Up: Issue 2 Previous: Paper 4

Voting matters - Issue 2, September 1994

Is a feedback method of calculating the quota really necessary?

R J C Fennell

Robin Fennell is a retired radar technician with the RAF and more recently, a Customs Officer. He has been a member of ERS since an abortive attempt to introduce STV into his union elections. He is currently active in transport and defence campaigning as well as electoral reform.

The March issue of Voting matters reprinted papers by B L Meek [1],[2], D R Woodall [3] and C H E Warren[4]. In this paper I will propose that their feedback method of calculating the quota is not necessary. To do this I will consider some of the basic principles of the Single Transferable Vote (STV) system.

One problem identified[5] is that if a candidate is elected any further preferences for that candidate are passed over. The question to be considered is 'are elected candidates continuing in the election or should they be considered as no longer available to receive votes'?

In other words is the purpose of a vote in the Single Transferable Vote system to try to elect candidates in the order the voter wishes or to place candidates in popularity order and have this order respected whatever the outcome of the rounds of the count? I suggest that it is the former. Once a candidate has been elected he has achieved the aim of participation in the election and, henceforth should take no further part in the election. Under these circumstances the manual counting method is satisfactory.

We will take a voting paper that shows preferences A,B,C,D and assume that B was elected on the first round. The transfer of B's surplus elects A on the second round. The question now arises on our paper, should the transfer of A's surplus go to B or C. Let us assume that our voter had future vision when deciding the preferences and knew that B would be elected in the first round would our voter put B as the second preference. I suggest that anyone so gifted would select the preferences A,C,D thus maximising the transfers to the candidates they wished to see elected. Of course this foresight is not available to voters so to cover all possibilities the voter will elect to keep to the original selection knowing that the counting system will not waste any part of a vote by transferring it to a candidate already elected.

In practice few voters would take the risk of excluding a candidate on the grounds that they are certain to be elected. If too many did then B would not be elected. Voters can be expected to behave in a rational fashion and vote for the candidates of their choice in the order they wish. When a candidate has been elected they have achieved the aim of both the candidate and the voter. The voter will now wish any surplus votes to be concentrated on the unelected choices.

Another problem identified by Meek[6] is how to treat unmarked candidates. He suggests that they should be considered either as being of equal merit, or that the voter wishes to leave the ordering of these candidates to others. Meek ignores the third possibility that the voter does not wish these candidates to have any part of the vote. The omission of the third alternative in Meek's paper is possibly due to the voting instructions that take a form similar to 'place the candidates in order until you can no longer differentiate between them'. If the instructions were changed to a form similar to 'place the candidates in order until you no longer wish the remaining candidates to have your vote' it would be clear how the voter required unmarked candidates to be treated. Under these circumstances the manual counting method is satisfactory.

If STV is to be used in local or parliamentary elections many voters will only want to vote for their particular party. They will not wish any proportion of their vote to go to candidates of a party with an opposing view to theirs. If votes are apportioned to all non-selected candidates, voters will have no way of ensuring that they do not vote for candidates of a party whose policies they cannot agree with.

The other problem foreseen by Meek[7] that I will consider is the possibility of voters indicating the same preference for two or more candidates. He suggests that this should be allowed and the counting system modified to accommodate it. The Electoral Reform Society (ERS) supports the Single Transferable Vote system, not the Transferable Multi-vote of Unity Value System. This second system may exist but it is not that supported by the Society and therefore should not be considered. The Single Transferable Vote system requires voters to cast a single vote, all or part of which may be transferred. That a multiple vote may have unitary value is irrelevant, it is a single vote which must be utilised.

D R Woodall[8] raises a different problem, that of the tactical voter. He postulates a situation where there are several Sensible Party candidates, say A,B,C and one Silly Party candidate, W. The tactical voter decides that W will be excluded and in order to maximise the transfer of votes after the first round he will vote W,A,B,C rather than A,B,C which is the real preference. The problem for the tactical voter comes when several voters take the same line. Assume in this election that the quota is 200. If 201 voters vote tactically and put W first then W will be elected reducing the vacancies available for Sensible Party candidates. In these circumstances the tactical voters will be as silly as W's party. The only way to avoid this is to place preferences in the order the voter wishes the candidates to be elected and not to attempt to vote tactically.

One of the main advantages of STV is that attempts to vote tactically are likely to end in a result that will not suit the tactical voter. The situation above could happen irrespective of the number of candidates or the size of the quota. The only safe way for voters to use their vote successfully is to vote according to preference.

The three works printed in the March issue of Voting matters may be mathematically rigorous but they are required? My contention is that if the basic principles of the Single Transferable Vote system are carefully considered then the feedback method of counting is unnecessary. The manual method used to date is satisfactory to ensure the correct result.

There is one further matter to be considered. If the feedback method is to be used, the constant recalculations necessary will require computers to be used. It is recognised in the papers supporting the method that it is too laborious to use hand counting. While the ERS has voted to use both computer and manual counting for its internal elections, I doubt if a system which cannot reasonably be counted by hand will be accepted by the general public. Computers are quick but they rely on the integrity of their programming. Computer technology is not yet at a state where incorrect programming, whether by accident or intent, will always be exposed. While it is not possible to say that the currently accepted Newland/Britton hand counting rules will always produce the correct result, they will produce a satisfactory result. I can see no reason to change the current system of counting.


  1. B L Meek, A New Approach to the Single Transferable Vote, Paper I, Voting matters, March 1994.
  2. B L Meek, A New Approach to the Single Transferable Vote, Paper II, Voting matters, March 1994.
  3. D R Woodall, Computer Counting in STV Elections, Voting matters, March 1994.
  4. C H E Warren, Counting in STV Elections, Voting matters, March 1994.
  5. As [1], section 3, item (iv).
  6. As [2], section 3, penultimate paragraph.
  7. As [2], section 6.8. As [3], second paragraph.
  8. As [3], second paragraph.

Up: Issue 2 Previous: Paper 4