Voting at the Northern Ireland Assembly Election, 2003.

Transcription

1 Voting at the Northern Ireland Assembly Election, Colin Rallings and Michael Thrasher Local Government Chronicle Elections Centre University of Plymouth This report for the Electoral Commission considers the second election to the Northern Ireland Assembly (NIA) held in November The election uses the 18 Westminster Parliamentary constituency areas for voting purposes but six Assembly members are elected for each of these constituencies using the Single Transferable Vote (STV) method of voting. Election results are those supplied by the Chief Electoral Officer for Northern Ireland. Those interested in the details of transfers at both the 1998 and 2003 should consult the relevant sheets published by the Chief Electoral Officer. The report is divided into two sections. The first section examines the elections in terms of support for parties and candidates and the distribution of seats, and is structured as follows: Distribution of votes and seats across Northern Ireland Proportionality and the electoral outcome Comparison of vote and seat shares, Parties and candidates Women candidates Incumbents seeking re-election Candidates and the electoral system Party competition Party quotas and the allocation of seats The personal vote A second section focuses upon electoral registration, methods of voting and voter turnout and is structured as follows: The impact of new electoral registration procedures Change in electorates, Methods for calculating electoral turnout Turnout in 2003 Changes in turnout, Constituency characteristics and election turnout Postal voting and turnout Rejected postal votes and turnout In-person voting Rejected in-person votes and turnout Reasons for rejecting votes Comparing rejected votes Some Tables are included in the body of the report but others are attached in the Appendix and are labelled with the prefix A. Party abbreviations are mainly used in the text but a full list may be found in the Appendix, Table A15.

2 Section 1: Support for candidates and parties and the distribution of seats Distribution of votes and seats across Northern Ireland Slightly fewer than 0.7 million valid first preference votes were cast in the Northern Ireland Assembly (NIA) election of 2003 (Table 1; but see also Table A1 for the constituency results and Tables A2-A3 for summary data for each party in each constituency). This is approximately 120,000 fewer votes than were recorded at the first NIA election in The most popular choice was the Democratic Unionist Party (DUP), whose candidates received more than one in four first preference votes. The second largest party in vote share was Sinn Fein (SF) with 23.5% of the vote. It finished fifteen thousand votes behind the DUP. Overall, fewer than six thousand votes separated SF and the Ulster Unionist Party (UUP), which received 22.7% of first preferences. The fourth largest party was the Social and Democratic Labour Party (SDLP), which polled 17% of first preference votes. The four main parties (DUP, UUP, SDLP and SF) were separated by a mere sixty thousand votes, with less than a nine percentage point gap separating the DUP and SDLP. Combined, these four parties captured 88.9% of the vote, compared with the 79% share that they obtained in The difference between the two main unionist parties, at just over twenty one thousand votes, was half that separating SF and SDLP. Votes for all unionist parties combined comprise 50.9% of the total (49.9% in 1998) while the combined share for SF and SDLP is 40.5% (39.6% in 1998). Table 1: Result of the Northern Ireland Assembly Elections, st pref votes Share % Cands (N=) Seats Seat % A 25, Con 1, DUP 177, Green 2, Ind 19, NIUP 1, NIWC 5, PUP 8, SDLP 117, SEA 2, SF 162, SP UKUP 5, UTW UUC 2, UUP 156, VFY WP 1, Total 692,

3 The largest number of Assembly seats was won by DUP, whose 30 seats comprise 27.8% of the total (Table 2 and Table A4). The second largest party in the new Assembly is UUP, which won three fewer seats than the DUP. Although SF received a larger percentage of first preferences than did UUP, the party won three fewer seats, and six fewer than DUP. The fourth largest party, in terms of both votes and seats is SDLP, which now has 18 Assembly members. Table 2: Seats by party at the Northern Ireland Assembly Election, change seats seats seats A DUP Ind NIWC PUP SDLP SF UKUP UUP Others The most successful party, measured by the ratio of successful to unsuccessful candidates, was DUP. Three-quarters (0.75) of its candidates were elected compared to 0.6 for both UUP and SF and 0.5 for SDLP. The four main parties returned 99 of the Assembly s 108 members, 91.7% of the membership. Among the smaller parties, the Alliance party performed best. It won six seats, slightly less than one in three fought by the party. Proportionality and the electoral outcome The electoral system, the Single Transferable Vote (STV), helped to ensure that there was a strong correlation between vote and seat shares. The largest party in vote share, DUP, also won the largest share of seats. This had not happened in Then, SDLP won the most first preference votes with UUP in second place, but these positions were reversed when seats were allocated. The combined vote share of minor parties, which did not win a seat and excluding Independents, was just 2.7%. One useful method for assessing the operation of the electoral system is to measure the ratio between votes and seats for each party. When that ratio equals one, a party s seat and vote shares are identical; when the ratio is above one then the party is relatively advantaged by the system, and when it falls below that figure it is disadvantaged. The seats to votes ratios for the four main parties were very close in Both the DUP and UUP had ratios of 1.1, but each of these parties benefited from transfers from among other unionist candidates. Although both SF and SDLP had seats/votes ratios 3

4 below one (0.94 and 0.98 respectively) the difference from equality was negligible. The Alliance party was the biggest beneficiary of the electoral system. Although it won 3.7% of first preference votes the party won six seats, 5.6% of the total. This is a seats to votes ratio of 1.5. A second useful measure for assessing the impact of the electoral system upon the Assembly s composition is to use an index of proportionality. The most commonly used measure of proportionality is referred to as the Loosemore-Hanby index. The index is calculated as follows: for any election the absolute values of the differences between vote and seat shares for each competing party are summed, with that total then divided by two. Thus, if every party s seat share is an exact reflection of its vote share then the value of the index would be zero; the outcome would be proportional. As a general rule of thumb an election result that has a Loosemore-Hanby index of above 10 is regarded as having an outcome that is disproportional. In the case of the 2003 Assembly result the calculated index is 6.4. This compares with an index score of 6.1 for the 1998 election outcome. For comparative purposes, the Loosemore-Hanby index scores for the last two UK general elections were 21.1 and 21.9 respectively. Comparison of vote and seat shares, A comparison of vote shares with 1998 shows that the two parties that advanced most were DUP and SF. The vote of the former rose by 7.6 percentage points whilst there was a 5.9 point rise for SF. Although the overall result for UUP was considered by some to be disappointing the party s vote share rose, albeit by a modest 1.4 percentage points. The biggest loss of vote share was that for the SDLP. Its vote, measured as a share of first preferences, declined by five percentage points. There was a drop in support also for UKUP, from 4.5% in 1998 to just 0.8% in 2003, although the party fielded half the number of candidates than it had done so in Another party in decline was Alliance whose vote fell by 2.8 percentage points. That said, Alliance did succeed in retaining all of its seats an achievement given the size of its vote. Undoubtedly, this outcome followed the receipt of transfer votes from those whose first preference was for another party. The DUP increased its share of seats by half (from 20 to 30) whilst SF s share increased by a third (from 18 to 24). The biggest loser amongst the major parties, in absolute terms, was SDLP, which lost a quarter of its Assembly strength (from 24 to 18). Despite the modest rise in support for UUP it was unable to prevent the loss of one of its seats (from 28 to 27). The fewer number of candidates and decline in support for UKUP contributed to the loss of four of its five seats. The two seats won in 1998 by the Northern Ireland Women s Coalition were both lost as was one of the two seats won by the Progressive Unionist Party. 4

5 Parties and candidates A total of 256 candidates contested the election an average of 14 candidates competing for the six seats in each constituency (Table A5). This was forty fewer candidates than had contested the election in The largest number of candidates was 19, in both East Antrim and North Down constituencies, a ratio of candidates to seats of more than 3:1 (Table 3). The fewest number of candidates was in Newry and Armagh, where just eleven competed for six seats, a ratio of slightly more than 2:1. In 1998 the largest number of candidates, 22, fought in Strangford, whilst the minimum was the 13 candidates that competed in Mid Ulster. Table 3: Candidates at the Northern Ireland Assembly Election, 2003 Constituency Total East Antrim 19 North Down 19 Belfast South 17 Belfast North 16 South Down 16 Belfast East 15 Belfast West 14 East Londonderry 14 South Antrim 14 Upper Bann 14 Foyle 13 Lagan Valley 13 Strangford 13 Fermanagh & South Tyrone 12 Mid Ulster 12 North Antrim 12 West Tyrone 12 Newry & Armagh 11 Total 256 The largest number of candidates, 43, stood on behalf of UUP, followed closely by DUP, which put 40 candidates into the field. SF and SDLP fielded 38 and 36 candidates respectively. The only other parties to have double digit numbers of candidates were Alliance (21) and the Progressive Unionist Party (11). A total of 22 candidates stood as Independents, with the greatest concentrations, four in total, in both East Antrim and North Down. 5

6 Women candidates There were 49 women candidates who fought the election, 19% of the total (Table A6). Women stood in all constituencies except West Tyrone, although in four cases (Belfast East, North Antrim, South Antrim and Strangford) there was just a single woman candidate competing (Table 4). At the other end of the range there were three constituencies where women candidates were more than 30% of the total. Foyle and Lagan Valley each had four women out of 13 candidates standing, whilst in Fermanagh and South Tyrone exactly a third of the candidates were women. Table 4: Candidate Characteristics at the Northern Ireland Assembly Election, 2003 Constituency female % male % incumbents re-elected defeated Belfast East Belfast North Belfast South Belfast West 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 Total Examining the number of women candidates standing for each party reveals some large differences (Table 5). DUP, for example, had 40 candidates but only four of these were women. Similarly, UUP also fielded four women candidates, 9.3% of its overall figure. There were six women out of 36 candidates (16.7%) that challenged on behalf of SDLP. Among the four major parties SF offered women candidates the most opportunities to stand. Twelve of the party s candidates were women, 31.6% of the total. Eight of the Alliance party s 21 candidates were women, giving it the highest proportion (38.1%) among those parties contesting 20 or more seats. Only one woman stood as an Independent candidate, compared to 21 men. Unusually, none of the Green party s six candidates were women, despite the fact that in Great Britain a significant proportion of its candidates at parliamentary, European and local elections are women. It is not unusual, perhaps, that all seven of the Northern Ireland Women s Coalition (NIWC) candidates should be women. 6

7 Table 5: Women candidates by party at the Northern Ireland Assembly Election, 2003 Party label female % male % Total A Con DUP Green Ind NIUP NIWC PUP SDLP SEA SF SP UKUP UTW UUC UUP VFY WP Total Of the 108 Assembly members elected in 2003, 18 (16.6%) were women (Table A7). This proportion compares with the UK parliament (17.9% women elected in 2001), the Scottish Parliament (39.5%) and Welsh Assembly (50%). In five of the 18 constituencies, no women were elected to the NIA. In addition to West Tyrone, where none had stood, the others were East Antrim, East Londonderry, North Antrim, and South Antrim. Five constituencies (Belfast West, Fermanagh and South Tyrone, Foyle, Lagan Valley and South Down) each returned two women to the Assembly, the maximum number at this election. Overall, the two main parties of the nationalist community, SF and SDLP returned seven and five women respectively, three-quarters of all women Assembly members. In percentage terms, women comprise 29% and 28% respectively of the parties legislative membership. Both DUP and UUP had two of their women candidates elected (6.7% and 7.4% of their total Assembly membership), as did the Alliance party (33.3%). Incumbents seeking re-election Of the 108 Assembly members elected in 1998, no fewer than 88 (83.0%) fought the 2003 election (Table A8). Two members switched to fight other constituencies. Sammy Wilson, elected to Belfast East in 1998, fought East Antrim in This brought the number of incumbents challenging in the 7

8 Antrim constituency to seven. Similarly, Alex Maskey switched from Belfast West to Belfast South but this move meant that in South there were now seven incumbents competing for the six available seats. The only constituency where six incumbents sought re-election was Belfast North. The constituency which potentially offered the best opportunity for a nonincumbent to win was Belfast East where only three incumbents sought reelection. In fact, all three incumbents in Belfast East were safely returned. A similar level of success was only achieved in one other constituency, Foyle, where all four members that stood again were re-elected. It was inevitable, of course, that one incumbent should be defeated in Belfast South, but a single incumbent defeat occurred in a further nine constituencies. The largest number of losses for incumbents occurred in Upper Bann where only David Trimble, among the four sitting Assembly members, was re-elected. Overall, 65 members won a second term of office but 23 (26.1%) were defeated (Table A9) leaving 43 new members of the Assembly (Table A10). Candidates and the electoral system One in four candidates (27) exceeded the electoral quota with first preference votes and was elected at the first count stage. By the end of the sixth counting stage some 59 candidates were elected, more than half the total. The perception of STV, certainly among parts of the UK media, is that the counting process may sometimes appear arcane and time consuming. It is certainly true that counting an STV election takes longer than one employing a simpler counting method, such as First Past the Post (Westminster elections) or PR-List (European Parliament elections). The 18 stage count undertaken in Strangford in 1998 was not surpassed in However, in North Down no winner at all was declared until the twelfth stage, then one more by the thirteenth stage, and the final four elected at the fourteenth stage. In East Antrim, although two seats were declared at the first stage, it was not until the fifteenth stage that the final four seats were determined. DUP and UUP had ten and nine of their candidates returned after the first count. Six SF candidates passed the electoral quota with first preference votes alone but only one from the SDLP was similarly successful. The sole victory for an Independent, Kieran Deeny in West Tyrone, was secured at the first stage. Only one woman, Iris Robinson (DUP, Strangford) was elected solely by first preferences. 23 of the 26 Assembly members elected by first preference votes alone were incumbents seeking a second term. UUP won the largest number of first places, eight from 18 constituencies. This was one more than achieved by DUP, while SF and SDLP each had one first place. Deeny, in West Tyrone, was the one remaining first-placed candidate. DUP won eight second and six third places. UUP won two second, but seven third places. While SF secured the second seat four times and the third seat a total of five times, SDLP won three second places. A third 8

9 of SDLP s 18 seats were the sixth and final ones to be allocated, although five of SF s 24 seats were also the last to be decided. By contrast, just three of DUP s and two of UUP s seats were the final ones settled. One Alliance candidate was the second to be elected but the remaining five were elected later once transfers became effective. Party competition The operation of STV means that a party may not benefit from fielding a large number of candidates, indeed in some circumstances that might prove counter productive. There were few constituencies where one of the main parties stood more than three candidates. In Lagan Valley, the UUP had four candidates and the SDLP fielded the same number in two constituencies, Foyle and South Down. In Mid Ulster, SF fielded four candidates but in Belfast West, the party leader s constituency, there were five candidates representing the party. In Belfast West SF won four of the six Assembly seats available in that constituency, while in both Mid Ulster and Newry and Armagh the party captured three seats (Table A4). Seven of SF s seats were drawn from three of Belfast s four constituencies. The SDLP s most successful constituency was Foyle where half the seats were won, and in each of Belfast South and South Down two SDLP candidates won through. Although DUP could not match SF in Belfast West it did win three seats in East Antrim, North Antrim and Strangford. Across Belfast it won one seat more than did SF. Alone among the parties, DUP was able to win at least one seat in all 18 constituencies (the values for UUP, SDLP and SF were 16, 14 and 12 constituencies respectively). UUP won half the seats in one constituency, Lagan Valley, but that feat was rather overshadowed by the subsequent defection of two of the three elected (Jeffrey Donaldson and Norah Beare) to the DUP. Across the Belfast constituencies, UUP won five out of 24 seats. At least one seat changed hands in 14 of the 18 constituencies (Table A4). In both East Antrim and North Down, DUP captured two new seats and there were single seat gains in a further six constituencies, including Belfast West and North as well as Upper Bann. Although UUP gained a seat in Lagan Valley, seats were lost in North Antrim and North Down. Two of the six gains made by SF were in Belfast (one each in North and South). Two more gains were made on the north coast (North Antrim and East Londonderry) while the remaining two were in constituencies that border the Irish Republic, Newry and Armagh and South Down. There was some, though not complete, overlap between SF gains and SDLP losses. SDLP now holds only one of six Assembly seats in Newry and Armagh, a constituency it holds at the Westminster parliament. In West Tyrone, SDLP was the victim of Deeny s success, while in East Antrim and West Belfast the seats lost by SDLP were won by DUP. Among the minor parties, UKUP was the biggest loser, declining from five seats to just one seat, while PUP s representation was cut from two to one seat. Assembly seats that had been won by NIWC, NIUP and UUC in 1998 all changed hands. 9

10 Party quotas and the allocation of seats One measure of the efficiency of a party s performance is the extent to which first preference vote support is finally translated into seats. When the size of the electoral quota for each constituency is known (the number of votes a candidate requires in order to be guaranteed of winning a seat) then it becomes simple to calculate the number of quotas won after first preference votes are summed for each party. Thus, if say the electoral quota is 5,000 and party A s three candidates together win 10,000 first preference votes, the party has achieved two quotas. This is not the same as saying that two of those three candidates will necessarily win seats, but rather that it might, once votes are transferred. Most of the main parties can highlight some constituencies as near misses and others as evidence of astute campaigning (Table A11). In Belfast East, for example, DUP won the equivalent of 2.7 quotas but finished with two seats. In neighbouring Belfast South it achieved 1.5 quotas but won a single seat. By contrast, it won three seats in East Antrim, despite polling the equivalent of 2.4 quotas. The UUP performed well in East Londonderry, winning two seats whilst polling 1.6 quotas. But in Antrim North it was unable to capitalise further on 1.5 quotas and won a single seat. Belfast West might have provided SF with five seats (4.6 quotas) but that was thwarted by DUP, which won one seat despite achieving just half a quota of first preferences. Similarly, in both Upper Bann and West Tyrone, SF fell short, polling one and a half quotas and one seat in the former constituency, and 2.7 quotas and two seats in the latter. SDLP converted 1.6 quotas into two seats in Belfast South but in South Down failed to improve on 2.5 quotas and finished with two seats. However, in Foyle and Lagan Valley, fractional quotas for SDLP were converted upwards in the allocation of seats. The personal vote What of the personal vote for newly elected Assembly members? Although first preferences under STV are not as clear a guide to a candidate s popularity as are votes in say a first past the post election, they do provide a measure of how support divided. We should not expect, however, that any candidate will win a substantial share of first preferences; there are many more candidates than one would expect to find in a conventional Westminster style election; it may not be in a party s interest to have all of its supporters voting for a single candidate. Nevertheless, there are some interesting aspects of the distribution of first votes for elected candidates (Figure 1). Just six of the 108 new Assembly members were supported by 20% or more of constituency voters. The clear leader in popular vote was Jeffrey Donaldson, with 34.2% of first preferences in Lagan Valley (Table A12). He was more than four percentage points ahead of the next individuals on the list, Peter 10

11 Robinson (Belfast East) and Nigel Dodds (Belfast North). Iris Robinson, DUP Strangford, was not only the sole woman elected at the first stage but also the only woman to receive more than one in five first preference votes. Reg Empey too won more than one in five first votes (20.9%) and together he and Peter Robinson received more than half of first preferences cast by Belfast East s voters. The most popular politician amongst nationalist supporters was Gerry Adams in Belfast West who won 18.9% of first votes. Share of First Preference Votes for Candidates (2003) % share first preference vote Assembly seats Figure 1: Share of 1 st preferences for Candidates There were 37 candidates elected who received less than 10% of first preference votes cast in their constituencies. The two most numerous parties in this category were UUP (12 elected with less than 10%) and SDLP (11 elected). By comparison, five DUP and three SF members were returned having won less than one in ten first preferences. All but one of the six successful Alliance party candidates was similarly supported. The lowest share of first preferences for a successful candidate, 3.7%, went to Donaldson s running mate, Norah Beare, UUP, who benefited from the transfer of intra-party surplus votes. More than half the women returned to the Assembly (10/18) received fewer than 10% of first votes in their constituencies. Another method for assessing the personal support given to elected candidates is to express first preference votes as a proportion of the electoral quota. Figure 1a shows, for all 108 elected candidates, first preference vote as a proportion of their constituency electoral quota. There were 27 candidates whose first preference vote exceeded the electoral quota (Table A13). Three candidates received more than twice as many first preferences as the electoral quota for their constituency (Donaldson, Lagan Valley; Robinson, Belfast East; Dodds, Belfast North). A total of ten elected 11

12 candidates received less than half as many first preference votes as the electoral quota for their constituency. Candidates' First Preference Vote as a Proportion of Electoral Quota Proportion of Electoral Quota Figure 1a: 1 st Preference Votes as Proportion of Electoral Quota 12

13 Section 2: Electoral Registration, methods of voting and voter turnout The impact of new electoral registration procedures The implementation of the Electoral Fraud (Northern Ireland) Act 2002 meant that electoral registration procedures in the province are different to that in the remainder of the UK. Instead of single household registration, as in England, Scotland and Wales, each individual of voting age is now required to complete and sign their own electoral registration form. Potential electors are asked for details of personal identification, including date of birth and National Insurance number. The first register, using these procedures, was published in December There was an 89% response from those canvassed. The new registration procedures corresponded with a decrease in the Northern Ireland electorate (Table 6). Compared with the eligible electorate recorded at the 2001 general election (1,191,009) the 2003 electorate (1,097,526) was smaller by 93,483 electors. The figure represents a 7.8% decline in the size of the 2001 register. The reduction of eligible electors was considerable in some constituencies. In Belfast North, for example, there was a decline of 9,588 electors (15.7% of the 2001 register) between the 2001 general election and the 2003 Assembly election. In both Belfast South and West the drop was almost nine thousand electors (14.7% of the old registers). In more rural areas the fall in registered numbers was not as large. Table 6: Change in electorate at the Northern Ireland Assembly Election, Constituency Electorate 2003 Electorate 1998 Change Change (%) of 1998 register Belfast East 51,937 60,562-8, Belfast North 51,353 62,541-11, Belfast South 50,707 61,209-10, Belfast West 50,861 60,669-9, East Antrim 55,473 59,313-3, East Londonderry 56,203 59,370-3, Fermanagh & South Tyrone 64,336 65,383-1, Foyle 65,303 68,888-3, Lagan Valley 67,910 71,661-3, Mid Ulster 60,095 59, Newry & Armagh 68,731 71,553-2, North Antrim 70,489 73,247-2, North Down 57,422 62,942-5, South Antrim 63,640 69,426-5, South Down 70,149 71, Strangford 66,308 70,868-4, Upper Bann 68,814 70,852-2, West Tyrone 57,795 59,081-1, Total 1,097,526 1,178,556-81,

14 In Mid Ulster, for example, the 2003 electorate fell by slightly more than a thousand (2.1%) and in Fermanagh and South Tyrone it was just over two thousand fewer than in 2001 (3.5%). According to the 2003 electoral register, the largest electorates were North Antrim and South Down, each of which had more than 70,000 electors registered. The four smallest constituencies were those in Belfast, none of which were above 52,000 electors. The gap between largest and smallest constituency electorate was almost 20,000 electors and the mean electorate was 60,974. Change in electorates, How did the electoral registers for the two Assembly elections compare? First, as expected, there was a significant fall in the overall electorate size, from 1,178,556 in 1998; a decline of 81,030 electors or 6.9% of the 1998 register (Table 6). However, the decline is far from uniform (standard deviation of the decline in electorate is 3,438). In Mid Ulster, for example, the number registered actually increased by 104 electors (0.2%). In South Down the register shrunk by just 851 electors, or 1.2% of the 1998 register. By contrast, there was a five figure decline in both Belfast North (-11,188) and Belfast South (-10,502). In percentage terms the registers for these two constituencies fell by 17.9% and 17.2% respectively. In fact, the four largest declines are all in Belfast and the combined reduction for the city, -40,123 electors, comprises almost half the total decline for the province. The consequence of this reduction is that whilst 20.8% of the Northern Ireland electorate in 1998 was registered in Belfast, by 2003 the city s share had fallen to 18.7%. Methods for calculating electoral turnout There are various methods used for calculating turnout. In the United States, for example, turnout is sometimes calculated in the conventional manner, i.e. the proportion of the eligible electorate casting a valid ballot turnout is simply, total valid votes divided by eligible electorate, multiplied by 100. At other times, however, US sources calculate turnout based on numbers voting as a proportion of the voting age population (VAP), ignoring those listed on electoral registers. This reflects a concern that the voluntary, and sometimes difficult registration process, serves to under-estimate real electoral turnout. In Northern Ireland the practice has been for the electoral authorities to calculate turnout by including not only valid ballots but also ballots that were rejected from the count in the vote total. There are a variety of causes to reject a ballot: the paper may be missing the official mark; the voter may have misunderstood the voting method and incorrectly marked the ballot paper; the voter may have deliberately spoiled their ballot; the voter s intentions were 14

15 unclear from the completed ballot; and it may be possible to identify the voter from the ballot paper, thereby compromising ballot secrecy. In the context of the Electoral Fraud Act it is particularly important that our analysis is sensitive to the number and proportion of rejected votes. Accordingly, we report two different calculations of turnout. The first, unadjusted turnout, takes no account of whether a ballot is accepted or rejected from the process of electing Assembly members but rather that an elector has either returned a postal ballot or voted in person. This statistic is calculated as, total valid vote plus total invalid vote, divided by electorate and then multiplied by 100. This method is the one preferred by the Chief Electoral Officer for Northern Ireland and the official returns report turnout using this procedure. The second method for calculating turnout is referred to in this report as adjusted turnout. This is the conventional method for calculating turnout and is simply, total valid votes cast divided by registered electorate and then percentaged. The first points to consider are, whether any significant differences arise from employing these two methods, and, if so, whether some constituencies are affected more than others in this regard (Table 7). Naturally, unadjusted turnout will always be higher than adjusted turnout because additional ballots (rejected ones) have been included in the calculation. In the case of the 2003 election, aggregate turnout (unadjusted) was 63.98% while turnout (adjusted) was 63.05% - less than a one percentage point difference. Table 7: Electorate, Vote and Turnout at the Northern Ireland Assembly Election, 2003 Constituency Seats Electorate Total vote Total valid vote Total rejected votes Electoral quota Unadjusted turnout % Adjusted turnout % Belfast East 6 51,937 31,524 30, , Belfast North 6 51,353 31,997 31, , Belfast South 6 50,707 31,737 31, , Belfast West 6 50,861 33,527 32, , East Antrim 6 55,473 31,343 30, , East Londonderry 6 56,203 34,703 34, , Fermanagh & South Tyrone 6 64,336 46,873 46, , Foyle 6 65,303 41,436 40, , Lagan Valley 6 67,910 41,724 41, , Mid Ulster 6 60,095 45,023 44, , Newry & Armagh 6 68,731 48,233 47, , North Antrim 6 70,489 44,632 44, , North Down 6 57,422 31,316 30, , South Antrim 6 63,640 37,858 37, , South Down 6 70,149 46,012 45, , Strangford 6 66,308 37,838 37, , Upper Bann 6 68,814 44,145 43, , West Tyrone 6 57,795 42,328 41, , Total 1,097, , ,028 10,

16 The largest difference between the two calculations for a single constituency was for Newry and Armagh (unadjusted 70.18%; adjusted 68.93%) a 1.25 percentage point difference. A simple test for the correspondence between the two turnout measures results in a Pearson correlation coefficient of 0.99 identical to that calculated for the 1998 election. We can conclude, therefore, that for the 2003 election it makes no difference to the general analysis of turnout whether we report unadjusted or adjusted figures. Turnout in 2003 The highest turnout (unadjusted) was in Mid Ulster where 74.9% of eligible electors participated (Table 7). Three other constituencies (West Tyrone, Fermanagh and South Tyrone and Newry and Armagh saw turnout exceed more than seven in ten electors. When turnout is measured by omitting invalid votes the number of constituencies where turnout exceeded 70% falls to three because Newry and Armagh s adjusted turnout is 68.9%. The lowest turnout was in North Down (54.5% unadjusted, 53.7% adjusted) with a further three constituencies recording turnouts of fewer than six in ten of eligible electors. However, if adjusted turnout is used there are five constituencies with turnouts less than 60%. There was a range of twenty percentage points between the highest and lowest turnouts and a standard deviation of 5.8 (standard deviation of adjusted turnout is 5.7). Changes in turnout, Adjusted aggregate turnout in 2003 fell by 5.7 percentage points when it is compared to 1998 (Table 8). For individual constituencies the largest decline of 9.2 points occurred in Mid Ulster. This was followed by Foyle (8.3 points) and Upper Bann (7.9 points). The smallest drop took place in Lagan Valley (4.2 points) but this was only marginally different from three of the four Belfast constituencies, Strangford and South Antrim. Although a decline in turnout might seem disappointing, it should be noted that from there was a 13 point decline in general election turnout and there was an eight point fall at the May 2003 elections for the Scottish Parliament and Welsh Assembly. There is the further point that at the time of the election in 2003 the NIA was suspended and electors were uncertain when that suspension would end. In the light of falling turnout for other types of election it is, perhaps, appropriate that the 2003 NIA election is compared with the 2001 general election. Across the province turnout declined by five percentage points only. The largest fall was in Newry and Armagh, 7.8 points, followed by West Tyrone 7.7 points and Mid Ulster 7.5 points. The narrowest gap in turnout between the two elections occurred in Belfast South, where the difference in favour of the general election was just 2.1 percentage points and Lagan Valley (2.5 points). Of course, for specific cases there might be explanations for the size of the decline but overall, given the circumstances of the Assembly election, it is perhaps surprising that there is so little difference 16

17 between the level of participation between this and the previous general election. Table 8: Change in turnout at the Northern Ireland Assembly Election, Constituency 2003 Turnout % Unadjusted 1998 Turnout % Difference Turnout % Adjusted 1998 Turnout % Difference Belfast East Belfast North Belfast South Belfast West 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 Total Constituency characteristics and electoral turnout Data from the 2001 census are available for parliamentary constituencies and these may be used to examine the relationship between some social and economic characteristics of the population and levels of voter turnout. Prior research of the association between social characteristics and voter turnout has shown that, inter alia, older people tend to vote more than do younger, social exclusion is associated with low turnout, and the better educated vote more than those without formal qualifications. In the context of Northern Ireland, of course, religious community background is a key variable that should be explored for its possible relationship with turnout. Accordingly, we conducted some simple bivariate analyses that considered the relationship between some social, economic and religious characteristics for each constituency and voter turnout in We did not find strong relationships between constituency voter turnout and social characteristics such as housing tenure, age, unemployment, and education (Figure 2). The scatterplots for these variables and turnout show no clear, statistically significant patterns, although it should be noted that there are only 18 cases available for analysis. 17

18 % turnout % turnout % turnout % turnout % owner occupied % pensioners % unemployed % with no education qualifications Figure 2: Constituency level voter turnout and some social/economic characteristics Moreover, in some cases, the direction of the relationship, if any, is opposite to that which might be hypothesised. For example, it appears that as the proportion of pensioners in a constituency increases so the level of electoral turnout declines. It also appears that turnout increases as the proportion of unemployed people increases, as it does also for the proportion with no formal educational qualifications. Even allowing for the fact that our observations are based on a relatively small number of cases and these particular bivariate relationships are not strong, these findings are certainly contrary to what might be expected from a review of the prior research literature. What of the relationship between turnout and religious community background (Figure 3)? When constituency turnout in 2003 is graphed against the proportion of people who, at the 2001 census, claimed a Protestant community background, there appears to be an inverse relationship: as the proportion of residents in a constituency claiming a Protestant community background increases, electoral turnout decreases. The relationship between turnout and the proportion claiming a Catholic community background is less apparent. As the proportion of the population claiming a Catholic community background rises upwards to approximately 50% the relationship with turnout is linear: the higher the proportion of Catholics the higher the turnout. Above 50%, however, that linear relationship disappears. Survey, rather than aggregate data, would be required to investigate such matters further. 18

19 turnout % turnout % protestant community background % catholic community background Figure 3: Constituency level voter turnout and religious characteristics Postal voting The 2002 Act also introduced new procedures for postal voting. These procedures might have had an impact on both the overall numbers opting to choose a postal ballot and also the ways in which those votes would be scrutinised and rejected or not. Throughout this section we shall compare the experience of 2003 with that of A total of 19,969 postal ballots were issued less than half as many that were issued in 1998 (Table A14). This comprises just 1.8% of the electorate (3.2% in 1998). There were three constituencies which had double the overall rate of postal ballot applications, viz., Mid Ulster (3.6% of electorate requested postal ballots), West Tyrone (4.0) and Fermanagh and South Tyrone (5.6%). The level of requests from these largely rural constituencies contrasts with the pattern in urban constituencies. In Belfast, for example, only around 1% of eligible electors applied for a postal vote. This urban/rural pattern of distribution of postal ballots was similar to that in Of the postal ballots issued some 17,563 were returned before the deadline (88.0% unadjusted turnout) (Table 9). In 1998 the comparable figure was 87.1% postal ballots returned. It is normal that the turnout amongst postal voters is higher than amongst people that choose to vote in person at a polling station. In two constituencies, Mid Ulster and West Tyrone, unadjusted turnout amongst postal voters exceeded nine in ten of those that had applied (Table 10). All other constituencies, with one exception, showed a turnout of between 80-89%. The sole exception to this, and a considerable outlier, was North Down where only 366 of 496 (73.8%) postal ballots issued were returned. In 1998 there were two additional constituencies to those already mentioned with postal voter turnout above 90% - Fermanagh and South Tyrone (91.2%) and Foyle (90.1%). Only one constituency in 1998 failed to 19

20 reach a turnout of more than eight in ten postal voters, Lagan Valley, but even here it was 79.4%. Table 9: Postal voting and turnout at the Northern Ireland Assembly Election, 2003 Constituency Postal ballots issued Postal ballots received Number included in count Number rejected Unadjusted turnout % Adjusted turnout % Belfast East Belfast North Belfast South Belfast West East Antrim East Londonderry 1,288 1,103 1, Fermanagh & South Tyrone 3,600 3,238 3, Foyle 1, Lagan Valley Mid Ulster 2,174 2,009 1, Newry & Armagh 1,416 1,252 1, North Antrim North Down South Antrim South Down 1,462 1,275 1, Strangford Upper Bann 1, West Tyrone 2,330 2,144 2, Total 19,969 17,563 16,437 1, Table 10: Postal voting at the Northern Ireland Assembly Election, Constituency Electorate with postal vote % Turnout postal voters % Electorate with postal vote % Turnout postal voters % Difference in postal voters Change in postal voter turnout Belfast East Belfast North Belfast South Belfast West 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 Overall

21 Rejected postal votes and turnout It does not follow that because a postal ballot has been received in time for an election that it will necessarily be included in the count. The change in legislation for Northern Ireland has tightened security with regard to postal ballots. The data show that 1,126 (6.4%) postal ballots received were subsequently rejected from the count (Table 11). This number is in marked contrast to the position in 1998 when just 54 (1.9%) postal ballots were later rejected. There was some variation across constituencies in the proportion of rejected papers. In Strangford just 11 (2%) postal ballots were excluded and in North Down, which had had the lowest take up of postal ballots, only 9 (2.5%) were rejected. Four constituencies, Belfast West, Lagan Valley, Newry and Armagh and South Down, all had more than one in ten postal ballots excluded from the count. Table 11: Invalid votes at the Northern Ireland Assembly Election, 2003 Constituency Invalid postal % Invalid in person % Total invalid % Belfast East Belfast North Belfast South Belfast West 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 Total 1, , , The variation in the number of postal ballots received but then rejected may have consequences for the calculation of turnout amongst postal voters. Indeed, when adjusted turnout is measured, i.e. rejected postal ballots are excluded from the calculation, we find that only 82.3% postal voters participated in the election proper (Table 12). Moreover, as we have shown above, the rate at which postal ballots were rejected varied considerably from constituency to constituency. Whereas the correlation between unadjusted and adjusted overall turnout was 0.99, when the comparison is limited to postal voter turnout only the correlation falls to In 1998 these correlations remained steady at 0.99 each. We conclude that in 2003 there was considerably more variation in the distribution of rejected postal ballots than had occurred in Of course, it should be noted that postal voters 21

22 still comprise a small minority of the voting population and this finding, therefore, does not undermine the earlier view that unadjusted and unadjusted turnout are virtually identical. Table 12: Difference in turnout between postal and in-person voting at the Northern Ireland Assembly Election, 2003 Unadjusted Adjusted Postal In person Postal In person Constituency % % Difference % % Difference Belfast East Belfast North Belfast South Belfast West 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 Overall Although the turnout amongst people voting by post is high, as a proportion of the total voting their numbers are rather small. In 2003 just 2.5% of all those that cast a ballot did so by post (3.9% in 1998). The largest number of postal voters, as a proportion of all voters, were in Fermanagh and South Tyrone (6.9%) followed by West Tyrone (5.1%). These two constituencies were, with Mid Ulster, the largest in In Belfast, the corresponding figures were close to 1.0%. In-person voting While 17,563 electors returned postal ballots almost 700,000 voted in person at a polling station (Table 13). The overall unadjusted turnout amongst those voting in person was 63.7% (adjusted turnout 62.7%). In 1998, over a million electors voted in person and turnout was 69.4% (adjusted 68.2%). Turnout amongst in-person voters fell by almost six percentage points (unadjusted 5.9; adjusted 5.5). The highest turnout in 2003 was Mid Ulster (74.6% unadjusted, 73.4% adjusted) and the lowest was North Down (54.4% unadjusted, 53.5% adjusted). These two constituencies were also the highest and lowest in There is a large gap in turnout, therefore, amongst postal and in- 22