Analysis of local election results data for Wales 2004 (including turnout and extent of postal voting)

Analysis of local election results data for Wales 2004 (including turnout and extent of postal voting) By Professors Michael Thrasher and Colin Rallings of the University of Plymouth Elections Centre

Introduction Elections for the 22 county councils in Wales took place on June 10 th, 2004. The elections, originally scheduled for May 2003, were postponed so that they did not coincide with the Assembly election. The postponement meant that the local elections were held at the same time as the European Parliament election. A total of 3,135 candidates competed for 1,262 seats in 879 divisions. Two elections, one in Bridgend, the other in Conwy, were postponed and although by-elections were held subsequently the results are excluded from this summary. Following the election the Conservative party and Plaid Cymru each had political control in one authority, three councils were controlled by Independents, eight were under Labour control and the remaining nine were hung councils with no single party in a majority. The battle for votes Table 1 shows the overall result for the more than twenty registered parties that contested the election. The Labour party polled the most votes (three in ten votes cast) with Independent candidates, supported by more than one in five voters, the next most popular choice. Plaid Cymru received support from one in six voters, closely followed by the Liberal Democrats. The Conservative party captured just over one in ten votes cast. The remaining parties won few votes overall (50,291, 5.5%) and for this reason all subsequent tables group their support under the general heading of Others. Labour won 478 seats, 37.9% of the total, enjoying a small winner s bonus, customary under the first past the post voting system. A similar bonus went to Independent candidates, 322 of whom were elected. One in four councillors elected to the new councils in 2004 stood as an Independent candidate. The remaining parties all won fewer seat than vote shares, although the percentage point differences were small. Few candidates from other parties were successful. The notable exceptions to this were Neath Port Talbot Ratepayers Association, People before Politics and Swansea Independent Association, none of which polled more than four figure vote totals but were successful nevertheless because support for them was concentrated in a relatively small number of divisions. Table 1: Summary results for each party PARTY votes %shares seats %seats British National 254 0.03 0 0 Party Caerleon 1,016 0.11 0 0 Independent Progress Party Cardiff 2,630 0.29 0 0 Independent Citizens Christian Peoples 266 0.03 0 0

Alliance Communist 198 0.02 0 0 Conservative 99,991 11.00 107 8.49 Democratic 784 0.09 3 0.24 Alliance Forward Wales 5,272 0.58 1 0.08 Green 10,799 1.19 0 0 Independent 206,394 22.71 322 25.46 Labour 277,599 30.54 478 37.91 Liberal 125,714 13.83 146 11.58 Democrats Llais Ceredigion 1,096 0.12 0 0 Liberal 1,139 0.13 2 0.16 Neath Port Talbot 7,027 0.77 9 0.71 Ratepayers New Millenium 522 0.06 0 0 Bean Party Plaid Cymru 148,974 16.39 174 13.8 People before 5,899 0.65 9 0.71 Politics People s 620 0.07 1 0.08 Representative Respect 210 0.02 0 0 Rhondda Peoples 596 0.07 0 0 Party Social 1,018 0.11 3 0.24 Democratic Party Social 388 0.04 0 0 Alternative Swansea 9,340 1.03 7 0.56 Independent Association United Kingdom Independence Party 1,217 0.13 0 0 Total 908,963 100 1262 100 Note: In single member divisions the calculation of each party s vote is straightforward but this is not the case in divisions where more than one seat is being decided. To calculate party vote in these cases we use votes cast for the highest placed candidate from each party. Where Independents stand we cluster these candidates into groups according to the number of seats at stake.

Party competition The number of seats in each division is termed district magnitude. For parliamentary elections district magnitude equals one because each constituency is represented by a single member of parliament. This is not universally so for local elections. Although in the majority of divisions district magnitude equals one that still leaves 163 (18.5%) that return two councillors, 76 (8.6%), returning three, 20 (2.3%) represented by four councillors and two divisions, both in Swansea, that have district magnitudes of five. In cases where district magnitude exceeds one, a party may field fewer candidates than there are seats available. Nevertheless, supporters of that party will have an opportunity to vote for a preferred candidate, either discarding any unused votes or casting them in favour of other candidates. Elections, contested and uncontested, took place in 879 divisions, excluding the two postponed contests. Table 2 takes account of the fact that some divisions elect more than a single councillor. The largest presence was that of the Labour party, whose candidates stood in two thirds of all divisions. In more than one in two divisions voters could vote for an Independent candidate whilst a slightly smaller proportion had the opportunity to vote for a Plaid Cymru candidate. Just over one in three divisions featured at least one Liberal Democrat but Conservative candidates stood in only one in four divisions. A party is deemed to have won a division when its candidate finishes at the top of the poll. This does not mean, necessarily, that it has captured all available seats; where district magnitude is greater than one seats may be shared between candidates that belong to different parties. Some 261 out of 879 divisions (29.7%), elected more than a single councillor. Of those 261 divisions with a district magnitude greater than one, a total of 93 (35.6%) returned councillors that had stood for different parties. Table 2: Divisions contested and won by party divisions % divisions % contested won Conservative 224 25.5 71 8.1 Labour 585 66.6 280 31.9 Lib Dem 320 36.4 98 11.1 Plaid Cymru 413 47.0 129 14.7 Independent 476 54.2 283 32.2 Other 131 14.9 18 2.0 Total 879 100 879 100 By contrast to Table 2, Table 3 takes account of varying district magnitudes, concentrating instead on the contests for 1,262 seats. According to this measure of party competition Labour contested 73.8% of seats, some distance ahead of Plaid Cymru and Independent candidates, who challenged in fewer than one in two seats. Labour candidates were successful in half the seats contested but the success rate of Independents was slightly higher (0.59). Approximately one in three candidates were

successful for the Conservatives, Liberal Democrats and Plaid Cymru parties respectively. Table 3: Seats contested and won by party seats seats contested % won % Conservative 349 27.7 107 8.5 Labour 931 73.8 478 37.9 Lib Dem 422 33.5 146 11.6 Plaid Cymru 546 43.3 174 13.8 Independent 548 43.5 322 25.5 Other 193 15.3 35 2.8 Total 1,262 100 1262 100 The operation of the electoral system A party benefits from the operation of the electoral system when its seats to votes ratio exceeds one. In such circumstances a party s overall share of seats is greater than its vote share. Of course, there may be different reasons for this. Large parties tend to have favourable seat/vote ratios because of electoral system effects. Smaller parties may also have a favourable ratio simply because their support is concentrated across a relatively small geographical area rather than fragmented across the country. Table 4 shows the seat/vote ratios for the 2004 election. The biggest advantage is Labour s with a ratio of 1.2 followed by 1.1 for Independents. The three other main parties, Conservative, Liberal Democrats and Plaid Cymru all have ratios of 0.8, indicating that all won a smaller proportion of seats than vote share. Although there are inequalities in the distribution of votes and seats, the overall outcome was reasonably proportional for an election conducted by first past the post. This can be demonstrated by employing the Loosemore-Hanby index of proportionality. This index first measures the absolute difference (i.e. ignores negative or positive signs) for each party s share of votes and seats. The differences are then summed and the total divided by two, leaving the index of proportionality as a single summary statistic. Another way of considering the index is that it captures what proportion of votes cast did not result in a share of seats wasted votes in the view of critics of the voting system. A rule of thumb suggests that an outcome with a Loosemore-Hanby index score of 10 or lower is proportional. The 2001 Westminster election, similarly fought under first past the post, results in an index score of 21.9 approximately the excess of Labour s share of seats over its vote share. For the Welsh Assembly election in 2003, fought using the Additional Member System, the index is 15.9. The index score for the 2004 Welsh local elections as a whole is 10.1, although that number would vary if the calculation was made on an authority by authority basis.

Table 4: Seat/Vote ratios Seat/Votes ratio Con 0.8 Lab 1.2 Lib Dem 0.8 Plaid Cymru 0.8 Independent 1.1 Other 0.5 Voting trends The previous local elections were held five years before in 1999. Boundary changes affecting 6 of 22 authorities mean that the number of divisions and seats has altered but comparisons at the aggregate level are still possible between the two elections. Table 5 shows vote shares for 2004 and 1999 and change in vote. Labour s vote declined by almost four percentage points while there was an almost two point fall for Plaid Cymru. There were only modest increases in vote share for both Conservative and Liberal Democrats but the largest increase was seen in support for Independent candidates whose overall vote share rose from 19% in 1999 to 22.7% in 2004. Table 5: Vote shares and change in vote share, 1999 and 2004 2004 1999 change % % Conservative 11.0 10.1 +0.9 Labour 30.5 34.5-3.9 Lib Dem 13.8 13.4 +0.4 Plaid Cymru 16.4 18.2-1.9 Independent 22.7 19.0 +3.7 Green 1.2 0.8 +0.3 Other 4.3 3.8 +0.5 Table 6 considers change in the proportion of seats gained in both 1999 and 2004. Although there was only a modest increase in the Conservative party vote share it did rather better in terms of increasing seat share, albeit from a low starting base. There was a greater improvement for the Liberal Democrats as the party s seat share rose from 7.7% to 11.6%. Among the main political parties the largest decline in seat share was that for Labour, a six-point fall from five years before.

Table 6: Seat shares and change in seat share, 1999 and 2004 2004 1999 change % % Conservative 8.5 5.9 +2.6 Labour 37.9 44.3-6.4 Lib Dem 11.6 7.7 +3.9 Plaid Cymru 13.8 16.1-2.3 Independent 25.5 23.2 +2.2 Green 0.0 0.1-0.1 Other 2.8 2.6 +0.2 Voting patterns and electoral contests Comparisons are also possible between votes cast at the 2004 local elections and at other elections. The local elections on June 10 th were held simultaneously with those to the European Parliament (EP). For the EP election Wales comprises a single electoral unit, returning four members elected by a PR-List voting system. Each voter has one vote. Table 7 compares the share of votes cast across Wales at each of these elections although care should be taken when interpreting the data because of the variable pattern of party competition for local council seats. Table 7: Vote shares at the Local and European Parliament elections, 2004 Local European Difference Conservative 11.0 19.4 +8.4 Labour 30.5 32.5 +1.9 Lib Dem 13.8 10.5-3.3 Plaid Cymru 16.4 17.4 +1.0 Independent 22.7 - -22.7 UKIP 0.1 10.5 +10.4 Other 5.6 9.7 +4.1 Since turnout is virtually identical the data suggest that a considerable proportion of voters divided their support amongst different parties at the two elections. Given the absence of an Independent slate of candidates fighting the European election, voters supporting Independent candidates at the local elections must have either reverted to one of the mainstream parties, most likely the Conservative Party, or possibly to one

of several minor parties, including the United Kingdom Independence party, Greens or British National Party. The Conservative vote almost doubled between the local and European election contests whilst there were smaller increases for Labour and Plaid Cymru. The EP vote share for the Liberal Democrats fell when compared to its standing at the local polls. Doubtless, some of these differences are a function of the variable patterns of party competition for local council seats but that does not wholly account for the variance in party support. Another opportunity for comparing voting patterns lies with local and Welsh Assembly elections. In 1999 the two elections were held simultaneously and the results showed then that a significant proportion of voters preferred different parties according to the type of contest. The Conservative vote, for example, was six percentage points lower at the local than for the Assembly contest whilst that for Plaid Cymru was over eleven points lower. Again, we should not overlook some effect on overall vote share caused by differential party competition. Table 8 brings the story up to date and shows that the vote share for the four main parties in 2004 is lower at the local than at the 2003 Assembly election. The largest gap is for Labour and Conservative parties, with only a very modest decline for the Liberal Democrats. There may be vote switching between the mainstream parties at different elections but undoubtedly the more than one in five voters that support Independents at local elections need to find another party to support when European and Assembly elections are held. Table 8: Vote shares at the local council and Welsh Assembly elections, 2003/2004 Local Election, 2004 Assembly election 2003* Difference Con 11.0 19.9-8.9 Lab 30.5 40.0-9.5 Lib Dem 13.8 14.1-0.3 Plaid Cymru 16.4 21.2-4.8 Ind 22.7 - +22.7 Others 5.6 4.7 +0.9 * Constituency ballots only Candidates and competition One important measure of the health of local democracy is the competition for seats. There are various methods for assessing the extent of electoral competition. First, we can examine the number of seats that remain uncontested because insufficient

nominations are received to trigger an election. Second, the ratio of candidates to seats may be studied and taken to be a proxy measure for the choice available to electors. In this section each method will be employed and the results compared with the previous local election. Table 9: Uncontested seats, 1999 and 2004 2004 1999 Seats Uncontested % Seats Uncontested % Anglesey 40 14 35.0 40 7 17.5 Blaenau Gwent 42 3 7.1 42 0 0.0 Bridgend 53 2 3.8 54 12 22.2 Caerphilly 73 0 0.0 73 2 2.7 Cardiff 75 0 0.0 75 0 0.0 Carmarthenshire 74 6 8.1 74 18 24.3 Ceredigion 42 5 11.9 44 11 25.0 Conwy 58 4 6.9 59 8 13.6 Denbighshire 47 3 6.4 47 2 4.3 Flintshire 70 14 20.0 70 14 20.0 Gwynedd 75 36 48.0 83 42 50.6 Merthyr Tydfil 33 0 0.0 33 0 0.0 Monmouthshire 43 1 2.3 42 2 4.8 Neath & Port Talbot 64 4 6.3 64 9 14.1 Newport 50 0 0.0 47 0 0.0 Pembrokeshire 60 5 8.3 60 16 26.7 Powys 73 45 61.6 73 39 53.4 Rhondda/Cynon/Taff 75 0 0.0 75 1 1.3 Swansea 72 0 0.0 72 3 4.2 Torfaen 44 7 15.9 44 0 0.0 Vale Of Glamorgan 47 0 0.0 47 0 0.0 Wrexham 52 3 5.8 52 15 28.8 Total 1262 152 12.0 1270 201 15.8 * The number of seats may vary between elections either because of boundary changes or postponed elections. Table 9 shows the number of seats at the 2004 and 1999 local elections. It is clear that the proportion of uncontested seats reduced from 15.8% overall in 1999 to 12%, or one seat in eight, in 2004. Of course, any decline in the number and proportion of uncontested seats is good for local democracy but the figures should be viewed perhaps in a broader context. For example, at the 2003 local elections in England, when a large majority of English districts voted, just 6% of more than ten thousand seats were uncontested. That same year Scottish local councils were re-elected simultaneously with elections to the Parliament. The proportion of uncontested council seats was exactly half the figure for Wales in 2004. However, a closer analysis reveals that the distribution of uncontested seats in Wales is rather skewed. As Table 9 reveals, just four local authorities, Isle of Anglesey, Flintshire, Gwynedd and Powys, are largely responsible for the volume of uncontested seats. If these particular authorities are excluded from the analysis then

the number of uncontested seats falls to just 43 from a total of 1004 seats, or 4.3%. Of course, a similar operation could be performed on the data for English and Scottish local elections, providing a different picture there too, but this does serve to show the contribution of these particular councils to the overall level of uncontested seats in Wales. Finally, some caution should be exercised in the interpretation of uncontested seats. Typically, it is Independent candidates in relatively small rural communities that are unchallenged. Our data do not allow us to consider whether such candidates are free from competition because the local communities concerned arrive at some pre-election consensus. Where competition for seats does occur we can measure its extent by considering the ratio of candidates to seats. Table 10 compares ratios in 2004 with the previous election, both overall and by local authority. It shows that the overall ratio in 1999 was 2.3 candidates for each vacancy rising to 2.5 at the most recent election. On this measure, therefore, local elections became more competitive. It is instructive also to compare these data with the position for both England and Scotland. In 2003 the candidate/seats ratio for English local authorities was identical to Wales in 2004 but the comparable figure for Scotland was higher at 3.4 candidates contesting each seat on average. It should be noted, however, that the Scottish Nationalist Party has adopted a policy of fighting local council seats across most parts of the country and that clearly affects the number of candidates overall. Table 10: Ratio of candidates to seats, 1999-2004 2004 1999 Candidates Ratio Candidates Ratio Anglesey 72 1.8 93 2.3 Blaenau Gwent 80 1.9 89 2.1 Bridgend 140 2.6 118 2.2 Caerphilly 176 2.4 167 2.3 Cardiff 297 4.0 264 3.5 Carmarthenshire 172 2.3 138 1.9 Ceredigion 89 2.1 82 1.9 Conwy 152 2.6 135 2.3 Denbighshire 125 2.7 122 2.6 Flintshire 129 1.8 140 2.0 Gwynedd 120 1.6 126 1.5 Merthyr Tydfil 80 2.4 67 2.0 Monmouthshire 142 3.3 129 3.1 Neath & Port Talbot 144 2.3 127 2.0 Newport 146 2.9 142 3.0 Pembrokeshire 157 2.6 123 2.1 Powys 115 1.6 122 1.7 Rhondda/Cynon/Taff 182 2.4 186 2.5 Swansea 250 3.5 186 2.6 Torfaen 85 1.9 68 1.5 Vale Of Glamorgan 137 2.9 125 2.7 Wrexham 145 2.8 112 2.2 Total/Overall 3135 2.5 2861 2.3

The more detailed figures that take account of individual local authorities shows variation in both scale and trend. In 2004, for example, three authorities, Cardiff, Swansea and Monmouthshire, saw candidate/seat ratios exceed three, whilst five authorities, Powys, Gwynedd, Isle of Anglesey, Flintshire, Blaenau Gwent and Torfaen saw fewer than two candidates on average contest each vacancy. In terms of the trend other differences emerge. Sixteen of the 22 authorities saw the competition for seats increase from 1999-2004, with Swansea showing the largest increase. Six authorities, however, saw a fall in the ratio, although in every case the decline was rather modest. It is difficult to explain such variations in competition and it might be that local circumstances in both 1999 and 2004 are responsible for both the level and trend in candidate competition. Women candidates The data allow us to consider two key characteristics of candidates gender and incumbency. The 2003 Welsh Assembly, with women comprising exactly half of its current membership, can properly claim to be amongst the most gender balanced elected bodies in the world. How do local authorities in Wales compare? Incumbency has also proved a lively issue in the context of Welsh local government; measures were sought to persuade long-serving councillors to retire, thereby permitting more opportunities for candidates to stand that are a more accurate reflection of the social make up of local communities. A total of 809 women candidates stood in 2004 (see Table 11), accounting for slightly more than a quarter of all candidates. This was an almost two percentage point increase on 1999. Wales lags behind both England and Scotland in the proportion of women candidates fighting local elections, although the gap is only five percentage points or less. In Cardiff more than one in three candidates was a woman while for six other authorities women numbered more than three in ten candidates. In most of the remaining councils women stood in reasonably large numbers but in five cases the proportion fell below one in five, and in the case of the Isle of Anglesey just six of the 72 (8.3%) candidates was a woman. The most dramatic increase in the number of women standing for election occurred in Wrexham where there was an eight point increase from 1999-2004. There were smaller increases in Neath and Port Talbot, Vale of Glamorgan and Pembrokeshire. In nine authorities the proportion of women candidates declined, the largest fall of four percentage points coming in Flintshire.

Table 11: Women candidates, 1999-2004 2004 1999 N= % N= % Anglesey 6 8.3 9 9.7 Blaenau Gwent 13 16.3 12 13.5 Bridgend 44 31.4 38 32.2 Caerphilly 53 30.1 52 31.1 Cardiff 101 34.0 93 35.2 Carmarthenshire 36 20.9 27 19.6 Ceredigion 13 14.6 10 12.2 Conwy 41 27.0 29 21.5 Denbighshire 25 20.0 28 23.0 Flintshire 28 21.7 36 25.7 Gwynedd 22 18.3 16 12.7 Merthyr Tydfil 15 18.8 12 17.9 Monmouthshire 45 31.7 38 29.5 Neath & Port Talbot 36 25.0 23 18.1 Newport 40 27.4 36 25.4 Pembrokeshire 37 23.6 21 17.1 Powys 35 30.4 31 25.4 Rhondda/Cynon/Taff 46 25.3 53 28.5 Swansea 65 26.0 49 26.3 Torfaen 27 31.8 23 33.8 Vale Of Glamorgan 42 30.7 30 24.0 Wrexham 39 26.9 21 18.8 Total 809 25.8 687 24.0 How successful were those women that did stand for election? Table 12 shows that a smaller proportion was successful overall - just 21.8% compared to 25.8% that stood as candidates. This difference may largely be related to the presence of male incumbent councillors rather than voter antipathy towards women as candidates, although a more detailed analysis would be required to confirm this. Nevertheless, more women were elected in 2004 than in 1999. At the 2003 Scottish local elections women comprised 22.5% of those elected, close to the Welsh figure, but higher proportions than that are elected for various types of local authority in England. Table 12: Elected women, 1999-2004 2004 1999 Elected % Elected % Anglesey 2 5.0 3 7.5 Blaenau Gwent 4 9.5 3 7.1 Bridgend 16 30.2 13 24.5 Caerphilly 18 24.7 17 23.3 Cardiff 28 37.3 24 32.0 Carmarthenshire 13 17.6 12 16.2 Ceredigion 4 9.5 3 7.1 Conwy 14 24.1 13 22.4 Denbighshire 10 21.3 9 19.1 Flintshire 13 18.6 11 15.7

Gwynedd 16 21.3 11 14.7 Merthyr Tydfil 1 3.0 5 15.2 Monmouthshire 10 23.3 8 18.6 Neath & Port Talbot 13 20.3 12 18.8 Newport 9 18.0 6 12.0 Pembrokeshire 11 18.3 8 13.3 Powys 20 27.4 19 26.0 Rhondda/Cynon/Taff 16 21.3 21 28.0 Swansea 22 30.6 17 23.6 Torfaen 15 34.1 15 34.1 Vale Of Glamorgan 14 29.8 9 19.1 Wrexham 6 11.5 9 17.3 275 21.8 248 19.5 Women comprise more than a third of councillors in just two of 22 councils. Unusually, in those two councils, Cardiff and Torfaen, women candidates enjoyed a higher success rate than male candidates. The same councils also occupied the top two positions at the 1999 election. By contrst, there are four councils, Merthyr Tydfil, Isle of Anglesey, Ceredigion and Blaenau Gwent where, after the 2004 election, women councillors comprise less than ten per cent of the council membership. The fate of incumbents The other noteworthy feature of candidates is the fate of sitting councillors seeking reelection. We make no comment here on the merits of incumbency, regarded in some quarters as an obstacle to change, for others necessary to preserve continuity and experience. Table 13 shows both the number of incumbents that sought re-election in 1999 and 2004 and incumbents as a percentage of all candidates. The actual number of incumbents seeking re-election increased from 932 to 947 from 1999-2004. However, because of the general increase in the number of candidates, whilst almost one in three candidates in 1999 was an incumbent that proportion reduced slightly to three in ten in 2004. Table 13: Councillors seeking re-election, 1999-2004 2004 1999 N= % N= % Anglesey 34 47.2 33 35.5 Blaenau Gwent 32 40.0 34 38.2 Bridgend 40 28.6 29 24.6 Caerphilly 63 35.8 53 31.7 Cardiff 54 18.2 51 19.3 Carmarthenshire 45 26.2 57 41.3 Ceredigion 24 27.0 27 32.9 Conwy 48 31.6 47 34.8 Denbighshire 30 24.0 34 27.9 Flintshire 46 35.7 49 35.0 Gwynedd 73 60.8 67 53.2 Merthyr Tydfil 28 35.0 23 34.3

Monmouthshire 31 21.8 29 22.5 Neath & Port Talbot 44 30.6 47 37.0 Newport 30 20.6 32 22.5 Pembrokeshire 40 25.5 43 35.0 Powys 57 49.6 65 53.3 Rhondda/Cynon/Taff 62 34.1 47 25.3 Swansea 58 23.2 57 30.7 Torfaen 33 38.8 36 52.9 Vale Of Glamorgan 33 24.1 34 27.2 Wrexham 42 29.0 38 33.9 Total 947 30.2 932 32.6 In terms of incumbents seeking re-election there is an extensive range between councils. In Gwynedd, for example, more than six in ten of all candidates were incumbents, many of whom were returned unopposed. In Powys and Isle of Anglesey too, there was a high proportion of incumbents seeking re-election. By contrast, fewer than one in five candidates fighting elections in Cardiff was a sitting councillor, a proportion only slightly less than those in Newport and Monmouthshire. It may be that Table 13 presents a distorted picture, focussing as it does on incumbents as a proportion of all candidates. It might be more instructive to look at the number of incumbents in terms of the proportion of seats contested (see Table 14). This shows that, in three-quarters of vacancies, voters had the choice of returning an existing councillor (although because of boundary changes, re-selection issues or party defections, the number may be slightly lower if two or more sitting councillors were competing for one seat). Despite the concerns about incumbents expressed prior to the election the proportion rose, albeit by a very small amount. In Gwynedd there were no fewer than 73 incumbents standing - the council has a total of just 75 seats. Seven further councils saw the number of incumbents rise above 80% of available seats. There was just one council, Ceredigion, where the proportion of incumbents to seats fell below 60%. Table 14: Councillors seeking re-election as a proportion of available seats 2004 1999 % % change Anglesey 85.0 82.5 +2.5 Blaenau Gwent 76.2 81.0-4.8 Bridgend 75.5 53.7 +21.8 Caerphilly 86.3 72.6 +13.7 Cardiff 72.0 68.0 +4.0 Carmarthenshire 60.8 77.0-16.2 Ceredigion 57.1 61.4-4.3 Conwy 82.8 79.7 +3.1 Denbighshire 63.8 72.3-8.5 Flintshire 65.7 70.0-4.3 Gwynedd 97.3 80.7 +16.6 Merthyr Tydfil 84.8 69.7 +15.1 Monmouthshire 72.1 69.0 +3.1 Neath & Port Talbot 68.8 73.4-4.6 Newport 60.0 68.1-8.1

Pembrokeshire 66.7 71.7-5.0 Powys 78.1 89.0-10.9 Rhondda/Cynon/Taff 82.7 62.7 +20.0 Swansea 80.6 79.2 +1.4 Torfaen 75.0 81.8-6.8 Vale Of Glamorgan 70.2 72.3-2.1 Wrexham 80.8 73.1 +7.7 Total 75.0 73.4 +1.6 Of course, it does not necessarily follow that council membership is static because there are incumbents seeking re-election those incumbents may subsequently be defeated. Therefore, in Table 15, we show the level of electoral success for incumbents. A total of 743 of the 947 incumbents standing in 2004 were re-elected 78.5% of the total. This figure compares with 76.7% of Assembly Members that were successfully returned at the 2003 election. When an incumbent wishes to continue in office, therefore, it is odds on that the desire is fulfilled. In three authorities, Vale of Glamorgan, Blaenau Gwent and Powys, there was better than a nine in ten chance that an incumbent would be re-elected. For incumbents standing in authorities undergoing a degree of political flux the chances of re-election were less. In Merthyr Tydfil, Rhondda/Cynon/Taff and Bridgend for example, local political factors go some way towards the explanation of why incumbents performed relatively badly. Table 15: The success rate of incumbents, 2004 Incumbents contesting Winning incumbents % Anglesey 34 30 88.2 Blaenau Gwent 32 30 93.8 Bridgend 40 25 62.5 Caerphilly 63 46 73.0 Cardiff 54 39 72.2 Carmarthenshire 45 40 88.9 Ceredigion 24 20 83.3 Conwy 48 36 75.0 Denbighshire 30 19 63.3 Flintshire 46 39 84.8 Gwynedd 73 60 82.2 Merthyr Tydfil 28 16 57.1 Monmouthshire 31 20 64.5 Neath & Port Talbot 44 36 81.8 Newport 30 25 83.3 Pembrokeshire 40 31 77.5 Powys 57 52 91.2 Rhondda/Cynon/Taff 62 37 59.7 Swansea 58 49 84.5 Torfaen 33 28 84.8 Vale Of Glamorgan 33 31 93.9 Wrexham 42 34 81.0 947 743 78.5

Electoral turnout Before turning to the analysis of electoral turnout there are some issues regarding data that should be clarified. At the time of writing of this report some local authorities could not provide complete and/or accurate information about some aspects of electoral turnout. In some cases the numbers of ballot papers issued for each division were not provided, in others the number of rejected votes, the number of postal votes issued and returned etc. are missing. However, we do not believe that the missing data are extensive, affecting a relatively small proportion of divisions and certainly should not prevent this part of the report from being prepared. Accordingly, we have produced commentary and prepared tables relating to electoral turnout but the reader should be aware that these observations and statistics should be regarded as indicative rather than definitive. Table 16 shows both unadjusted and adjusted turnout for each local authority. Unadjusted turnout is calculated as the sum of valid and invalid votes divided by the eligible electorate multiplied by 100. Adjusted turnout is the same formula but invalid votes are excluded from consideration. Adjusted turnout will always fall below unadjusted turnout. Unadjusted turnout is 41.8% and adjusted turnout is only slightly lower, 41.0%, a difference of less than one percentage point between the two methods of calculation. In two authorities, Ceredigion and Pembrokeshire, turnout rose above half the eligible electorate. The lowest turnout, in Bridgend, saw one in three electors voting and in a further five authorities, including Cardiff, Newport and Swansea, less than four in ten electors voted. The general picture changes very little when adjusted turnout is examined. Only one authority, Ceredigion, has a turnout of above 50% on this measure and ten authorities have turnout below 40%. Table 16: Electoral turnout in 2004 Unadjusted % Adjusted % Difference Anglesey 49.4 49.0 0.5 Blaenau Gwent 46.0 45.2 0.8 Bridgend 33.6 32.9 0.6 Caerphilly 40.4 39.7 0.7 Cardiff 38.7 38.3 0.4 Carmarthenshire 47.5 46.7 0.8 Ceredigion 53.8 53.2 0.6 Conwy 38.4 37.3 1.1 Denbighshire 44.4 43.8 0.6 Flintshire 36.2 35.4 0.8 Gwynedd 48.1 47.3 0.8 Merthyr Tydfil 40.2 39.8 0.4 Monmouthshire 48.3 47.8 0.6 Neath & Port Talbot 43.4 42.2 1.3 Newport 37.9 37.0 0.9 Pembrokeshire 50.6 49.8 0.8

Powys 49.1 48.4 0.7 Rhondda/Cynon/Taff 42.2 40.8 1.4 Swansea 36.9 36.3 0.5 Torfaen 40.1 39.5 0.6 Vale Of Glamorgan 44.4 43.6 0.7 Wrexham 40.0 39.2 0.8 Overall 41.8 41.0 0.8 It is difficult to make detailed comparisons of turnout for 1999-2004. This is because the only data that we have for 1999 is based upon our own calculations rather than information from local authorities regarding the actual number of ballot papers issued. In some cases our estimates may be unreliable because they are based on the number of votes cast for each party and cannot take into account the number of voters who did not use their full quota of votes in divisions whose district magnitude exceeds one or of voters with multiple votes who supported more than one party. That said, our best estimate suggests that turnout declined by about eight percentage points from the 1999 figures. As noted earlier, there were 152 seats in 145 divisions that were uncontested. The total electorate in these uncontested divisions was 208,496. This means that some 9.4% of the electorate did not have the opportunity to vote at the 2004 elections. In 1999 some 195 divisions were uncontested, containing over three hundred thousand electors, and 13.7% of the local electorate. Postal voting Table 17 shows the numbers of postal votes issued. More than two hundred thousand electors were issued with a postal vote, 10.6% of the eligible electorate. Although Cardiff saw the largest number such voters, in percentage terms the authority with the highest proportion is Neath and Port Talbot where almost one in five electors was issued a postal vote. Nine authorities saw postal vote applications rise to above one in ten electors. The smallest number of applications was in Ceredigion where just 2,272 postal votes were issued, covering less than one in twenty electors. The range between highest and lowest postal vote authorities is 13.5 percentage points. Table 17: Postal votes issued postal votes electorate % Anglesey 2,329 32,619 7.1 Blaenau Gwent 5,488 48,948 11.2 Bridgend 8,694 97,167 8.9 Caerphilly 11,428 124,662 9.2 Cardiff 29,116 238,017 12.2 Carmarthenshire 9,599 122,589 7.8 Ceredigion 2,272 47,078 4.8 Conwy 6,598 80,551 8.2 Denbighshire 9,349 64,863 14.4 Flintshire 6,586 93,639 7.0 Gwynedd 3,205 49,715 6.4 Merthyr Tydfil 2,442 42,803 5.7

Monmouthshire 7,250 66,465 10.9 Neath & Port Talbot 18,625 101,588 18.3 Newport 16,823 104,505 16.1 Pembrokeshire 4,567 82,433 5.5 Powys 3,285 40,442 8.1 Rhondda/Cynon/Taff 23,383 157,418 14.9 Swansea 20,104 176,966 11.4 Torfaen 8,235 59,657 13.8 Vale Of Glamorgan 4,791 89,230 5.4 Wrexham 8,274 88,922 9.3 Total 212,443 2,010,277 10.6 Table 18 shows the number of postal votes returned prior to the election and those that were returned late and were therefore ineligible for the election. Percentage figures are first, the unadjusted turnout amongst postal voters and second, an indication of the proportion that tried to vote, successfully and unsuccessfully. There are some authorities that either did not receive any late postal votes or did not record the numbers received after the closing date for receipts of postal votes had passed. Overall, unadjusted turnout amongst postal voters was 65.3%, rising to 66.0% if we include those whose forms were either delayed by the post or were returned late. There were three authorities, Pembrokeshire, Merthyr Tydfil and Anglesey, where more than three quarters of those with a postal vote returned it before the deadline. There were no authorities where fewer than half the postal votes issued were returned. Table 18: Postal voter turnout Postal issued Returned before election % Returned late % Anglesey 2,329 1,752 75.2-75.2 Blaenau Gwent 5,488 3,674 66.9 60 68.0 Bridgend 8,694 4,945 56.9 113 58.2 Caerphilly 11,428 7,031 61.5 166 63.0 Cardiff 27,988 18,137 64.8 70 65.1 Carmarthenshire 9,599 6,114 63.7 74 64.5 Ceredigion 2,272 1,581 69.6 22 70.6 Conwy 6,598 4,839 73.3 21 73.7 Denbighshire 9,349 5,956 63.7-63.7 Flintshire 6,586 4,740 72.0 66 73.0 Gwynedd 3,126 2,309 73.9-73.9 Merthyr Tydfil 2,442 1,889 77.4 27 78.5 Monmouthshire 7,250 4,911 67.7 57 68.5 Neath & Port Talbot 18,625 12,035 64.6 61 64.9 Newport 16,823 8,809 52.4 219 53.7 Pembrokeshire 4,535 3,509 77.4 44 78.3 Powys 3,285 2,082 63.4-63.4 Rhondda/Cynon/Taff 23,383 15,683 67.1 204 67.9 Swansea 20,104 13,229 65.8 133 66.5 Torfaen 8,235 5,585 67.8 69 68.7 Vale Of Glamorgan 4,791 3,319 69.3 29 69.9 Wrexham 8,274 5,710 69.0 50 69.6 Total 211,204 137,839 65.3 1,485 66.0

Rejected postal votes Postal votes may be rejected before the count because they breach one or other procedural rule, for example, the declaration of identity is unsigned. Data were collected on the number (but not the reason) of rejected postal votes and these are shown in Table 19. In three authorities no information regarding rejected postal votes is available and so the overall figure of 5.0% rejected, one in twenty, is likely to be a slight under-estimate. Relative to the number of votes rejected at the count the proportion of postal votes rejected is rather high. In both Rhondda/Cynon/Taff and Newport the proportion of rejected postal votes is 9.3% (almost one in ten) whereas in Monmouthshire the figure is below 2%, or less than one in fifty. Table 19: Rejected postal votes N= % Anglesey* - - Blaenau Gwent 183 5.0 Bridgend 217 4.4 Caerphilly 343 4.9 Cardiff 454 2.5 Carmarthenshire 443 7.2 Ceredigion 61 3.9 Conwy 383 7.9 Denbighshire* - - Flintshire 146 3.1 Gwynedd 120 5.2 Merthyr Tydfil 96 5.1 Monmouthshire 76 1.5 Neath & Port Talbot 876 7.3 Newport 818 9.3 Pembrokeshire 222 6.3 Powys* - - Rhondda/Cynon/Taff 1,465 9.3 Swansea 519 3.9 Torfaen 175 3.1 Vale Of Glamorgan 93 2.8 Wrexham 259 4.5 Sum 6,949 5.0 * Either no postal votes were returned late or data are not recorded Postal and in-person voters It is possible to compare the unadjusted turnout amongst postal voters with that for electors who opted instead to vote at a polling station. Table 20 calculates the unadjusted turnout for both postal and in person voters. Amongst postal voters turnout was 65.3% overall but for polling station voters it was 39.0%. Put another way, postal voters are 1.6 times more likely to vote than those electors whose vote is cast at a polling station.

The third column of Table 20 shows the percentage point difference between the two calculations, which across the country as a whole was over twenty-six points. The most extreme difference in turnout between the two voting methods was in Merthyr Tydfil where postal voters were twice as likely to vote than electors required to attend a polling station. Smaller differences, though still above thirty percentage points, could be found in Flintshire and Conwy. By contrast, in Powys the gap between postal and in person voters was not large and differences of a similar scale can be found in Ceredigion, Newport and Carmarthenshire. The final column in Table 20 describes the ratio of polling station voters to those voting by post. For all authorities the ratio is 5.1:1 in favour of in-person voters but there are some marked differences in the application for and casting of a postal vote. In Ceredigion, for example, in-person voters out-number postal voters by fifteen to one and in both Pembrokeshire and Vale of Glamorgan the ratios are eleven to one. However, in Neath and Port Talbot, Newport and Torfaen the ratio between postal and in person voters falls to just one to three. We have no data that would help to explain why there are such marked differences between local authorities. Table 20: Turnout for postal and in-person voters turnout postal turnout in person diff ratio of in person to postal voters Anglesey 75.2 47.5 27.8 8.2 Blaenau Gwent 66.9 43.4 23.6 5.1 Bridgend 56.9 31.3 25.6 5.6 Caerphilly 61.5 38.3 23.2 6.2 Cardiff 64.8 35.0 29.8 3.9 Carmarthenshire 63.7 46.1 17.5 8.5 Ceredigion 69.6 53.0 16.6 15.0 Conwy 73.3 35.3 38.1 5.4 Denbighshire 63.7 41.2 22.5 3.8 Flintshire 72.0 33.5 38.4 6.2 Gwynedd 73.9 45.9 28.0 8.9 Merthyr Tydfll 77.4 38.0 39.4 8.1 Monmouthshire 67.7 46.0 21.8 5.5 Neath & Port Talbot 64.6 38.7 25.9 2.7 Newport 52.4 35.2 17.2 3.5 Pembrokeshire 77.4 49.7 27.7 10.9 Powys 63.4 47.8 15.5 8.5 Rhondda/Cynon/Taff 67.1 37.9 29.2 3.2 Swansea 65.8 33.2 32.7 3.9 Torfaen 67.8 35.7 32.1 3.3 Vale Of Glamorgan 69.3 43.0 26.3 10.9 Wrexham 69.0 37.0 32.0 5.2 Total 65.3 39.0 26.2 5.1 Another method for examining the relationship between postal and in-person voting is to compare the unadjusted turnout amongst each type of voter in every division where

an election was held. Figure 1 shows electoral divisions organised in rank order of postal vote turnout (unadjusted) and plotted alongside that line is the turnout of inperson voters in the same division. Had the two figures been closely correlated then we might have expected a more or less constant sized gap between the two lines. However, since the two lines are not parallel it follows that there is only a weak correlation between the two measures of turnout. This suggests that turnout amongst postal voters in a division is not necessarily the best guide to how many people might vote in person in that division or vice versa. 120 100 80 60 Percentage turnout 40 20 0 Postal voters In-person voters Divisions Figure 1: Postal and In-person voters by electoral division Conclusions Although Independent councillors have all but disappeared from many parts of English local government they continue in Wales. While the Labour party remains dominant it does not have the same prominence in local government as it does in parliamentary elections. Both Plaid Cymru and Liberal Democrats have pockets of strength but the Conservative party remains as the fourth largest political party. There was some modest success also for smaller parties. Wales remains notable for the relatively large number of uncontested council seats but when contests do occur there is often no shortage of candidates from which to choose, although the pattern of competition may not resemble that which is presented to electors at a general election. Labour won four times the number of seats captured by the Conservatives and around three times more than the Liberal Democrats and Plaid Cymru. To a degree the distribution of seats is skewed in favour of the dominant party but the extent of electoral bias at the aggregate level is not great. Indeed, the index of proportionality compares favourably to the outcome found in some elections conducted by PR. This would seem to be largely because Labour s main competitors restrict their challenge to areas where there is at least some prospect of victory.

Although Labour emerged on top in terms of both vote and seat share this was not a particularly successful election for the party. Its vote share, 30.5%, and seat share, 37.9%, were the lowest at any local election since 1976. By contrast, the steady decline in support for Independents appears now to have reversed 2004 saw the highest vote for non-party candidates since the early 1970s. The Liberal Democrats too crossed an important threshold, safely defending a vote share which is more than twice as large as anything achieved during the 1990s. The Conservative party, however, still polls below its support of two decades ago. The coincidence of local and European elections demonstrated further the loosening of partisan ties amongst parts of the electorate. The United Kingdom Independence party captured more than one in ten votes at the European election but did not field local election candidates. The converse is true for Independents, capturing more than one in five local votes. Naturally, electors with these European or local preferences would have had no choice but to vote for another party or otherwise not use their vote. Given that turnout at the two elections was virtually identical it is clear that very few voting opportunities were spurned, rather that electors switched support according to the contest under consideration. Although the proportion of uncontested seats remains comparatively high the trend continues downwards. In 2004 the proportion of contested seats was 88% of the total; at the previous election it was 78% and in 1991 it was just 63%. Moreover, there is a concentration of uncontested seats in a handful of local authorities and if the figures are recalculated without these then the pattern of contestation improves dramatically. When seats are contested there does not appear to be a shortage of candidates willing to challenge for office. Women accounted for slightly more than one in four candidates, a modest increase compared with five years before. In general, women had a slightly lower chance of being elected than men 21.8% of successful candidates were women. Incumbent councillors featured prominently. Three in ten candidates was a councillor seeking re-election. In three out of four divisions the voters had an opportunity to return a councillor for a further term of office. Almost eight in ten councillors elected in 2004 had held a seat on the previous council. Notwithstanding the proportion of uncontested seats it remains true that when local elections are held then electors in Wales appear more likely to vote than do their counterparts in England. That said, turnout at 41% fell to an all time low. In 1999, of course, local and Assembly elections coincided, driving turnout higher as a result. In 1995 the newly constituted unitary councils held their first elections. But at other, less significant times, local turnout has been higher than that seen in 2004. But it is wrong to look at this phenomenon in isolation. Turnout has been falling for all types of election and the factors contributing to that decline affect a wide range of institutions and not just local authorities in Wales. As well as fewer people voting it is also true that more and more electors are opting to vote by post. Although historical data for local elections are difficult to find it is certainly true that the more than one in ten electors who were issued a postal vote in 2004 is a record. Interestingly, there appears to be considerable variation between local authorities in the level of postal votes issued. It is impossible to say whether this

reflects local authority policy or is a product of variable efforts made by party activists to persuade supporters to register by post. There can be no doubt about the outcome however postal voters are more likely to vote than are those electors who remain registered to vote at a polling station. Approximately two in three postal voters cast a ballot although more than one in twenty of these votes were subsequently rejected for some breach of procedure. Fewer votes are rejected that have been cast in person but the turnout amongst such voters is lower than amongst postal voters. Of course, polling stations remain the favoured option for voters but the trend is certainly towards greater postal voting.