top of page
Tom Monto

STV - PR - Malta - Edmonton. A real hodgepodge!

Updated: May 5

There does seem to be a "sweet spot" in reform efforts.

say the eight-hour day movement

when workers were working 10 or 12 hours, they (themselves) did not have time or energy to fight for shorter hours.

the need was such though that others (say church ministers) took up the fight.


then when hours were shortened to say nine-hour days Mon to Friday and a half day on Saturday, workers had enough free time to fight for the eight-hour day and for two full days off on week ends.


and two or three weeks off a year.


political culture then dictated that that was enough for Canadian workers.


France somehow continued on to achieve 6-week holidays each year, I believe.


Right now concerning electoral reform, we are in the 10 to 12 hour day stage. Things are so bad we can't get it fixed without something unusual happening. (although they could actually get worse before they get better!)


Winnipeg got STV for city and provincial election after a general strike.

Belgium got PR due to rising civil unrest and to avoid worse - civil war.

Ireland got STV due to rising civil unrest and to avoid civil war.


The efficiency of PR/STV is perhaps shown by history of less disorder in the PR-elected Republic of Ireland versus the disorder and violence in FPTP-elected Northern Ireland. The Brit. government imposed STV on Ireland and Northern Ireland after partition, but as soon as it could (1929) the dominant group in Northern Ireland stopped it, leading to decades of strife and murder. It returned to STV in 1973, almost 50 years ago.


if we can get a bit of ER or PR, say in just one city or province, then it should be easy to get more because

- problems are so visible;

- our elections are so dis-proportional, and

- need for ER is clear to some or at least to many in "most every place," and

- ER/PR/STV produces immediate and self-evident improvement as soon as first fair election:

- the first woman councillor elected in Calgary and in Winnipeg was in first STV election in that city, 1917 and 1920 respectively.*

- the first woman MLA elected in Winnipeg was in Manitoba's first STV provincial election, 1920

- first Liberal Dem elected in 50 years was elected in State of Victoria's first STV election in 2006.

- Edmonton's first Labour MLA was elected in Alberta's first provincial STV election, 1926.

and so on.


Each of these happened while men and Liberals and Conservatives were elected as well so results were balanced and fair.


And the fair, balanced rep was repeated in next election and the next as long as voters voted for that.


PR is more dependable than FPTP.


FPTP may produce accidental election of a single iconoclastic or opposition or minority or Labour or Independent candidate in some obscure corner due to local vote splitting or local conditions.


but usually these minority successes under FPTP are flash events.

Often these minority politicians lose their seat in next election, even if they get more votes in the next election, which they often do. The old-line parties (Liberals and Conservatives) close ranks to end the person's stay in office.


under PR, generally no such flash:

Edmonton's Labour MLA was re-elected in 1930, finally dying in office in 1934 - when will Prairie boys learn to swim?


When Edmonton used STV in city elections (1923-1927), Labour consistently took four seats each time.


Compare that with what we see after STV ended for Edmonton city elections (1928).

Labour took majority control with minority of the vote, then a few years did not have one seat in city hall. Votes were not cast all that differently each time.


So if we can get a bit of the good (real PR) somewhere, wildfire or domino theory will prevail and it will spread (or so I expect and hope).

But we just need first success...

is it to be Yukon or PEI? And is it to be soon?


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

Some have the idea that if a party has half or more than half of a quota needed to win a seat, it should be rounded off to one seat. so .5 is effective quota.


But of course if 1/11th or 1/10th of the vote is quota for a 10-seat district, which normally is the case, and half of that is sufficient to win a seat, you might need potentially 20 seats if you have winners win with just a half-quota.


So the idea is mathematically correct but also wrong.


Perhaps it is not too much to say it is like chaos theory - we actually don't know how large Canada is -- no one has actually measured every square foot or square inch that is inside Canada land area. it is estimated - rounded off - to nearest 1000 sq. kms. Canada is said to be 9.985M sq. kms. but that is clearly rounded off.

every inlet has little inlets

every peninsula has little peninsulas and so on.


Rounding has to stop somewhere.


Quota should not be thought of as minimum required to win a seat but as the amount that for sure means you win a seat.


say you have .45 of quota, say you mathematically round it up to .5 then you round that up to one as is normally done for .5. and a mere fraction becomes everyone - and before you know it, you sound like someone defending plurality FPTP elections!


You have to be careful how you round up, for accuracy - unless you want to inflate numbers, which when you have only so many seats in a district, you don't want to do.


under PR it is possible to win a seat with less - perhaps with half or less than half of quota even - due to use of largest remainder method to allocate the last seats.


This happens because some votes are exhausted - they run out of marked back-up preferences - and so the number of votes drops sometimes so much so, that if Droop quota was required to win a seat, it would be impossible to fill all the seats.


Largest remainder method explained in 1924

John D. Hunt (author of 1924 A Key to P.R.) laid out how to allocate seats under STV.


STV mostly allocates seats based on quota - anyone with quota gets a seat for sure.


In his example it came down to three candidate remaining with one seat remaining to fill. The least-popular candidate was eliminated and his votes transferred based on back-up preferences to the other two or to the exhausted pile. (Transfers are done just by simply moving them from one pile to another - no great math skills required).


The most popular of the remaining two had majority of the remaining vote and wins the seat.


But Hunt's vote count process did envision use of the largest remainder method.


Earlier in the count when first person (A) took a seat with a surplus of votes (over and above the Droop quota) and there were still seats to fill, Hunt got into how to transfer A's surplus votes, and mentioned largest remainder.


say A wins with 199-vote surplus.

You derive fractions of votes marked for B, for example, compared to overall votes won by A. and multiply the surplus by this fraction to get the number of votes to go to B. And do this with all candidates whose name was marked as next available choice on any votes originally placed on A.


For example,

B 8/9 X 199 = 176 8/9ths of votes

C 1/18 X 199 = 11 1/18

D 1/36 X 199 = 5 19/36

E 1/36 X 199 = 5 19/36

(Nowadays with calculators we use decimal fractions, but back in 1920s they used old-school fractions!)


Hunt's proposed STV, used in Canada STV elections, uses whole numbers of votes, thus avoiding the completed fractional math of vote transfers under Australia's STV,


How many votes go to B, C, D and E of A's 199-vote surplus?

first look at the whole votes indicated above:

B 176

C 11

D 5

E 5

total 197

two votes still remaining

So you look at fractional votes and give the two remaining votes to the two candidates with the largest fraction left over (largest remainder if you will)

B with 8/9 of a vote gets one vote.

E and D are tied, but E had more first count votes so gets the last vote. (Hunt threw in an extra complication there!)


So surplus transfers are the only use of "largest remainder" in STV, (and surplus transfers are only done after a seat is filled - and if there are still seats to fill and/or transfers still to make. The winner at the end does not have his or her surplus transferred, and there are other cases too where surplus is not transferred, so not as much need for such intense math as some think in STV.

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


With party-list top-up, you use the largest remainder to establish seats. Rounding off is needed as seats must be in whole numbers and party vote percentages are fractions, producing fractional seats.


We just have to have care when we do the rounding. Relative popularity (not just sheer adherence to a mathematical quota) is still important even in quota-based system.


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


Minority rule

A party with majority of seats may not have support of majority of voters.

Say to win a majority of seats, you need 51 seats in a 100-seat house.

Each of those 51 members could be elected with as few as 24 percent (or even only 18 percent) of the votes cast in their district.

so 51 times .24 = 12

Thus, 12 percent of votes cast overall may be enough to take majority power in the chamber.


This number becomes less where candidates of that party run in districts where number of voters are less than elsewhere. Rural sparsely-settled districts usually have fewer voters than densely-packed urban districts (despite the widely different geographic sizes involved). Rural district may have a half or a third as many voters as a big-city downtown urban district although having perhaps a thousand size the geographic size.


And voter turn-out can vary too, even if districts all had similar number of voters on the list.


So it is technically possible for a party to take majority power with less than 12 percent, but so far, in Canadian federal elections never has everything aligned in such a way as to allow a party to take majority control with less than 38 percent.


38 percent of the vote was how much Chretien's Liberals had in 1997 when he took a majority of seats in the HofC.


This was almost equalled in 2015 when Trudeau took 39 percent and a majority of seats in House of Commons. That is, less than two out of five votes cast supported the party in power.


Where multiple-member districts are used, and scientific basis of election is used such as MMP top-up or STV, the minimum needed for majority control become much less variable and much closer to a majority of the votes cast.


And voters' awareness of unfairness rises.


in Malta, where STV has been used since 1921, high level of fairness is expected.

in 1981 Labour took 49.1 percent of the vote and a slight majority of seats. although National party took 50.9 percent of the vote.

Such unfairness caused controversy and a constitutional change to prevent its repetition.

(it happened even under perfectly sound STV due to voters across the small country being separated into 13 districts,)


Such odd result was considered so serious that a constitutional amendment was brought in to allow supplemental members to be added to ensure a party with majority of votes always takes a majority of seats.


But this result would not even be seen as odd in Canadian elections.


Often parties with fewer votes take more seat than other parties with more votes.


Often a party with minority of vote takes a majority of the seats. (in federal Canadian elections this has happened six times since 1984.)


Here (Canada) where we have voters split among 338 districts, that sort of odd result would be common, even if we did not use FPTP where the leading candidate -- and only the leading candidate -- takes the district seat.


Multi-member districts reduce splintering of the vote while also allowing district-level PR (in such system as STV or regional MMP, as in Scotland).


The fewer the districts, potentially the more fair the election result.

Fewer odd bits of votes being left behind when counting ends in each district.

(see footnote)

The more members per district, potentially the more fair and balanced would be the district result.


FPTP in single-member districts is potentially the worst combination possible.

Block Voting, in multi-member districts, is likely as bad.

It is both worse and better in different ways:

- worse in that we can't know how many voters actually support the winning party or the losing party, for that matter. Casting multiple votes opens the door to this confusion. How many voters cast all their votes for a whole party slate is un-stated by sheer overall vote tallies.


- better, in that sometimes mixed representation is produced and thus a higher number of voters who agree with at least some of the elected members, if we say a voter who votes for a Liberal is happy to see any Liberal elected, that a voter who voted for a Conservative is happy to see any Conservative elected, etc.


I should add also that the more members in a district, the more choices the voter has to vote not only for party but also for individual candidate within that party. That is true if voting system is set up to allow vote to go directly to a candidate, instead of just to a party slate

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


Two different ways to analyze fairness of an election

- does the seats taken by a party reflect that party's portion of the valid votes?

This measure performed on all parties is how the Gallagher index is derived.

But a poor result ion one region may be balanced against a poor result in another place so overall the Gallagher index may be good but the represention could be highly regionalized with many voters ignored in each place.


- how many votes are used to elect someone.

This is not often used but may be as important as the other measure.

Where a large proportion of votes are used to elect the successful candidate(s), that means the result must be proportional, no matter how parties' seat standings compare to their vote proportions.


Under FPTP, effective votes are barely over half, and often not even half, of votes cast.

in 2013 Edmonton city election, 114,000 votes went to successful candidates and 92,000 to unsuccessful ones.

2017 86,000 votes went to successful candidates and 105,000 were cast for unsuccessful ones.

(more info here:


The measure of "Effective votes" can be done even where transferable votes might - perfectly democratically - create poor Gallagher Index derived from First Count preferences.

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

looking at Effective Votes in Malta's STV elections

In Malta in March 26, 2022 election,

say we look at two districts picked at random:


District 1 22,000 votes cast

quota was 3615

successful candidates were marked with these tallies after surplus, if any, was transferred away:

Azzopardi 3615 votes

Carabott 3615

Demarco 3615

Debatista 3200

Farrugia 3500

total of 17,545 votes

Thus, 80 percent of votes went to successful candidates. These votes might have been directed to these candidates by preferences other than first choice but they did go to a candidate who was favoured by voter over others.


District 10

votes cast: 21,667

quota: 3529

successful candidates were marked with these tallies after surplus, if any, was transferred away

Giglio 3529

Arregio 3529

Falzon 3529

Grima 3529

Sammut 2973

total of 17,000

Thus, 79 percent of votes went to successful candidates. These votes might have been directed to these candidates by preferences other than first choice but they did go to a candidate who was favoured by the voter over others.


So Effective Votes under Malta's STV were about 30 percentage points more than we saw in Edmonton's city FPTP elections.


As votes can cross party lines under STV, the final result might vary from party proportionality based on First Count preferences,

but about 80 percent of voters saw their vote actually used to elect someone they preferred.


As well there are about 1700 exhausted votes at the end -- some of these might have been marked for candidate(s) who were elected before the preferences on the ballot were used to transfer the vote. So potentially another 8 percent of the voters might have seen one of their choices elected, even if their vote was not used to elect them.


That has got to be democratic, however you look at it.


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

Footnote:

The fewer the districts, the less voters are separated from candidates they may favour. If "A free election is one in which all citizens are able to vote for the candidate of their choice,"* then having voters split up into 338 ridings means that many voters do not have right to vote as freely as they might want. A voter's own daughter or father might be running and they can not vote for them because they reside in a different district, for example. ======================================= Explainer "Free and Fair Elections" online* states: "A free election is one in which all citizens are able to vote for the candidate of their choice, and a fair election is one in which all votes have equal power." The instructional website lists 8 aspects of free and fair elections. Only #7 of the 8 aspects listed even refers to how votes translate into seats. The other numbers refer to access to voting booths; to right to run as candidate; etc. -- all important enough but not enough to ensure fairness if vote tallies practically do not matter. In #7, emphasis is put on accurate counting:

"Ballots must be counted correctly in order for an election to reflect the will of voters." But no mention is made of fact that votes once counted should be used to count for something. You can count votes as accurately as you want, but if a majority of them are ignored when seat(s) are allocated, as happens often under FPTP -- whether votes are examined overall or in specific districts -- what is the good of that? *https://www.facinghistory.org/educator-resources/current-events/explainer/free-fair-elections


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

*for information on the election of Winnipeg's first woman city councillor in 1920,

see


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

In Australia it sometimes happens that someone with less than 1 percent of the vote will be elected in the end after vote transfers.


This happened once when a candidate who got just .51 percent of the vote won.


But that .51 percent is based only on initial votes. After vote transfers were conducted that candidate had many more than .51, something like 25 percent of the vote, so that is what really got him elected.


Group ticket voting takes power out of hands of the voters, - although that is voluntary choice by voters) and i think some Australia jurisdictions are dropping it.

and also dropping their requirement that votes rank all parties or all candidates.


That is likely why NZ did not adopt STV when it brought in PR - due to scare stories of overly strict Australian STV rules.


Ireland and Malta have long history of STV and there no voter is forced to mark more candidates than he or she wants to.


In fact many votes go to first preference and most go to first or second so any later marked back-up preferences are often not even referred to, even when marked.


This is not failure of the system - single voting in MM district alone (without vote transfers) produces balanced mixed rep, the hallmark of PR.


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

3 views

Recent Posts

See All

Comments


bottom of page