Public Elections: introduction of STV and AV.

STATES OF JERSEY

PUBLIC ELECTIONS: INTRODUCTION OF STV AND AV

Lodged au Greffe on 24th December 2013 by Deputy M. Tadier of St. Brelade

STATES GREFFE

2013   Price code: C  P.171

PROPOSITION

THE STATES are asked to decide whether they are of opinion

to request the Privileges and Procedures Committee to bring forward plans in time for the 2014 elections for the implementation of –

1. a  single  transferable  voting  system  (STV)  for  multi-member constituencies; and
2. an alternative voting (AV) system for single-member constituencies.

DEPUTY M. TADIER OF ST. BRELADE

REPORT

During their work on electoral reform, the Electoral Commission employed the expert advice  of  Dr. Alan Renwick,  University  of  Reading.  Dr. Renwick  is  a  reader  in comparative politics and Director of Postgraduate Research Studies in Politics at the University of Reading.

Part of his work for the Commission was to look into voting systems and to compare Jersey's  current  system  of  first-past-the-post  in  both  single-  and  multi-member constituencies with alternative methods of voting.

Dr. Renwick was remarkably clear in his advice on the changes that needed to be made –

There  can  be  no  justification  for  maintaining  multi-member  plurality  in Jersey in preference to STV. It is rarely possible for an electoral system expert to  give  such  a  definite  judgement:  in  most  cases,  one  electoral  system performs better on some criteria, while another performs better on other criteria; the final decision then depends on which of these criteria one values more. In Jersey's case, however, all the plausible criteria point the same way: STV performs better on all criteria.'

Similarly, for single seat constituencies he says –

It would clearly be advantageous to introduce AV also for the elections in single-member districts.'

An extract from Dr. Renwick's report is attached at the Appendix.

The  Electoral  Commission  was  obviously  convinced  by  these  findings  as  they recommended  on  page 37  of  their  final  report,  under  SUBSIDIARY RECOMMENDATIONS that –

A Single Transferable Vote System should be introduced in elections for Deputy in 2018 and should the Constables remain as members of the States, an Alternative Vote System should be introduced in respect of their election.'

As things currently stand, Islanders will be electing one Connétable , up to 4 Deputies and 8 Senators each. STV and AV are desirable anyway, but they become even more necessary when choosing 8 Senators if we are to stop it becoming a complete lottery', as some members have quite correctly called it.

The  answer  given  by  the  Electoral  Commission  as  to  why,  in  the  face  of  such seemingly compelling expert advice, they would wait until 2018 to adopt the new voting model was that it would be too confusing to the public to bring in this changes this quickly, especially when coupled with other changes.

I do not agree with this logic, personally. There is a strong case to suggest that it is an appropriate change to bring it with all the other reforms. Currently, it is unclear whether there will be any reforms for 2014, other than those already agreed.

However, on the subject of complexity¸ Dr. Renwick gives the following advice –

The only credible criticism of STV in the Jersey context is that it is complex. There is no doubt that the process of counting votes under STV is often complex. But ordinary voters do not need to understand that complexity in order to understand how to exercise their vote and why the election result is as it is.'

Finally,  Dr. Renwick  highlights  3 important  disadvantages  of  our  current  system (multi-member plurality), whilst making the case for STV –

First, it can seriously misrepresent opinion. If groups of voters tend to vote for  the  same  set  of  candidates,  the  largest  group  can  secure  all  the representation even if is in the minority of the population as a whole. The groups here might be partisan, but need not be: for example, they could be ideological or geographical.

Second, as a corollary of the first point, multi-member plurality can lead to large numbers of wasted votes, an effect that is likely to depress electoral turnout. There is clear evidence that greater proportionality in elections leads to  higher  turnout.  In  non-partisan  Jersey,  standard  measures  of proportionality have little meaning. But wasted votes are strongly associated with  non-proportionality.  Thus,  it  is  safe  to  surmise  that  Jersey's  non- proportional voting system harms electoral turnout.

Third, multi-member plurality can do a bad job of choosing the most popular candidates, as vote-splitting between candidates with similar constituencies can allow a less popular candidate through. Such problems are more likely to arise than under single-member plurality because of the greater number of candidates.

STV would significantly reduce each of these difficulties. It would be wholly compatible  with  Jersey's  non-partisan  politics:  it  is  used  in  many  non- partisan elections, such as trade union elections and elections within the Church of England.'

Given the clear and compelling evidence, I would argue that there is no valid reason not to change from our current system of voting to the suggested AV and STV systems.

I would add that it will be necessary to educate the public, the counters and members on the exact workings of the new system. I would envisage that we would invite Dr. Renwick to come and speak to members prior to the debate so members can ask him questions.

Financial and manpower implications

There will be consequential work for both the States Greffe and Law Draftsman's Office if this proposition is adopted. It is envisaged that these costs would be met from pre-existing budgets. It is also worth noting that there was also a £59,863 under-spend from the Electoral Commission Budget. This may be able to be redirected towards the implementation of this proposition.

APPENDIX

Extracts from: "The Jersey States Assembly in Comparative Perspective. A Report for the States of Jersey Electoral Commission"

Alan Renwick, University of Reading 9th August 2012

Summary of Principal Findings

The Electoral System

• It would be impossible for the Commission to fulfil its tasks completely without considering the electoral system itself.
• Like most other small democracies, Jersey uses a plurality electoral system. It uses single-member plurality (also known as first past the post') to elect the Connétable s and the Deputies in some parishes and districts. It uses multi-member plurality (also known as the block vote') to elect the remaining Deputies and all of the Senators.
• Most alternative systems would not be appropriate for Jersey. But the alternative vote (AV) offers a viable alternative for single-member contests and the single transferable vote (STV) presents a viable alternative for multi-member contests.
• The  choice  between  single-member  plurality  and  AV  is  one  of  limited significance, but AV is clearly (if only marginally) superior in Jersey's context.
• The  choice  between  multi-member  plurality  and  STV  is  important.  STV  is superior on all criteria: it would allow better representation of opinion, reduce the number of wasted votes (which could be expected to improves turnout), and make it more likely that the most popular candidates are elected.
• Thus, if the current structure of Deputies, Connétable s, and Senators is in broad terms maintained, AV should be introduced for single-member contests and STV for multi-member contests.
• There is a strong case for extending the use of STV to all the Deputies. A sensible reform would replace the current Deputies and Senators with a single class of member elected by STV in districts of three-to-five seats. Such a system could operate alongside the Connétable s.

Types of Electoral System

This final section turns to consider the core of the electoral system itself: the set of rules determining the nature of the votes that voters can cast and the mechanisms by which those votes are translated into seats in the legislature. Jersey currently uses plurality voting in all three parts of the electoral system. Connétable s are elected using single-member plurality (commonly, though misleadingly, known as first past the post'), as are the Deputies in the parishes and districts that elect one Deputy . Multi- member  plurality  (commonly,  but  again  misleadingly,  known  as  the  block  vote' system) is used to elect the Deputies in multi-member parishes and districts as well as the ten Senators.

While  consideration  of  the  electoral  system  is  not  explicitly  included  in  the Commission's terms of reference, it is impossible to consider whether the current arrangements – including the division into three classes of member – are appropriate unless the core of the electoral system is examined. In addition, the issue of low electoral turnout is one of the concerns frequently raised in Jersey, and reform of the electoral system offers the prospect of some improvement on this score.

Figure 5 briefly  outlines  the  main  categories  of  electoral  system  that  are  used  in legislatures around the world. While there are as many electoral systems in the world today as there are elected bodies, these systems can be classified into three broad families. Majoritarian systems follow the basic principle that whoever wins most support – either a relative majority (a plurality) or an absolute majority – should win all  the  spoils.  Proportional  systems  enshrine  the  principle  that  seats  should  be distributed in proportion to levels of support. Intermediate systems combine elements of both logics or use procedures that yield outcomes between the extremes.

Each of these families can be broken down in turn into a variety of narrower types. Even within these types, considerable variation can remain. For example, Jersey uses multi-member plurality to elect both the Senators and the Deputies in the two-, three-, or four-member districts, but the system operates very differently when used island- wide to elect ten officeholders compared to when it is used in, say, St. Lawrence to elect two. Similarly, proportional systems vary from barely proportional two- or three- member systems, as in Chile and parts of Spain, to highly proportional systems in, for example, Israel and the Netherlands, where the whole country forms a single district.

Looking across the democratic world as a whole, the commonest electoral systems are proportional systems. Among European sovereign democracies, for example, only France and the UK have no proportional elements in their national electoral systems, though several other countries – including Greece, Hungary, Italy, and Lithuania – have systems that fall into the intermediate family.

Figure 5. Types of electoral system

Among our sample of small democracies, the pattern is rather different: here, single- and multi-member plurality systems predominate. Figure 6 summarizes the systems in place across these polities. In part, the preponderance of majoritarian systems revealed in Figure 6 arises for reasons that are not interesting. Major electoral reform being rare,  most  former  British  colonies  and  current  overseas  territories  or  crown dependencies retain the plurality systems bequeathed to them by the UK. Our sample includes only British overseas territories: were those of France and the Netherlands also included, a somewhat different picture would arise. Furthermore, independent small democracies are disproportionately former British colonies. Thus, the pattern revealed in Figure 6 is in significant part no more than an artefact of the British legacy.

Nevertheless, two factors may tend to perpetuate the survival of majoritarian electoral systems in these polities. First, the combination of small populations and, in many cases, strong community attachments within the polity tends to favour single-member districts: if there is a strong desire that particular islands or parishes should have their own representation and those islands or parishes have populations numbered in the hundreds or low thousands, there is little space for districts large enough to justify several members. Second, several – though by no means all – if these polities have non-partisan systems in which elections based on party lists would be incongruous.

As in the previous sections, the preponderance of a particular system does not imply its desirability for Jersey. Some of the alternatives to the current plurality system can confidently be ruled out:

• Politics in Jersey being non-partisan, list-based systems – closed-, flexible-, or open-list  proportional  systems,  mixed-parallel  or  mixed-compensatory systems, or bonus-adjusted systems – would be incongruous. Though some in Jersey may advocate the development of a party system, it would be quite inappropriate to seek to force that precipitately through the design of the electoral system.
• Limited  vote,  single  non-transferable  vote,  and  Borda  count  systems  can represent different strands of opinion within the community effectively, but they can also leave candidates' success or failure dependent as much open tactical calculations as actual popularity.

This leaves the options of the alternative vote (AV) for single-member contests and the single-transferable vote (STV) for multi-member contests. These two systems are described in Appendices 1 and 2 (pp. 21–3 of this report).

The choice between AV and single-member plurality is one of marginal importance: rarely would anything significant be changed by AV's introduction, either in the election  campaign  or  in  the  outcome.  In  a  non-partisan  context,  AV  should  be marginally preferred: in contests with more than two candidates, it is slightly more likely  than  single-member  plurality  to  lead  to  the  election  of  the  most  popular candidate. That is why AV or multi-round systems employing the same logic as AV are used very widely in non-partisan elections – perhaps most notably, in internal elections within even those political parties, such as the UK Conservative Party, that strongly oppose AV's use in parliamentary contests between political parties. Still, this is not a reform that should spark any excitement on either side of the debate.1

1 For a detailed discussion of the comparison between AV and single-member plurality, see

Alan Renwick, The Alternative Vote: A Briefing Paper (London: Political Studies Association, 2011).

Figure 6. Electoral systems in small democracies

Rather more important is the choice between plurality and STV in the multi-member contests. Multi-member plurality has three important disadvantages:

• First, it can seriously misrepresent opinion. If groups of voters tend to vote for the same set of candidates, the largest group can secure all the representation even if is in the minority of the population as a whole. The groups here might be  partisan,  but  need  not  be:  for  example,  they  could  be  ideological  or geographical.
• Second, as a corollary of the first point, multi-member plurality can lead to large numbers of wasted votes, an effect that is likely to depress electoral turnout. There is clear evidence that greater proportionality in elections leads to  higher  turnout.2  In  non-partisan  Jersey,  standard  measures  of proportionality have little meaning. But wasted votes are strongly associated with  non-proportionality.  Thus,  it  is  safe  to  surmise  that  Jersey's  non- proportional voting system harms electoral turnout.
• Third, multi-member plurality can do a bad job of choosing the most popular candidates, as vote-splitting between candidates with similar constituencies can allow a less popular candidate through. Such problems are more likely to arise than under single-member plurality because of the greater number of candidates.

STV  would  significantly  reduce  each  of  these  difficulties.  It  would  be  wholly compatible with Jersey's non-partisan politics: it is used in many non-partisan elections, such as trade union elections and elections within the Church of England. STV is best suited  to  elections  for  up  to  around  seven  members:  the  voting  process  becomes complex for voters if the number increases much further. While it would be suitable for multi-member Deputy elections, therefore, it might, at present, pose challenges for the election of ten Senators. If the number of Senators is reduced, as planned, to eight, however, these difficulties would diminish. There is no evidence at all that voters find STV confusing or cumbersome where the number of seats per district is low: in both Ireland and Malta, the two countries that use STV for national lower-house elections today, the proportion of ballot papers that are spoilt is around 1 per cent of the total – exactly the same figure as in the UK. More information on the guidance that might be offered to voters to ensure that they understand how to use their vote under AV or STV is provided in Appendix 3 of this report (p. 24).

In a partisan context, STV could be criticized on the grounds that proportional systems, because  they  tend  to  produce  coalition  government,  harm  accountability  and government effectiveness. In the non-partisan context of Jersey, however, this argument does not apply.

STV  is  sometimes  also  criticized  for  weakening  political  parties.  In  Jersey's  non- partisan context, however, this again clearly does not apply. Those who favour the development of political parties might have concerns on this ground. But there is no clear difference between multi-member plurality and STV in the degree to which they create incentives for or against such a development.

2 See especially James W. Endersby and Jonathan T. Krieckhaus, "Turnout around the Globe:

The Influence of Electoral Institutions on National Voter Participation, 1972–2000", Electoral Studies 27 (2008), pp. 601–10.

All in all, therefore, there can be no justification for maintaining multi-member plurality in Jersey in preference to STV. It is rarely possible for an electoral system expert to give such a definite judgement: in most cases, one electoral system performs better on some criteria, while another performs better on other criteria; the final decision then depends on which of these criteria one values more. In Jersey's case, however, all the plausible criteria point the same way: STV performs better on all criteria.

Given that STV is desirable, it would clearly be advantageous to introduce AV also for the elections in single-member districts. Otherwise, voters would have to place an x' next to single candidate to elect some Assembly members while ranking candidates in order of preference for other members, which could create confusion.

Thus, if something like the current structure of Deputies, Connétable s, and Senators is retained, the electoral system should be reformed so as to replace the plurality system with AV in the single-member contests and STV in the multi-member contests.

STV also has advantages relative to AV: it fosters more accurate representation of opinion and reduces the number of wasted votes. As before, arguments that in most circumstances would work against STV – particularly relating to the disadvantages of proportionality – do not operate in a non-partisan context. Thus, simply looking at the merits of the electoral system, there is a clear case for replacing the current Deputies and Senators with a single class of Assembly member elected using STV in districts of three to five members. Such districts would need to span several parishes in some cases, but  the  interests  of  the  parishes  could  be  protected  through  the  retention  of  the Connétable s within the Assembly if that were thought desirable. If parish identities were considered to justify some malapportionment in favour of the smaller parishes, the apportionment of STV Assembly members could be done using the simple Sainte- Laguë method. If such malapportionment were not thought justified, the Connétable s could  be  retained,  but  the  STV  apportionment  could  be  conducted  using  the compensatory Sainte-Laguë method, thereby minimizing discrepancies.

Of course, whether a reform of this kind is preferred should depend on additional considerations, such as whether the Deputies and Senators are considered to perform usefully  different  roles.  Looking  simply  at  the  operation  of  the  electoral  system, however, such an arrangement would clearly be superior to the current rules. It would also be simpler than the current system and therefore less likely to cause confusion. It ought therefore to be given serious consideration.

Appendix 1: The Alternative Vote

The alternative vote electoral system (AV) is similar to the single-member plurality or first past the post' system in that it is used in districts that elect one member each. Under AV, voters can rank the candidates according to preference, placing a 1' by their favourite candidate, a 2' by their next most favoured candidate, and so on. In most of the elections where AV is used, voters can express as many or as few preferences as they wish. Australia uses a version of AV where a vote is valid only if all candidates are ranked.

In the first stage of the vote count, only first preferences are counted. If a candidate has more than 50 per cent of first preferences, that candidate is elected and the counting process is over.

If no candidate wins more than 50 per cent of first preferences, the candidate with fewest first preferences is knocked out of the race. The ballot papers in their pile are looked at again and added to the piles for the remaining candidates according to the second preferences that these voters have indicated. If a candidate has now passed 50 per cent of all votes, that candidate is elected. If still no one has met this mark, the process continues  until  someone  does pass  50 per cent or  until  there  are  only  two candidates left, in which case the one with more votes is elected.

The logic underlying AV can be seen from a simple example. Suppose that a club is planning to redecorate its meeting room and wants to choose a new carpet. Three carpets receive nominations from society members. A vote is taken to choose among these, with the following result:

Light Green  Dark Green  Light Blue Votes received  33  58  65

If single-member plurality is employed, the blue carpet wins. But is this actually the most popular carpet? If most society members are mainly concerned about whether the carpet is light or dark, then light blue is probably the most popular choice. But if they are  more  interested  in  colour,  it  seems  that  the  election  has  produced  the  wrong outcome: more members want a green carpet than a blue carpet, but the split in the green vote has allowed the blue carpet to win. We cannot know which is actually the most popular carpet just by looking at the plurality result.

AV avoids this. After the first round of counting, it is clear that the light green carpet has least support. It is therefore eliminated and we look at the second preferences of the people who voted for it. If most of them care primarily about colour, then the dark green carpet will win. If most care more about shade, then the light blue carpet will win. Either way, we can be confident that the result reflects what voters actually want.

The process is equivalent to an election with multiple rounds of voting in which the bottom candidate is eliminated after each round until one candidate passes 50 per cent. The difference is that AV allows voters to express all their preferences at once, rather than having to trudge to the polls multiple times.

Such systems – either AV or systems with multiple rounds of voting – are standardly used  by  societies,  trade  unions,  political  parties,  and  others  to  elect  individual officeholders.

Appendix 2: The Single Transferable Vote

The single transferable vote system (STV) applies basically the same logic as AV to an election in which multiple candidates are to be elected. As under AV, voters can rank the candidates in order of preference. The normal version of STV again allows voters to rank as many or as few candidates as they wish, while the Australian version requires them to rank all the candidates.

The first step of the counting process is to work out how many votes a candidate needs to secure election – the so-called electoral quota'. Under AV, as we saw, candidates need  to  pass  50 per cent  of  the  vote  to  secure  election.  The  logic  is  that,  once  a candidate has passed 50 per cent, it is clear that no other candidate can meet them. Similarly, if there are two positions to be filled, candidates have to pass a third of the vote to secure election: it is impossible for more than two candidates to pass a third of the vote, so we know that the candidates who achieve this are the most popular. The general rule is that candidates must reach what is called the Droop quota to secure election:

where v is the total number of valid votes cast and n is the number of seats to be filled.

The count begins by counting only voters' first preferences. If no candidate meets the Droop quota, then, as under AV, the bottom candidate is eliminated and the second preferences  of  that  candidate's  supporters  are  added  to  the  piles  of  the  remaining candidates.

The difference from AV is that, even after a candidate has met the quota, the counting process continues: there are still some seats to fill. One of the goals of STV is to ensure so far as possible that every vote should count equally. To achieve this, it is necessary to  redistribute  a  winning  candidate's  surplus  votes  to  the  remaining  candidates. Consider, for example, a case where one candidate captures two-thirds of all the first preferences in a four-seat district. This candidate has clearly won many more votes than were needed to secure election. If we do not redistribute the surplus, the outcome will be that one person will represent two thirds of the voters, while the remaining third of voters will determine the victors in three seats. In order to ensure that the favourite candidate's voters are fairly represented, therefore, the winning candidate's surplus votes (the votes they have won in excess of the quota) are transferred to remaining candidates according to the next preference marked.

For example, suppose that our club is planning three social excursions for the coming year and five possible venues have been suggested. The members rank the options in order of preference. Their first preferences are shown in the first row of Table 6, on the next page.

Using the STV system, the first step is to determine the quota needed for election. 100 votes have been cast and three venues are to be chosen, so the Droop quota is

So a venue needs 26 votes to be chosen. This makes sense, as 26 is the smallest number of votes that only three venues can secure: if the quota were 25, it would (just) be possible for four venues to hit this mark.

Table 6. A hypothetical application of STV

Kew  Blenheim  Tate  Alton  Legoland Gardens  Palace  Gallery  Towers  

First preferences  52  15  5  20  8 Transfer of Kew surplus  –26  +15  +8  +2  +1 Second round totals  Chosen  30  13  22  9 Transfer of Blenheim surplus  –4  +3  +1  +0 Third round totals  Chosen  16  23  9 Transfer of Legoland votes  +2  +7  –9 Fourth round totals  18  30

Chosen

It is clear from the first preferences on the top row of the table that Kew Gardens is, by some margin, the most popular choice. In fact, it has just over half of all the votes cast, and twice the Droop quota. In order to ensure that half of Kew's supporters have not wasted their votes, we need to consider their second preferences. We therefore take the surplus  of  26 votes  from  the  Kew  pile  and  redistribute  them  to  the  other  venues according to the voters' second preferences. (The simplest way to do this is to pick out 26 votes at random from Kew's pile. The most accurate way is to look at all of the Kew votes and to redistribute them all at the value of half a vote.)

This redistribution is shown in the second line. It turns out that most garden lovers also like country houses, and some like art galleries, while few are so keen on theme parks. The effect is that Blenheim, with 15 extra votes and 30 votes in total, now also passes the quota and is confirmed as the second venue to be chosen. Blenheim has a surplus of four votes, which is redistributed to the remaining venues. Again, the more sedate pleasures of the Tate Gallery secure more of these votes than the theme parks.

We have now reached the row labelled Third round totals'. One more venue remains to be chosen, but none of the three remaining venues has met the quota, so we eliminate the venue with fewest votes – namely, Legoland. As would be expected, if those who want to go to Legoland don't have that option, most choose the alternative theme park as their second preference. The transfer of these preferences pushes Alton Towers over the quota, so Alton Towers is the venue of the third excursion.

If we look at how the votes have panned out, we can see that the majority of the club's members prefer sedate excursions, but a significant minority would prefer something more lively. The STV system ensures that both these preferences are reflected in the final choice. If, by contrast, the multi-member plurality method currently used in Jersey elections had been employed, each member would have had three votes, and those preferring  sedate  pleasures  might  well  have  captured  all  three  slots.  That  may  be appropriate in some contexts. But if you want the outcome to reflect the spread of opinion among voters, it is not.

Such an example may seem rather frivolous. But it illustrates the sorts of preference flows that might exist between candidates in a non-partisan context. If it is considered important that the election outcome should reflect the spread of preferences in the electorate, then, in a non-partisan context, STV is a good way of achieving that.

Appendix 3: Guidance to Voters for AV and STV Elections

One of the concerns sometimes expressed about AV and STV is that they are complicated for voters to understand. There is certainly some complexity in the process of counting votes, particularly under STV. Voters do not, however, have to understand all the details of that counting process in order to understand how best to use their vote.

We can take the example of Scott ish local council elections. Since 2007, these have been held using STV in districts each electing three or four members. AV is used for by-elections where only one position is being filled. For the 2012 elections, the legally prescribed text at the top of the ballot paper was as follows:

[Number to be elected] of the candidates listed below will be elected. You can make as many or as few choices as you wish.

Put the number 1 in the voting box next to your first choice.

Put the number 2 in the voting box next to your second choice.

Put the number 3 in the voting box next to your third choice. And so on.[3]

In addition, local councils and the UK Electoral Commission produced a range of materials offer voters guidance on the election. The City of Edinburgh Council, for example, had a page on its website giving basic information and including links to further information.[4] The main link was to a booklet produced by the UK Electoral Commission, which gave slightly more information on the voting process but no information on the counting process.[5] There was also a link to a detailed description of the count, though this was lengthy and clearly not designed to be accessible to the average voter.

There is no evidence that voters in fact find STV confusing. As noted in the main text, the proportion of ballots cast that are invalid is around 1 per cent in both Ireland and Malta – the two countries using STV to elect their national lower or single chamber. This is exactly the same as the figure in recent UK elections. In Northern Ireland, all elections except Westminster elections are conducted using STV, and the share of spoilt ballot papers at recent elections has ranged from less than 1 per cent to fractionally over 2 per cent.[6]  There  is  some  evidence  that  this  figure  has  been  higher  when  several elections have been held simultaneously. This also occurred in Scotland in 2007, after which it was decided not to hold local council and Scott ish Parliament elections at the same time.