Main menu

Online publications

Other

The Arbuthnott Commission
Submission From The De Borda Institute
March 2005

Notes and abbreviations for this article are at the bottom of this page.

Abstract
In a democracy which aspires to be plural, any decision-making voting procedure should be preferential, and any electoral system should be both preferential and proportional. The British government imposed PR-STV on NI as part of the 1920 Settlement; it was quickly replaced by Stormont but re-introduced under direct rule. It has worked well. But it could be better. The paper outlines some of its merits and demerits, and suggests some improvements.

Introduction
Firstly, a few general comments:

i) It is at least unfortunate that, in some voting systems, voters use a cross, r. As children, we received this mark when we got our sums wrong. It would be better, surely, in first-past-the-post elections and two-option referendums, if voters were asked to use a tick, 4, or a plus :, or even the number, 1.

ii) In the system you call "first past the post, or 'simple majority'" (para 3.8), you rightly put 'simple majority' into inverted commas, presumably because, in many contests, the winner does not get a majority at all but only the largest minority. But why don't you put "post" into commas as well? After all, as Professor Sir Michael Dummett has pointed out, there isn't one.1 In a straight two-candidate contest, the winner needs 50% +1 of the votes. In a three cornered contest, victory may depend on only 33% + 1. When there are four candidates, as Scotland knows only too well, the 'post' may be but 25% + 1. And the world record, as far as I know, was a candidate in Papua New Guinea, who won an election on only 6.3% of the vote!2

iii) Is such a result fair? Maybe not. For maybe 93.7% of the voters thought this particular candidate was the worst!

iv) And is it fair to insist that a voter may cast only one preference? One candidate good, other candidates bad? One candidate, and one party, right, and all right, about only everything? All the other candidates, and all the other parties, only always wrong?

Any voting system which so restricts the voter's powers of expression is obviously not fair! In other words, voting should be preferential.

v) We welcome your "commitment to maintain the benefits of proportional representation", (para 1.5).
vi) In places like Northern Ireland and Bosnia, however, and maybe Scotland as well, there is also a need for a preferential procedure. We are often told that the democratic process is part of the peace process. Sadly, however, the democratic process is not always 'peace-ful', {see para d) below}. This certainly applies to any 'first-past-the-post' system where the voter, in effect, votes 'for' someone and, implicitly, 'against' every other candidate.

vii) It also applies to decision-making, however, because, unfortunately, nearly every political question, no matter how complex, is eventually but invariably reduced to a dichotomy. Such was the case in Northern Ireland in the 1973 border poll; it was worse in the Balkans, where "all the wars in the former Yugoslavia started with a referendum"3 But such was also the case with the Jenkins Commission, and this despite the fact that his commissioners had visited New Zealand which, in a 1992 poll on the electoral system, had held a five-option referendum.4 We trust, in your own deliberations, that on all points of contention, any conclusions will be expressed, not in terms of "one option good, other options bad", but rather as your collective preferences.

viii) Indeed, if any multi-option decisions are resolved by a multi-option preference vote, you could express your collective opinions in terms of average preference scores. The procedure is the modified Borda count, (MBC).5

PR-STV

So now to the body of this submission, a discussion on the way PR-STV operates in Northern Ireland, and how it might be improved.

a) Firstly, it is undoubtedly true that using two different electoral systems, with two different voting and counting procedures, causes confusion.6 The same is true in Bosnia-Herzegovina, where AV was used (with 1-2-3 voting) alongside a PR-closed list system (with crosses). This could perhaps be alleviated a little if, as implied above, {para i)}, plurality votes were marked by a tick or a 1. But better change the system: 'first-past-the-post' is hopelessly unfair. AMS is an improvement, of course, but it is still, fundamentally, a two-party system or, as in Germany, one which tends to produce two big parties plus a couple of little others.7

b) PR-STV is a relatively high threshold voting system. In a 3-seater, the quota is 25% + 1. In a 4-seater, it is 20% + 1. And in a 6-seater,8 it's about 14% + 1. Proportionality can be improved, however, by using a top-up system. AMS, after all, is a top-up version of 'first-past-the-post'. And a rather unusual sort of top-up is used with PR-STV in Malta.9 Your own proposals, therefore, - PR-STV in 3/4-seaters - involve thresholds which, for a plural democracy, are too high. In such a scenario, a top-up is essential.
And given the need for both a proportional and preferential system, it would seem logical to suggest that any top-up should be based, not just on a count of the 1st preferences, but rather on, say, the first three 'party preferences'.10
It should also be noted that, for obvious reasons, the larger political parties prefer smaller constituencies, and sadly, in the Republic of Ireland, over the years, "Average constituency size was steadily reduced, to the clear benefit of the larger parties."11 A minimum threshold is essential, lest…

c) … "the power of the party political machine to be strengthened at the expense of fairness to the citizen," (para 1.4). The way political parties use PR-STV in NI, however, means that fairness is often sacrificed. In theory, the voter is entitled to cast his/her preferences for whomsoever, across gender, across party, and across the sectarian divide. Happily, many do.
Unhappily, however, some political parties play the system. They field as many candidates as they reckon they have quotas.12 Then, in order to ensure that all of their candidates get roughly the same 1st preference count, they issue "instructions" to the voters, telling those in one part of the constituency to vote, say, A-B-C; in another part, B-C-A; and in the third sector, C-A-B; and they do this by issuing dummy ballot papers to the voters, as the latter pass though a 'gauntlet' of party activists at the entrance to the polling station.13 This could quite easily be obviated, at least for the first part of the day, by arranging the ballot paper in random, as opposed to alphabetical order.
As an electoral system, then, PR-STV allows voters to vote across the divides, but it does not necessarily encourage them so to do. Furthermore, the bigger parties tend to ask their voters to vote DUP or SF 1-2-3 or whatever, and in a sectarian society likes ours, it would be rather naïve to suggest that PR-STV actually encourages inter-party co-operation, as is perhaps the case with AV in Australia.
In the Republic, meanwhile, and partly because of the high threshold and the absence of a top-up, there has been a tendency for the parties various to nevertheless fall into two opposing camps: FF and the PDs on the one hand, FG, GP and Labour on the other, (while SF is still out in the cold, of course, for other reasons). This tendency to split the country into two is, of course, regrettable, not least because it somewhat negates the very purpose of having a PR system. But such are the dangers of having a high threshold.

d) In contrast to PR-STV, the quota Borda system (QBS)14 actually encourages the voter to vote across the divides. Indeed, QBS could be regarded as a 'peace-ful' system, {para vi) above}, and the very antithesis of plurality voting. The latter disallows preference voting. QBS, however, gives those who participate fully in the democratic process - i.e., those who express their preferences, those who cross the divide, those who vote for both a Catholic and a Protestant, those who vote for a Catholic and a Moslem and an Orthodox - a correspondingly full influence on the result. In effect, the voter can regard the election as a real part of the peace process, and treat his/her own vote as a positive act of reconciliation.

e) In a PR-STV count, the vote may be transferred, literally, from, say, a Protestant candidate to a Catholic one. In PR-STV, therefore, the voter may vote for both a Protestant and a Catholic, but in the count, it may be treated as a vote for one or the other. In QBS, it is 'and'.

f) PR-STV remains a very good system. There are, however, a number of other disadvantages. For example, the candidate who is the perfect compromise candidate - someone like Ante Markovic, to take a 1990 Bosnian example - could be the 2nd preference of everybody but the 1st preference of none. In which case, he could get a first round score of zero and could well be eliminated. (An MBC top-up based on the voters' first 3 'party preferences' - see footnote 10 - would ensure the system, overall, could not produce such an unfair result.)

g) Furthermore, PR-STV can be very capricious. Consider, for example, the situation when, with but one seat remaining, 100 voters have preferences C-B-A, 101 have preferences B-C-A and 102 have preferences A-C-B. By any understanding, C is the most popular of the three; yet C is the first to be eliminated. Such anomalies could again be obviated by a "three-party preference" top-up, and eliminated in QBS. It should also be pointed out that PR-STV is not monotonic, unlike any Borda count, BC or MBC.

h) Because the electoral law is itself incorrect, PR-STV counts are sometimes counted improperly; this applies to both jurisdictions, Ireland North and South. When candidates are eliminated because their totals are too small, their votes are transferred as required, or, if no further preferences have been indicated, rendered non-transferable. The numbers involved, however, are relatively tiny. Nevertheless, this might mean, at the end of the day, that the final candidate to be elected falls short of the quota and is elected by default. But that is OK.
When a candidate is elected because his/her total of 1st preferences is in excess of the quota and the surplus is transferred, the numbers of votes involved may be considerable. Now in Northern Ireland (unlike the Republic), an examination takes place of all the candidates 1st preferences. Let us say the successful candidate, W, has a surplus of 300 votes, and that 50% of the voters voted W-X, 25% W-Y and 25% W-Z. If such is the case, then the surplus is divided accordingly, X gets 150 votes, while Y and Z both get 75. Which is fair enough.
If, however, 50% voted W-X, 25% W-Y and 25% were no-transferable, then what sometimes happens is that the 300 is divided in the ratio of the transfers, so X gets 200 votes and Y gets 100. In other words, some voters' votes are transferred, without the voters' knowledge let alone their consent!
The reason why is probably obvious: if large numbers of votes were non-transferable, then maybe many candidates would be elected by default and not just the last one, in which case the whole basis of PR-STV, the quota, would not be seen to be functioning. We don't think the law, originally drafted as part of the 1920 Settlement, is malicious; nevertheless, this effect of the mathematical rules, if and when it happens, should be regarded as an infringement of the voters' human rights.15 To change this anomaly would actually involve a simplification of the law.

i) Because PR-STV relies entirely on the quota, the benefits of a PR system in society at large are not as cohesive as they might be. Basically, in any sectarian election, those who are standing need to get only a quota. If elected, like Nigel Dodds (DUP) or Gerry Kelly (SF), for example, they then say they represent North Belfast. In practice, however, they represent the ghetto of their quota. At best, neither bothers to canvass the other area. At worst, they 'slag off' each other, and he who 'slags' the more often benefits at the expense of the more moderate (yet still perhaps sectarian) candidates. Meanwhile, as if life wasn't difficult enough already, the totally non-sectarian candidates have to try to canvass both areas. Again, an MBC top-up would tend to obviate this disadvantage.16

j) Once elected, those concerned tend to confine their operations to their own particular areas. That is not to say that a proportional election system was not absolutely essential for NI. We were also lucky to have imposed upon us a form of PR in which the voters themselves determine the basis of proportionality. In PR-list systems, proportionality is based only on the parties. In PR-STV and QBS, it is the voters who decide, and if, for example, a quota of voters casts their 1st and subsequent preferences for anti-GM food candidates, then, sure enough, an anti-GM candidate will get elected. As far as Northern Ireland is concerned, therefore, such a system allows for the day when sectarianism is no longer such a potent force in the province.

k) Sadly, on decision-making, the Belfast Agreement still insists on dichotomous forms of voting: either a consociational vote on certain 'key' issues in the Assembly, or (amazingly) a straight majority vote on that which is, for many, the most 'key' subject of all - the question of the constitutional links. As suggested above, we believe a non-majoritarian form of preference voting would be fairer.
May we therefore ask the Arbuthnott Commission to resolve any of its own multi-optional questions by preference voting, if only for the sake of those who live in conflict zones.

Conclusion

If fairness for the voters is the priority, there should be only one system of voting. And only one system of counting. The electoral system should be both preferential and proportional, and the overall threshold of proportionality should be of the order of 5%, but certainly not much higher. Accordingly, we would recommend a 6-seater form of PR-STV (or, better still, QBS) + top-up (an MBC count of the voters' first three 'party preferences'), in which the voter casts his/her preferences on just one ballot paper.17 The count would take place in two stages: in the constituencies, and then in regional top-ups.

This electoral system, hopefully with a local/regional top-up, could be used for both the local councils and the Scottish parliament. Like NI, Scotland could also use PR-STV, now with a national top-up, for the Euro-elections. And one day, when fairness for the voter eventually comes to Westminster, the same system could apply to general elections.

Notes

1 Principles of Electoral Reform, Michael Dummett, OUP, 1997, p 39.

2 The International Handbook of Electoral System Design, IDEA, 1997, p 42.

3 Oslobodjene, 7.2.1999, p 11.

4 Defining Democracy, Emerson, The de Borda Institute, pp 119-120. This Institute also made submissions to the Jenkins Commission, one written and one in a public hearing, where we argued that any referendum on such a multi-optional question should itself be multi-optional.

5 See the author's The Politics of Consensus. See also footnote 10.
Reforming the House of Lords, for example, is obviously a multi-option question. Indeed, they had five options. Lord Desai suggested a "rankings" system, which is, of course, a BC; (Hansard, 22.1.2003). But no. Politicians are obsessed with dichotomies. They took five majority votes, lost the lot, and then said there was a crisis. Crazy!

6 It was even worse in 1996, when the British government introduced a closed list PR system for the Forum elections. The d'Hondt procedure is pretty accurate when used in large multi-member constituencies. The Forum elections, however, involved only five-member constituencies - almost a world record for smallness in d'Hondt counts and therefore a possible world record for guaranteed anomalies… which, sure enough, and as predicted, occurred.

7 The choice of electoral system has a huge impact on what then becomes the democratic system. 'First-past-the-post' leads to a two-party state; AMS tends to result in 2 + 2; a low-threshold form of PR, like the Dutch PR-list system with a threshold less than 1%, tends to produce a multi-party state. And so on. Patterns of Democracy, Arend Lijphart, Yale, 1999.
It must be emphasised, of course, that another major cause of both the two-party system and the tendency of multi-party societies to devolve into two-party blocs, is the western obsession with binary voting, the widespread belief in what is, in fact, the most inaccurate measure of collective opinion ever invented, the 2,500-year-old majority vote.

8 NI Assembly elections use 6-seaters. In our local councils, the constituency sizes vary somewhat, usually between 6 and 3. While in the Euro-elections, the entire province is just the one 3-seater.

9 Defining Democracy, p 48.

10 The GP suggested such a system for use in the NI Assembly - see "A Green Party/Comhaontas Glas Discussion Document". Copies are also available from the de Borda Institute.
One 'party preference' is defined as one or more preferences for just one party. If, then, a voter votes 'UUP-1, UUP-2', he only expresses one party preference. If another votes 'UUP-1, SDLP-2', she expresses two party preferences. If a third votes 'UUP-1, UUP-2, SDLP-3', he also expresses two party preferences. And if a fourth votes 'UUP-1, GP-2, SDLP-3, UUP-4', she expresses three party preferences.
A BC proceeds as follows. In a vote on n options, a 1st preference gets n points, a 2nd preference gets n-1 points … and so on.
In an MBC, in a vote on n options, a voter may submit a partial vote by voting for only m options. In which case, a 1st preference gets m points, a 2nd preference gets m-1 points, and so on.
Accordingly, in an MBC top-up:
he who votes UUP-1, UUP-2 gives the UUP 1 point
she who votes UUP-1, SDLP-2 gives the UUP 2 points and the SDLP 1 point,
he who votes UUP-1, UUP-2, SDLP-3 also gives the UUP 2 points and the SDLP 1 point,
while
she who votes UUP-1, GP-2, SDLP-3, UUP-4 gives the UUP 3 points, the GP 2 points, and the SDLP 1 point.

11 A New Electoral System for Ireland, Michael Laver, The Policy Institute, 1998, p 7. In 1923, Ireland even had one 9-seater, in Galway. By 1969, however, the total picture had been reduced to 2 five-seaters + 14 four-seaters + 26 three-seaters, and it was still getting worse when, thankfully, an independent commission took over.

12 This is actually an advantage, and both PR-STV and QBS have this property. Other systems, like PR-list, do not, and Bosnia was a classic example where tiny parties nevertheless produced huge lists of candidates, like virility symbols!

13 In many ways, the conduct of Northern Ireland elections falls short of international standards. See the author's article, Get Party Activists off the Backs of Voters, published in Fortnight, May 2004.

14 Details are on the de Borda website: www.deborda.org

15 Again, more info on the web-site.

16 An even better system would be the one already mentioned, QBS. Another system which is very attractive in this regard is used in Lebanon. See Defining Democracy, pp 57-58.

17 If a top-up was introduced into NI, the voting procedure would not change at all. For the voters, it is just as simple. For the returning officers, a little more complicated to count. And for society as a whole, more fair.

Abbreviations

• AMS additional member system
• BC Borda Count
• FF Fianna Fáil
• GM genetically modified
• IDEA Institute for Democracy and Electoral Assistance
• NI Northern Ireland
• PR proportional representation
• SDLP Social Democratic and Labour Party
• STV single transferable vote
• AV alternative vote
• DUP Democratic Unionist Party
• FG Fine Gael
• GP Green Party
• MBC modified Borda count
• PD Progressive Democrats
• QBS quota Borda system
• SF Sinn Féin
• UUP Ulster Unionist Party
  

Back to top