SciELO - Scientific Electronic Library Online

vol.26 número1Reformas a los sistemas electorales: Algunas reflexiones desde la prácticaElectoral Reform Doesn't Matter-or Does It?: A Moderate Proportional Representation System for Chile índice de autoresíndice de materiabúsqueda de artículos
Home Pagelista alfabética de revistas  

Servicios Personalizados




Links relacionados


Revista de ciencia política (Santiago)

versión On-line ISSN 0718-090X

Rev. cienc. polít. (Santiago) v.26 n.1 Santiago  2006 


revista de ciencia política / volumen 26 / Nº 1 / 2006 / 212 - 215

Evaluating Electoral Systems

Gary W. Cox

Universidad de California, San Diego, Estados Unidos


La academia especializada ha sugerido que los sistemas electorales tienen como propósito promover (1) la representación justa de los partidos, (2) la ejecución adecuada de las políticas gubernamentales, y/o (3) la representación correspondiente a los intereses de los votantes de cada circunscripción electoral. Estos tres criterios pertenecen a tres diversas clases de rendición de cuentas _la que existe de los partidos con sus seguidores; la del gobierno con la ciudadanía, y la de los legisladores con su circunscripción electoral_ que no puede ser maximizadas simultáneamente. Este ensayo discute cada uno de estos criterios en abstracto para luego hacer referencia específica a las implicaciones del sistema electoral chileno en estos y a las propuestas para reformarlo.


Scholars have suggested that electoral systems should be designed to promote (1) fair representation of parties, (2) good governmental performance and/or (3) adequate local representation. These three criteria pertain to three different kinds of accountability—the accountability of parties to their supporters; the accountability of governments to their citizens; and the accountability of MPs to their supporters—which cannot be simultaneously maximized. In this essay, I discuss each criterion in the abstract and with specific reference to the current Chilean electoral system and proposals to reform it.

PALABRAS CLAVE • Ingeniería electoral • Reforma electoral


Scholars have suggested a wide variety of criteria by which to evaluate electoral systems. In this essay, I review three of the most commonly advocated criteria, according to which electoral systems should be designed to ensure (1) fair representation of parties, (2) good governmental performance and/or (3) adequate local representation. I point out that these three criteria pertain to three different kinds of accountability—the accountability of parties to their supporters; the accountability of governments to their citizens; and the accountability of MPs to their supporters—which cannot be simultaneously maximized. Inevitably, there are trade-offs and any given electoral system must balance concerns for governmental accountability against concerns for partisan or personal accountability. I conclude with a brief comment on Chile's current situation.


Many scholars view fair representation, both actual and perceived, as the most important standard by which to judge an electoral system. Simple and transparent systems are sometimes advocated because they are more likely to be judged fair and thus help the losers understand that they have lost. Of course, it is possible to imagine systems that are simple and transparently unfair, so other criteria -that more directly address fairness- typically come into play. Of these criteria, proportionality is the most widely known. Scholars promoting proportional representation typically argue that it defines, or is the only way to ensure, fair representation.

A weaker standard of fairness is that a system should be unbiased. Partisan bias can be defined as follows. Let (v1,…,vn) be the aggregate national vote shares and (s1,…,sn) be the aggregate national seat shares received by parties 1,…,n. An electoral system is unbiased only if, were any two parties to exchange vote shares, then they would also exchange seat shares. Violations of this zero-bias condition arise in multiple ways: malapportionment, gerrymandering, and so on.

Note that perfect proportionality ensures zero bias but zero bias does not require perfect proportionality. For example, if two parties compete in a single-member district system, party A may get 70% of the seats when it wins 55% of the vote, and party B may likewise get 70% of the seats when it wins 55% of the vote. In this case, the system is not proportional and yet it is unbiased. Thus, zero bias is a weaker requirement of fairness than is proportionality. It merely requires that the electoral system treats all parties similarly, conditional on the aggregate vote shares garnered by all parties.

To further clarify the concept of bias, consider two features of the Chilean system as of 2000: the use of binominalismo; and the presence of institutional senators. Binominalismo, by itself, does not necessarily entail bias. It is true that the second-largest party will get a larger seat return per vote in any given district, as long as its vote is not doubled by the largest party. It is also true that the Unión has generally been the second-largest party in more districts than the Concertación and thus has benefited more from this district-by-district bonus for the second-largest party. However, consider a longer-term view, in which the Unión eventually supersedes the Concertación as the largest party (one way to think of them "switching votes"). At this point, the Unión would likely suffer just as much as had the Concertación. If so, then the system is unbiased, by the definition given above.1

In contrast, the institutional senators introduced a clear element of pro-Unión bias. Even if the Unión and Concertación had exchanged votes on the elected Senators, the institutional Senators would still have remained in the Unión column.

Everyone endorses unbiasedness—both those preferring majoritarian and those preferring proportional systems. However, those preferring majoritarian systems have an additional burden of proof to bear. It is not enough to point out that majoritarian systems can in principle be unbiased, because political parties in such systems have proved adept at malapportioning and gerrymandering districts. Thus, those who advocate majoritarian systems should include, as an integral part of their design, features that address and mitigate threats to create bias via redistricting.


Another group of scholars view good governmental performance -variously defined- as the most important standard by which to judge an electoral system. To Hermens (1941), the most important thing was to secure stable governments. The work of Huber and Powell (1994) and Powell and Vanberg (2000) suggests evaluating electoral systems in terms of their ability to produce governments whose policies are congruent with the center of electoral opinion.

Both government stability and government policy are of course influenced by factors other than the electoral system. However, it is possible to measure stability and policy congruence, and relate these measures to electoral system characteristics-controlling for other factors as well as one can (see, e.g., Powell and Vanberg, 2000).


A third group of observers value local representation. The notion is that parties and their platforms tend to be anonymous and abstract. In order to induce citizens to get involved with the process of self-government, what is needed is a personal connection with "my" representative in Washington, Tokyo or Valparaíso. The desire to foster such connections is an important motivation for the current Dutch reform movement, for example.


The three criteria articulated above essentially refer to three different entities that voters might wish to hold accountable: parties, governments and individual MPs. Each form of accountability requires that voters be able to express approval or disapproval on the ballot of each entity. I would argue that holding governments accountable is the most important power to give voters. Thus, the ability of voters to express a choice between governments at elections is paramount.

One might argue that holding governments accountable in presidential systems such as Chile is easy, since voters have the option to vote separately for the presidential candidate of their choice. Even institutionally powerful presidents, however, have reason to seek a stable majority support in their congresses. Thus, voters have an interest in knowing not just who the president will be but also in knowing who his legislative allies will be.

The more important it is for voters to have a clear choice between governments at election time, the more important it is to ensure that presidents campaign with the legislative majority (or group) that they intend to govern with. How can one ensure that presidents will declare their allies? The most important factors encouraging clear presidential alliances are concurrent presidential and legislative elections; and low-magnitude districts in the legislative races. The low-magnitude districts produce either a few large parties (the US, the UK) or a few large alliances (such as the Polo in Italy, or the Concertación in Chile). The concurrent elections then make presidential candidates the most important figures in the campaign and facilitate clear alliances.2


The current Chilean system affords a clear choice of government. Two large coalitions oppose one another both in the legislative districts and in the presidential race. The winning presidential candidate then governs in conjunction with the legislative leaders of the coalition parties. At the same time, since the removal of the institutional Senators, the system is unbiased. Finally, the system allows voters some -albeit limited- choice between parties and persons.

Whatever reform is undertaken should preserve the clarity of governmental choice that the current system affords. Getting rid of the "second party bonus," thereby upgrading the system from a merely unbiased one to a more proportional one, would be fine, assuming that a substantial portion of Chileans favor it. One could increase proportionality while preserving governmental choice in various ways-for example, by introducing a mixed proportional system with the current districts (adding a national party list vote separate from the district vote for a candidate); or a mixed proportional system with single-member districts (as in Germany); or perhaps a tier of compensatory seats.


1 Of course, if one adopts proportionality as one's standard of what is fair, then binominalismo remains unfair by that (more demanding) standard.

2 There are several other factors that might in principle affect the emergence of clear presidential alliances. One is the fused vote, which tends to produce presidential dominance. Another would be allowing the president to be dismissed by a vote of no-confidence, as in the recently abandoned Israeli experiment with a separately elected prime minister.


Hermens, F.A. 1941. Democracy or Anarchy? A Study of Proportional Representation. Notre Dame: University of Notre Dame Press.         [ Links ]

Huber, John D., and G. Bingham Powell. 1994. "Congruence between citizens and policymakers in two visions of liberal democracy". World Politics 46 (3): 291-326.         [ Links ]

Powell, G. Bingham, and Georg Vanberg. 2000. "Election laws, disproportionality and median correspondence: Implications for two visions of democracy". British Journal of Political Science 30 (3): 383-411.         [ Links ]

Gary W. Cox es profesor benemérito de Ciencia Política de la Universidad de California, San Diego.  Es autor de The Efficient Secret (ganador del premio George H. Hallett en el 2003), y Making Votes Count (ganador del premio Woodrow Wilson Foundation en 1998). Además, es coautor de Legislative Leviathan (ganador del premio Richard F. Fenno Prize en 1993), y Elbridge Gerry's Salamander. Su último libro, Setting the Agenda, fue publicado en 2005. Después de haber sido Guggenheim Fellow, Cox fue distinguido por la American Academy of Arts and Sciences en 1996 y la National Academy of Sciences en 2005. 



Creative Commons License Todo el contenido de esta revista, excepto dónde está identificado, está bajo una Licencia Creative Commons