JOINT COMMITTEE ON THE CONSTITUTION debate -
Wednesday, 31 Mar 2010

The committee continues its hearings on the electoral system. An earlier meeting examined the Gregory method for transferring an elected candidate's surplus votes. Under this system all ballot papers are sorted and transferred according to the next available preference. Under the current rules on vote counting for Dáil elections, a random sample of ballot papers is selected for the transfer of surplus votes. In the course of our hearings we heard evidence which suggested that under certain circumstances the current system can result in the election of the "wrong" candidate as it involves an element of chance that may affect the outcome of the count. Professor Farrell recommended to the committee the adoption of the Gregory method for transferring surplus votes in Dáil elections as a solution to the arbitrariness of random selection.

Today the committee will consider the Gregory method as used in Northern Ireland elections and its application in a manual vote counting system. In Northern Ireland, the single transferable vote electoral system is used for all elections except parliamentary general elections, and all counts are conducted manually.

I welcome Mr. Douglas Bain, chief electoral officer for Northern Ireland and Ms JocelynMcCarley, assistant chief electoral officer, and thank them for attending. We are very grateful to them for taking the time to meet with us to discuss how this system works in practice in the sorting and counting of votes at election counts. Their presentation has been circulated to members and we are very grateful for that.

Before we commence, I inform the witnesses that members of the committee have absolute privilege but that same privilege does not apply to witnesses. I ask them to commence their presentation.

I thank the Chairman. We welcome this opportunity to address the committee on the Northern Ireland system of proportional representation and, in particular, on the provisions for the transfer of surpluses and the exclusion of candidates. At the outset, I wish to make clear that neither of us have any remit for any particular voting system nor are either of us experts in any other system on which the committee has already taken evidence. Rather we hope to give members a practitioner's view of how the Northern Ireland system works at a manual count.

I apologise for the absence of a paper. I am sure members will appreciate that when an election may be called within the coming seven days there are one or two other rather more pressing matters. We will be delighted to take any questions members may have.

We propose that I shall take the committee very briefly through the system to set the scene and all the difficult bits on surpluses will be dealt with by my colleague who is much more experienced in the practicalities.

As the Chairman said, the system is used in Northern Ireland for all except parliamentary elections. In the presentation, I provide a reference to the current version of the legislation as it applies to Assembly elections. Exactly the same provisions appear in other statutes for European Parliament and local government elections.

As in most similar systems, voters rank the several candidates. In the case of Assembly elections six candidates are selected for each seat and election is by reaching a quota or being deemed to have reached a quota. Single votes are transferred in accordance with the voter's expressed preferences. The system starts in a very straightforward way. Papers are sorted according to the first preference and a quota for the election is established. Candidates who reach or gain more than that are deemed elected and then one gets on to the issue of transferring votes to bring other candidates up to the surplus. My colleague will deal with that matter.

Generally speaking, candidates with the fewest votes are excluded, in turn, and their votes are distributed to the remaining candidates in accordance with the next available preference given on the ballot papers. The transfer of surpluses and exclusions continues until the number of seats remaining equals the number of candidates remaining.

The next slide in the document shows the method of calculating the quota and there is nothing particularly magical about it. That sets the background but the committee is probably interested in hearing how we deal with the transfer, and my colleague will deal with that.

I will speak about transfer of surplus. Votes are transferred at a transfer value in Northern Ireland. There is a transfer of surplus and then a transfer of consequential surpluses. A consequential surplus, which comes about further on in the count, is a transfer of the last packet of papers that the candidate received. It does not take in the whole amount.

The votes are transferred with a transfer value and this reduces the value of each vote so the value does not exceed the surplus available for transfer. We calculate a transfer value by dividing the surplus of the candidate for whom the votes are being transferred by the number of ballot papers on which those votes are given. Further down the count that can bring a consequential surplus, which is the last packet received by the candidate.

I will go through some issues in the Northern Ireland system. As we are doing manual counts, we find there are some advantages. It is quicker than some other systems because we can defer a transfer of surplus if it will not make a difference to the rank order of the remaining candidates. That applies to the lowest ranked candidates.

We also have multiple exclusions. As an example, there was a European Parliament election in June 2009. There are three seats to be filled in such an election and with the first preference votes we had one candidate who reached the quota, the Sinn Féin candidate, Bairbre de Brún. The quota at a European Parliament election is approximately 150,000 votes and her vote was approximately 6,000 more than that, which meant there was a surplus of that amount. The lowest candidates had approximately 300 to 600 votes so even if we had transferred the surplus votes to our lowest candidates, there would have been no difference and it would not have helped them overtake the next candidates.

We deferred transferring her surplus, which is allowed under Northern Ireland rules, and excluded two of the lowest candidates. That helps us in a manual count and makes the process quicker and more manageable. When I indicate how long it takes, the committee might not agree. We only transfer the packet of papers that led to the quota being obtained rather than all the papers held by a candidate.

Northern Ireland rules allow for sub-stages when we have exclusions. We must transfer in an exclusion the first-preference votes first, and we can have somebody deemed elected after that sub-stage of an exclusion. The stage of the count relates to the exclusion of all papers but we consider the candidates after transferring first-preference votes to see if anybody has reached the quota. As these are in our rules, they can lead to a quicker manual count which could be more accurate because it is hoped there would be fewer stages.

The accuracy point is important with a manual count. As can be seen from the length of a count, people are engaged for 20 or so hours and, being human, mistakes will be made. Anything that can be done to keep the process simple and as short as possible is likely to increase accuracy. The longer people are there, the less accurate the counting will become. There are certainly some advantages in keeping the length of the count down but there are many other factors which affect this, possibly more significantly than the differences my colleagues spoke about.

First, there is the number of staff employed on the count. It is not a case of double the number of staff leading to a halving of the count time. It is inherent in this system of the single transferable vote, STV, that the votes will be sorted by candidate, and large numbers of staff can be employed in that work. Inevitably, there are periods when only a small number of staff are involved in doing the number crunching and others are apparently doing nothing.

What is perhaps more important than the number of staff is the quality. Quality, like most things, depends on paying sufficient rates for the job and testing the people to ensure they are of the right grade. We test all the people who apply to work for us on counts both on the speed and accuracy of counting. In the process leading to the upcoming election, there was a failure rate of approximately 25% of people who applied as they were not suitable for the work. The number of candidates is also a major factor, as is the number of voters.

Perhaps the most important factor of all is how the votes fall between particular candidates. There is a detailed slide in the document to illustrate to the committee that there is absolutely no way of predicting how long a count will take. It relates to the Assembly elections in 2007, where one can see that in the 18 constituencies there was a range of 12 to 18 candidates, with six to be elected in each constituency. The number of stages varied between six and 13 and the total counting time, including the verification of votes, ranged from 13 to 22 hours. We have streamlined the process since and the quality of staff has been improved; even with that I doubt we will shave more than an hour from these times. If an STV count of our kind is being done manually, it is lengthy, especially if there is a large number of candidates.

I suspect we have said more than enough and the value of this session will be in the questions the members will ask. We are very happy to take them.

It is very interesting to hear about the operations in Northern Ireland. We understand the normal proportional representation system because some of us have gone through the process many times. There is no problem with eliminations and the essential difference is in the transfer of surpluses. I have not fully got my head around the question of the transfer value and how it works. Are the witnesses aware of our system and the differences between it and the Northern Ireland system with regard to the transfer of surpluses? That is the crucial issue in the consideration of whether we should change our system to what is happening in Northern Ireland. I do not have a clear picture of the essential difference in the transfer of surpluses.

I will try to answer the query. For example, if there is a quota of 1,000 votes and a candidate polled 1,200, there would be 200 votes over the quota. To transfer the surplus we look at the 1,200 votes giving the candidate the surplus and look to the next available preference. We would check that they would all be transferable but only 1,000 may be transferred because some people will not vote for more than one candidate.

What happens if one person particularly gains from the transfer and it puts him or her over the quota? In our system the package is redistributed. Can the witnesses give an example of a person on 990 votes, ten short of a quota, whose transfers bring him or her up to 1,050? What happens to the rest of the package?

Yes, if it is a consequential surplus. It is the packet which gives him or her the quota. The surplus is divided by the number of votes transferable, which will have gone down considerably because there will now be two candidates, and this gives the transfer value. The number of papers to be transferred multiplied by the transfer value should equal the number of papers transferred as a surplus.

It varies from election to election, for reasons about which I am unclear. At elections to the European Parliament the votes are face down at the verification so that, in theory, no one can see them. At Assembly elections they are face up. We have counting agents, the equivalent of tallymen, but the impression we get from speaking to our candidates is that they are nothing like as expert as tallymen at predicting the result.

I may have confused the situation. It is at the verification stage that papers are face down. As they are sorted by candidates and counted, they are face up. In practice, because of the way our count centres are laid out, candidates are sorted into racks as in postal sorting offices in the old days. It is very difficult for counting agents to ensure papers are put into the correct box.

I will return to the European elections. The main difference between our systems is the randomness of the final parcel. In most cases, depending on the surplus and size of the parcel, the count centre may go through the whole parcel. In European elections it could be a significant parcel of votes which puts a candidate over the quota. If a candidate received a parcel of 50,000 votes, putting him or her over the quota by 5,000, which would make a difference to the last two remaining candidates, do you go through all 50,000 ballot papers?