR system commonly envisioned for Canada. Majority of PEI voters voted for A in 2016. So far in history of Canada, no MMP system has been used at any level. J (STV) (or a form thereof) is the BC-STV system for which a majority of BC voters voted in 2005. J (STV) was used in Medicine Hat's STV city election in 1928. All city councillors were elected in one at-large district using STV. The mayor was elected by city council later. L (STV/FPTP) was used in Manitoba in 1920. STV was used to elect Winnipeg MLAs. M (STV/AV) was used in Alberta in eight elections from 1924 to 1955, M (STV/AV) was used in eight elections in Manitoba from 1922 to 1953. STV was used to elect MLAs to represent Edmonton, Calgary and Winnipeg, and Medicine Hat and St. Boniface briefly. M (STV/AV) was used in many of the STV city elections in the 19 Canadian cities that used STV in their elections in the 1917-1971 period. STV in one at-large district elected the city councillors and AV in one at-large district elected the mayor. In the historic use of L and M in Alberta and Manitoba, in every district contest except one, mixed representation was produced in each district where STV was used. A (MMP), E (RUPR) and H (Dual-Member PR) were the three alternative options for BC voters in the 2018 BC referendum. Retention of the FPTP system received a majority of votes. A (MMP) and E (RUPR) are favourites of many electoral reformers. As well, systems using STV partially or fully have many strong proponents. The reasons for this include that - it is candidate based and is the main system of PR that does not use party lists; - that it is only PR system used in Canada to any great extent (Toronto's brief use of Limited Voting is STV's only contender.), - has been used successfully in countries close to Canada for more than a hundred years - Ireland, Malta and Australia. STV systems listed above are: with adjustment seats -- B, C, D, E, and G with no adjustment seats -- J, L and M. SNTV, a system never used in Canada history, is similar to STV. SNTV is used in several countries and with adjustment seats would yield proportional results about as proportional as some SNTV and STV elections (sometimes SNTV elects the same as STV) and would likely be as proportional as MMP systems. My ideas to make up a model electorate to test how results of the 13 systems would compare. 42 districts: 24 rural districts two cities of nine seats each MMP systems mean the number of districts has to be reduced. Although the reduction could moderated by increasing the overall number of members compared to the number used previously under FPTP. In the case of New Zealand, the total number of members was increased by 20 percent. The 1979 Task Force on Canadian Unity called for expansion of total members by 60 seats at the time of the change to MMP, about the same proportional increase as occurred in New Zealand.) In the model, if we don't increase the total number of members, systems that have a large number of adjustment seats could mean only 21 districts, down from 42, each district being twice as large as the ones previously used under FPTP A medium number of adjustment seats (14) could mean 28 districts (two districts where there were three under FPTP). A small number of adjustment seats (8) could mean 34 rural districts (three districts where there had been four under FPTP) and eight districts in each city instead of nine. With a fairer system in place for district elections, a certain level of proportionality could be produced through a small number of adjustment seats. With FPTP system in use in district elections, about 21 adjustment seats would be needed to achieve a certain level of proportionality.
With STV or SNTV or regional MMP, as few as 8 (20 percent of overall seats) might be enough, certainly 20 percent would be enough to have much more fair and proportional representation than we have now.
As a trade-off, though, where there are much fewer districts covering the same territory (due to many adjustment seats), there is more adjustment needed to be made to achieve a certain level of proportionality due to the larger districts. In larger districts more voters are ignored (usually) in each district if FPTP is used.
New Zealand has had MMP since 1996.
New Zealand now has 49 adjustment seats and 71 district seats. 120 in total
the 1993 election (the last pure-FPTP election) NZ had 99 seats -- 95 geographical district and four Maori seats (each covering a quarter of the country)
in changing to MMP, districts increased in size, 71 covering what had been covered by 95 seats before, Maori distrcits were increased in number to 5.
and adjustment seats were established, to the number of 65, about half of the total seats.
The larger districts produced less proportional representation but the new adjustment seats addressed the dis-proportionality -- at least in the case of parties who had more than five percent of the vote (the threshold).
1993 FPTP
two parties took 95 percent of the seats with 70 percent of the votes
one party took majority of seats with only 35 percent of the votes
four parties in total took all the seats.
1996 MMP (65 district seats, 55 list seats)
overall results, after adjustment seats
1996 two parties took 68 percent of the seats with 62 percent of the votes
no party took majority of seats (no false-majority)
five parties in total took all the seats.
in seat contests:
two parties took 56 district seats (86 percent of the seats) with 65 percent of the votes
(so less proportional results than when districts were more numerous and smaller in size)
one party took majority of seats with only 34 percent of the votes
six parties in total took all the seats (more parties represented than under FPTP)
1996 adjustment seats
(large number of adjustment seats and adjustment seats only going to parties who had more than 5 percent of the vote meant no party took more district seats than it was due proportionally overall. There was no overhang.)
every party with more than 5 percent of the vote received some adjustment seats
more than enough adjustment seats to give the smallest eligible party enough seats to have proportional rep.
due to 5 p.c. threshold, Christian Coalition with 4.33 percent of the party votes got no seats. (it won no district seats and was awarded no adjustment seats). due to this, other parties got more seats than they were due proportionally.
ACT, the least-popular party to win seats, with 6.1 percent of the party vote got 8 seats, more than the 7.3 seats it was actually due.
New Zealand First captured all five Maori seats plus one other.
with 13 percent of the party vote, it was awarded 11 adjustment seats to bring its total past the 15.6 seats it was due (13 percent of the 120 seats).
So I will work out these 13 models to show how more adjustment seats create less proportional results in district elections and thus how even MMP systems are improved if more fair system (STV, or SNTV or regional MMP/RUPR) is used in district elections.
Adding members when switching to PR was identified in Canada as the way to go, years before New Zealand did it.
Back in 1979 the Task Force on Canada Unity reported: [...] The simple fact is that our elections produce a distorted image of the country, making provinces appear more unanimous in their support of one federal party or another than they really are. Quebec, for instance, has for years given an overwhelming proportion of its Commons seats to Liberals: in the 1974 federal election, that party won 81 per cent of the seats though it got only 54 per cent of the popular vote. In the same election the Progressive Conservatives gained the second highest popular support while, with less total support across the province, the Social Credit Party won four times as many seats. In the elections of 1972 and 1974 two Alberta voters out of five favoured other parties but every elected member was a Progressive Conservative. Nor are these examples exceptional. Under our current electoral system, which gives the leading party in popular votes a disproportionate share of parliamentary seats in a province, the regional concentration in the representation of political parties is sharply accentuated. This makes it more difficult for a party’s representation in the House of Commons to be broadly representative of all the major regions. In a country as diverse as Canada, this sort of situation leads to a sense of alienation and exclusion from power. Westerners in particular increasingly resent a disproportionate number of Quebec members in a Liberal caucus which has very few of their own. If there were more Quebec members in the Progressive Conservative caucus representing more accurately the popular vote in that province, that caucus would be in a better position to reflect and understand the concerns of Quebecers. To correct the existing situation with its corrosive effect on Canadian unity, we propose a major change in the electoral system. We would continue the current simple-majority single-member constituency system because of the direct links it establishes between the voter and his MP, but would add to it a degree of proportional representation. We would increase the overall number of Commons seats by about 60 and these additional seats would be awarded to candidates from ranked lists announced by the parties before the election, seats being awarded to parties on the basis of percentages of the popular vote. [bold added later] We have opted for these additional seats being assigned to those on party lists announced before an election rather than to candidates who have run and placed second in individual constituencies in order to avoid any connotation that these additional members are second-class representatives and to encourage parties to use this means to attract candidates who might otherwise be difficult to entice into politics. We have examined in some detail various ways in which this could be done, although we would prefer to leave the final choice in this matter to Parliament in consultation with experts. One method would base the allocation of the 60 seats on the basis of the vote in each province won by a party, the additional seats being awarded to those parties which otherwise would be proportionately under-represented. Another method would be to allocate the 60 seats on the percentage of the country-wide vote received by each party and apply what is known as the d’Hondt formula for allocating seats provincially among parties. The procedure for allocating seats in the second method is more complex and difficult for electors to understand, but reduces the likelihood of minority governments resulting. [PR-produced Minority governments] Canadians have traditionally expressed a fear that a system of proportional representation would produce frequent minority governments and hence weak and unstable cabinets. An analysis of how our proposal might have worked in each federal election since 1945 suggests that the combined electoral system we are proposing, with about 280 single-member constituencies plus 60 additional seats to make representation more proportionate, would not only have produced a more broadly based representation within each party in the Commons but would not have significantly increased the incidence of minority governments over that period. [...] (https://primarydocuments.ca/task-force-on-canadian-unity-a-future-together/) ================================================== Would PR produce many more minority governments than FPTP? There have been incidences of false majorities since 1945 but there have also been many minority governments produced by FPTP. My analysis of federal elections tells me that PR would have given us minority governments instead of majority governments elected under FPTP, in a few cases: 1949, 1953, 1968, 1974, 1988, 1993, 1997, 2000, 2011 and 2015.
(see my blog: https://montopedia.wixsite.com/montopedia/post/canada-s-minority-rule-federal-fptp-produces-odd-results-and-always-has) The others were actual majority governments elected with support of majority of votes (1958 and 1984) or in the case of all the others, minority governments elected with only a minority of the vote. Those ten instances listed were cases where a party was elected to government since 1945 who only had support of a minority of the voters. Obviously under PR, the party would not have won majority power (But PR often is not exactly proportional so it is possible that still the party would have won majority of the seats even under PR.) Under PR, if not elected to majority of seats, the leading party would have had to get co-operation of one or more other parties to stay in power, just as minority governments do in most of the other cases. Lester Pearson's minority governments of the 1960s are generally considered the most effective and progressive of any governments in our history. So although it is simpler to have a majority government and it is democratic to have majority government elected by a majority of voters, we cannot always count on any one party taking the votes of a majority of voters. In most cases a minority government elected by a minority of voters is the most democratic result we can hope for. The minority government to stay in power would have to have support of a majority of members, and those members in a PR system would reflect the votes cast by a majority of voters. Under FPTP, majority government merely means that a party had more votes than any other party in a majority of districts - it could easily be fewer than a majority of votes overall, it cold also receive a minority of votes in each district it wins.
As few as 17 percent has proven to be enough to win a seat in a FPTP contest. If that was common, a party winning 17 percent of the vote in a scant majority of districts would have total power. A scant 9 percent of the vote would be enough for a majority government in the worst case. =====================================================
Comments