Evidence submission to

the

Independent Review into the 2007 Scottish

Elections

from the

Electoral Reform Society

1. Introduction

The Electoral Reform Society welcomes this independent inquiry into the Scottish elections of 3 May 2007. We hope that the Inquiry will be able to throw light on:

ƒ What went wrong in the conduct of these elections; ƒ What needs to be done to reduce the risk of the same problems arising in future elections; ƒ What other changes should be made to our electoral arrangements.

Although in many ways the elections of 3 May were a step forwards for Scottish democracy, serious problems were nevertheless encountered. If with hindsight it is concluded that there were deficiencies in the planning and preparation for the elections, we hope the focus will be on the changes that need to be made before the next round of elections and not on the apportionment of blame. In our work in over the past year, we have not encountered a single person with responsibility for the elections not totally committed to making the elections a success and not entirely professional in their approach.

The Society has already contributed materials to this Inquiry, including research notes on ballot rejection in: • Rejected ballots in UK General Elections • Rejected ballots in Welsh and Scottish elections • Rejected ballots in London elections.

We draw on these in this submission, and attach copies of the research notes for ease of reference. We would be delighted to provide further copies of any other Society publications on request: many are available in both electronic format and as hard copies.

2. The Role of the Electoral Reform Society

The Electoral Reform Society promotes systems that lead to representative institutions (both in terms of party and the diversity of society) and which offer the voter the widest choice and most influence over the result. The Society holds the view that STV as the system that best achieves these ends. The Society therefore took a keen interest in Scotland’s moves towards a better voting system for its local elections and when opportunities arose advanced the case for STV.

The Society took the initiative in establishing Fairshare, a campaign group in Scotland with representation from all major parties and others with special interests in electoral reform. It has provided Fairshare with funding and assisted it in obtaining funding from other sources and it continues to play a major role in Fairshare, although it should not be assumed that the views of Fairshare and of the Society coincide on all issues.

Following the decision to use STV in the local elections of May 2007, the Society opened an office in Edinburgh with six staff to run educational campaigns to prepare for the new system, aimed at political parties, the media and voters in general. Our activities are described in section 8 below.

The Society has prepared reports and press briefings about both the and local government elections. In particular:

ƒ Our report on the Scottish Parliament elections is at present being printed and a draft copy is enclosed;

ƒ Our final report on the Scottish Local Government elections will be a much more detailed one and at present we anticipate a launch date in September. Our preliminary report has, however, been made available to the Inquiry.

This submission will address all the issues that the Inquiry is investigating. However, the Society is an organisation that campaigns for changes that will improve our democracy. Our contribution on matters of policy is therefore fuller than on matters logistical, managerial and operational.

3. Legislation

Amidst packed timetables at both the Westminster and Scottish Parliaments, many legislative changes were made in recent years which affected the 3 May 2007 elections. Each change required research and consultation. Many required subsequent secondary legislation in either or both Parliaments before implementation was possible

The Electoral Reform Society welcomes many of the changes that have taken place - in particular, the change of voting system for the Scottish Local Government elections.

The key questions appear to be:

Was legislation on the redesign of the AMS ballot paper enacted early enough to allow sufficient time for the development and testing of voter information materials?

Here we need to consider whether the information campaign would have been more effective if the ballot paper design had been agreed earlier and voter information materials produced earlier. That the ballot paper designs were not finalised until January 2007 was certainly a problem for the Society, but we understand an even greater one for the Commission. Although the Scottish Executive and Electoral Commission took the view that voter information would be much more effective if delivered close to the election – a view from which we did not dissent – the legislation was much later than would have been desirable. It is possible that a longer voter education campaign might have reduced the number of spoilt ballots, but we are not convinced that this was a major reason for the high number.

Did the legislative changes, and in particular those made to the arrangements and timetable for postal voting, result in voters being disenfranchised through them not receiving their postal votes in time?

Here the answer is almost certainly ‘yes’. We refer the Inquiry to the report from the Association of Electoral Administrators (‘Official Post Election Report, May 2007 elections’, AEA, July 2007). Although this report is on elections in England and Wales, Scottish electoral administrators clearly experienced similar problems (see section 6 below). While politicians would be wrong to accept the advice of groups such as the

AEA without serious questioning, one can ask whether sufficient weight was given to those with professional experience of electoral administration.

Did the number of legislative changes put unacceptable pressure on returning officers resulting in errors in the administration of the elections?

It appears that the number of changes did put great pressure on electoral administrators but, other than the problems in the printing and distribution of postal ballots, we are not aware of any evidence to support the view that the number of changes in itself resulted in errors in electoral administration or in spoilt ballots.

4. Combined Elections

The Electoral Reform Society argued that, with plans for the introduction of STV, the Scottish Parliament and local government elections should be held on separate days. The combined elections see most media and public attention focussed on the contests for the Scottish Parliament. Local government and the issues for which it is responsible are given scant attention. Holding the elections on separate days would encourage the media and the public to attend to each separately, thereby giving more scope for local councillors to be held to account on their records.

We are of course aware of the problems of voter turnout. Combining the elections produces a higher turnout, particularly for local government elections, and that is a significant benefit. Nevertheless, our preference would be for separate elections where different electoral systems are used.

Whether combining the elections contributed to the level of spoilt ballots is an issue that the Inquiry will want to consider. In a separate paper (appendix 1) we have compared levels of spoilt ballots in Scotland and Wales. The evidence appears to suggest:

1. Holding two elections on the same day is likely to lead to some increase the number of spoilt ballots. However, combined elections have not always produced a significantly greater level of spoilt papers, and in the past the relatively small percentage of voters completing their ballots incorrectly has not been sufficient reason to abandon combined elections where they are perceived to bring other benefits. Spoilt ballots were not a major problem in Scotland’s combined elections of 1999 and 2003. Although the level rose in 2003, this was perhaps because voter information campaigns focused on encouraging people to vote while in 1999 the emphasis had been on how to vote. In Wales the percentage of spoilt ballot fell in 2003 when the local elections were separated from those of the Assembly, but this might have been a consequence of the relatively high level in 1999 when voter interest in a devolved Assembly was not as high as in Scotland (unfortunately we have no information on whether the focus of information campaigns changed as it did in Scotland). With a single election in Wales in 2003, the level of spoilt ballots was still higher than in Scotland where combined elections were held.

2. Holding two elections on the same day using different types of is likely to increase the level of spoilt ballots. Much of the voter information campaign in Scotland aimed to alert electors to the need to use numbers rather than a cross on the local government ballot paper, and it is not surprising that at counts

throughout Scotland we saw voters also using numbers on the AMS ballot paper (as well as crosses on the STV paper). A voter education campaign that needed to explain two quite different systems was necessarily going to be complex and some voter confusion was perhaps inevitable.

This suggests that to reduce the level of spoilt ballots in future elections, either:

a) The elections should be held on different days, even if this reduces turnout for either or both elections; or

b) Both elections should use the same electoral system.

However, even if elections were held on different days, either a year or two years apart, it is still possible that the level of spoilt ballots would be higher if the two elections used different systems.

When deciding whether it is best to hold the Scottish Parliament and local government elections on the same day or on separate days, there is a judgement to be made between maximising turnout and minimising rejected ballot papers. However, whether people go to vote in two elections every four years or in a Scottish Parliament election one year and a local government election the next, leaving only a three year gap until the next Scottish Parliament polling day, they would be aided by the consistency of using the same voting system for each election. If people have to do the same thing each time they vote, the repetition will reinforce their awareness of how to vote validly.

Our strong preference would be for both elections to held using STV. While we have noted arguments for and against combined elections, we have no strong views one way or the other provided the same voting system is used. The evidence from 3rd May shows that voters had few problems using STV while the spoilt-ballot problem was mainly one of the AMS elections. However, we have many other reasons for advocating a change to STV for the Scottish Parliament elections – these we present in section 10 below.

5. Ballot Paper Design

The Electoral Reform Society supported, in principle, the move to a single ballot paper for the Scottish Parliament elections. The change was recommended by the Commission on Boundary Differences and Voting Systems (the Arbuthnott Commission) which was concerned at the perceived ‘second class’ status of regional members: the Commission argued that if, rather than the regional vote being referred to as the ‘second vote’, it were made the first vote, this perception might be partly overcome. Moreover, it is the regional vote which generally determines the overall party composition of the Parliament, so there were good reasons for making this the ‘first’ vote.

The Arbuthnott Commission also noted that a single ballot paper worked well in the mixed-member elections in New Zealand and Germany.

Moreover, the Scotland Office consulted all Scottish political parties and other interested groups on the move to a single ballot paper, and we are not aware that any objected to the change in principle. Alternative ballot paper designs were tested on a sample of

electors and the most popular design chosen (although it is possible that questions need to be asked about the adequacy of the tests).

However, what appears clear is:

a) The single ballot paper was not tested as part of a combined election in which voters were also required to complete an STV ballot paper using numbers rather than crosses.

b) Changes were made to the final version of the ballot paper, truncating the instructions, to accommodate the large number of regional parties. This final design was not adequately tested on a sample of electors.

Testing of the final ballot paper, along with an STV ballot paper, may well have identified the problems that led to the high number of spoilt ballots.

It is reasonable to believe that if two ballot papers had been retained for the Scottish Parliament election, the level of spoilt ballots would have been very much lower. Each paper could have had the clear instruction, “Put one cross against ...”.

Nevertheless, there were good reasons for the change to a single ballot paper and we do not believe it should be abandoned without further consideration of how it might be improved. We therefore recommend a review of the ballot paper design, with further tests using a revised single ballot paper alongside an STV ballot paper. However, if alternative designs do not lead to significant reductions in the level of spoilt ballots, then we would support a return to separate papers for the regional and constituency votes.

We have, however, argued for a change in the electoral system for the Scottish Parliament to STV. It would retain the broad proportionality of the Parliament, it would make all constituency contests competitive, it would remove the need for tactical voting, and it would result in all MSPs being elected with the same democratic mandate. It voters were to be presented with two ballot papers, one for the Parliament and the other for their local council, each to be completed in the same way (i.e. by numbering preferences), then we believe the number of spoilt ballots in both elections would be considerably lower.

One further issue regarding the appearance of the ballot paper are the rules about party and candidate description. The broad, inclusive rules about how political parties could describe themselves on the ballot paper did not help to make it clear to the voters which vote, in which column, was which. The SNP labeled themselves ‘Alex Salmond for First Minister’ on regional lists, allowing the party to appear at the top of the ballot in all of the regions apart from the Lothians due to alphabetical order. It meant that the ‘SNP’ appeared only on the constituency side of the ballot paper, and the side of the ballot paper which was meant to be devoted to a vote for a party, not a person, began, for most voters, with a person’s name. Had there been separate ballot papers, there would have been a greater distinction made between the two votes, so a party using a person’s name to describe themselves might have been less of a red herring. That apart, it is hugely debatable whether the ballot paper is the place for campaigning or sloganising and it is likely some voters had difficulty finding the party they favoured.

6. Postal Voting

The expansion of postal voting was an initiative designed to increase participation in elections. Voters no longer need a specific reason to register for a postal vote – anyone can request one. This postal voting option was first available in 2001 and take-up has been steady.

People were able to register as postal voters right up to the voter registration deadline, which for this election was moved to only three weeks before polling day. The Electoral Reform Society welcomes the extended registration deadline – moving it to after the start of the official campaign capitalises on the increased awareness of the need to register that there is after that point, as awareness of the election rises. We recognise that this posed administrative challenges for the printing and issuing of postal ballots, and that this election saw delays and mistakes in those operations.

We refer the Inquiry to the report of the Association of Electoral Administrators (see section 3 above) which fully documents the similar problems faced by electoral administrators in England and Wales and which makes recommendations for changes in the electoral timetable.

We have supported the changes to allow later registration, but believe the present timetable may be over-ambitious. While we do not have specific proposals to make, it appears to us that the AEA recommendations are a reasonable compromise.

However, our Society does have concerns over the increase in postal voting. Any form of voting outside polling stations is vulnerable to fraud, and although steps can be taken to minimise the risks – such as a move to individual registration and the use of personal identifiers – they cannot reduce the risk to the secrecy of the ballot. With postal voting there can be no guarantee that voters have not been unduly influence in casting their votes by household members or over-enthusiastic canvassers, intimidated or bribed. If postal voting were to produce significantly higher turnouts, we would accept that a balance should be struck between the greater legitimacy of the results obtained by that higher turnout and the increased risk of malpractice. At present, however, it is not clear to us that the increased turnouts, other than when local government elections are held on their own, justify the risks to the security of the ballot.

The Society notes that early voting and weekend voting have been piloted in England and Wales in an attempt to increase participation and decrease the number of people requiring a postal vote. While we have supported the piloting of these approaches, we note that they have not been hugely successful.

In our view, low turnouts will not be overcome by changes in the method of voting. Electors do not abstain from voting because voting is difficult, but because they lack confidence in the efficacy of the process. Our style of politics is not attractive to them, and they do not believe that the outcome of elections will make a great difference to their lives. We do not see any single solution to this problem, but we believe that the use of an electoral system such as STV, creating competitive elections and more representative institutions, is essential if we are to change to a more participatory political culture.

7. Counting the Votes

The election counts for the 3 May 2007 elections were a new experience for all involved. It was the first count of a public election by STV in Great Britain for more than half a century; it was the first time that electronic vote counting was used in Scotland (although it had been used in England, most importantly in London) and it was the first time that optical character recognition technology was used in a public election in Britain.

With several substantial changes being made in the nature of election counts, the risks of things going wrong were greater. It is unfortunate that all did not work perfectly, but that requires an investigation into how errors might be avoided in future rather than an abandonment of the changes introduced.

Many official observers – from around the UK and from other countries – were in attendance at counts across Scotland to view the electronic counting process. The Electoral Reform Society had a sizable team of observers from the UK, USA and Canada, divided around all but a few of the 31 count centres. Our observers reported several issues with the counting process that need to be considered.

Was the decision to use electronic counting justified?

Electronic counting was recommended by Scottish returning officers after examining hand counts of STV elections in Northern Ireland and after studies of the technologies available. While we did not take the view that a hand count would have been too difficult or too long a process, the Scottish Parliament’s decision to use the Weighted Inclusive Gregory version of STV (rather than the rules used in Northern Ireland) greatly strengthened the case for electronic counting of the STV election.

Hand counting STV elections requires the sorting and re-sorting of ballot papers at each stage of the count. At each stage there is scope for human error. We therefore expect that electronic counting would have produced a more accurate result.

Some have argued that the Scottish Parliament elections might have been counted faster by hand. If the elections had run in the same way as in 2003 that might have been the case, but in 2007 delays in counting the Parliament elections arose not just from technical problems with the counting equipment but also from the number of ballot papers that required adjudication. Moreover, it is possible that the use of a single ballot paper would have made a hand count a longer process in 2003.

If a decision were taken to separate the Parliament and local government elections, we would see not great need for electronic counting for the Parliament elections. With combined elections, not using the equipment already in place in counting halls would have been a strange decision (and in some cases one required alternative venues to accommodate tables for hand-counting).

We therefore conclude that the decision to use electronic counting was quite reasonable.

Was testing and preparation of equipment adequate?

With hindsight it is easy to argue that the counting equipment was not adequately tested. We note, however, that a number of tests and demonstrations were held in Scotland

prior to the Scottish Executive’s decision to recommend electronic counting and prior to local authorities’ decisions to use it. It seems clear that these tests did not sufficiently subject the equipment to the volume of ballots, and particularly ballots requiring adjudication, and the operating conditions of a real count, but we have no evidence to suggest that those concerned were other than diligent and professional in their preparation and planning for the elections.

Was the process sufficiently secure?

Although Scotland used electronic counting of paper ballots rather than , that does not guarantee the security of the system. If someone had tampered with equipment in such a way as to mis-record votes or totals, can we be sure that their actions have been detected. The answer is surely ‘no’.

The Inquiry will be aware of the report of the Open Rights Group on e-voting and e- counting in the UK, including on the Scottish elections, as well as numerous allegations of ballot-rigging through electronic counting equipment elsewhere in the world.

We have no reason to doubt the integrity of Returning Officers or their staff or their contractors. We have no evidence of malpractice by electoral administrators in Scotland and we are not aware of any allegations of malpractice. In Britain, unlike many other countries, including part of the US, across the political parties there is trust in the integrity of those who administer our elections. However, that confidence is at least partly based on confidence in the rules and procedures and in the transparency of the process. If that confidence were to be lost, regaining it would be very difficult.

We therefore hope that the Inquiry team will make recommendations, based on the best international practice, on matters such as ensuring that counting equipment has not been tampered with and will accurately read ballot papers and correctly allocate votes to candidates. This may involve procedures for testing equipment after installation but before the start of the count, during the count and after completion of the count. If may involve the random sampling of paper ballots and checks that they have been properly recorded and processed.

Were returning officers and their staffs adequately prepared?

Returning officers were required to deal with counts quite unlike any they had been involved in before. We are not aware of any significant problems having arisen.

We are, however, aware of allegations of inconsistency in the adjudication of doubtful ballot papers and we have observed staff who were, at least in the early stages of counts, not as confident as they might have been in following the guidelines provided by the Electoral Commission. We are not in a position to assess whether inconsistency was as serious a problem as some suggest but it is clearly an area that would merit further study with a view to improving training and performance in future elections.

Did the counts give a negative perception of the elections?

In a traditional hand-count observers, including candidates, agents and the media, see all ballot papers and not just the small percentage that require adjudication. It is possible that the electronic count process created a situation in which the number of

spoilt ballot papers was seen out of context. The problems with spoilt ballots were undoubtedly serious, but not as serious as some of the early media reporting suggested. What observers were shown were the problems, and reports of problems drowned out the positive stories of a new electoral system providing better representation of voters’ views.

Counts, and particularly STV counts in Ireland, can be occasions of excitement and drama as observers can watch the increasing piles of votes for the candidates. In many of the Scottish counts, however, observers knew little about how the count was progressing because of difficulties with monitoring screens and in some cases returning officers could have done more to keep people informed.

Negative perceptions did not cause ballots to be spoilt or lead to inaccuracies in the count, but there is a danger that they could unnecessarily undermine confidence in the elections. The provision of information to those attending counts should receive more attention than it did in some places on 3 and 4 May.

Should the counts have started shortly after the close of polls, or should all counting have been left to the following day?

Elections should be counted as accurately and as quickly as reasonably possible. We appreciate that overnight counts put extra pressure on returning officers and their staff, but: ƒ The counting equipment was not sensitive to the time of day; ƒ Any significant technical problems that arose were handled by technical staff who were not the people who had spent their day in polling stations; ƒ The number of spoilt ballot papers had nothing to do with the timing of the count; ƒ While the performance of staff in the adjudication of ballot papers might have been affected by the timing of the count, we are not aware of any errors having been made as a result of sleep deprivation.

We nevertheless sympathise with electoral administrators who must work excessively long hours, even if only once every four years, and recognise that they have a strong case for arguing for change on the grounds of administrative efficiency, cost and decency to their employees. We also understand that those involved in elections, whether as candidates, activists or journalists, want results as early as possible.

We therefore take no position on when the counts should have been held – to us it appears a political and administrative matter than does not affect the conduct of the election.

8. Public Information

The largest scale public information campaign in the lead-up to the 3 May 2007 Scottish elections was the joint Scottish Executive and Electoral Commission campaign, VoteScotland. This was a well-resourced, wide ranging campaign, motivating and educating the public about registration and voting. The task was huge, and much of the resulting campaign was a credit to the staff involved in managing it.

An evaluation of any such campaign is always going to find areas that could have been improved upon and to set priorities for future campaigns it will be important to research which components were most effective in reaching target audiences. We question the apparently heavy reliance on the provision of information through a website: although the VoteScotland site was excellent, many of those who were less likely to vote were also less likely to be internet users. More could have been done if more resources were available, particularly for groups less likely to vote such as young people and members of ethnic minorities, and we have no reason to believe that the problems encountered in the elections were primarily the result of inadequacies in the information campaign.

However, task facing campaign planners was formidable. Introducing a new electoral system based on preference voting to an electorate that had only experienced ‘X’ voting would have been hard enough, but at the same time electors had to be reminded how to use AMS, and AMS with a single ballot paper when in the past there had been two. Surveys indicated that voter understanding of AMS dropped from 1999 to 2003, possibly because in 2003 the focus of information campaigns changed from introducing the voting system to the promotion of turnout. In 2007 planners therefore faced one new system and another that was not well understood. And at the same time there was concern that a further decline in turnout could damage the perceived legitimacy of the outcome.

These elections also saw returning officers invested with a new duty to promote the election and educate the public about it. The returning officers received dedicated resources from the Scottish Executive to facilitate carrying out this new duty. Different local authorities implemented different approaches, as is appropriate when they have such variety in, for example, geography, demographics or political culture. Many were imaginative and engaging. It is important that these programmes are properly evaluated and the good practice captured so that it can be replicated elsewhere.

For the first time in UK elections there were dedicated information officers at the polling stations, as well as large posters explaining how to vote. We give the Scottish Executive great credit for these innovations. However, whether information officers were as effective as they might have been has been questioned. Some of these officers certainly did their job excellently, proactively offering help to all voters and communicating what to do in the polling booth clearly, accurately and succinctly. Some observers and commentators have, however, reported that other information officers were providing a less satisfactory service:

• One issue raised has been a culturally ingrained reluctance of some voters to admit to ignorance or uncertainty and to ask for help. Some voters who needed help may not have asked for it, hence the need for information officers to take a proactive approach. Not all information officers did so on 3 May.

• Of more serious concern are the reports of polling station information officers providing incomplete or inaccurate information. Examples reported include voters being told to mark two crosses on the Scottish Parliament ballot paper, without the necessary qualification being given that these two crosses had to be in separate columns. From what we know of the rejected ballot papers thanks to our observation teams at the counts, we know that marking two crosses in one column on the Scottish Parliament ballot paper and none in the other column, was a significant cause of ballot paper invalidation and consequent rejection.

Other inaccurate information reported includes voters being told that they could number up to three or four candidates on the local government election ballot paper. This is incorrect and while not invalidating the ballot paper, does make that vote potentially less likely to count than if the voter had been accurately informed that they could mark as many preferences as they chose, up to the number of candidates listed.

Inaccurate information coming at such a critical point is unacceptably damaging to the election process. Even voters who had previously received and taken in the correct information may have been misled into invalid voting by a mistaken information officer.

The recruitment and training processes for polling station information officers need to be reviewed and thoroughly evaluated. The review should include discussions with trainers and people who were information officers to establish how well equipped the officers were to provide clear, accurate information and whether or not they did so. If necessary, recruitment and training processes should be revised for future elections to make sure that these staff are able to carry out their role effectively and supervision needs to be adequate to establish confidence that they are doing so.

The Electoral Reform Society and a number of other organisations carried out their own voter education work, including trade unions and voluntary, community and campaigning organisations. The Electoral Reform Society’s own programme was designed to complement and reinforce the work that others were doing and differed from other organisations’ work with a clear, explicitly focus on the local government, STV election. It included:

ƒ Publications, and in particular ‘Campaigning under STV’ and ‘The : A media guide to Scotland’s local elections, 2007’;

ƒ The production and distribution of leaflets, including ‘Why STV’?, ‘What is STV?’, ‘Questions and Answers on STV’, and ’STV – a myth busting guide’;

ƒ The production and distribution of 200,000 postcards aimed at different sectors of the electorate;

ƒ Distribution of a CD-ROM with presentations on the merits of STV and how it works;

ƒ Recruitment and training of about 40 volunteers who spoke at meetings throughout Scotland;

ƒ Work with the media, including a presentation on STV to a dinner gathering of Scotland’s leading political commentators;

ƒ Promotion of a prize draw through Scotland’s largest circulation tabloid newspaper with respondents receiving text messages reminding when and how to vote;

ƒ ‘Democrazy’, a website aimed at young people: over 100 young Scottish groups uploaded songs with a political content and from a shortlist the winners were chosen by an on-line STV election;

ƒ Poster vans and poster bikes in the few days prior to the elections reminding people to vote by numbers in the local government elections.

The Society’s education programme, whether with the media, other opinion formers, or members of the public, focussed on the local government elections. While we did not ignore the Scottish Parliament election by any means, our objective was to do what we could to ensure that the change to STV went smoothly and was as beneficial to democracy in Scotland as it could be.

9. Roles and responsibilities

Responsibilities for legislation and planning

Running two elections on the same day with each election the responsibility of different bodies is not ideal. While the conduct of the Scottish Parliament elections is a key concern of MSPs, responsibility lies at Westminster. There the general responsibility for elections lies with what was the Department of Constitutional Affairs – now part of the new Ministry of Justice – while the Scotland Office had responsibilities for the Scottish Parliament elections. The local government elections, however, were the responsibility of the Scottish Executive (whose experience in these matters is necessarily limited). The Electoral Commission had a statutory role in the Parliament elections, but was only involved in the local government elections by invitation of the Scottish Executive.

However, although the spread of responsibilities may have been messy and may merit review before further Scottish elections are held, we are not aware of any failures in the decision-making process resulting from confusion over where responsibilities lay. In our view it would be a mistake to believe that the spread of responsibilities contributed in a major way to the problems of 3 May.

Returning Officers’ responsibilities for the conduct of elections

One other group plays a huge role in elections in Scotland: returning officers. They manage nominations, postal votes, polling stations, voter registration, vote counting, and now are also significant in ensuring that voters are aware of the election, are encouraged to vote, and know how to do so. Although the Scottish Executive is responsible for the local government elections, and the Electoral Commission and the Scotland Office are responsible in their different ways for the Scottish Parliament elections, no returning officer has any accountability to any of these bodies.

The Electoral Reform Society has every confidence in the competence and integrity of all of Scotland’s 32 returning officers. However, we note changes are being made to give the Electoral Commission responsibility for setting standards for the performance of returning officers and other electoral administrators. Although the conduct of local elections has been devolved to the Scottish Parliament, the Commission should also have a role in setting standards for these elections.

10 The Voting System

Not to consider the voting systems used on 3 May would be to ignore a key issue.

The

AMS worked tolerably well in 1999 and 2003 when two ballot papers were used and the local elections also required voting with a single ‘X’. But, although AMS has had advantages in giving Scotland a parliament that broadly reflects the views of the Scottish electorate, it is a system with many deficiencies. These include:

ƒ AMS produces two classes of elected members – although they have the same status in the Parliament, regional members are perceived by many electors as being of secondary importance.

ƒ Regional members owe their positions to their standing in their parties rather than to the electorate. A bad candidate at the top of a list might be almost certain to be elected, while a good candidate at the bottom has no hope.

ƒ AMS is not well understood by many Scottish voters. There is evidence that it 2003 there was even more misunderstanding than in 1999 when it was first introduced. It appears that some saw the regional vote – then the second vote – as a second preference.

ƒ In its constituency sections, AMS suffers from the problems of First-Past-the-Post: o many constituency contests are uncompetitive, the seats being quite safe for one party or another giving parties little incentive to campaign and electors little incentive to vote; o many votes are wasted in the sense that they are cast for losing candidates or merely add to already large majorities; o to make their votes count, many voters need to vote tactically; o winners very often have the support of less than 50% of voters (and can even be the most unpopular of the candidates!).

AMS, however, has an additional complication in Scotland. The 2005 reduction in the number of Scottish MPs at Westminster has resulted in Scotland being divided into 59 constituencies for Westminster elections but 73 for Scottish Parliament elections.

For a much fuller description of experience of AMS in Scotland and Wales, we refer the Inquiry to chapter 4 of ‘Britain’s experience of electoral systems’, published by the Society in April of this year.

The Arbuthnott Commission noted many of the difficulties inherent in AMS. The change to a single ballot paper with the regional vote as the first vote was one of the Commission’s recommendations for ameliorating some of these difficulties, but the Commission recommended that, if revisions to the system did not produce adequate improvement then a change to STV should be considered.

It appears to us that a change advocated by the Arbuthnott Commission, possibly because of the manner of its implementation, introduced a new problem in the number of spoilt ballots.

The Single Transferable Vote

While it appears that use of a single AMS ballot paper has caused considerable confusion (in spite of this having been Scotland’s third AMS election), the voters of Scotland took to STV with relative ease. It is an intuitive system; ranking choices in order of preference is an everyday activity for most people, in any number of mundane contexts. In contrast to the problems with rejected ballot papers in the Scottish Parliament election, the spoilage rate for local government ballot papers, using the new STV system, averaged only 1.8 per cent across Scotland. Where the council has provided the information, it is also apparent that the vast majority of voters expressed more than one valid preference (reference). Voters grasped the opportunity provided by STV to show several preferences to ensure that they got the councillors they wanted.

We would not advocate choosing an electoral system on the basis only of the level of spoilt ballots it produces, but quite apart from the problems of 3 May we find the case for STV over AMS overwhelming.

If the system for the Scottish Parliament were to be changed to STV:

ƒ If constituencies of, say, 4 members were used, Scotland would have a parliament as least as representative as with AMS (see “The Implications and Consequences of Introducing STV for the Scottish Parliament Elections”, John Curtice, published by the Centre for Scottish Public Policy, 2006).

ƒ As well as producing proportionality by party, outcomes under STV would be broadly proportional by any factors which significant numbers of voters considered important (e.g. age, gender, ethnicity, policies on the environment, Europe, etc.);

ƒ All members would be elected with the same democratic mandate – all would have faced the electorate in the same way;

ƒ None would be elected simply because of their positions in their parties;

ƒ There would be no safe seats and all elections would be competitive;

ƒ Most electors would have a constituency member of their preferred party;

ƒ The need to attract second and subsequent preferences from voters whose primary allegiance is to other candidates can promote less negative forms of campaigning;

ƒ Independents would not be disadvantaged by need to fight elections with a system based on the quest for party proportionality.

The difficulties that voters encountered on 3 May to us appear yet another reason, and a very substantive reason, for changing the system for the Scottish Parliament elections to STV. With such a change we are confident that future Scottish elections, whether or not conducted on the same day, would not see so many Scottish voters losing their say in their choice of representatives because of ballot paper errors.

Appendices

Appendix 1: Rejected ballots in UK general elections Research Note, June 2007 Lewis Baston (ERS)

UK general election rejected ballots as % of ballot papers

0.4

0.35 0.38 0.38 Rejected % 0.3 0.3 0.31 0.25

0.2

0.15 0.18 0.17 0.15 0.15 0.1 0.13 0.13 0.11 0.12 0.05

0 1964 1966 1970 1974 1974 1979 1983 1987 1992 1997 2001 2005

UK General election rejected ballots 1964-2005 (by official category)

140000 Want of mark 120000 Identified Uncertain 100000 Overvote

80000

60000

40000

20000

0 1964 1966 1970 1974 1974 1979 1983 1987 1992 1997 2001 2005

Source: House of Commons Library Research Paper 04/61 (2004) and Electoral Commission statistics.

Overvote Uncertain Identified Want of mark 1964 15486 15974 6787 2826 1966 11519 27794 8525 2061 1970 14144 17544 7541 2118 1974 12214 19542 7320 3176 1974 13492 15267 6009 2938 1979 72515 36250 5801 3282 1983 27938 15654 4693 2819 1987 14216 15345 4976 2408 1992 14538 17587 5008 2593 1997 25234 61415 4421 2169 2001 22590 69910 3760 2548 2005 20595 54377 4439 2971

The table above includes the four statutory reasons for rejection of a ballot at the count. The bulk of rejections are for two causes, voting for more than one candidate and ‘blank or void for uncertainty’. The two minor causes are means by which a voter can be identified, or want of official mark. The absolute numbers of voters identifying

themselves seem to have been on a declining trend over the previous 40 years (although there also seems to be some relationship with turnout, with more identified ballots in high turnout elections). Want of official mark, which is nearly always the result of administrative error by polling station officials, has fluctuated without much of a trend. Recent changes to the design of ballot papers (such as the replacement of the official mark by a barcode) are intended to reduce this source of error and in some elections there are no rejections (although if a voter obliterated the barcode that would invalidate the ballot).

In recent years the expansion of postal voting has created ambiguity in measuring ballot spoilage. There are additional factors that can cause returned ballots not to be entered into the count, for instance failure to complete a statement of identity and return it with a postal ballot. Strictly, 0.69 per cent of returned ballots were rejected in 2005, but a comparison using the traditional categories of rejected ballots produces the figure of 0.31 per cent. This is complicated by the errors and inconsistencies in the data available. In assessing rejected/ spoiled ballots, it is important to be precise so that data is comparable, but this is regrettably not always possible. An early task for the Review may be therefore to establish with clarity what is meant in each instance by ‘rejected’ or ‘spoilt’ ballots to ensure comparability between different areas and different elections.

Concurrent elections: the evidence from past UK General Elections

The four highest recorded rates of rejected ballots are those elections in which there have been local government elections taking place on the same day.

In 1979 there were elections in all of England outside London for district councils and metropolitan boroughs, and a few areas in Wales. At that time there was a uniform two- tier structure of local government and the lower tier authorities were up for election. Many of these elect all-out once every four years, while some (notably the metropolitan boroughs) elect using partial replacement (by thirds). In 1979 there were also a number of authorities which normally elect by thirds that had all-out elections following boundary changes and voters were electing three councillors from one ballot paper using 3 X- votes. In most authorities electing all-out, there were multi-member elections using the same method. It is not surprising therefore that the 1979 general election saw the largest number by far in recent elections of ballots ruled invalid because of voting for too many candidates. The number of blank or void for uncertainty ballots (‘uncertain’) also increased.

In 1997 there were county council elections in many areas of England. This did not include London or the metropolitan boroughs. Nor, because of changes in local government structure, did it include many of the larger provincial towns and cities because they were now ‘unitary’ authorities and had no county councils. The areas with county council elections were therefore a more rural and affluent, and less young and ethnically diverse, section than average of England. Overvoting increased somewhat from 1992, but the biggest increase was in uncertain ballots. County council elections in 1997 involved only single member local authority elections, the same FPTP system as the General Election, so this increase seems an indicator of inherent confusion in running two elections at a time rather than confusion caused by different electoral system or voter experiences. In Northern Ireland there were local government elections

using STV on the same day, and a relatively high level of overvoting in the Westminster election.

In 2001 the pattern was the same as in 1997 and the total level of rejected ballots was higher, despite the much lower turnout of that general election than 1997 or its immediate predecessors. As a result, the proportion of ballots rejected increased markedly to 0.38 per cent, level with the peak reached in 1979.

In 2005 the pattern was as in 1997 and 2001, with the addition of direct elections for mayor in four English local authorities using the ‘Supplementary Vote’, a system resembling Instant Runoff Voting. Voters are faced with a ballot paper with two columns, the first to indicate their first preference and their second to indicate their second preference. The voter marks an X in each column. There were also boundary changes in the county council elections in England, introducing multi-member wards for the first time in a few areas. The classical categories of spoilage fell, particularly uncertain ballots, although there was an overall increase in the numbers and proportion of returned ballots that were rejected.

This pattern is strongly suggestive of a causal relationship between having more than one election on the same day and ballot spoilage, even if the same system is used. The very high number of overvotes in 1979 also suggests that when the mode of voting is different in two contemporaneous elections, there is a relationship between the mode of election and the type of spoilage that affects the Westminster election. The instructions to use more than one vote in 1979 for local government ballots seems to have resulted in confusion on the General Election ballot leading to a spike in the number of overvotes.

Overvoting, uncertainty and the 2005 General Election

Overvoting and uncertainty by region, 2005

0.9 0.8 Uncertain % 0.7 Overvote % 0.6 0.5 0.4 0.3 0.2 0.1 0 Wales East West London Eastern Scotland Humber Midlands Midlands N Ireland Yorkshire North East North South East North West South West

Source: Calculated from Electoral Commission data.

It must be noted that the quality of data varies from area to area depending on the returns made by EROs. Data was unavailable for a small number of constituencies, and not broken down by reason for rejection in a few more. There also seems to have been varying practice in allocating rejected ballots to each category, particularly between overvotes and uncertain ballots. For instance, the London seat of Bethnal Green & Bow claimed 140 overvotes and no uncertain, while another London seat (Hampstead & Highgate) claimed no overvotes but 144 uncertain. Different EROs seem to classify ballots differently – for instance there are valid arguments for a Westminster election for categorising a ballot filled out STV-style by the voter with a 1 and a 2 as either being an overvote or void for uncertainty.

A quick look at the two main classical sources of spoilage (OVUC for the rest of the paper) by region in the 2005 election shows that Northern Ireland is far and away the region that is most prone to OVUC spoilage, and the difference with other regions is nearly all because of greater overvoting in Northern Ireland.

Overvoting The 18 Northern Ireland constituencies comprise 18 of the top 19 constituencies in the UK for overvoting in 2005. The reason is not difficult to establish. Local elections took place using STV on the same day as the Westminster election, and the multiple marking of the ballot paper used under STV seems to have overflowed into the Westminster ballots.

The same happened, to a somewhat lesser extent, in areas of England where there were concurrent multi-member council elections or SV mayoral elections in 2005. The one Great Britain constituency to exceed any Northern Ireland seat’s rate of overvoting

was Wycombe, where double-member county wards were introduced in nearly the whole of the constituency in 2005 and the number of overvotes went up from 11 in 2001 to 156 in 2005. Overvotes tended to increase in the areas with new multi-member county wards. However, fluctuations in overvotes can also be ascribed to changes in interpretation of some ballot papers from one election to another, and this administrative reason seems to have caused some apparent movements in this variable. For this reason it might be better to look overall at OVUC, combining overvotes and uncertain or blank ballots.

Overvoting and uncertainty The highest rate of spoilage (1.27 per cent of ballots, the only mainland constituency exceeding 1 per cent), was in Norwich South, the then Home Secretary’s constituency, showing a big increase on 2001 – it seems possible that much of this was deliberate protest rather than accidental spoilage.

While Northern Ireland is top for overvoting, the level of blank and uncertain papers is not particularly high, with the result that total OVUC spoilage does not show such a marked regional bias. In some predominantly Nationalist constituencies in Northern Ireland, spoilage fell significantly from 2001 to 2005, perhaps reflecting a fall in deliberate Republican spoilage of Westminster ballots. Some other rises in OVUC spoilage have no obvious explanation, such as the neighbouring Welsh constituencies of Wrexham and Clwyd South. While concurrent local elections, and particularly multi- member elections, seem to have an effect this would need to be more thoroughly tested, allowing for the demographic variables that also affect spoilage (i.e. education, ability in English, income and so on).

However, one particular cluster is very apparent from the 2005 general election – the group of constituencies where there was a mayoral election in 2005 but not in 2001.

OVUC 2001 OVUC 2005 Change in Note % % OVUC % Doncaster Central 0.183 0.357 +0.174 Doncaster North 0.204 0.380 +0.176 Don Valley 0.199 0.244 +0.045 North Tyneside 0.154 0.577 +0.422 Tynemouth 0.184 0.549 +0.365 Stoke-on-Trent Central 0.281 0.448 +0.166 Stoke-on-Trent North 0.271 0.339 +0.067 Partial Stoke-on-Trent South 0.177 0.407 +0.230 UK overall 0.349 0.280 -0.069 (In Stoke-on-Trent North most but not all of the constituency had a mayoral election).

OVUC spoilage in areas with 2005 mayoral elections, 2001 and 2005 %

0.7

0.6 Doncaster Central Doncaster North 0.5 Don Valley 0.4 North Tyneside Tynemouth 0.3 Stoke Central 0.2 Stoke North Stoke South 0.1 UK 0 2001 2005

Conclusion

The evidence from UK general elections suggests that the rate of spoilage tends to increase whenever two elections are run concurrently on the same day. The effects seem to increase when the systems are different, i.e. multi-member FPTP voting in local elections in England, STV voting in local elections in Northern Ireland, and SV voting for elected mayors. The pattern of spoilage is related to the nature of the system being used alongside the FPTP general election, as seen by more overvoting when there are other ballots where voters are expected to make several marks on the paper.

However, standards and definitions of each sort of spoilage seem not to be uniform, which makes analysis difficult, and it is likely that some variation in recorded levels of rejected ballots is caused by differences in administrative practice and local standards of what is considered acceptable. It is therefore desirable that a rigorous definition of spoilage is drawn up for future analysis, and it would help if media coverage of elections were to be clear about the matter as well.

Appendix 2: Rejected ballots in Wales and Scotland

Research note Lewis Baston and Hywel Nelson, ERS July 2007

Compared to Scotland, there has been remarkably little discussion of rejected ballots in the elections for the Welsh Assembly, despite the similarity between the electoral systems in use and the fact that Wales is something of a ‘control’ case given that the conditions remained unchanged between 2003 and 2007, with no concurrent elections and no change in ballot design.

Rejected ballots and ballot factors in Wales

The available data for Wales, however, seems to be very sketchy and not centrally collated. In June 2007 the Electoral Reform Society attempted to contact Welsh local authorities to obtain data for the three Assembly elections. While some local authorities provided full data, there were often missing numbers, particularly for the list PR regional ballot, and it was very difficult to obtain a breakdown by reason for rejection. While it is obviously beyond the Scottish Elections Review’s terms of reference, better reporting and collection of data in Wales would seem to be desirable.

Given the limitations of the Welsh data, some comparisons are still possible. The following table summarises what we could establish from ten fully comparable constituencies.

Rejected constituency ballots in 10 Welsh constituencies (per cent) 1999 2003 2007 Aberavon 1.62 0.79 0.51 Blaenau Gwent 2.16 0.72 0.42 Brecon and Radnorshire 0.82 0.72 0.71 Carm. West and Pemb. South 0.93 0.54 0.46 Ceredigion 0.93 0.57 0.41 Neath 1.03 0.81 0..77 Newport East 1.07 0.60 0.55 Newport West 1.37 0.57 0.75 Preseli Pembrokeshire 0.93 0.76 0.56 Ynys Mon 0.81 0.41 0.32 WALES 1.14 0.61 0.54 sample of 10

The rejection rate in Wales was therefore highest in 1999, the first Assembly election. The pattern across different constituencies reflects a number of factors, including the usual demographic indicators, but it seems possible from the statistics that ballot size in the concurrent local elections also played a part. Local elections in Wales sometimes involve very long ballot papers as the number of seats per ward ranges from 1 to 5 – in some wards the list of candidates is longer than 20 names and voters are expected to place 5 crosses. The highest proportions in 1999, and the largest falls in 2003, seem to be associated with authorities where there are the most multi-member wards (such as

Blaenau Gwent) and least where the pattern is for single member or uncontested elections (such as Brecon & Radnorshire and Preseli Pembrokeshire).

The Welsh Assembly Government decided, principally on the grounds of enabling electors to consider local issues, to separate polling days for Assembly and local government elections. A by-product of this decision seems to have been a marked reduction in the rate of ballot spoilage in the Assembly election in 2003 and 2007 in all the constituencies listed above.

The Sunderland Commission recommended STV for local elections in Wales and followed the Kerley Working Group both in this recommendation and in recommending separate local and Assembly election days. It noted that if STV local elections were introduced in Scotland in 2007:

Scottish voters in May 2007 will at one and the same time be asked to vote in Scottish Parliamentary elections using AMS and in local government elections using STV. It is not a comment on the intelligence of the Scottish electorate if we say that this looks like a recipe for confusion among voters. (Paragraph 6.38)

However, the view of both Sunderland and Kerley was strongly in favour of STV for local elections and neither report can be taken to imply that administrative difficulties in running STV elections alongside other elections is a valid argument against STV.

Wales and Scotland compared

Rejected constituency ballots in devolved elections in Scotland and Wales Scotland Wales

Ballot Other Rejection Ballot Other Rejection type elections rate % type elections rate % Single Multi member 1999 Double member 0.33 Double 1.30* FPTP local FPTP local Single 2003 Double member 0.64 Double None 0.70* FPTP local Multi member 2007 Single 4.07 Double None 0.60* STV local Sources: Wales (ERS survey), Scotland SPICE briefing 07/36. *Based on incomplete available data

The fact that the first election in Wales saw higher spoilage than Scotland is interesting. A number of factors may be at work. The ‘how to vote’ public information campaign in Scotland in 1999 was by common consent a very effective and wide-reaching one, and the campaign in Wales may not have been quite as strong. This might be no fault of the organisers of that campaign, because Wales does not have the national media and other institutions that characterise Scotland and communicating with the electorate would have been more difficult. Also, the narrow margin in the Welsh Assembly referendum and the existence of a substantial body of anti-Assembly feeling in Wales may have resulted in

some voters deliberately choosing to spoil their ballots as a gesture of hostility to the institution that was being elected. Then, perhaps most relevantly for present purposes, the Welsh local elections were taking place concurrently and in some areas involved very long ballot papers.

The steady fall in spoilage in Wales is a marked contrast with the growth in spoilage in Scotland. To some extent, the growth of spoilage in Scotland 2003 was probably connected with the greater effort made in the public information campaign in that election to encourage turnout rather than providing technical ‘how to’ information as in 1999. While there may be many factors at work, and it is hard to measure the intensity and effectiveness of the public education that went on in the two countries in the three sets of elections, the evidence does seem persuasive that multi-member local government elections alongside MMP/AMS elections for devolved bodies do drive up the spoilage rate.

One may also note that ensuring high turnout for local elections is a reason cited for combined elections.

Date Turnout Scotland combined 1999 58.2* Wales combined 1999 46.3* Scotland combined 2003 49.4* Wales Assembly 2003 38.2* Wales local government/ Europe 2004 41.9** Scotland combined 2007 51.7* Wales Assembly 2007 43.4* *Valid constituency votes, source http://www.scottish.parliament.uk/msp/elections/2007/documents/Table13.pdf or http://www.assemblywales.org/mrs071094_2007_assembly_election_results.pdf **In areas with contested local elections, source Rallings & Thrasher Local Elections Handbook 2004 page xv. T

Direct comparison between Scotland and Wales is misleading because of the greater powers of (and greater consensus around) the Scottish Parliament compared to the Welsh Assembly. In 1999 Scottish turnout was 11.9 percentage points higher than Welsh, and the difference in 2003 and 2007 was 11.2 points and 8.3 points. Uncoupling the local elections has not seemed to do any damage to the attractiveness of voting in the Assembly elections.

Turnout in the Welsh local elections is difficult to compare because the only set so far since devolution, in June 2004, took place on the same day as the election, but the relatively high turnout compared to the Assembly election of 2003 suggests that participation might not be unduly harmed by uncoupling the elections in Wales. Whether this would apply in Scotland, where the Parliamentary contest might be presumed to exercise a greater pull than the Assembly in Wales for voters who might not otherwise turn out for local elections, is debatable.

Conclusions

The available data from Wales supports the following observations:

• Moving the date of local government elections was associated with lower rates of ballot rejection in the 2003 and 2007 Welsh Assembly elections.

• Multi-member local elections running alongside an MMP/AMS election generate more confusion and rejected ballots than single-member local elections.

• Welsh voters have found the double ballot format relatively easy to use in three elections, as indeed the Scots did in two.

Appendix 3: Rejected ballots in the London elections 2000 and 2004

Research note Gertrud Malmersjo, ERS, and Lewis Baston, ERS June 2007

The Scottish elections of May 2007 were not the first occasion on which electors were required to multi-task and a high rate of ballot spoilage resulted. The London Mayor and Assembly system introduced in 2000 involved two concurrent elections, and the electoral systems for each institution involve two choices for voters to make. In 2004 the elections coincided with the 5-year term of the European Parliament, adding a fifth voting choice.

Institution System Representative Voting method Constituency Assembly X for a person GLA member (2000 and MMP/ AMS London-wide 2004) X for a party GLA members First vote Mayor X for a person Supplementary for Mayor (2000 and Vote (SV) Second vote 2004) X for a different person for Mayor Europe X for a party List PR MEP (2004)

% Rejected ballots in London elections

18 16 14 12 Constituency GLA 10 List GLA 8 Mayor 1 6 Mayor 2 4 Europe 2 0 2000 2004

% rejected 2000 2004 Constituency GLA 9.4 6.2 List GLA 5.0 9.4 Mayor 1 2.2 3.0 Mayor 2 17.1 14.6 Europe - 0.9

In the 2004 multiple elections the highest proportion of spoiled ballots (6.2%, well over the spoilage rate in the 2007 election for Scottish Parliament constituency seats) was recorded in the London Assembly elections, particularly in the constituency member election. The constituency element of the AMS/MMP election was, as in Scotland, the most prone to spoilage and also the only one that used the traditional first-past-the-post mode of voting.

Mayoral Elections

First preferences 2000 was the first Mayoral election in London and saw a rather high spoilage rate (2.2%). Most of the spoilage (65.3% of rejected ballots) was from ‘more votes than entitled’ suggesting that there might have been some difficulty understanding the system. The Supplementary Vote (SV) system uses two columns side by side on the same ballot paper. The voter marks an ‘X’ for his or her first preference in the first column, and an ‘X’ for second preference in the second column.

At constituency level, the highest and lowest spoilage rates were consistent at both the 2000 and the 2004 elections: City and East had the highest spoilage rates and Bexley and Bromley had the lowest rate.

Looking at the type of spoilage per constituency, City and East had among the highest rates of spoilage for ‘more votes than entitled’ although Lambeth and Southwark and Enfield and Haringey had slightly higher rates. Interesting to note is the Croydon and Sutton spoilage pattern which was different to all other constituencies with 28.1% ‘more votes than entitled’ and 70.9% ‘uncertain or blank’ in 2004 and a similar pattern in the 2000 election although not so striking. This may reflect a different administrative practice.

Disallowed second preferences The second preference vote spoilage was reported at the very high level of 17.1% in 2000 and 14.6% in 2004. However, although some voters will have failed to understand the system, the bulk was most probably not spoilage but voters choosing not to indicate a second preference. This argument is strengthened by looking at the reasons for spoilage where 99.7% were due to the reason ‘uncertain or blank’. Failing to record a second preference vote does represent an understandable voter choice and should not be equated with spoilage, in the same way that voters casting only a single preference in an STV election are making a valid choice. Second preferences indicated by supporters of the leading candidates in an SV election are particularly unlikely to affect the result

and not completing the second vote might be considered entirely rational behaviour by voters.

One set of disallowed second preferences was the result of ‘double voting’, i.e. voting first and second preferences for the same candidate. This was widespread in London.

Overall in London, 7.7 per cent of valid first-preference votes had the same candidate as their second preference, something that could not possibly help that candidate. The incidence of double voting was related to social exclusion and not having English as a second language, as is apparent from the distribution between the borough-based GLA constituencies.

Double vote (as % of total valid first vote) Bexley and Bromley 5.9 Croydon and Sutton 6.4 South West 6.4 Merton and Wandsworth 6.8 Havering and Redbridge 6.9 Barnet and Camden 7.2 Greenwich and Lewisham 7.3 West Central 7.5 Enfield and Haringey 7.8 Ealing and Hillingdon 8.6 Brent and Harrow 8.8 North East 8.7 Lambeth and Southwark 9.4 City and East 11.7

Double voting was particularly prevalent among Respect and Labour voters. Particularly in the context of an election with as many ballot papers as in London in 2004 (mayoral, GLA constituency, GLA list and European), voters who are not particularly aware of the different systems may decide to vote for the party they support at every opportunity. The second column of the mayoral ballot may seem an extension of that principle. In communicating with voters, parties may logically respond to a multiple election by urging voters simply to support them across the board.

The first preference vote spoilage of the 2004 election was higher than the 2000 election (3.3%). however, the spoilage rate for ‘more votes than entitled’ had decreased, now only representing 55.8% of the spoilage. The second preference vote showed a similar pattern to 2000, as 99.6% of ‘spoiled’ second preference votes were marked ‘uncertain or blank’.

Assembly elections

The London Assembly election has, both in 2000 and 2004, used a single ballot paper for the constituency and list votes.

(BBC, 2000)

The high rate of spoilage in the constituency section in particular seems to suggest that single ballot papers do not help voters to understand what their votes do in MMP elections. Spoilage fell in 2004 from the very high levels reached in 2000, but was still overall in excess of the Scottish results in 2007.

London Assembly % spoilage 2000 and 2004

18 Seat 2000 List 2000

16 Seat 2004 List 2004 14 12 10 8 6 4 2 0 LONDON North East North South West South West Central City and East and City Brent and Harrow and Brent Bexley and Bromley and Bexley Croydon and Sutton and Croydon Barnet and and Camden Barnet Enfield and Haringey and Enfield Ealing and Hillingdon and Ealing Havering and Redbridge and Havering Southwark and Lambeth Merton and Wandsworth and Merton Greenwich and Lewisham

% rejected ballots Seat 2000 List 2000 Seat 2004 List 2004 Barnet and Camden 6.5 4.3 5.3 2.2 Bexley and Bromley 6.9 4.9 3.5 1.9 Brent and Harrow 11.6 5.9 7.2 2.9 City and East 16 6.8 10.5 3.6 Croydon and Sutton 8 4.7 4.7 2.2 Ealing and Hillingdon 8.4 5.2 5.9 3.0 Enfield and Haringey 10.2 5.3 6.4 2.5 Greenwich and Lewisham 10,1 4.7 5.9 2.3 Havering and Redbridge 7.9 5.4 5.3 2.7 Lambeth and Southwark 9.5 4.6 8.3 2.8 Merton and Wandsworth 6.3 5.7 5.3 2.4 North East 12.8 5.0 7.7 2.5 South West 7.3 3.8 4.5 1.9 West Central 9.6 4.7 6.7 2.4 LONDON 9.3 5.0 6.2 2.5

There are major differences in the categories of spoilage between the constituency election and the list election. The vast majority of spoilage at the Assembly Constituency elections came from the category ‘uncertain or blank’ votes, (97.4% in 2000 and 95.7% in 2004). City and East had the highest spoilage rate in both 2000 (16%) and 2004 (10.5%).

In 2004 Havering and Redbridge had the highest spoilage on ‘more votes than entitled’ and West Central had the highest rate of ‘uncertain or blank’ (96.8%). In 2000 Merton and Wandsworth had the highest spoilage on ‘more votes than entitled’ and Croydon and Sutton had the highest rate of ‘uncertain or blank’. The spoilage rate for ‘more votes than entitled’ increased from 2.2% in 2000 to 3.7% in 2004.

The spoilage pattern for the Assembly list election looks different. The ‘more votes than entitled’ is 15.7% of spoils in 2000 and 30% of the lower total in 2004.

So although the total spoilage rate was reduced in 2004, the spoilage categories varied and the category ‘more votes than entitled’ increased in 2004. The experience of another list PR election alongside the Assembly in 2004 seems to have led to more understanding of list voting, although the constituency section was still characterised by a very high level of spoilage.

European Election

A proportional list system was introduced for the first time in the 1999 European election. There was a relatively low level of spoilage in the 1999 and 2004 European Parliament elections in London, only 0.9% in the multiple election of 2004, but City and East still present higher figures of spoilage than other constituencies (1.4%). The highest rate of spoilage for ‘more votes than entitled’ (53%) can again be found in City and East.

1.6

1.4

1.2

1.0

0.8 1.4 0.6 1.2 1.1 1.1 1.1 1.0 1.0 0.9 0.9 0.9 0.4 0.8 0.8 0.7 0.7 0.6 0.2

0.0 LONDON Sutton Bromley Camden Haringey Hillingdon Ealing and North East Barnet and Bexley and Enfield and South West Redbridge Southwark Merton and Wandsworth Lewisham Croydon and West Central Lambeth and Havering and City and East Greenwich and Brent and Harrow

Comparative % rates of spoilage in London elections 2004

12.0

Euro 10.0 Mayor 1 GLA seat 8.0 GLA list

6.0

4.0

2.0

0.0 LONDON North East North South West South West Central City and East and City Brent and Harrow and Brent Bexley and Bromley and Bexley Croydon and Sutton and Croydon Barnet and and Camden Barnet Enfield and Haringey and Enfield Ealing and Hillingdon and Ealing Havering and Redbridge and Havering Southwark and Lambeth Merton and Wandsworth and Merton Greenwich and Lewisham

Euro Mayor 1 GLA seat GLA list Barnet and Camden 1.0 2.5 5.3 2.2 Bexley and Bromley 0.6 2.3 3.5 1.9 Brent and Harrow 1.1 3.4 7.2 2.9 City and East 1.4 5.0 10.5 3.6 Croydon and Sutton 0.7 2.6 4.7 2.2 Ealing and Hillingdon 0.7 3.2 5.9 3.0 Enfield and Haringey 1.1 3.0 6.4 2.5 Greenwich and Lewisham 0.9 2.6 5.9 2.3 Havering and Redbridge 0.8 2.9 5.3 2.7 Lambeth and Southwark 1.1 3.0 8.3 2.8 Merton and Wandsworth 0.9 2.6 5.3 2.4 North East 1.0 3.1 7.7 2.5 South West 0.8 2.4 4.5 1.9 West Central 1.2 2.7 6.7 2.4 LONDON 0.9 3.0 6.2 2.5

Tentative observations

The definition of rejection of votes appears as one of the main issues in the analysis of spoilage. The categories for spoilage changed from 2000 to 2004 as more categories were added, however, the main categories for defining spoilage continued to be ‘more votes than entitled’ and ‘uncertain or blank’.

Analysing the data assuming that ‘more votes than entitled’ is a category which exclusively includes rejected ballots where voters failed to understand the process, while the category ‘uncertain or blank’ includes both erroneously completed ballots and deliberately blank votes, the following observations can be made:

The pattern of spoilage seems to indicate that when a new voting system is introduced mistakes on the ballot paper are more easily made. The Mayoral election, the London list Assembly and the European election all introduced relatively new ways of voting to the electorate quite recently. The elections were held on the same date, but the voting system used in each election was different.

• Novelty The London Assembly constituency member election used first-past-the-post and although the overall spoilage rate was higher than the list election the category ‘more votes than entitled’ was significantly higher in the list election in both the 2000 and 2004 elections suggesting that ‘the novelty factor’ had an impact. The fact that the overall spoilage was reduced between the elections might support this observation.

• Confusing ballot papers The constituency and the list candidates appeared on the same ballot in the London Assembly elections which could have lead to some confusion among votes leading to more spoilage. The fact that the category ‘more votes than entitled’ increased in 2004 might support that view, and it is one with relevance to Scotland. The advice of London Elects on their experiences with ballot paper designs would be useful.

• Categories and definitions The definitions of spoilage, particularly the category ‘uncertain or blank’ is misleading as it may include voters who made a conscious decision to vote blank or only give one preference (Mayoral elections). This lack of information can lead to the wrong conclusion, for example, the 2004 Mayoral election had a higher spoilage rate than the 2000 election but the category ‘more votes than entitled’ was reduced in 2004. This could in fact mean that the spoilage rate did not really increase in 2004 but more people simply chose to vote blank. The category ‘uncertain or blank’ is far too broad and it makes it more difficult to analyse spoilage due to for example two different lists on the same ballot paper.

• Deprivation Looking at spoilage from individual constituencies there is a clear pattern emerging with City and East consistently having high levels of spoilage in all elections both in 2000 and 2004. City and East not only has the highest overall spoilage rate, but also one of the highest numbers of ballots rejected on the basis of ‘more votes than entitled’.

City and East includes City of London, Tower Hamlets, Newham and Barking and Dagenham, which apart from City of London, present some of the highest figures of deprivation in London. Comparing with the 2001 census these areas have a high number of people with low levels of education and both Newham and Tower Hamlets have high levels of BME population. The spoilage rates for Newham and Barking and Dagenham in the London Borough election in 2006 were among the highest in London, 1.3% and 1.7% respectively.

Comparing spoilage categories, Barking and Dagenham had their spoilt ballots mainly from the category ‘uncertain or blank’ while Newham and Tower Hamlets presented very high levels of spoilage because of ‘more votes than entitled’. Tower Hamlets had 33.9% and Newham had by far the highest number of rejected ballots in London in this category (58.6%).

This finding is also reminiscent of the Scottish experience, particularly the high rates of spoilage in the East End of Glasgow.

• Spoilage is a complex phenomenon The fact there were so many nuances in the different types of spoilage in London emphasises the point that voter confusion is often multi-dimensional and this is very likely to be the case in Scotland and 2007. Getting as much information as possible from the individual returning officers about their classifications and categories for defining spoilage should provide a valuable source of further knowledge about the situation, and help illuminate the different sources of ballot rejection, including deliberate and accidental causes.