Here's a look at how elections really work...
First Past The Post one seat to fill
say there are only two candidates running for the seat --
A takes 54 percent of votes and takes the seat
B takes 46 percent of the votes and is not elected.
majority of voters are satisfied with the result,
54 percent of valid votes are used to elect the winner.
But many districts = scope for dis-proportionality
With one seat to fill in each district, the electorate is divided into maximum number of districts.
Districts may not be same size (contain the same number of people) but even if they are the same size, voter turnout is likely to vary from district to district.
Even if number is the same and voter turnout is the same and only two candidates run for the seat in each district, the percentage of votes received by the winner varies from district to district.
so the result may be A 54 percent of votes cast, B 46 percent of votes cast
or it might be A 70 percent of vote cast, B 30 percent of votes cast.
Even if we assume districts have same number of votes cast (which is not the case in real life), the variance in winners' percentage makes for wide range of vote tallies.
I'll put this into terms of concrete numbers of votes, through a simple model.
A 1,00,000-vote election model
Let's say there are 100 elected members in the chamber, and thus under FPTP there would be 100 districts.
And let's pretend that in each district, 10,000 votes are cast.
(1M votes cast overall)
A can win 51 percent of the districts with 275,000 votes if he takes 54 percent in each district where he wins (51 X 5400 votes)
or
A can win 60 percent of the districts with 420,000 votes if he takes 70 percent in each district where he wins. (60 X 7,000)
If party A takes 300,000 votes, it may win majority government (275,000 is enough to take 51 seats)
or it may take only 40 seats if its voters are packed into just those districts
or it may take 20 seats with 6,500 in each and have its other 170,000 votes spread without chance of victory over the remaining 80 districts.
So even if only two candidates run in each district, there are all kinds of variability possible in election outcomes.
But seldom do only two candidates run to fill a seat.
so the result in the district may be:
Party A 70 percent of votes cast B 14 C 10 D 5 E 1 percent
or
it may be Party A 18 percent of votes cast B 17 C 16 D 15 E 12 F 10 G 6 H 6 percent.
In both cases A is the winner.
18 percent of votes in a district is only 1800 votes so with such a winning vote tally repeated across many districts, it is possible to get majority government -- 51 seats in a 100-seat chamber, with 1M votes cast -- with only 100,000 votes.
in such a worst-case scenario, it is possible for party B to win the other 49 seats say with wide majorities in each of them, perhaps 70 percent in each, making a total of 350,000 votes, plus just a bit less than 18 percent in the other 51 districts - an additional 100,000 votes.
so overall Party B may receive 450,000 votes and yet be denied government by Party A which has taken much fewer votes, perhaps as few as only 100,000.
But districts are not the same size - the policy on number of residents may be plus or minus 20 percent so size ratio may be 80 to 120.
And voter turnout fluctuates from district to district.
and just like above, a winner in a district can win with anything from 18 to 82 percent.
A range from 29 percent to 72 percent is common.
When you combine unequal sizes of districts, variable voter turnout and winner percentages that normally range from 29 percent to 72 percent, you can see how in almost every election a party can take most of the seats in the House of Commons or in any provincial legislature with less than half the valid votes.
Unlike the First Past The Post contest shown at the top of this page, where two candidates run to fill one seat, in a general election when party A takes most of the seats without taking a majority of votes, it is not proven that party A is the choice of the majority of voters.
When perhaps candidates of five parties, plus independent candidates, run in all or many of the various districts and the winner is often elected with less than a majority of votes in the district, and no party takes a majority of votes cast overall, it is not clear that Party A is the choice of most voters even if it takes majority of seats.
In fact if the choice for the voter is reduced to just two parties, Party A and Party B, then it might very well be that Party B is the choice of a majority of voters, despite the fact that Party A might take a majority of seats.
The election method is partly to blame for Party A getting majority of seats with just 200,000 votes. A district winner receiving less than half the votes cast in a district, perhaps as few as 18 percent, is not a fair election system.
The districting is at least patly to blame, and not just due to variation in population size between the different districts.
Variations in voter turnout are to be expected whenever the electorate are divided into separate groups.
But if winners of Party A are elected with fewer votes than winners of Party B, Party A will take more seats than is its due based on votes received, and Party B will take fewer seats than is its due.
In each district where only one seats is filled, the winner is elected and all other votes are cast aside, electing no one.
But if there is just one district covering the whole electorate, if each voter has just one vote and if votes are cast efficiently, you will have more fairness to each party.
Say the 1M voters are collected in various voting blocks of unanimous sentiment, and each of the 100 seats are allocated fairly.
Any party with 10,000 votes will take a seat and will take an additional seat for each 10,000 votes it has more than 10,000.
Party B with 450,000 votes (as mentioned above) will take 45 seats.
Party A with 100,000 (as mentioned above) will take 10 seats.
Other parties will also take seats based on their vote total divided by 10,000.
But if we want the elected members to know the situation within their own areas, lumping all the electorate into one district may not work.
So perhaps there would be 20 districts and each district elects five members.
And let's assume fair voting is used - PR.
Each voter has just one vote and votes are cast efficiently, you are still likely to to have good fairness to each party even if the electorate is now divided into 20 districts.
If each district is the same size and 50,000 valid votes are cast in each, any party with 10,000 votes will take a seat and will take additional seat for each 10,000 it has more than 10,000.
Party B with 450,000 votes (as mentioned above), 22,000 in each district, will take 40 seats.
Party A with 200,000 (as mentioned above), 10,000 in each, will take 20 seats.
Other parties will also take seats based on their vote total divided by 10,000. Some variation will be created when parties have vote tallies of less than 10,000 or odd fractions of 10,000.
But if we want voters to vote for candidates, each party will run two or more candidates in each district. The votes cast are not only split up among the different parties but also among candidates of the same party. To prevent votes from being wasted and to allow parties to take their due share of seats, the type of vote used can be transferable.
Such is done in the system known as Single Transferable Voting (STV).
To be guaranteed election, a candidate must take number of vote equal to quota. Quota (Droop) in a five-seat district would be 1/6th of votes cast so with 50,000 cast in a district, quota would be 8333.
If a candidate has more votes than quota, his or her surplus is passed onto the candidate marked as the next preference. The vote is not going to go to the voter's first choice but will go to someone else whom the voter endorses.
As well, if a candidate is unelectable, the vote can be transferred to the next usable preference marked on the ballot.
The most-popular candidates are elected, proven by surpassing quota when other candidates did not, or by simple plurality at the end of the vote counting process.
Due to electing say five members in a district, about half the votes will be initially cast for the candidates who are elected in the end.
An additional 30 or so percent of the valid votes will be transferred from less-popular candidates to more-popular candidates who are elected.
Thus 80 percent of valid votes, or more or less, are used to actually elect someone, not always the voter's first choice but someone whom the voter marked a choice for.
Under party-list PR, the largest parties win seats in large numbers and all but the smallest parties win some number of seats, so again most votes (about 80 percent or more) are used to elect someone.
Under FPTP, as we saw above, parties may take much more or much less than their due share of seats.
As well, in each district, the winner often takes less than half the votes, with votes for all other candidates not being used to elect anyone. So under FPTP it is common for only half the valid votes or less to be used to elect someone.
The other votes are wasted, and those wasted votes are not drawn from each party equally, thence arises potential for a party to take much fewer seats than its due.
And thus for a majority of voters to be dis-satisfied with the election result.
Proportional Representation depends on:
- one overall district being used or as few districts as possible
- multiple members being elected in the district
- each voter having just one vote.
Increased fairness can be accomplished by use of:
- transferable votes if voters vote directly for candidate of choice, as in STV
- large District Magnitude, if 100 members are elected in a single contest, any party with about 1 percent of votes could be accorded a seat. The imposition of an electoral threshold could however restrict this fairness.
Or parties could be able to receive votes from other parties if each does not have enough to get their due share of seats, if voters vote directly for party of choice as in party-list PR, such as through a coalition receiving combined parties' votes.
Under open-list list PR, voters have choice of voting for candidate of choice or party of choice. Even without transferable votes and "sharing" of votes between parties, there is more fairness than under FPTP.
Block Voting is not a fair election system
Multi-member districts are used sometimes without PR or STV. In many of these cases, each voter is allowed to cast multiple votes. The party that has the most supporters in the district usually takes all the seat in the district with votes cast for others being cast aside. This system, called Block Voting, suffers the same bad characteristics as FPTP.
SNTV
With multi-member districts, if each voter casts just one vote and it is not transferable, as in the system called Single Non-Transferable Voting (SNTV), some votes are wasted, perhaps as many as under FPTP, but each district, usually a city or like area, elects a range of representatives, so most voters have someone elected in their district whom they feel represents their sentiment, even if their vote itself was not used to elect anyone.
STV
With multi-member districts, if each voter casts just one vote and it is transferable when found to be cast for someone who is overly popular or unelectable, you have the system called Single Transferable Voting (STV). Where the district has four or more members, always a range of representation is elected and most voters have someone elected in their district whom they feel represents their sentiment. As well, 80 percent or so of the votes cast are used to elect someone so parties get their due share of seats, within the district and also across the electorate if STV is used in all the election contests.
List PR can be used at the district level or overall or both. Voting for individual candidates may be allowed as under open-list list PR, or voter may have just choice of parties as under closed-list PR. For many voters, voting for a party of choice may be just as good as voting for an individual candidate.
Here is a list of election systems described above:
First Past The Post - used currently in Canada in federal elections and each province.
Block Voting - used extensively at the federal and provincial levels in the past.
SNTV - never used in Canadian government elections, used in former British Dominion Vanuatu
STV - used to elect MLAs in Alberta and Manitoba from 1920s to 1950s
Open-list list PR - never used in Canadian government elections, used in many European countries
Closed-list list PR - never used in Canadian government elections, used in many European countries.
As well, Mixed Member Proportional is a combination of FPTP and Closed-list PR. after district results are announced, top-up is allocated according to party vote shares across the country.
In the above discussion, I have looked at elections as five levels:
the vote cast by the voter -
is the vote transferred as under STV?
the result in the district or districts as per votes cast
are additional seats filled such as under MMP?
the make-up of the chamber - which party or group of parties wields majority control to pass legislation?
Confusion is created where people think to win a district seat a candidate must have majority of votes when the requirement to have a majority is true to pass legislation in a chamber.
Or where people think for a party to win majority of seats it must have majority of votes across the whole electorate;
Or when people think a voter may be satisfied with the result in his or her district even if their vote itself was not used to elect someone - the only way to know is to give the voter a full range of candidates to choose from and to have a voting system that gives the voter liberty to vote for whom they want,
either by the ability to transfer the vote if found to be cast ineffectively as under STV, or for the percentage of votes cast ineffectively to be as few as possible as under list PR.
Those are some of my ideas anyway.
Hope they were of interest.
=============
Comments