I am involved in several grassroots organizations in Israel whose goals are to improve the electoral process (and thus democracy) in Israel.  One of our main goals is to implement electoral districts for greater accountability to the voter. We are currently trying to address the question of electoral district design and how one minimizes the manipulation of the process for the political benefit of one or another of the political parties. 

• How can electoral districts be designed to minimize the manipulation of the electoral process?
• How can voter accountability be best achieved?
• Are you aware of existing computer software that can create districts algorithmically, or at least with minimal human intervention?

Today the 120 members of the Knesset are elected through a party-list proportional representation system to serve 4-year terms.

Einführung

The question presented is closely related to how electoral districts are designed and decided upon within the broader Electoral System Design a given country chooses to adopt. Most of the so called plurality/majority electoral systems use single-member districts (SMDs), where the candidate with the most votes is elected and thus maximum accountability is being achieved. The most common electoral system using SMDs is by far First Past the Post, a system in which the candidate with the most votes takes the single seat from the electoral district at hand. All proportional representation electoral systems however use what is known as multi-member electoral districts (MMD), where more than one representative is elected to office. The decision whether to adopt a majority system or a proportional system thus concerns various trade-offs such as for example strong majority government capacity versus representation and proportionality. As we shall see, these issues are present also for the debate on drawing of electoral boundaries and increasing voter’s ability to throw out unsatisfactory representatives. To see more information about SMDs and MMDs, please click here.

In the following text we try to relate to the three questions asked by the user by tackling them in a similar order as posed: first we present some ideas on how to contemplate improving voter accountability in electoral systems. Secondly, we discuss the concrete process of drawing electoral boundaries and how to try to minimize the possibilities for parties to manipulate that process. Lastly, and outlook on possible software and technical details which may help in the process of boundary delimitation is provided.

Summary of responses

As our expert Deborah Schein puts it, the question regarding electoral districting is fundamental to the composition of government in all parliamentary systems. Electoral district magnitude is the first important decision that has to be taken. District magnitude refers to the number of representatives to be elected from one constituency, and thus determines an electoral system's ability to translate votes casted into seats won proportionally. In proportional electoral systems, it is important to find the balance between accountability and proportionality taking into account the political situation and party system in the respective country.

Using the whole country as one electoral district does indeed give the greatest degree of proportionality. But when the districts are made larger the problem is becomes that the linkage – especially geographical - between elected members and his or her district grows weaker. Because of this paradox there has been a lively debate on the ideal electoral district magnitude. Most scholars agree that district magnitudes of between three and seven seats tend to work quite well, and it has been suggested that odd numbers work better than even numbers, particularly in a two-party system. (Reynolds et. al. 2005, 82.)

Designing and drawing of electoral districts

When the suitable electoral system and district magnitude have been defined, the next step is boundary delimitation which usually refers to the concrete process of drawing electoral boundaries. According to ACE Comparative Data section about boundary delimitation approximately two thirds of all countries delimit constituencies for electoral purposes, whereas in less than a third of all countries the existing “natural” regional, provincial or other boundaries are being used.

The authority responsible for drawing the boundaries as well as the degree of independence granted to this authority vary considerably from country to country. Traditionally, legislatures have been allowed to draw their own districts. It is, nevertheless, worth noting that incumbent politicians are often reluctant to change the existing electoral district boundaries if it even hypothetically reduces their chances of re-election. Increasingly, countries are turning the process over to independent commissions. This is part of an international movement to eliminate "politics" from the redistricting process. According to ACE Comparative Data section about drawing electoral boundaries, in 34.4 per cent of the countries, the electoral districts are defined by the 1st or 2nd chamber of the parliament, whereas an independent boundary commission is responsible in 19.4 percent of the countries.

Reforms have also included provisions for identifying formal criteria to consider when drawing districts. These criteria are usually listed in the electoral laws of the country, and they often include factors such as equality across districts regarding the size of population, respect for local administrative boundaries and other geographic features such as natural (physically-defined) boundaries, and recognition of communities of interest. Equality in population size, respect for natural boundaries and communal interests do not always harmonize. For example if one wishes to enforce size, it could be at the expense of splitting communal and/or natural physical boundaries. Each country thus needs to make its own priorities of how to proceed. (Reynolds et. al. 2005, 82.)

Perhaps the most common type of manipulation or corruption related to the process of defining electoral districts is gerrymandering. This usually occurs in electoral systems with single-member constituencies, and it can be defined as “the deliberate manipulation of electoral district boundaries so as to advantage or disadvantage a particular political interest” (Reynolds et. al. 2005, 177). The term originally comes from the USA, where political parties in the government have been able to draw the constituency boundaries in order to ensure maximum support from certain ethnic minorities (Reynolds et. al. 2005, 123). Horation Boneo's paper on Observing Delimitation illustrates how manipulation could be done in practice and discusses ways to observe it.

One way to reduce gerrymandering and political influence over districit design, is using existing natural administrative or cultural boundaries. Deborah Schein points out that the homogeneity of these existing regions should hence be taken into account when drawing district boundaries. The more politically homogeneous – ideologically or culturally - the region, the more likely the elected party representatives would reflect that ideology. And vice versa, in mixed communities, according to Schein, “the swing vote then becomes more important and competition for votes by parties increases”. In a mixed electoral constituency the vote could also be divided between the two most extreme parties.

Both PN experts Horacio Boneo and Tim Meisburger emphasize independence of the authority responsible for boundary delimitation. Boneo adds that electoral districts can also be a mix of multi- and single-member models, as is the case in Ireland. All in all, drawing electoral district boundaries is a complex process which may appear to be incomprehensible in the eyes of an average voter. To educate the citizens in terms of electoral districting, a redistricting game has been developed for example in the USC School of Cinematic Arts.

Ways to improve voter accountability

Boneo points out that in a proportional system, one can try to improve accountability by opening up the process of selecting candidates (such as using open primaries). Another way to improve accountability is to use open lists where voters are able to vote single candidates in party lists. As Deborah Schein points out, to avoid parties’ internal democracy to dilute the influence of the ordinary voter, open lists could allow the actual voter – not the party – to influence the order of the list. She also mentions the possibility of using both national and regional lists in a mixed process. This process could ensure that both powerful national parties and more specific regional interests would be represented in the parliament. Tim Meisburger suggests two other options: reducing the length of representatives' terms and/or limiting the number of consecutive terms the MPs can compete for re-election. Limiting the number of consecutive terms might make the race for seats more competitive, as incumbents often have some advantage in the media whilst holding a position.

Technologie

The ACE article about Technology for Boundary Delimitation gives a thorough introduction to different technological solutions to consider in designing suitable electoral districts. To mention some examples, in New Zealand the parliamentary electoral districts are delimited every five years following the national census by an independent Commission. David Henry describes that the criteria used there include community of interests, topography, communication links, and if possible local government boundaries. Debashis Sen summarizes the procedure of boundary delimitation by informing that first step is to identify the lowest administrative unit and then decide that the electoral districts will be built out of addition of these “invisible” blocks. Both Henry and Sen emphasize the importance of equality of population between the different constituencies. However, according to these two experts and Carl Dundas, a deviation margin of +/- 5 per cent or +/- 10 percent may be necessary in order to take care of geographical continuity and/or natural barriers such as mountains, lakes and so on.

If the procedure followed is as above, GIS software can be used to help drive logical conclusions. But as Sen points out, a digitized large scale map (preferably on 1:5000 scale) would be a prior requirement. If population figures are available for the smallest administrative unit and these units are plotted in a GIS map, it should be possible to carve out districts in the most objective manner. Dr. Noor Mohammad points out that India is using GIS in its ongoing delimitation exercise. More information can be found here.

In New Zealand, as Henry notes, sophisticated mapping software is used based on the population in each mesh block (small statistical unit). For example MapInfo, ArcInfo and ArcView applications are available for GIS software. Some attempts have been made to develop open source software for delimitation purposes for example in the state of Virginia.

Ray Kennedy mentions the Mexican algorithm developed in 2004-2005 redistricting process. It is a function of population, geometric compactness of districts, integrity of municipalities (i.e., how often municipal boundaries had to be ignored in the districting), and time of movement within the district. The Federal Electoral Institute’s report “Mexican Electoral Districting” can be found here.

Conclusions

As highlighted in our experts’ responses and introduction, electoral district design is a crucial element in turning votes cast into seats won in the parliament. The number of representatives elected from each district also directly relates to whether voter accountability or proportional representation is prioritized in respective country. Some experts point out that the most accountable are single-member constituencies, which usually however require a majority/plurality electoral system. Our question was posed by a person from Israel, where the electoral system is based on proportionality and the whole country forms one electoral district. In this case, according to some experts, it could be possible to divide the country into constituencies electing fewer representatives, and this way decrease proportionality and the possibilities of smaller parties to get their representatives to Knesset, the national parliament. Our experts also recognize the problem of possible manipulation in the process of drawing electoral boundaries. However, they find independent delimitation bodies and respect for existing administrative boundaries to be useful ways to decrease the influence of political parties in boundary delimitation. The experts’ replies also show that GIS computer software and its different applications are widely acknowledged as practical tools for drawing electoral boundaries with minimal human intervention.

Related sources

Instituto Federal Electoral (IFE), Mexican Electoral Districting (México: IFE, 2009).

Reynolds, Andrew, Reilly, Ben and Ellis, Andrew, Electoral System Design: The New International IDEA Handbook (Stockholm: International IDEA, 2005).

Lijphart, Arend, Constitutional Choices for New Democracies. In Diamond, Larry and Plattner, Marc F. (Eds.) Electoral Systems and Democracy (Maryland: John Hopkins UP, 2006) pp. 73-85.

ACE Electoral Knowledge Network:

Electoral Systems and Their Consequences

Electoral Systems, Institutional Framework & Governance

Boundary Delimitation

Technology for Boundary Delimitation

The consolidated reply was written together with Maija Karjalainen

Input received, with thanks from:

Horacio Boneo

Carl Dundas

David Henry

Ray Kennedy

Ronan McDermott

Tim Meisburger

Dr. Noor Mohammad

Deborah Schein

Debashis Sen

1. One way is to identify the lowest administrative unit (for example, a ward of a town municipality or a census village) and decide that the electoral districts will be built out of addition of these indivisible blocks. Thereafter, the number of voters per district can be kept as close to the notional average as possible, provided that a +/- 10% margin may be allowed to take care of geographical continuity and natural barriers. But for each deviation, a reasoned decision may be kept on record for public scrutiny.

2. Issuing of photo-identity cards to the voters cuts down the possibility of impersonation to a great extent. But the best way is to have images of the electors printed in the electoral lists, as is being done in countries like India.

3. If the procedure followed is as at (1) above, a GIS software can help greatly in arriving at logical conclusions. But a digitised large scale map (preferably on 1:5000 scale) would be a prior requirement. Mapinfo or Arcinfo softwares are easily available for GIS applications though open source GIS softwares are also there.

 The questions as placed open up a whole range of issues related to the design of electoral systems.  As it seems that those who placed the question are specially interested in exploring the possibilities of moving from a PR-national constituency system to single member (SM) constituencies to improve accountability of Knesset members to voters, I will try to frame the answers in that context.

1) On the relation between design of electoral districts and manipulation of the electoral process: The electoral districts that minimize the possibility of manipulation via district design are those whose boundaries coincide with existing administrative boundaries (the whole country, as in Israel; or provinces, as in Argentina). The moment in which you introduce electoral districts that require specific design (as in the case of SM constituencies) you are introducing an additional opportunity for manipulation, which has been frequently used (numerous examples can be found in the USA experience). There are approaches to the delimitation of electoral district boundaries that diminish the possibility of manipulation (see for instance the approach used in Canada). There are also intermediate options between pure PR and SM constituency systems that are not as susceptible to district design manipulation as SM constituency systems. For instance, the Irish approach with a relatively small district magnitude (3 to 5 members per constituency) are less prone to manipulation in design and, if combined with Single Transferable Vote, you would have a much greater degree of proportionality than with a pure SMC.

2) Achieving accountability of representatives to voters: Undoubtedly, SMC systems significantly improve the ability of voters to identify their representatives and hold them accountable. However, accountability is not the only dimension to consider. SM constituency systems may introduce significant disproportionality, depending on the distribution of votes. It would also practically eliminate the possibilities of representation of small parties. In the case of PR systems, you can try to improve accountability by opening up the process of selecting candidates (open primaries, etc.)

3) Delimitation software: There are quite a few, already mentioned in a previous answer. Lisa Handley is a specialist on the subject and might provide you with additional advice. (Send an e-mail to [email protected] if you would like to contact her.)

Mexico, in its 2004-2005 redistricting process, developed an algorithm that was a function of population, geometric compactness of districts, integrity of municipalities (i.e., how often municipal boundaries had to be ignored in the districting), and time of movement within the district. The Federal Electoral Institute has published a report on the process entitled "Distritación 2004-2005: Camino para la Democracia".

ESRI have an extension for ArcView (a major GIS package) that addresses districting, albeit for a US paradigm.
A useful starting point, provided you have a GIS to begin with - a challenge in itself.

Keywords "electoral districting GIS software" into Google will return some useful links.

The question posed regarding electoral districting is fundamental to the composition of government in a parliamentary system. Reducing the proportionality would reduce the number of parliamentary parties and in principle make the formation of governments easier and reduce the possibility of a small party with little national support pulling out of government and bringing down the government or the threat of pulling out dictating policy so that it reflects the minority rather than the majority view. It should be possible to reduce the proportionality while maintaining the party list system by reducing the number of seats in each electoral constituency. To maintain a national character and also reflect regional interests there could be a number of seats reserved for a national list; which would represent the more dominant political parties and regional lists that represent more specific interests. Such a system could help reduce the number of parties represented in the Knesset. 

In answer to your question regarding manipulation of electoral constituencies, it should be understood that the more politically homogeneous the region, the more likely the elected party representatives would reflect that ideology, the more mixed the communities, either ideologically or culturally the swing vote then becomes more important and competition for votes by parties increases and a broader based appeal is necessary. At the same time, in a mixed electoral constituency the vote could also be divided between the two most extreme parties, which could receive the majority of votes. In designing constituencies all these factors need to be considered including demographics. While an electoral system may not fundamentally change the outcome of the elections, the electoral system will help shape the composition of government.

That said; it is also possible to increase the nexus and accountability between the elected representative and voter by adopting a system that would allow the voter to vote for the entire list or mark his/her preference for a candidate. Such a system allows the voter and not just the party to influence the order of the list. Such a system increases accountability while maintaining a high degree of proportionality.

While this does not specifically answer your question, it does raise other issues that need to be considered while sensitizing the population to changes in the electoral system. Some of these changes may also require constitutional reform, and in any system politicians elected under one system are generally reluctant to adopt changes that could reduce their chances for re-election.

Having read the previous responses, I think the three issues raised have been adequately covered. I would just like to underscore the point that a quota for each electoral district based on the national electorate, with tolerance levels of 5%-10% above or below the quota, would set the stage for equitable districts, although it would not eliminate the possibility of gerrymandering.

I understand from your question that you are primarily interested in electoral system reform to enhance accountability, and I agree with you that the single most important change you could make to achieve this is to replace national list PR with 120 single member districts. Although in theory PR seems the perfect system to maximise representation, in actuality (as I am sure you have discovered in Israel) people can seldom be defined by a single ideology. On some issues they might support the party line, but on others they may prefer another approach, and in PR systems they would have no representative to turn to. Also, filtering constituent interest through a party structure enhances the power of the party hierarchy while diluting the influence of the ordinary voter. In practice this means that accountability and the overall quality of representation is often much lower in PR systems than in systems that use single member districts.

In majoritarian systems greater accountability and more proportional representation is achieved by reducing the size of the constituency. So, by going from one constituency to 120 constituencies you will enhance accountability. Increasing the number of constituencies increases the representativeness of the body, just as increasing the number of respondents in a survey increases the accuracy (representativeness) of the results.

Other methods of increasing accountability include reducing the length of representatives' terms. In countries that use single member districts everyone has noticed that the road tends to get fixed before an election. Increasing the number of elections (going from four year terms to two year terms) will enhance accountability.

One issue you raise is how to define districts so they are less subject to manipulation. What I think you mean is how can we prevent parties from defining districts in their own interest, as happens in the US when Republican and Democrat incumbents collude to create safe seats for incumbents. In a safe seat, a representative is less accountable, so accountability can be enhanced by making seats more competitive. In several states in the US, and in other countries, this is done by having districting done by an independent (non-political) commission, or increasingly through computer software that designs districts based solely on population and compactness, without regard to historical voting patterns.

Another method of increasing the competitiveness of seats is through term limits. Incumbents always have an advantage in media coverage, and often will (consciously or unconsciously) create conditions or legislation that tends to favor the re-election of incumbents. Limiting the number of consecutive terms a representative can compete for will make races more competitive and enhance accountability.

Hope this is useful. Tim

If population figures are available for the smallest administrative unit and these units are plotted in a GIS map, it should be possible to carve out districts in the most objective manner. India is using GIS in its ongoing delimitation exercise. Click here to find a website which may be of interest.

In New Zealand the parliamentary electoral districts are delimited every 5 years following the national census by an independent Commission. The criteria used include the normal ones of community of interests, topography, communication links, and if possible local government boundaries. Importantly the total population of each electorate has to be the same or near enough the same - the Commission can operate a tolerance plus or minus of 5 %. Thus if the target for each electorate is say 55000 - being the population divided by the number of seats, the Commission can in the case of any one electorate go down to a minimum of  52750 and a maximum of 57250. The Commission publishes its proposals and holds hearings on objections and counter-objections. Its decisions are then final.

Sophisticated mapping software is used based on the population in each mesh block (small statistical unit) which allows boundary lines to be easily drawn and redrawn, what if scenarios explored and the knock on effects of a boundary change in one electorate to be instantly calculated for another electorate.

This is a simplified version of the NZ position and if you want further details let me know.