STV, the form of Proportional Representation used in Alberta from 1920s to 1950s, was fair. It was much fairer - it yielded more democratic results - than the non-proportional First Past The Post system we use today to elect MLAs, MPs and many of our city councillors.
In FPTP, as many as 82 percent of the voters in a district are ignored, elect no one. (In Toronto in 2014, under FPTP, a city councilor was elected with just 18 percent of the votes in her district. In that election contest, 82 percent of valid votes were ignored.)
But the number of votes that are used to elect someone in STV elections is usually around 80 percent. Only 20 percent of valid ballots were not used to elect someone.
You might say though that the votes used to elect people under STV are not first preferences but only later choices. Even if this was true, the votes would be used effectively to elect someone preferred over other candidates, so the voter would be well pleased by such eventuality.
But actually a great number of the votes (often around 60 to 70 percent) are used to elect first-preference candidates in each multi-member district used in STV,
and an even greater number are used to elect the first or second choice of the voter, and so on.
(In the 2007 Glasgow city election, in each four-member ward, around 90 percent of the voters saw one or more of their first, second, third or fourth preferences elected in their district.)
(for more info on Scotland's STV, see
====================================================================
To see how many voters are represented under STV, let's take the STV election of Calgary MLAs in 1926 as an example.
Any candidate elected on the first count was elected with only first preferences. So that is clearly pleasing to those voters. And when two or three MLAs are elected in the First Count, they are of a variety of parties so that is obviously fair, mixed representation that truly reflects the mixed votes cast by the voters. Great stuff all around.
As well, we can "reverse-engineer" in a matter of speaking, and look at winnrrs in the end and go back to first count and see how many votes were placed on the winners as first preferences.
1925 election of Calgary MLAs
(quota was 3290) Five members elected
First-round vote tallies:
McGillivray received 5928 votes in the first count (elected on first count)
Webster received 2941 votes in the first count
Irwin received 1662 votes in the first count
White received 1222 votes in first count
Parkyn received 2467 votes in first count.
These five -- all winners in the end -- thus received 14,220 first-choice votes out of 20,000 votes cast. This is 71 percent.
McGillivray was elected on the first count (quota being 3290, he had quota on the first count). He did not pick up more votes as he was immediately declared elected.
But the other four received more votes through transfers so that in the end, they had an additional 4063 votes.
Through votes transfers, by the time the votes transfers were finished, these five winners were known to have support from an additional 4063 voters. They were not the first choice of these 4063 voters but were more preferred than any others still remaining in the running when the additional votes were lodged with them.
Webster received another 349 votes;
Irwin got another 1628 votes;
White got another 1701 votes;
Parkyn got another 385 votes.
(Some of these votes were votes transferred as McGillivray's surplus, so the confirmed number of voters who marked one or more of their preferences to members of the five is a bit of a question.
But at the same time, any transfers that were marked for McGillivray never came to him as he could not receive more votes after his election. Nor to Webster and Irwin who were elected relatively early in the process. So they might have been the second choce on many more ballots than they actually received.
Next section attempts to arrive at the lowest possible number of votes who saw their vote used to elect one of those five.
=============================================
looking at Edmonton in 1926
Lymburn declared elected after first count. he had quota just based on first preferences. His 20-vote surplus was not transferred until later.
Leedy was eliminated in the second count.
His 140 votes were quickly transferred.
Once votes are transferred in the 2nd Count, usually through transfer of surplus votes of a First-Count winner, but in this case through Leedy's elimination, then some murkiness creeps in. But if we carefully eliminate the possibility of faulty logic, even at expense of our argument, then we should be fine.
Starting in the second count, I am going to assume that all possible votes are of the lowest possible preference.
At the time when Robertson was eliminated in the 3rd Count, he held 365 votes. 361 of his votes were marked with him as first preference; and there were four that were marked "Leedy 1 Robertson 2". These had been transferred to Robertson when Leedy was eliminated. (Many were marked with names of other candidates for 3rd, 4th preferences, etc. but that is not important.)
When Robertson votes were transferred to others, they went to 15 other candidates. Which of those are the four second preferences is not known so we have to assume that any four could be the second preferences. Therefore to be on the safe side I assume that four of each transfer are third preferences marked Leedy 1, Robertson 2, and 3 for the recipient candidate. (This means that 60 votes are assumed to be marked this way while I know that actually only four are. but this is the only way not to be accused of fudging the figures to make my case.)
As well, the four second-preference votes might not be in any of the transfers at all, but might bear no third preference. If so, they were among the 10 that were placed in the exhausted pile when Robertson was eliminated. On the other hand, those exhausted votes might have been votes that were marked "Leedy 1, Robertson 2," or they might be votes where Robertson was marked 1, ones where the voter marked just one preference. So that raises more uncertainty.
Four votes went to Farmilo in the 2nd Count, so we have to assume that when Farmilo is eliminated later, that when his votes were transferred to others, each transfer might have contain those four votes and that therefore the recipient of them could not be more preferred than the third preference. Farmilo votes were transferred to four candidates so if we calculated that the four votes are in each of the four transfers, then for our calculations those four become 16 at least as possibility..
It is known that Farmilo's votes when he was eliminated contained 973 first preference votes and four second-preference votes, and others. But in the four transfers of his votes I have counted 16 second-preferences in total just to be on the safe side. That is to say, of 1891 Farmilo's votes at the time he was eliminated, 973 were first preferences and four had "Leedy 1, Farmilo 2". But how many were transferred to other candidates and how many were among the 461 that went to the exhausted pile is unknown.
======================
(Here's more explanation which may be safely skipped over when reading this piece)
So when Leedy was eliminated in the 2nd Count, there are votes whereon Leedy was marked as first preference and Farmilo as second preference. If four votes go to Farmilo in the Second Count, which they did, then we have to assume that when Farmilo is eliminated later, that when his votes were transferred to others, each transfer might contain those four votes and that therefore the recipient of them could not be more preferred than the third preference.
Farmilo votes were transferred to four candidates so the four votes by being counted in each of the four transfers then became 16. It is known that Farmilo's votes when he was eliminated contained 973 first preference votes and four second preference votes, and others.
But in the four transfers of his votes I have counted 16 second preferences in total just to be on the safe side. That is to say that of 1891 Farmilo's votes, 973 were first preferences and at least 918 may not have been.
======================================================
1926 Edmonton
Lymburn was elected with 3046 votes, all first preferences.
Weaver was elected with 3065 votes, consisting of 2200 first preferences and a minimum of 371 second preferences. He won after receiving 863 in vote transfers, all of which could have been second preferences. At minimum, 84 percent of his total votes received were from first and second preferences.
Gibbs was elected with 3543 votes, consisting of 879 first preferences and a minimum of 1337 second preferences. He won with 2664 vote transfers, all of which might have been second preferences. At minimum, 63 percent of his total votes received were from first and second preferences.
Prevey elected with 2940 votes, consisting of 1517 first preferences and a minimum of 460 second preferences. He won with 1423 vote transfers, all of which might have been second preferences. At minimum, 67 percent of his total votes received were from first and second preferences.
Duggan elected with 2265 votes, consisting of 857 first preferences and a minimum of 504 second preferences. He won with 1408 vote transfers, all of which might have been second preferences. At minimum, 60 percent of his total votes received were from first and second preferences.
Overall, winning candidates received 15,000 votes, 82 percent of the 18,000 valid votes.
Overall, 8501, 57 percent of the winning candidates' vote totals, were made up of first preferences.
(8501 was 47 percent of the 18,154 valid votes in Edmonton.)
(These votes were not transferred, except if determined to be surplus votes.)
Overall, at least 11,200, at least 75 percent of the winning candidates' vote totals, were made up of first and second preferences.
(11,200 was 62 percent of the 18,154 valid votes in Edmonton.)
Overall, at least 2700 votes, at least 18 percentage points of the winning candidates' vote totals, were made up of second preferences.
Overall, 79 percent of those who cast valid votes actually helped to elect members. As well, an unknown number of the exhausted votes had been marked with preferences indicating support for the winners. If so, they were exhausted because those preferences did not become usable until afte the preferred candidate had already been elected.
In 1935, more than 80 percent of votes cast were used to elect the Edmonton MLAs. As well, an unknown number of the exhausted votes had been marked with preferences indicating support for the winners.
The large percentage of 79 PERCENT-PLUS and 80-PERCENT-PLUS are usual for STV elections -
in all STV elections the winners take a very large proportion of the votes cast.
BUT under FPTP as much as 82 percent of the votes may be ignored in a district.
=======================================================================
Other stats that confirm STV's practicality and ability to produce a high number of Effective Votes, and other STV matters
comparing Ireland's STV with our old STV in Canada
When Edmonton, Calgary and Winnipeg used STV to elect their MLAs, they used whole-vote surplus transfer and simply left the winners with quota right there on paper. They did not mark them down as having 0 votes.
Quota stays the same in Irish elections so what is purpose of the votes still in play number?
In Edmonton/Calgary/Winnipeg STV, the tables showed the candidates' votes, (exhausted votes usually recorded separately). This was confirmed by the total of these two numbers, which was the same as the original valid votes. There was no actual statement of the votes still in play.
Calgary, for sure in many sources, was said to use a more complicated system than whole-vote transfer so I imagine it must have been Gregory, but I have not seen the actual vote tables to confirm that surmise.
Perhaps Calgary was the only one to use Gregory. Calgary did pretty much for sure, and perhaps other cities did as well, but maybe no others did.
(I did find a newspaper page showing Vancouver's vote count in 1922. You will find it in a separate Montopedia blog.)
The "pragmatic contingency plan" used in Edmonton/Calgary/Winnipeg STV when there were fewer transferable votes than surplus was simply to leave the winner with all the non-transferable votes (as quota+) and transfer the transferable ones without need for any math.
But it is difficult to see if this happened in Irish results because the vote count table does not actually say how many are left behind with the winner, and because some counts do two things - a surplus transfer and a transfer of eliminated candidate's votes. This double purpose obscures the thing.
(The Wikipedia article for the 1927 Manitoba election presents the vote count process in Winnipeg but like Ireland, it meshes differnt transfers in one count in its vote count table. A separate list (in a Montopedia blog) shows which counts in the table correspond to which counts in the actual vote count because the actual count only did one thing in each count, not two.)
In Alberta STV elections, there were many instances of not having enough transferable votes to cover the surplus. But this seems not to be the case in Ireland's elections.
=========================
Alberta prov STV elections
Edmonton and Calgary 90 MLAs 1926-1955
(Medicine Hat used STV in 1926 but not worthy of attention -- only three candidates went for two seats)
90 MLAs elected in Edm and Calg 1926-1955
(not including by-elections where AV was used)
theoretically each time a candidate is elect his or her surplus votes are to be transferred but this is not done if the surplus votes are not enough to make difference or if all the seats are already filled (so never was the surplus of the last person to be elected transferred, for instance)
31 times the surplus was transferred using math (the whole vote method)
9 Calgary/22 Edmonton
19 times the surplus was transferred without math
(transferable votes did not exceed the surplus) 16C/3Edmonton
40 times no transfer was done (either would have had no effect, or seats already filled or certain to be) 19C/21Edmonton
so these numbers tell us that in 16 STV contests, 8 in Calgary and 8 in Edmonton, only 31 times was mathematical reduction resorted to. Thus an average of only two times in each city each election did mathematical reduction have to be used.
(I have the tables for all of these STV elections except one (I collected them from various files at the Provincial Archives) but they are all different sizes and shapes so not easy to reproduce.)
Here is more explanation of this point:
I looked at Alberta provincial STV elections and found that actually the "complicated" math-reduction method for surplus vote transfers had to be used actually much fewer times than some imagine:
Edmonton and Calgary 90 MLAs elected through STV 1926-1955
(Medicine Hat used STV in 1926 but not worthy of attention -- only three candidates went for two seats)
90 MLAs elected in Edm and Calg 1926-1955
(not including by-elections where AV was used)
theoretically each time a candiate is elected, his or her surplus votes are to be transferred but this is not done if the surplus votes are not enough to make a difference or if all the seats are already filled (so never is the surplus of the last elected person transferred, for instance)
31 times the surplus was transferred using mathematical reduction (in this case the the whole-vote method was used, not the Gregory fractional method.) 9 Calgary/22 Edmonton
19 times the surplus was transferred without math (simply looking at the ballot and moving it to next valid marked preference)
(transferable votes did not exceed the surplus) 16C/3Edmonton
40 times no transfer was done (either would have had no effect, or seats already filled or certain to be) 19Calgary/21Edmonton
so these numbers tell us that in 16 STV contests, 8 in Calgary and 8 in Edmonton between 1926 and 1955, only 31 times was mathematical reduction resorted to.
An average of only two times in each city in each election did the "complicated" mathematical reduction method have to be used.
it might be different in a system where voters are required to give very complete rankings but in Canada the "one-third of seats filled through math" stat was the case.
The point being that if math-reduction surplus transfers are used as seldom as that, it seems to not matter what method is used to derive them, whether Gregory or semi-random whole-vote method is used, as far as most of the seat results go.
And anyways the math for math-reduction surplus transfer is quite simple:
say B is successful candidate and we need to find how many of B's votes are to be transferred to Candidate A
a/b X C = A
where
a is number of votes marked with next preference for candidate A among B's votes
b is number of Candidate B's votes
C is number of votes that B has more than quota (B's surplus)
A = number of votes that are transferred from B to A
======================
Much is made of the vote transfers and the quota.
But Harris pointed out that in many cases candidates were elected with less than full quota.
We see this in the use of STV in Calgary.
From 1924 to 1956 44 MLAs were elected in Calgary.
Eleven of them, a full quarter of them, were elected with partial quota.
We also see this in the use of STV in Edmonton.
From 1924 to 1956 46 MLAs were elected in Calgary.
Twenty of them, just less than half of them, were elected with partial quota.
===================================
PR was used in Edmonton, Calgary and Winnipeg for 30 years, and to elect MLAs to represent two other cities provincially, as well
and STV was used for municipal elections in those cities and 15 others as well
This is said to be the deepest experience of PR in all of Canada and the U.S. too.
That is said by those who discount Illinois's use of Cumulative Voting for about 100 years
and by those who put Canada's use of STV ahead of the use of municipal STV in U.S.'s largest city, New York City, in five elections, 1937-1945.
in its time of use of the STV-PR system, Calgary held more STV elections than the City of Cambridge, Mass. has had so far. (Cambridge has now used STV since the 1940s.)
And in each election contest/district each time (usually the district was city-wide), mixed balanced rep was elected.
Labour (of diff. tints), Conservatives, Liberals, almost always got rep.
Farmer, Communist, Social Credit MLAs and Independents also were elected whenever they had approximately a quota (17 percent or so) of support.
A city's first woman MLA and first woman councillor was elected under STV in many cases
Winn. city elections used STV 1920 to 1970
Winn. city elections suffered from low DM -- only three seats in a ward were filled at a time.
but still the contests mostly produced mixed rep.
one-party sweeps of a district's seats did occur in Ward 1 of Winnipeg.
The other two wards always produced rep for two or more parties.
But yes, our STV experience can be seen as depressing, perhaps a failure, in that that fairness did stop and we don't have PR now.
provincially
Man. -- Winnipeg, St. Boniface STV stopped in 1955 (last STV election was 1953)
Alberta -- Edmonton and Calgary) STV stopped in 1956 (last STV election was 1955)
municipally
all cities dropped STV by 1930 except Calgary and Winn (and a few small Winn area municipalities)
Calgary 1960 mostly dropped STV, then intermittent use in 1960s, 1970 - all aldermanic seats filled through STV, then it stopped
Winn. 1970 stopped
Prov. PR was always "PR light," in that most members in the legislature were elected through single-winner contests with no overall proportionality. (other members elected either through FPTP or IRV)
But still it was most seen in North America, leaving aside Illinois's CV.
As pointed out in by another person in a share on list serve, Canada also has had experience with Cumulative Voting and Limited Voting.
STV did not come to forefront until it proved itself in Australia (Tasmania 1890s) and in city elections in Ireland (Sligo 1918) - in Britain!) and Calgary (1917-), BC cities (1918-).
and in Britain:
in 1918 STV was adopted for the university constituencies of Cambridge, Oxford, Combined English Universities, Combined Scottish Universities and Dublin University.(friom Wiki "History and use of STV")
Malta and Ireland brought STV in about 1920 nationally.
it proved itself to work in British and British Empire settings, and momentum carried it - for a while anyway
Never has list PR or SNTV been used in North America
(SNTV esp with slight PR top-up, could work admirably,
but STV is not much more difficult than SNTV -- ranked voting, the only distinction between it and STV, is not difficult especially if STV is "optional-preferential method".
Vanuatua elections show us what SNTV can do, still with X voting but MMDs.
Canada has long experience with MMDs, even longer than its use of STV.)
mixture of district members and at-large members is used in some city elections in North America.
This is nice combination of overall proportionality with micro-local (sub-city) rep for those who think a city is too large for proper accountable rep.
"nice" and proportional is only the case though if single voting is used -
Block voting allows one-party sweep -- under block voting one voting block often takes all the seats -- so it does nothing for PR other than hold fragile promise of good things.
PR of a sort can be produced even with block voting if the most-popular party (voluntarily) does not run full slate.
if that happens, mixed rep is produced. but same could be accomplished though single voting without expense and bother of counting so many votes, and without having to depend on the most-pop. party voluntarily restraining itself.
=============================
Vancouver's recent election did produce mixed rep and the most-pop party did not run full slate - apparently on account of worrying about diluted votes or just poltical tradition (this is perhaps a case of who knows how these things start?)
but if it ran full slate of ten candidates it still might not suffer.
Sure voters cast only seven of their ten votes - on average - some might cast ten; some might cast only five or fewer--
but perhaps that low casting is merely due to no party running more than 7 candidates.
in addition to ABC Vancouver, Greens, Forward Together, Progressive and all others also ran seven or fewer.
chicken and egg kind of thing.
===================================
Labour/CCF/NDP did relatively well under STV, not so under FPTP until 1980s
1926 -– Alberta used proportional representation (STV) to elect MLAs in Edmonton and Calgary.
CLP's Lionel Gibbs elected in Edmonton;
DLP's Fred White and Independent-Labour candidate Robert Parkyn elected in Calgary.
Use of STV to elect Edmonton MLAs produced the election of a Labour/CCF MLA every election from 1926 to 1955, excepting 1935 and 1940, and 1955.
In Calgary under STV, a Labour/CCF MLA was elected in 1921, 1926, 1930 and 1944.
BUT after the change to First Past the Post in 1956,
no CCF/NDP was elected in Edmonton until 1982,
None was elected in Calgary until 1986.
====================
Comments