Suggestions for Conducting
A DEMONSTRATION ELECTION
Using Single Transferable Voting (Proportional Representation)
Preparations
Count several batches of ballots for practice before undertaking to carry out even a demonstration election.
Necessary Materials
A large blackboard.
Sheets of plain paper for ballots. Scratchpads, about 7cm x 11cm in size, can be used.
Provide no more than 125 ballots regardless of the size of the audience.
Arrangements
If possible, have the blackboard easily accessible from the front row of seats occupied by the audience. If you are to speak from a platform separated from the audience, put the blackboard on the platform and also as many chairs as there are to be candidates in the election.
Introductory Remarks
Include in your introductory remarks, which should be brief, such matters as the following:
(1) The purpose of proportional representation is to elect a legislative or policy-determining body truly representative of all who vote to elect it, each like-minded group among the voters receiving the same share of the members elected that it has of the votes cast.
(2) Such a result cannot be brought about if any voter is allowed to help elect more than one candidate. If, for example, each voter is allowed to have his ballot counted for as many candidates as there are seats to be filled, the largest organized group — frequently not a majority — will usually be able to elect its entire slate, depriving the rest of the voters of all representation whatever. The existing First Past The Post system with single-member districts — our commonest method of electing representatives – is correct in giving voters only one vote. However it forms the wrong kind of constituencies behind the candidates. It puts together a group of people who live together instead of a group of people who want the same representative, so our existing method of electing representatives is wrong.
(3) In order to secure the correct kind of constituency the single-member district must be abolished. If the municipality electing the representative body is not too large, the entire municipality should be polled as one district. If the community is too large for that, it should be cut up into several large districts, each of which is to elect several members. The voter still retains the one vote. Each casts just one vote in a district that elects several members.
(4) Explain to what sort of body members are to be elected by the ballots about to be marked. It is usually best to elect a national representative body, for example delegates to represent Canada in an international conference or members of an imaginary council to govern the country. The election of a body for which persons in the audience will be suitable candidates may sometimes be advisable, but there is often too much danger of arousing personal feeling and thus distracting attention from the principle of the election system.
Distribute the ballots. If there are fewer than 25 in the audience, give each person two ballots with permission to impersonate two voters.
Nominations
In regard to nominations, bring out the following points. When the STV system is used, preliminary nomination contests are entirely unnecessary. In only a single election the STV system not only gives the correct number of seats to each party, where parties exist, but also determines which candidates in each party are preferred by its voters.
For the demonstration election ask for people to volunteer to serve as candidates.
Call for nominations from the floor.
As people volunteer, write their names very plainly at the left-hand edge of the blackboard. (They might identify themselves in pretend or actual parties.)
Five is a good number of spots to fill; nine is a good number of candidates. But if it seems important to permit one additional nomination, permit it. The greater the difference between the number of candidates nominated and the number of candidates to be elected, the longer it will take to count the ballots.
Two methods of voting:
1. Ask the voters to copy the names on their ballots as fast as they are written on the board and in the same order. They should mark their preferences with figures.
2. Voters should write the candidates' names in order of preference.
The indication of choices by figures is in accordance with general usage in actual elections that use the STV system.
Either way, ask the voters to copy their choices on to another piece of paper. This will be useful later (in the "Testing the results" section below).
======================
Directions to Voters for Voting Method Number 1.
Explain the method of voting, about as follows:
Put the figure 1 opposite the name of your first choice. If you want to express also second, third, and other back-up choices, do so by putting the figure 2 opposite the name of your second choice, the figure 3 opposite the name of your third choice, and so on. You may express thus as many choices as you please, without any regard to the number being elected.
Your ballot will be counted for your first choice if it can help him. If it cannot help him, it will be transferred to the first of your choices whom it can help.
Directions to Voters for Voting Method Number 2.
Voters indicate their choices by the order in which they write the names of candidates on the ballot. This method makes the transfer of ballots a little easier. Nothing should be written on the ballots until the nominations are finished.
==============================
Inform voters at time of voting:
You cannot hurt any of your favorites by marking back-up preferences for others. The more choices you express, the surer you are to have your ballot count for one of them. But do not feel obliged to express choices that you do not really have.
If figures are used, a ballot is spoiled if the figure 1 is put opposite more than one name. Same for the other numbers.
Give an opportunity for questions about the method of voting only. Ask the voters not to fold the ballots. While voters are voting, start ruling rows on the blackboard next to the candidates' names.
The Count of First Choices
Have the ballots collected and brought to you.
If the blackboard is on the platform, separated from the audience, have enough persons occupy the row of chairs on the platform to represent each candidate as "tally clerks." If the blackboard is down in front of the audience, ask persons in the first row of seats to serve thus.
Beginning at one end of the row of the tally clerks, ask the first of them to receive ballots for the first candidate, the second to receive ballots for the second candidate, etc.
Explain:
you are going to distribute the ballots according to the first choices marked, paying no attention at present to any other choice that may be indicated. Add that in a public election this preliminary sorting is usually done at the several voting precincts, later vote transfers would be carried out by staff at a central counting place, where all the ballots are assembled and where the candidates, reporters, and others may watch every step.
Distribute the ballots to the tally clerks, each according to the voter's first choice. Ask the tally clerks to assist you by answering to their candidates' names and by examining each ballot received to see that no mistake is made. If you think best, have one or two others help you in this distribution.
If any ballots are invalid, tell the audience how they were spoiled and lay them in a separate pile.
When all the ballots have been sorted, ask the tally clerks to count their ballots carefully.
Have them report the numbers received for their candidates as you read the names from the blackboard. As each number is given, enter it after the name on the blackboard.
Ask the tally clerks to recount their ballots or have the count of each one checked by another.
Add the column of first choices, asking the audience to check your work. In this operation disregard invalid ballots. Enter the total number of valid ballots. at the bottom of the column.
The Quota
Next, determine the quota on the blackboard.
Explain each step carefully to the audience. The points treated in the following sample explanation may well be included.
We must now determine how many ballots or votes are required to ensure a candidate's election. The number of candidates to be elected is 5. The number of valid ballots cast is 70. Each ballot counts as one, and only one, vote. It is obvious that if any candidate receives a full fifth of the 70 votes, that is, 14, he is sure of being one of the five strongest and therefore sure of election.
But would not a somewhat smaller number ensure a candidate's election? Suppose a candidate has received barely more than one-sixth of the 70 votes. Will not even that number ensure his election? Let us test it. A sixth of 70 is 11 2/3. 12 is therefore just more than a sixth of the whole number of valid ballots. Suppose, then, that a candidate has received 12 ballots: is he sure of election? He is, for though four others might receive 12 votes each, that would use up, in all, 60 votes, so that there would be only 10 votes left for any sixth candidate. A candidate who received 12 votes would therefore be sure of being one of the five most popular and therefore sure of being elected. A candidate with only 11 votes, on the other hand, would not be sure of being one of the 5 highest, for 6 candidates might each get 11 of the 70 votes cast.
12, the smallest number that will certainly elect, is called the quota.
Here we were electing five and found the quota to be just more than one-sixth of the total vote. If we had been electing 9 we should have found it to be just more than one-tenth. If we had been filling a single office,we should have found it to be just more than half, or a majority.
The quota is one more than the result of dividing the number of valid ballots by one more than the number of seats to be filled.'It would not be fair to voters to take a full 5th of 70 as the quota: no voter wants his ballot to accomplish nothing; and since any candidate who has received 12 votes must inevitably be elected, a 13th ballot put in his pile would accomplish nothing.
Transfer of Surplus Ballots
Run down the column of first-choice votes received by the candidates and put a ring around the numbers that equal or exceed the quota. Announce that the candidates thus indicated, having received the quota, are elected.
Suppose, for example, that only one candidate, C, has received more than the quota, and that he has received 22 ballots.
Explain:
The 12 voters whose ballots happen to be left to elect C are treated justly, for they are represented by their first choice. And the 10 voters whose ballots happen to be taken for transfer are treated justly: having had no share in the election of C — who, however, has been elected by others — they are given the opportunity of helping elect later choices instead. The general result will be the election of candidates who represent the voters truly.
Explain the next step somewhat as follows :C required for election only 12 votes. He has received 22. It is evident that the 10 ballots that he has received in addition to the 12 required would be wasted if allowed to remain in his pile. The voters who cast these surplus 10 ballots for C are delighted, of course, to have C elected. But if each of them were present and allowed to act for himself he would not let his ballot stay with C; he would want it transferred to the next back-up preference — provided, of course, that candidate could be helped to election by his ballot. Ask C's tally clerk to count off 10 of C's ballots from the top of his pile and hand them to you for distribution. That will leave 12, the quota, in his hands to elect C. Count the ballots yourself to make sure you have the correct number.
At this point someone may ask what right you have to take the particular 10 ballots you have taken rather than 10 others. Whether this question is asked or not explain briefly that you will answer this question fully as soon as you have transferred the ballots: you want those present first to see how the transfers are made. (The answer is in "The Element of Chance" section below.)
Explain that votes should be transferred to those who have already been elected. Request the tally clerks to let you know the moment their candidate has achieved quota. Request the tally clerks to keep the ballots they receive by transfer separate, for the present, from the ballots received in the previous sorting.
Read off the back-up preferences from one of the surplus ballots in order until you reach one who is not yet elected. Pass the ballot to the tally clerk of that candidate, reminding him to keep it separate from those he has already. Do the same with the other ballots. Have someone help you in this distribution if it seems best. It is not wise, however, to have more than one person read off the choices aloud.
Ask the tally clerks to look carefully at every ballot handed to him and hand back any that are not marked for his candidate as next choice.
Explain to the audience that in every case you are simply following the voter's marked instructions and passing the ballot to the candidate whom he wants to help under the circumstances.
When you come to the first ballot that is marked with the figure 2 for a candidate who has already received the quota, read the name of the second choice aloud, point out that that candidate does not need the ballot, and then read the third choice and if necessary the fourth, etc., until you reach the name of the candidate to whom the ballot should be passed.
If one of the surplus ballots shows no effective back-up preference, none marked for a candidate who is not yet elected, give it back to the tally clerk from whom you took it originally, and ask him to put it on the bottom of his pile and give you another from the top in its place.
Explain to the audience why you do this: such a ballot, though it cannot help elect any other candidate, will serve with full effect as one of the 12 that elects C.When C's surplus ballots have all been transferred, read the candidates' names one by one and have the tally clerks announce the number of ballots received by their candidate during the transfer (not the total numbers thus far received).
Record the transfers and later the new totals.Check the transfer by making sure that the numbers in the transfer column after the plus signs add up to the number after the minus sign.
Mention the further check, always used in public elections, of adding the new totals to make sure that their sum is the total number of valid ballots.
When the transfer has been checked, tell the tally clerks to put their ballots last received on top of their first-choice ballots. Tell them on each future transfer they should also keep ballots received separate until the transfer has been checked and then put them with their other ballots.
=============================
The Element of Chance that arises in transfers of surplus votes
When C's surplus has been disposed of, it is necessary to consider the important question raised above — which ballots should be taken from C's pile for transfer. It is known that 10 of them are transferable.
The following points should be shared:
It is obvious that the 22 ballots in C's pile may differ in respect to second choices, so that it might make a difference which 10 of them we take for transfer as C's surplus.
Various methods of selecting votes to be transferred:
#1. Mathematical method. It is possible to eliminate this element of chance by examining the next choices on all 22 of C's ballots and taking for transfer 10 that are perfectly representative, so far as next choices are concerned, of all of them that are transferable. This method is the one used in most public elections under the STV system in other countries.
#2. Semi-random method: In some cities using STV, a simpler method was adopted. The surplus ballots of an elected candidate is taken by chance except that an equal number, as nearly as possible, is taken from each of the voting divisions of the electoral district.
#3. Random method Most private organizations that used the STV system in the early 1900s took surplus ballots entirely by chance. This method is entirely defensible. However, the chance method might possibly give different results if applied twice to the same set of ballots. In a very small election this method might cause a variety of different results. But in a large election, mathematicians find, the amount of possible change is very small.
Mathematical analysis of method #3
Suppose that, in an election in which the ballots number 59,999 and the quota is 6000, Candidate A receives 10,000 first choices, and suppose that B is marked as second choice on 6000 of those ballots, C on 4000 of them. Proportionally, 4000 surplus votes need to be transferred -- 2400 ballots to B and 1600 to C.
How large a variation from this standard is likely to result from taking the 4000 surplus ballots by mere chance as in method #3? The chances are even that neither B nor C would gain or lose more than 11 votes and that the odds are more than 3 to 1 that neither would gain or lose more than 20 votes, just 50 to 1 that neither would gain or lose more than 40 votes, and about 2000 to 1 that neither would gain or lose more than 60 votes. But suppose, as would happen about once in 2000 times, the error did amount to 60 votes. Even in that case it would be only one vote in each thousand, or one-tenth of one per cent.)
Element of chance under STV is less important than it is under FPTP
In considering the practical importance or unimportance of the element of chance that is produced under STV, we may also bear in mind an incomparably greater element of chance that is connected with our usual method of electing representatives. Under that method the voters are arbitrarily divided into districts, each of which elects one member of the representative body.
Under such a system it may make all the difference in the world on which side of your house the division line is drawn. In one district your vote may turn the scale between two candidates and so be of great importance. In another it may be thrown away every time because your candidate is sure to have far more than he needs for election even if he did not have your vote, or it may be thrown away because your favoured candidate always has too few for election even with your vote. In comparison with such an element of chance, that involved in taking surplus ballots by chance under the STV system is of no importance.
Moreover, how the district line is drawn under the old system of election is frequently determined with the deliberate purpose of gerrymandering, that is, the government drawing the lines in such a way as to waste as many as possible of the votes that are in favour of the party not in power.
Any of the three methods of STV vote transfers listed above, on the other hand, do not lend themselves to deliberate manipulation, the count being conducted in the presence of many witnesses.
Though the random method of selecting surplus ballots (#3 above) is good enough, it should not be favoured over #1, the mathematical method, unless those who make the choice, clearly understanding the issue, decide that the extra trouble involved in carrying out the mathematical method is not worthwhile.
Where the number of ballots is very small and rivalry among the candidates is great, the mathematical method should be used without question.
The Rest of the Count
As all the chief difficulties have now been passed, the rest of the count can be carried through very quickly.
If there are any more surpluses to be transferred, transfer them, one by one, just as you did the first one, entering the result on the blackboard after the transfer of each surplus.
Elimination of lowest-ranking candidates
Explain that even now, after the transfer of all surpluses, it might be unfair to declare the five highest elected. (We are assuming, of course, that not all the five highest have received the full quota.) This can usually be made evident by pointing to two or more of the candidates, say A and B, below the five highest, who together have at least a full quota of votes. What if all who cast those votes preferred both A and B to all the other unelected candidates? In that case, evidently, either A or B should get one of the seats. We must therefore eliminate the lower-ranking candidates one at a time and transfer each candidate's ballots to others, according to the next back-u preference marked on them, before moving on to eliminate another candidate.
This feature of the STV system enables the voters of one party to vote for various candidates of the party without any danger of thereby "splitting the party vote" and helping other parties. If the party voters together compose the quota, they will eventually be accumulated together to elect one of their party if they mark their back-up preferences along party lines.
If there are candidates who have no ballots at all, declare them defeated and write the word "defeated" after every zero in the last result column.
Point out that even these weakest candidates were left in the running as long as possible: they had the opportunity to receive ballots by transfer from those who received quota and surpluses.
Next, declare defeated the candidate now having fewest votes to his credit.
Explain that several of the candidates must eventually be defeated, and it seems fairest to defeat the lowest one first.
Take the ballots of the eliminated candidate and transfer them, one by one, each to the voter's next choice among the unelected candidates.
Read out the back-up preferences on each ballot so that the audience can follow.
If any of the ballots cannot be transferred to any remaining candidate, because not marked for any one of them, lay them aside as "exhausted." This will be caused by a ballot bearing no back-up preference, or only preferences marked for those already eliminated or only marked for those already elected.
Comment on the advantage to the voter of marking all the choices he has. You will probably be able to say also that many of those who have cast these exhausted ballots have seen one or more of their choices elected by others.
When all of an eliminated candidate's ballots have been transferred, enter the results in the next column on the blackboard.
Compare the number of votes other candidates received to the number lost by the eliminated candidate. Any difference should be equal to the number of added exhausted votes.
Check total number of votes held by each candidate plus exhausted. This should equal the number of votes cast.
Eliminate the candidate who is now the lowest ranking. Transfer the ballots in the same way.
Proceed thus until five candidates have received the quota or until the defeat of some candidate leaves only five undefeated.
In case five receive the quota, cease transferring ballots when the last quota is completed and treat all the undistributed ballots as "exhausted."
In case through eliminations the number of undefeated candidates is reduced to five before five candidates receive the quota, do not distribute the ballots of the last candidate eliminated, but declare the remaining candidates elected at once.
Explain that transferring the ballots of the last defeated candidate, though it might very likely complete the quotas of all the remaining candidates, could not alter the result, as the five have been elected – the five have already been shown to have a better right to the five seats than any others.
If you encounter a tie during the count, explain the rule for ties and mention the fact that ties very seldom occur in large elections. In the case of a tie it is fair to consider that candidate stronger who got his support earlier in the count.
Someone may ask why the tie should be decided by reference to accumulated votes (first and later back-up preferences) rather than just to first choices. The answer is that the decision should be made according to the last possible tabulation so that as large a number of ballots as possible may have a voice in making it.
If an error in the counting should come to light during the demonstration you can usually point out that its discovery was due to some feature of the STV system, as, for example, the necessity of handling some of the ballots more than once. In a real election an error discovered would, of course, have to be traced out and corrected. In a demonstration election it is often unwise to take the time to do this.
========================================================
Schedule 45 minutes for this demonstration election
The time required for an expert to carry through a demonstration election with a hundred ballots, with necessary explanations, is about half an hour. The beginner will need a somewhat longer time. The demonstrator should plan also, if possible, to leave ten or fifteen minutes after the election for testing the results, pointing out the effects of such an election system on a private organization or a community that uses it, and answering questions.
=========================================
Testing the Results
(A) If the candidates were arranged in pretend or actual parties, point out that the successful candidates represent the different parties and different points of view within the parties. Usually it can be shown that each party received its fair share of the seats, as nearly as that share can be estimated by the first choices. It must not be inferred, however, that the final result should correspond exactly with the party grouping indicated by first choices: some voters cross party lines when they mark their back-up preferences, and the full and true story of the voters' real wishes is shown only by the final result.
(B) Point out how many of the voters have had a share in the election of the representatives. The number is found by subtracting the number of wasted votes from the total number of ballots cast. Wasted votes are spoiled votes, exhausted ballots, votes left undistributed with the candidate or candidates last defeated, and votes left with surviving unelected candidates.
(C) Ask all who saw their first choice elected to raise their hands. The number can be determined exactly by reference to the first count results posted on the blackboard.
(D) Best choice used. Take up the ballots that have elected one of the last candidates to receive the quota and read out the choices of two or three of them to show that each one has been counted to help elect the candidate whom, under the actual circumstances existing, the person who marked it most wanted to help. That is, leaving out those who have been elected or eliminated, demonstrating that the vote had gone to the first back-up preference who was elected.
Un-used votes
Examine the un-used votes. These are the few ballots that have not helped to elect anyone. Show the audience that even they were cast by voters who are in most cases satisfied. Many of the back-up preferences marked on them are for candidates who were elected.
(E) Read aloud the first choice, second choice, etc., on each one of the un-used votes until you come to a candidate who has been elected. As you do this, arrange the ballots in piles.
When done, announce:
-how many of the "unrepresented" voters saw their second choice elected,
-how many their third choice, etc.
(There will be none that have their first choices elected - they would not be unrepresented.)
(F) Un-used votes (compared to Block Voting)
Read off the first five choices — the group that the voter would have picked if he had been able to cast five votes under Block Voting— and note how many of them have been elected.
Announce:
how many of the "unrepresented" voters saw none of their five elected,
how many saw one of their five elected,
how many saw two of their five elected,
how many saw three of their five elected,
how many saw four of their five elected.
(There will be none that have all five of their top five choices elected - any with five elected would not be among the "wasted" votes.)
(G) Un-used votes
Read off all the choices marked on the un-used ballots.
Sort the exhausted votes into two piles - one for those who saw none of their candidates elected and one for those that did. So we see that quite a few of the exhausted votes actually did see one or more of their choices elected.
Note how many of them marked preferences for all of the five winners,
how many voted for four of the winners,
how many voted for three,
how many voted for two,
how many voted for one of the winners.
If there are too many wasted votes to handle them all expeditiously, examine a few of them only.
Try to leave plenty of time for questions.
================
Some Effects of the System
If you have time, mention some of the beneficial results of the system, especially the following:
(1) The discouragement of "machine rule." Both the preferential and the proportional features of STV contribute to this most important result.
The preferential feature makes it possible to vote against some candidates without the least danger of wasting your vote and perhaps helping to elect those you like least of all by so doing. It enables you to nominate and vote for your real favorites, knowing that if they are unsuccessful you will still have a chance to help the least objectionable among the remaining candidates.
Important as this freedom of expression is, however, it might not always be sufficient to solve the problem of party-machine rule by itself. If a single block of votes could elect all the seats in a district, such as occurs under Block Voting, a party-machine might control a block of votes large enough to take all the seats without the support of a majority of the voters and thus wield a power out of proportion to its numbers. The proportional feature of the STV system enables every substantial group to obtain its own separate representation without the aid of any party-machine, and restricts any such party-machine to its fair proportion of the members on the basis of the votes cast.
(2) The election of representatives who can safely be trusted with the power needed for the greatest efficiency. The STV system makes it certain that the body to which such power is given will be truly representative.
Quicker substitute for demo election
If you have too little time to carry through an actual election, a possible substitute is to pass around copies of a report of a real election and explain the election given in it.
(These suggestions were described as based on the experience of Secretaries of the Proportional Representation League in some hundreds of elections. They were first published in 1922, in Proportional Representation Review, April 1922 Supplement.
Other info is from J. H. Humphreys's book Proportional Representation, and other was previously published by Tom Monto (myself) on the Montopedia blogsite.)
==========================
From Effective Voting, a 1898 publication of the Proportional Representational Committee of Ontario:
"Small elections at meetings are a valuable means of diffusing a knowledge of PR. So to ensure random-ness in small meeting elections, it is well for each voter in a small meeting to vote in duplicate or triplicate - that is, each voter casts two or even three ballots - which gives the same advantage in counting as if there were twice or three times the number of votes."
Other excerpts from the pamphlet Effective Voting are available in the blog : "Effective Voting - Pro-Rep, 1898 Ontario"
=========================
See blog "list of Montopedia blogs concerning electoral reform" to find other Tom Monto blogs on this important subject.
=========================================
Here's some notes on a demonstration STV election
I worked through a pretend drink election, showing math formulas, vote transfer table, etc.
nice thing about electing drinks is immediately voters either are happy or not with choice.
here's link to Catherine Helen Spence's demonstration STV article
Here is the original article
I say we do three separate elections as the time needed for one is about 40-45 minutes
The first ones may take longer than that until we get it down to a science...
Spence: "The time required for an expert to carry through a demonstration election with a hundred ballots, with necessary explanations, is about half an hour. plus 10 or 15 minutes afterwards.
(Spence:) The demonstrator should plan to use 10 or 15 minutes after the election to test the results, point out the effects of using such an election system on the community using it, and to answer question.
Thus, if voters cast votes in all three at once, we could have opportunity to do three without having to stop to collect votes again.
(each election could use different-colour ballots to aid in sorting)
But maybe one or two separate elections is enough.
three elections?
vote on drink
vote on snack, and
vote on desert or something else.
For each vote we put a chart on the wall with nine names running down left side
for drink election:
perhaps these "candidates":
milk
orange juice
grapefruit juice
coke
pepsi
tea
buttermilk
water
root beer
the ballot is distributed to each "voter"
each ballot has same choices.
(perhaps ballot has three separate sections - one to determine drink, one to determine snack, one to determine desert or each voter casts separate ballots
Voter is invited to rank the candidates but to rank only as many as he or she wants to.
voter is instructed not to fold the ballot (to save time, apparently).
===============================================
the first "election" determines drink to be served
say 125 votes are cast. (a bit fewer is fine too, likely)
(five tables, 25 votes in each table perhaps
(if we don't get 25 people out to the event, perhaps the number of people at each table can be multiplied by say three, then other pre-marked votes added to bring table count up to 25.
or just the number at the table is multiplied by some factor to get a total number close to 25.
the number of votes at each table does not have to be the same, just as single-member districts vary in size.)
first-preferences are counted at each table separately.
first, pretend FPTP election
each table is separate single-winner "district"
the choice of drink with most votes at the table is served to each person at a table.
much unhappiness with what is served
then STV election
electing five drinks that suit the tastes of voters, fairness achieved by large number of effective votes (votes that actually elect a drink) and each drink being elected with the same number of votes (in most cases)
The number used to elect the drink (in most cases) is the quota.
Quota is valid votes cast divided by (5 plus 1), plus 1.
(hopefully there are not many spoiled ballots! there are zero in the pretend election shown here)
99 valid votes quota is 17
(if 125 valid votes, quota would be 21)
Any drink with quota or more is certain to be elected.
Any drink elected is made available to all who want it, or just in the exact amount to replicate effective votes?
if we make drinks available to all, then those voters who like two of the winners can enjoy both (which is what would happen if two MLAs whom a voter likes are elected in the district)
Either way, those whose preferences were not elected would not have drink they voted for. But like in real STV elections, likely with the variety elected, they would find something they are somewhat happy with.
First Count
the table tallies for each of the nine "candidates" are added together.
The vote tallies are posted on wall chart.
Later Counts
Subsequent eliminations or transfers of surplus votes of elected candidates are done until all seats filled, or until the number of remaining candidate is thinned to the number of remaining open seats.
Chart showing pretend numbers and pretend transfers
99 votes Quota 17
First 2nd 3rd 4th 5th 6th
Count Count Count Count Count Count
Grapefruit Water Buttermilk Beer Coke
surplus elimination elimin elimin surplus.
Milk 10 10 2 12 12 12 3 15 (Elected) Orange juice 11 +1 12 12 12 1 13 0 13 Grapefruit juice 18 -1 17 ELECTED----------------------------------------------------------- Coke 16 16 16 16 4 20 -3 17 ELECTED Pepsi 14 14 14 14 4 18 ELECTED Tea 12 12 2 14 3 17 ELECTED --------------------------- Buttermilk 5 5 5 -5 0 ELIMINATED -------------------------- Water 4 4 -4 0 ELIMINATED--------------------------------------------- Root Beer 9 9 9 9 -9 0 ELIMINATED Exhausted 0 0 0 2 2 2 1 3 Total 99 99 99 99 99 99
At end of 6th Count, one seat remain, two candidates remain.
Orange juice is less popular; Milk is more popular.
Milk is given the last seat.
Winners:
Grapefruit Juice,Tea, Coke, Pepsi and Milk
Final vote tallies of winners:
Grapefruit Juice 17
Tea 17
Coke 17
Pepsi 18
Milk 15
Total effective votes: 84
Action taken after result announced:
all voters have their choice of one of those drinks (or drinks issued in proportion as per vote - but some votes may be collected in advance (equivalent of mail-in votes)
just as a voter has right to consult with just one of the MLAs who would represent their district.
Note that transfers were performed according to voters' instruction (Best choice used)
When Beer was eliminated, votes might have been transferred to OJ, but instead they went to Milk or Coke or Pepsi (whichever was next in the voter's heart).
If the vote is used to elect someone, the vote is used to elect someone preferred by the voter over others.
==========================================
Calculation of surplus transfers
(only next usable marked preference is referred to)
Grapefruit juice surplus transfer
exhausted = 0 18 - 0 = 18 transferable votes
surplus 1
candidate's transferable votes 18
with just one vote to be transferred, only the most popular preference gets the one vote.
Tea
no transfer required as Tea did not have more than quota - no surplus.
Coke surplus transfer
exhausted 6 votes transferable votes= 15
surplus 4 cand's transferable votes 15 4/15 = .27 marked preferences (next usable preference) milk 11 X .27 = 2.93 Round Off to 3 orange juice 4 X .27 = 1.08 Round Off to 1 TOTAL 4 transfer three to milk and transfer one to orange juice (as done in the above table in the 6th Count column) the 6 exhausted votes stay with Coke also 8 milk 3 OJ TOTAL 17 Pepsi was elected but no transfer necessary as any transfer cannot change order of popularity of Milk and OJ ========================== in this example, 9 candidates to start with by 6th Count 3 were eliminated 4 were elected two are left, Milk elected (even though it does not have quota) ==================================================== Check fairness of election (see below for details) (A) party proportionality (are we grouping drinks in "parties"?) But we can make general statement of how many are happy with the result. As well, voter themselves can judge the result by what drinks won. (B) First preference elected: ask all who had their first choice elected to raise their hands. number of voters whose first preference was elected: Grapefruit Juice 18 Tea 12 Coke 16 Pepsi 14 Milk 10 Total FP voters: 70
(C) Effective votes (those not wasted) Wasted votes are: spoiled votes, zero in my pretend election exhausted ballots 3 votes left undistributed with the candidate last defeated 13 (OJ) votes left with candidates neither elected nor eliminated zero TOTAL 16 Effective votes 99 total - 16 wasted = 83 (83 is the same as total of vote tallies of winners (as above). 84 percent were effective votes (much larger percentage than almost all FPTP elections). And most - or all - of the un-used votes/"unrepresented" voters actually had their choices elected, just with the vote not being used to do so, as we prove next. ============= D to G: Un-used votes/ "unrepresented" voters - marked preference(s) were elected (although vote is not used to do so) D. at least one marked preference are elected E. many of the highly-preferred marked preference(s) are elected F. how many of the top five choices on the un-used vote are satisfied? G. if we look at all the marked preferences. most (all?) of the un-used ballots have many choices elected Examine the few un-used votes, votes that were not used to elect anyone. Un-used votes are: - exhausted ballots (these votes recorded as exhausted but back-up preferences marked on them were elected in many cases) 2 - votes left undistributed with the candidate last defeated (these votes not used to elect anyone, but candidates marked as back-up preferences on the ballots were elected in many cases, without the help of the ballot) 13 (OJ) TOTAL 15 un-used votes (D) preferred choices elected Despite being "un-used votes", all (or most) of un-used votes bear back-up preferences for elected candidates. =============== Spence: Read off all the choices marked on the un-used ballots. Sort the un-used votes into two piles: - one for those who saw none of the marked candidates elected, and - one for those who saw one or more of the marked candidates elected.So we see that quite a few of the exhausted votes actually did see one or more of their choices elected. ========== Un-used votes are exhausted votes and votes left undistributed with the last-defeated candidate last defeated. Exhausted votes Exhausted votes will be of three types - those that have first and back-up preferences for elected candidates; those that have first and back-up preferences for both eliminated and elected candidates. those that have first and back-up preferences only for those who were eliminated; Votes of the first two types are voters who are in most cases satisfied, because one or more of their choices were elected even though their vote was not used to elect anyone. We can sort the exhausted votes into these three piles (Spence did not suggest this but I think it is good idea) We can then see that many exhausted votes actually saw their choice elected. Two who wanted Buttermilk were happy to see Grapefruit juice elected, even though the vote itself was not used to elect Grapefruit juice. Grapefruit juice was already elected and there were no other back-up preferences so the votes were moved to exhausted. Votes left undistributed with the last-defeated candidate Orange Juice was the last candidate defeated. The votes were not transferred. 11 of the 13 OJ votes at the end were never transferred at any time but likely bore marked preferences for Grapefruit Juice, Tea, Coke, Pepsi and/or Milk. (But in my pretend election we can only guess - in our demo STV election we can look right at the ballots and see the back-up preferences, to test the result.) (E) How many un-used votes saw the voter's most preferred candidate(s) elected? Spence: "Read aloud the first choice, second choice, etc., on each one of the un-used votes until you come to a candidate who has been elected. As you do this, arrange the ballots in piles: those whose first choice was elected, those whose second choice was elected, those whose third choice was elected, etc. When done, announce how many of the "unrepresented" voters saw their second choice elected, how many saw their third choice elected, etc. [This sounds a bit time-consuming but only a small fraction of the votes cast are not used to elect anyone.]
(F) Un-used votes/"unrepresented" voters -- how many of the first five choices are satisfied? (comparing STV to Block Voting) Un-used votes are Spoiled votes, zero in my pretend election Exhausted ballots 2 Votes left undistributed with the candidate last defeated * 13 (OJ) and Votes left with surviving un-elected candidate zero TOTAL 15 On each of these wasted votes, read off the first five choices — the group that the voter would have picked if he had been able to cast five votes under Block Voting— and note how many of them have been elected. Put them into piles sorted by how many of the top five preferences were elected. Announce how many of the "unrepresented" voters saw none of their five marked preferences elected, how many saw one of their five marked preferences elected, how many saw two of their five marked preferences elected, how many saw three of their five marked preferences elected, how many saw four of their five marked preferences elected. how many saw five of their five marked preferences elected. (G) Un-used votes -- Was the elected preference a second, third, fourth etc. choice? Examine the pile of votes for those that saw one or more of the marked candidates elected,Note: how many of them voted for all of the five winners, how many voted for four of the winners, how many voted for three of the winners, how many voted for two of the winners,how many voted for one of the winners. If there are too many wasted votes to handle them all expeditiously, examine a few of them only. Try to leave plenty of time for questions. ====================================================== Proving fairness
(From American PR League's 1922 article "Demonstration STV election") [the original wording (approx.)] Testing the Results (A) If the candidates were arranged in pretend or actual parties, point out that the successful candidates represent the different parties and different points of view within the parties. Usually it can be shown that each party received its fair share of the seats, as nearly as that share can be estimated by the first choices. It must not be inferred, however, that the final result should correspond exactly with the party grouping indicated by first choices: some voters cross party lines when they mark their back-up preferences, and the full and true story of the voters' real wishes is shown only by the final result. (B) Point out how many of the voters have had a share in the election of the representatives. The number is found by subtracting the number of wasted votes from the total number of ballots cast. Wasted votes are spoiled votes, exhausted ballots, votes left undistributed with the candidate or candidates last defeated, and votes left with surviving un-elected candidates. (C) Ask all who saw their first choice elected to raise their hands. The number can be determined exactly by reference to the first count results posted on the blackboard. (D) Best choice used. Take up the ballots that have elected one of the last candidates to receive the quota and read out the choices of two or three of them to show that each one has been counted to help elect the candidate whom, under the actual circumstances existing, the person who marked it most wanted to help. That is, leaving out those who have been elected or eliminated, demonstrating that the vote had gone to the first back-up preference who was elected.[This is to show that the election official does not produce the transfer but instead the transfer is done according to the voter's instructions] (E) [5a in original article]Examine the un-used votes.These are the few ballots that have not helped to elect anyone. Show the audience that even they were cast by voters who are in most cases satisfied. Many of the back-up preferences marked on them are for candidates who were elected.
Read aloud the first choice, second choice, etc., on each one of the un-used votes until you come to a candidate who has been elected. As you do this, arrange the ballots in piles and at the end announce how many of the "unrepresented" voters saw their second choice elected, how many their third choice, etc.(There will be none that have their first choices elected - they would not be unrepresented.)
(F) [5b in original article] [STV wasted votes compared to Block Voting's effective votes]
Examine the "wasted" ballots as listed in B above. Read off the first five choices — the group that the voter would have picked if he had been able to cast five votes under Block Voting— and note how many of them have been elected.Announce how many of the "unrepresented" voters saw none of their five elected, how many saw one of their five elected, how many saw two of their five elected, how many saw three of their five elected, how many saw four of their five elected.(There will be none that have all five of their top five choices elected - any with five elected would not be among the "wasted" votes.) (G) [5c in original article] [Exhausted votes] [Votes that could be called used up but satisfied] Read off all the choices marked on the un-used ballots Sort the exhausted votes into two piles - one for those who saw none of their candidates elected and one for those that did. So we see that quite a few of the exhausted votes actually did see one or more of their choices elected. Note how many of them marked preferences for all of the five winners, how many voted for four of the winners, how many for three, how many for two,how many voted for one of the winners. If there are too many wasted votes to handle them all expeditiously, examine a few of them only.
Try to leave plenty of time for questions.
===============================================
SUGGESTIONS
for voting
make sure each "district" has about same number of votes by say having five votes available for each and only after all 25 of those have been issued, five to each district, start to give out next round of votes, next group being maybe three for each district, and so on. if that makes sense
for vote counting
first announce the results as per the first preference, with candidates listed in order of popularity (number of first votes received)
already any candidate(s) with quota are known to be elected.
the one at the bottom is certain to be eliminated
audience should be told that:
those at the bottom are likely to be eliminated
those near the top are likely to be elected.
and that the vote count and vote transfers will determine the exact results.
And I think do the vote transfers in front of the audience.
but if not, announce the results round by round as the vote count progresses.
==========
Comments