We all believe that elections should see the elected representatives reflect a majority of the voters. So it is confusing to see legislators elected with just a minority of the votes.
But it happens in many elections that use non-proportional First Past the Post. This is sometimes obscured by the use of the term "majority" when no such thing is happening.
I notice that in New Zealand - yes the same place where very fair MMP is in place, the touchstone of our world-wide movement - it is said that receiving a majority of votes is reason for the election of a candidate when nothing of the sort is meant.
The official results announces that successful candidates won with so much "majority" when what is meant is lead over the nearest competitor.
Such as in the Northland electorate. Willow-Jean Prime is said to have been elected by a majority of 163 but actually she only got 38 percent of the vote, not a majority at all.
Her lead over her nearest contender was 163, but she needed 5000 more votes than she received if she was to get a majority of the valid votes.
In its defence, If there are only two candidates, the winning candidate will take a majority of the votes, and the amount that he or she gets more than 50 percent will be something close to the winner's lead over the other contender.
But where there are more than two candidates, such as in Christchurch electorates or in Invercargill, the winner may not take a majority of the votes at all. Under non-proprotional FPTP, it is common for a candidate to be elected despite not getting a majority of the votes at all.
In Invercargill, for another example, the Wikipedia gives the winner's majority as 224, but she actually did not have a majority at all - Penny Simmonds received less than 45 percent of the vote. It seems that this happened in many districts, just as it happens all too often in all FPTP elections. But usually elsewhere, unlike in NZ, there are no supplemental members to offset the distortions. 37 of the district winners apparently won with less than a majority of the vote in the district. (source: online - Spinoff "A better visual breakdown of the 2020 election results")
Although in each case, each is said to have won with majority of so many hundreds or thousands of votes.
The wide variance needed to be elected in systems that do not use quota A big difference between proportional systems and non-proportional that may be worth stressing is the wide variance needed to be elected in systems that do not use quota.When mere plurality is needed to win, when votes are split in variety of ways from district to district, when voter turn out varies, when size of district varies, there is little standardization, and thus little relationship between vote share or vote tallies to representation from candidate to candidate or party to party.
Say in the last Edmonton election conducted with 12 single-member plurality ward elections, one Councillor was elected with less than 3500 votes; another with almost 12,000 votes.As well, there were nine people who received more than 3500 votes who were not elected. This includes one who got more than 5200 votes, half again more than the least-popular successful candidate.
(Other problems: more than a third of the time - a candidate with the support of only a minority of district voters is elected and no others are, thus leaving the majority in the district un-represented. These wasted votes sometimes exceed two-thirds of the votes cast in a district and is seldom less than a third of the votes cast.) Alberta provincially 2019
Premier Rachel Notley, one of the most popular candidate anywhere, got 9500 votes, 65 percent of the votes cast in the district.
In a different district, UCP leader Kenny, who would replace Notley as premier, got 12,000 votes, 66 percent of the vote.
In Athabasca... the UCP candidate did even better 17,000 votes, 69 percent.
In Innisfail the UCP candidate got 19,000 votes, 75 percent
Olds-Didsbury the successful candidate (UCP) got 21,000 votes, 79 percent of the vote
But meanwhile compared to these titans,
In Calgary Falconridgem, the successful candidate (UCP) got 6800 votes, 48 percent of the district vote.
In Calgary McCall, the successful candidate (NDP) got 6600 votes, 52 percent of the vote in the district.
in the 2015 Alberta provincial election, Calgary Buffalo was a slow race. 4602 was enough to win, while in Calgary Glenmore it was a fast race where 7015 votes were not enough to win.
Federally, there is wide variation as well under non-proportional FPTP. One riding in Saskatchewan, Desneth, had only 26,000 votes cast in 2019 while the average riding in Toronto had 50,000 votes cast. SMALL (sparsely-populated) rural ridings (but large in size)
In Desneth, SK 12,000 votes (42 p.c.) was enough to win.
Labrador (NL) (extremely rural -- 294,000 square kilometres in size) 12,000 votes cast. 4900 votes was enough to be elected (43 p.c.)
St. Laurent, Quebec (rural 42 square kilometres) 41,000 votes cast. the successful candidate got 24,000 votes (59 percent)
LARGE (urban) ridings
(miniscule in size - sometimes less than 6 square kilometres in land area)Peterborough 70,000 votes cast. the successful candidate got 27,000 votes (39 percent) Toronto Centre (6 square kilometres) 55,000 votes cast. the successful candidate got 31,000 votes (57 percent)
Toronto St. Pauls 60,00 votes cast. the successful candidate got 32,000 votes (54 percent)
Calgary Shepard 79,000 votes cast. the successful candidate got 59,000 votes (75 percent) (Note that these are size-sorted by urban and rural, not East or West, although the North is more rural than the South. Alberta, oddly enough, is one of the more urbanized provinces, so is actually under-represented by population across the country.)
Scientific voting
It is interesting to notice the contrast between St. Laurent and Peterborough.the two successful candidates received about the same number of votes; each took one seat,but one was elected with a large majority of the district votes, the other with a minority of the district votes.In one the majority of the district voters was represented; in the other the majority was un-represented and disregarded. It is comparisons like this that make the need for scientific representation, as PR used to be occasionally called, so important.
In STV, by comparison, the variance is much less.
There are no variance in the size of the arbitrary districts. Voters form themselves into similarly-sized constituencies to each elect its own rep.
And there is no variance in turnout. At least none is measured if the area is a single grouped district.a single standard quota is used across the broader district, usually a city-wide district.and that quota is used to declare most candidates elected, a couple or so possibly being elected at the end by mere survival.
in five-member STV district, the quota would be 17 percent.
likely three members would have that number of votes and be declared elected. They would each have this exact number, no more and no less, any surplus being transferred away.The last two would have something less than that, but likely more than half the quota each.
The exhausted votes make exact prediction impossible.
But these basic facts are known:
With three seats filled, half the votes have been used. 3 times 17 percent
The remaining half of the valid votes (minus any exhausted votes) would be spread over the remaining candidates, as they steadily are eliminated and the vote transferred.
At the end this might come down to the remaining votes in play being gathered on three candidates, mostly on the two most popular of these. So at least two thirds of the remaining votes would be gathered on these two, who would be elected.
A pretty standard structure of vote tallies compared to what happens under non-proportional winner-take-all FPTP.
And note that about 80 to 90 percent of the votes (barring exhausted votes) were used effectively to elect someone. and note that the proportion that were used to elect the representatives is likely far more than 70 percent of the vote (exhausted votes seldom exceeding 20 percent of the votes), and note that the proportion used to elect the representatives is never less than a majority.
This is a far cry from the result possible under non-proportional FPTP where often 65 percent of the votes are ignored.
Regionalism
Often a city give its seats to candidates of just one party --- say Calgary to Conservatives and Edmonton to NDP --- while the large minority of supporters of the other party in each city has little or none representation.
And to think that Edmonton Conservatives can count on sympathetic treatment by Calgary MLAs or that Calgary NDP-ers can count on sympathetic treatment by Edmonton MLAs is putting a lot of trust in that party (or class) connection.
or Montreal with its Liberal reps, Toronto with its NDP reps (in the old days I would have said Conservative), etc.or federally when Alberta elected 33 Conservatives (all but one of its MPs), while BC elected 17 Conservatives (out of its 42 MPs), or Quebec elected only 10 Conservatives (out of 78 of the province's MPs).
District non-proportionality breeds regionalism
FPTP is non-proportional at the district level. Through natural process (weakness begetting weakness and strength begetting strength; disregarded voters staying home, happy voters participating in greater number) and to un-natural processes (such as gerrymandering and party allocation of resources), FPTP produces regional disparities.
Multi-member districts (such as under STV) produce mixed representation in each district (that is, in each city and in each province and in each region). And it thus prevents regional disparity --- naturally, organically and synergistically.
===============================================
Addition added in August 2023:
Equal representation does not mean equal-sized districts
Looking at Chartists in 1838 Britain, they asked for "equal representation." Apparently they wanted districts of equal sizes, each electing just one member.
I am frankly surprised such revolutionaries/reformers/utopians wanted so little as that. In those days there were many MMDs and some SMDs with some SMDs larger than some MMDs, so having equal-sized districts with consistent DM would have meant inequality.
Doubtless their goal was for each voter to have due representation --
if Chartist candidates take ten percent of the vote, they should get ten percent of the seats - that is equal representation.
and it seems they thought that would be achieved through equal sized FPTP districts. (we have only to look at Canadian elections to see the fallacy of that belief)
Being able to vote in a FPTP district of equal sizes would have likely not given the Chartist movement its due share of seats, just as it denies fair number of seats to the NDP almost always and Greens almost totally since forever, and the People's Party as well in the last two elections, so it is hardly what they were seeking -
but that district-equality is the only fairness that FPTP can do. (but not what MMDs (even two-seat districts) could do - with fair voting
FPTP cannot guarantee minority representation.
FPTP cannot even guarantee that each winner will have same number of votes because distribution of the vote in the district varies, and does not guarantee that each district will have same number of votes because turn-out varies.
FPTP does not guarantee that each successful candidate will have same percentage of votes even.(even IRV simply gives a minimum percentage - must have at least 50 percent of votes, votes still in play at that point in time not necessarily votes cast.
under FPTP
percentage of votes varies from 18 percent (2014 Toronto city election) to 82 percent
under FPTP in 2021 election
winner's percentage varied from 29.5 percent to 76 percent (Souris Moose Mountain)
winner's number of votes in last federal elections varies from 44,000 to 13,000 in standard districts (much lower than that in Territories or far north, I think)
so even equal-sized district is not foolproof at ensuring winners are elected with same number of votes.
New York city when it used STV 1937-1945 used the boroughs as districts - so no arbitrary districts - and made fairness by allocating different number of seats to each borough - not as per population but as per voter turn-out
so that is fairness to the "almost-nth" degree.
(the nth degree (in practical terms) would have been city-wide list PR and a DM of 100 or more.)
As the number of seats varied with voter turnout (as much as was allowed by gradients of 50,000 to 75,000 votes), so the number of votes that were quota stayed pretty constant (at about 65,000 by my reckoning, assuming Droop quota was used)
using voter turnout to set seat count meant that quota was about the same in each borough despite different number of votes in each.
the vote count no matter what it was was divided by number of seats plus one (if Droop was used) (I think so anyway - the exact form of STV used is still murky)
and so in 1937 you had something like this
I know the seat count for each borough but not the vote total. I assume Droop quota was used but if Hare was used, number would vary but there would be almost the same range of quota.):
Richmond one seat 75,000 to 125,000 votes (actually this is IRV as there is only a single winner) quota is 50 percent plus one, so 37,500 to 62,000
Queens and the Bronx each with five seats350,000 to 425,000 votes quota was between 58,333 and 70,833.
Manhattan six seats 425,000 to 500,000 votes quota was between 60,714 and 71,428.
Brooklyn nine seats
650,000 to 725,000 votes. quota was between 65,000 and 72,500.
so we see that irrespective of vote tallies and seat counts in each district, the number of votes that guaranteed success ranged from just 58,333 to 72,500.a factor of 1 to 1.24.
as each member has an equal one vote in the chamber, it makes sense for each to be elected with the same number of votes.
That is also really the only way to ensure that each party gets its fair due.
In New York, each party got its fair number of seats based on that organically-uniform rate (quota).
while the quota for a seat stayed the same from election to election, voter turn-out varied from election to election so the number of city councillors on council under this system varied from term to term depending on voter turnout.
However 24 years earlier, John H. Humphreys, part of the brain trust of the PR movement at the time,
pointed out how a uniform quota could have drawbacks and produce disproportional results.
Humphreys discusses the drawback of using a uniform quota (say 1000 votes) when number of valid votes vary from district to district in the Mercury, Oct. 21, 1913
THE LIMITED APPLICATION OF THE UNIFORM QUOTA. - To the Editor of Sir,—The House of Assembly in declining,to limit the application of the uniform quota has, 1 venture to suggest,...
in Humphrey's example
district seats varied from 5 to 7
districts' votes totals vary from 5000 to 7000 (much less than the range we see under FPTP)
party totals were 2000, 4000, 12,000 and 14,000.
party totals in districts varied from 100 to 3100.
In his example,
uniform quota (1000, actually 1001 as back then (apparently) you had to pass quota to be elected) and
use of the electoral threshold (apparently requirement to have one district seat to be eligible for the second allotment) was taken as given),
this situation produced party dis-proportionality.
despite fair "averages" system used to set seats after "first allotment" (equivalent of the overall top-up),
the point being, districting caused vote splitting and the uniform quota did not make sense.
2000, about a third of votes in a district with 6 seats, gave one party just one seat.
But that is pretty artificial result -- that vote tally is just two votes less than two quotas.
And in another district the same party got just one vote less than one quota.
thus almost three times quota produced one seat.
this apparently was what PR was about in old days -- discussing whether uniform quota or STV's organic quota makes more sense (the question of a set number of votes versus a set fraction of the vote)
we are at least past that -- I think no one is saying use uniform quota, and district DM is taken as being set in advance - no need for the fluctuating district size method of old New York City. (perhaps NYC's strict STV added fuel to the anti-STV/anti-PR feeling that gradually formed a majority in the city.)
STV -- normally now -- does not use "uniform quota" -
the quota used is a set ratio or fraction --
in a five-seat district, for example, Droop quota is one sixth of the votes, plus one, however many that is
And we see that STV as used today produces much more fair results than FPTP, although not as cleanly as the fluctuating seat count system in old New York City.
in 2020 Ireland election
in five member districts:
highest quota 12,992 Tipperary lowest quota 10,057 Galway West lowest vote tally of successful candidate 8340 Dublin Fingal (this is the final count, not that candidate's first count tally) the range under STV in five-seat districts was a factor of about 1 to 1.5. Examples of districts of DM of 3 and 4 (I don't have time to look at each district) three-member districts Cork SW quota 11,085 least-popular successful candidate had 10,078 Cork NW quota 11,593 least-popular successful candidate had 11,173 Galway East quota 10.361 least-popular successful candidate had 10,022 four-member districts Dun Laghoire quota 12,459 least-popular successful candidate 11,071 Limerick City quota 9226 least-popular successful candidate had 8207 Mayo quota 12,871 least-popular successful candidate had 10,977 The range of these districts was from 8340 as the least for a successful candidate to 12,992 as the largest vote tally taken by a successful candidate (this is the vote tally counted for the candidate after any surplus votes were taken away from him/her and his/her vote tally was lowered to the quota in that district.) Thus the range here too was a factor of about 1 to 1.5. The number of votes received among successful candidates is much more consistent than the range under FPTP in Canada's last federal election. Canada's federal 2021 election conducted using FPTP successful candidates largest vote tally 44,456 Foothills (Alberta ) lowest vote tally (in provinces) 4119 Labrador (NL) the range under FPTP was a ratio of 1 to 11. This range was as large as it was despite each district having the same number of members - one member each - and each having about the same number of voters, and thus theoretically having about the same number of votes cast. The percentage of the votes cast received by successful candidates varied from 76 percent (Souris Moose Mountain) to 29 percent (Trois Riviere). This was due to plurality being enough to be elected and the vote structure in each district varying widely due to vote splitting, different numbers of candidates competing, etc. while under STV in Ireland, quota - as a percentage of votes and as number of votes - varied much less: five-seat district 17 percent (a median of about 11,500 votes) four-seat district 20 percent (a median of about 11,000 votes) three-seat district 25 percent (a median of about 11,000 votes). The number of seats in the district varied, the number of votes cast varied but the function of quota meant that quota in each district was about the same number of votes irrespective of size or voter turnout. However while it is clear that normal STV today produces much more fair results, perhaps there is reason to use old New York City's organic (fluctuating) seat count system. If Canada switches to RUPR, where cities are multi-member districts and institutes fair voting in each city (STV or SNTV or list PR),. the voter turn-out will be a large unknown - will more voters get out and vote under fair system? would turn-out stay at about 50-65 percent or dramatically increase? will boundary changes affect turn out? so bearing that in mind, guessing at total vote tallies in each district will be difficult but essential if seat count is to reflect votes cast, a necessity if quota is to be about the same from district to district. so we could use the NY city method (at least for first couple RUPR elections until things settled out): after votes are counted (the first count), in each province separately a quota could be established - the number of votes cast in the province divided by number of seats in the province (established by the Constitution) say a 16 percent fluctuation up and down allowed for cities versus rural ridings - so 5 to seven ratio. (or the votes could be counted in two groups -- those cast in rural districts and those cast in city districts -- and average (quota) established for each using 5 to 7 ratio of rural versus urban. that would set two quota, one for rural districts and one for city districts. each district would be guaranteed to have at least one seat. each district would be given additional seats based on the quota. with "STV quota" calculated for each district as per vote total in the district. that would encourage people to vote as the district would have more rep if it had more votes cast (within limits of seats available) because rep should not be about representing an area or a city (irrespective of votes cast) but representing the voters who cast votes within the area or city. and party proportionality depends on each member being elected with about same number of votes. because even if were have fair voting and use districts, there would be some degree of unfairness if winners were elected with different number of votes from district to district. say for example in Alberta ridings (in 2021) Banff voter turn-out was 77,000 Foothills 64,000 Red Deer- Mountain 63,000 Red Deer 62,000 Lethbridge 59,000 Battle River 59,000 Lakeland 53,000 Bow river 51,000 Yellowhead 51,000 Medicine Hat 48,000 Peace River 47,000 Fort McMurray 43,000 Grande Prairie Mackenzie 43,000 so even just in Alberta rural ridings there is wide variation in votes, from 43,000 to 77,000. all the seats listed here were taken by candidates of one party so there is no party proportionality issue but there is a fairness issue, and if there comes a time when the Conservatives do not take all the rural seats again and again, then the wide variation might cause disproportionality. say if Liberal or NDP voters take one seat with more than half of 77,000 versus a Conservative taking a different seat with just more than half of 44,000, or visa versa. not to mention taking a seat with just 24 percent of the votes in a 43,000-vote district versus 76 percent in a 77,000 vote district, as can happen under FPTP.
Thanks for reading.
==================================
Comments