The Electoral Commission Northern Ireland Desk Research - Final Report July 2003 Electoral Commission Desk Research: Final Report

Contents

I Background ...... 1 INTRODUCTION...... 1 OBJECTIVES ...... 2 APPROACH ...... 3 STRUCTURE OF THE REPORT...... 7 II The August 2002 Register ...... 8 INTRODUCTION...... 8 THE TREND AT NI LEVEL...... 8 PARLIAMENTARY CONSTITUENCIES ...... 13 CENSUS WARDS ...... 18 CONCLUDING REMARKS...... 31 III The December 2002 Register ...... 33 INTRODUCTION...... 33 THE NEW SYSTEM...... 33 THE NI LEVEL...... 35 PARLIAMENTARY CONSTITUENCIES ...... 36 CENSUS WARDS ...... 38 CONCLUDING REMARKS...... 47 IV The May 2003 Register ...... 48 INTRODUCTION...... 48 ROLLING REGISTRATION...... 48 CORRELATIONS ...... 54 CONCLUDING REMARKS...... 61 V Conclusions...... 62 INTRODUCTION...... 62 EXTENT OF NON-REGISTRATION ...... 62 THE PATTERN OF NON-REGISTRATION ...... 63 DEPRIVATION...... 64

PricewaterhouseCoopers July 2003 Electoral Commission Desk Research: Final Report

I Background

Introduction

1.1 This report has been commissioned by the Electoral Commission in Northern Ireland to help in understanding the extent and nature of the changes in the patterns of voter registration following the introduction of the Electoral Fraud Act 2002.

1.2 The Electoral Fraud Act received Royal Assent on 1 May 2002. It pertains only to Northern Ireland and applies to all future elections. The Act was introduced primarily in order to overcome widely held perceptions of personation and electoral abuse in Northern Ireland, fundamentally changing the voter registration and identification processes.

1.3 Under the new arrangements, persons who are eligible to vote in Northern Ireland must now register on an individual basis and they must provide certain items of information for identification purposes at the time of registration. Previously, registration had been on a household basis.

1.4 The first Electoral Register to be based on the Act was published on 2 December 2002. This followed the September-October annual canvass undertaken by the Electoral Office for Northern Ireland (EONI). The December 2002 Register contained a total of 1,072,346 entries. This represented a fall of 10 per cent compared to the number of entries on the previously published August 2002 register (Table 1.1).

Table 1.1 The change in the Electoral Register Registered to Change compared to August 2002

vote Register

Number Number %

August 2002 1,192,136 - -

December 2002 1,072,346 -119,790 -10.0

May 2003 1,098,726 -93,410 -7.8

Source: EONI.

1.5 Anyone not on the Register may apply to have their name added through the system of Rolling Registration. As a result of this process, the number on the Electoral Register rose to 1,098,726 as of May 2003. This represented an increase of 26,380 (+2.5 per cent) compared to the December 2002 Register, but the Register published in May 2003 was still eight per cent below the August 2002 Register (Table 1.1).

PricewaterhouseCoopers July 2003 Page 1 Electoral Commission Desk Research: Final Report

Objectives

1.6 The primary objective of this desk research is to compare and contrast the current Northern Ireland Electoral Register with previous Registers and relevant socio- economic datasets, with a view to better understanding how the new registration processes have affected the voter franchise in Northern Ireland.

1.7 The desk research reported here is intended to complement and augment the findings from a separate opinion research exercise that has been commissioned to gauge the profile of those who have not registered and the reasons for their non-registration.

1.8 The specific requirements of this research are to help the Electoral Commission in understanding:

• The current extent of non-registration in Northern Ireland.

• The patterns of non-registration across Northern Ireland.

• Any correlation between decreased registration rates and indicators of deprivation.

1.9 In order to meet these objectives, the May 2003 Register is compared with the December 2002 Register, the August 2002 Register and the 2001 Census of Population results. The extent of non-registration is to be established for each of these Registers, with analysis and detailed commentary provided for the following levels:

• Wards

• Parliamentary constituencies.

• Northern Ireland as a whole.

1.10 A particular point of interest is to compare registration rates and changes in the Register with indicators of deprivation at Census Ward level, in order to help in assessing if there is a correlation between deprivation and registration.

1.11 Other issues to be addressed in the research include:

• The degree to which demographic changes since the 2001 Census may have affected registration rates, including population changes due to people moving and changes due to the balance between young people ageing into the eligible population compared to the death rate.

• The degree to which residential arrangements may have affected registration, for example, students living away from home, armed forces, long-term patients in residential care, and so on.

PricewaterhouseCoopers July 2003 Page 2 Electoral Commission Desk Research: Final Report

• Whether the new registration processes have had a disproportionate impact on people of different religious backgrounds.

1.12 The research also includes an analysis of the effect of the Rolling Registration process. This is the process by which individuals who are eligible to be on the Register, but who were not included during the annual electoral canvass, can apply to be included. Those who wish to change their details, for example, if they have moved house, can also avail of rolling registration. The process enables the Register to be updated on a monthly basis. The Registers for January to May 2003 will be compared to identify if there has been any attenuation of the 10 per cent difference between the December 2002 Register and the August 2002 Register.

Approach

1.13 The programme of work undertaken for this desk research exercise comprised four steps, which are summarised in Figure 1.1 overleaf. The first step was to collate the various datasets that are required for the research. Information on the August 2002, December 2002 and May 2003 Registers were obtained from the EONI. These data show the number of registered electors by Census Ward and Parliamentary Constituency. Unfortunately, other relevant items of information were not available, such as the age distribution of the electorate on each Register. Rolling Registration data were also compiled in respect of January, February, March and April 2003.

1.14 As noted above, a key point of interest in this study is whether there is a correlation between registration rates at Census Ward level and deprivation. The requirement is to compare registration rates with a range of indicators of deprivation, including the NI Measures of Deprivation developed by the Department of Social Policy and Social Work at Oxford University. The NI Measures of Deprivation are described in Annex A below.

1.15 Unfortunately, the existing NI Measures of Deprivation are based on the 1984 Census Ward boundaries. In contrast, the ward-level data in the Electoral Register are based on the 1992 boundaries. These are related, but distinct geographies. For example, there are 582 Wards on the 1992 boundaries compared to 566 on the 1984 definitions. In addition, the 2001 Census Ward data are all published on the 1992 boundaries1.

1.16 For that reason, the approach taken in this report is to examine correlations between registration rates at ward level and indictors that are strongly related to those used for the NI Measures of Deprivation, for example, unemployment rates, benefit claimants, and so on.

1 It will be possible to construct an approximate set of 1984 Wards from the 2001 Census data, using what are called Census Output Areas. However, the Census 2001 data are not yet available at that level of geographical detail.

PricewaterhouseCoopers July 2003 Page 3 Electoral Commission Desk Research: Final Report

Figure 1.1 Electoral Commission Desk Research: Work Programme

A. Voter Registration Data B. Census of Population 2001 preparation C. DSD Social Security dataset D. Election results

Defining non-registration rates: Defining the % of canvassed problem % of eligible voters (Census 2001)

1. Extent 2. Patterns Analysis 3. Correlations Socio-demographic Deprivation

Reporting

1.17 The socio-economic datasets that have been compiled for this research are listed in Figure 2.1. They include the Ward level data published for the 2001 Census, which is available via the web-site of the Northern Ireland Statistics and Research Agency (NISRA). The main groups of Census variables used in this report are listed in Box 1.A along with the relevant population. These data encompass a number of the indicators used in the NI Measures of Deprivation, and also include variables such as socio-economic grouping, which is correlated with income.

1.18 The range of deprivation-relevant variables used for the analysis was augmented by using Ward-level information on social security claimants, extracted from the web- site of the Department for Social Development (DSD).

1.19 Finally, information was obtained for a number of the more recent Elections, both local and for Westminster. The main point of interest was to calculate turnout rates, as an indicator of the extent of competition for votes, which may affect the propensity to register. Election data are only available for Parliamentary Constituencies and District Electoral Areas (DEAs), which are aggregates of 5-6 contiguous Census Wards. Election data are not available for Census Wards.

PricewaterhouseCoopers July 2003 Page 4 Electoral Commission Desk Research: Final Report

Box 1.A Main groups of Census 2001 variables analysed to Ward level Heading Relevant population

• Usually resident population • All ages

• Age groups • Sixteen age bands

• Qualifications and students • Aged 16-74

• Economic activity • Aged 16-74

• Community background • All ages

• Stated religion • All ages

• Country of birth • All ages

• Ethnic background • All ages

• Students • 16-17 and 18-74

• Tenure • Households

• Dwelling type • Households

• Socio-economic grouping • 16-74

• Health • All ages

• Communal establishment • All ages residents

• Household composition • Households

1.20 The second stage of the research was concerned with the problem of defining ‘non- registration’ at the various geographic levels for which analysis is required. This is discussed in further detail in Section 2. A key point made in that discussion is that it is not possible from the available information to calculate ‘true’ non-registration rates for any geographic level. This does not mean that no analysis of the issue is possible and a framework is developed that enables progress to be made in addressing the research requirements.

PricewaterhouseCoopers July 2003 Page 5 Electoral Commission Desk Research: Final Report

1.21 The third stage of the approach comprised the statistical analyses of correlations at Census Ward level between voter registration and socio-economic indicators, including deprivation measures. It is, however, necessary to issue a strong word of warning at the very outset concerning the inferences that can be drawn by comparing Ward-level variations in registration rates with socio-economic indicators. Such comparisons may, for example, suggest that areas with an above-average concentration of young people aged 18-24 also tend to have lower than average registration. The observed spatial relationship may be significant, in the statistical sense. Such geographical relationships do not necessarily imply that people aged 18- 24 are less likely to be registered than are other age groups. That is, analysis of spatial data can highlight interesting geographical associations between registration and different socio-economic and demographic indicators, but considerable care and caution needs to be exercised in drawing conclusions regarding cause-and-effect processes as these affect individuals.

1.22 Partly for that reason, but also to help in better understanding the socio-economic and demographic relationships, the statistical analysis has been augmented by a number of case studies of individual wards. The case study approach has been implemented by preparing detailed statistical profiles for a selected set of wards. The statistical correlations identified in the analyses for all wards are then illustrated for the case study wards, for example, to show what effect relative deprivation may have on registration rates.

1.23 The following wards were chosen for the case study approach:

1.24 Whiterock and Island in Belfast. These are two wards with significant concentrations of deprivation. In other respects, the two wards provide a number of sharp contrasts. Whiterock is almost exclusively Catholic (99 per cent) while Island is overwhelmingly Protestant (90 per cent). In the Island ward, 30 per cent of the population aged over 18 are aged 60 or over, compared to 21 per cent in Whiterock. Though both wards saw sharp falls in registration levels from August 2002 to December 2002, there was a significant ‘bounce-back’ in Whiterock but not in Island.

1.25 Massereene in South Antrim. This is a more affluent ward located in the suburbs of Belfast. The ward was chosen specifically to extend the geographical spread of the case studies and to inject a contrast on the deprivation issue. Also, it has been growing relatively quickly in population in recent times. The religious mix in this ward is 57 per cent Protestant and 39 per cent Catholic, which is reasonably close to the Northern Ireland average (53:44).

1.26 Tempo in Fermanagh and South Tyrone. This is a classic instance of a ward where there is intense electoral competition. The recent Census of Population shows the Protestant and Catholic communities are very finely balanced, with an almost even split (49 per cent Protestant and 51 per cent Catholic). The ward also extends the geographical spread into a more rural area west of the Bann.

PricewaterhouseCoopers July 2003 Page 6 Electoral Commission Desk Research: Final Report

Structure of the Report

1.27 The report is structured as follows.

1.28 Section 2 focuses on the August 2002 register. It does this partly in order to provide a platform for the discussion of the use of the Electoral Register to calculate registration rates using Census data on the resident population. A second objective is to establish the baseline situation regarding the pattern and extent of ‘registration rates’ that pertained prior to the introduction of the Electoral Fraud Act. The reason for this is that, in order to understand what impact the change in registration procedures may have had, it is first necessary to understand the pre-existing system.

1.29 Section 3 discusses the December 2002 register. The first part of the Section considers the pattern of registration rates, again at various spatial scales. The issue of interest is the extent of non-registration, and the factors associated with this. The Section then considers the geographic pattern of change in registration, particularly the relationship with indicators of deprivation.

1.30 Section 4 analyses the May 2003 Register. The Section commences with a discussion of the Rolling Registration process, examining the consequent change in the Register that occurred from December 2002 to May 2003. Following this, the Section addresses the issue of whether this has had an impact on the geographic pattern of voter registration, with a particular focus on any relationship that may exist with relative deprivation.

1.31 Section 5 presents a summary of the main conclusions to be drawn from the research.

PricewaterhouseCoopers July 2003 Page 7 Electoral Commission Desk Research: Final Report

II The August 2002 Register

Introduction

2.1 This Section of the report analyses the August 2002 Electoral Register, the last to be published under the old system.

2.2 The Section commences by considering the trend in the Electoral Register at Northern Ireland level, relative to the published estimates for the population aged 18 and over. The main point of interest in this discussion is whether it is feasible to use the Electoral Register under the old system as a measure of the extent of non- registration in Northern Ireland. This is very important in understanding whether the fall in the Register following the introduction of the new system of individual registration represents an increase in non-registration, or conversely leads to a truer and more accurate picture.

2.3 The conclusion drawn is that the old Electoral Register is unlikely to give an accurate measure of the extent of non-registration in Northern Ireland. This is even more the case at sub-NI geographical scales, including Parliamentary Constituencies and Census Wards. Nonetheless, some useful insights can be gleaned by considering the trends at Parliamentary Constituency level and also the pattern in the Register at Ward level relative to the population as at the 2001 Census. The latter analysis includes a statistical approach designed to provide a better understanding of geographical variations in the Register at Census Ward level. The main purpose of the statistical analysis is to identify socio-economic and demographic correlates of registration, particularly as these relate to relative deprivation.

The Trend at NI Level

2.4 The ‘true’ registration rate for the population as a whole is given by the following formula:

Persons registered to vote

Registration Rate equals divided by

The population qualified to vote

PricewaterhouseCoopers July 2003 Page 8 Electoral Commission Desk Research: Final Report

2.5 Thus, if the number of persons who are registered to vote is known as well as the size of the population that is qualified to vote through meeting the entitlement criteria (Box 2.A), then it is possible to calculate the registration rate for a given area. Subtracting the registration rate from 100 gives the non-registration rate2.

Box 2.A The Entitlement to Vote

A person was qualified to register at the annual canvass in autumn 2002 if he or she:

• Is a British citizen, Irish citizen, Commonwealth citizen or citizen of a member state of the European Union.

• Is 18 years of age or over, or will become 18 on or before 30 November 2003.

• Has been resident in Northern Ireland during the whole of the three month period 16 July - 15 October 2002.

• Is not subject to any legal incapacity to vote (this mainly covers convicted prisoners and certain Peers of the realm). There are other grounds which affect a very small number of people.

• Public Servants and members of the defence forces who are on duty overseas for long periods are qualified to register, as are members of their family. Also persons who would be qualified if they lived in Northern Ireland but now live abroad, provided that they have not been away for more than fifteen years. Such persons must be British citizens and can only vote at Westminister elections. They form a very small portion of the Register.

Source: EONI website.

2.6 The simplest way of calculating the registration rate for Northern Ireland as a whole is to use the total on the Electoral Register as a measure of the number of persons registered to vote and the official estimates for the population aged 18 and over as the measure for the size of the population qualified to vote. Doing this gives the picture shown in Figure 2.1, that is, a relatively stable estimated ‘registration rate’ of around 94-96 per cent.

2 In the discussion that follows, it is more convenient to focus on the registration rate, though clearly the non- registration rate is simply the obverse of the registration rate.

PricewaterhouseCoopers July 2003 Page 9 Electoral Commission Desk Research: Final Report

2.7 Does this then mean that the ‘non-registration’ rate in Northern Ireland under the old system was in turn fairly constant at about 4-6 per cent of the population? Unfortunately, it does not. This is readily seen by examining the components of the Electoral Register under the old system. Specifically, a number of features of that system indicate that the number of entries on the Register was in excess of the total number of persons eligible to vote by dint of being registered.

Figure 2.1 Electoral Register as per cent of Population 18+

100 98 96 94 92 90

Per cent 88 86 84 82 80 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003

Sources: EONI; NISRA

2.8 As noted in Section 1 above, the old Register was based on an annual canvass of households. Prior to the canvass, each household would be issued with a registration form on which the names of qualified persons would be entered. Not everybody returned their forms. The non-return rate was typically in the order of one in ten.

2.9 Under the old system, if a household did not return its form, its existing details would be carried forward on the Register for one year. In some cases, households may simply have neglected to return their forms. But there would also be instances where one or more members of the household may have moved away. Thus, the practice of carrying household details forward undoubtedly inflated the Register somewhat, though the effect is not known.

2.10 A second way in which the Register may have held more entries than there were persons actually entitled to vote was that a person could, quite legitimately, be registered at more than one location, for example, a person with a second home. Considered in terms of calculating a registration rate as described above, this inflates the Register because a person is only entitled to vote once, even though they may be registered more than once.

PricewaterhouseCoopers July 2003 Page 10 Electoral Commission Desk Research: Final Report

2.11 Finally, the Register may have been inflated through instances of fraudulent registration. Clearly, this was part of the rationale for the introduction of a new system of registration. But the extent of fraud on the old Register is simply not known. The 1998 report by the Northern Ireland Office (NIO) on the administration of elections in Northern Ireland stated that “the Review suspects that there is a level of false registration which is carried out with the express intention of abusing the system”, but no concrete evidence was adduced as to the incidence of false registration. Indeed, much of the discussion of registration in the NIO report centred on ways of addressing non-registration.

2.12 Overall, therefore, it is likely that the old Electoral Register over-stated the number of persons entitled to vote. As illustrated below, each of the three factors discussed above must be subtracted from the Register in order to get an accurate picture of the number of persons eligible to vote:

The electoral register in the old system

Fraudelent entries Carried forward e.g. moved away

M ultiple entries

Persons eligible to vote

2.13 The problem is, however, that none of these factors are observable in such a way that their cumulative impact on the Register can be calculated. Regarding the size of the population qualified to vote, the use of the official estimates of the usually resident population in Northern Ireland is also problematic, for the reasons shown below:

PricewaterhouseCoopers July 2003 Page 11 Electoral Commission Desk Research: Final Report

The Usually Resident Population

Minus deaths, out-migrants

Minus persons not qualified The Resident measured Population - population Mid-year or Usually Census (March) Resident Population The population qualified Plus Attainers, to vote In-migrants Plus qualified persons living away

2.14 There are two main sets of problems with the use of official estimates for the usually resident population as a measure of the population qualified to vote. First, these estimates will generally refer to a different, earlier time period than the one in which the annual canvass for the Register takes place. The 2001 Census, for example, refers to March of that year. As their name implies, the mid-year population estimates refer to June of each year. Some population changes would be expected in the period between the population estimates being prepared and the canvass for the Electoral Register. These changes include losses due to deaths and out-migration and gains due to in-migration and young people ageing into the eligible population, that is, those attaining the age of 18.

2.15 These components of population change, it should be said, are likely to have only a minor effect at the Northern Ireland level3. In recent years, the population aged 18 and over has been growing at an annual rate of 0.7 per cent. This has been almost entirely due to the excess of persons attaining the age of 18 over deaths, as the numbers of those migrating into Northern Ireland has been roughly in balance with the numbers leaving (NISRA, NI Annual Abstract of Statistics, 2001).

3 As discussed below, this is not the case at sub-regional level.

PricewaterhouseCoopers July 2003 Page 12 Electoral Commission Desk Research: Final Report

2.16 Even at NI level, however, the usually resident population may have diverged from the eligible population at the time of the annual canvass for two further reasons. First, as can be seen from Box 2.A, not all those who are usually resident are eligible to vote in Northern Ireland. Again, this is likely to have a fairly minor influence. In the 2001 Census, the proportion of the population born outside the EU was 1.2 per cent, and some of these would be Commonwealth citizens. HM Forces stationed in Northern Ireland were also counted as part of the resident population in the 2001 Census, but they would have comprised less than 0.5 per cent of the population aged 18+. As with the components of population change, these effects are more pronounced in some local areas, for example, in those Census Wards where the Forces are stationed. This is discussed further below.

2.17 Second, some individuals living away from Northern Ireland, and hence not included in any estimate of the usually resident population, will have an entitlement to vote and may be registered. This effect, it can be noted, will tend to offset the inclusion in the resident population of people who are not eligible to register for a vote in Northern Ireland.

2.18 Overall, therefore, at the NI level, the factors that result in the estimate of the usually resident population being different from the eligible population tend to offset each other. This is in contrast to the position regarding the Electoral Register under the old system, where all of the major effects result in deductions from the Register. It is therefore not unreasonable to conclude that the most recent estimate of the NI resident population provides a reasonable basis for calculating registration rates. It must, however, be appreciated that while this may be the case at NI level, it is certainly not the case at sub-NI level. This can clearly be seen by considering ‘registration rates’ for Parliamentary Constituencies.

Parliamentary Constituencies

2.19 From 1997 to 2001, NISRA estimates that the overall NI population aged 18 and over expanded by 2.8 per cent, or 0.7 per cent per annum. The distribution of that growth was, however, very uneven at District Council level4. As can be seen from Figure 2.2, Belfast suffered a 1.6 per cent population decline while Banbridge recorded an increase of 8.4 per cent (see also Table B2.1).

4 This is the most detailed geographical level for which such estimates are published. It should also be noted that mid-year estimates are not produced for the Parliamentary Constituencies.

PricewaterhouseCoopers July 2003 Page 13 Electoral Commission Desk Research: Final Report

Figure 2.2 Population Change by Local Government District Persons aged 18+, 1997-2001

Antrim Ards Armagh Ballymena Ballymoney Banbridge Belfast Carrickfergus Castlereagh Coleraine Cookstown Craigavon D e rry Down Dungannon Fermanagh Larne Lima va dy Lisburn Magherafelt Moyle Newry & Mou Newtownabbey North Down Omagh Strabane N. Ireland -2-10123456789 Per cent change

Source: NISRA, Mid-year Population Estimates

2.20 In very broad terms, a similar pattern of change can be seen in the Electoral Register over the same period. Thus, comparing the August 2002 register with that used for the 1997 Westminster elections, the total number on the Register grew by one per cent. However, reflecting the population movements shown in Figure 2.2, there was a decline in each of the Belfast constituencies accompanied by growth in all other areas. Thus, the change in the Register varied from -8 per cent in Belfast South to +7 per cent in South Down (Figure 2.3. See also Tables B2.2 and B2.3).

2.21 Indeed, the decrease in the Belfast constituencies was in excess of what would have been expected from the change in the population aged 18 and over (compare Figures 2.2 and 2.3). This indicates that the geographical pattern of changes in the Register is influenced by more than simply population movements. This issue is explored further later in this Section when the distribution of the Register by Census Ward is compared with various socio-economic indicators from the 2001 Census of Population.

PricewaterhouseCoopers July 2003 Page 14 Electoral Commission Desk Research: Final Report

Figure 2.3 Difference between August 2002 Electoral Register and Westminister 1997 Register: Per cent

Belfast East Belfast North Belfast South Belfast W est East Antrim East Londonderry Fermanagh & South Tyrone Foyle Lagan Valley Mid Ulster Newry & Armagh North Antrim North Down South Antrim South Down Strangford Upper Bann West Tyrone

N. Ireland

-9-8-7-6-5-4-3-2-101234567 Per cent

Source: EONI.

2.22 It is also of interest to compare the Register within each Parliamentary Constituency with the size of the resident population aged 18 and over. As shown in Figure 2.4, the August 2002 Register as a percentage of the resident population varies widely, from 77 per cent in South Belfast to 103 per cent in Mid Ulster and Fermanagh and South Tyrone (see also Table B2.4). Clearly, the difficulties involved in using the Electoral Register to calculate ‘true’ non-registration rates are greatly accentuated at sub-NI geographical scales.

Figure 2.4 August 2002 Electoral Register as per cent of resident population aged 18+

Belfast East Belfast North Belfast South Belfast W est East Antrim East Londonderry Fermanagh & South Tyrone Foyle Lagan Valley Mid Ulster Newry & Armagh North Antrim North Down South Antrim South Down Strangford Upper Bann West Tyrone

N. Ireland

0 20406080100 Per cent

Sources: EONI, 2002; Census, 2001

PricewaterhouseCoopers July 2003 Page 15 Electoral Commission Desk Research: Final Report

2.23 Taken at face value, the Register for Mid Ulster, Fermanagh and South Tyrone or Newry and Mourne would suggest that more people are registered in these Constituencies than actually live there. Part of the reason that the Register exceeds 100 in these Constituencies may be the tendency noted above for the old Register to give an inflated estimate of the true number of registered eligible voters, though there is no evidence as to how or why the inflation effect should vary geographically. The more significant factor explaining the variation in raw ‘registration rates’ by Parliamentary Constituency is a dislocation between the geography of the electorate and the distribution of the population. That is, people on the Register are not necessarily registered in the area in which they usually reside.

2.24 A very good example of this is the student factor. Under the old system of registration, many students attending third-level institutions but living away from the family home were registered at their home rather than their term-time address. By contrast, in the most recent Census of Population, which counted people at the place where they were living on 31 March 2001, the usual place of residence for such individuals would have been their term-time address. Thus, the very large share of students amongst the resident population in Belfast South (Figure 2.5) is clearly a major component of the apparently low registration rate in that constituency.

Figure 2.5 Full-time students aged 18-74 Per cent of population aged 18+

Belfast East Belfast North Belfast South Belfast West East Antrim East Londonderry Fe rma na gh & South T yrone Foyle La ga n Va lle y Mid Ulster Newry & Armagh North Antrim North Down South Antrim South Down Stra ngford Upper Bann West Tyrone N. Ireland

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Per cent

Source: Census, 2001.

2.25 But the student factor does not explain all of the variation in raw registration rates between Belfast South and the other Constituencies. If registration rates were to be equalised across all constituencies, while holding the resident population constant in each constituency, an additional 15,000 registrations would need to be re-allocated to Belfast South. But in 2001, according to the Census, there were 11,000 full-time students aged 18-74 in South Belfast. This was 8,000 in excess of what would be expected if such students were to be distributed pro-rata on the population aged 18+, meaning that another 7,000 or so non-students would have to be ‘re-allocated’ to Belfast South in order to equalise its registration rate with the NI average.

PricewaterhouseCoopers July 2003 Page 16 Electoral Commission Desk Research: Final Report

2.26 In other words, the student factor potentially explains, at the very most, a little over 50 per cent of the variance between the South Belfast registration rate and the average rate for Northern Ireland as a whole. It is not possible at this juncture to say what these other factors might be. There may, for example, be a tendency for people moving into the area from outside Belfast to remain registered at their family home address, at least initially5. Further, the area contains a relatively high proportion of people who were born outside Northern Ireland, especially from outside the EU (Figure 2.6). Though it should be noted that not all such persons lack an entitlement to vote in Northern Ireland.

Figure 2.6 Population born outside Northern Ireland

Belfast East Belfast North Belfast South Belfast West East Antrim East Londonderry Fe rma na gh & South T yrone Foyle Great Britain La ga n Va lle y Mid Ulster Rep. of Ireland Newry & Armagh Other EU North Antrim North Down Elsewhere South Antrim South Down Stra ngford Upper Bann West Tyrone

N. Ireland

0 5 10 15 Per cent of population

Source: Census, 2001.

2.27 Such considerations would suggest that the variations at Parliamentary Constituency level in raw or unadjusted registration rates are not solely a function of spatial differences in Register ‘inflation’ under the old system or the dislocation between the distribution of the population and the geographical pattern of registrations. The differences are very likely to also reflect other factors. For example, it may be that some age groups are more likely to register than are others. Or registration rates may be associated with the geography of deprivation.

2.28 Such putative influences on the propensity to register cannot be analysed at the level of Parliamentary Constituencies. It is necessary to go further down the spatial scale and consider variations in registration rates at Census Ward level. This is the focus of the next part of this Section.

5 Perhaps during a ‘settling-in’ period, when living in rented accommodation; 17 per cent of dwellings in the constituency are privately rented, compared to an NI average of 7 per cent, so population mobility and household turnover would be higher than average.

PricewaterhouseCoopers July 2003 Page 17 Electoral Commission Desk Research: Final Report

Census Wards

2.29 The problems involved in identifying actual or true registration rates are considerably magnified at Census Ward level. There are a number of reasons for this, including:

• Wide variations between wards in the mix of demographic and socio- economic factors that may be correlated, negatively or positively, with the propensity to register.

• Local concentrations of ‘special’ factors, such as students, army bases, communal care establishments, and so on.

• Unobservable systematic influences. For example, the influence of the factors discussed above that may tend to inflate the register.

• Unobservable random influences. For example, a dislike of filling in forms and providing personal information to the Government may not be predictable from socio-economic and demographic factors.

2.30 Reflecting all of the above, there are large disparities between wards in raw or unadjusted registration rates calculated using the Electoral Register in each Census Ward as a percentage of the local population aged 18+. This is clear from Figure 2.7, which shows raw registration rates ranging from 126 per cent (Forkhill in Newry and Mourne) to 41 per cent in the Botanic ward in Belfast South (see Annex Table D.1 for a full listing of the registration rates shown in Figure 2.6, arranged by District Council and Census Ward. See also Maps A.1 and A.2 in Annex C). Indeed, 44 per cent of the 582 Wards showed a ‘registration rate’ of over 100 using the August 2002 Register.

Figure 2.7 August 2002 Electoral Register compared to Population 18+ Census Wards

9 8 7 6 5 4 3 2 Electoral Register ('000s) 1 0 0123456789 Population aged 18+ ('000s)

Sources: EONI; Census, 2001.

PricewaterhouseCoopers July 2003 Page 18 Electoral Commission Desk Research: Final Report

2.31 Some insight into the range of factors associated with variations in registration rates at Ward level can be obtained by looking at the ‘extremes’ of the distribution. Table 2.1 gives the 20 wards with the lowest raw registration rates. The student factor is immediately obvious in a number of Wards, including Botanic, Windsor, Stranmillis, Strand in Derry and Strand in Coleraine, Ballynafeigh, Rostulla, University and Portstewart. In these Wards, over one in four persons aged 18 and over is a full-time student, compared to a NI average of four per cent.

Table 2.1 August 2002 Register as per cent of Population: The 20 Census Wards with the lowest raw registration rates Parliamentary Registration District Council Ward Constituency rate %

Belfast South Belfast Botanic 42

Belfast South Belfast Windsor 44

South Antrim Antrim Aldergrove 46

Belfast South Belfast Stranmillis 55

North Down North Down Loughview 56

Lagan Valley Lisburn Wallace Park 62

West Tyrone Omagh Lisanelly 63

Foyle Derry Ebrington 64

East Londonderry Limavady Gresteel 67

Foyle Derry Strand 67

East Londonderry Coleraine Strand 71

Belfast South Belfast Ballynafeigh 71

South Down Down Killough 73

East Antrim Newtownabbey Rostulla 73

East Londonderry Coleraine University 77

West Tyrone Omagh Strule 78

East Londonderry Coleraine Portstewart 79

East Londonderry Limavady Magilligan 80

Belfast South Belfast Blackstaff 81

East Londonderry Limavady Roeside 82

PricewaterhouseCoopers July 2003 Page 19 Electoral Commission Desk Research: Final Report

2.32 Other influences are also manifest. Thus, for example, there are army bases in Aldergrove, Loughview, Wallace Park, Ebrington, Lisanelly, Gresteel, Killough and Strule. Magilligan contains one of Northern Ireland’s two prisons.

2.33 Beyond the 80 per cent mark, however, it becomes increasingly difficult to identify any one individual location-specific factor as providing at least a partial explanation for some degree of variance from the average. Thus, for example, it is not immediately obvious why Roeside and Blackstaff should be in the bottom 20 for raw registration rates. The former is a largely Protestant area while the latter is a relatively deprived inner-city Protestant area in Belfast (it encompasses the Sandy Row area).

2.34 These variances are analysed below using a multivariate statistical model designed to take account of, or control for, a range of socio-economic and demographic characteristics. Prior to discussing how such an approach can be of benefit, it is useful to briefly consider the 20 Wards with the highest registration rates based on the August 2002 Register. These are shown in Table 2.2.

2.35 The 20 wards listed in Table 2.2 bear certain similarities, in that they are mostly located in largely rural constituencies in the west and south of Northern Ireland. Unlike the 20 wards with the lowest rates, however, it is difficult to identify a set of specific local factors that can potentially contribute to an explanation.

2.36 One reason why a Ward may have a registration rate greater than 100 is that there may have been significant population change due to, for example, new housing developments or a youthful population profile with an above-average rate of entry into the Register by those attaining the qualifying age. It is possible to take account of the latter factor, by including relevant demographic variables. The impact of new building developments is much more difficult. Though it should be noted that the August 2002 Register was essentially based on the October-November 2001 canvass, so the March 2001 Census data are unlikely to have changed significantly.

2.37 Alternatively, these 20 Wards may be ones in which a combination of the above influences is married with a high proportion of students living away from the family home, but registered in their home area. This factor is impossible to assess, since information is only available on where students live, not where they hail from.

2.38 Overall, therefore, it is rather more difficult to ‘explain’ the Wards at the upper end of the spectrum of registration rates. This is an issue to which we return later in this Section.

PricewaterhouseCoopers July 2003 Page 20 Electoral Commission Desk Research: Final Report

Table 2.2 August 2002 Register as per cent of Population: The 20 Census Wards with the highest raw registration rates Parliamentary Registration District Ward Constituency rate % Fermanagh & South Fermanagh Donagh 109 Tyrone Mid Ulster Dungannon Donaghmore 109 Fermanagh & South Dungannon Caledon 109 Tyrone Newry & Armagh Newry and Mourne Fathom 109 Fermanagh & South Dungannon Castlecaulfield 110 Tyrone Fermanagh & South Dungannon Augher 110 Tyrone West Tyrone Strabane East 110

South Down Newry and Mourne Rostrevor 110

Mid Ulster Magherafelt Swatragh 111

Mid Ulster Dungannon Altmore 111

Mid Ulster Cookstown Killycolpy 111

West Tyrone Omagh Termon 111

East Londonderry Limavady Enagh 112

West Tyrone Omagh Coolnagard 112

Newry & Armagh Armagh Killeen 113

West Tyrone Omagh Drumnakilly 115

Newry & Armagh Newry and Mourne Creggan 116

Newry & Armagh Newry and Mourne Silver Bridge 116

Mid Ulster Dungannon Washing Bay 118

Newry & Armagh Newry and Mourne Forkhill 129

PricewaterhouseCoopers July 2003 Page 21 Electoral Commission Desk Research: Final Report

2.39 As noted above, it is necessary to adopt a multivariate statistical approach in seeking to ‘explain’ or account for variations between Wards in raw registration rates. The basic approach is illustrated in Figure 2.8 below. Thus, following the discussion above, it is possible to identify a set of socio-economic and demographic factors that can be measured at Ward level and that are hypothesised to be associated in some way with the propensity to register to vote. Information on a range of such factors is available from the datasets that were collated for this report.

Figure 2.8 Statistical modelling approach to voter registration rates at Census Ward level

Systematic influences e.g. deprivation, age profile, Statistical model communal establishments

Randomly Expected voter distributed Actual voter registration influences e.g. registration rates, adjusted 'big brother' rates for systematic attitude? influences

Non-observables Compare e.g.fraud v. actual with 'dead-wood' expected

2.40 An important point to note from Figure 2.7, and which follows from the discussion at the beginning of this Section, is that a number of potential influences on the size of the Register in each Ward are simply not observable. This means that it is impossible to distinguish between possible sources of ‘inflation’ of the Register at Ward level. Thus, for example, the inflationary effect of carry-over from one Register to another cannot be distinguished from any influence that may be due to fraud or other practices that may artificially increase the number on the Register. Hence, this report does not, because it cannot, seek to distinguish different sources of inflation of the Register. For that reason, no comment is made on whether any particular set of results implies a higher or lower likelihood of fraud or otherwise.

PricewaterhouseCoopers July 2003 Page 22 Electoral Commission Desk Research: Final Report

2.41 Prior to presenting the multivariate results, it is useful to first explore some of the relationships that may exist between the systematic factors listed in Figure 2.8 and the geographical pattern of registration rates at August 2002. This can be done by calculating simple pairwise correlations between the variables of interest and the August 2002 registration rates.

2.42 As was noted in the Introduction above, the NI Measures of Deprivation are not available for the set of Wards used for the Electoral Register. Nonetheless, there are a number of variables on the 2001 Census and also available from the DSD datasets that are highly correlated with the NI Measures of Deprivation. A number of these deprivation-related indicators are listed in Table 2.3. They are grouped according to the domain of the NI Measures of Deprivation to which they most directly relate (see Annex A for a summary of the domains of the NI Measures of Deprivation).

Table 2.3 Correlations between deprivation indicators and August 2002 registration rates Variable Correlation1 Labour market Unemployment rate -0.07 Long-term unemployed (% of population aged 16-74) -0.07 Unemployed or unable to work because sick/disabled (%) -0.05 Receipt of benefits Income Support/JSA (% of population aged 16-74) 0.01 Incapacity benefit (% of population aged 16-74) 0.09 Disability benefit (% of population aged 16-74) 0.09 Qualifications None (% of population aged 16-74) 0.17 Health status Long-term limiting illness (% of population aged 16-74) -0.02 1. Correlations highlighted in bold are statistically significant (p<0.01).

2.43 Thus, the employment domain, which is intended to measure job deprivation, is indicated by a selected number of labour market variables derived from the Census Ward-level dataset, that is, the unemployment rate, the long-term unemployment rate and a measure of employment deprivation. The last of these is most directly related to the indicator in the employment domain of the NI Measures of Deprivation. The other variables map to the NI Measures as follows:

• Income deprivation – receipt of benefits.

• Education – lack of qualifications.

• Health – the incidence of long-term illness.

PricewaterhouseCoopers July 2003 Page 23 Electoral Commission Desk Research: Final Report

2.44 For each of these variables, Table 2.3 shows the simple correlation with the August 2002 registration rate, at Census Ward level. There are two points of interest in the values of the correlations shown in the Table:

• What is the direction of the relationship?

• How strong is the relationship?

2.45 The first of these questions is answered by looking at the sign of the correlation. A negative value for the correlation implies that, the higher the value of the relevant indicator, the lower will be the registration rate, on average. Conversely, a positive correlation implies that, the higher the value of the relevant variable, the higher will be the registration rate. The absolute values of the correlations shown in Table 2.3 can vary between 0 and 1. The closer a correlation is to zero, the weaker the relationship, and vice versa. Though it should be appreciated that such pairwise correlations can be misleading if they exhibit a correlation more because they are themselves correlated with some other variable affecting registration, such as age.

2.46 The main point to emerge from Table 2.3 is that, on a pairwise basis, deprivation was hardly correlated at all with the geographical pattern of the August 2002 Register. Only one of the correlations is statistically significant (qualifications), and this has the opposite sign to what would be expected (it is positive, whereas the expectation is that registration rates are negatively related to deprivation).

2.47 The August 2002 registration rates did, however, display rather more significant relationships when correlated with a range of other demographic and socio-economic variables. A selected set of pairwise correlations is shown in Table 2.4 below. As can be seen, variables such as housing tenure and population age composition are significantly correlated with the August 2002 registration rates at Census Ward level.

2.48 As would be expected from the discussion above regarding the tendency for students to register at their home rather than term address, the geographical pattern in the August 2002 registration rates is negatively correlated with the proportion of the resident population that are full-time students aged over 18. These students tend to be concentrated in Census Wards that contain a university, or that are adjacent to such Wards.

2.49 Similarly, the larger the proportion of persons aged 18-24 or 25-29 that live in a Ward, the lower is the registration rate, on average. This probably reflects a combination of the student factor and the lower propensity of younger people to register to vote. A further point of interest in the age composition correlations is that they indicate an inverted U-shaped relationship, where registration rates tend to rise with age before peaking in the age group 45-59 and then starting to decline again. This suggests that there is not a simple straight-line relationship between age and the propensity to register to vote.

2.50 The tenure pattern in the correlations with registration rates is also interesting, suggesting that Wards with higher than average concentrations of owner-occupied housing tend to have above-average registration rates.

PricewaterhouseCoopers July 2003 Page 24 Electoral Commission Desk Research: Final Report

Table 2.4 Correlations between selected demographic and socio-economic indicators and August 2002 registration rates Variable Correlation1 Students Full-time aged 18+ -0.45 Age composition Aged 18-24 -0.43 Aged 25-29 -0.23 Aged 30-44 0.23 Aged 45-59 0.46 Aged 60-74 0.04 Aged 75-84 -0.11 Aged 85+ -0.18 Housing Tenure Owner-occupation 0.40 Social rented -0.21 Private rented -0.45 Other rented -0.37 Community background Protestant -0.18 Catholic 0.21 Other/none -0.55 Other Detached house 0.45 Average number of persons aged 18 or over per household 0.44 Lone-parent w/dependent children -0.10 Residual other household type -0.32 1. Correlations highlighted in bold are statistically significant (p<0.01).

2.51 By contrast, the higher the share of rented housing, be this in the social sector, the private rented sector or other rented, the lower will be the registration rates. This is interesting also because tenure tends to be correlated with deprivation indicators. For example, there is a strong positive correlation between the proportion of households living in social rented dwellings and indicators such as the local area unemployment rate, the incidence of long-term illness and the proportion with no qualifications (see Table B2.7 in Annex B).

PricewaterhouseCoopers July 2003 Page 25 Electoral Commission Desk Research: Final Report

2.52 Table 2.4 also shows the correlation between the community background composition of the population and registration rates at Census Ward level. The results suggest a negative correlation between registration rates and the proportion of the population that is from the Protestant community, with the opposite relationship holding for the Catholic share at Census Ward level. These relationships are statistically significant, though they are not especially strong.

2.53 Indeed, though there are many significant relationships shown in Table 2.4, as a general rule the correlations are in the range 0.2 to 0.45. That is, no single demographic or socio-economic characteristic stands out as being particularly strongly correlated with the overall pattern of registration rates by Census Ward. This reflects the wide variation that exists in registration rates and the multiplicity of factors influencing their distribution at Census Ward level, some of which were discussed above in relation to the 20 Wards with the lowest registration rates.

2.54 The wide variation in registration rates in turn suggests the use of a multivariate approach in order to take the analysis beyond the simple pairwise approach used in Tables 2.3 and 2.4. The particular technique used for this purpose is known as multiple linear regression. This is a methodology whereby, in principle, the separate and independent effects of a set of ‘explanatory’ variables can be isolated and the significance of their effect on the variable of interest can be measured, though as discussed below this is often difficult to achieve in practice.

2.55 The detailed statistical results are reproduced in Table B2.8, along with a brief technical note on interpretation. This Section first summarises the main findings and then illustrates the implications of the model results using the case study Wards listed in Section 1 above.

2.56 Prior to discussing the results, it is necessary to issue some cautionary remarks. It was noted above that the August 2002 registration rates and various demographic and socio-economic variables were often not very strongly correlated. Indeed, many of the demographic and socio-economic variables that would be considered useful to include in a statistical model are more highly correlated with each other than with registration rates (see Table B2.7). The problem that often arises in such a situation is that it can become difficult to establish the precise nature and scale of the effect of any one individual variable. This poses quite substantial difficulties in selecting variables to be included in the model and deciding which variables to exclude.

2.57 This has been especially problematic in identifying measures of multiple deprivation to include in the model, as the variables listed in Table 2.3 above are all very highly correlated with each other. The approach taken has been to select one variable under each of the following deprivation headings, as follows:

• Unemployment - the long-term unemployment rate.

• Education - the proportion of the population with no qualifications.

• Health - the incidence of long-term limiting illness.

PricewaterhouseCoopers July 2003 Page 26 Electoral Commission Desk Research: Final Report

2.58 An income deprivation variable was not selected because the available indicators are simply too highly correlated with the unemployment variables. They are effectively substitutes for each other in any exercise seeking to explain registration rates. For that reason, it is possible to use either an unemployment variable or an income variable, but not both together.

2.59 As it transpired, each of the three deprivation variables listed above was insignificant in the August 2002 model, either singly or when taken in combination. This is consistent with the findings reported above for the simple pairwise correlations. Nonetheless, the deprivation results are reported in Table B2.8 and further discussed below in relation to the case studies. The reason for this was to ensure that the model estimated for August 2002 could be compared with a similar model estimated for registration rates under the new system, to identify if there had been any change in the effect of deprivation. This is one way of examining whether the change to the system of voter registration may have had a greater (or lesser) effect on areas of deprivation.

2.60 The more significant influences on registration rates were broadly in line with the pattern of correlations shown in Table 2.4 above, including:

• Population age composition. The effect increases with age up to the age group 45-59, and declines thereafter.

• The specific local factors discussed in relation to the 20 Wards with the lowest registration rates all have significant negative effects on local area registration rates. These include the student factor and the presence of an army base, as well as communal care establishments.

• Tenure – the higher the proportion of persons in owner-occupied dwellings, the higher the registration rate. This is likely to at least partly reflect an affluence factor (which is the obverse of a deprivation effect), as this tenure category tends to be negatively correlated with indicators of deprivation.

• Detached dwellings – the proportion of such dwellings in a Ward had a significant positive effect on the registration rate. This may be reflecting primarily a rural and suburban effect.

2.61 A final point to note is that constituency effects were absent. That is, after controlling for the demographic and socio-economic variables listed above, there were no significant differences in registration rates between Wards depending on their location by Parliamentary Constituency.

2.62 The final part of this Section illustrates the model results by undertaking a components of difference analysis for each of the four case study Wards. The objective is to identify how each of the model variables contributes to the observed or actual difference between the August 2002 registration rate for each Ward and the Northern Ireland average registration rate.

PricewaterhouseCoopers July 2003 Page 27 Electoral Commission Desk Research: Final Report

Table 2.5 Components of the difference in registration rates between case study Wards and the Northern Ireland average: August 2002

Whiterock Island Massereene Tempo Registration rate, August 2002 (%) 107.6 83.5 103.1 104.4 Difference from NI1 (pps) Actual 11.0 -13.1 6.5 7.8 Predicted 2.8 -5.1 -0.9 8.5 • Attainers (16-17 year olds) 2.8 -0.8 -0.8 1.1 • Age composition -1.8 0.4 -0.2 0.8 • Median age 2.4 -0.4 0.0 0.0 • Household size & type -1.3 -1.8 -0.1 0.9 • Place of birth & ethnic origin 1.1 -0.3 -1.0 0.9 • Community background 0.9 -0.6 0.0 -0.2 • Detached dwellings -1.6 -1.6 0.1 1.8 • Owner-occupied -2.1 -1.8 0.8 0.6 • Social class -0.1 -0.2 -0.1 0.4 • Deprivation 0.6 0.2 -0.1 -0.0 • Special factors 1.6 3.1 0.9 1.2 • Turnout (Council, 2001) 0.0 -0.1 -0.1 0.1 • Constituency effect 0.4 -1.1 -0.3 0.9 Unexplained residual 8.2 -8.0 7.4 -0.7 1 Northern Ireland rate = 96.6.

2.63 To understand the results, consider the Tempo ward. In August 2002, the registration rate in Tempo was 104.4 per cent. This was 7.8 percentage points above the Northern Ireland average. Given the demographic and socio-economic characteristics of the Tempo ward, such as the age distribution of the local population and the proportion in owner-occupied housing, the predicted difference between Tempo and Northern Ireland based on the statistical model discussed above is +8.5 percentage points. This figure is shown in the row labelled ‘predicted’ in Table 2.5.

2.64 This predicted difference is positive partly because Tempo has a favourable age composition in terms of voter registration. Thus, when Tempo’s age composition is compared to that for Northern Ireland, the statistical model estimates that this adds +1.1 percentage point to the difference between Tempo’s registration rate and the NI average (the demographic and socio-economic data used for these calculations are reproduced in Table B2.9).

PricewaterhouseCoopers July 2003 Page 28 Electoral Commission Desk Research: Final Report

2.65 The main reason for this is that people aged 45-59 have an above-average representation in Tempo (24.9 per cent of people aged 18+, compared to 23.5 per cent for Northern Ireland as a whole). This is the age group which, according to the statistical model, has the highest propensity to register to vote, after controlling for other factors.

2.66 Tempo also has an above-average concentration of households living in detached dwellings (72.4 per cent compared to an NI average of 36.5 per cent). According to the model, this difference between Tempo and Northern Ireland added 1.8 percentage points to its registration rate over and above the average effect of this variable. Similarly, a higher than average proportion of households living in owner- occupied households added 0.6 percentage points to Tempo’s registration rate compared to the Northern Ireland average.

2.67 The ‘special factors’ row summarises the adjustments to the predicted value arising from the fact that Tempo (and the other wards) tend not to contain such specific influences as a large student population or an army base, though both Whiterock and Massereene contain an above-average proportion of their resident populations living in medical and care establishments.

2.68 Adding all of these effects together, the model predicts a +8.5 percentage point difference between Tempo’s August 2002 registration rate and the Northern Ireland average. In other words, the model predicts an August 2002 registration rate for Tempo of 96.6 (the NI average) plus 8.5, that is, 105 per cent. This is not significantly different from the actual Tempo rate of 104 per cent. The percentage points difference between the actual Tempo rate and the predicted rate is given by the unexplained residual row.

2.69 Regarding the community background variable, this is best illustrated by comparing Whiterock and Island, since Tempo has a very balanced religious mix. According to the model, the difference in the August 2002 registration rate between Whiterock and Northern Ireland is increased by one percentage point as a result of the fact that the Ward is almost exclusively Catholic.

2.70 By the same token, for the Island Ward, the model predicts that the difference between its registration rate and Northern Ireland’s was reduced by 0.6 percentage points as a consequence of the fact that Island is 90 per cent Protestant. The reason for this is that the statistical model calculates that, after controlling for other demographic and socio-economic variables, the larger the Protestant share of a Census Ward, the lower was the August 2002 registration rate. Though, the effect is not large6, and is only significant at the 10 per cent level.

6 The calculated effect is -0.023 percentage points per one percentage point increase in the Protestant share.

PricewaterhouseCoopers July 2003 Page 29 Electoral Commission Desk Research: Final Report

2.71 Turning to the differences across the case study wards, the first point of note is the very large contrasts in the raw or unadjusted August 2002 registration rates. The highest registration rate amongst the four cases was in Whiterock. At 108 per cent, this was 11 percentage points above the Northern Ireland average and exceeded the rates found in the more affluent Massereene ward and the rural Tempo ward7. Indeed, though they are both inner-city Belfast wards, with high levels of employment and income deprivation, the Whiterock registration rate in August 2002 was almost 25 percentage points above the Island rate.

2.72 It is possible to ‘explain’ about one-third of this difference with reference to the factors included in the statistical model. The community background factor has already been alluded to. Regarding the influence of demographic factors, while Whiterock has an adverse population age composition from the perspective of voter registration (an overall effect of -1.8 pps), its predicted registration rate is boosted by a high proportion of attainers (people aged 16 to 17 at the time of the 2001 Census8), giving rise to an above-average in-flow to the eligible population in the period since the March 2001 Census.

2.73 The effects of a number of other variables on Whiterock and Island are broadly comparable. Most noticeably, their low proportions of detached and owner-occupied dwellings serve to reduce predicted registration rates, with a combined effect of minus 3-3.5 percentage points in each of the two wards. As was noted in the earlier discussion, geographical differences in these two factors at least partly reflect relative affluence. This is particularly the case with regard to owner-occupied dwellings, the share of which tends to be negatively correlated with a number of deprivation indicators such as unemployment and receipt of benefits. To that extent, the negative effects of this factor on Whiterock and Island may be capturing a deprivation effect, albeit indirectly. The change in this factor will be a point of interest in the analysis of the May 2003 Register in particular.

2.74 The results for Massereene are interesting in that the model predicts that the Ward’s August 2002 registration rate should have been at or very close to the Northern Ireland average. There is no one component of difference that is above one percentage point, even though this is a relatively more affluent area compared to the other case study areas. Indeed, what the model is saying is that it cannot distinguish between Massereene and the Northern Ireland average. This is despite the fact that the Ward had an actual registration rate that was almost 7 percentage points above the NI average, and very close to the Tempo rate. One reason for this is likely to be a missing variable. In Massereene’s case this may be the absence of an indicator for the more rapid growth that appears to have taken place in the Ward in recent times.

7 It should be recalled from the discussion of Parliamentary Constituencies above that areas outside Belfast tend to have higher registration rates due in part to the tendency for students to register at their home addresses.

8 A number of variables were tested to seek to capture the attainer effect, including a variable derived by surviving the March 2001 population in the attainer age range through to the relevant canvass period. In practice, however, the approach that worked best was to use the simple measure based on the population aged 16-17 at the time of the 2001 Census, relative to the population aged 18 and over.

PricewaterhouseCoopers July 2003 Page 30 Electoral Commission Desk Research: Final Report

2.75 There are also relatively large unexplained residuals in the case of Whiterock and Island. Even though the model correctly predicts that Whiterock should have an above-average registration rate, it under-predicts the actual Whiterock rate by 8 percentage points. Similarly, while the model points to a below-average rate in the Island ward, it still over-predicts the Island rate, again by eight percentage points. These are clearly differences that cannot be explained by Ward-level data alone, either from the Census or administrative sources.

Concluding Remarks

2.76 This Section has analysed the August 2002 Electoral Register at three different levels, that is:

• Northern Ireland as a whole.

• Parliamentary Constituency.

• Census Ward.

2.77 The analysis at NI level indicates that, under the old process for voter registration, the number of entries on the Register is likely to have been an over-estimate of the actual number of persons registered to vote. Conversely, the most recent estimate for the resident population is likely to have been a reasonably good predictor of the number of persons entitled to vote in Northern Ireland. This means that the 3.4 per cent non-registration rate indicated by the August 2002 Register is most probably an under-estimate of the extent of non-registration in Northern Ireland at that time. Unfortunately, the sources of Register ‘inflation’ are not quantifiable.

2.78 The analysis by Parliamentary Constituency showed wide variations, even at such a relatively large spatial scale. The raw registration rates varied from 77 per cent in Belfast South to 103 per cent in Fermanagh and South Tyrone and Mid Ulster. Taken at face value, the former suggests a non-registration rate of 23 per cent while the latter indicates more people registered in the constituency than actually reside there.

2.79 But the Belfast South rate of ‘non-registration’ is undoubtedly affected by factors such as the tendency for students living away from the family home to be registered at their place of origin. That is, there is a dislocation between the geographical distribution of the population and the geography of the Register which limits the use of raw registration rates to impute the geographical pattern of non-registration.

2.80 All of these factors are even more pronounced at Ward level, where raw registration rates range from 41 per cent to 126 per cent, with a pronounced clustering in the range 94 per cent to 104 per cent.

2.81 These variations in Ward-level registration rates were analysed using three different methods:

PricewaterhouseCoopers July 2003 Page 31 Electoral Commission Desk Research: Final Report

• Simple pairwise correlations between August 2002 registration rates and various indicators for deprivation as well as demographic and socio- economic variables.

• A more sophisticated analysis that seeks to identify the most significant influences on registration rates, holding all other factors constant.

• A case-study approach based on more detailed analysis of a cross-section of Census Wards.

2.82 Specific factors can be adduced at the lower end of the registration rate spectrum. Wards with high concentrations of the student population tend to have very low rates of registration, for the reason stated above. Similarly, the presence of an army base or a communal establishment also tends to depress registration rates.

2.83 Overall, however, no single factor emerged from the analyses as having the most significant influence on the geographic pattern of registration rates at Census Ward level. One consistent finding that did emerge from each of the analyses undertaken, however, was that indicators of relative deprivation did not have a significant effect on the geography of the August 2002 registration rates.

2.84 Community background was found to have a small, albeit significant effect, such that the larger the Protestant share of a Census Ward, the lower was the August 2002 registration rate.

2.85 Nonetheless, the more important explanatory variables included demographic influences, such as population age composition, and the proportion of households living in owner-occupied and detached dwellings. The latter variables were positively associated with registration rates calculated at Census Ward level.

PricewaterhouseCoopers July 2003 Page 32 Electoral Commission Desk Research: Final Report

III The December 2002 Register

Introduction

3.1 This Section of the report analyses the December 2002 Electoral Register. It commences with an overview on the changes that have occurred in the process by which the Register is compiled. This leads to the conclusion that the requirements of the new system may well have resulted in a Register that more accurately reflects the number of persons eligible to vote by dint of being qualified and having completed and returned the appropriate form. To that extent, it should, in principle, provide a better estimate of the extent of non-registration amongst the Northern Ireland population as a whole.

3.2 The Section then turns to look at spatial variations, at both Parliamentary Constituency and Census Ward levels, in the pattern of registration rates produced by the new system and also changes in the numbers on the December 2002 Register compared to the August 2002 Register. These analyses indicate that the implementation of the new system has produced a geographical pattern in registration rates in the December 2002 Census that is highly correlated with its predecessor. Thus, the December 2002 Register largely preserves the dislocation between the distribution of the population and the geography of the Register that was observed and described in Section 2 above.

3.3 Nonetheless, the pattern of change has not been entirely even. As in the previous Section, the variations across Census Wards are examined with reference to a range of socio-economic and demographic factors.

The New System

3.4 Considered from the perspective of calculating true non-registration rates, the new system introduced on foot of the Electoral Fraud Act (2002) is fundamentally different from its predecessor in at least three important respects:

• Registration is on an individual rather than a household basis.

• Persons registering under the Act must provide more detailed information for identification purposes. The requirements include a personal signature and also the individual’s National Insurance Number.

• The carry-forward system described in Section 2 above no longer applies.

PricewaterhouseCoopers July 2003 Page 33 Electoral Commission Desk Research: Final Report

3.5 These changes are likely, albeit to an unknown degree, to have reduced the extent of ‘inflation’ to which Registers published under the old system had been prone. In particular, the third factor listed above is very likely to have had a discernible impact on the number of entries recorded under the new system. Previously, up to one in ten households did not return their forms at the time of the annual canvass in October-November. These households would be given a year’s grace through the carry-forward system. But, as noted in Section 2 above, this is likely to have inflated the number of entries on the old Registers over and above the number of persons actually entitled to vote because they were both qualified and registered.

3.6 In principle, the more stringent identification requirements may have induced a reduction in the number of irregular entries, though since the extent of registration fraud on the old system was not quantified it is impossible to know what effect this would have had on the new Register compared to its predecessor.

3.7 Nonetheless, given the above, it ought to be the case that the new system should produce a more accurate count of the number of persons who are entitled to vote, that is, they are qualified and have gone through the registration process. The impact of the changes in that regard can be summarised as follows:

The Electoral The Electoral Register in the old Register in the new system system

Fraudelent entries Carried forward & no Reduced? Fraudelent entries longer eligible e.g. moved away Reduced? Multiple entries Multiple entries

Persons Persons eligible to vote eligible to vote

3.8 Though the precise impact of the changes on the Electoral Register cannot be quantified from the evidence to hand, they should have the effect of facilitating a better estimate of the actual or true rate of non-registration, at least at the NI level.

PricewaterhouseCoopers July 2003 Page 34 Electoral Commission Desk Research: Final Report

The NI Level

3.9 As noted in the introduction to this report, the number on the December 2002 Register was ten per cent lower than its predecessor, the August 2002 Register. If the changes described above did in fact result in a more accurate Register, this would suggest an overall NI rate of non-registration in the order of 13-14 per cent, depending on the assumption that is made regarding the increase in the population aged 18 and over (Figure 3.1)9.

Figure 3.1 Electoral Register as per cent of Population 18+

100

98 96

94 92 90

Per cent 88 86 84 82 80 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003

Sources: EONI; NISRA

3.10 It is, however, a moot point as to whether the switch to the new system per se also had the effect of reducing the ‘true’ registration rate underlying the August 2002 Register. Some individuals may have found the requirements of the new system more off-putting. For example, it has been suggested that the request for a National Insurance number may have acted as a disincentive to some individuals for various reasons. Though, there is no direct quantitative evidence to hand on the effect that this may have had.

9 The registration rate for December 2002 in Figure 3.1 is calculated on the basis of a 0.7 per cent increase in the population aged 18 and over in the period since the 2001 Census of Population. This is in line with the recent trend in population growth rates (see Section 2 above).

PricewaterhouseCoopers July 2003 Page 35 Electoral Commission Desk Research: Final Report

3.11 It is also possible that there may have been some reduction in the Register due to some individuals being unaware of the changes or not fully understanding what was required. This type of bedding-in effect from the changeover should, in principle, be resolved over some period of time by the system of Rolling Registration instituted on foot of the new Act. It is not possible for a report of this nature to directly examine people’s reaction to the new process. Clearly, however, it will be interesting to consider the effect of rolling registration on the size of the Register. This is considered in Section 4 below.

Parliamentary Constituencies

3.12 There was considerable variation in the pattern of changes by Parliamentary Constituency, ranging from an 18 per cent reduction in Belfast West to six per cent in Mid Ulster (Figure 3.2. See also Table B2.5). The overall pattern shows a clear divide between the Belfast constituencies and the rest. Each of the Belfast constituencies registered an above-average percentage reduction in the Electoral Register under the new system.

Figure 3.2 Difference between August 2002 Register and December 2002 Register: Per cent

Belfast East Belfast North Belfast South Belfast W est East Antrim East Londonderry Fermanagh & South Tyrone Foyle Lagan Valley Mid Ulster Newry & Armagh North Antrim North Down South Antrim South Down Strangford Upper Bann West Tyrone

N. Ireland

-20 -15 -10 -5 0 5 Per cent

Source: EONI

3.13 This may be due, at least in part, to the pattern of population change described in Section 2 above at District Council level. It will be recalled that the Belfast area has been losing population, so the population base for the December 2002 Register would have been less than that for the August 2002 Register, which was compiled from the 2001 canvass. But the rate of population decline in Belfast has been less than 0.5 per cent per annum, and this factor will not therefore account for all of the difference between Belfast and the rest. Another possible factor is that the implementation of the new system may have been more difficult in Belfast, where there is higher residential mobility and turnover of households in dwelling units, partly due to the above-average share of people living in the rented sector. There is, however, no direct evidence on this factor.

PricewaterhouseCoopers July 2003 Page 36 Electoral Commission Desk Research: Final Report

3.14 Whatever the reason or mix of reasons, the larger percentage changes in the Register in the Belfast area have had the effect of reducing the raw or unadjusted registration rates in the four Belfast constituencies to a greater degree than in other areas. In Belfast South, for example, the December 2002 Register as a percentage of the population aged 18+ fell from 77 per cent in the August 2002 Register to 65 per cent for December 2002, a drop of -12 percentage points compared to the NI average of minus 10 percentage points (Figure 3.3. See also Table B2.4).

Figure 3.3 Electoral Register as per cent of Population aged 18+ December 2002 compared to August 2002

Belfast East Belfast North Belfast South Belfast W est East Antrim East Londonderry Fermanagh & South Tyrone Foyle Lagan Valley Mid Ulster August 2002 Newry & Armagh North Antrim December 2002 North Down South Antrim South Down Strangford Upper Bann West Tyrone

N. Ireland

0 20 40 60 80 100 Per cent

Sources: EONI, 2002; Census, 2001

3.15 But even within Belfast there were sharp differences, as the largest fall was in Belfast West which experienced a reduction of 18 percentage points, falling from a 99 per cent raw registration rate in August 2002 to 81 per cent in December 2002. This clearly demonstrates that population shifts are only one part of the story behind the geographical differences in the change in the Register.

3.16 Regardless of the variations in percentage differences in the change in the Register, it is also very striking that the overall geographical ranking of raw registration rates was largely preserved under the new system. Belfast South remained the constituency with the lowest registration rate, 22 percentage points below the overall NI average of 87 per cent10. Similarly, Fermanagh and South Tyrone (96 per cent) and Mid Ulster (97 per cent) continue to record the highest registration rates. This in turn would suggest that many of the factors underlying geographical variations in raw registration rates in the August 2002 Register continued to operate on the December 2002 Register. This is best examined by looking at the registration rates by Census Ward.

10 This is the NI average measured relative to the population aged 18+ at the 2001 Census of Population, which provides the most up-do-date information for Parliamentary Constituencies and Census Wards. It differs slightly from the 86 per cent NI average shown in Figure 3.1 above. This is because the latter is based on an

PricewaterhouseCoopers July 2003 Page 37 Electoral Commission Desk Research: Final Report

Census Wards

3.17 Similar to the situation for the August 2002 Register, there were large disparities in registration rates at ward level in the December 2002 Register (Figure 3.4). The lowest rate of registration was in the Botanic ward (23 per cent) with the highest being in the Forkhill ward in Newry and Armagh (117 per cent) (see also Table D.1). These are the same Wards for which the lowest and highest rates were recorded on the August 2002 Register. Indeed, the broad pattern of rates shown in Figure 3.4 is quite similar to the August 2002 rates (see Figure 2.7 above), the main difference being that there are far fewer wards with rates in excess of 100 on the December 2002 Register).

Figure 3.4 December 2002 Electoral Register compared to Population 18+ Census Wards

10

9

8

7

6

5

4

3

2 Electoral Register ('000s)

1

0 123456789 Population aged 18+ ('000s)

3.18 Certainly, there was little change in the pattern of registration rates at the lower end of the spectrum. This can be seen by considering the 20 wards with the lowest registration rates. With just three exceptions, these are the same Wards as those that comprised the lower end of the range in the August 2002 Register. By December 2002, the Shaftesbury Ward had entered the bottom 20, though this was not a major change as the Ward had any event been on the fringe of the bottom 20 Wards, with a ranking of twenty-third in August 2002. Another new entrant was the Island Ward, which is one of the case studies on which particular attention is focussed. It is discussed in detail later in this Section. The Wards moving out of the bottom 20 were Roeside, Strule and Magilligan.

estimate of the 2002 mid-year population and incorporates a 0.6 per cent population increase compared to 2001.

PricewaterhouseCoopers July 2003 Page 38 Electoral Commission Desk Research: Final Report

3.19 Overall, therefore, there was remarkably little change in the bottom end of the range. This suggests that those factors associated with lower registration rates in the August 2002 register were again manifest under the new system, that is, students registering at their family home addresses, army bases, communal establishments and so on.

Table 3.1 December 2002 Register as per cent of Population: The 20 Census Wards with the lowest raw registration rates Parliamentary District Ward Rate Constituency %

Belfast South Belfast Botanic 23

Belfast South Belfast Windsor 35

South Antrim Antrim Aldergrove 43

Belfast South Belfast Stranmillis 47

North Down North Down Loughview 49

West Tyrone Omagh Lisanelly 55

Lagan Valley Lisburn Wallace Park 58

Belfast South Belfast Ballynafeigh 58

Foyle Derry Strand 59

Foyle Derry Ebrington 60

East Londonderry Limavady Gresteel 62

East Londonderry Coleraine Strand 63

South Antrim Antrim Springfarm 65

East Antrim Newtownabbey Rostulla 66

South Down Down Killough 66

Belfast South Belfast Shaftesbury 67

Belfast South Belfast Blackstaff 69

East Londonderry Coleraine Portstewart 69

Belfast East Belfast Island 69

East Londonderry Coleraine University 69

PricewaterhouseCoopers July 2003 Page 39 Electoral Commission Desk Research: Final Report

Table 3.2 December 2002 Register as per cent of Population: The 20 Census Wards with the highest raw registration rates Parliamentary District Ward Rate Constituency % Fermanagh & South Fermanagh Donagh 103 Tyrone Mid Ulster Magherafelt Draperstown 103

North Antrim Ballymena Ardeevin 103 Fermanagh & South Dungannon Caledon 103 Tyrone Fermanagh & South Fermanagh Tempo 103 Tyrone West Tyrone Omagh Trillick 104

West Tyrone Omagh Clanabogan 104

West Tyrone Omagh Drumnakilly 104 Fermanagh & South Fermanagh Brookeborough 104 Tyrone West Tyrone Omagh Newtownsaville 104

West Tyrone Omagh Coolnagard 105 Fermanagh & South Dungannon Augher 105 Tyrone Mid Ulster Cookstown Lissan 106

Newry & Armagh Armagh Killeen 106

Mid Ulster Magherafelt Swatragh 108

Mid Ulster Dungannon Altmore 108

Newry & Armagh Newry and Mourne Creggan 109

West Tyrone Omagh Termon 110

Mid Ulster Dungannon Washing Bay 113

Newry & Armagh Newry and Mourne Forkhill 117

PricewaterhouseCoopers July 2003 Page 40 Electoral Commission Desk Research: Final Report

3.20 The situation was not as stable at the upper end of the spectrum, though some patterns did persist. Thus, as shown in Table 3.2, the largest registration rates continued to be found in the more rural District Council areas. In the December 2002 Register, there were eight new entrants to the 20 Wards with the highest rates (Draperstown, Ardeevin, Tempo, Trillick, Clanabogan, Brookeborough, Newtownsaville and Lissan). Those dropping out included Donaghmore, Fathom, Castlecaulfield, East, Rostrevor, Killycolpy, Enagh and Silver Bridge.

3.21 Interestingly, a number of the new Wards in the top 20 (notably Tempo and Brookeborough), as well as Caledon from the August 2002 Register, are finely- balanced between the two communities. In these areas, every vote counts at election time and this may have provided an additional impetus to the registration process. In other Wards, population growth may have been a factor, for example, Coolnagard.

Figure 3.5 The December 2002 Register compared to August 2002 6

5

4

3 Thousands 2 December 2002 Register 1

0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 August 2002 Register

3.22 The broad pattern of change in the number of persons on the Register over the period from August to December 2002 is shown in Figure 3.5. In general, the actual number of persons on the Register by Census Ward in December 2002 was highly positively correlated with the number of the August 2002 Register. Notwithstanding the switch to individual registration, this was to be expected given that the Register as at September 2002, that formed the basis for the October-November canvass from which the December Register was derived. The pattern shown in Figure 3.5 does, however, indicate a tendency for the larger wards, which are mostly located in urban areas, to exhibit a steeper decline from August to December.

3.23 It is useful therefore to examine the 20 Wards that recorded the largest percentage decline in the actual number of person on the Electoral Register in the period from August to December 2002. These Wards are listed in Table 3.3. As can be seen, the Wards showing the largest decline in the Register were indeed located in and around the Belfast area.

PricewaterhouseCoopers July 2003 Page 41 Electoral Commission Desk Research: Final Report

Table 3.3 December 2002 Register as per cent of August 2002 Register: The 20 Census Wards with the largest percentage decline

Parliamentary District Ward % change Constituency

Belfast South Belfast Botanic -45

Belfast West Belfast Falls -24

Belfast West Lisburn Twinbrook -24

Belfast West Belfast Whiterock -24

Belfast West Lisburn Poleglass -23

Belfast West Lisburn Collin Glen -22

Belfast West Belfast Glencolin -22

South Antrim Antrim Springfarm -21

Belfast North Belfast Water Works -21

Belfast North Newtownabbey Coole -21

Belfast West Lisburn Kilwee -20

Belfast North Belfast Ardoyne -20

Belfast North Newtownabbey Dunanney -20

South Antrim Antrim Fountain Hill -20

Belfast North Belfast Bellevue -20

Belfast South Belfast Woodstock -20

North Down North Down Harbour -20

Belfast South Belfast Windsor -20

Belfast West Belfast Clonard -19

Belfast South Belfast Shaftesbury -19

3.24 The largest percentage decline was in Botanic. This supports the conclusion drawn above that the tendency for students and other individuals to register at their home address persisted under the new system. It also hints at the greater difficulties that may have existed in implementing the new system in urban areas where population mobility and dwelling turnover tend to be greater than in rural areas.

PricewaterhouseCoopers July 2003 Page 42 Electoral Commission Desk Research: Final Report

3.25 Considering the remaining Wards with the largest decline, many of these would also be considered to be amongst Northern Ireland’s most deprived areas, for example, Whiterock, Twinbrook, Ardoyne and Falls. Indeed, for the group of 20 Wards shown in Table 3.3, the average percentage of the 18+ population in receipt of Income Support or JSA was 33.5 per cent, almost double the Northern Ireland average of 17 per cent. Similar contrasts can be drawn in terms of other deprivation indicators such as the unemployment rate (an average of 15 per cent, compared to the NI average of 6.9 per cent).

3.26 A further interesting feature of these 20 Wards is that, taken collectively, they have a substantial Catholic majority (an average Catholic share of 63 per cent) and many are in Nationalist areas. In these 20 Wards, the average Nationalist share of the vote in the 2001 local elections was 55 per cent, though this is measured at DEA rather than Ward level.

3.27 Overall, when considering the shifts into and out of the set of Wards with the highest registration rates, and also those Wards where the decline in registration levels have been highest, it is clear that there have been some changes in the geography of registration rates, albeit the December 2002 pattern remained highly correlated with the August 2002 distribution both in levels and registration rates. The simple correlation between the level or size of the August and December Registers was 0.99. The correlation between the registration rates was less than this, at 0.91.

3.28 An interesting question is whether these changes reflect a new set of factors at work or whether there have simply been changes in the mix of factors that shaped the August 2002 geography of registration rates. This Section considers this issue by examining pairwise correlations between the change in the registration rates and selected indicators of deprivation and also demographic and socio-economic characteristics. The estimation of a multivariate statistical model is deferred to Section 4 below. At least some of the variation in the December Register will have reflected any difficulties there may have been in implementation. But the system of Rolling Registration means that, by May 2003, the system will have been better ‘bedded in’.

3.29 In the following analysis, two separate sets of correlations are reported:

• Similar to Section 2 above, the correlation between socio-economic and demographic factors and the registration rate at December 2002, which is compared to its counterpart as of August 2002.

• The correlation between socio-economic and demographic factors and the percentage change from August 2002 to December 2002 in the number of persons on the Register.

PricewaterhouseCoopers July 2003 Page 43 Electoral Commission Desk Research: Final Report

3.30 Turning first to the correlation between deprivation indicators and registration rates, the main point to note is that, whereas for the August 2002 Register there was no correlation at Census Ward level, by December 2002 a number of the deprivation indicators were showing an increase in the size of their correlation with registration rates (Table 3.4). This was particularly marked in the labour market indicators and also receipt of benefit. These correlations between deprivation and registration rates were not especially strong, with all remaining below 0.3. The correlations with the percentage change in the size of the Register were somewhat larerg, reaching 0.45 in the case of the unemployment rate.

Table 3.4 Correlations between deprivation indicators and the August 2002 and December 2002 registration rates Variable August December Change1 20021 20021 Labour market

Unemployment rate -0.07 -0.26 -0.45 Long-term unemployed (% of population aged 16-74) -0.07 -0.24 -0.41 Unemployed or unable to work because sick/disabled (%) -0.05 -0.23 -0.41 Receipt of benefits Income Support/JSA (% of population aged 16-74) 0.01 -0.18 -0.40 Incapacity benefit (% of population aged 16-74) 0.09 -0.07 -0.30 Disability benefit (% of population aged 16-74) 0.09 -0.08 -0.32 Qualifications None (% of population aged 16-74) 0.17 0.09 -0.07 Health status Long-term limiting illness (% of population aged 16-74) -0.02 -0.17 -0.31 1. Correlations highlighted in bold are statistically significant (p<0.01).

3.31 The change in the pattern of correlation between registration and deprivation may, however, have simply been reflecting changes in the relationship between registration and other socio-economic and demographic factors. As can be seen from Table 3.5, the correlations with these variables also showed some change. Most notably, the relationship between registration rates and owner-occupation and social housing was stronger in the December 2002 Register as compared with its predecessor, as was the relationship with the proportion of households living in detached dwellings. These variables are correlated with deprivation, but also with other factors. For example, social sector housing is predominantly urban and the incidence of detached dwellings is higher in more rural areas. Thus, the pattern of change in the correlations may be an artefact of urban-rural differences in the implementation of the new system of registration rather than being specifically related to deprivation. This issue is examined in more detail in Section 4 below when the results of a multivariate analysis are presented.

PricewaterhouseCoopers July 2003 Page 44 Electoral Commission Desk Research: Final Report

Table 3.5 Correlations between selected demographic and socio-economic indicators and voter registration rates: August 2002 to December 2002 Variable August December Change1 20021 20021 Students

Full-time aged 18+ -0.45 -0.42 -0.27 Age composition Aged 18-24 -0.43 -0.45 - 0.35 Aged 25-29 -0.23 -0.22 -0.14 Aged 30-44 0.23 0.19 0.06 Aged 45-59 0.46 0.46 0.28 Aged 60-74 0.04 0.05 0.08 Aged 75-84 -0.11 -0.05 -0.10 Aged 85+ -0.18 -0.11 0.06 Housing Tenure Owner-occupation 0.40 0.54 0.53 Social rented -0.21 -0.40 -0.51 Private rented -0.45 -0.39 -0.21 Other rented -0.37 -0.18 0.22 Community background Protestant -0.18 -0.09 0.12 Catholic 0.21 0.12 -0.09 Other/none -0.55 -0.55 -0.31 Other Detached house 0.45 0.64 0.63 Average number of persons aged 18 or over per 0.44 0.56 0.46 household Lone-parent w/dependent children -0.10 -0.35 -0.59 Residual other household type -0.32 -0.27 -0.16 1. Correlations highlighted in bold are statistically significant (p<0.01).

3.32 Population age composition was found to be a significant factor in registration rates in the multivariate analysis of Section 2 above. For the most part, the correlations with registration rates were broadly stable across the different age categories. There was, however, a significant negative correlation between the proportion of the population aged 18-24 and the percentage change in the registration rates. This may reflect a lower propensity on the part of young people to register individually, or it may simply be a function of their greater mobility, particular in areas such as Botanic.

PricewaterhouseCoopers July 2003 Page 45 Electoral Commission Desk Research: Final Report

3.33 The final point of note from Table 3.5 is the pattern of correlations by community background. Similar to the situation in August 2002, the Protestant and Catholic shares are only weakly correlated with registration, though by December 2002 the correlation with the Protestant share is not significant. Further, the percentage change in the Register was not correlated with the Catholic share, and only weakly related to the Protestant share. The change in the number on the Register was negatively correlated with the proportion saying they were not from a Catholic or Protestant community background or who had no religion, but the correlation with the registration rate remained unchanged. Partly, this is because such persons comprise just three per cent of the total population. From such a small base, changes at the margin (that is, the percentage change in the number on the Register) did not noticeably feed through to the overall average (that is, the registration rate as a proportion of the population aged 18+).

3.34 The foregoing analysis points up some interesting patterns in the change in the Register from August 2002 to December 2002. These are well reflected in the case study Wards. Thus, the two urban Wards, Whiterock and Island, both experienced very large percentage declines in the Register, 24 per cent and 17 per cent respectively (Table 3.6). As a consequence, the Whiterock registration rate fell below the Northern Ireland average

Table 3.6 Changes in the Electoral Register, August to December 2002: The case studies White- Masser- N. Variable Island Tempo rock eene Ireland Population aged 18+ 3,450 3,297 2,564 1,718 2,4751

Electoral Register (number)

August 2002 3,713 2,754 2,643 1,793 2,329

December 2002 2,834 2,282 2,522 1,776 2,067

Percentage change in register

August 2002 to December 2002 -23.7 -17.1 -4.6 -1.0 -10.1

Registration rates

August 2002 107.6 83.5 103.1 104.4 96.6

December 2002 82.1 69.2 98.4 103.4 86.9

1. Average population aged 18+ per Census Ward in Northern Ireland

3.35 The smallest decline was in the more rural Tempo ward, a fall of just one per cent. The size of the disparity between the change in Tempo and the urban Wards should warn against attributing all of the declines in Whiterock and Island to deprivation. This is assessed in Section 4 below.

PricewaterhouseCoopers July 2003 Page 46 Electoral Commission Desk Research: Final Report

3.36 Massereene occupied an intermediate position between these two extremes, though its decline of five per cent was half the Northern Ireland average. This may reflect the fact that it is a growing suburban area. Its household base would therefore be less stable than in Tempo, but it does have a very high incidence of owner- occupation (83 per cent) which is positively correlated with registration.

Concluding Remarks

3.37 This Section has examined the first Electoral Register produced under the new system of individual, rather than household, registration. Compared to the Census 2001 data, at the Northern Ireland level the Electoral Register as a proportion of the population aged 18+ fell from 96 per cent under the old system to 87 per cent under the new system as of December 2002.

3.38 Comparing the two processes, the conclusion is drawn that the new Register is likely to provide a better estimate of the extent of non-registration. At December 2001, this stood at 13-14 per cent, depending on the assumptions drawn regarding population change since the 2001 Census.

3.39 The main reason for this is that the new system is less likely to suffer from ‘list inflation’ due to the abolition of the carry-forward system. Not all of the change will have been due to this factor and it is likely that some of those who were registered on the old system did not register under the new system. It is, however, impossible to say what part of the difference between the August 2002 and December 2002 Registers is due to a reduction in list inflation, and what part is due to an actual decline in the number of persons on the Register.

3.40 The pattern of change in the Register varied both at Parliamentary Constituency and Census Ward level. At Parliamentary Constituency level, the largest percentage declines occurred in the four Belfast Constituencies. Nonetheless, the ranking of Parliamentary Constituencies according to registration rates remained largely unchanged, with the more rural Constituencies continuing to record the highest registration rates. Similarly, at Census Ward level, the broad pattern in registration rates as of December 2002 was highly correlated with the August 2002 pattern. This leads to the conclusion that factors such as the tendency for students living away from the family home to be registered in their place of origin continue to produce a dislocation between the distribution of the population and the geography of the Electoral Register. For example, there was very little change in the composition of the 20 Wards with the lowest registration rates.

3.41 Nonetheless, the change in the Register was not uniformly distributed across the 582 Census Wards. The main point of interest to emerge from the analysis of changes in the number of persons on the Register was that this was correlated with the geography of deprivation. That is, the greater the incidence of deprivation, the larger was the percentage decline in the Register. This may, however, be reflecting influences from other factors, such as an urban-rural contrast in the effect of the switch to individual registration. The next Section considers these issues in greater detail, as some further ‘bedding-in’ will have occurred due to Rolling Registration.

PricewaterhouseCoopers July 2003 Page 47 Electoral Commission Desk Research: Final Report

IV The May 2003 Register

Introduction

4.1 This Section analyses the May 2003 Register, commencing with the change in the size of the Register compared to the situation at December 2002. This change will have reflected the system of Rolling Registration, whereby anyone not on the Register may apply to have their name added or a change recorded, such as a new address or a change of name.

4.2 This enables a calculation of the effect of Rolling Registration on the Northern Ireland registration rate as a percentage of the population aged 18+, giving an estimate of the overall non-registration rate as of May 2003.

4.3 The Section then considers the net change in the Electoral Register at Parliamentary Constituency and Census Ward level. Clearly, changes in the Register at Constituency and especially local level would not be expected to match the overall Northern Ireland average, as there will be geographical variations due to household mobility and also entries to the Register due to people attaining the age of 18 as well as exits due to migration and death.

4.4 The system of Rolling Registration obviously provided an opportunity for the new registration process to ‘bed in’ and hence it is of interest to re-examine the relationships between socio-economic and demographic factors and voter registration rates. Two issues are of particular interest:

• The relationship with deprivation.

• Community background.

Rolling Registration

4.5 The changes in the Register at Northern Ireland level due to Rolling Registration are summarised in Table 4.1 overleaf (see also Table B2.9 in Annex B). Considering the period to April 2003, the total number of additions amounted to 30,512. A large proportion of these (43 per cent) came in the first month of Rolling Registration, perhaps as people realised that they had missed the deadline for the December 2002 Register. A further 29 per cent of the additions came in April 2003, perhaps in anticipation of the May Assembly Elections, since postponed.

4.6 The monthly total of deletions was less erratic, though there was somewhat of a surge in April 2003, which accounted for 35 per cent of the total of 7,961 deletions. Overall, there was a positive net change in the Register to April 2003 of 22,551 persons. A further net change of 3,829 occurred in the subsequent month, giving a total net change from December 2002 to May 2003 of +26,380.

PricewaterhouseCoopers July 2003 Page 48 Electoral Commission Desk Research: Final Report

Table 4.1 Rolling Registration: January 2003 to May 2003 Registra- Variable Additions Deletions Net change tion Rate1 % of Dec. No No No. % 2002 January 2003 13,139 1,617 11,522 1.1 87.9

February 2003 3,238 1,586 1,652 0.2 88.0

March 2003 5,235 1,981 3,254 0.3 88.2

April 2003 8,900 2,777 6,123 0.6 88.7

Totals 30,512 7,961 22,551 2.1

May 2003 n.a. n.a. 3,829 0.4 89.1

December 2002 to May 2003 n.a. n.a. 26,380 2.5

1 Per cent of population aged 18+ from March 2001 Census of Population (NISRA).

Source: EONI

4.7 As a consequence of the Rolling Registration process, the number of people on the Electoral Register increased by 2.5 per cent between December 2002 and May 2003. This resulted in a 2.1 percentage points rise in the Register as a proportion of the population aged 18+, suggesting that the non-registration rate measured on that basis fell from 13 per cent to 11 per cent.

4.8 At Parliamentary Constituency level, the net change in the number of persons on the Register ranged from +1.3 per cent in North Antrim to +5.8 per cent in Belfast West (Figure 4.1). There was therefore a degree of ‘bounce-back’ in the Belfast West constituency, where the largest percentage fall (-18 per cent) had occurred in the period from August 2002 to December 2002, when the new registration system was introduced.

4.9 This did not, however, occur in the other Belfast Constituencies to any noticeable degree. Belfast North and Belfast South showed net changes that were not much above the Northern Ireland average while Belfast East was slightly below the average.

4.10 The second largest increase (+3.8 per cent) was in Foyle. Taken in conjunction with the Belfast West net change, this may suggest that there was somewhat of an urban effect in the change in the number on the Register. This may have been the case to a certain extent, perhaps representing a degree of ‘catch-up’ by areas that had seen the largest fall in the period August to December 2002. This is best examined by looking at the pattern of net changes by Census Ward.

PricewaterhouseCoopers July 2003 Page 49 Electoral Commission Desk Research: Final Report

Figure 4.1 Net change in the Electoral Register Per cent change on December 2002 Register

Belfast East Belfast North Belfast South Belfast West East Antrim Ea st Londonde rry Fe rma na gh & South T yrone Foyle Lagan Valle y Mid Ulster Newry & Armagh North Antrim North Down South Antrim South Down Stra ngford Upper Bann West Tyrone N. Ireland

0123456 Per cent

Source: EONI

4.11 Considering the pattern by Census Ward, it would appear that there was some degree of ‘catch-up’, with the largest percentage net change from December 2002 to May 2003 tending to occur in the Wards that had seen the largest percentage declines from August to December 2002 (Figure 4.2). The relationship is not, however, especially strong. Other factors besides catch-up were clearly at work.

Figure 4.2 Net change from December 2002 to May 2003 compared with the percentage change in the Register August-December 2002

15

10

5

0 Netchange Dec. 02 to 03 May -5 -50 -40 -30 -20 -10 0 Per cent change August-December 2002

Source: EONI.

PricewaterhouseCoopers July 2003 Page 50 Electoral Commission Desk Research: Final Report

4.12 On further examination, it can be seen that, for the vast majority of Census Wards, the net change due to Rolling Registration was concentrated in a fairly narrow band, between zero and five per cent (Figure 4.3). A small number of Wards did better than this, registering net changes of over six per cent, rising to double digits in the case of a handful of Wards.

Figure 4.3 Census Wards ranked by net change in the Electoral Register December 2002 to May 2003 as % of December 2002

15

10

5 Per cent

0

-5 1 51 101 151 201 251 301 351 401 451 501 551 Rank

Source: EONI.

4.13 The 20 Wards that ranked highest in terms of net change from December 2002 to May 2003 are shown in Table 4.2. Interestingly, these Wards bear some resemblance to the 20 Census Wards experiencing the largest percentage declines from August 2002 to December 2002. The overlap comprises Whiterock, Falls, Glencolin, Clonard, Ardoyne and Collin Glen (see Table 3.3 above).

4.14 Apart from the overlap, there are further similarities in terms of deprivation levels. In the 20 Wards shown in Table 4.2, the average proportion of the population aged 18+ in receipt of Income Support or JSA was 32 per cent, compared to a Northern Ireland average of 17 per cent. The average unemployment rate in these Wards was 16 per cent, well above the Northern Ireland average of 6.9 per cent.

4.15 Nonetheless, what is perhaps most striking about these 20 Wards is that they are overwhelmingly Catholic and nationalist. The average Catholic share of the population in these Wards is 92 per cent, while the average nationalist share of the vote in the DEAs in which they are located was 84 per cent at the 2001 Council elections. This would suggest that the net change figures in these 20 Wards are reflecting some extra effort to register the local population rather than any specific effect from deprivation.

PricewaterhouseCoopers July 2003 Page 51 Electoral Commission Desk Research: Final Report

Table 4.2 Rolling Registration: The 20 Census Wards with the largest percentage net change, December 2002 to May 2003

Parliamentary District Ward % change Constituency

Belfast West Belfast Whiterock 14.1

Belfast West Belfast Falls 11.9

South Down Newry and Mourne Burren and Kilbroney 10.3

Lagan Valley Lisburn Derryaghy 9.9

Foyle Derry Creggan Central 9.5

Belfast West Belfast Glencolin 7.7

Newry & Armagh Newry and Mourne Camlough 7.6

Foyle Derry Crevagh 7.6

Belfast West Belfast Upper Springfield 7.0

Foyle Derry Culmore 6.9

West Tyrone Strabane North 6.7

Mid Ulster Magherafelt Bellaghy 6.5

South Down Newry and Mourne Derryleckagh 6.4

Belfast West Belfast Clonard 6.3

Foyle Derry Springtown 6.3

Belfast North Belfast Ardoyne 6.2

Lagan Valley Lisburn Dunmurry 6.2

Foyle Derry Brandywell 6.2

Belfast West Lisburn Collin Glen 6.1

Newry & Armagh Armagh Abbey Park 5.9

PricewaterhouseCoopers July 2003 Page 52 Electoral Commission Desk Research: Final Report

4.16 This conclusion is strengthened by considering the case study Wards. Whiterock was amongst the 20 Wards that recorded the largest percentage net changes due to Rolling Registration. In fact, Whiterock’s net change of +14.1 per cent was the highest in Northern Ireland.

4.17 This represents a marked contrast with the Island Ward, where the net change was less than two per cent (Table 4.3). Compared to the Whiterock Ward, the number of additions was much lower in the Island Ward, whereas the number of deletions was broadly comparable. Thus, the very low registration rate in Island (amongst the bottom 20 Wards in Northern Ireland) has persisted in the period since Rolling Registration was introduced.

4.18 As expected, Massereene more closely tracked the Northern Ireland average, with a slightly higher net change of three per cent compared to the 2.5 per cent average net change.

4.19 The Tempo Ward showed very little change, with a net increase of less than one per cent. But it will be recalled that the registration rate was already very high in Tempo, standing at 103 per cent in December 2002, only marginally below the 104 per cent rate as of August 2002. That is, there was very little scope in Tempo for an additional rise in registration over and above what may be generated by natural increase in the population. Interestingly, in that regard, the net change in Tempo is approximately what would be expected due to natural increase (it will be recalled from Section 2 above that Northern Ireland’s population is expanding at about 0.5- 0.7 per cent per annum).

Table 4.3 Rolling Registration: January 2003 to May 2003 - The case studies White- Masser- N. Variable Island Tempo rock eene Ireland Additions, Jan ’03-April ‘03

Number 333 47 95 20 30,512

Per cent of December 2002 Register 11.8 2.1 3.8 1.1 2.8

Deletions, Jan ’03-April ‘03

Number 22 30 25 11 7,961

Per cent of December 2002 Register 0.8 1.3 1.0 0.6 0.7 Net change, December 2002 to May

2003 Number 401 40 75 11 26,380

Per cent of December 2002 Register 14.1 1.8 3.0 0.6 2.5

Source: EONI

PricewaterhouseCoopers July 2003 Page 53 Electoral Commission Desk Research: Final Report

Correlations

4.20 As noted in the Introduction, the Rolling Registration process provided an opportunity for the new system to ‘bed-in’, for example, by enabling people who had missed the deadline for the December 2002 Register to have their names entered on the Register. For that reason, the May 2003 Electoral Register ought to be more ‘settled’ and free from the influence of such unobservable factors that may have affected the December 2002 Register. Thus, the remainder of this Section re- examines the correlations between registration rates and demographic and socio- economic influences. The approach is similar to that employed for the analysis of the August 2002 Register in Section 2 above.

4.21 The pattern of correlations between various deprivation indicators and registration rates is shown in Table 4.4. There are two main points to note:

• The increase in the size and significance of the correlations for the labour market indicators that occurred between August 2002 and December 2002 persisted through to May 2003.

• Perhaps reflecting the pattern of Rolling Registration, there was some attenuation of the correlations between December 2002 and May 2003. For example, the correlation between registration rates and the unemployment rate declined from -0.26 to -0.21.

Table 4.4 Correlations between deprivation indicators and registration rates Variable August December May 2003 Change, 20021 20021 Aug ’02- May’03 Labour market

Unemployment rate -0.07 -0.26 -0.21 -0.36 Long-term unemployed (% of population -0.07 -0.24 -0.18 -0.32 aged 16-74) Unemployed or unable to work because -0.05 -0.23 -0.17 -0.32 sick/disabled (%) Receipt of benefits Income Support/JSA (% of population aged 0.01 -0.18 -0.12 -0.31 16-74) Incapacity benefit (% of population aged 16- 0.09 -0.07 -0.02 -0.21 74) Disability benefit (% of population aged 16- 0.09 -0.08 -0.02 -0.21 74) Qualifications None (% of population aged 16-74) 0.17 0.09 0.10 -0.06 Health status Long-term limiting illness (% of population -0.02 -0.17 -0.13 -0.27 aged 16-74) 1. Correlations highlighted in bold are statistically significant (p<0.01).

PricewaterhouseCoopers July 2003 Page 54 Electoral Commission Desk Research: Final Report

4.22 The correlations between deprivation and registration rates shown in Table 4.3 were in turn due to the correlation between deprivation and the percentage change from August 2002 to May 2003 in the number of persons on the Electoral Register by Census Ward. But, as was noted in Section 3 above, the correlation with deprivation following the introduction of the new system is not especially strong. This issue is explored further below in the discussion of a multivariate statistical model for the May 2002 registration rates.

4.23 Before considering that issue, it is useful to also look at the trend in the correlation between registration and community background. The main point of note is that the simple correlations between community background and registration rates were almost identical in May 2003 as compared with August 2002 (Table 4.5). This was because, in respect of the Protestant and Catholic communities, there was no simple correlation between community background and the change in the number on the Register from August 202 to May 2003.

Table 4.5 Correlations between community background and voter registration rates Change from August August December 1 2002 to: May 2003 20021 20021 Dec. 2002 May 2003

Protestant -0.18 -0.09 -0.16 0.12 -0.05

Catholic 0.21 0.12 0.19 -0.09 0.07

Other/none -0.55 -0.55 -0.58 -0.31 -0.41 1. Correlations highlighted in bold are statistically significant (p<0.01).

4.24 The simple correlations discussed above do not, however, control for the influence of other factors that affect registration. Thus, as in Section 2, the pattern of registration rates as of May 2003 is explored more fully by estimating a multivariate statistical model. The detailed statistical results are shown in Table B4.1 in Annex B.

4.25 In order to facilitate comparisons with the August 2002 results (see Table B2.8), the same set of socio-economic and demographic variables are used for the May 2003 analysis. The only difference is the use of three additional variables. The first two of these are what may be called ‘transition’ variables. They are:

• A variable to indicate if the Ward was canvassed, in whole or part, by post. This happened in 14 of the 582 wards.

• The percentage difference between the August 2002 Register and the population canvassed for the autumn 2002 canvass, on which the December 2002 Register was based.

PricewaterhouseCoopers July 2003 Page 55 Electoral Commission Desk Research: Final Report

4.26 According to the statistical model, Wards in which a postal canvass was used showed a two percentage point decrease in their registration rate compared to other Wards with similar socio-economic and demographic characteristics.

4.27 The population that was canvassed for the autumn 2002 canvass was based on the electors listed on the September 2002 Register. Each of the persons on that Register was issued with personal registration forms, which were progressively distributed by about 1,000 canvassers. The September Register differs from the Autumn Register, which was the last published Register, by +1 per cent, representing some 12,412 electors. The percentage difference varied considerably by Census Ward, from +8.9 per cent (Derryaghy) to -5.9 per cent in Blackcave. The variance in the Island Ward was -4.9 per cent, putting it only three places above Blackmore in the ranking by Census Ward. Given these variances, and the fact that the September Register was the source of names to be canvassed, it is very important to take account of the canvassed population in a model for registration under the new system. In the statistical model, the effect of the variable is directly proportional, that is, a ±1 percentage point difference in the canvassed population relative to the August 2002 Register leads to a ±1 percentage point increase or decrease in the May 2003 registration rate.

4.28 The third additional variable that was used was based on information from the Valuation and Lands Agency (VLA) showing, for each Census Ward, the number of notified additions of properties to the VLA’s list and also the number of deletions. This provides a proxy indicator for where new households may be locating. Unfortunately, the available data only ran from January to April 2003, giving a partial picture of trends by Census Ward since many Wards had no notifications in that period.

4.29 Regarding the socio-economic and demographic variables, the model results for May 2003 were very similar to those for August 2002. There were, however, some differences of detail, which are discussed below.

4.30 As in the August 2002 model, demographic variables proved to be significant predictors of geographical variations in registration rates. Regarding age composition effects, the main point of difference is that the profile by age group for May 2003 was ‘stretched’ by comparison with the August 2002 results. This is illustrated in Figure 4.4 overleaf, which compares the model effects for each age group. One point of interpretation is that, for technical reasons, the effect for the 18- 24 age group is set to zero, so that what the model produces for each of the other age groups is the effect on registration rates relative to the 18-24 age group.

4.31 Bearing that point in mind, the notable feature of the comparison is that the effect at each age band in May 2003 is stronger than was the case in August 2002. This may have arisen because younger people aged 18-24 were less likely to register using the new system than were people aged 25 and over. If that was the case, areas with a comparatively high concentration of young people would, holding all other factors constant, be predicted to have lower registration rates than the average.

PricewaterhouseCoopers July 2003 Page 56 Electoral Commission Desk Research: Final Report

Figure 4.4 Voter registration and the age composition effect

0.8

0.7

0.6

0.5 August 2002 0.4 May 2003 0.3 Percentage points 0.2

0.1

0.0 18-24 25-29 30-44 45-59 60-74 75-84 85+

Source: PwC

4.32 The second point of note in relation to demographic effects is that the household size effect increased by a significant margin. That is, Wards where the average household size was above the Northern Ireland average tended to register above- average registration rates. This may be reflecting a household type effect. For example, Wards containing relatively high proportions of their population in pensioner or lone-parent families tend to have lower average household sizes compared to the norm for Northern Ireland.

4.33 Some other differences include:

• In the August 2002 model, the proportion of the population born in the Republic of Ireland was positively correlated with registration. This effect was positive but not significant in the May 2003 model.

• The effect of the election turnout variable was higher in the May 2003 model, being significant at 90 per cent. This provides some evidence that electoral competition gives a boost to registration rates.

• Community background was not a significant influence.

• Social class effects remained absent.

• The effect on registration rates due to the proportion of households living in owner-occupied dwellings increased, from 0.05 percentage points per one point increase in the rate of owner-occupation to 0.08 percentage points in the May 2003 model.

PricewaterhouseCoopers July 2003 Page 57 Electoral Commission Desk Research: Final Report

4.34 Regarding the deprivation effects, these remained insignificant, though in contrast to the August 2002 model, each of the variables was found to have a small but negative influence.

4.35 This would tend to confirm the conclusions drawn from the simple correlations discussed above. As was noted in the discussion of Section 2, deprivation effects tend to be difficult to isolate since the predictor variables are also correlated with other variables such as tenure. This point is illustrated in Table 4.6 which shows the results from estimating a model comprising all of the effects discussed above but with an alternative specification for deprivation. The alternative deprivation predictor is a synthetic variable obtained by adding up the values for the three separate deprivation variables used in the basic model discussed above. As can be seen, when owner-occupation is included in the model, the deprivation effect is insignificant and incorrectly signed.

Table 4.6 Deprivation effects and voter registration rates at May 2003: Model estimates with and without owner-occupation With owner- Excluding owner-

occupation occupation

Owner-occupation Coefficient 0.091 -

Significance ***

Deprivation1 Coefficient 0.004 -0.047

Significance n.s. * 1. Calculated as the sum of the long-term unemployment rate, the proportion with a long-term limiting illness and the percentage with no qualifications.

4.36 When owner-occupation is excluded from the model, the deprivation indicator is estimated to have a negative effect, but this is not very strong and is only marginally significant. This suggests that, because it is correlated with deprivation, owner- occupation is ‘masking’ the effects of direct measures of deprivation in the basic model. But this ‘masking’ effect is only partial, as owner-occupation is correlated with other factors besides deprivation, including the rural-urban composition of the population and also age and family type. In short, owner-occupation is a better predictor of registration rates and this is why it comes through in the model results as being highly significant, in contrast to the deprivation indicators.

4.37 The implications of the modelling results at Census Ward level are illustrated with reference to the case study Wards in Table 4.7 overleaf. As in Section 2, the analysis is based on the components of difference between the registration rate in each Ward and the Northern Ireland average. The interpretation of the results is the same as for the comparable Table 2.5 in Section 2 above. The main point to bear in mind is that the analysis is designed to provide an insight into why the registration rate in a Ward may differ from the Northern Ireland average.

PricewaterhouseCoopers July 2003 Page 58 Electoral Commission Desk Research: Final Report

Table 4.7 Components of the difference in registration rates from Northern Ireland: Case study Wards – May 2003

Whiterock Island Massereene Tempo Registration rate, May 2003 (%) 93.8 70.4 101.3 104.0 Difference from NI1 (pps) Actual 4.7 -18.7 12.2 14.9 Predicted -6.4 -13.8 1.9 10.9 • Attainers (16-17 year olds) 2.5 -0.7 -0.7 1.0 • Age composition -2.5 1.5 0.3 1.3 • Median age 2.4 -0.4 0.0 0.0 • Household size & type -1.5 -3.2 -0.1 1.9 • Place of birth & ethnic origin 1.2 -0.1 -0.8 0.6 • Community background 0.8 -0.5 -0.2 0.4 • Detached dwellings -1.8 -1.9 0.1 2.1 • Owner-occupied -2.8 -2.4 1.1 0.7 • Social class 0.2 0.0 -0.2 0.7 • Deprivation -0.9 -0.5 0.3 -0.2 • Special factors -0.1 1.0 -0.1 -0.3 • Turnout (Council, 2001) 0.3 -0.6 -1.0 0.5 • Constituency effect -2.7 -1.2 -1.9 1.9 • Transition -1.4 -4.9 5.2 0.4 Unexplained residual 11.1 -4.9 10.3 4.0 1 Northern Ireland rate = 89.1, when calculated from the 2001 Census of Population results.

4.38 Compared to August 2002, the results for the components of the difference in registration rates are broadly in line with what would be expected from the foregoing discussion of the statistical modelling results. For example, with its younger population age profile, the age composition effects for Whiterock are more negative in May 2003 (-2.5 percentage points) as compared with August 2002 (-1.8 percentage points). Similarly, the owner-occupied effect is more pronounced, at -2.8 percentage points in May 2003 compared with -2.1 percentage points in August 2002. In contrast to August 2002, the deprivation effect for Whiterock is negative in May 2003, but still less than one percentage point. Regarding the community background effect, this is again relatively unimportant in each of the Wards.

4.39 There are, however, two main points of difference. In the August 2002 model, constituency effects were noticeably absent. In the May 2003 models, these are present to varying degrees. In the Whiterock Ward, the constituency effect changed from +0.4 to -2.7 percentage points.

PricewaterhouseCoopers July 2003 Page 59 Electoral Commission Desk Research: Final Report

4.40 These constituency effects are difficult to interpret. One possibility is that they may reflect urban-rural differences in the implementation of the change. As was discussed in Section 3 above, population mobility is higher in urban areas and this could have hampered the registration process in the autumn 2002 canvass. Thus, for example, the constituency effect for Tempo is positive, at +1.9 percentage points, but negative in Whiterock and Island.

4.41 Second, the effect of the transition variables was significant and varied markedly from one Ward to another. This effect is due to the difference between the canvassed population and the published August 2002 Register, as no postal canvass was undertaken in any of the case study Wards.

4.42 In the Island Ward, the variance between the canvassed population and the August 2002 Register had a -5 percentage points effect on the registration rate. In Massereene, the effect was +5.2 percentage points. While it is only possible to speculate on these points, the contrasts may reflect variations in population growth by Census Ward. As was noted in Section 2 above, Massereene is an expanding suburban Ward, and this effect is difficult to capture using Census 2001 data, which refer to a point in time. Island, by contrast, is a declining inner-city ward.

4.43 The final point to note is the pattern of unexplained residuals, that is, the portion of the difference from the Northern Ireland average registration rates that cannot be explained by the factors included in the statistical model. In the case of Whiterock the residual is +11 percentage points. In other words, the actual registration rate in Whiterock, the most deprived of the four case study Wards, was significantly in excess of what would be expected from its socio-economic and demographic composition. A similar result held for the August 2002 model.

4.44 In the Island Ward, the unexplained residual was -5 percentage points. This is a little less than the -8 percentage points residual estimated for August 2002. Whether this May 2003 residual effect is significant or not depends on what is to be made of the transition effect, but this is an unknown quantity.

4.45 In the Massereene Ward, the residual effect is +10.3 percentage points, about the same as in the August model. This residual arises because the model predicts that Massereene’s registration rate should be in line with the Northern Ireland average, but it is 12.2 percentage points above the average. Again, this is likely to be due to factors that are not included in the model. One such factor may be faster than average population growth, though as was noted above the Rolling Registration rate for Massereene was only slightly above the Northern Ireland average.

4.46 Finally, the residual for Tempo increased slightly, from -0.7 in August 2002 to four in the May 2003 model. This is not statistically significant and it remains the case that the model performs reasonably well in ‘explaining’ why Tempo has a registration rate that is well above the Northern Ireland average, a rural area with broadly favourable demographic characteristics.

PricewaterhouseCoopers July 2003 Page 60 Electoral Commission Desk Research: Final Report

Concluding Remarks

4.47 This Section has analysed the outcome from Rolling Registration in the period up to May 2003. The main findings are as follows.

4.48 The Northern Ireland registration rate increased by 2.1 percentage points, suggesting a non-registration rate of about 11 per cent.

4.49 Much of the change occurred in the first month of Rolling Registration, accounting for about 43 per cent of the net change to May 2003. There were some variations across Parliamentary Constituencies, with Belfast West recording a 5.8 per cent rise in the number of persons on the Electoral Register. This suggests some degree of ‘bounce-back’ in areas with the larger falls in the Register from August to December 2002.

4.50 At Census Ward level, the vast majority of Wards fell into a relatively narrow range, from zero to +5 per cent. A relatively small number of Wards recorded larger increases, ranging up to +14 per cent in Whiterock. The 20 Wards with the highest percentage increases due to Rolling Registration shared some common characteristics, including high levels of deprivation. Most noticeably, however, they were predominantly Catholic and nationalist, suggesting that some local variations reflect registration efforts being made ‘on the ground’.

4.51 The Section also examined the pattern of registration at May 2003 and compared this with the August 2002 pattern. This was done using statistical correlations and also with reference to the case studies. Though there were some changes of detail, the main finding is one of broad stability in the socio-economic and demographic factors that are correlated with registration rates.

4.52 Regarding community background, this was found not to be significant after controlling for influences such as population age composition and so on.

4.53 Various indicators for deprivation were found to be correlated with the percentage change between August 2002 and May 2003 in the number of persons on the Register. However, the correlation was not strong enough to have a significant effect on registration rates calculated relative to the population aged 18+ by Census Ward, when other factors are taken into account.

PricewaterhouseCoopers July 2003 Page 61 Electoral Commission Desk Research: Final Report

V Conclusions

Introduction

5.1 This final Section presents the conclusions from the various analyses that have been undertaken with reference to the following topics:

• The current extent of non-registration in Northern Ireland.

• The patterns of non-registration in Northern Ireland.

• Any correlation between decreased registration rates and indicators of deprivation.

Extent of Non-registration

5.2 Expressed as a percentage of the population aged 18 and over, the Northern Ireland registration rate stood at about 88 per cent in May 2003, giving a non-registration rate of approximately 12 per cent (Figure 5.1). Taken at face value, this would appear to represent a sharp decline compared to August 2002, just prior to the introduction of the new system of individual registration. Then, the Register as a percentage of the voting-age population was almost 96 per cent.

Figure 5.1 Electoral Register as per cent of Population 18+

100

98

96

94

92 Per cent 90

88

86

84 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004

Sources: EONI; NISRA.

5.3 It is, however, impossible to tell what part of the change was due to an actual increase in non-registration following the new system, and what part was due to the elimination of various sources of ‘inflation’ in the Electoral Register under the old system of household registration.

PricewaterhouseCoopers July 2003 Page 62 Electoral Commission Desk Research: Final Report

5.4 To the extent that the change in the system of registration may have resulted in an increase in non-registration, at least some of that has been ‘clawed-back’ through the system of Rolling Registration whereby people who are not on the Register can apply to have their names added.

5.5 As a result of Rolling Registration, the registration rate under the new system increased by 2.5 percentage points in the period from December 2002 to May 2003. This is in excess of the rate of natural increase in the Northern Ireland population aged 18+, currently about 0.5-0.7 per cent per annum.

5.6 Whether the registration rate will continue to increase at that pace is a moot issue, as much of the change occurred in the first month after the introduction of the new system and in the month leading up to the postponed May Assembly Elections. In other months, the Register has been growing at a rate of about +0.4 percentage points.

The Pattern of Non-registration

5.7 A key point of interest in establishing the pattern of non-registration at Census Ward level is that this would provide a basis for assessing the influence of various factors that might be correlated with non-registration, such as deprivation. It is not, however, possible to calculate non-registration rates at Census Ward level.

5.8 The reason for this is that there is a dislocation between the geography of the Register and the distribution of the population. For example, students aged 18 and over tend to be registered at their home address rather than their term address. But they are recorded on the Census of Population as being resident at their term address. This effect was apparent under both the old system and the new system.

5.9 Reflecting this dislocation, the ‘registration rate’ at Census Ward level in May 2003 varied from a low of 24.3 per cent in the Botanic Ward to 122 per cent in Forkhill.

5.10 A range of factors affect the pattern of registration at Census Ward level. Consistent with the dislocation factor noted above, registration rates measured relative to the resident population tend to be higher outside the urban areas. Partly reflecting this, wards with higher concentrations of owner-occupier households and households living in detached dwellings tend to have above-average rates of registration. Similarly, population age composition affects the pattern in registration rates. The effect of these factors on registration rates did not change greatly between August 2002 and May 2003.

5.11 Overall, however, there is no one single factor that emerges as the major determinant of the geographical pattern in registration rates. Indeed, the case study evidence in this report indicates that a complex set of factors affect registration rates. Not all of these are captured by the demographic and socio-economic data on which this report is based.

PricewaterhouseCoopers July 2003 Page 63 Electoral Commission Desk Research: Final Report

Deprivation

5.12 The Northern Ireland Measures of Deprivation are presently only available on the 1984 Census Ward geography, but the 2001 Census of Population is published using the 1992 Census Ward boundaries. For that reason, this report has assessed the relationship between deprivation and registration rates using various indicators that are highly correlated with deprivation in the domains of employment, income, education and health.

5.13 There are two main findings. First, the percentage change in the Register in the period from August 2002 to May 2003 is negatively correlated with a range of deprivation indicators. This would suggest that areas with an above-average incidence of deprivation experienced larger than average percentage falls in the Electoral Register that was compiled under the new system compared to the Register that existed under the old system.

5.14 The second main finding is that, when other relevant influences such as population age composition are taken account of, indicators of deprivation were not correlated with the registration rate either in August 2002 or in May 2003. In other words, deprivation is not a good predictor of the rate of registration by Census Ward. This conclusion is consistent with the more detailed case studies undertaken for this report. In two relatively deprived areas in Belfast, the May 2003 registration rate ranged from a below-average 70 per cent to an above-average 94 per cent.

5.15 These two findings can be reconciled because the first is an effect at the margin while the latter refers to average effects. Thus, the registration rate is an average measure, while the percentage change measures effects at the margin. Whether an effect at the margin feeds through to the average depends both on the magnitude of the effect at the margin and the weighting of that factor in the overall average.

5.16 The conclusion to be drawn is that, while it bore some relation to the changes seen in the period from August 2002 to May 2003, deprivation was not sufficiently significant as a predictor of voter registration as to exert a strong and separate influence on the geography of registration rates by Census Ward.

PricewaterhouseCoopers July 2003 Page 64

ANNEXES:

Annex A The N.I. Measures of Multiple Deprivation

Annex B Statistical Tables

Annex C Maps

Annex D Statistical Tables – Census Ward

Electoral Commission Desk Resesrch: First Report

Annex A The NI Measures of Multiple Deprivation

The NI Measures of Deprivation were published in July 2001 to contribute to the Government’s New Targeting Social Need (TSN) initiative. They were developed by the Department of Social Policy and Social Work at Oxford University, led by Professor Noble.

The NI Measures of Deprivation are spatially-based; they are designed to contribute to New TSN by attempting to identify areas in greatest need. This was done by constructing various measures of deprivation at three different geographical levels:

• 1984 Census Wards – there are 566 of these in Northern Ireland. The average number of people per ward is about 3,000, but they range quite widely in size from under 1,000 to over 11,000.

• Enumeration Districts used for the 1991 Census – there are 3,729 of these, containing 200 people on average.

• District Councils – there are 26 of these, ranging in size from almost 300,000 in Belfast to 15,300 in Moyle.

• District Councils are each comprised of a number of different wards which are in turn comprised of a number of enumeration districts.

Essentially, then, the NI Measures of Deprivation provide a means of mapping differences in deprivation between various geographical areas. As the name implies, the concept of objective need that underpins the NI Measures is that of multiple deprivation, or unmet need due to a lack of resources. In order to capture different aspects of deprivation, the NI Measures are built around a set of seven ‘domains of deprivation’. These are listed in Figure 1.

Figure 1 Domains of the NI Measures of Deprivation

Domains

Geographical Employ- Social Income Health Education access to Housing ment services environment

Multiple Deprivation

PricewaterhouseCoopers July 2003 Annexes - Page 1 Electoral Commission Desk Research: First Report

Figure 2 The NI Measures of Deprivation: Wards

Domains

Geographical Employ- Social Income Health Education access to Housing ment services environment

Indicators

Lack of Post office SMR 75 qualifications GP surgery Receipt of All living in Destinations disability A&E hospital Crime - Housing in families in Unemployed of school related Dentist various disrepair or in receipt leavers receipt of benefit Optician recorded W/out of Incapacity GCSE/GNVQ a means Cancer Pharmacist offences central Benefit (% of performance heating Child dental Library Local Area tested Absenteeism 16-59 year health Museum Problem Lacking benefit 11-12 year Score insulation olds) Prescriptions Social Security (%) olds not in for anxiety or grammar or T&EA depression school office

Weights: 25% 25% 15% 15% 10% 5% 5%

Multiple Deprivation measure

Measurement of deprivation

For each 1984 Census Ward, deprivation in the domains shown in Figure 2 is measured by combining information on various indicators that are felt to reflect some major aspect of the relevant domain. So, for example, the extent of income deprivation within a ward is measured by calculating the proportion of the ward’s population in receipt of a means-tested benefit. Similarly, the employment domain is intended to measure exclusion from the world of work. It is measured by using information on indicators such as the numbers of people unemployed and those aged under-60 receiving Incapacity Benefit.

Within each domain, the various indicators are combined to give a single measure for that domain. In some cases, the component indicators are simply added up (e.g. in the income and employment domains). In other instances, more sophisticated statistical methods are required since the components use different measurement scales (the health, education and social environment domains).

These domain deprivation measures, seven in all, are produced for each ward. The next step is to produce an overall summary measure of multiple deprivation for each ward. However, because the domain measures are calculated on different scales, it is not possible to derive such a summary measure for each ward simply by adding up the domain measure scores for that ward.

PricewaterhouseCoopers July 2003 Annexes - Page 2 Electoral Commission Desk Research: First Report

To get around this, the Noble team devised the following procedure. First, the individual domain deprivation measures are ranked by ward, from most deprived to least deprived. Second, the ward rankings for each domain measure were transformed to a common distribution. These transformed ward rankings for each domain were then added together, using the weights shown in Figure 2, to get an overall multiple deprivation measure score. According to the Noble report, the criteria for selecting a set of weights for combining the standardized domain scores within each ward were:

• The importance of their contribution to an overall concept of multiple deprivation.

• Robustness of the indicators comprising the domain.

Finally, the resulting summary scores are used to rank wards from most deprived (Crumlin in Belfast) to least deprived (Wallace Park in Lisburn).

Accompanying the above, a separate ward-based Child Poverty measure has been produced. For each ward, this is the percentage of the local population aged 0-15 that lives in families in receipt of means-tested benefits. This indicator is one of the components of the income deprivation domain measure, so it does not feature in the construction of the overall summary measure of multiple deprivation.

PricewaterhouseCoopers July 2003 Annexes - Page 3 Electoral Commission Desk Research: First Report

Annex B Statistical Tables

PricewaterhouseCoopers July 2003 Annexes - Page 4 Electoral Commission Desk Research: First Report

Table B2.1 Population aged 18+ in Local Government Districts: Per cent change

District Council 1991-2001 1997-2001 %%

Antrim 10.4 2.3 Ards 16.9 6.2 Armagh 9.0 3.3 Ballymena 7.5 2.7 Ballymoney 15.6 7.6 Banbridge 27.8 8.4 Belfast -4.1 -1.6 Carrickfergus 16.4 4.8 Castlereagh 5.9 0.2 Coleraine 9.5 2.5 Cookstown 11.5 3.7 Craigavon 10.1 5.0 Derry 15.7 4.2 Down 13.8 5.0 Dungannon 11.6 4.3 Fermanagh 11.4 4.6 Larne 6.1 1.6 Limavady 18.4 4.6 Lisburn 10.8 1.0 Magherafelt 16.3 6.7 Moyle 13.3 4.3 Newry and Armagh 9.9 5.0 Newtownabbey 8.3 2.0 North Down 8.4 2.6 Omagh 11.4 3.9 Strabane 12.5 4.9

NI Average 8.7 2.8

Sources: NISRA, Mid-year Population Estimates

PricewaterhouseCoopers July 2003 Annexes - Page 5 Electoral Commission Desk Research: First Report

Table B2.2 Electoral Register by Parliamentary constituency: 1997-2002

Parliamentary Constituency Westminister Assembly Westminister August 1997 1998 2001 2002

Number Number Number Number

Belfast East 61,745 60,562 58,455 57,592

Belfast North 64,577 62,541 60,941 59,199

Belfast South 63,890 61,209 59,436 58,778

Belfast West 61,781 60,669 59,617 58,588

East Antrim 58,982 59,313 60,897 60,851

East Londonderry 58,827 59,370 60,215 60,613

Fermanagh and South Tyrone 64,602 65,383 66,640 67,499

Foyle 67,621 68,888 70,943 70,911

Lagan Valley 71,226 71,661 72,671 72,611

Mid Ulster 58,835 59,991 61,390 62,219

Newry and Armagh 70,657 71,553 72,466 73,215

North Antrim 72,415 73,247 74,451 75,211

North Down 63,006 62,942 63,212 63,285

South Antrim 69,409 69,426 70,651 70,597

South Down 69,856 71,027 73,519 74,516

Strangford 69,983 70,868 72,192 72,354

Upper Bann 70,399 70,852 72,574 73,210

West Tyrone 58,171 59,081 60,739 60,887

Northern Ireland 1,175,983 1,178,583 1,191,009 1,192,136

Sources: EONI, 2002.

PricewaterhouseCoopers July 2003 Annexes - Page 6 Electoral Commission Desk Research: First Report

Table B2.3 Electoral Register by Parliamentary constituency: 1997-2002 – Per cent of Westminister 1997 Register

Parliamentary Constituency Westminister Assembly Westminister August 1997 1998 2001 2002

%% % %

Belfast East 100 98 95 93

Belfast North 100 97 94 92

Belfast South 100 96 93 92

Belfast West 100 98 96 95

East Antrim 100 101 103 103

East Londonderry 100 101 102 103

Fermanagh and South Tyrone 100 101 103 104

Foyle 100 102 105 105

Lagan Valley 100 101 102 102

Mid Ulster 100 102 104 106

Newry and Armagh 100 101 103 104

North Antrim 100 101 103 104

North Down 100 100 100 100

South Antrim 100 100 102 102

South Down 100 102 105 107

Strangford 100 101 103 103

Upper Bann 100 101 103 104

West Tyrone 100 102 104 105

Northern Ireland 100 100 101 101

Sources: EONI, 2002.

PricewaterhouseCoopers July 2003 Annexes - Page 7 Electoral Commission Desk Research: First Report

Table B2.4 Electoral Register by Parliamentary constituency: August 2002, December 2002 and May 2003 - Per cent of population aged 18+

August 2002 December May 2003 Difference 2002 compared to NI % % % pps

Belfast East 94 83 85 -4.4

Belfast North 94 79 81 -7.7

Belfast South 77 65 67 -22.5

Belfast West 99 81 86 -3.1

East Antrim 97 87 88 -1.0

East Londonderry 93 85 86 -2.8

Fermanagh and South Tyrone 103 96 98 9.0

Foyle 97 86 89 0.4

Lagan Valley 96 87 90 0.7

Mid Ulster 103 97 99 10.3

Newry and Armagh 102 94 96 7.2

North Antrim 100 93 94 4.8

North Down 95 85 87 -2.5

South Antrim 95 84 86 -3.0

South Down 100 92 95 5.5

Strangford 98 89 90 0.9

Upper Bann 98 91 93 3.6

West Tyrone 100 93 95 6.3

Northern Ireland 97 87 89 pps Percentage points

Sources: EONI, 2002; Census 2001.

PricewaterhouseCoopers July 2003 Annexes - Page 8 Electoral Commission Desk Research: First Report

Table B2.5 Electoral Register by Parliamentary constituency: December 2002 compared to August 2002

August 2002 December 2002 Difference

No No %

Belfast East 57,592 50,914 -11.6

Belfast North 59,199 49,899 -15.7

Belfast South 58,778 49,303 -16.1

Belfast West 58,588 48,131 -17.8

East Antrim 60,851 54,577 -10.3

East Londonderry 60,613 55,475 -8.5

Fermanagh and South Tyrone 67,499 63,013 -6.6

Foyle 70,911 62,817 -11.4

Lagan Valley 72,611 65,939 -9.2

Mid Ulster 62,219 58,808 -5.5

Newry and Armagh 73,215 66,802 -8.8

North Antrim 75,211 69,728 -7.3

North Down 63,285 56,610 -10.5

South Antrim 70,597 62,343 -11.7

South Down 74,516 68,317 -8.3

Strangford 72,354 65,562 -9.4

Upper Bann 73,210 67,596 -7.7

West Tyrone 60,887 56,512 -7.2

Northern Ireland 1,192,136 1,072,346 -10.0

Sources: EONI, 2002; Census 2001.

PricewaterhouseCoopers July 2003 Annexes - Page 9 Electoral Commission Desk Research: First Report

Table B2.6 Electoral Register by Parliamentary constituency: May 2003 compared to December 2002

December 2002 May 2003 Difference

No No %

Belfast East 57,592 51,983 2.1

Belfast North 59,199 51,476 3.2

Belfast South 58,778 50,804 3.0

Belfast West 58,588 50,909 5.8

East Antrim 60,851 55,529 1.7

East Londonderry 60,613 56,238 1.4

Fermanagh and South Tyrone 67,499 64,405 2.2

Foyle 70,911 65,213 3.8

Lagan Valley 72,611 68,041 3.2

Mid Ulster 62,219 60,155 2.3

Newry and Armagh 73,215 68,786 3.0

North Antrim 75,211 70,628 1.3

North Down 63,285 57,536 1.6

South Antrim 70,597 63,654 2.1

South Down 74,516 70,230 2.8

Strangford 72,354 66,373 1.2

Upper Bann 73,210 68,883 1.9

West Tyrone 60,887 57,883 2.4

Northern Ireland 1,192,136 1,098,726 2.5

Sources: EONI, 2002; Census 2001.

PricewaterhouseCoopers July 2003 Annexes - Page 10 Electoral Commission Desk Research: First Report

Table B2.7 Selected correlations between socio-economic variables Variable Unemp. L-T No Social Detach- rate limiting quals. rented ed illness D. Unemployment Unemployment rate - 0.78 0.63 0.76 -0.50 Long-term unemployed 0.95 0.70 0.59 0.69 -0.44 Unemployed or unable to work 0.93 0.88 0.72 0.78 -0.52 because sick/disabled E. Receipt of benefits Income Support/JSA 0.93 0.86 0.72 0.77 -0.49 Incapacity benefit 0.80 0.82 0.72 0.69 -0.42 A disability benefit 0.81 0.86 0.70 0.68 -0.42 F. Qualifications None 0.63 0.77 - 0.67 -0.29 F. Health status Long-term limiting illness 0.78 - 0.77 0.77 -0.55 G. Housing Tenure Owner-occupation -0.78 -0.75 -0.62 -0.94 0.72 Social rented 0.76 0.77 0.67 - -0.72 Private rented 0.18 0.09 -0.08 -0.04 -0.19 Other rented -0.18 -0.20 -0.13 -0.26 0.36 H. Students Full-time aged 18+ 0.02 -0.23 -0.38 -0.11 -0.02 I. Age composition Aged 18-24 0.36 0.07 0.03 0.21 -0.15 Aged 25-29 0.21 0.06 0.17 0.18 -0.21 Aged 30-44 -0.14 -0.29 -0.19 -0.15 0.14 Aged 45-59 -0.31 -0.29 -0.19 -0.32 0.44 Aged 60-74 0.00 0.31 0.25 0.13 -0.17 Aged 75-84 -0.06 0.27 0.10 0.05 -0.15 Aged 85+ -0.10 0.16 -0.11 -0.06 -0.07 J. Community background Protestant 0.47 -0.29 0.16 0.08 0.04 Catholic -0.47 0.30 -0.13 -0.08 -0.03 Other/none -0.30 -0.31 -0.47 -0.02 -0.30 K. Other correlates Detached house -0.50 -0.55 -0.29 -0.72 - Persons aged 18+ per household -0.32 -0.45 -0.18 -0.63 0.79 Lone-parent w/dependent children 0.78 0.66 0.50 0.83 -0.65

PricewaterhouseCoopers July 2003 Annexes - Page 11 Electoral Commission Desk Research: First Report

Table B2.8 Accompanying Notes

Multiple linear regression is a methodology whereby, in principle, the separate and independent effects of a set of ‘explanatory’ variables can be isolated and the significance of their effect on the variable of interest can be measured, though for the reasons discussed in the main body of the report this is often difficult to achieve in practice.

In the absence of previous research of a similar nature to this study, the selection of variables for inclusion in the model was based on testing a range of alternative specifications. Initially, over 70 variables were tested in assessing the combination of variables that gave the best results. The final specification is shown in Table B2.8 below.

The interpretation of the predictor variables is straightforward. For example, the population aged 45 to 59, entered as a proportion of the population aged 18 or over in each Ward, has a positive coefficient. This tells us that, the higher the proportion of people aged 45 to 59 in a Ward, the higher will be the registration rate for that Ward relative to the average for all Wards. The result does not say if people aged 45-59 are more likely than the excluded age groups to register to vote. It should be recalled that a model based on Census Wards simply tells us what characteristics of the Wards are more or less correlated with the variable of interest. Note also that the model does not include the population aged 18-24. For technical reasons, at least one age group group must be assigned a coefficient value of zero, against which the coefficients for the included age groups can be compared.

The significance of the coefficient can be judged from the t-ratio, a statistic that enables us to say if the coefficient is more or less likely to be different from zero. The level of significance of each predictor variable is indicated in the Table. Note that a number of insignificant variables were retained in the model. This is both to illustrate some points of particular interest (e.g. the deprivation indicators, the absence of a social class gradient and so on) and also to facilitate comparisons with a similar model estimated for May 2003 (see Table B4.1 below).

A final point to note is that the model includes Constituency-specific effect. Such location- specific effects do not necessarily tell us anything about the underlying processes generating variations in registration rates between Wards. They are included to assess the extent to which registration rates may vary by Constituency over and above what is predicted from the socio-economic and demographic variables.

PricewaterhouseCoopers July 2003 Annexes - Page 12 Electoral Commission Desk Research: First Report

Table B2.8 Regression model results – August 2002 Register as per cent of population aged 18+, by Census Ward Standard Coefficient t-statistic Significance error Attainers (% of population 18+) • Aged 16-17 0.903 0.266 3.395 *** Age composition (% of population 18+) • Aged 18-24 0.000 - - - • Aged 25-29 0.097 0.228 0.425 • Aged 30-44 0.223 0.152 1.463 • Aged 45-59 0.456 0.204 2.240 ** • Aged 60-74 0.516 0.191 2.708 *** • Aged 75-84 0.401 0.242 1.653 * • Aged 85+ 0.275 0.343 0.801 Median age -0.408 0.157 -2.595 *** Household size - persons 18+ 5.370 3.021 1.777 * Where born (% of population) • N. Ireland 0.000 - - - • Great Britain -0.404 0.162 -2.488 ** • Republic of Ireland 0.306 0.127 2.413 ** • Other EU 0.051 0.599 0.085 • Elsewhere -0.027 0.550 -0.050 Ethnic background (% of population) • White 0.000 - - - • Non-white -0.047 0.515 -0.091 • Irish Traveller -0.805 0.612 -1.316 Community background (% of population) • Catholic 0.000 - - - • Protestant -0.023 0.013 -1.802 * • Other/none 0.110 0.227 0.486 Communal establishments (% of population) • Medical and care -0.392 0.131 -3.001 *** • Other -0.504 0.180 -2.798 *** Army base in Ward -5.439 1.938 -2.807 *** Prison in Ward -6.338 3.466 -1.829 * Second/holiday homes (% of dwellings) -0.198 0.079 -2.517 ** Full-time students (% of population 18+) -0.722 0.161 -4.485 *** Other rented (% of dwellings) -0.315 0.138 -2.274 ** Residual household type (% of households) -0.669 0.171 -3.909 ***

PricewaterhouseCoopers July 2003 Annexes - Page 13 Electoral Commission Desk Research: First Report

Table B2.8 Regression model results – August 2002 Register as per cent of population aged 18+, by Census Ward Standard Coefficient t-statistic Significance error Local election turnout, 2001 (%) 0.013 0.055 0.233 Detached dwellings (%) 0.050 0.019 2.594 *** Owner-occupied (%) 0.064 0.036 1.765 * Social Class • Managerial and professional -0.016 0.094 -0.174 • Intermediate occupations (incl. small 0.039 0.091 0.428 employers) • Lower supervisory and technical -0.014 0.164 -0.084 occupations • Routine manual 0.000 - - Deprivation indicators (%) • No qualifications -0.008 0.092 -0.087 • Long-term unemployed 0.063 0.324 0.195 • Long-term limiting illness 0.024 0.079 0.301 Constituency effects • Belfast East -0.985 1.512 -0.652 • Belfast North -1.937 1.354 -1.430 • Belfast South -1.833 1.552 -1.181 • Belfast West 0.516 1.509 0.342 • East Antrim -0.221 1.441 -0.154 • East Londonderry -0.735 1.146 -0.641 • Fermanagh & South Tyrone 0.954 0.945 1.009 • Foyle 0.474 1.166 0.407 • Lagan Valley 0.288 1.308 0.220 • Mid Ulster 1.065 0.946 1.125 • Newry & Armagh 1.286 0.916 1.404 • North Antrim 0.597 1.166 0.512 • North Down 0.538 1.549 0.348 • South Antrim -0.216 1.331 -0.163 • South Down 0.100 1.088 0.092 • Strangford -0.330 1.413 -0.234 • Upper Bann -0.194 1.057 -0.183 • West Tyrone 0 - - - Constant 68.062 17.371 3.918 0.000

Number of observations 582 Adjusted R-square 0.812 Standard error of the estimate 3.652

PricewaterhouseCoopers July 2003 Annexes - Page 14 Electoral Commission Desk Research: First Report

Table B2.8 Regression model results – August 2002 Register as per cent of population aged 18+, by Census Ward Standard Coefficient t-statistic Significance error F-ratio 50.356 Significance of F 0.000 *** Significant at 99 per cent. ** Significant at 95 per cent. * Significant at 90 per cent.

PricewaterhouseCoopers July 2003 Annexes - Page 15 Electoral Commission Desk Research: First Report

Table B2.9 Summary statistical profile: Case study Wards Variable White- Island Mass- Tempo N. Ireland rock ereene Population aged 18+ 3,450 3,297 2,564 1,718 2,4751 A. Registration Electoral Register (number) August 2002 3,713 2,754 2,643 1,793 1,192,136 December 2002 2,834 2,282 2,522 1,776 1,072,346 May 2003 3,235 2,322 2,597 1,787 1,098,726 Registration rates August 2002 107.6 83.5 103.1 104.4 96.6 December 2002 82.1 69.2 98.4 103.4 86.9 May 2003 93.8 70.4 101.3 104.0 89.1 Change in register from August 2002 to: December 2002 -23.7 -17.1 -4.6 -1.0 -10.1 May 2003 -12.9 -15.7 -1.7 -0.3 -7.8 Rolling Registration: Additions January 2003 76 15 45 3 13,139 February 2003 36 7 5 3 3,238 March 2003 103 9 20 4 5,235 April 2003 118 16 25 10 8,900 Total 333 47 95 20 30,512 Per cent of December 2002 11.8 2.1 3.8 1.1 2.8 Deletions January 2003 2 3 3 1 1,617 February 2003 5 6 1 3 1,586 March 2003 9 6 4 2 1,981 April 2003 6 15 17 5 2,777 Total 22 30 25 11 7,961 Per cent of December 2002 0.8 1.3 1.0 0.6 0.7 Net change (as % of December 2002) January 2003 2.6 0.5 1.7 0.1 1.1 February 2003 1.1 0.0 0.2 0.0 0.2 March 2003 3.3 0.1 0.6 0.1 0.3 April 2003 4.0 0.0 0.3 0.3 0.6 Total 11.0 0.7 2.8 0.5 2.1 B. Demography Age composition of voting-age population (18+)

PricewaterhouseCoopers July 2003 Annexes - Page 16 Electoral Commission Desk Research: First Report

Table B2.9 Summary statistical profile: Case study Wards Variable White- Island Mass- Tempo N. Ireland rock ereene Aged 18-24 15.9 10.7 9.4 11.1 12.8 Aged 25-29 8.8 12.1 11.7 8.8 9.3 Aged 30-44 31.9 27.3 33.7 30.2 30.4 Aged 45-59 22.3 20.1 21.6 24.9 23.5 Aged 60-74 12.8 17.7 15.4 15.5 15.9 Aged 75-84 6.0 9.1 6.0 7.2 6.2 Aged 85+ 2.3 2.9 2.1 2.3 1.9 Attainers Aged 16-17 7.4 3.4 3.4 5.5 4.3 Average age Median 28 35 34 34 34 C. Origins and background Place of Birth (% of population) N. Ireland 96.2 92.9 89.0 93.0 91.0 Great Britain 1.7 4.5 7.0 3.1 4.8 Republic of Ireland 1.2 0.6 1.9 3.1 2.3 Other EU 0.7 0.7 0.5 0.5 0.6 Elsewhere 0.2 1.3 1.6 0.3 1.2 Ethnic background (% of population) White 99.7 99.0 98.7 99.8 99.2 Irish Traveller 0.0 0.0 0.0 0.2 0.1 Other 0.3 1.0 1.3 0.0 0.8 Community background (% of population) Catholic 99.0 5.3 38.8 51.5 43.8 Protestant 0.7 89.5 56.9 48.1 53.1 Other/none 0.3 5.2 4.3 0.4 3.1 D. Employment and unemployment Economic activity rate 41.4 54.9 71.5 63.9 62.0 Employment rate 30.3 48.4 69.3 59.6 57.5 Unemployment rate 26.4 11.8 3.0 6.6 6.9 Long-term unemployed 4.7 2.6 0.9 2.2 1.8 Unemployed - never worked 3.6 1.0 0.1 0.4 0.6 Unemployed or unable to work because 35.5 21.3 9.1 12.6 14.6 sick/disabled Full-time students 2.4 1.6 2.9 2.9 4.1 E. Receipt of benefits Income Support/JSA 55.8 25.2 12.3 18.0 17.0 Incapacity benefit 18.8 10.2 5.5 6.1 8.3

PricewaterhouseCoopers July 2003 Annexes - Page 17 Electoral Commission Desk Research: First Report

Table B2.9 Summary statistical profile: Case study Wards Variable White- Island Mass- Tempo N. Ireland rock ereene A disability benefit 45.9 18.3 11.6 12.6 14.8 F. Qualifications (% of population 16- 74) None 64.6 55.6 34.5 49.4 41.6 Level 1 16.5 16.0 21.0 18.3 17.2 Level 2 10.1 10.4 18.1 15.5 16.4 Level 3 5.3 5.9 9.4 6.8 9.0 Level 4 2.4 8.0 2.5 7.3 10.9 Level 5 1.2 4.2 4.6 2.7 4.9 F. Health status Long-term limiting illness 50.8 40.0 23.1 27.6 27.8 G. Housing Household size – persons 18+ 1.9 1.6 1.9 2.1 1.9 Tenure (% of households) Owner-occupation 36.1 41.1 82.5 78.6 69.6 Social rented 60.2 47.6 12.0 9.8 21.2 Private rented 3.4 9.9 3.1 8.5 6.7 Other rented 0.2 1.4 2.5 3.1 2.5 Type of dwelling (% of households) Detached 5.1 3.6 38.2 72.4 36.5 Semi-detached 17.6 11.5 29.1 14.3 27.3 Terraced 74.4 70.9 26.8 8.7 27.4 Purpose built flat 2.2 12.7 4.8 1.7 7.2 Bedsits 0.8 1.0 0.8 0.4 1.0 In commercial building 0.0 0.3 0.3 1.7 0.3 Caravan 0.0 0.0 0.0 0.8 0.3 Second/holiday homes Per cent of households 0.0 0.1 0.0 1.2 0.8 Communal establishments (% of population aged 18+) Medical and care 2.8 0.1 2.4 0.5 1.1 Educational 0.0 0.0 0.0 0.0 0.3 Other 0.4 0.0 0.0 0.0 0.5 H. Car ownership No car/van 67.8 55.2 21.7 17.6 26.3 1 car/van 29.1 37.6 47.2 42.0 44.5 2 or more cars/vans 3.1 7.2 31.1 40.4 29.2 I. Social Class (NS-SEC) (% of population aged 16-74, excl. students, long-term unemployed and never

PricewaterhouseCoopers July 2003 Annexes - Page 18 Electoral Commission Desk Research: First Report

Table B2.9 Summary statistical profile: Case study Wards Variable White- Island Mass- Tempo N. Ireland rock ereene worked) Managerial and professional 12.0 20.5 34.3 21.5 29.5

Intermediate occupations (incl. small 12.7 14.8 22.6 29.9 23.2 employers) Lower supervisory and technical 9.9 12.8 11.1 10.7 10.3 occupations Routine and manual occupations 65.4 51.9 32.0 38.0 37.1 J. Household composition Pensioner 12.5 27.2 19.5 18.1 20.0 Couples with no children 4.1 10.3 17.8 11.7 12.7 Couples, other households with 25.1 12.8 29.9 35.0 28.4 dependent children Couples, all children non-dependent 6.9 4.9 8.1 9.5 8.2 Lone parent, dependent children 24.3 12.6 4.1 5.4 8.1 Lone parent, all children non-dependent 9.2 5.1 3.7 6.3 4.6 All student 0.0 0.0 0.0 0.0 0.3 One person non-pensioner 13.0 23.7 13.8 10.7 14.5 Other 5.1 3.4 3.2 3.3 3.1 Elections Parliamentary constituency Belfast Belfast South Fermana - West East- Antrim gh & South Tyrone Westminister, 2001 Turnout, 2001 69.9 63.4 62.8 79.0 68.3 Unionist vote 12.7 77.5 73.7 46.6 52.6 Nationalist vote 85.4 6.1 21.4 52.1 42.5 District Election Area (DEA) Lower Victoria Antrim Enniskill Falls Town en Council vote 2001 Turnout 68.9 59.8 54.7 71.3 66.0 Unionist vote 0.0 70.2 66.3 47.9 Nationalist vote 100.0 1.8 27.4 43.1 1. Average population aged 18+ per Census Ward in Northern Ireland

PricewaterhouseCoopers July 2003 Annexes - Page 19 Electoral Commission Desk Research: First Report

Table B4.1 Regression model results – May 2003 Register as per cent of population aged 18+, by Census Ward Standard Coefficient t-statistic Significance error Attainers (% of population 18+) • Aged 16-17 0.8062 0.2656 3.0350 *** Age composition (% of population 18+) • Aged 18-24 0.000 - - - • Aged 25-29 0.2465 0.2268 1.0870 • Aged 30-44 0.3727 0.1515 2.4610 ** • Aged 45-59 0.5901 0.2030 2.9080 *** • Aged 60-74 0.7394 0.1895 3.9020 *** • Aged 75-84 0.7095 0.2413 2.9400 *** • Aged 85+ 0.6174 0.3420 1.8050 * Median age -0.3945 0.1566 -2.5190 ** Household size - persons 18+ 10.0471 3.0390 3.3060 *** Where born (% of population) • N. Ireland 0.000 - - - • Great Britain -0.3946 0.1619 -2.4370 ** • Republic of Ireland 0.2106 0.1263 1.6680 * • Other EU 0.2893 0.5977 0.4840 • Elsewhere -0.3464 0.5488 -0.6310 Ethnic background (% of population) • White 0.000 - - - • Non-white 0.5452 0.5135 1.0620 • Irish Traveller -0.6924 0.6146 -1.1270 Community background (% of population) • Catholic 0.000 - - - • Protestant -0.0076 0.0127 -0.6010 • Other/none -0.1266 0.2263 -0.5600 Communal establishments (% of population) • Medical and care -0.4263 0.1301 -3.2770 *** • Other -0.4115 0.1801 -2.2850 ** Army base in Ward -3.6911 1.9304 -1.9120 * Prison in Ward -6.4455 3.4581 -1.8640 * Second/holiday homes (% of dwellings) -0.2452 0.0782 -3.1340 *** Full-time students (% of population 18+) -0.5068 0.1604 -3.1590 *** Other rented (% of dwellings) -0.2543 0.1376 -1.8480 * Residual household type (% of households) -0.7543 0.1704 -4.4270 ***

PricewaterhouseCoopers July 2003 Annexes - Page 20 Electoral Commission Desk Research: First Report

Table B4.1 Regression model results – May 2003 Register as per cent of population aged 18+, by Census Ward Standard Coefficient t-statistic Significance error Local election turnout, 2001 (%) 0.0920 0.0545 1.6890 * Detached dwellings (%) 0.0580 0.0199 2.9150 *** Owner-occupied (%) 0.0833 0.0360 2.3130 ** Social Class • Managerial and professional -0.0447 0.0941 -0.4750 • Intermediate occupations (incl. small 0.0516 0.0903 0.5720 employers) • Lower supervisory and technical 0.0254 0.1631 0.1560 occupations • Routine manual 0.000 - - Deprivation indicators (%) • No qualifications -0.0225 0.0916 -0.2460 • Long-term unemployed -0.0514 0.3226 -0.1590 • Long-term limiting illness -0.0097 0.0795 -0.1220 VLA data • Additions 0.5017 0.2809 1.7860 * • Deletions -1.2067 1.2039 -1.0020 Transition variables • Postal canvass -2.2229 1.0567 -2.1040 ** • Canvassed electorate - % difference 0.9455 0.0895 10.5640 *** from August Register Constituency effects • Belfast East -1.7939 1.5151 -1.1840 • Belfast North -4.7813 1.3624 -3.5090 *** • Belfast South -3.6033 1.5873 -2.2700 ** • Belfast West -3.3339 1.5181 -2.1960 ** • East Antrim -1.0294 1.4458 -0.7120 • East Londonderry -1.1280 1.1529 -0.9780 • Fermanagh & South Tyrone 1.2582 0.9431 1.3340 • Foyle -0.2361 1.1756 -0.2010 • Lagan Valley 0.1534 1.3116 0.1170 • Mid Ulster 1.5998 0.9418 1.6990 * • Newry & Armagh 0.4078 0.9121 0.4470 • North Antrim 0.0927 1.1662 0.0800 • North Down -0.5595 1.5589 -0.3590 • South Antrim -2.4932 1.3295 -1.8750 * • South Down -0.1118 1.0909 -0.1020 • Strangford -1.3333 1.4189 -0.9400

PricewaterhouseCoopers July 2003 Annexes - Page 21 Electoral Commission Desk Research: First Report

Table B4.1 Regression model results – May 2003 Register as per cent of population aged 18+, by Census Ward Standard Coefficient t-statistic Significance error • Upper Bann -0.3608 1.0608 -0.3400 • West Tyrone 0 - - - Constant 30.0498 17.2937 1.7380 *

Number of observations 582 Adjusted R-square 0.868 Standard error of the estimate 3.627 F-ratio 70.391 Significance of F 0.000 *** Significant at 99 per cent. ** Significant at 95 per cent. * Significant at 90 per cent.

PricewaterhouseCoopers July 2003 Annexes - Page 22 Electoral Commission Desk Research: First Report

Annex C Maps

PricewaterhouseCoopers July 2003 Annexes - Page 23 Electoral Commission Desk Research: First Report

Annex D Statistical Tables: Census Ward

PricewaterhouseCoopers July 2003 Annexes - Page 24 Electoral Commission Desk Research: First Report

Table C.1 Registration rates by Census Ward

Parliamentary Per cent of Population aged 18+ District Council Ward constituency August 2002 December 2002 May 2003

% % % South Antrim Antrim Aldergrove 46 43 44 South Antrim Antrim Balloo 94 84 85 South Antrim Antrim Ballycraigy 99 84 85 South Antrim Antrim Clady 105 93 95 South Antrim Antrim Cranfield 100 92 93 South Antrim Antrim Crumlin 100 82 87 South Antrim Antrim Drumanaway 98 85 87 South Antrim Antrim Farranshane 98 82 85 South Antrim Antrim Fountain Hill 101 81 82 South Antrim Antrim Greystone 99 84 85 South Antrim Antrim Massereene 103 98 101 South Antrim Antrim Parkgate 102 91 94 South Antrim Antrim Randalstown 99 92 94 South Antrim Antrim Shilvodan 99 91 93 South Antrim Antrim Springfarm 82 65 66 South Antrim Antrim Steeple 98 79 80 South Antrim Antrim Stiles 95 77 78 South Antrim Antrim Templepatrick 98 87 89 South Antrim Antrim Toome 103 90 94 Strangford Ards Ballygowan 104 94 96 Strangford Ards Ballyrainey 96 88 89 Strangford Ards Ballywalter 99 89 90 Bradshaw's 99 93 94 Strangford Ards Brae Strangford Ards Carrowdore 95 91 92 Strangford Ards Central 93 81 83 Strangford Ards Comber East 96 86 87 Strangford Ards Comber North 95 87 88 Strangford Ards Comber West 99 90 91 Donaghadee 93 87 87 North Down Ards North Donaghadee 93 82 84 North Down Ards South Strangford Ards Glen 98 85 86

PricewaterhouseCoopers July 2003 Annexes - Page 25 Electoral Commission Desk Research: First Report

Table C.1 Registration rates by Census Ward

Parliamentary Per cent of Population aged 18+ District Council Ward constituency August 2002 December 2002 May 2003

% % % Strangford Ards Gregstown 102 91 93 Strangford Ards Killinchy 101 94 95 Strangford Ards Kircubbin 97 87 89 Strangford Ards Lisbane 101 94 94 Strangford Ards Loughries 94 85 86 North Down Ards Millisle 101 94 95 Strangford Ards Movilla 95 88 89 Strangford Ards Portaferry 103 95 95 Strangford Ards Portavogie 97 89 91 Strangford Ards Scrabo 92 83 84 Strangford Ards Whitespots 103 94 95 Newry & Armagh Armagh Abbey Park 88 79 84 Newry & Armagh Armagh Ballymartrim 104 99 102 Newry & Armagh Armagh Callan Bridge 97 88 90 Newry & Armagh Armagh Carrigatuke 103 94 97 Newry & Armagh Armagh Charlemont 107 99 101 Newry & Armagh Armagh Demesne 99 95 97 Newry & Armagh Armagh Derrynoose 106 97 101 Newry & Armagh Armagh Downs 101 88 89 Newry & Armagh Armagh Hamiltonsbawn 105 102 103 Newry & Armagh Armagh Hockley 101 98 98 Newry & Armagh Armagh Keady 104 90 94 Newry & Armagh Armagh Killeen 113 106 109 Newry & Armagh Armagh Killylea 106 99 101 Newry & Armagh Armagh Laurelvale 105 100 102 Newry & Armagh Armagh Loughgall 104 100 101 Newry & Armagh Armagh Markethill 100 95 97 Newry & Armagh Armagh Milford 104 97 101 Newry & Armagh Armagh Observatory 103 97 99 Newry & Armagh Armagh Poyntz Pass 105 101 103 Newry & Armagh Armagh Rich Hill 98 89 91 Newry & Armagh Armagh Tandragee 98 92 94 Newry & Armagh Armagh The Mall 86 80 79

PricewaterhouseCoopers July 2003 Annexes - Page 26 Electoral Commission Desk Research: First Report

Table C.1 Registration rates by Census Ward

Parliamentary Per cent of Population aged 18+ District Council Ward constituency August 2002 December 2002 May 2003

% % % North Antrim Ballymena Academy 101 96 97 North Antrim Ballymena Ahoghill 101 95 96 North Antrim Ballymena Ardeevin 104 103 104 North Antrim Ballymena Ballee 95 81 82 North Antrim Ballymena Ballykeel 97 81 82 North Antrim Ballymena Ballyloughan 105 94 96 North Antrim Ballymena Broughshane 99 94 96 North Antrim Ballymena Castle Demesne 95 82 82 North Antrim Ballymena Craigywarren 103 100 102 North Antrim Ballymena Cullybackey 100 97 98 North Antrim Ballymena Dunclug 92 78 80 North Antrim Ballymena Dunminning 102 97 97 North Antrim Ballymena Fair Green 97 80 83 North Antrim Ballymena Galgorm 102 97 98 North Antrim Ballymena Glenravel 105 100 102 North Antrim Ballymena Glenwhirry 102 96 97 North Antrim Ballymena Grange 105 102 103 North Antrim Ballymena Harryville 96 86 86 North Antrim Ballymena Kells 102 94 95 North Antrim Ballymena Moat 95 88 88 North Antrim Ballymena Park 100 89 89 North Antrim Ballymena Portglenone 100 95 96 North Antrim Ballymena Slemish 104 97 98 North Antrim Ballymena Summerfield 103 89 91 Ballyhoe and 101 94 96 North Antrim Ballymoney Corkey North Antrim Ballymoney Benvardin 101 96 96 North Antrim Ballymoney Carnany 98 92 92 North Antrim Ballymoney Clogh Mills 96 92 93 North Antrim Ballymoney Dervock 101 92 92 North Antrim Ballymoney Dunloy 102 97 100 North Antrim Ballymoney Fairhill 102 100 101 North Antrim Ballymoney Glebe 104 99 99 North Antrim Ballymoney Killoquin Lower 97 92 92

PricewaterhouseCoopers July 2003 Annexes - Page 27 Electoral Commission Desk Research: First Report

Table C.1 Registration rates by Census Ward

Parliamentary Per cent of Population aged 18+ District Council Ward constituency August 2002 December 2002 May 2003

% % % North Antrim Ballymoney Killoquin Upper 96 94 95 North Antrim Ballymoney Knockaholet 104 95 96 North Antrim Ballymoney Newhill 100 91 94 North Antrim Ballymoney Route 96 94 95 North Antrim Ballymoney Seacon 101 91 93 North Antrim Ballymoney Stranocum 101 95 97 North Antrim Ballymoney The Vow 100 94 95 Upper Bann Banbridge Ballydown 97 91 94 South Down Banbridge Ballyward 100 92 94 Upper Bann Banbridge Banbridge West 102 90 91 South Down Banbridge Bannside 101 95 97 Lagan Valley Banbridge Dromore North 99 91 95 Lagan Valley Banbridge Dromore South 98 95 97 Upper Bann Banbridge Edenderry 96 86 87 Upper Bann Banbridge Fort 98 88 90 Upper Bann Banbridge Gilford 97 91 91 Lagan Valley Banbridge Gransha 103 100 102 South Down Banbridge Katesbridge 105 103 105 Upper Bann Banbridge Lawrencetown 100 94 96 Upper Bann Banbridge Loughbrickland 107 102 104 Lagan Valley Banbridge Quilly 106 100 102 South Down Banbridge Rathfriland 100 99 99 Upper Bann Banbridge Seapatrick 102 94 96 Upper Bann Banbridge The Cut 96 84 86 Belfast West Belfast Andersonstown 97 84 88 Belfast North Belfast Ardoyne 93 74 79 Belfast East Belfast Ballyhackamore 93 82 84 Belfast East Belfast Ballymacarrett 94 83 85 Belfast South Belfast Ballynafeigh 71 58 60 Belfast North Belfast Ballysillan 92 77 80 Belfast West Belfast Beechmount 92 77 81 Belfast North Belfast Bellevue 95 76 78 Belfast East Belfast Belmont 96 84 86

PricewaterhouseCoopers July 2003 Annexes - Page 28 Electoral Commission Desk Research: First Report

Table C.1 Registration rates by Census Ward

Parliamentary Per cent of Population aged 18+ District Council Ward constituency August 2002 December 2002 May 2003

% % % Belfast South Belfast Blackstaff 81 69 71 Belfast East Belfast Bloomfield 89 73 76 Belfast South Belfast Botanic 42 23 24 Belfast North Belfast Castleview 93 84 87 Belfast North Belfast Cavehill 100 90 92 Belfast East Belfast Cherryvalley 88 79 81 Belfast North Belfast Chichester Park 95 82 85 Belfast North Belfast Cliftonville 95 81 85 Belfast West Belfast Clonard 98 79 84 Belfast North Belfast Crumlin 87 77 77 Belfast North Belfast Duncairn 88 73 74 Belfast West Belfast Falls 105 80 89 Belfast West Belfast Falls Park 99 86 90 Belfast South Belfast Finaghy 96 84 87 Belfast North Belfast Fortwilliam 92 80 82 Belfast West Belfast Glen Road 99 85 88 Belfast West Belfast Glencairn 93 77 78 Belfast West Belfast Glencolin 100 78 84 Belfast West Belfast Highfield 95 85 87 Belfast East Belfast Island 84 69 70 Belfast East Belfast Knock 96 87 88 Belfast West Belfast Ladybrook 100 84 88 Belfast North Belfast Legoniel 92 78 80 Belfast South Belfast Malone 92 78 81 Belfast South Belfast Musgrave 95 80 82 Belfast North Belfast New Lodge 99 81 86 Belfast East Belfast Orangefield 97 87 89 Belfast South Belfast Ravenhill 87 75 78 Belfast South Belfast Rosetta 92 83 84 Belfast South Belfast Shaftesbury 83 67 70 Belfast West Belfast Shankill 93 83 84 Belfast East Belfast Stormont 103 93 94 Belfast South Belfast Stranmillis 55 47 49

PricewaterhouseCoopers July 2003 Annexes - Page 29 Electoral Commission Desk Research: First Report

Table C.1 Registration rates by Census Ward

Parliamentary Per cent of Population aged 18+ District Council Ward constituency August 2002 December 2002 May 2003

% % % Belfast East Belfast Sydenham 89 77 79 Belfast East Belfast The Mount 84 73 74 Belfast South Belfast Upper Malone 93 81 82 Upper 102 84 90 Belfast West Belfast Springfield Belfast North Belfast Water Works 95 75 79 Belfast West Belfast Whiterock 108 82 94 Belfast South Belfast Windsor 44 35 37 Belfast South Belfast Woodstock 90 72 74 Belfast North Belfast Woodvale 90 77 78 East Antrim Carrickfergus Blackhead 92 82 85 East Antrim Carrickfergus Bluefield 87 76 78 East Antrim Carrickfergus Boneybefore 98 93 94 East Antrim Carrickfergus Burleigh Hill 102 90 93 East Antrim Carrickfergus Clipperstown 95 82 83 East Antrim Carrickfergus Eden 107 98 100 East Antrim Carrickfergus Gortalee 90 82 83 East Antrim Carrickfergus Greenisland 95 89 90 East Antrim Carrickfergus Killycrot 89 79 81 East Antrim Carrickfergus Knockagh 102 92 94 East Antrim Carrickfergus Love Lane 94 82 82 East Antrim Carrickfergus Milebush 96 85 86 East Antrim Carrickfergus Northland 97 78 81 East Antrim Carrickfergus Sunnylands 93 83 84 East Antrim Carrickfergus Victoria 101 91 93 East Antrim Carrickfergus Whitehead 97 83 84 East Antrim Carrickfergus Woodburn 103 93 94 Strangford Castlereagh Ballyhanwood 101 90 93 Belfast South Castlereagh Beechill 85 74 75 Belfast South Castlereagh Cairnshill 100 88 90 Strangford Castlereagh Carrowreagh 96 82 83 Strangford Castlereagh Carryduff East 97 88 89 Strangford Castlereagh Carryduff West 97 87 89 Belfast East Castlereagh Cregagh 92 84 84

PricewaterhouseCoopers July 2003 Annexes - Page 30 Electoral Commission Desk Research: First Report

Table C.1 Registration rates by Census Ward

Parliamentary Per cent of Population aged 18+ District Council Ward constituency August 2002 December 2002 May 2003

% % % Belfast East Castlereagh Downshire 95 85 85 Strangford Castlereagh Dundonald 92 83 84 Strangford Castlereagh Enler 95 81 82 Belfast South Castlereagh Galwally 94 81 85 Belfast East Castlereagh Gilnahirk 100 92 93 Graham's 97 90 90 Strangford Castlereagh Bridge Belfast East Castlereagh Hillfoot 99 91 92 Belfast South Castlereagh Knockbracken 92 86 87 Belfast East Castlereagh Lisnasharragh 105 96 97 Belfast East Castlereagh Lower Braniel 99 89 91 Belfast South Castlereagh Minnowburn 93 86 88 Strangford Castlereagh Moneyreagh 101 93 94 Belfast South Castlereagh Newtownbreda 97 84 86 Belfast East Castlereagh Tullycarnet 96 83 85 Belfast East Castlereagh Upper Braniel 99 91 92 Belfast East Castlereagh Wynchurch 94 83 85 East 100 95 96 Coleraine Agivey Londonderry East 86 75 76 Coleraine Atlantic Londonderry East 93 78 78 Coleraine Ballysally Londonderry East 93 87 88 Coleraine Castlerock Londonderry East 93 83 83 Coleraine Central Londonderry East 98 83 85 Coleraine Churchland Londonderry East 97 81 83 Coleraine Cross Glebe Londonderry East 85 79 79 Coleraine Dundooan Londonderry East 93 86 87 Coleraine Dunluce Londonderry East 99 93 94 Coleraine Garvagh Londonderry East 94 87 89 Coleraine Hopefield Londonderry East 101 93 95 Coleraine Kilrea Londonderry East 101 96 98 Coleraine Knocklynn Londonderry

PricewaterhouseCoopers July 2003 Annexes - Page 31 Electoral Commission Desk Research: First Report

Table C.1 Registration rates by Census Ward

Parliamentary Per cent of Population aged 18+ District Council Ward constituency August 2002 December 2002 May 2003

% % % East 102 95 97 Coleraine Macosquin Londonderry East 96 85 86 Coleraine Mount Sandel Londonderry East 79 69 70 Coleraine Portstewart Londonderry East 104 102 103 Coleraine Ringsend Londonderry East 83 74 74 Coleraine Royal Portrush Londonderry East 71 63 64 Coleraine Strand Londonderry East 99 89 90 Coleraine The Cuts Londonderry East 77 69 70 Coleraine University Londonderry East 103 99 101 Coleraine Waterside Londonderry Mid Ulster Cookstown Ardboe 107 101 103 Mid Ulster Cookstown Coagh 99 94 96 Mid Ulster Cookstown Dunnamore 106 100 102 Mid Ulster Cookstown Gortalowry 98 90 92 Mid Ulster Cookstown Killycolpy 111 102 105 Mid Ulster Cookstown Killymoon 100 88 91 Mid Ulster Cookstown Lissan 107 106 107 Mid Ulster Cookstown Moneymore 96 94 96 Mid Ulster Cookstown Newbuildings 99 90 92 Mid Ulster Cookstown Oaklands 100 97 99 Mid Ulster Cookstown Oldtown 102 96 98 Mid Ulster Cookstown Pomeroy 99 92 95 Mid Ulster Cookstown Sandholes 100 97 99 Mid Ulster Cookstown Stewartstown 104 97 99 Mid Ulster Cookstown The Loop 103 99 100 Mid Ulster Cookstown Tullagh 104 97 99 Upper Bann Craigavon Aghagallon 105 97 101 Upper Bann Craigavon Annagh 90 85 86 Upper Bann Craigavon Ballybay 94 85 86 Upper Bann Craigavon Ballyoran 104 97 99 Upper Bann Craigavon Bleary 104 99 101 Upper Bann Craigavon Brownstown 88 81 82

PricewaterhouseCoopers July 2003 Annexes - Page 32 Electoral Commission Desk Research: First Report

Table C.1 Registration rates by Census Ward

Parliamentary Per cent of Population aged 18+ District Council Ward constituency August 2002 December 2002 May 2003

% % % Upper Bann Craigavon Church 92 84 85 Upper Bann Craigavon Corcrain 93 81 83 Upper Bann Craigavon Court 86 77 79 Upper Bann Craigavon Derrytrasna 103 99 101 Upper Bann Craigavon Donaghcloney 103 95 98 Upper Bann Craigavon Drumgask 96 86 88 Upper Bann Craigavon Drumgor 93 77 80 Upper Bann Craigavon Drumnamoe 103 88 92 Upper Bann Craigavon Edenderry 95 86 88 Upper Bann Craigavon Kernan 106 101 103 Upper Bann Craigavon Killycomain 94 88 88 Upper Bann Craigavon Knocknashane 101 94 96 Upper Bann Craigavon Magheralin 103 101 102 Upper Bann Craigavon Mourneview 96 87 88 Upper Bann Craigavon Parklake 98 92 93 Upper Bann Craigavon Taghnevan 105 100 102 Upper Bann Craigavon Tavanagh 98 91 90 Upper Bann Craigavon The Birches 103 99 100 Upper Bann Craigavon Waringstown 99 94 95 Upper Bann Craigavon Woodville 96 89 91 Foyle Derry Altnagelvin 95 82 84 Foyle Derry Ballynashallog 106 94 97 Foyle Derry Banagher 101 98 99 Foyle Derry Beechwood 98 85 90 Foyle Derry Brandywell 102 88 94 Foyle Derry Carn Hill 97 82 86 Foyle Derry Caw 92 84 88 Foyle Derry Clondermot 103 96 99 Foyle Derry Claudy 105 98 100 Foyle Derry Creggan Central 103 85 93 Foyle Derry Creggan South 106 94 99 Foyle Derry Crevagh 100 85 91 Foyle Derry Culmore 99 84 90

PricewaterhouseCoopers July 2003 Annexes - Page 33 Electoral Commission Desk Research: First Report

Table C.1 Registration rates by Census Ward

Parliamentary Per cent of Population aged 18+ District Council Ward constituency August 2002 December 2002 May 2003

% % % Foyle Derry Ebrington 64 60 61 Foyle Derry Eglinton 98 91 93 Foyle Derry Enagh 101 94 96 Foyle Derry Foyle Springs 104 91 96 Foyle Derry Holly Mount 102 94 97 Foyle Derry Kilfennan 103 90 92 Foyle Derry Lisnagelvin 90 82 84 Foyle Derry New Buildings 100 92 93 Foyle Derry Pennyburn 101 88 91 Foyle Derry Rosemount 89 78 81 Foyle Derry Shantallow East 102 90 93 Shantallow 105 87 91 Foyle Derry West Foyle Derry Springtown 105 89 94 Foyle Derry Strand 67 59 62 Foyle Derry The Diamond 91 81 83 Foyle Derry Victoria 102 91 92 Foyle Derry Westland 102 87 92 South Down Down Ardglass 102 90 93 South Down Down Audley's Acre 105 94 96 South Down Down Ballymaglave 97 88 91 South Down Down Ballymote 88 79 82 Ballynahinch 98 84 86 South Down Down East South Down Down Castlewellan 105 94 97 South Down Down Cathedral 98 88 90 South Down Down Crossgar 102 96 98 Strangford Down Derryboy 100 90 92 South Down Down Donard 99 92 94 South Down Down Drumaness 103 98 100 South Down Down Dundrum 101 96 97 South Down Down Dunmore 105 99 101 South Down Down Killough 73 66 69 Strangford Down Killyleagh 97 92 92 South Down Down Kilmore 105 97 98

PricewaterhouseCoopers July 2003 Annexes - Page 34 Electoral Commission Desk Research: First Report

Table C.1 Registration rates by Census Ward

Parliamentary Per cent of Population aged 18+ District Council Ward constituency August 2002 December 2002 May 2003

% % % South Down Down Murlough 100 91 92 South Down Down Quoile 98 89 90 Strangford Down Saintfield 102 93 94 South Down Down Seaforde 105 99 100 South Down Down Shimna 101 93 95 South Down Down Strangford 103 95 97 South Down Down Tollymore 97 92 94 Mid Ulster Dungannon Altmore 111 108 110 Fermanagh & 110 105 107 Dungannon Augher South Tyrone Fermanagh & 104 99 101 Dungannon Aughnacloy South Tyrone Fermanagh & 107 97 99 Dungannon Ballygawley South Tyrone Fermanagh & 106 92 98 Dungannon Ballysaggart South Tyrone Fermanagh & 105 99 101 Dungannon Benburb South Tyrone Fermanagh & 109 103 104 Dungannon Caledon South Tyrone Fermanagh & 110 102 106 Dungannon Castlecaulfield South Tyrone Fermanagh & 105 101 102 Dungannon Clogher South Tyrone Mid Ulster Dungannon Coalisland North 109 99 103 Coalisland 100 91 95 Mid Ulster Dungannon South Coalisland West 106 98 99 Mid Ulster Dungannon and Newmills Fermanagh & 94 85 86 Dungannon Coolhill South Tyrone Mid Ulster Dungannon Donaghmore 109 100 102 Fermanagh & 101 85 88 Dungannon Drumglass South Tyrone Fermanagh & 101 95 97 Dungannon Fivemiletown South Tyrone Fermanagh & 105 100 101 Dungannon Killyman South Tyrone Fermanagh & 108 98 100 Dungannon Killymeal South Tyrone Fermanagh & 106 98 99 Dungannon Moy South Tyrone Fermanagh & 103 97 99 Dungannon Moygashel South Tyrone Fermanagh & 88 79 82 Dungannon Mullaghmore South Tyrone

PricewaterhouseCoopers July 2003 Annexes - Page 35 Electoral Commission Desk Research: First Report

Table C.1 Registration rates by Census Ward

Parliamentary Per cent of Population aged 18+ District Council Ward constituency August 2002 December 2002 May 2003

% % % Mid Ulster Dungannon Washing Bay 118 113 115 Fermanagh & 103 97 99 Fermanagh Ballinamallard South Tyrone Fermanagh & Belcoo and 104 94 98 Fermanagh South Tyrone Garrison Fermanagh & 98 92 95 Fermanagh Belleek and Boa South Tyrone Fermanagh & Boho Cleenish 105 100 102 Fermanagh South Tyrone and Letterbreen Fermanagh & 109 104 105 Fermanagh Brookeborough South Tyrone Fermanagh & 103 96 98 Fermanagh Castlecoole South Tyrone Fermanagh & 107 100 101 Fermanagh Derrygonnelly South Tyrone Fermanagh & 104 99 102 Fermanagh Derrylin South Tyrone Fermanagh & 89 81 81 Fermanagh Devenish South Tyrone Fermanagh & 109 103 106 Fermanagh Donagh South Tyrone Fermanagh & 93 86 86 Fermanagh Erne South Tyrone Fermanagh & Florence Court 104 98 101 Fermanagh South Tyrone and Kinawley Fermanagh & 105 94 97 Fermanagh Irvinestown South Tyrone Fermanagh & Kesh Ederney 102 95 97 Fermanagh South Tyrone and Lack Fermanagh & 99 94 96 Fermanagh Lisbellaw South Tyrone Fermanagh & 100 97 98 Fermanagh Lisnarrick South Tyrone Fermanagh & 99 94 97 Fermanagh Lisnaskea South Tyrone Fermanagh & 101 99 100 Fermanagh Maguires Bridge South Tyrone Fermanagh & 106 96 100 Fermanagh Newtownbutler South Tyrone Fermanagh & 99 89 91 Fermanagh Portora South Tyrone Fermanagh & 108 103 106 Fermanagh Rosslea South Tyrone Fermanagh & 101 94 96 Fermanagh Rossorry South Tyrone Fermanagh & 104 103 104 Fermanagh Tempo South Tyrone East Antrim Larne Antiville 97 81 82 East Antrim Larne Ballycarry 98 89 91 East Antrim Larne Ballyloran 94 78 80

PricewaterhouseCoopers July 2003 Annexes - Page 36 Electoral Commission Desk Research: First Report

Table C.1 Registration rates by Census Ward

Parliamentary Per cent of Population aged 18+ District Council Ward constituency August 2002 December 2002 May 2003

% % % East Antrim Larne Blackcave 96 82 83 East Antrim Larne Carncastle 104 95 96 East Antrim Larne Carnlough 101 96 96 East Antrim Larne Central 90 80 81 East Antrim Larne Craigy Hill 95 82 83 East Antrim Larne Gardenmore 101 90 92 East Antrim Larne Glenarm 95 86 88 East Antrim Larne Glynn 100 91 92 East Antrim Larne Harbour 99 92 92 East Antrim Larne Island Magee 100 88 90 East Antrim Larne Kilwaughter 103 98 99 East Antrim Larne Town Parks 99 87 88 East 99 96 99 Limavady Aghanloo Londonderry East 94 86 88 Limavady Ballykelly Londonderry East 92 86 86 Limavady Coolessan Londonderry East 101 92 93 Limavady Dungiven Londonderry East 112 101 104 Limavady Enagh Londonderry East 101 95 97 Limavady Feeny Londonderry East 88 80 81 Limavady Forest Londonderry East 99 95 96 Limavady Glack Londonderry East 67 62 64 Limavady Gresteel Londonderry East 99 85 85 Limavady Greystone Londonderry East 80 74 75 Limavady Magilligan Londonderry East 103 94 95 Limavady Rathbrady Londonderry East 82 74 75 Limavady Roeside Londonderry East 99 95 97 Limavady The Highlands Londonderry East Upper 105 97 99 Limavady Londonderry Glenshane Lagan Valley Lisburn Ballinderry 101 93 95 Lagan Valley Lisburn Ballymacash 94 85 87

PricewaterhouseCoopers July 2003 Annexes - Page 37 Electoral Commission Desk Research: First Report

Table C.1 Registration rates by Census Ward

Parliamentary Per cent of Population aged 18+ District Council Ward constituency August 2002 December 2002 May 2003

% % % 104 96 98 Lagan Valley Lisburn Ballymacbrenna n Lagan Valley Lisburn Ballymacoss 97 86 90 Lagan Valley Lisburn Blaris 101 93 94 Belfast West Lisburn Collin Glen 101 79 84 Lagan Valley Lisburn Derryaghy 96 88 97 Lagan Valley Lisburn Dromara 101 96 98 Lagan Valley Lisburn Drumbo 104 95 97 Lagan Valley Lisburn Dunmurry 97 82 87 Lagan Valley Lisburn Glenavy 102 90 94 Lagan Valley Lisburn Harmony Hill 95 88 90 Lagan Valley Lisburn Hilden 92 80 83 Lagan Valley Lisburn Hillhall 97 86 88 Lagan Valley Lisburn Hillsborough 101 95 98 Belfast West Lisburn Kilwee 102 81 85 Lagan Valley Lisburn Knockmore 91 80 82 Lagan Valley Lisburn Lagan Valley 93 81 83 Lagan Valley Lisburn Lambeg 96 83 85 Lagan Valley Lisburn Lisnagarvey 99 91 93 Lagan Valley Lisburn Maghaberry 93 86 89 Lagan Valley Lisburn Magheralave 83 74 76 Lagan Valley Lisburn Maze 100 90 94 Lagan Valley Lisburn Moira 100 93 96 Lagan Valley Lisburn Old Warren 97 82 85 Belfast West Lisburn Poleglass 99 76 79 Lagan Valley Lisburn Seymour Hill 89 82 84 Lagan Valley Lisburn Tonagh 98 83 85 Belfast West Lisburn Twinbrook 101 77 80 Lagan Valley Lisburn Wallace Park 62 58 59 Mid Ulster Magherafelt Ballymaguigan 102 97 100 Mid Ulster Magherafelt Bellaghy 103 98 105 Mid Ulster Magherafelt Castledawson 93 87 88 Mid Ulster Magherafelt Draperstown 105 103 106

PricewaterhouseCoopers July 2003 Annexes - Page 38 Electoral Commission Desk Research: First Report

Table C.1 Registration rates by Census Ward

Parliamentary Per cent of Population aged 18+ District Council Ward constituency August 2002 December 2002 May 2003

% % % Mid Ulster Magherafelt Glebe 100 94 95 Mid Ulster Magherafelt Gulladuff 103 98 100 Mid Ulster Magherafelt Knockcloghrim 98 92 94 Mid Ulster Magherafelt Lecumpher 102 102 103 Lower 108 103 106 Mid Ulster Magherafelt Glenshane Mid Ulster Magherafelt Maghera 98 92 96 Mid Ulster Magherafelt Swatragh 111 108 110 Mid Ulster Magherafelt Tobermore 103 101 102 Town Parks 96 88 90 Mid Ulster Magherafelt East Town Parks 101 94 97 Mid Ulster Magherafelt West Mid Ulster Magherafelt Upperlands 98 97 98 Mid Ulster Magherafelt Valley 100 93 96 North Antrim Moyle Armoy 102 93 94 North Antrim Moyle Ballylough 99 94 95 Bonamargy and 94 85 86 North Antrim Moyle Rathlin North Antrim Moyle Bushmills 99 84 85 North Antrim Moyle Carnmoon 98 94 95 North Antrim Moyle Dalriada 99 89 91 North Antrim Moyle Dunseverick 95 89 90 North Antrim Moyle Glenaan 92 89 90 North Antrim Moyle Glenariff 101 94 97 North Antrim Moyle Glendun 101 90 92 North Antrim Moyle Glenshesk 96 92 92 North Antrim Moyle Glentaisie 99 92 94 North Antrim Moyle Kinbane 95 81 83 North Antrim Moyle Knocklayd 97 84 87 Moss-side and 100 98 97 North Antrim Moyle Moyarget Newry and 99 91 93 South Down Annalong Mourne Newry and 100 84 86 Newry & Armagh Ballybot Mourne Newry and 97 90 91 Newry & Armagh Bessbrook Mourne Newry and 101 92 97 South Down Binnian Mourne

PricewaterhouseCoopers July 2003 Annexes - Page 39 Electoral Commission Desk Research: First Report

Table C.1 Registration rates by Census Ward

Parliamentary Per cent of Population aged 18+ District Council Ward constituency August 2002 December 2002 May 2003

% % % Newry and Burren and 106 88 97 South Down Mourne Kilbroney Newry and 105 91 98 Newry & Armagh Camlough Mourne Newry and 100 93 95 South Down Clonallan Mourne Newry and 116 109 113 Newry & Armagh Creggan Mourne Newry and 108 96 101 Newry & Armagh Crossmaglen Mourne Newry and 98 83 87 Newry & Armagh Daisy hill Mourne Newry and 102 90 96 South Down Derryleckagh Mourne Newry and 100 89 92 Newry & Armagh Derrymore Mourne Newry and 101 95 97 South Down Donaghmore Mourne Newry and 99 90 93 Newry & Armagh Drumalane Mourne Newry and 98 83 87 Newry & Armagh Drumgullion Mourne Newry and 109 100 102 Newry & Armagh Fathom Mourne Newry and 129 117 122 Newry & Armagh Forkhill Mourne Newry and 100 91 93 South Down Kilkeel Central Mourne Newry and 102 95 95 South Down Kilkeel South Mourne Newry and 103 96 97 South Down Lisnacree Mourne Newry and 103 95 100 South Down Mayobridge Mourne 107 98 101 Newry and Newry & Armagh Newtownhamilto Mourne n Newry and 110 102 104 South Down Rostrevor Mourne Newry and 97 82 86 South Down Seaview Mourne Newry and 116 102 108 Newry & Armagh Silver Bridge Mourne Newry and 105 97 100 South Down Spelga Mourne Newry and 95 81 84 Newry & Armagh St Mary's Mourne Newry and 89 78 81 Newry & Armagh St Patrick's Mourne Newry and 105 99 101 Newry & Armagh Tullyhappy Mourne

PricewaterhouseCoopers July 2003 Annexes - Page 40 Electoral Commission Desk Research: First Report

Table C.1 Registration rates by Census Ward

Parliamentary Per cent of Population aged 18+ District Council Ward constituency August 2002 December 2002 May 2003

% % % Newry and 97 84 87 Newry & Armagh Windsor Hill Mourne Belfast North Newtownabbey Abbey 92 78 80 South Antrim Newtownabbey Ballyclare North 96 83 85 South Antrim Newtownabbey Ballyclare South 98 87 89 South Antrim Newtownabbey Ballyduff 99 86 88 South Antrim Newtownabbey Ballyhenry 100 90 92 South Antrim Newtownabbey Ballynure 103 93 95 South Antrim Newtownabbey Ballyrobert 100 93 95 South Antrim Newtownabbey Burnthill 98 87 88 South Antrim Newtownabbey Carnmoney 96 83 84 East Antrim Newtownabbey Cloughfern 99 85 87 South Antrim Newtownabbey Collinbridge 98 86 89 Belfast North Newtownabbey Coole 96 76 78 South Antrim Newtownabbey Doagh 100 90 92 Belfast North Newtownabbey Dunanney 94 75 78 South Antrim Newtownabbey Glebe 98 88 89 South Antrim Newtownabbey Glengormley 87 76 77 South Antrim Newtownabbey Hawthorne 95 84 85 South Antrim Newtownabbey Hightown 98 88 89 East Antrim Newtownabbey Jordanstown 101 95 97 South Antrim Newtownabbey Mallusk 101 94 96 East Antrim Newtownabbey Monkstown 96 84 85 South Antrim Newtownabbey Mossley 95 85 86 East Antrim Newtownabbey Rostulla 73 66 68 Belfast North Newtownabbey Valley 94 78 81 Belfast North Newtownabbey Whitehouse 92 82 83 North Down North Down Ballycrochan 103 90 92 North Down North Down Ballyholme 99 88 90 North Down North Down Ballymaconnell 102 92 92 North Down North Down Ballymagee 96 88 89 North Down North Down Bangor Castle 94 87 86 North Down North Down Bloomfield 97 86 88 North Down North Down Broadway 99 92 93

PricewaterhouseCoopers July 2003 Annexes - Page 41 Electoral Commission Desk Research: First Report

Table C.1 Registration rates by Census Ward

Parliamentary Per cent of Population aged 18+ District Council Ward constituency August 2002 December 2002 May 2003

% % % North Down North Down Bryansburn 96 87 88 North Down North Down Churchill 96 87 87 North Down North Down Clandeboye 94 81 84 North Down North Down Conlig 95 87 90 North Down North Down Craigavad 96 86 88 North Down North Down Crawfordsburn 104 96 97 North Down North Down Cultra 103 90 91 North Down North Down Dufferin 94 82 84 North Down North Down Groomsport 102 97 98 North Down North Down Harbour 92 74 76 Holywood 93 82 84 North Down North Down Demesne North Down North Down Holywood Priory 99 91 93 North Down North Down Loughview 56 49 49 North Down North Down Princetown 97 89 89 North Down North Down Rathgael 96 85 86 North Down North Down Silverstream 91 78 79 North Down North Down Springhill 102 92 94 North Down North Down Whitehill 96 81 83 West Tyrone Omagh Beragh 107 99 100 West Tyrone Omagh Camowen 94 84 85 West Tyrone Omagh Clanabogan 109 104 106 West Tyrone Omagh Coolnagard 112 105 108 West Tyrone Omagh Dergmoney 92 87 87 West Tyrone Omagh Dromore 104 100 102 West Tyrone Omagh Drumnakilly 115 104 109 West Tyrone Omagh Drumquin 105 100 103 West Tyrone Omagh Drumragh 92 88 89 West Tyrone Omagh Fairy Water 105 101 102 West Tyrone Omagh Fintona 99 90 94 West Tyrone Omagh Gortin 98 92 94 West Tyrone Omagh Gortrush 93 79 81 West Tyrone Omagh Killyclogher 97 89 92 West Tyrone Omagh Lisanelly 63 55 57

PricewaterhouseCoopers July 2003 Annexes - Page 42 Electoral Commission Desk Research: First Report

Table C.1 Registration rates by Census Ward

Parliamentary Per cent of Population aged 18+ District Council Ward constituency August 2002 December 2002 May 2003

% % % West Tyrone Omagh Newtownsaville 109 104 107 West Tyrone Omagh Owenkillew 103 101 105 West Tyrone Omagh Sixmilecross 109 100 104 West Tyrone Omagh Strule 78 74 75 West Tyrone Omagh Termon 111 110 113 West Tyrone Omagh Trillick 107 104 106 West Tyrone Strabane Artigarvan 105 98 101 West Tyrone Strabane Ballycolman 100 90 92 West Tyrone Strabane Castlederg 103 93 95 West Tyrone Strabane Clare 104 97 100 West Tyrone Strabane Dunnamanagh 107 102 105 West Tyrone Strabane East 110 98 99 West Tyrone Strabane Finn 98 91 92 West Tyrone Strabane Glenderg 102 97 98 West Tyrone Strabane Newtownstewart 101 93 96 West Tyrone Strabane North 102 90 96 West Tyrone Strabane Plumbridge 102 95 99 West Tyrone Strabane Sion Mills 96 86 87 West Tyrone Strabane Slievekirk 99 93 94 West Tyrone Strabane South 99 87 91 West Tyrone Strabane Victoria Bridge 103 95 96 West Tyrone Strabane West 96 89 90

PricewaterhouseCoopers July 2003 Annexes - Page 43 Electoral Commission Desk Research: First Report

Table C.2 The change in the number of persons on the Electoral Register by Census Ward

% change in number registered: Parliamentary District Council Ward From August 2002 to: constituency December 2002 to May 2003 December 2002 May 2003

% % % South Antrim Antrim Aldergrove -7 -5 2 South Antrim Antrim Balloo -11 -10 1 South Antrim Antrim Ballycraigy -15 -14 1 South Antrim Antrim Clady -12 -9 3 South Antrim Antrim Cranfield -8 -6 2 South Antrim Antrim Crumlin -19 -14 6 South Antrim Antrim Drumanaway -13 -11 2 South Antrim Antrim Farranshane -17 -14 3 South Antrim Antrim Fountain Hill -20 -19 1 South Antrim Antrim Greystone -16 -14 2 South Antrim Antrim Massereene -5 -2 3 South Antrim Antrim Parkgate -11 -8 3 South Antrim Antrim Randalstown -7 -5 2 South Antrim Antrim Shilvodan -8 -7 2 South Antrim Antrim Springfarm -21 -20 2 South Antrim Antrim Steeple -19 -18 1 South Antrim Antrim Stiles -19 -17 2 South Antrim Antrim Templepatrick -11 -9 2 South Antrim Antrim Toome -12 -9 3 Strangford Ards Ballygowan -9 -7 2 Strangford Ards Ballyrainey -8 -7 1 Strangford Ards Ballywalter -11 -9 2 Bradshaw's -6 -6 1 Strangford Ards Brae Strangford Ards Carrowdore -4 -3 1 Strangford Ards Central -13 -11 1 Strangford Ards Comber East -10 -10 1 Strangford Ards Comber North -9 -8 1 Strangford Ards Comber West -10 -8 2 North Down Ards Donaghadee -7 -7 0

PricewaterhouseCoopers July 2003 Annexes - Page 44 Electoral Commission Desk Research: First Report

Table C.2 The change in the number of persons on the Electoral Register by Census Ward

% change in number registered: Parliamentary District Council Ward From August 2002 to: constituency December 2002 to May 2003 December 2002 May 2003

% % % North Donaghadee -12 -10 2 North Down Ards South Strangford Ards Glen -13 -12 1 Strangford Ards Gregstown -11 -9 2 Strangford Ards Killinchy -7 -6 1 Strangford Ards Kircubbin -10 -9 1 Strangford Ards Lisbane -7 -7 0 Strangford Ards Loughries -10 -8 1 North Down Ards Millisle -7 -6 1 Strangford Ards Movilla -8 -7 1 Strangford Ards Portaferry -8 -8 0 Strangford Ards Portavogie -8 -6 2 Strangford Ards Scrabo -10 -9 1 Strangford Ards Whitespots -9 -8 1 Newry & Armagh Armagh Abbey Park -10 -5 6 Newry & Armagh Armagh Ballymartrim -5 -2 4 Newry & Armagh Armagh Callan Bridge -10 -8 2 Newry & Armagh Armagh Carrigatuke -9 -6 3 Newry & Armagh Armagh Charlemont -7 -6 2 Newry & Armagh Armagh Demesne -5 -3 2 Newry & Armagh Armagh Derrynoose -9 -5 4 Newry & Armagh Armagh Downs -13 -11 2 Newry & Armagh Armagh Hamiltonsbawn -3 -2 1 Newry & Armagh Armagh Hockley -3 -3 0 Newry & Armagh Armagh Keady -13 -10 4 Newry & Armagh Armagh Killeen -6 -4 2 Newry & Armagh Armagh Killylea -6 -4 2 Newry & Armagh Armagh Laurelvale -5 -2 2 Newry & Armagh Armagh Loughgall -4 -3 2 Newry & Armagh Armagh Markethill -5 -3 2

PricewaterhouseCoopers July 2003 Annexes - Page 45 Electoral Commission Desk Research: First Report

Table C.2 The change in the number of persons on the Electoral Register by Census Ward

% change in number registered: Parliamentary District Council Ward From August 2002 to: constituency December 2002 to May 2003 December 2002 May 2003

% % % Newry & Armagh Armagh Milford -7 -3 4 Newry & Armagh Armagh Observatory -6 -4 2 Newry & Armagh Armagh Poyntz Pass -4 -2 2 Newry & Armagh Armagh Rich Hill -8 -7 1 Newry & Armagh Armagh Tandragee -6 -5 1 Newry & Armagh Armagh The Mall -8 -8 -0 North Antrim Ballymena Academy -5 -4 1 North Antrim Ballymena Ahoghill -6 -5 1 North Antrim Ballymena Ardeevin -1 0 1 North Antrim Ballymena Ballee -14 -13 1 North Antrim Ballymena Ballykeel -16 -15 1 North Antrim Ballymena Ballyloughan -10 -8 2 North Antrim Ballymena Broughshane -5 -3 1 North Antrim Ballymena Castle Demesne -13 -14 -0 North Antrim Ballymena Craigywarren -3 -2 1 North Antrim Ballymena Cullybackey -4 -2 2 North Antrim Ballymena Dunclug -15 -14 2 North Antrim Ballymena Dunminning -4 -4 0 North Antrim Ballymena Fair Green -17 -14 3 North Antrim Ballymena Galgorm -5 -4 1 North Antrim Ballymena Glenravel -5 -3 2 North Antrim Ballymena Glenwhirry -6 -5 1 North Antrim Ballymena Grange -3 -2 2 North Antrim Ballymena Harryville -11 -10 1 North Antrim Ballymena Kells -7 -6 1 North Antrim Ballymena Moat -8 -7 1 North Antrim Ballymena Park -11 -12 -1 North Antrim Ballymena Portglenone -5 -4 1 North Antrim Ballymena Slemish -7 -6 1 North Antrim Ballymena Summerfield -14 -12 2

PricewaterhouseCoopers July 2003 Annexes - Page 46 Electoral Commission Desk Research: First Report

Table C.2 The change in the number of persons on the Electoral Register by Census Ward

% change in number registered: Parliamentary District Council Ward From August 2002 to: constituency December 2002 to May 2003 December 2002 May 2003

% % % Ballyhoe and -7 -5 2 North Antrim Ballymoney Corkey North Antrim Ballymoney Benvardin -5 -5 0 North Antrim Ballymoney Carnany -6 -6 0 North Antrim Ballymoney Clogh Mills -5 -4 1 North Antrim Ballymoney Dervock -9 -8 1 North Antrim Ballymoney Dunloy -5 -2 3 North Antrim Ballymoney Fairhill -2 -2 1 North Antrim Ballymoney Glebe -5 -5 0 North Antrim Ballymoney Killoquin Lower -6 -5 0 North Antrim Ballymoney Killoquin Upper -1 -1 1 North Antrim Ballymoney Knockaholet -9 -8 1 North Antrim Ballymoney Newhill -9 -6 3 North Antrim Ballymoney Route -2 -2 1 North Antrim Ballymoney Seacon -10 -8 2 North Antrim Ballymoney Stranocum -6 -4 2 North Antrim Ballymoney The Vow -6 -5 2 Upper Bann Banbridge Ballydown -6 -3 3 South Down Banbridge Ballyward -8 -6 3 Upper Bann Banbridge Banbridge West -12 -11 1 South Down Banbridge Bannside -6 -3 3 Lagan Valley Banbridge Dromore North -8 -4 5 Lagan Valley Banbridge Dromore South -4 -1 2 Upper Bann Banbridge Edenderry -11 -9 2 Upper Bann Banbridge Fort -11 -9 2 Upper Bann Banbridge Gilford -6 -6 1 Lagan Valley Banbridge Gransha -2 -1 2 South Down Banbridge Katesbridge -2 -0 2 Upper Bann Banbridge Lawrencetown -6 -4 2 Upper Bann Banbridge Loughbrickland -5 -3 2 Lagan Valley Banbridge Quilly -6 -4 2

PricewaterhouseCoopers July 2003 Annexes - Page 47 Electoral Commission Desk Research: First Report

Table C.2 The change in the number of persons on the Electoral Register by Census Ward

% change in number registered: Parliamentary District Council Ward From August 2002 to: constituency December 2002 to May 2003 December 2002 May 2003

% % % South Down Banbridge Rathfriland -1 -1 0 Upper Bann Banbridge Seapatrick -8 -6 2 Upper Bann Banbridge The Cut -13 -11 2 Belfast West Belfast Andersonstown -14 -10 5 Belfast North Belfast Ardoyne -20 -15 6 Belfast East Belfast Ballyhackamore -12 -10 3 Belfast East Belfast Ballymacarrett -12 -9 2 Belfast South Belfast Ballynafeigh -19 -16 4 Belfast North Belfast Ballysillan -16 -13 4 Belfast West Belfast Beechmount -17 -12 6 Belfast North Belfast Bellevue -20 -18 3 Belfast East Belfast Belmont -13 -11 2 Belfast South Belfast Blackstaff -15 -12 3 Belfast East Belfast Bloomfield -18 -14 4 Belfast South Belfast Botanic -45 -42 5 Belfast North Belfast Castleview -9 -7 3 Belfast North Belfast Cavehill -10 -8 2 Belfast East Belfast Cherryvalley -11 -8 3 Belfast North Belfast Chichester Park -14 -11 3 Belfast North Belfast Cliftonville -14 -11 4 Belfast West Belfast Clonard -19 -14 6 Belfast North Belfast Crumlin -12 -11 1 Belfast North Belfast Duncairn -17 -15 2 Belfast West Belfast Falls -24 -15 12 Belfast West Belfast Falls Park -13 -9 4 Belfast South Belfast Finaghy -12 -9 3 Belfast North Belfast Fortwilliam -13 -11 2 Belfast West Belfast Glen Road -14 -11 4 Belfast West Belfast Glencairn -17 -16 2 Belfast West Belfast Glencolin -22 -16 8

PricewaterhouseCoopers July 2003 Annexes - Page 48 Electoral Commission Desk Research: First Report

Table C.2 The change in the number of persons on the Electoral Register by Census Ward

% change in number registered: Parliamentary District Council Ward From August 2002 to: constituency December 2002 to May 2003 December 2002 May 2003

% % % Belfast West Belfast Highfield -10 -8 2 Belfast East Belfast Island -17 -16 2 Belfast East Belfast Knock -9 -8 1 Belfast West Belfast Ladybrook -16 -11 5 Belfast North Belfast Legoniel -16 -14 3 Belfast South Belfast Malone -15 -12 4 Belfast South Belfast Musgrave -15 -13 2 Belfast North Belfast New Lodge -18 -13 6 Belfast East Belfast Orangefield -10 -8 2 Belfast South Belfast Ravenhill -13 -10 4 Belfast South Belfast Rosetta -10 -8 2 Belfast South Belfast Shaftesbury -19 -16 5 Belfast West Belfast Shankill -11 -10 2 Belfast East Belfast Stormont -10 -8 1 Belfast South Belfast Stranmillis -14 -11 3 Belfast East Belfast Sydenham -14 -11 3 Belfast East Belfast The Mount -13 -12 1 Belfast South Belfast Upper Malone -13 -11 2 Upper -18 -12 7 Belfast West Belfast Springfield Belfast North Belfast Water Works -21 -18 4 Belfast West Belfast Whiterock -24 -13 14 Belfast South Belfast Windsor -20 -16 4 Belfast South Belfast Woodstock -20 -18 2 Belfast North Belfast Woodvale -15 -14 2 East Antrim Carrickfergus Blackhead -11 -8 4 East Antrim Carrickfergus Bluefield -12 -10 2 East Antrim Carrickfergus Boneybefore -5 -4 1 East Antrim Carrickfergus Burleigh Hill -11 -9 3 East Antrim Carrickfergus Clipperstown -14 -12 2 East Antrim Carrickfergus Eden -9 -7 2

PricewaterhouseCoopers July 2003 Annexes - Page 49 Electoral Commission Desk Research: First Report

Table C.2 The change in the number of persons on the Electoral Register by Census Ward

% change in number registered: Parliamentary District Council Ward From August 2002 to: constituency December 2002 to May 2003 December 2002 May 2003

% % % East Antrim Carrickfergus Gortalee -10 -8 1 East Antrim Carrickfergus Greenisland -6 -5 1 East Antrim Carrickfergus Killycrot -11 -9 2 East Antrim Carrickfergus Knockagh -10 -9 2 East Antrim Carrickfergus Love Lane -13 -12 1 East Antrim Carrickfergus Milebush -12 -10 2 East Antrim Carrickfergus Northland -19 -17 3 East Antrim Carrickfergus Sunnylands -10 -9 1 East Antrim Carrickfergus Victoria -10 -9 2 East Antrim Carrickfergus Whitehead -14 -13 1 East Antrim Carrickfergus Woodburn -10 -9 1 Strangford Castlereagh Ballyhanwood -10 -8 2 Belfast South Castlereagh Beechill -13 -11 3 Belfast South Castlereagh Cairnshill -12 -10 3 Strangford Castlereagh Carrowreagh -15 -13 1 Strangford Castlereagh Carryduff East -9 -9 1 Strangford Castlereagh Carryduff West -10 -9 2 Belfast East Castlereagh Cregagh -9 -9 0 Belfast East Castlereagh Downshire -11 -10 1 Strangford Castlereagh Dundonald -10 -9 1 Strangford Castlereagh Enler -14 -14 1 Belfast South Castlereagh Galwally -13 -10 4 Belfast East Castlereagh Gilnahirk -8 -7 1 Graham's -7 -7 0 Strangford Castlereagh Bridge Belfast East Castlereagh Hillfoot -9 -7 1 Belfast South Castlereagh Knockbracken -7 -5 2 Belfast East Castlereagh Lisnasharragh -8 -7 1 Belfast East Castlereagh Lower Braniel -10 -7 2 Belfast South Castlereagh Minnowburn -7 -6 2 Strangford Castlereagh Moneyreagh -8 -7 1

PricewaterhouseCoopers July 2003 Annexes - Page 50 Electoral Commission Desk Research: First Report

Table C.2 The change in the number of persons on the Electoral Register by Census Ward

% change in number registered: Parliamentary District Council Ward From August 2002 to: constituency December 2002 to May 2003 December 2002 May 2003

% % % Belfast South Castlereagh Newtownbreda -13 -11 2 Belfast East Castlereagh Tullycarnet -14 -12 2 Belfast East Castlereagh Upper Braniel -8 -7 1 Belfast East Castlereagh Wynchurch -12 -10 2 East -5 -3 1 Coleraine Agivey Londonderry East -13 -12 1 Coleraine Atlantic Londonderry East -17 -16 1 Coleraine Ballysally Londonderry East -6 -6 0 Coleraine Castlerock Londonderry East -10 -10 -0 Coleraine Central Londonderry East -15 -14 1 Coleraine Churchland Londonderry East -16 -15 2 Coleraine Cross Glebe Londonderry East -7 -7 0 Coleraine Dundooan Londonderry East -8 -6 1 Coleraine Dunluce Londonderry East -6 -5 1 Coleraine Garvagh Londonderry East -7 -6 1 Coleraine Hopefield Londonderry East -8 -6 2 Coleraine Kilrea Londonderry East -5 -3 2 Coleraine Knocklynn Londonderry East -7 -5 1 Coleraine Macosquin Londonderry East -11 -10 1 Coleraine Mount Sandel Londonderry East -13 -12 1 Coleraine Portstewart Londonderry East -2 -1 1 Coleraine Ringsend Londonderry East -12 -11 1 Coleraine Royal Portrush Londonderry East -11 -9 2 Coleraine Strand Londonderry East -10 -9 1 Coleraine The Cuts Londonderry East -9 -8 1 Coleraine University Londonderry

PricewaterhouseCoopers July 2003 Annexes - Page 51 Electoral Commission Desk Research: First Report

Table C.2 The change in the number of persons on the Electoral Register by Census Ward

% change in number registered: Parliamentary District Council Ward From August 2002 to: constituency December 2002 to May 2003 December 2002 May 2003

% % % East -4 -2 2 Coleraine Waterside Londonderry Mid Ulster Cookstown Ardboe -5 -3 2 Mid Ulster Cookstown Coagh -5 -3 2 Mid Ulster Cookstown Dunnamore -5 -3 2 Mid Ulster Cookstown Gortalowry -8 -6 2 Mid Ulster Cookstown Killycolpy -8 -6 2 Mid Ulster Cookstown Killymoon -12 -9 4 Mid Ulster Cookstown Lissan -1 -0 1 Mid Ulster Cookstown Moneymore -2 -0 1 Mid Ulster Cookstown Newbuildings -9 -7 2 Mid Ulster Cookstown Oaklands -3 -2 1 Mid Ulster Cookstown Oldtown -6 -4 2 Mid Ulster Cookstown Pomeroy -7 -4 3 Mid Ulster Cookstown Sandholes -3 -1 2 Mid Ulster Cookstown Stewartstown -7 -5 3 Mid Ulster Cookstown The Loop -4 -3 1 Mid Ulster Cookstown Tullagh -6 -4 2 Upper Bann Craigavon Aghagallon -8 -4 4 Upper Bann Craigavon Annagh -6 -4 2 Upper Bann Craigavon Ballybay -10 -8 2 Upper Bann Craigavon Ballyoran -7 -5 2 Upper Bann Craigavon Bleary -5 -3 2 Upper Bann Craigavon Brownstown -7 -6 1 Upper Bann Craigavon Church -9 -8 1 Upper Bann Craigavon Corcrain -14 -11 3 Upper Bann Craigavon Court -10 -8 2 Upper Bann Craigavon Derrytrasna -4 -2 3 Upper Bann Craigavon Donaghcloney -7 -5 2 Upper Bann Craigavon Drumgask -10 -8 3 Upper Bann Craigavon Drumgor -17 -14 3

PricewaterhouseCoopers July 2003 Annexes - Page 52 Electoral Commission Desk Research: First Report

Table C.2 The change in the number of persons on the Electoral Register by Census Ward

% change in number registered: Parliamentary District Council Ward From August 2002 to: constituency December 2002 to May 2003 December 2002 May 2003

% % % Upper Bann Craigavon Drumnamoe -14 -11 4 Upper Bann Craigavon Edenderry -9 -7 2 Upper Bann Craigavon Kernan -4 -2 2 Upper Bann Craigavon Killycomain -7 -7 0 Upper Bann Craigavon Knocknashane -7 -5 2 Upper Bann Craigavon Magheralin -2 -2 1 Upper Bann Craigavon Mourneview -9 -9 1 Upper Bann Craigavon Parklake -6 -5 2 Upper Bann Craigavon Taghnevan -5 -2 3 Upper Bann Craigavon Tavanagh -7 -8 -1 Upper Bann Craigavon The Birches -4 -3 1 Upper Bann Craigavon Waringstown -5 -4 2 Upper Bann Craigavon Woodville -8 -6 2 Foyle Derry Altnagelvin -13 -11 3 Foyle Derry Ballynashallog -11 -9 3 Foyle Derry Banagher -3 -2 1 Foyle Derry Beechwood -14 -9 6 Foyle Derry Brandywell -13 -8 6 Foyle Derry Carn Hill -16 -12 5 Foyle Derry Caw -8 -4 4 Foyle Derry Clondermot -7 -4 3 Foyle Derry Claudy -7 -5 2 Foyle Derry Creggan Central -17 -9 9 Foyle Derry Creggan South -11 -7 5 Foyle Derry Crevagh -15 -8 8 Foyle Derry Culmore -15 -9 7 Foyle Derry Ebrington -6 -5 2 Foyle Derry Eglinton -7 -5 1 Foyle Derry Enagh -7 -5 2 Foyle Derry Foyle Springs -12 -8 5

PricewaterhouseCoopers July 2003 Annexes - Page 53 Electoral Commission Desk Research: First Report

Table C.2 The change in the number of persons on the Electoral Register by Census Ward

% change in number registered: Parliamentary District Council Ward From August 2002 to: constituency December 2002 to May 2003 December 2002 May 2003

% % % Foyle Derry Holly Mount -8 -5 3 Foyle Derry Kilfennan -12 -11 2 Foyle Derry Lisnagelvin -9 -7 1 Foyle Derry New Buildings -8 -7 1 Foyle Derry Pennyburn -12 -10 3 Foyle Derry Rosemount -13 -9 4 Foyle Derry Shantallow East -11 -8 3 Shantallow -17 -14 5 Foyle Derry West Foyle Derry Springtown -15 -10 6 Foyle Derry Strand -12 -8 5 Foyle Derry The Diamond -11 -9 2 Foyle Derry Victoria -11 -10 2 Foyle Derry Westland -15 -10 6 South Down Down Ardglass -12 -9 3 South Down Down Audley's Acre -10 -8 2 South Down Down Ballymaglave -9 -7 3 South Down Down Ballymote -10 -6 4 Ballynahinch -14 -12 3 South Down Down East South Down Down Castlewellan -10 -7 3 South Down Down Cathedral -10 -8 2 South Down Down Crossgar -5 -3 2 Strangford Down Derryboy -10 -8 2 South Down Down Donard -7 -5 2 South Down Down Drumaness -5 -3 2 South Down Down Dundrum -5 -3 2 South Down Down Dunmore -5 -4 2 South Down Down Killough -9 -5 4 Strangford Down Killyleagh -6 -5 0 South Down Down Kilmore -8 -7 1 South Down Down Murlough -9 -8 1

PricewaterhouseCoopers July 2003 Annexes - Page 54 Electoral Commission Desk Research: First Report

Table C.2 The change in the number of persons on the Electoral Register by Census Ward

% change in number registered: Parliamentary District Council Ward From August 2002 to: constituency December 2002 to May 2003 December 2002 May 2003

% % % South Down Down Quoile -9 -8 2 Strangford Down Saintfield -9 -8 1 South Down Down Seaforde -6 -5 1 South Down Down Shimna -8 -6 2 South Down Down Strangford -7 -5 2 South Down Down Tollymore -6 -3 3 Mid Ulster Dungannon Altmore -3 -1 2 Fermanagh & -5 -3 2 Dungannon Augher South Tyrone Fermanagh & -5 -4 2 Dungannon Aughnacloy South Tyrone Fermanagh & -9 -7 2 Dungannon Ballygawley South Tyrone Fermanagh & -12 -8 6 Dungannon Ballysaggart South Tyrone Fermanagh & -6 -3 2 Dungannon Benburb South Tyrone Fermanagh & -5 -4 1 Dungannon Caledon South Tyrone Fermanagh & -7 -4 4 Dungannon Castlecaulfield South Tyrone Fermanagh & -3 -3 0 Dungannon Clogher South Tyrone Mid Ulster Dungannon Coalisland North -9 -6 3 Coalisland -9 -5 4 Mid Ulster Dungannon South Coalisland West -8 -6 2 Mid Ulster Dungannon and Newmills Fermanagh & -9 -8 1 Dungannon Coolhill South Tyrone Mid Ulster Dungannon Donaghmore -9 -6 2 Fermanagh & -16 -13 3 Dungannon Drumglass South Tyrone Fermanagh & -6 -4 2 Dungannon Fivemiletown South Tyrone Fermanagh & -4 -3 1 Dungannon Killyman South Tyrone Fermanagh & -9 -7 2 Dungannon Killymeal South Tyrone Fermanagh & -8 -7 1 Dungannon Moy South Tyrone Fermanagh & -6 -4 2 Dungannon Moygashel South Tyrone

PricewaterhouseCoopers July 2003 Annexes - Page 55 Electoral Commission Desk Research: First Report

Table C.2 The change in the number of persons on the Electoral Register by Census Ward

% change in number registered: Parliamentary District Council Ward From August 2002 to: constituency December 2002 to May 2003 December 2002 May 2003

% % % Fermanagh & -11 -7 4 Dungannon Mullaghmore South Tyrone Mid Ulster Dungannon Washing Bay -5 -3 2 Fermanagh & -6 -4 2 Fermanagh Ballinamallard South Tyrone Fermanagh & Belcoo and -10 -5 5 Fermanagh South Tyrone Garrison Fermanagh & -7 -3 4 Fermanagh Belleek and Boa South Tyrone Fermanagh & Boho Cleenish -5 -4 2 Fermanagh South Tyrone and Letterbreen Fermanagh & -4 -3 1 Fermanagh Brookeborough South Tyrone Fermanagh & -8 -5 2 Fermanagh Castlecoole South Tyrone Fermanagh & -7 -5 1 Fermanagh Derrygonnelly South Tyrone Fermanagh & -4 -2 3 Fermanagh Derrylin South Tyrone Fermanagh & -9 -9 0 Fermanagh Devenish South Tyrone Fermanagh & -5 -3 3 Fermanagh Donagh South Tyrone Fermanagh & -8 -7 1 Fermanagh Erne South Tyrone Fermanagh & Florence Court -5 -3 3 Fermanagh South Tyrone and Kinawley Fermanagh & -10 -7 3 Fermanagh Irvinestown South Tyrone Fermanagh & Kesh Ederney -7 -5 3 Fermanagh South Tyrone and Lack Fermanagh & -5 -3 2 Fermanagh Lisbellaw South Tyrone Fermanagh & -2 -1 1 Fermanagh Lisnarrick South Tyrone Fermanagh & -5 -2 4 Fermanagh Lisnaskea South Tyrone Fermanagh & -2 -1 1 Fermanagh Maguires Bridge South Tyrone Fermanagh & -9 -5 4 Fermanagh Newtownbutler South Tyrone Fermanagh & -10 -9 1 Fermanagh Portora South Tyrone Fermanagh & -5 -1 3 Fermanagh Rosslea South Tyrone Fermanagh & -8 -5 3 Fermanagh Rossorry South Tyrone

PricewaterhouseCoopers July 2003 Annexes - Page 56 Electoral Commission Desk Research: First Report

Table C.2 The change in the number of persons on the Electoral Register by Census Ward

% change in number registered: Parliamentary District Council Ward From August 2002 to: constituency December 2002 to May 2003 December 2002 May 2003

% % % Fermanagh & -1 -0 1 Fermanagh Tempo South Tyrone East Antrim Larne Antiville -16 -15 1 East Antrim Larne Ballycarry -9 -7 2 East Antrim Larne Ballyloran -17 -16 1 East Antrim Larne Blackcave -15 -14 1 East Antrim Larne Carncastle -9 -8 1 East Antrim Larne Carnlough -5 -5 1 East Antrim Larne Central -11 -10 2 East Antrim Larne Craigy Hill -13 -12 1 East Antrim Larne Gardenmore -10 -8 2 East Antrim Larne Glenarm -9 -7 2 East Antrim Larne Glynn -9 -8 1 East Antrim Larne Harbour -7 -7 0 East Antrim Larne Island Magee -11 -9 2 East Antrim Larne Kilwaughter -5 -4 2 East Antrim Larne Town Parks -12 -11 1 East -3 -1 2 Limavady Aghanloo Londonderry East -8 -7 2 Limavady Ballykelly Londonderry East -6 -7 -0 Limavady Coolessan Londonderry East -9 -7 2 Limavady Dungiven Londonderry East -10 -7 2 Limavady Enagh Londonderry East -5 -4 2 Limavady Feeny Londonderry East -9 -7 2 Limavady Forest Londonderry East -5 -4 1 Limavady Glack Londonderry East -7 -5 2 Limavady Gresteel Londonderry East -14 -14 -0 Limavady Greystone Londonderry East -8 -7 1 Limavady Magilligan Londonderry

PricewaterhouseCoopers July 2003 Annexes - Page 57 Electoral Commission Desk Research: First Report

Table C.2 The change in the number of persons on the Electoral Register by Census Ward

% change in number registered: Parliamentary District Council Ward From August 2002 to: constituency December 2002 to May 2003 December 2002 May 2003

% % % East -9 -8 2 Limavady Rathbrady Londonderry East -10 -8 2 Limavady Roeside Londonderry East -4 -2 2 Limavady The Highlands Londonderry East Upper -7 -5 2 Limavady Londonderry Glenshane Lagan Valley Lisburn Ballinderry -8 -6 2 Lagan Valley Lisburn Ballymacash -9 -7 3 -7 -5 3 Lagan Valley Lisburn Ballymacbrenna n Lagan Valley Lisburn Ballymacoss -11 -8 4 Lagan Valley Lisburn Blaris -7 -6 1 Belfast West Lisburn Collin Glen -22 -17 6 Lagan Valley Lisburn Derryaghy -8 1 10 Lagan Valley Lisburn Dromara -6 -3 2 Lagan Valley Lisburn Drumbo -8 -6 2 Lagan Valley Lisburn Dunmurry -16 -11 6 Lagan Valley Lisburn Glenavy -11 -8 4 Lagan Valley Lisburn Harmony Hill -7 -5 2 Lagan Valley Lisburn Hilden -13 -10 4 Lagan Valley Lisburn Hillhall -12 -10 2 Lagan Valley Lisburn Hillsborough -6 -3 3 Belfast West Lisburn Kilwee -20 -17 4 Lagan Valley Lisburn Knockmore -11 -10 2 Lagan Valley Lisburn Lagan Valley -12 -11 2 Lagan Valley Lisburn Lambeg -13 -12 2 Lagan Valley Lisburn Lisnagarvey -8 -6 2 Lagan Valley Lisburn Maghaberry -7 -4 4 Lagan Valley Lisburn Magheralave -11 -8 3 Lagan Valley Lisburn Maze -10 -6 4 Lagan Valley Lisburn Moira -8 -5 3

PricewaterhouseCoopers July 2003 Annexes - Page 58 Electoral Commission Desk Research: First Report

Table C.2 The change in the number of persons on the Electoral Register by Census Ward

% change in number registered: Parliamentary District Council Ward From August 2002 to: constituency December 2002 to May 2003 December 2002 May 2003

% % % Lagan Valley Lisburn Old Warren -15 -13 3 Belfast West Lisburn Poleglass -23 -19 5 Lagan Valley Lisburn Seymour Hill -9 -6 3 Lagan Valley Lisburn Tonagh -15 -13 2 Belfast West Lisburn Twinbrook -24 -20 5 Lagan Valley Lisburn Wallace Park -6 -5 2 Mid Ulster Magherafelt Ballymaguigan -5 -2 3 Mid Ulster Magherafelt Bellaghy -5 1 6 Mid Ulster Magherafelt Castledawson -6 -5 1 Mid Ulster Magherafelt Draperstown -2 1 3 Mid Ulster Magherafelt Glebe -6 -6 1 Mid Ulster Magherafelt Gulladuff -5 -3 1 Mid Ulster Magherafelt Knockcloghrim -6 -4 3 Mid Ulster Magherafelt Lecumpher 0 1 1 Lower -5 -2 3 Mid Ulster Magherafelt Glenshane Mid Ulster Magherafelt Maghera -6 -3 4 Mid Ulster Magherafelt Swatragh -3 -1 3 Mid Ulster Magherafelt Tobermore -2 -0 2 Town Parks -8 -7 1 Mid Ulster Magherafelt East Town Parks -7 -5 3 Mid Ulster Magherafelt West Mid Ulster Magherafelt Upperlands -1 -0 1 Mid Ulster Magherafelt Valley -6 -4 3 North Antrim Moyle Armoy -9 -8 1 North Antrim Moyle Ballylough -5 -4 2 Bonamargy and -9 -8 2 North Antrim Moyle Rathlin North Antrim Moyle Bushmills -15 -14 1 North Antrim Moyle Carnmoon -4 -4 0 North Antrim Moyle Dalriada -10 -8 2 North Antrim Moyle Dunseverick -7 -6 1

PricewaterhouseCoopers July 2003 Annexes - Page 59 Electoral Commission Desk Research: First Report

Table C.2 The change in the number of persons on the Electoral Register by Census Ward

% change in number registered: Parliamentary District Council Ward From August 2002 to: constituency December 2002 to May 2003 December 2002 May 2003

% % % North Antrim Moyle Glenaan -3 -2 1 North Antrim Moyle Glenariff -6 -4 3 North Antrim Moyle Glendun -11 -8 3 North Antrim Moyle Glenshesk -4 -4 1 North Antrim Moyle Glentaisie -7 -5 2 North Antrim Moyle Kinbane -14 -12 2 North Antrim Moyle Knocklayd -14 -10 4 Moss-side and -2 -3 -0 North Antrim Moyle Moyarget Newry and -8 -6 3 South Down Annalong Mourne Newry and -15 -14 2 Newry & Armagh Ballybot Mourne Newry and -8 -7 1 Newry & Armagh Bessbrook Mourne Newry and -9 -5 5 South Down Binnian Mourne Newry and Burren and -17 -8 10 South Down Mourne Kilbroney Newry and -14 -7 8 Newry & Armagh Camlough Mourne Newry and -7 -5 2 South Down Clonallan Mourne Newry and -6 -2 4 Newry & Armagh Creggan Mourne Newry and -12 -7 6 Newry & Armagh Crossmaglen Mourne Newry and -15 -11 4 Newry & Armagh Daisy hill Mourne Newry and -12 -6 6 South Down Derryleckagh Mourne Newry and -11 -7 4 Newry & Armagh Derrymore Mourne Newry and -6 -4 3 South Down Donaghmore Mourne Newry and -9 -6 3 Newry & Armagh Drumalane Mourne Newry and -15 -11 4 Newry & Armagh Drumgullion Mourne Newry and -9 -7 2 Newry & Armagh Fathom Mourne Newry and -9 -5 4 Newry & Armagh Forkhill Mourne

PricewaterhouseCoopers July 2003 Annexes - Page 60 Electoral Commission Desk Research: First Report

Table C.2 The change in the number of persons on the Electoral Register by Census Ward

% change in number registered: Parliamentary District Council Ward From August 2002 to: constituency December 2002 to May 2003 December 2002 May 2003

% % % Newry and -9 -7 2 South Down Kilkeel Central Mourne Newry and -7 -7 1 South Down Kilkeel South Mourne Newry and -7 -6 1 South Down Lisnacree Mourne Newry and -7 -3 4 South Down Mayobridge Mourne -8 -6 3 Newry and Newry & Armagh Newtownhamilto Mourne n Newry and -7 -5 2 South Down Rostrevor Mourne Newry and -15 -12 4 South Down Seaview Mourne Newry and -12 -7 6 Newry & Armagh Silver Bridge Mourne Newry and -8 -5 3 South Down Spelga Mourne Newry and -15 -12 3 Newry & Armagh St Mary's Mourne Newry and -12 -10 3 Newry & Armagh St Patrick's Mourne Newry and -5 -4 2 Newry & Armagh Tullyhappy Mourne Newry and -13 -11 3 Newry & Armagh Windsor Hill Mourne Belfast North Newtownabbey Abbey -16 -13 3 South Antrim Newtownabbey Ballyclare North -14 -11 3 South Antrim Newtownabbey Ballyclare South -11 -10 2 South Antrim Newtownabbey Ballyduff -13 -11 2 South Antrim Newtownabbey Ballyhenry -10 -8 1 South Antrim Newtownabbey Ballynure -9 -8 2 South Antrim Newtownabbey Ballyrobert -7 -6 1 South Antrim Newtownabbey Burnthill -11 -10 0 South Antrim Newtownabbey Carnmoney -14 -13 1 East Antrim Newtownabbey Cloughfern -14 -12 2 South Antrim Newtownabbey Collinbridge -12 -9 3 Belfast North Newtownabbey Coole -21 -19 2 South Antrim Newtownabbey Doagh -10 -8 2

PricewaterhouseCoopers July 2003 Annexes - Page 61 Electoral Commission Desk Research: First Report

Table C.2 The change in the number of persons on the Electoral Register by Census Ward

% change in number registered: Parliamentary District Council Ward From August 2002 to: constituency December 2002 to May 2003 December 2002 May 2003

% % % Belfast North Newtownabbey Dunanney -20 -17 4 South Antrim Newtownabbey Glebe -11 -10 1 South Antrim Newtownabbey Glengormley -13 -12 2 South Antrim Newtownabbey Hawthorne -12 -11 1 South Antrim Newtownabbey Hightown -11 -9 2 East Antrim Newtownabbey Jordanstown -6 -4 2 South Antrim Newtownabbey Mallusk -7 -5 2 East Antrim Newtownabbey Monkstown -12 -11 2 South Antrim Newtownabbey Mossley -11 -9 1 East Antrim Newtownabbey Rostulla -10 -7 2 Belfast North Newtownabbey Valley -16 -14 3 Belfast North Newtownabbey Whitehouse -11 -10 1 North Down North Down Ballycrochan -13 -11 2 North Down North Down Ballyholme -11 -9 3 North Down North Down Ballymaconnell -10 -10 0 North Down North Down Ballymagee -8 -7 1 North Down North Down Bangor Castle -8 -9 -0 North Down North Down Bloomfield -11 -9 2 North Down North Down Broadway -7 -6 1 North Down North Down Bryansburn -10 -9 1 North Down North Down Churchill -10 -9 1 North Down North Down Clandeboye -14 -11 3 North Down North Down Conlig -8 -5 3 North Down North Down Craigavad -10 -9 2 North Down North Down Crawfordsburn -7 -6 1 North Down North Down Cultra -13 -11 2 North Down North Down Dufferin -13 -11 3 North Down North Down Groomsport -5 -4 2 North Down North Down Harbour -20 -18 3 Holywood -12 -10 2 North Down North Down Demesne

PricewaterhouseCoopers July 2003 Annexes - Page 62 Electoral Commission Desk Research: First Report

Table C.2 The change in the number of persons on the Electoral Register by Census Ward

% change in number registered: Parliamentary District Council Ward From August 2002 to: constituency December 2002 to May 2003 December 2002 May 2003

% % % North Down North Down Holywood Priory -7 -6 2 North Down North Down Loughview -14 -13 1 North Down North Down Princetown -9 -8 0 North Down North Down Rathgael -12 -10 2 North Down North Down Silverstream -14 -14 1 North Down North Down Springhill -9 -8 1 North Down North Down Whitehill -16 -14 2 West Tyrone Omagh Beragh -7 -7 1 West Tyrone Omagh Camowen -10 -9 1 West Tyrone Omagh Clanabogan -4 -2 2 West Tyrone Omagh Coolnagard -7 -3 3 West Tyrone Omagh Dergmoney -6 -6 0 West Tyrone Omagh Dromore -5 -2 2 West Tyrone Omagh Drumnakilly -9 -5 5 West Tyrone Omagh Drumquin -4 -2 2 West Tyrone Omagh Drumragh -4 -3 1 West Tyrone Omagh Fairy Water -3 -3 0 West Tyrone Omagh Fintona -10 -5 5 West Tyrone Omagh Gortin -6 -4 2 West Tyrone Omagh Gortrush -15 -13 3 West Tyrone Omagh Killyclogher -8 -5 3 West Tyrone Omagh Lisanelly -13 -10 4 West Tyrone Omagh Newtownsaville -4 -2 2 West Tyrone Omagh Owenkillew -2 2 3 West Tyrone Omagh Sixmilecross -8 -5 3 West Tyrone Omagh Strule -5 -4 1 West Tyrone Omagh Termon -1 2 3 West Tyrone Omagh Trillick -3 -1 2 West Tyrone Strabane Artigarvan -7 -4 3 West Tyrone Strabane Ballycolman -11 -8 3

PricewaterhouseCoopers July 2003 Annexes - Page 63 Electoral Commission Desk Research: First Report

Table C.2 The change in the number of persons on the Electoral Register by Census Ward

% change in number registered: Parliamentary District Council Ward From August 2002 to: constituency December 2002 to May 2003 December 2002 May 2003

% % % West Tyrone Strabane Castlederg -10 -8 2 West Tyrone Strabane Clare -6 -4 3 West Tyrone Strabane Dunnamanagh -4 -2 2 West Tyrone Strabane East -11 -10 1 West Tyrone Strabane Finn -7 -6 1 West Tyrone Strabane Glenderg -5 -4 1 West Tyrone Strabane Newtownstewart -7 -5 3 West Tyrone Strabane North -12 -6 7 West Tyrone Strabane Plumbridge -6 -3 4 West Tyrone Strabane Sion Mills -10 -9 2 West Tyrone Strabane Slievekirk -7 -6 1.1 West Tyrone Strabane South -12 -8 3.8 West Tyrone Strabane Victoria Bridge -8 -7 1.2 West Tyrone Strabane West -7 -7 0.6

PricewaterhouseCoopers July 2003 Annexes - Page 64