top of page
Tom Monto

STV - reflections and observations in its defence

Updated: Jan 26, 2023

Criticism of STV includes that it is complex, slow and not proportional, and either produces too much local representation or not enough.

I hope the following essay puts the readers’ mind to rest by reassuring them that STV elections strike a happy medium, are easily conducted by voters and election officials, and the fairness and balance of STV elections are - and would be - far superior to those conducted under the present FPTP system.

STV is slow - results are not known for days or weeks

Under STV, the first count gives you a very good idea of who actually will be elected in each district. Thus, overall party seat count could be prognosticated already on election night, by assembling a preliminary guess that would be very close to the final result.

Haven't seen evidence of the use of STV first counts to announce early results in old Canadian STV elections, because things were slower back then and electricity/TV/radio/public expectations were simpler, and perhaps no one thought to do that.

But no reason why the first count in a STV election could not work that way.

And the prognosis would be based on certain or almost-certain election of individual candidates (some of which would already be elected by taking quota on first count). Early MMP results based on overall party seat proportions would not clearly indicate which individual candidates are elected.

And early results announced with such ballyhoo on election night of FPFP elections are just preliminary and subject to change through the count of advance votes, recounts, etc.

STV is basically simple - but it is elegant in its simplicity

In their suspicion of STV, some emphasize PR-STV's complexity, not its simplicity.

The basis of PR-STV is is Single Voting in a MM district, with transferable votes that are transferred if and when needed to be and if they bear back-up preferences marked by voters that are still usable.

Note that not all votes will be transferred - because there is no need or they bear no back-up preferences or the back-up preferences they bear have expired by the recipient already being elected or eliminated, etc.

Any votes bearing first preferences marked for candidate who are elected will not be transferred.

Any back-up preferences are only used if the first-choice candidate is eliminated, or to transfer surplus votes if the candidate is elected. In both cases back-up preferences are consulted and voters transferred only if seats are still not yet filled. They are contingency votes - used only when the first preference proves ineffective.

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

Don’t blame STV for no PR today

If STV has not come to pass, it is likely not the fault of STV being offered as the alternative.

The one BC referendum where a majority of voters voted for Electoral reform was on the question of a shift to STV. This was in 2005. The Government ignored the result.

A majority of voters of BC have never voted for change to MMP. (But MMP or STV would be great. (I guess I am saying not to rule out STV without good grounds, and so far I have not seen any.)

Would ranking of numerous candidates be mandatory?

Under Canada's STV, voters did not have to rank more candidates than they wanted to.

In Australia, they do although I think that country is changing away from mandatory full ranking.

So what NZ sees objectionable in Australia, STV is not anything to worry about in the kind of whole-vote surplus transfer, no mandatory full ranking system as was used in Canada.

STV does not produce exact proportionality

Richard Cartwright, one of the longest-serving MPs in Canadian history, was a big fan of STV. He thought it should be called proportionate representation as it was not exactly proportional.

Under STV, there would not be proportionality to the degree of single digits.

A party with 8 percent of the vote, evenly spread across Canada, would likely not have seats under STV, except in a district that had 11 seats. It would though likely take one seat in every district with 11 seats.So a non-issue.

But currently and historically we see under FPTP:

- a party with 25 percent of the vote getting only two seats (CCF in Alberta in 1944) and

- Conservatives with 30 percent of the vote in Toronto getting only 8 percent of the seats (2021 fed election)* and

- up to 83 percent of the voters in a district getting no representation at all (look at Toronto municipal election 2014, Ward 16).

The need for PR is great. STV or MMP would be great.

(* as per Real's research

Maybe in FPTP people gaze over several districts and say proportionality is okay (Some do, as difficult as it may be to believe.)

But in STV we look at the specific districts where STV is used (or in the city where a city-wide at-large district is used) and see in each election, where party labels are used and the seat count is more than three, that a mixed crop of reps are elected. Multiple parties are represented; and the bulk of voters have hand-picked the successful candidates. No one is elected through party slates alone, no one is elected who does not have more votes than the unsuccessful candidates.

A high proportion of votes are actually used to elect the reps, 75 to 80 percent or more.

That is the kind of proportionality that we see under STV.

Specific stats are available as well.

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

Is STV complex and confusing to voters?

In their suspicion of STV, some look at vote transfers and seem to think that voters find voting under STV to be confusing.

They are mixing together two different things -- how to vote and how to transfer votes as part of the vote count.

Voters don't have to know the mechanics of vote transfers - it is no secret, just not important.

How many can picture how a car engine works, or mail is delivered, or car batteries work, or emails work?

Voting.

In Canada elections voters did not have to mark more preferences than they desired.

Due to this liberty, the difference between voting in FPTP and PR-STV boils down to this:

Under FPTP each voter marks one X and stops

under PR-STV each voter marks “1” (or an X) for his or her first preference and if he or she wants to, the voter can go on to use figures for back-up preferences starting with 2.

Even if no back-up preferences are marked, the vote could go to elect someone - to elect the first preference.

As well, even if back-up preferences are marked, the vote may not be used to elect anyone. Just like under SMP, not all voters can see their choice elected.

Marking back-up preferences does not guarantee the vote will be used to elect someone.

Not marking back-up preferences does not guarantee that the vote will not be used to elect someone.

The presence or absence of back-up preferences does not affect the first count results.

Even with no back-up preference marked by any voters, the result under STV would be different from the winner in a FPTP contest.

The front runners in the 1st Count in a PR-STV contest are different from SMP. Single voting in MM district is very different from single voting in single-member district.

All or most of them will be elected in the end.

What are the front runners?

Look at the list of candidates arranged in order of popularity in the 1st Count, take the same number of top candidates as the number of seats to be filled. Those are the front runners that I mean.

All or most of them will be elected. Only the bottom one or two or so of the front runners are at risk of not being elected, They are at higher risk of not being elecged if the nearest one below them is close to them in vote totals. That was noticed even a hundred years ago.

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

There are different types of STV

Any electoral system is a family of systems. There is not just one MMP system either.

The underlying similarities of all STV systems is Single Voting in a MM district, with transferable votes that are transferred if and when needed to be and if they bear back-up preferences marked by voters that are still usable.

Some say that STV uses complex formulas and the first transfers are already complex - the assumption being that the transfers will be more complex as the vote count progresses.

But the opposite is true - at least the STV vote count process does not become more complex as it deepens.

The point to note is the first transfers are transfers of elected candidates' surpluses.

By definition, there are fewer of these than the number of seats to be filled so there are not a large number of the "complex" transfers in any district contest.

These complex formulas are sometimes presented confusingly but are basically of the form A/B X S. As explained on page 9 of my booket When Canada had PR.

Not so complex at all.

And despite how it is presented, even that "complex" formula is only used if there are more transferable votes than the number of surplus votes, not always the case at all, at least not always the case in historic Canadian STV elections. (That is because in historic Canadian STV elections, voters did not need to rank more than one candidate when casting the ballot. Still even if voters did not mark any back-up preferences at all, the result of STV elections would still be different and more fair than SMP by virtue of Single Voting in MM districts, which is a part of STV.)

Often the surpluses of the final elected candidates are not transferred at all as their election themselves fills the last seats and that ends the vote count. So the "complex" formula is often not used in the case of the filling of those seats.

The other transfers (the transfers of votes formerly held by eliminated candidates, as opposed to transfer of surplus votes) are performed without math other than addition of transferred votes to candidates' running totals. Not complex at all.

As mentioned, voters in historic Canadian STV elections did not need to mark more preferences than they desired. No pressure. No trouble. Still as mentioned, STV elections even without marked back-up preferences elect differently and more fairly than Single-member Plurality (FPTP) does.

STV does ensure local representation

Certainly, local issues are not national issues. But local elections are a part of British tradition. People do think nationally --- and do vote locally.

At least under the present SMP (FPTP), people do vote along party lines. Federal party policy is set at the national level.

However, I am hearing from others that they would like to have choice within a party slate and that was shown on several occasions in historic Canadian STV elections as well.

In any case, in Canada constitutionally we have separate seats within provinces and territories so no national PR is possible.

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

Winning candidate’s surplus votes

There is variation in how different forms of STV deal with the transfer of a winning candidate’s surplus votes.

Most Can. PR-STV elections used the whole vote method of transferring surplus votes. That is the type I mostly discuss.

All STV systems use the same system for transfer of eliminated candidates.

There is scrutiny of the count as it happens. the math of transfers' effect on candidates' running totals can be tested any time. we trust the result of FPTP without seeing the actual votes. We should be willing to do same with STV, and it is not that the "opaque" vote transfers chanfethe order of most of the candidates already listed by vote tallies in the first count.

Legislators establish the process in advance. It is not ad-hoc or last minute thing.

Number of transfers can never be more than number of candidates in the running.

Transfers are of two main types:

A. transfers based strictly on back-up preferences. (Used for transfers of eliminated candidates

(and for transfers of elected candidate's surplus when there are fewer transferable votes than surplus.)

each ballot is consulted to find next usable marked back-up preference, if any.

the vote is moved to the candidate who is marked as next choice of voter.

if no back-up preference, the vote is moved to a pile of exhausted votes (AKA non-transferable votes.

if next back-up preference is marked for someone already elected or eliminated, move on to next preference, if any.

B. fractional vote surplus transfer method

transfers based on back-up preferences modulated by the number of surplus votes to be transferred.

(Used for transfers of elected candidate's surplus when there are more transferable votes than surplus.)

packets of votes marked for each candidate are reduced so that transfers are miniaturized version of all the votes held by the elected candidate. quota left behind is copy in small as well.

One way is to first take out all non-transferable votes.

if there are not more non-transferable votes than the surplus, go to A above.

If there are more transferable votes than surplus, either of two main methods used

whole vote surplus transfer method -- consult only next usable preference if any.

fractional vote transfers - make separate pile for each permutation of preferences marked on candidate's votes, derive fractions and transfer. overly-complicated but work-able. I won't say more on it.

whole vote method

derive X/Y and multiply the number of surplus votes.

X is number of votes held by candidate that are marked next for recipient.

Y is number of total transferable votes held by candidate who won.

Not so difficult at all.

done for years without help of computers or calculators.

Note that this is done for surplus votes of elected candidates so happens at max the same number of times as the number of seats in the district, so say only three times or five times or 8 times or whatever. and that is the maximum possible.

It is possible for this most-complicated of STV transfers to happen only three times in a district electing five.

The method used to fill seats under STV is verifiable

The back-up preferences are consulted one by one each in turn, with no later preference having an impact on earlier stronger preference. The various permutations are not important - not all marked preferences on all votes will be used. used in whole vote surplus transfers are done by looking at net usable backup preference only.

All permutations on the votes held by the eliminated or elected candidate are used to compose transfers in Gregory system, but that is not the type of STV I am talking about - (whole vote surplus transfers - and the end result of the Gregory method is much the same as end result under the whole vote surplus transfer method so the extra work and exactitude of Gregory does not seem worth it to me and to many others (including John D. Hunt who brought STV into use in Alberta in 1924).

PR-STV gets criticized both ways for having too little local representation and sometimes for having too much, by those who want proportionlaity calculated at the national level for example.

But I think STV allows proportionality based on the city-wide districts or at least in multi-seat districts, so is proportional enough while producing local representation (at the city scale).

Multi-seat districts - how many seats?

Five seats is a common number of seats for STV multi-seat districts.

Malta uses five-seat districts and only five-seat districts.

Edmonton and Calgary had five seats when STV was used in those cities in 1926.

But five is no necessary limit.

Ten seats worked well in Winnipeg for 6 elections.

21 seats work well in Legislative Council election in New South Wales in 2018.

Tensets means any candidate with 8 percent of the votes in that city will be elected. Thus that any party whose supporters rank only its party candidates will take at least one seat if it has 8 percent oghe vote.

This would deny representation to any party that has less than eight percent spread evenly across the country, but -

There are no parties with less than 8 percent of support spread evenly across the country. A party with say more than four percent of the votes would likely have at least 8 percent of the vote in one or two cities in Canada, and thus would elect at least two members.

Are there significant parties that don't have at least 9 percent in any city in the country?

And if a city had five seats and all were in a single multi-seat district, any party that has 15 to 20 percent of the district votes will have at least one seat in that district.. That is very fair.

Such a party would take two or more in many cities and perhaps none in others. But Under FPTP a majority of votes in a city, split among two or more parties, may get no seats, while perhaps a minority of perhaps a third of the voters get all the seats.

There are no parties with 10 percent of support spread evenly across the country.

Small parties may get an easier start under STV than under FPTP

Due to transferable votes, people are free to vote for whom they truly want, knowing that if their first choice is un-electable, if back-up preferences are marked and are usable, the vote will (usually) be transferred to assist someone to be elected. That is the assurance that can help get a new party off the ground.

Although PR in Canada based on national figures may be most proportional (as measured by party vote totals and seat totals),PR in Canada cannot be based on national figures. voters in one province cannot influence elections in other provinces, constitutionally.

What if voters don’t want to mark secondary preferences?

Some seem to think that STV demands that voters rank numerous candidates. But many votes don’t know all the candidates and do not want to rank them all.

Under STV as used successfully in Canada in the past, voters are not forced to rank more canaite than they desired.

If STV was broughtinot use in Canada, some voters would want to mark only one or two candidates. Fine.

At worst STV creates no more waste than would happen under FPTP.

Single voting in MM district creates good even without transfers and without any marked back-up preferences.

But where preferences are marked and used if needed, they can be useful to voters to ensure the result is something they are happy with.

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

from The Single Transferable Vote by Nicolaus Tideman Journal of Economic Perspectives—Volume 9, Number 1—Winter 1995—Pages 27–38 The Single Transferable Vote (aeaweb.org) STV is actually an evolving family of vote-counting rules rather than a single rule. The different varieties of STV share the following features: a quota of votes is established, and any candidate who attains the quota is elected; surplus votes of elected candidates are transferred to other candidates favored by those who voted for the elected candidates; candidates are eliminated sequentially and their votes transferred to other candidates, with the candidate eliminated at each stage generally being the one with the fewest current votes. Differences arise with respect to the computation of the quota, whether the surplus votes are ones picked at random or fractions of all votes, the way in which ties are resolved, and whether account is taken of already-elected candidates to whom surplus votes would be transferred if they were not already elected. The central attraction of the single transferable vote compared to other voting procedures is that voters are able to rank candidates in whatever ways they wish, and the vote-counting process sorts the voters into equal-sized groups that are suitably represented by particular candidates (with the possibility that some voters will be split among two or more groups [if fractional transfers are used]. (Explanatory note: In more formal terms, the attraction of the single transferable vote is expressed by the condition that Dummett (1984, p. 282) has called "proportionality for solid coalitions" (PSC). The solid coalition for the set of candidates C consists of those voters who rank all candidates in C ahead of all other candidates. If a vote counting rule satisfies "proportionality for solid coalitions," then for any set of candidates C, whatever percentage of the voters are in the solid coalition for C, at least as large a percentage of the elected candidates (rounded down to an integer) will be from C, as long as C contains at least that many candidates. It is the fact that STV satisfies PSC that justifies describing STV as a system of proportional representation) The single transferable vote selects a set of candidates that replicates the diversity of opinion within the electorate. Systems of proportional representation based on party lists are also supposed to do this. In these systems, voters vote for parties, positions are awarded to parties in proportion to the votes they receive, and individual candidates are then declared elected based on position on party lists. The deficiencies of these systems compared to STV are that they presuppose that what voters care about is captured in party definitions, and they give tremendous power to party officials. When systems of proportional representation are not used, democracies usually select multi-member bodies through single-member districts with plurality voting. This election method tends to create uniformity within the elected body, with diversity arising primarily from whatever variation exists in the characteristics of districts. This method also produces great conflict over the drawing of district boundaries. Sometimes the individual members of multimember bodies are elected "at large," [in one or more MMDs] with each voter usually given more than one vote. [Block Voting] If each voter has as many votes as there are positions to be filled, then a majority coalition is able to secure all of the positions. If each voter has only one vote [in an MMD (SNTV)], then there is significant potential for diversity in the elected body, but there is a great premium on organizing among voters, so that votes are not wasted on candidates who will not be elected. When voters have an intermediate number of votes [(Limited Voting) or where the voter can cast one vote in an MMD but also mark back-up preferences as in STV], the system has characteristics between the two polar cases. The single transferable vote removes much of the need for discipline and strategy in voting. It allows voters to classify candidates in terms of whatever characteristics they believe to be relevant. And it selects a group of elected candidates that reflects the diversity of views in the electorate.

Tideman wrote : As the name "single transferable vote" suggests, STV is a system in which each voter casts one vote, and under prescribed conditions the vote is transferred from one to another of a voter's ranked list of candidates.

I would say As the name "single transferable vote" suggests, STV is a system in which each voter casts one vote, and under prescribed conditions the vote is transferred from being placed on one candidate to being placed on another according to the voter's marked ranking of candidates. Such transfers potentially arise if the vote is placed on a candidate who is declared un-electable due to being the least-popular candidate, or it is placed on a candidate who is elected with a surplus of votes (over and above quota).

Transfers may cross party lines. Votes are only ever transferred to a candidate who is marked as a preference by the voter.

Some votes are never transferred, even if the voter marked back-up preferences.

Even if a voter marks no back-up preferences, the vote may still be used to elect a candidate - if the voter's first preference is elected, the vote is used to elect that candidate.


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

Quota does not apply to Cumulative Voting


Where X voting is used, some people disregard the large portion of wasted votes and try to apply quota same as under CTV.


But if you look at old X CV elections, you see that no quota actually is used, just simple plurality, which means high or low is enough to be elected, whichever, just depending on the opponent's score!


In Illinois State House of Representative elections, in the 1st District, conducted using CV, the proportion of voters taken by successful candidates varied from 39 percent to 19.88 percent in the 1956-1966 period,. The district elected three members so the Droop quota was 20 percent. But there was no such consistency.

(Wiki Illinois State House of Representatives 1st District; "Downloadable Vote Totals". Illinois State Board of Elections. Illinois State Board of Elections. Retrieved 2021-04-11.)


Variation occurs due to the large number of wasted votes and due to unequal distribution of a party's vote over its two candidates.


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





2 views

Recent Posts

See All

Early Labour culture

Clarissa Mackie "Elizabeth's Pride A Labor Day story"    Bellevue Times Dec. 5, 1913

Comments


bottom of page