top of page
Tom Monto

PR-STV defended and explained

Updated: Jan 11, 2023

I have tried to talk more about the importance of multi-member districts than about preferential voting.


I have been saying that MM districts are necessary and integral to PR and that

transfers as conducted under STV, made possible by preferential voting, always make little or no difference to the front runners in the first count of the PR-STV vote counts.


These front runners would have been elected with no preferential voting at all, and most or all of them are elected even after vote transfers are conducted under PR-STV so vote transfers actually made small or no effect on those elected under PR-STV.


Single voting in MM districts is the backbone of STV-PR and alone usually produces mixed, balanced and party-proportional results in the front runners in the first count. PR-STV always elects most or all of these first-count front runners.


Some might say that "Multi-member preferential voting" (PR-STV) is a very good system, but it gets bogged down in complexity after about 5 members. But I don't see it.


Historically, there was no "bogged down" happening in Winnipeg's 10-member district from 1920 to 1945, nor in Edmonton and Calgary's 6-seat districts in 1930s, nor in their 6- or 7-seat districts in the 1950s. (The drop in the 1940s to 5-seat districts was done as part of an across-the-board reduction in seats in the province overall and not due to PR-STV being too complex with 6 seats.)

In these cities, PR-STV was not cancelled because of problems conducting PR-STV vote transfers, but for political reason alone.


Actually, the fact that PR-STV districts in Malta or Ireland having five or less seats each is not likely due to complexity of STV but instead likely due to desire to keep district sizes small or due to city-wide districts only needing five or less seats.


Both Malta and Ireland's STV systems were established a hundred years when there were fewer people, when technology and transportation were more primitive than today's computers and cars, and when STV itself was an unknown quantity, never being used very much even at the city level when those two countries were the first in the world to adopt PR-STV for national elections. so likely the relatively small District Magnitude was more due to natural timidity than to the actual limits of PR-STV.


Among the largest districts in the world to use PR-STV in 1910s and 1920s were the STV elections held in Canada:


Winnipeg's first provincial election to use PR-STV (1920) elected 10 MLAs in one contest.

Calgary's first city election to use PR-STV (1917) elected nine city councillors in one contest.

(the count took days but it never bogged down. The officials kept doggedly at it in long days and produced fair and accurate results.

Part of the length of the process was that the transfers were conducted using Gregory method for surplus transfers where extra effort was made to prevent any possibility of random-ness to creep in.


By contrast, provincial PR-STV election counts went quicker than that. Transfers of surplus votes under the Alberta provincial elections were conducted with reference only to the next usable back-up preference. This made the vote count simpler but meant that it was (at least statistically) possible for later transfers, if any, to vary slightly due to variation in the back-up preferences carried forward (piggy-backed) with the next usable back-up preference.


But the actual chance of such "random" piggybacking having any effect was minimal,

for these reasons:


- The number of times surplus transfer were done in any election was never very many - they cannot exceed the number of seats filled, and not even that many - when the final seat was filled, no transfer of surplus was done at all, randomly or otherwise.


- the effect of un-balanced piggyback preferences cannot have large effect due to there not always being any subsequent transfers after the suspect transfer. Eventually the count came to an end (usually in less than 21 counts) and even where un-balance crept in due to un-balanced "piggybacking", it would only have an effect at all if the piggy-backed preferences were referred to. Thus the unbalance would only possibly have an effect if the votes bearing the un-balanced piggybacking were transferred two or more times after the unbalanced piggybacking.


- It was not likely that the small variation in the "random" piggy-backed preferences (under the provincial election method) would have an effect on who was subsequently elected even in cases where there were subsequent transfers and where un-balanced piggybacking came to light in subsequent transfers - the last unsuccessful candidate, dropped at the end, usually had a few hundred or even a couple thousand votes fewer than the nearest successful candidate so likely would have lost whether or not the piggybacking had been perfectly and scientifically balanced.


Since those far-off days, PR-STV elections have been conducted in Australia where as many as 21 seats are successfully filled in one contest.


There is no bogging down due to having more than five seats, as some seem to think.


The computation and announcement of election results were/are slower under PR-STV than the as-it-happens preliminary results (often presented as final) we get on our election night nowadays, but even on election night in STV elections, with the completion of the first count, some candidates knew they were elected and others knew it was not at all likely they would be elected. Later vote transfers seldom made much difference to those front runners - and sometimes no change at all to any of the front runners in the first count. So the final result was fairly well known even on the first count conducted on election night.


in the end.And under STV, party seat tallies are mostly established even on the first count. Even on election night of PR-STV elections you can tell whether or not a party has majority of the votes and therefore it is pretty clear where any party will have majority of the seats or that no party has majority of seats. Which party has the most seats can actually be pretty accurately forecast just by looking at the overall vote count, which is mostly known on the first count done on election night.


Unlike FPTP where a less popular party can have more seats than another party. Under PR (whether it is PR-STV or MMP) there is no way a party with the second-most votes would take 40 more seats than the most popular party, as happened under FPTP in Canada in 2021

So under PR, unlike FPTP, just party vote tallies actually pretty much tell you the seat tallies of the parties or at least the relative break-down in party seat numbers.


Even where vote transfers under PR-STV determined that different people were elected than were leading in the first count, the changes were sometimes just different people of the same party, so the party seat tallies are mostly established in the first count, based on votes cast.


Under FPTP, party seat tallies are set by accidental simple plurality victories made possible by one-time vote splitting among other voters, or by slight shift in votes, or capturing all the seats in a district, city or province despite majority of voters in the district, city or province voting against the winner(s) But these strange little twists are not seen in PR, and therefore the overall vote tallies pretty much tell you who has won and the rough seat tallies of the different parties. And that is known or can be known on election night even under PR-STV.


I think if you look closely at MMP systems, you will see dis-proportionality to a degree in all. Parties with less than threshold take no seats at all or fewer than they are due. This can add up to perhaps 20 percentage of voters being ignored and with no recourse of transfer to other candidates or parties, unlike what we see in STV, these votes have no effect and so are simply disregarded.


But any move away from single-winner FPTP, with its fragmenting of the voters into 338 or 100 or 87 separate contests or whatever, would be welcome.


And any sort multi-member districts with fair voting or any sort of list PR top-up pooling of votes would be a positive - and necessary - step.

=========================================

There is confusion in the terms: preferential voting/ranked ballots/transferable votes/ ranked choice voting are used in both multiple-winner and single-winner contests. Preferential voting (under any of its names) is where the voter marks a first choice and then provides back-up preferences to be used if just having the first choice alone would see the vote be wasted. (Confusingly, ranked voting can also be used where all the choices are considered at once (perhaps with the rankings weighed differently and perhaps valued equally), and in ranked voting it is often not required that the voter mark back-up preferences (or to vote at all) so back-up preferences may have no effect at all because they simply were not marked.) But the number of members elected in each district or contest is very important thing. Without multiple-members elected, only one party or candidate can be elected so no proportionality. The diversity of opinion held in the district cannot be represented by just one member.

In city elections the election of mayor or the city councillors if they are arranged in single-member wards cannot be proportional. (see below for more on this) So multiple members in a district or contest are necessary to have PR and to use the PR system known as STV in particular -- more on this below. As well, a fair voting system is also required if PR result is to be assured time after time. A fair system is one based on proportionality, that is, one that denies a single party or group from taking all the seats. Such is Single Transferable Voting (note the capitalizations), which some call PR-STV or STV-PR. Back in the day in the British Commonwealth and the U.S., it was simply called Proportional representation because no (or few) countries used party-list PR circa 1920 and none of them were in the Commonwealth. To prevent just one group from taking all the seats, under STV-PR, each voter can only cast one vote (with back-up preferences marked if the voter chooses). Even this is not fool-proof against one-party sweep unless the district or contest has more than three seats. But with four or more seats no party can take all the seats unless it takes more than 80 percent of the vote and supporters mark their back-up preferences for that party's candidates and only that party's candidates (hardly ever the case), and if that party runs enough candidates to fill all the seats - not always the case. STV-PR uses quota which says that any candidate that receives votes in the proportion of the valid votes divided by the number of seats plus one (the Hare Quota) or divided by the number of seats plus one, plus one (Droop quota), will be elected. it is possible to be elected without quota but quota produces immediate election. Thus in four-seat district any candidate receiving more than 20 percent of the vote (Droop quota) will be elected. It always happens that at least two candidates of different parties get quota - if not in first count then through combination of first choice votes plus the addition of vote transfers allowed by other voters' ranked ballots. But if it happens that there are seats to be filled, by the time the number of candidates has been thinned through elimination to the number of remaining seats, then those surviving to that point are declared elected, even if they don't have quota. The field of candidates is thinned by elimination of the least popular candidate, sort of like plurality but in reverse. Under FPTP the most popular is elected even if he/she does not have majority of votes; in PR-STV, if needed, the least-popular candidate is eliminated. The rest may not all win but that one candidate is taken out of the equation. His or her votes though are not necessarily taken out of the equation. If the voter had marked back-up preference(s), the vote will transferred to the candidate marked as the next back-up preference, if that candidate is still in the running. Once a candidate is eliminated (or elected), they do not receive more votes. (that is how it worked in Canada anyway).

So that is PR-STV in a nutshell. (a differently-worded explanation is below) Terms are used in confusing ways perhaps on purpose by those who are opposed to electoral reform. Some say a system that uses ranked votes to elect just a single winner in a district or contest is "ranked ballot" or "single transferable voting" when more correct term at least in Canada is Instant-runoff Voting or Alternative Voting, or in the U.S. "Ranked Choice Voting." Some use the term "Proportional Representation" for only party-list PR systems, ignoring the value of PR-STV and its historical proven effectiveness in Canada in provincial elections in two provinces and 20 municipalities. PR-STV is not majoritarian if by majoritarian you mean a candidate needing to take a majority of the valid votes cast. Quota used in PR-STV to determine some winners, as explained above, is less than a majority (50 percent plus 1), if the district is three or more, and no candidate in history of PR-STV has ever taken majority of votes cast, the usual distribution of two or more popular candidates plus less-popular outliers being the norm.

And often some in PR-STV are elected without taking the quota even. In party-list PR, it can happen that 20 percent or more of valid votes cast are ignored, due to the threshold of five percent.

In STV about the same can be ignored -due to voters not marking back-up preferences or not enough back-up preferences and their vote being eligible for transfer one more time, or - due to the votes of the last eliminated candidate not being transferred as that elimination brought the number of candidates down to the number of remaining open seats and therefore the remaining candidates are simply declared elected, filling all the seats and the process ends. The term majoritarian itself is used confusingly. some say FPTP is majoritarian, that to be elected you must have majority. but actually whether you have majority of the votes or not is not important in FPTP. To be elected under FPTP, you must have the most votes -- that is, have more votes than the second-place candidate, no matter if the number of votes you take is large (82 percent is not unknown) or low (as few as 18 percent is enough sometimes to be elected under FPTP). If only two candidates are running for one spot, one or the other will take majority (more than half the votes) so FPTP in that limited case is majoritarian.

Example: in the 2016 U.S. presidential election, neither Trump nor Clinton received majority of the votes across the U.S.

How could this be? Because the election was not held using only two candidates. (and of course their percentage of the vote or even relative popularity was not what put Trump in the White House. it was the electoral college). PR-STV is not possible in city elections where single-member wars are used or a mayoral election. But back when STV was used in Canada in provincial elections in two provinces and 20 municipalities, MLAs and city councillors were already being elected in multi-seat districts, through Block voting (voters casting multiple votes). Block voting may seem exotic but today most of the city councillors in Alberta are elected in multi-seat districts, a most backward and un-democratic system. To change to PR-STV from Block voting, really took just two changes - decrease the number of votes each voter could cast to just one, and teach voters and election officials about marking ranked votes and how to do a STV vote count. This STV vote count is more complicated than the vote tally comparison of FPTP but is nowhere so complicated that computers are needed. STV was used in 15 provincial elections to elect from 5 to 10 members in a district, then in two more, where 2 and 4 were elected in each district. and city elections between 1917 and 1961 often saw 5 to 8 elected in STV contests. Calgary and Winnipeg held last SV elections in Canada in late 1960s and 1971, electing two or three in each contest. But if a city uses single-member wards, it also just takes two changes to go to PR-STV: - Simply group the existing district together to form multiple member districts electing the same as the total of the old districts (or more or less if it is decided to change the number of seats to address long-time urban-rural l disparity or other inequity) and - teach voters and election officials about marking ranked votes and how to do a STV vote count. The existing situation - with all of our provincial and federal politicians being elected in single-member districts - is recent, and we should not assume it was always that way. Every province used multi-seat districts to elect at least some of their members over all or some of the 1867-1996 period, except Quebec. The last federal multi-member district was abolished about 1957, I think, and the last provincial multi-member district as recently as 1996. I have blog on this:

Canadian elections - federal and provincial - used multi-member districts more than we think (montopedia.wixsite.cpr_discussion_group@googlegroups.com ========================================================== Some say electing multiple members is not sufficient for proportionality. that you need something that uses some form of quota.

that STV uses preferences.

while MMP measures party support and allocates seats in proportion to that. STV uses quota to fill some of the seats and others are filled by comparative plurality. The preferences are taken by one by one so are really just successive votes, each not different from vote cast under FPTP. I agree with part of what these people might be saying - I believe it is necessary to have quota to have assurance of getting proportional election result. But in a couple STV elections in Alberta the result was the same as if the election had been held using Single Non-Transferable Voting - that is, the STV winners were the same as the leaders in the first count who would have been elected under SNTV. The STV vote transfer merely confirmed that choice. So the quota - and the trasfers - had no effect.

If you accept STV as producing proportional results (which some perhaps don't but I do), then on occasion even a system without transfers and thus without quota produces just as proportional results.

Such cannot be said of FPTP , where there is no such direct relationship with supporters of multiple parties to the elected members. Only one can win in each district with all others ignored. Proportionality as measured overall where a lobsided victory in one district for Party A is balanced against a lobsided victory for Party B in another district is not true proportionality as the un-represented voters in each district simply have no representation - no elected member holding their views has been elected in their district.

Multiple-winner districts (or a pool to determine top-up seats in MMP) are critical to having PR. Some might say that one vote cannot be used to set different proportionalities. But I disagree.

if a person votes for a man who is Conservative while another votes for a woman who is a Liberal - just as examples - each of them in their own way has helped proportionality based on two different things, two different kinds of proportionality. Proportionality is short-hand for representation of diverse representation, with each part being in proportion to how votes are cast, based on party or any other aspect you want.

Recently in Canada we are looking more at the proportion of women elected (or in cabinet or nominated as candidates), being, for so long and still now, usually under-represented proportionally. for example.

The proportionality is based on the female proportion of the population in one direction (not female voters or votes cast for women candidates) and the proportionality of representation on the other. You cannot have proportionality in a single-winner district. You cannot have diverse representation in a single-winner district. ======================================================== Here is a link to a 2011 NZ comment about STV which I think explains how STV is not proportional in the view of that writer https://tryingtoreason.wordpress.com/2011/11/12/on-stv-and-proportionality/ On Ireland: <https://en.wikipedia.org/wiki/Elections_in_the_Republic_of_Ireland> and click on the date, or for example see <https://en.wikipedia.org/wiki/2020_Irish_general_election> Someone once said that Multi-seat constituencies are not enough for proportionality. One also needs a specific mechanism, otherwise the bigger the district magnitude the worse the proportionality as in Block Voting or the multi-seat AV system which the Australian federal Senate used early in the 20th century. There is a reason why the Australians often call PR-STV "quota-preferential voting."

==================

I recently wrote an electoral reformer acquaintance in Scotland: Thank you for contributing, as one who actually lives under PR and "STV-PR".

It would be fascinating to read a history of Scotland's successful drive for STV-PR. Is there something you would recommend? [Apparently there is none available right now.] PR is a fair means to establish representation, and thus fairly establish the majority in the chamber, a legislative majority, that is supported by majority of voters and not a way to bring in consensus government or the like. .....

To measure the satisfied voters under STV, I generally look at the number of votes received by the successful candidates on the count where they are declared elected to see what proportion of the voters were satisfied by result.

Some are looking just at first count vote tallies but that is not enough to measure STV effectiveness. ================================================

Another explanation of STV-PR


Voters rank candidates in order of preference, with the later preferences being used just as contingency votes (back-up votes). Each district elects multiple members, usually elect three to seven representatives in each; each voter casts just one vote. The count is cyclic, conducted in stages, each stage involving the elimination of candidates and the transfer of their votes, or the transfer of surplus votes of a previously-elected candidate, until all seats are filled. The transfers are conducted in line with the next marked preference marked by the voter, if any.


Any candidate whose tally reaches a quota, the minimum vote that guarantees election, is declared elected.


After a candidate is elected, the candidate's surplus votes (those in excess of the quota) are transferred to other candidates. (This transfer can be conducted using whole votes (some of the surplus votes are transferred) or all surplus votes can be transferred at a fraction of their value, proportionate to make-up of the candidate's votes or set part thereof.)


If there are no elected candidates whose surplus votes have not been transferred and there are still seats to fill, the least popular candidate is eliminated, those votes being transferred to their next preference at full value. The votes are transferred according to the next marked preference on each ballot. Any votes that cannot be transferred are moved to a pile labelled exhausted or non-transferable.


The count ends when all the seats are filled. If at any point after all surplus votes have been transferred, it is found that there is only one more candidate than the number of remaining open seats,all of them except the least popular candidate are declared elected, even if they do not have quota.


The transfer of votes of eliminated candidates is simple -- the transfer of surplus votes is more involved. There are various methods for transferring surplus votes.


Manual methods used in early times transfer surplus votes according to a randomly selected sample, Manual methods still today in places where STV was adopted in early 20th Century (Ireland and Malta) transfer only a segment of the surplus selected based on the next usable marked preference.


Other more recent methods transfer all votes at a fraction of their value (the fraction derived by the surplus divided by the candidate's tally) and with reference to all the marked preferences on the ballots, not just the next usable preference. They may need the use of a computer.


The different methods may not produce the same result in all respects. But the front runners in the first count, before any transfers are conducted, are all or mostly are elected in the end, so the various methods of transfers all produce much the same result.

(from Wiki "Proportional Representation")


=================


0 views

Recent Posts

See All

Comments


bottom of page