RE: dual member plurality system
I have heard it said that the outcome of a dual member plurality system (where both the first and second place candidates are elected) would be close to proportional.
This bears closer scrutiny. I will use Dennis Pilon's book The Politics of Voting, or at least the portions of it I have read so far, as sounding board for the scrutiny.
A dual member plurality system implies a two-seat Multi-Member district. This means representation could be mixed (elected members could belong to two different parties), but does it ensure mixed rep?
Does it ensure proportional? Is it democratic enough?
In a dual member plurality system (where both the first and second place candidates are elected) means that the two candidate with the most votes -- vote tallies relative to the others, the two leading candidates -- are elected.
If our goal is "to create more accurate and inclusive representation in our legislatures with governments that really do reflect a majority of the Canadian electorate" (Pilon, The Politics of Voting, p. xii-xiii), then is a dual member plurality system good enough?
As Pilon first states "there is an important slippage between the theory and practice of the majoritarian vision under plurality: few legislative majority governments in the system ever get a majority of the votes cast... fail in terms of effective representation of our political divisions by over- and under-representing different positions"
"They also violate the notion of majority rule by awarding an electoral minority the majority of legislative voting power." (Pilon, p. 8-9) The majority of voters (perhaps spread over several parties) is sometimes unrepresented -- even getting no seats sometimes -- and the largest single group, often only a minority of the voters, is always over-represented. That is the case under single-member plurality. But is dual-member plurality much different?
Pilon states "social factors contribute to the adoption of a particular voting system and that, once adopted, the voting system conditions the nature of political competition." (p. 13). Thus, how a country's politics is structured when it is adopts a new system is not as important as the need (or not) for it to change - the system can be changed again later if the new system is not seen as proportional enough or a flaw emerges.
Pilon states "voting-system effects are dynamic, shaping not just static results but the very process of contestation from which results emerge."
"Plurality, for instance, [election of the leader in a collection of separate district contests] will reward voting support that is concentrated geographically over voting support that is more spread out", say in the case of Social Credit (in Alberta in 1930s) or Bloc Quebecois.
This is the result of elections and becomes accepted behaviour, as Pilon put it, "because plurality systems tend to discriminate against voting support that is dispersed, some voters may decide not to support political options that they fear are not popular in their geographic area." (p. 14) The results of district elections become regionalized. "First, the voting system affects the incentive structure for both parties and voters. Both make decisions about how to act based on how effectively they think they can get their message across and achieve results through the existing institutional arrangements. Second, the voting system affects the kind and degree of diversity that is represented in our legislatures, from the diversity of party viewpoints to the social diversity embodied in the elected members." That is the case under single-member plurality, but is dual-member plurality much different? Pilon drew a connection between how legislatures operate and how districts should elect. "People deliberate, negotiate, and change their positions based on the actions and reactions of others" (p. 17); Decision making is not just yes or no, a choice between black or white. So representation should be as balanced as possible. Under our present system, the purpose of elections is different from the behaviour of legislatures. But this means un-representation. Is dual-member plurality more representational than SMP? Pilon says voting systems are defined by three things --
- ballot design (X voting or ordinal (figures/numbers)),
- districting (one at-large district or more than one district; single-member or multi-member districts) and
- voting formula. Pilon defines voting formula as "how the aggregated votes will be added up [and used] to determine a winner or winners." Formula are of three types (Pilon, p. 18): 1. Plurality: "the winner must simply gain a plurality of the votes, or more votes than any other individual candidate. In other cases, the winner must receive a majority of the votes cast. 2. In other cases, the winner must receive a majority of the votes cast. [It is implied that the word "winner" can mean "winners" in cases where multi-member districts are used.] 3. Finally, in other cases, the formula "allots representation in proportion to the votes cast for different competitors." I interpret this to mean proportional based on party vote tallies, not vote tallies received by individual candidates. Voting formula looks at how members are elected both as goals and as mechanics. the chart on page 19 of Pilon's book categorizes "major voting systems"
but a different chart could look like this.
Major Voting Systems District elections Election through a single election contest single multi member member district district X voting single voting SMP* SNTV (a form of Multi-member plurality)* multiple voting XXX Limited Voting*** Cumulative Voting*** Multi-member plurality (AKA Block Voting), one form of this is the Dual Member Plurality* Ranked voting single voting AV** STV*** Election through Multiple elections double ballot (runoff) France a second election follows elimination of all but the two leading candidates.**
(This system is also used for party nomination contests to ensure majority support is needed to win. This is used in the cases where Alternative Voting is not used, which is the case nowadays for most of the most important party leadership contests.)
Proportional Party List at-large party based *** Mixed systems Mixed Member Proportional
a form of Mixed Member Proportional
SMP district elections, with top-up seats allocated in compensation to create overall seat representation as per Proportional party-list system *** STV/MMP
a form of Mixed Member Proportional
STV district elections, with top-up seats allocated in compensation to create overall seat representation as per Proportional party-list system [not used anywhere] ***
Parallel (Mixed Member Majority) Japan, Italy 1994-2004) *** SMP district elections with other members allocated based on party standings. (This is a straightforward system proposed for Canada federal elections that Liberal MP Clifford Sifton proposed back in the old days. [info on this can be seen online: https://www.winnipegregionalrealestatenews.com/publications/real-estate-news/1327]
Pilon (p. 24): "The parallel system offers something to smaller parties in that it has a party-list pool of seats." [It just does not give the smaller parties as many top-up seats as the MMP system would, under the same party vote percentages.]
Mixed STV/AV system Another type of mixed system is where election of members in one part of a province is done by STV; in another part by Alternative Voting. Alberta and Manitoba used that kind of mixed system from 1920s to 1950s.
Type of results * plurality ** majoritarian *** mixed semi-proportional or proportional results [SNTV is described as only semi-proportional, although it may produce the same result as STV (which is described as a proportional system) (such as in cases where vote transfers do not have produce any change to First Count front runners). And even a Proportional system is not strictly proportional where parties with less than five percent of the vote (perhaps altogether receiving 20 percent of the vote) receives no representation at all.] ===================================================== Formula are of three types (from Pilon, p. 18): (Plurality -- "the winner must simply gain a plurality of the votes, or more votes than any other individual candidate. In other cases, the winner must receive a majority of the votes cast. Majoritarian -- "In other cases, the winner must receive a majority of the votes cast." [It is implied that the word "winner" can mean "winners" in cases where multi-member districts are used.] Proportional -- "Finally, in still other cases the formula allots representation in proportion to the votes cast for different competitors." I interpret this to mean proportional based on party vote tallies, not vote tallies received by individual candidates. The first two formula elect just one each in each district, based on candidate's vote tallies. Proportional elects mixed rep across the country or province or in a district, based on party vote tallies, reflecting mixed sentiment of the voters. Dual member plurality is/was used commonly. Pilon says it is used in "Canada, U.S. (some local elections)" p. 19. For example, Edmonton used dual member plurality in wards to elect city councillors from 1980 to 2010. By then, Edmonton was not using party labels in city elections so rep in some district elections may have been mixed as regards party affiliation of successful candidates (therefore proportional in regards mixed rep (if party labels had been evident) but with each voter casting two votes in those aldermanic elections a single voting block - the largest single voting group - could have taken both seats, thus producing un-mixed rep and thus un-proportional rep (even if party labels had been used). [STV -- I consider STV to be proportional in its results but its mechanics does not involve party vote tallies so is not Proportional as under the terminology here. (I will discuss STV below.)] The number of votes that each voter can cast is not mentioned but is important both as to goal of the formula and to set the base for the formula. ===================================== Dual member plurality ballot design: X voting, not preferential districting: two members multi-member district (not at-large, unless legislature (impracticably) has just two members) voting formula: plurality, majoritarian or proportional? obviously dual member plurality is plurality-based but is it proportional? Multiple Voting or Single Voting? Proportionality depends on the number of votes that each voter can cast. A two-seat district does not necessarily mean each voter can cast two votes (although a one-seat district does practically mean that each vote would cast just one vote.) But the number of seats a district elects makes big difference in setting the number of votes cast. The number of seats is the difference that produces either our existing system (Single Member Plurality) or Dual Member Plurality -- a form of multi-member plurality (as it is named by Pilon on p. 19). The number of seats a district elects leads to decision on the number of votes that each voter can cast. (Voters practically never cast more votes than the number of seats a district elects but often cast fewer. This is due to parties not running more candidates than the number of seats to be elected. There is no sense in a party running more candidates that each voter can cast, except if party is so heterogeneous that a particular candidate would get votes that another of same party brand would not . We see that when a candidate runs as an" Independent Conservative" or "Independent Liberal" etc. This mostly comes up when there are multiple seats in a district, party discipline is looser. Especially so when the number of parties exceeds the number of seats in a district.) For a district electing multiple members, the number of votes cast by a voter makes the difference between Multi-Member Plurality or Single Non-Transferable Voting (which Pilon names multi-member majority (p. 19), not because it means that each candidate gets a majority of the votes but instead because altogether the elected members reflect the support of a majority of voters). (Single Non-Transferable Voting is a system that combines multi-member districts and Single Voting. It was used in Japan 1947-1993). It is a form of Limited Voting, which was used in Toronto elections in 1880s, and in the U.S. in some municipal elections today). Multi-member plurality, unusually for a X voting ballot design, allows a voter to cast more than one X. (Usually where voters can mark more than one preference, the voter has to use numbers (ordinals) to indicate preference.) Multi-member Plurality, such as DMP, allows variety of results depending on number of parties, number of candidates and importantly the break down of voting groups/voter behaviour. Rep elected can be majority or can be minority. It can be very fair (majority) or very unfair/un-mixed/un-proportional. Ballot design: X voting Districting: multiple members Voting formula: Plurality "If just two candidates are running (for a single empty seat), the winner will have more than 50% of the vote. But if three candidates are running, the winner could succeed with as little as 34% of the vote." The number of candidates running depends on the number of parties and also on the number of votes each voter can cast. There is no sense in a party running more candidates that each voter can cast (as explained above) Under Limited Voting and in SNTV, no one party can take all the seats - this produces mixed representation. in Pilon's Limited Voting case (p. 23) "Party A cannot run three candidates and hope to win all three seats because their 28 votes will not be enough to elect three candidates." that is, where each Party A candidate gets no more than 14 votes each, and where Party B supporters are casting 20 votes (with no more than 10 going to each candidate). Unlike SNTV and Limited Voting, where each voter can cast as many votes as the number of open seats, the result may be un-mixed - the largest party - even one with minority of the voters - can take all the seats. Does dual member plurality produce majority government -- is the majority always represented? If each voter can cast two votes, then the largest group - which could be a minority - could take all the seats -- not proportional or even a mixed result. If each voter can cast just one vote, no one party can take all the seats so result is mixed but is it proportional? is majority represented? Say 50 voters in a two-seat district each party puts forward enough candidates to fill at least all the seats. each voter casts two votes say there are two parties Red 28; Blue 22 the two Red candidates receive 56 votes each and take both seats. majority rep three parties Red 28 blue 14 green 14 the two Red candidates receive 56 votes each and take both seats. majority rep three parties Blue 22 Red 14 green 14 the two Red candidates receive 44 votes each and take both seats. minority rep each voter casts one vote two parties Red 28 Blue 22 each party puts forward just one candidate; each party elects one. elected by acclamation, no contest (no excitement). 100 percent rep. each party puts forward two candidates four in total. Only two can win. Each party elects one unless one party has more than twice the number of votes of the other one. 100 percent rep. based on party, but some voters of each party would not see their candidate elected. Three parties each party puts forward one candidate; two of the parties elects one each. one party gets no rep. Majority representation, as only the least popular candidate (with less than third of vote) is not elected. each party puts forward two candidates, six in total. the two most popular candidates are elected. They will be of different parties unless one party has more than twice the number of votes of the next most popular party. Possible representation varies. based on candidate: DMP could elect two candidates each with little more than 16 percent of the vote. based on party: both elected members could be from one party - the successful party merely getting little more than 32 percent of the vote. the elected members could be from two different parties, so party representation could be easily around 90 percent of the vote, with the least popular party getting only 10 percent of the vote. So in answer to question first put to us, Dual Member Plurality could be proportional (where majority is represented) but could also be un-mixed and un-proportional. Plurality systems -- Multi-Member Plurality or Single-member plurality -- is not scientific representation. It is relative (and not in a good way). STV is example of a scientific electoral system. Under STV, support of a smallest-possible mathematically derived portion of the voters are required to be elected is the most proportional. It ensures that the winners accumulate support from a majority of voters, the litmus test of our democratic representation system, that only minority if un-represented, the opposite of many district contests under SMP elections. Quota Quota under Droop (the very least mathematically possible to almost fill one more than the number of seats to be filled) is the most proportional. But the Hare quota (which is more easily explained) probably works good enough, especially as STV elections, as conducted in Canada history, saw a good number elected through no quota at all but merely through survival of candidates until the field of candidates is thinned through elections and eliminations to the number of remaining open seats - and as the number of votes is decreased by the growing number of exhausted votes, to a point where getting a quota of votes (Hare or even Droop ) is not physically possible. The Australia STV system where votes rank at least a set portion of the candidates does not suffer the high rate of exhausted votes Canada experienced. As Pilon points out SMP is defended on grounds of local representation (as well as other rationales). from The Politics of Voting, p. 6: "Another impact of this different politicization is that it has tended to sever the consideration of institutional values from the concrete political realm, contributing to a mismatch between the laudable goals of the reformers and political reality in terms of organized political forces and their interests. For instance, in 2004 the BC Citizens’ Assembly (BC-CA) decided in favour of the single transferable vote (STV) form of PR, which was opposed by all three of the province’s major parties and most key groups in civil society, over the mixed-member proportional (MMP) form of PR, which was favoured by two of three major parties and many forces in civil society, including most of the public submissions to the BC-CA itself. Even in terms of the deliberation process, the values approach tends to limit critical inquiry precisely because it converts what should be seen as potentially questionable attributes of different systems into morally equivalent value commitments. Thus local representation is characterized as a question of value (i.e., do we like or prefer it) rather than as something that should be investigated and assessed in terms of its real functioning within the political system (i.e., does it really matter in political outcomes)." Pilon makes great point here. local representation -- or fairness or any other goal -- should not be taken as must-have. STV uses multi-member districts. It does not defend "local representation" as currently used in SMP. Note that "local representation" as done by MPs is different from "local representation" as done by MLAs and that "local representation" as done by today's MPs with a riding with 77,000 voters (Edmonton Strathcona) is different from a time when in 1904 fewer than 6000 voters voted to elect the "local" MP but over a massive territory -- the fewer voters were spread across a riding that stretched 70 kilometres north south (at its greatest) and stretched more than the full breadth of the present province of Alberta. So "local representation" meant something different back then than it does today. and it means something different even today in different places in Canada. STV puts "local representation" at a pre-set and understandable level - historically, a district used in STV sometimes was a city, electing five to ten MLAs. at other times it was a third or sixth of a city. But today's SMP sees a riding being one-eighth of the city -- at least, eight ridings have the prefix Edmonton. This one-eighth of Edmonton is different from how Edmonton is divided among 20 provincial districts for the election of MLAs through SMP or the division of the city into 12 wards for election of 12 councillors through SMP. (Using a city as basis of a large MM district prevents gerrymandering as city corporate limits are set by body separate from parties in power and change little over time. STV means less tinkering with boundaries. The fewer larger districts mean that voting blocks are not as split and voters are not barred from voting for whom they want to. the choices/range that STV offers voters (of parties and of candidates within parties) is much broader than under SMP. STV's ranked ballot design means that voters can make fine distinctions with their marked preferences. of course just the first preference will be even consulted, if that choice is elected. and it is possible for none of a voter's marked preferences to be used. But the transferable vote ballot design means that it is possible for a vote to be re-directed to where it will do good - a larger proportion of the votes were used to elect someone under Canada's historic STV elections than in today's SMP elections. STV is more scientific than SMP or even Dual Member Plurality. I see that Pilon does not use term "scientific voting" in his book The Politics of Voting. So the term is not widespread, but I have seen the term "Scientific Suffrage" in a 1898 book entitled Effective Voting. (see https://montopedia.wixsite.com/montopedia/post/effective-voting-pro-rep-in-1898-ontario-book) "Scientific suffrage," under such systems as MMP or STV, produces more certainty, compared to plurality systems.
Thanks for reading.
=============================
Commentaires