DD Papers
Elections in Northern Ireland: systems for stability and success

Draft, February 1998

This is the second of a series of working papers being published by Democratic Dialogue to work through otherwise apparently intractable problems associated with negotiating a settlement of the Northern Ireland conflict. Democratic Dialogue welcomes comment on its contents, which are not intended to be definitive but to stimulate constructive discussion and debate.

© Democratic Dialogue 1998

Acknowledgements

Democratic Dialogue is indebted to Carmel Roulston who undertook this research at short notice, and executed it with speed. We recognise her as the principle author. We also acknowledge Sydney Elliot, who was, as ever, on hand to offer invaluable advice.

Introduction

Electoral systems are the key variables in the political process in a democracy, because to a large extent they determine who gets what, when and how.

(Reeve & Ware, 1992)

Taking the view that electoral outcomes in Northern Ireland which include small parties, and more women endows the pro process which the electoral system governs with greater stability than would otherwise have been the case, this paper undertakes a review of electoral systems elsewhere which are targeted at:

(a) the inclusion of small political parties and

(b) increasing the number of women in elected positions representing political parties

It makes recommendations based on this review which could be beneficial to any future Northern Ireland electoral systems. We refer to a discrete part of these, viz. an assembly.

The paper firstly considers the purpose of elections, before describing the key elements of election systems, and their relationship with electoral outcomes. We note that a huge number of mechanisms for converting votes cast into seats won by parties and individual candidates exists throughout the world. We explore the technical detail of a number of these and then assess how they work in practice. We turn attention to how systems are tailored to include those groups who generally find themselves outwith the outcome. But it is not electoral systems alone which will achieve proportional representation for women. Parties are the gatekeepers to candidature, and we consider their responsibilities to women in this respect.

In conclusion, we suggest six principles which electoral systems in Northern Ireland should adhere to. We present possible parameters for such structures, and we sketch five specific systems which give practical expression to these principles and parameters.

Democratic Dialogue

February 1998

I. Introduction: studying elections

Voting in elections is for most people in modern liberal-democracies the most significant, indeed the only, form of participation in politics. Elections are a key link between citizens and policy-makers. They have both practical and symbolic importance in the many states which call themselves democracies, playing a vital role in the legitimisation of the political system and contributing to political stability and order. The composition of legislative assemblies is, in the last instance, determined by popular vote in elections; for this reason, political parties have paid increasing attention to electoral strategies, cultivating ever more sophisticated techniques for influencing, measuring and responding to public opinion. Indirectly, elections influence the composition of governments at local, regional and national levels and therefore have some bearing upon the policies of states. Even if we accept Schumpeter's view (1975) that in liberal democracies citizens choose between competing policy-makers rather than decide upon policies, it is hard to deny that accountability through the ballot box has the effect of concentrating the minds of political elites keen to be returned to office.

For all these reasons, every aspect of the conduct of elections attracts the attention of all those with an interest in politics and policy-making. A vast and growing literature has been created on electoral rules and institutions, while the conduct and outcomes of individual elections are studied and interpreted in considerable detail. Often in such studies it is the citizens' engagement with the processes which is the object of attention; in particular, in recent years, their shifting political allegiances and the extent to which these can be reflected in the available party system. As new political identifications begin to intersect with or replace older alignments and cleavages, the mechanisms for expressing and aggregating preferences have become more significant. Put simply, sometimes the electoral system is based on the expectation that a society is divided by social class or attitude into two roughly equal groups, which will be represented by competing parties or blocs of parties. In reality, social and attitudinal changes may have brought on to the scene new issues —such as the environment, or parties—such as the Green Party, which cannot find a space given the constraints of the system. Adjustments to the electoral system can, and have been, introduced with the aim of achieving certain specific outcomes: more parties in the legislature, fewer parties elected, more women and members of minority groups and so on. This can be illustrated by developments in a number of European states—both in the EU and in former Communist-led states—in recent years. In Britain, France and Italy, for example, the emergence of new parties or rejection of traditions associated with existing dominant parties have brought demands for changes to the electoral system. In the former Communist states, the framers of electoral laws have found themselves obliged to balance the objective of encouraging an open pluralistic system against the desire to limit both political fragmentation and the influence of the successors to the former ruling parties.

It has become widely accepted, then, that the electoral system is an important variable in the determination of how citizens participate and are represented. The details of the electoral system —from the structure of the ballot paper to the size of constituencies—can have an influence upon the behaviour of both voters and parties, favouring some choices and discouraging others. The exact nature of these effects is still open to debate, but a strong consensus has been reached on the importance of the electoral system. Sartori commented that the "electoral system is the most specific manipulative element in politics." (Lijphart, 1994b:2 ) Putting this more strongly, Blackburn (1995:1) tells us that the "crucial democratic link between politicians and people— or government and the governed—is the electoral system. The quality of that electoral system itself determines the quality of our democracy." It is not clear, however, that simple causal links can be identified between, for example, electoral system and party system. The 'law' that Duverger (1954) expounded—"the simple-majority single-ballot system favours the two-party system"—with its corollary that proportional representation favours multi-party systems, has come to seem less obvious. Other factors—the nature of political cleavages, size of territory, level of economic development—have come to be seen as influencing the nature of party-systems. Which aspects of the electoral system have what effects has also been the subject of much discussion, and there has been growing resistance to the idea that it is possible to establish a 'best possible' system. (Gallagher et. al , 1995, Reeve & Ware, 1992). Changing the rules may change some aspects of the nature of the process, but questions still remain as to whether absolute judgements can be made about how "democratic" any given system might be. Nevertheless, even if we accept that there is no perfect electoral system, we can still attempt to define the potential effects of key elements of the electoral system on the fortunes of political parties, and ultimately, on the policy agenda in any future representative assembly in Northern Ireland.

II. Key elements of election systems

To the general public, the key distinguishing feature between electoral systems is whether they can be described as 'proportional representation' or 'first past the post' ('majoritarian' or 'simple plurality') systems. Blackburn (1995:362) points out that, strictly speaking, proportional representation is "not in itself a system for elections, but rather a criterion upon which to evaluate the working of any one of a range of electoral systems which can be used for voting purposes. It is a principle or yardstick by which to test the degree of representative proportionality between citizens' votes and successful party candidates." What we can say is that some systems have been devised in order to achieve a closer match between votes cast for a party and seats won by it. As Gallagher observes, however, it is not accurate to see PR and plurality systems as polar opposites. "After all, even plurality systems are 'proportional' in their own way in that they award the seat(s) within each constituency to the party with the largest number of votes. The real opposite of proportional representation would be a 'perverse' system that awarded all the seats to the party with the fewest votes." (1995:275) The different systems, according to this view, can be seen as belonging to a spectrum, with different degrees of proportionality arising from the presence or absence of certain features. Some are, however, considerably more accurate in relating seats won to votes cast than others. The British simple plurality system, for example, has been described as so disproportional as to be "a distortion and falsification of democracy".(Blackburn, 1995:363) The most important features identified over the years by scholars such as Bogdanor (1984), Blais (1988), Carstairs (1980), Rae (1971), Gallagher (1991) and Lijphart (1994) are:

• assembly size,
• electoral formula,
• ballot structure,
• district magnitude and
• thresholds of representation.

There is continued debate about the relative importance of these elements and to what extent they should be viewed as independent variables, but all can be seen as having some impact upon the eventual outcome of elections.

Assembly size might seem to be the least significant factor, but Lijphart points out that while it makes sense for small countries to have relatively small assemblies, "when assemblies are made extremely small, the chances of proportional election results are severely reduced." In assemblies of over approximately 100 seats, differences in size appear to have little influence on proportionality, but smaller assemblies may be very disproportional. (1997:74)

Electoral formula and ballot structure, identified as being of great importance in Rae's influential work (1971) are part of the process of 'aggregating votes', that is drawing together the individual choices made by voters into collective outcomes. (Reeve & Ware, 1992: 83) Ballot structure refers to the range of choices which a voter can express: does she have more than one vote, can she cast votes for more than one party, how many preferences can she register and so on. The electoral formulas translate votes into seats. Plurality and majoritarian systems have relatively simple formulas - the candidate with the biggest number of votes wins the seat, even if he has not won the votes of a majority, or the candidate with a majority of preferences takes the seat. The various PR systems, which allow for a greater range of preferences and have multi-member constituencies, have more complex formulas and ballot structures. All PR systems have multi-member constituencies—this is a defining feature of PR, though a few plurality/majority systems have more than one representative—and district magnitude refers to the number of legislators elected from each district. In PR systems, as a rule, the greater the district magnitude the more proportional the system. Finally, threshold of representation refers to the percentage of votes required to have a chance of winning a seat. Sometimes this refers to a minimum percentage laid down by law in party list or mixed systems (e.g. in Poland, since 1994, only those parties winning at least 4% of the national poll will be allocated seats) and sometimes it refers to the effective percentage required in order to have a chance of winning a seat . In this case, it is then linked to district magnitude (the lower the district magnitude, the higher the effective threshold). In general, the higher the threshold the less proportional will be the outcome and the greater the number of 'wasted votes'. This underlines again the degree to which electoral systems are a continuum; in some PR systems a relatively high percentage of votes may be 'wasted', though not as many as is the case in plurality systems.

III. Common electoral formulas

An enormous variety of mechanisms for turning votes cast by citizens into seats won by parties and candidates exists throughout the world. As noted above, the adoption of one mechanism in favour of another, or the retention of a system in use for many years, may be intended to achieve a specific outcome or correct a perceived actual or potential imbalance. In the newly created states and democracies of Eastern and Central Europe, we have been able to witness the creation of systems in relation to certain values and objectives which the leading reformers considered important. In contrast to the one-party systems which had been overthrown, free and open party competition was widely agreed upon as a fundamental principle. However, this was constrained in some cases by the desire to limit the influence of the former ruling communist party and the concern to achieve stable and secure majorities in order to manage economic transformation. In some cases efforts were also made to revive or recreate pre-communist traditions and to learn from the mistakes of, or emulate the success of more established parliamentary democracies. With all these factors coming into play, Holmes notes that the systems adopted were often based on compromises between competing principles and have required subsequent adjustment and amendment. (1997:152) Electoral systems have become a current issue again in the UK, with discussions about the system to be used in the Scottish Parliament and Welsh Assembly, as well as reform of the electoral system for Westminster being investigated. In the latter case, pressures and doubts arise from the need to keep some of what has been valued in the old system—territorial link between constituents and representatives, for example—while removing its most obvious shortcomings—wasted votes, for instance.

If we limit ourselves to the British and East European cases, we can find examples of the most popular and respected formulas in use today.

For UK Westminster elections, of course, the system used is single-member plurality. This system, or a derivative, has also been employed in countries which were formerly British colonies. (New Zealand, however, recently adopted a PR system). This is a relatively straightforward system: voters, organised in territorial constituencies, are presented with a ballot paper containing a list of names of candidates, most of whom are attached to political parties. Voters choose one of these candidates only and the candidate whom the largest number of voters identify as their choice wins the seat, even if the largest number is a minority of the total number of votes cast. In the country as a whole, the party which wins the largest number of constituencies (again, even if this party has not won a majority of votes cast) will be the one to form a government. Voters are then potentially influencing two outcomes, namely, the result in their constituency and the result in the country as a whole. The criticisms of this system are well-known, but before we enumerate them it may be useful to outline some of the reasons for its survival. It is simple to operate, particularly for those counting the votes, it retains a clear link between representative and constituency and it allows local issues to predominate where there is strong feeling about them. It is argued that disproportionalities in any one constituency may be 'ironed out' over the country as a whole, thus 'safe' Labour seats balanced by 'safe' Tory seats, for example. The major disadvantages are that, by denying voters the opportunity to express a range of preferences rather than a straight choice, it leads to 'wasted votes', where a voter can feel that he or she has little effect on the outcome of the election, either at the local or national levels. The benefits of being able to highlight local issues and on occasion vote for independent candidates are offset by the strength of the party system in modern states. It strongly disadvantages smaller parties, especially those whose support is dispersed throughout the country, such as the Liberal-Democrats. Further, it can lead to a situation where a relatively small shift in public opinion can lead to a significant change in the composition of the parliament. The misgivings about this system are illustrated by comments on British elections in recent years, when both Conservative and Labour parties won disproportionate numbers of seats.

To remove some of these faults while keeping most of the benefits of the system, modified plurality or majority systems have been proposed and tried. In Australia, there is the alternative vote system, where voters elect a single constituency representative but are allowed to rank candidates in order of preference. When all first preferences are counted, provided no-one has a majority, lower-placed candidates are eliminated and their votes redistributed in accordance with second-preferences until one candidate has a clear majority over the others. In France, a similar principle is employed in the second-ballot system, where lower-ranked candidates are eliminated and the voters return to the polls to decide as to which they prefer among the remaining choices. In some opinions, the French system is fairer because voters have more information when making their second choice. Numerous variations on this basic formula are possible: sometimes voters can only offer two preferences, sometimes only the top two candidates go into the second stage and so on. It does not eliminate disproportionality across the whole country, but it does allow voters more influence over the range of choices offered to them. In this respect, it can be said to have some of the features of the US primary system, where party members take part in the process of selecting candidates.

The single-transferable vote system, which is also popular in English-speaking countries, also combines the features of a primary with an election. It allows voters to convey a lot of information about their preferences, and in particular to select among a range of candidates offered by the same party. Voters can also express support for more than one party, or for parties plus independents, on their ballot-papers. Although counting the votes is complex, this is also a relatively simple system to use in most contexts. Voters are presented with a list of candidates competing to win a variable number of seats in multi-member constituencies. They rank their preferences in order, and again are both influencing the choice of constituency member as well as the composition of the government. In counting the votes, the key factor is the 'quota' or proportion of the votes which any candidate must reach in order to be elected. In Ireland, the 'Droop quota' is used; the total votes cast are divided by the total number of seats plus one. It is possible to use other methods of calculating the quota; the 'Hare' system simply divides votes cast by seats, which produces a smaller 'effective threshold'. To be elected, a candidate must win this number of votes plus one and does not require any more votes than this. Consequently, the processes of redistribution involve both votes for those candidates who have been eliminated and the surplus votes of those who have been elected, so in theory, there are no 'wasted votes'. It is this system which the Liberal-Democrats favour for British Westminster elections and it does have certain advantages, including the likelihood that it would give them seats in parliament in proportion to their support in the country. Voters are given considerable control over the way their vote is used, they can influence the direction in which their party will develop, they can highlight issues which they consider important and it is much less disproportional than the single-member plurality system. However, this appears to be a system which might not work as well in larger constituencies and it can allow local issues to have too great a significance.

If STV is favoured because of the power it gives electors, party-list systems are weighted in favour of the parties. As with the other mechanisms described, there are many possible variations: in the degree of choice given to voters, in the size of the constituencies and in the precise formula used to calculate how seats relate to votes. The basic principle is that parties are fundamental to the representation of opinion and that party representation in the legislature should be closely proportional to party support in the country. Accordingly, parties offer voters lists of candidates, usually containing as many names as there are seats to be filled. In some countries there are lists for more or less large constituencies (Spain has seven-member, Finland has twelve-member constituencies), in others (Holland, for example) there is one nation-wide constituency. Voters then decide which party they prefer and seats are allocated to the names on the list (usually in the order decided by the party) in proportion to the support among the electorate. In some countries, voters can change the order of the candidates (usually to very little effect); in others (Switzerland, Luxembourg) electors can vote for more than one list, or can use more than one vote for the same candidate. However, even where an elector votes for an individual candidate, as in Finland, the vote cast may in some circumstances be counted as part of the overall vote for the party and be used to elect a candidate who might not ever have featured in that voter's preferences. In some systems, a second-tier of seats is reserved, allocated to parties on the basis of total votes cast to iron out any disproportionalities in the constituency rounds. The precise formula used to allocate seats in party list systems is based on the principle that as far as possible the average number of votes needed to elect candidates is the same for each party. According to the d'Hondt method, the total votes cast for each list are divided and redivided so that the seats go to the parties with the highest average of votes. Other 'highest average' methods are used where it is regarded as desirable to 'overcompensate' smaller parties. 'Largest remainder' systems involve dividing the total number of votes cast by the number of seats to be won and tend to be less likely to penalise smaller parties. Party list systems are usually very good at achieving proportionality (though this varies according to the size of the constituencies and the level of threshold), but can be said to sacrifice voter choice to some extent in favour of party control. Parties choose the candidates and can determine or strongly influence which candidates take seats by putting favoured candidates higher on the lists. Some decentralisation of control over such decisions can be achieved by allowing local party organisations to influence the lists, or by having primaries, as has happened recently in Israel. These procedures still leave control in the hands of party activists, however, limiting the voters choice to a degree.

Finally, there are additional member systems. The best known of these is the German system, variants of which have been adopted by some Eastern European states, including Hungary and Bulgaria. To some extent the law-makers in these states were influenced by similar considerations to those which applied in Germany, aiming to find a proportional system which would give voters a wide choice but would also avoid the multiplication of smaller parties which many saw as responsible for the paralysis of the Weimar parliamentary system. In Germany, the country is divided into 328 single-member constituencies, where candidates are elected using a plurality system. On the ballot-paper, however, the voter has another vote which is cast for a party list, and which results in the election of a further 328 members. The object of this second vote is to compensate for disproportionalities at the constituency level; the number of seats going to party list candidates depends upon how many they have won at constituency level and on how proportional that number is compared to overall support for the party in the country as a whole. 'Additional' seats are awarded to parties who have won fewer constituency seats relative to their overall share of the vote. It sometimes happens that a few extra seats are required in the assembly to ensure that no party receives fewer seats than their share of the vote indicates as necessary. The systems proposed for the Scottish parliament and Welsh assembly are similar to the German one. All sorts of variations are possible; higher or lower thresholds for the party list vote; fewer members elected on this basis, regional or nation-wide lists. The advantages of this system are that it allows voters to express complex preferences and is reasonably proportional. Compared to other systems, however it is complex both for voters and for counters of votes.

IV. Electoral systems in practice

Broad generalisations are often made about the differences between PR and 'first past the post' systems. These are grounded in fact to some extent, though they tend to be qualified after closer examination. To return to Duverger's 'law', this is based on the assumption that the 'fairness' of PR systems—the closer correlation of seats won by parties to votes cast for them— encourages and rewards smaller parties, leading to a fragmented or multi-party system, with a tendency to have coalition governments—which change frequently—and a high degree of political bargaining. The reverse of this is that plurality systems, by disproportionately advantaging large parties and underrepresenting smaller parties produce a stable two-party system, with alternating governments and adversarial politics. These generalisations are then extended to encompass the following assumption: that the adoption of PR involves opting for fairness at the expense of political stability. It has then been argued that PR is more suited to smaller countries, in part because with their smaller burden of decision-making they will be less adversely affected by frequent governmental changes and in part because multi-member constituencies are thought to lead to a potential loss of connection between voters and representatives. Plurality/majoritarian systems are based upon a strong territorial principal, in which a bond is created between the single representative and her constituents. In a large population, with multi-member constituencies, this bond would be sacrificed.

A number of reservations can be entered about these assumptions. Firstly, there is the question of the link between electoral formula and party system. As Rae noted, party systems are influenced by many forces, one of which is the electoral law. In addition, "electoral laws are themselves shaped by party systems." (Rae, 1957:141) So, determining which force causes what results will require more specific information about individual countries. Many writers have observed that PR systems have been introduced in response to changes already taking place in the party system. (Carstairs, 1980, Rokkan, 1970) Likewise, Bogdanor and Butler (1983) demonstrated that reductions in the numbers of effective parties can take place under PR. Plurality systems have sometimes coincided with one-party dominance, or with one large party confronting a shifting alliance of smaller parties (both of these have been a feature of Indian politics since independence and at one time were thought possible for the UK).

Secondly, it is not at all clear that two-party systems are in any case necessarily more stable or effective. A number of writers (Finer, 1975 and Powell, 1982 for example) have argued that on certain key indicators of success the consensual-coalition democracies perform better than adversarial systems. Finer's arguments are well known, and widely contested (see Dearlove and Saunders, 1984). Adversarial politics in his view lead to governments which manipulate economic policy for electoral advantage, leading to artificial disruptions in the business cycle. In the longer term, coalitions tend to be more centrist and consensual with fewer sharp changes of policy.

Other writers have noted that on matters such as voter participation and control of violence, consensual-coalition governments also perform better. Growing electoral volatility - with voters changing parties in an unpredictable manner might also increase instability in some plurality/majoritarian systems. Famously, Arend Lijphart argued (1977) that consensual-coalition government was particularly appropriate for societies deeply divided by ethnic, linguistic or religious differences. Of course, to put the other point of view, it is argued that multi-party systems are inherently less fair than two-party systems as they tend to produce governments dominated by relatively unpopular centrist parties, which can 'hold the balance of power' and be 'permanently in government'.

Using the systems

Showing the difficulty of finding simple causal relationships between electoral systems and party systems, or the danger of drawing hasty conclusions about electoral systems and stable government, is not to argue that electoral systems have no effects at all. On the contrary, in keeping with Sartori's claim, quoted above, about their potential use as an instrument for political manipulation, studies of electoral systems and their histories have shown that changes and adaptations have been made in order to achieve certain outcomes. There have been cases of ruling parties changing the rules of the game in order to achieve some advantage for themselves. In the Fourth and Fifth French Republics there have been shifts from alternative vote to PR and back again in order to limit the power of the Communists or enhance the success of the dominant party. Constituency boundaries have been manipulated in order to consolidate the vote for particular parties or individuals. The US provides some examples. In the 1800s, the Governor of Massachusetts, Elbridge Gerry, constructed a salamander-shaped congressional district ( hence 'gerrymander' ) in order to favour his own party. In the 1960s, the US Supreme Court began to require states to construct electoral boundaries in conformity with the principle that "every vote should count equally" so that the vote of a citizen in one territorial unit should not count for more than that of a voter in another district. This principle was then applied to allow boundaries in certain areas to be redrawn in order to ensure the election of representatives from certain ethnic minorities—to consolidate the black and Hispanic vote, in effect. (See Phillips, 1996 for a discussion of this.)

Bringing in excluded groups

The original introduction of PR systems was intended to offset the disadvantages of minority groups—in Denmark in the 1850s, to help the German minority in Schleswig, for example. Belgium introduced PR in the 1890s as a response to the problems of accommodating a three-party system. As Lijphart and others have demonstrated, in societies which are divided in complex ways, PR has been introduced to systems to ensure that minority communities are represented in proportion to their numbers in the population. It is important to recall the caveats entered above; there is no guarantee that the effects sought will be achieved simply by changing electoral laws. Other factors—public resistance, party policies, demographic shifts—will also have an impact, sometimes reinforcing, sometimes undermining the consequences of electoral change. Nevertheless, certain patterns can be discerned.

If we take as an example the effects of electoral systems on the percentages of women elected to legislative assemblies, there are indications that, as Gallagher puts it, (Gallagher et al, 1995: 294) PR systems affect not only the representation of parties but also the "profile of the individuals who sit on the parliamentary benches." A substantial literature on women in politics has, since the 1970s, shown that women candidates find it more difficult to be elected into parliaments under plurality systems than under PR. (Darcy, Welch & Clark, 1987, Lovenduski & Norris, 1993, Phillips, 1995, Randall, 1987) The features most conducive to electoral success for women candidates are: larger district magnitudes and a party list electoral system. In addition, rules which increase the representation of smaller parties—using largest remainder calculations, having the lowest possible thresholds—may in some cases also lead to greater numbers of women in assemblies, as smaller parties, with weaker competition for seats, tend to put forward more women candidates. It appears to be the "single-member" feature of plurality systems—combined with cultural attitudes which discriminate against women—that is the key to explaining their under-representation in parliaments elected under such rules. The presence of women as candidates in such systems attracts attention which leads selectors to fear they may win fewer votes than a male candidate. In PR systems with their larger numbers of candidates, the absence of women from lists or ballot papers attracts attention and requires justification. All of these factors can also be seen to apply to other excluded or minority groups.

Party responsibilities

The electoral system alone will not achieve proportional or even radically improved representation for women. The willingness of parties to select women for winnable seats is also important, and in the places where women have been most successful, parties have responded to pressure from women's organisations and/or from governments. A recent study of provincial elections in Argentina seems to indicate that a combination of larger district magnitude, a centralised party list system and a law requiring parties to field a minimum number of women candidates in the higher portions of the lists leads to an improvement in the numbers of women elected. Similarly, improvements in the numbers of women in parliament are possible under plurality/majority systems if party leaderships or selectors are encouraged to put women candidates forward in winnable seats. (There is in fact no evidence showing that women candidates are likely to lose safe seats.) Hence we can compare the UK and Italy. Before the 1997 election in the UK, 9.2% of parliamentary seats were held by women—a very slight improvement on the previous situation when the numbers of women for years hovered around 6%. Italy, however, even with its additional member system, returned women to only 8% of the seats. The 1997 British election saw women take an astonishing 18% of seats, bringing the UK above many European PR systems and bringing it close to the German figure of 20%. (Germany has a system similar to that in Italy, but with a greater number of members elected via the party list.) The significant difference in 1997 was the decision by the Labour Party to instruct constituencies to select a women candidate in a high proportion of safe seats. Although this policy was abandoned after a challenge in the courts, no women candidates already in place were 'deselected'. Combined with the landslide victory for Labour which led to victories for candidates in seats not usually considered winnable, the result is that of the 129 seats in Westminster held by women, 101 are Labour seats.

While there is no guarantee that future elections to Westminster using the simple-plurality system will produce the same successes for women Labour candidates, there is some reason to believe that the 1997 election could be a watershed as far as women candidates are concerned. Firstly, the success of Labour women will lead to pressure on other parties who do not already do so to select a fair proportion of women for winnable seats. This would conform to a pattern observed in Scandinavia, where parties on the left of the spectrum initiated a process of selecting women which was then followed by Centre and Right-wing parties. Secondly, recent research from the US indicates that a greater number of women representatives leads to an increase in the general interest in politics among women, (Verba et al 1997) which in turn may produce more women prepared to be candidates. Finally, it is possible that electoral reform in Britain may lead to the introduction of a system which—other things being equal—will at least facilitate the entry of larger numbers of women into parliament. Social attitudes which lead to the expectation that the political elite should be male-dominated may already have been transformed in Britain, so that women candidates should enjoy even greater success under a proportional system.

V. Conclusions and recommendations

A recent survey of electoral systems in democratic states throughout the world (Blais & Massicote, 1997) revealed that, of 77 states designated as "strong democracies", 34 had PR systems, 10 had "mixed' systems which include a PR element, and 26 were plurality -based. (One country, Chile, had a hybrid system too difficult to classify.) The survey confirmed the view that "Europe is the heartland of PR"; of 33 European states, 27 had some form of PR system. South America, with its European influences, has also shown a preference for PR elections. The authors concluded that the "debate over the vices and virtues of various electoral formulas is still very much alive, but proportional representation may be closely associated, in the minds of many, with the ideal of democracy." (116) This association of proportional representation with democracy is borne out by the fact that many of the emerging post-communist states, especially in Central and Eastern Europe, have opted to include at least elements of PR in their new electoral systems, in spite of pressure to resist fragmentation in the interests of building 'strong government'. The exclusion that accompanies advantaging large parties, can, in the long run, lead to greater instability.

For any future Northern Ireland assembly, therefore it would seem appropriate to recommend that elections be conducted under the fairest—that is, the most conducive to proportionality—electoral system. There are good arguments in favour of this. Not only are people already familiar with the STV method from local and European elections, but the more complex formula for the May 1996 elections. A PR system allows for the incorporation of values which are of great importance to future stability and successful government in Northern Ireland.

Possible systems: principles

The principles any electoral system should subscribe to should thus involve:

• proportionality and inclusiveness: the fair representation of all sections of the population and of all shades of democratic opinion are desirable both in terms of social justice and because they promote identification with and a sense of ownership of political institutions and processes.
• accountability: giving a greater weight to the interests and values of all sections of the community means that political executives must take some account of them in setting policy agendas and decision-making
• effectiveness: having a broader range of opinions represented means that policy areas and ideas previously not considered will be heard and acted upon
• accommodation: assemblies elected on the basis of PR tend to lead to a situation where parties are encouraged to find points of commonality on which they can build rather than stressing division and difference; far from rewarding extremists, as is often alleged, these systems offer opportunities for consensus building
• straightforwardness and intelligibility: it should be clear to voters what will be the possible or likely outcomes of casting their votes in particular ways
• act to minimise divisiveness: the system should balance party- positional and individual-territorial rooted representation

Insofar as arguments against PR have validity in any context, it is hard to see how they apply to Northern Ireland. For a regional assembly, in a small territory, with a relatively small population, arguments about strong territorial links between electors and representatives would seem not to apply. Counting and casting votes under any system would not be too cumbersome or complicated given the small electorate. The limited decision-making load would also appear to make PR particularly suitable in this context.

Possible systems: parameters

As the descriptions of various systems indicates, there is potentially unlimited choice of electoral system for a Northern Ireland assembly. Finding the 'best' possible system depends upon a number of factors, some of which have not yet been determined, such as how many members would be elected. It would seem desirable to have around 100 members, if the principle of proportionality is to be preserved. The nature of the government to be created by the assembly and the powers which it will have are likewise unclear at present, making very firm conclusions about the electoral process difficult to reach. Any comments about likely options are thus offered in this context.

Familiarity, intelligibility and voter choice would be ensured if a straightforward STV system were put in place. Existing Westminster constituencies could be retained, returning equal or variable numbers of candidates, or larger constituencies could be created. Creating larger constituencies returning greater numbers of representatives would allow more opportunities for smaller parties to have a chance of winning seats. It would also make it possible for a greater number of women to be elected, perhaps correcting the present relative underrepresentation of women in all elected bodies. To a greater or lesser extent, such systems would tend to favour the existing larger parties.

Adopting a party list system would also satisfy the requirements of clarity and simplicity to use, but, as we have seen would reduce voter choice to some extent, depending on how many preferences voters were allowed to express. The more voter choice is allowed, of course, the more complex such systems become. In Northern Ireland, a party list election based on one, two or three large constituencies would achieve reasonable proportionality. It would seem acceptable to have relatively low thresholds, given the fact that the overall electorate is relatively small and the importance of principles like proportionality and inclusivity. Party list systems permit parties to bring into the assembly members of groups likely to be underrepresented or excluded, for example, members of ethnic minorities or women. Party activists and voters can exert pressure to achieve such representativeness, or electoral laws can be drafted requiring parties to place set numbers of such candidates higher in the lists to increase their chances of success. As with STV systems, the bigger the constituency, the weaker the link between citizens and representatives, but again this might not be seen as too significant given the relatively small area and electorate.

The benefits of both systems might be retained if there were to be a mixed system based on STV and additional-member formulas, which would result in a percentage of representatives coming from constituencies and the remainder from party lists. The size of constituencies and the numbers elected from each sector would be determined according to how much importance was placed on particular values. For example, if voter choice and a strong link between representative and constituents are given highest priority, then there would be a higher proportion of members coming from constituencies. Having a two-section ballot paper as in the German system—with electors both choosing constituency candidates and expressing a preference for a party—gives reasonable choice while enhancing proportionality. Increasing the proportion coming from party lists (to 50%, as in Germany) would lead to greater inclusivity and proportionality. There would also be less likelihood of list members being perceived as having a lesser status than constituency members. Greater proportionality would also follow if the list members were elected from a Northern Ireland-wide constituency, with low thresholds and a 'largest-remainder' formula for allocating seats. Such a system would be similar, though not identical, to that proposed for Scottish and Welsh regional elections. While less straightforward than STV or simple party-list systems, it is not confusing or difficult for voters to understand.

Possible systems: practice

Any number of permutations are possible. Below we outline a number, and provide some comment on likely outcomes, bearing in mind the caevat entered above in respect of 'unknown' factors such as powers etc.

• Single constituency closed party list system

An assembly of 120 members, returned from a single Northern Ireland constituency. The ballot paper would consist of a list of the parties competing in the election; there would be no candidate names on the paper. Parties would have to register, and deposit with the electoral office a list of the candidates who would occupy seats won by the party. The order of candidates would be decided by the party concerned. Seats would be allocated by dividing the number of votes cast for a party by the electoral quota. If the number of quotients turns out to be less than the number of seats available, the remaining seats would be allocated using the largest remainder method.

• Mixed system A

An assembly of 120 members, electing 72 individual members from 4 constituencies by STV, and 48 members from nominated parties on a regional list. The ballot paper in each constituency would thus contain two sections:

(a) Constituency candidates to be selected by preferential voting; and

(b) The regional list to be selected by voting 'X' for a party.

There should be no limit on the number of nominated parties entitled to contest any election. The seats in the regional sector being allocated by 'greatest remainder'. The four constituencies might emerge by grouping current Westminster constituencies. Thus Newry and Armagh, South Down, Upper Bann, Lagan Valley, Strangford and North Down could be one such constituency; the four Belfast districts another; South, East and North Antrim with Mid-Ulster a third, and Foyle, West Tyrone and Fermanagh and South Tyrone the fourth.

• Mixed system B

An assembly of 110 members, electing 90 individual members from the current 18 Westminster electoral areas by STV, and 20 members from a regional list, with an award of 2 seats to each of the ten parties who cumulatively secured the highest number of votes. The ballot paper in each constituency would thus contain two sections:

(a) Constituency candidates to be selected by preferential voting; and

(b) The regional list to be selected by voting 'X' for a party.

There should be no limit on the number of nominated parties entitled to contest any election, and no threshold should be necessary, given the size of the voting population.

• Mixed system C

An assembly of 108 members, returning 9 individual members from 6 constituencies by STV, and 54 party representatives from 1 Northern Ireland wide constituency. The ballot paper in each constituency would thus contain two sections:

(a) Constituency candidates to be selected by preferential voting; and

(b) The regional list to be selected by voting 'X' for a party.

The 6 electoral areas could be composed in two ways. One, electoral areas might become congruent with existing county boundaries, or they might be fashioned by grouping the existing Westminster constituencies in bunches of three. Thus the constituencies envisaged are:

1. Foyle, East Londonderry, North Antrim;
2. West Tyrone, Mid Ulster, Fermanagh and South Tyrone;
3. East Antrim, South Antrim, Belfast North;
4. North Down, Strangford, Belfast East;
5. Belfast West, Belfast South, Lagan Valley;
6. Upper Bann, Newry and Armagh, South Down.

• Mixed System D

Similar to Mixed System C, with six members being returned from 9 constituencies, composed of the following pairings of Westminster constituencies:

1. Belfast West, Belfast North;
2. Belfast East, Belfast South;
3. North Down, Strangford;
4. Lagan Valley, South Down;
5. Upper Bann, Newry and Armagh;
6. South Antrim, East Antrim;
7. North Antrim, Mid Ulster;
8. East Londonderry, Foyle;
9. West Tyrone, Fermanagh and South Tyrone.

Possible systems: Outcomes

In keeping with that asserted elsewhere in this paper, returning representatives from a single constituency, if no thresholds are imposed, will present the most proportional outcome. If the results of the May 1996 election are extrapolated and applied to such a system, Richard Sinnot calculates the following: the UUP and DUP would each have obtained three fewer seats; the SDLP would have received two more seats; Sinn Fein, the APNI and the UDP would have remained the same. The PUP would have gained two seats, the UKUP one, and the Labour Coalition and NIWC one less. Further, the Greens, Conservatives and Worker's Party would each have won a seat.

Any mixed system allows for a balance between party policy and personality interests. System B would result (assuming a similar turnout to the May 1996 election) in the voice of women being heard via the NIWC, (assuming also that the NIWC contests any forthcoming elections) and presents an opportunity for parties to adopt an 'alternative list'—placing male and female candidates alternately on the party list. It is difficult to extrapolate the local constituency results, as the system here envisaged engages STV. Systems C and D would, according to our earlier reasoning, and research elsewhere, allow for greater electoral impact by smaller parties and by women, but it is difficult to predict numerically the outcomes.

Different systems will lead to somewhat different outcomes and will require different strategies from voters and parties. Whatever system is adopted by those drawing up the rules, it should be clearly related to identifiable principles which are spelled out in some detail for the electorate. It should also be possible to keep the system under review and introduce adjustments or complete changes after a set period of time had elapsed. Such tasks might be put in the hands of an independent electoral commission, reporting to the assembly and the public whose functions would also include supervision of the conduct of elections and review of constituency boundaries.

We have before us much literature, theory and practice of how to create manageable and fair election processes and outcomes. In crafting constitutional change, we cannot ignore them.

References

Blais, Andre (1988) The Classification of Electoral Systems, European Journal of Political Research, No. 16 pp. 99-110

Blais, A and Massicote, L (1997) Electoral Formulas: A macroscopic perspective, European Journal of Political Research, Vol. 32, No. 1 pp 107-129

Blackburn, R (1995) The Electoral System in Britain, Macmillan, London

Bogdanor, Vernon (1984) What is Proportional Representation?, Robertson, Oxford

Bogdanor, V and Butler, D (eds.) (1983) Democracy and Elections: electoral systems and their political consequences, Cambridge University Press, Cambridge

Carstairs, A M (1980) A Short History of Electoral Systems in Western Europe, Allen & Unwin, London

Darcy, R, Welch, S and Clark, J (1994) Women, elections and representation, University of Nebraska Press, Lincoln

Dearlove, J and Saunders, P (1984) Introduction to British Politics: Analysing a capitalist democracy, Polity Press, Cambridge

Duverger, Maurice (1954) Political Parties, Methuen, London

Gallagher, Michael (1991) Proportionality, Disproportionality and Electoral Systems, Electoral Studies Vol. 10, No. 1 pp. 33-51

Gallagher, M, Laver, M and Mair, P (1995) Representative Government in Modern Europe, New York and London, McGraw Hill

Holmes, Leslie (1997) Post-Communism: an introduction, Cambridge, Polity Press

Jones, Mark (1998) Gender Quotas, Electoral Laws and the Election of Women: Lessons from the Argentine Provinces, Comparative Political Studies, Vol. 31, No. 1, February 1998 pp. 3-21

Lijphart, Arend (1997) Democracy in Plural Societies: a comparative exploration, Yale University Press, New Haven, Connecticut

Lijphart, Arend (1994) Electoral Systems and Party Systems, Oxford and New York, Oxford University Press

Lijphart, A (1994b) Democracies: Forms, performance and constitutional engineering, European Journal of Political Research, 25, pp. 1-17

Lijphart, A (1997) The Difficult Science of Electoral Systems, Electoral Studies, 16, pp 73-77

Lovenduski, J and Norris, P (1993) Gender and Party Politics, Sage, London

Phillips, Anne (1996) The Politics of Presence, Clarendon, Oxford

Rae, D W (1971) The Political Consequences of Electoral Laws, Yale University Press, New Haven

Randall, V (1987) Women and Politics: An International Perspective, Macmillan, Basingstoke

Reeve, A and Ware, A (1992) Electoral Systems: A Comparative and Theoretical Introduction, Routledge, London

Rokkam, Stein (1970) Citizens, Elections, Parties, Universitetsforlaget, Oslo

Schumpeter, J (1975) Capitalism, Socialism and Democracy, Harper and Tow, New York

Sinnot, Richard (1997) Choosing an Electoral System for Elections to Conflict-resolution Negotiations: The Case of Northern Ireland Representation Vol. 34, No. 3 &4 Autumn/Winter 1997 pp. 190-199

Verba, S, Burns, N and Schlozman, K L (1997) Knowing and Caring about Politics: Gender and Political Engagement, The Journal of Politics, Vol. 59, No. 4 pp. 1051-1072