Content uploaded by Richard Snyder
Author content
All content in this area was uploaded by Richard Snyder on Sep 06, 2018
Content may be subject to copyright.
Content uploaded by Richard Snyder
Author content
All content in this area was uploaded by Richard Snyder on Sep 06, 2018
Content may be subject to copyright.
B.J.Pol.S. 31, 651–671 Copyright @ 2001 Cambridge University Press
Printed in the United Kingdom
The Value of a Vote: Malapportionment in
Comparative Perspective
DAVID SAMUELS AND RICHARD SNYDER*
Comparative studies of electoral institutions have largely neglected a fundamental characteristic
of most of the world’s electoral systems: malapportionment. This article provides a method for
measuring malapportionment in different types of electoral systems, calculates levels of
malapportionment in seventy-eight countries, and employs statistical analysis to explore the
correlates of malapportionment in both upper and lower chambers. The analysis shows that the
use of single-member districts is associated with higher levels of malapportionment in lower
chambers and that federalism and country size account for variation in malapportionment in upper
chambers. Furthermore, African and especially Latin American countries tend to have electoral
systems that are highly malapportioned. The article concludes by proposing a broad, comparative
research agenda that focuses on the origins, evolution and consequences of malapportionment.
The remarkable advance of democracy around the globe during the last
twenty years has focused much attention on how different kinds of democratic
institutions work. Building on an earlier generation of studies that addressed
this issue by analysing a relatively small number of long-standing democracies,
scholars have increasingly exploited an expanded dataset that includes the
more than sixty countries comprising what Huntington has called the ‘third
wave’ of democratization.
1
These efforts to broaden our understanding of the
varied dynamics and institutional formats of democratic systems have been
* Department of Political Science, University of Minnesota and Department of Political Science,
University of Illinois at Urbana-Champaign, respectively. The authors are listed in alphabetical order
and share equal responsibility for this work. We thank Lars Alexandersson, Andre´ Bouchard, John
Carey, Khatuna Chitanava, Michael Coppedge, Andre´s Mejia Costa, Gary Cox, Brian Crisp, Chris
DenHartog, Brian Gaines, Uwe Gehring, Edward Gibson, Allen Hicken, Simon Hug, Mark Jones,
Charles Kenney, Edward Kolodziej, Gunar Kristinnson, James Kuklinski, Michael Kullisheck,
Fabrice Lehoucq, Arend Lijphart, Juan Linz, Eric Magar, Rene´ Mayorga, Mikitaka Masuyama, Jose´
Molina, Scott Morgenstern, Shaheen Mozaffar, Trudy Mueller, Gerardo Munck, Jairo Nicolau,
Guillermo O’Donnell, Ingunn Opheim, Michael Patrickson, David Peterson, Daniel Posner, Andrew
Reynolds, Phil Roeder, Jean-Franc¸ois Roussy, James Schopf, Miroslav Sedivy, Matthew Shugart,
Peter Siavelis, Wilfred Swenden, Rein Taagepera, Michelle Taylor-Robinson, Michael Thies,
Laimdota Upeniece, David Wall, Jeffrey Weldon, Assar Westerlund and the anonymous referees of
this Journal for their help and suggestions. Scholars interested in obtaining data should contact David
Samuels at dsamuels@polisci.umn.edu.
1Samuel P. Huntington, The Third Wave: Democratization in the Late Twentieth Century
(Norman: Oklahoma University Press, 1991). See Arend Lijphart, Democracies (New Haven, Conn.:
Yale University Press, 1984), and G. Bingham Powell, Contemporary Democracies: Participation,
Stability, and Violence (Cambridge, Mass.: Harvard University Press, 1982), for exemplars of the
earlier generation of studies. See Gary W. Cox, Making Votes Count (Cambridge: Cambridge
University Press, 1997), for a recent work that employs an expanded dataset.
652 SAMUELS AND SNYDER
especially vigorous in the subfield of electoral studies, where a burgeoning
cross-national literature has emerged on the political consequences of electoral
laws.
2
Curiously, however, this literature has largely neglected a fundamental
characteristic of many of the world’s electoral systems: malapportionment, or
the discrepancy between the shares of legislative seats and the shares of
population held by geographical units.
3
This discrepancy has important political
ramifications. From the standpoint of democratic theory, for example,
malapportionment violates the ‘one person, one vote’ principle
4
which,
according to Dahl, is a necessary condition for democratic government.
5
Taagepera and Shugart regard malapportionment as a ‘pathology’ of electoral
systems, and Gudgin and Taylor conclude that it may be ‘ethically un-
justifiable’.
6
Given these strong normative claims about malapportionment, it
is surprising that students of democracy have devoted so little attention to this
topic.
In addition to these normative issues, malapportionment can have important
consequences for policy making. Recent work on cases as distinct as the United
States, Brazil, Russia, Argentina, Japan and Mexico shows that malapportion-
ment shapes the decision-making contexts of incumbent executives and
legislators in ways that have major effects on policy choices and coalitional
2See, for example, Cox: Making Votes Count; Lijphart: Democracies; Arend Lijphart, Electoral
Systems and Party Systems: A Study of Twenty-Seven Democracies, 1945–1990 (Oxford: Oxford
University Press, 1994); Rein Taagepera and Matthew S. Shugart, Seats and Votes: The Effects and
Determinants of Electoral Systems (New Haven, Conn.: Yale University Press, 1989).
3Exceptions include Burt L. Monroe, ‘Disproportionality and Malapportionment: Measuring
Electoral Inequity’, Electoral Studies,13(1994), 132–49; Burt L. Monroe, ‘Bias and Responsiveness
in Multiparty and Multigroup Representation’ (paper presented at the Political Methodology Summer
Meeting, UC San Diego, 1998); Burt L. Monroe, ‘Mirror Representation in the Funhouse: Systematic
Distortions in the Legislative Representation of Groups’ (unpublished paper, University of Indiana,
1998); Lijphart, Electoral Systems and Party Systems, pp. 124–30; Bernard Grofman, William
Koetzle and Thomas Brunell, ‘An Integrated Perspective on the Three Potential Sources of Partisan
Bias: Malapportionment, Turnout Differences, and the Geographic Distribution of Party Vote
Shares’, Electoral Studies,16(1997), 457–70; Alfred Stepan, ‘Toward a New Comparative Analysis
of Democracy and Federalism’ (paper prepared for the Conference on Democracy and Federalism,
Oxford University, 1997); and Ron Johnston, David Rossiter and Charles Pattie, ‘Integrating and
Decomposing the Sources of Partisan Bias: Brookes’ Method and the Impact of Redistricting in Great
Britain’, Electoral Studies,18(1999), 367–78.
4Michael Balinski and H. Peyton Young, Fair Representation: Meeting the Ideal of One Man,
One Vote (New Haven, Conn.: Yale University Press, 1982).
5Robert A. Dahl, Polyarchy: Participation and Opposition (New Haven, Conn.: Yale University
Press, 1971), p. 2.
6Taagepera and Shugart, Seats and Votes: The Effects and Determinants of Electoral Systems,
pp. 17–18; Graham Gudgin and Peter J. Taylor, Seats, Votes, and the Spatial Organisation of
Elections (London: Pion Limited, 1979). In addition Grofman et al. (‘An Integrated Perspective on
the Three Potential Sources of Partisan Bias’) have recently shown that malapportionment, along with
turnout differences and the geographic distribution of party votes, can contribute to partisan ‘bias’,
that is, asymmetry in how party vote shares are translated into seat shares in the national legislature.
See also Johnston et al., ‘Integrating and Decomposing the Sources of Partisan Bias’.
The Value of a Vote 653
dynamics.
7
These single-country studies suggest that malapportionment can
have an important impact on executive–legislative relations, intra-legislative
bargaining and the overall performance of democratic systems.
Despite the normative and practical importance of malapportionment, we
know of no cross-national, comparative study of this key dimension of
electoral systems.
8
In this article, we seek to fill this gap by providing a
method for calculating malapportionment cross-nationally and by measuring
the degree of malapportionment in seventy-eight countries.
9
We show that
levels of malapportionment vary significantly in both upper and lower
chambers and across regions, with African and especially Latin American
countries having comparatively high levels of malapportionment. We also
explore how important institutional variables, such as federalism, bicameral-
ism and the structure of electoral districts, are associated with malapportion-
ment. The final section concludes and provides suggestions for future
research.
7United States: Mathew D. McCubbins and Thomas Schwartz, ‘Congress, the Courts, and Public
Policy: Consequences of the One Man, One Vote Rule’, American Journal of Political Science,32
(1988), 388–415; Brazil: Barry Ames, Political Survival: Politicians and Public Policy in Latin
America (Berkeley: University of California Press, 1987); Russia: Daniel Treisman, After the
Deluge: The Politics of Regional Crisis in Post-Soviet Russia (Ann Arbor: University of Michigan
Press, 1999); Argentina: Kent Eaton, ‘Party, Province, and Coparticipacio´n: Tax Reform in
Argentina’ (paper prepared for the 1996 Meeting of the American Political Science Association, San
Francisco, 1996); Edward L. Gibson and Ernesto Calvo, ‘Federalism and Low-Maintenance
Constituencies: Territorial Dimensions of Economic Reform in Argentina’, Studies in Comparative
International Development, forthcoming; Japan: Michael Thies, ‘When Will Pork Leave the Farm?
Institutional Bias in Japan and the United States’, Legislative Studies Quarterly,23(1998), 467–91;
Mexico: Alberto Diaz-Cayeros, ‘Federal Resource Allocation under a Dominant Party: Regional
Financial Transfers in Mexico’ (paper prepared for the 1996 Meeting of the American Political
Science Association, San Francisco, 1996); Scott Morgenstern, ‘Spending for Political Survival:
Elections, Clientelism, and Government Expenses in Mexico’, Divisio´n de Estudios Polı´ticos, No.
69 (Centro de Investigaciones y Docencia Econo´micas, Mexico, 1996).
8Many studies explore apportionment in the United States (such as Gary W. Cox and Jonathan
N. Katz, ‘The Reapportionment Revolution and Bias in US Congressional Elections’, American
Journal of Political Science,43(1999), 812–41), and a handful of single-country studies analyse
other cases (e.g., Hiroyuki Hata, ‘Malapportionment of Representation in the National Diet’, Law
and Contemporary Problems,53(1990), 153–70, on Japan; and Jairo M. Nicolau, ‘As Distorc¸o˜es
na Representac¸a˜o dos Estados na Caˆ mara dos Deputados Brasileiros’, DADOS: Revista de Cieˆncias
Sociais,40(1997), 441–64, on Brazil). Lijphart’s bivariate coding of countries (in Electoral Systems
and Party Systems,p.130) as either ‘malapportioned’ or ‘not malapportioned’ illustrates the
impediments to cross-national research posed by the lack of an index for measuring levels of
malapportionment.
9Most of our cases would be classified as ‘free’ according to Freedom House (Freedom House.
2000. ‘Country Ratings’, on-line at http://www.freedomhouse.org/ratings/, accessed 14 April 2000).
We do include some cases with ‘semi-free’ elections, however, because if fully democratic
procedures are ever established in these cases, it is likely that current patterns of malapportionment
will shape the dynamics of the emerging democratic regime.
654 SAMUELS AND SNYDER
MEASURING MALAPPORTIONMENT
At the broadest level, we can describe electoral systems as either perfectly-
apportioned or malapportioned.Inaperfectly-apportioned system, no citizen’s
vote weighs more than another’s. The Israeli Knesset exemplifies such a system:
its 120 members are elected in a single, national at-large district. Four other
lower chambers in our sample are perfectly-apportioned: The Netherlands, Peru,
Namibia and Sierra Leone.
In a malapportioned system, by contrast, the votes of some citizens weigh
more than the votes of other citizens. In the most extreme case, a malapportioned
system might allocate to a single person an electoral district (say, that person’s
home), one vote, and all the legislative seats. The rest of the population would
hold all votes except one and would receive no seats. Although no real-world
electoral system approximates this extreme, nearly all exhibit some malappor-
tionment. Malapportionment thus seems a normal characteristic of most
electoral systems.
How can we measure the degree of malapportionment across electoral
systems? Ratios of largest-to-smallest districts might seem an obvious means
for assessing malapportionment. However, such ratios actually prove poor
indicators of malapportionment. First, district size on the basis of population
tells us little about the degree to which districts are underrepresented or
overrepresented: we also need to know how many seats are allocated to each
district. Furthermore, even if we know how many seats are held by the largest
and smallest districts and can therefore calculate ratios of ‘worst represented’
to ‘best represented’ districts, such ratios tell us little about overall degrees of
malapportionment. For example, even if this ratio is 50:1 (e.g., a single-member
district system in which the largest district has a population fifty times greater
than the smallest district),
10
all other districts may have nearly-equivalent
populations, and, hence, the largest and smallest districts could be extreme
outliers in a system with a low degree of average malapportionment. Although
it may be tempting to interpret wide gaps between the best and worst represented
districts as signs of high overall levels of inequality in electoral systems, a better
measure is required.
With one important modification, the Loosemore–Hanby index of electoral
disproportionality (D)
11
provides such a measure.
12
To calculate malapportion-
10 This is the case in India, where in 1991 the largest district for the national lower chamber (Thane)
had a population of 1,744,592, whereas the smallest district (Lakshadweep) had a population of just
31,665. See David Butler, Lahiri Ashok and Roy Prannoy, India Decides: Elections 1952–1995, 3rd
edn. (New Delhi: Books & Things, 1995), p. 16.
11 Disproportionality arises when ‘political parties receive shares of legislative seats that are not
equal to their shares of votes’. By contrast, malapportionment ‘occurs when geographical units have
shares of legislative seats that are not equal to their shares of population’ (Monroe, ‘Disproportion-
ality and Malapportionment’, p. 138).
12 This measure was previously employed to measure malapportionment in Brazil by Nicolau (see
Nicolau, ‘As Distorc¸o˜esnaRepresentac¸a˜o dos Estados na Caˆmara dos Deputados Brasileiros’). We
recognize that the Loosemore–Hanby index has certain shortcomings (Taagepera and Shugart, Seats
The Value of a Vote 655
ment (which we call ‘MAL’toavoid confusion with the widely-used ‘M’ that
refers to district magnitude), one takes the absolute value of the difference
between each district’s seat and population shares, adds them, and then divides
by two.
13
Thus, the formula is:
MAL ⫽(1/2)兺兩s
i
⫺v
i
兩
where sigma stands for the summation over all districts i,s
i
is the percentage
of all seats allocated to district i, and v
i
is the percentage of the overall population
(or registered voters)
14
residing in district i. The example in Table 1 illustrates
how to apply the formula.
15
For each district, the deviation from perfect apportionment is the difference
between the district’s share of seats (s) and voters (v). To calculate overall
malapportionment for the four districts, we first add the absolute values of the
differences between seats and voters for each district. We then divide the total
by two. So MAL ⫽(1/2)(兩36 ⫺40兩⫹兩24 ⫺30兩⫹兩23 ⫺20兩⫹兩17–10兩)⫽10 per
TABLE
1Distribution of Voters and Seats
District
1234
Percentage of voters 40 30 20 10
Percentage of seats 36 24 23 17
(F’note continued)
and Votes; Vanessa Fry and Iain McLean, ‘A Note on Rose’s Proportionality Index’, Electoral
Studies,10(1991), 52–9; Michael Gallagher, ‘Proportionality, Disporportionality and Electoral
Systems’, Electoral Studies,10(1991), 33–51; Monroe, ‘Disproportionality and Malapportion-
ment’), principally that it does not respect Dalton’s Principle of Transfers (see Monroe,
‘Disproportionality and Malapportionment’, p. 139). However, because the Loosemore–Hanby
index is widely-used and relatively straightforward to calculate, we decided to employ it. The
Loosemore–Hanby index has an additional advantage: it is far easier to interpret than the alternative,
‘Equal Proportions’ index suggested by Monroe (‘Disproportionality and Malapportionment’, Table
1) that does satisfy the principle of transfers. The Equal Proportions index is a measure of
distributional deviation that indicates the average difference between districts’ expected and actual
shares of seats. The Loosemore–Hanby index, by contrast, measures how many actual seats are not
apportioned equitably as a proportion of all seats in the legislature. Moreover, we calculated levels
of malapportionment for all the cases in our sample using both indices, and we found that the resulting
scores correlated strongly (at the 0.78 level). We also ran the regressions analysed below using both
indices, and the same variables were significant to nearly the same degree. Readers interested in
obtaining the data based on the Equal Proportions index should contact the authors.
13 See Taagepera and Shugart, Seats and Votes, pp. 104–5.
14 We use population per district whenever available. Most countries apportion seats on the basis
of population rather than registered voters.
15 The example is adapted from Taagepera and Shugart, Seats and Votes, who discuss
disproportionality rather than malapportionment.
656 SAMUELS AND SNYDER
cent, in this case. This score means that 10 per cent of the seats are allocated
to districts that would not receive those seats if there were no malapportion-
ment.
Multi-Tier Systems
Calculating malapportionment in cases with single-tier electoral systems, like
the United States, is straightforward with the above formula. However,
measuring malapportionment in cases with multi-tier systems, such as
Germany, Mexico or Japan, is more complex because territorial units are
allocated seats on different bases according to the rules for each tier.
16
We
propose the following five-step solution to this dilemma:
17
(1) Calculate the percentage of seats awarded to each district without
including any upper-tier seats in the total number of seats. Using the above
example of a four-district system, each district has respectively 36, 24, 23
and 17 per cent of the seats. For illustrative purposes, it helps to assume that
the percentage of seats equals the number of seats, and thus the districts are
awarded thirty-six, twenty-four, twenty-three and seventeen seats.
(2) Multiply the percentage of the country’s total population residing in each
district by the number of upper-tier seats. Suppose in our example that a
hundred legislators are also elected in a nationwide upper-tier district by list
proportional representation. To calculate malapportionment, we assume
that each lower-tier district is entitled to a proportion of upper-tier seats
equal to that district’s proportion of the national population. Thus, the four
districts in our example would receive respectively 40, 30, 20 and 10 per
cent of the upper-tier seats, or forty, thirty, twenty and ten seats.
(3) Add the number of upper-tier seats allocated to each district to the number
of lower-tier seats allocated to each district. In our example, each district
would thus receive a total of seventy-six, fifty-four, forty-three and
twenty-seven seats.
16 In single-tier systems, all electoral districts are primary, that is, they cannot be divided into
smaller districts to which seats are allocated. By contrast, multi-tier systems contain secondary
districts that can be partitioned into two or more primary districts (Cox, Making Votes Count,
pp. 48–9). Multi-tier systems can have both a single-member district system as well as a proportional
representation (PR) system laid ‘on top’ of the single-member (SMD) districts. See Luis Massicotte
and Andre´ Blais, ‘Mixed Electoral Systems: A Conceptual and Empirical Survey’, Electoral Studies,
18 (1999), 341–66, for a discussion of such mixed, mult-tier electoral systems.
17 Our proposed solution to the problem of measuring malapportionment in multi-tier systems
rests on the following assumptions. First, we assume that in systems where voters cast votes in each
tier, they do not vote strategically. Furthermore, although upper tiers tend to reduce the localistic
nature of elections, we assume that both voters and candidates ‘think locally’ in all tiers, and thus
voters do not reason differently when casting votes in distinct tiers. Although they may not accurately
reflect reality, these assumptions help us develop a simple, cross-national measure of malapportion-
ment.
The Value of a Vote 657
(4) Calculate the new percentage of seats allocated to each district using the
total number of seats in the national assembly. The addition of the upper-tier
to the lower-tier districts yields the results shown in Table 2.
(5) Calculate malapportionment using the new percentage of seats for each
district. Using the above example, MAL ⫽(1/2)(兩38 ⫺40兩⫹兩27 ⫺30兩⫹
兩21.5 ⫺20兩⫹兩13.5 ⫺10兩)⫽5%.
Our example suggests that upper tiers tend to reduce, but not eliminate,
overall malapportionment. The case of Germany’s lower chamber illustrates
this point. If the Bundestag did not have an upper tier, malapportionment would
double, increasing from 3 per cent to 6 per cent.
Three factors condition how upper tiers affect malapportionment: the degree
of malapportionment in the lower tier, the number of seats allocated to the upper
tier, and whether the upper tier allocates seats at the national or subnational level.
Lijphart claims that national-level upper tiers will ‘entirely eliminate’
malapportionment.
18
However, this is not necessarily the case. Although an
inverse relationship does in fact exist between malapportionment and the
number of seats allocated to national-level upper tiers (that is, as the number of
seats increases, malapportionment declines), a national upper tier with few seats
will only minimally reduce overall malapportionment, especially if the lower
tier has a high degree of malapportionment.
Although national-level upper tiers necessarily reduce malapportionment,
provincial, state or regional upper tiers can actually increase overall malappor-
tionment if the seats in the tiers are themselves malapportioned. For example,
Mexico’s upper tier has five forty-seat districts with proportional representation
(PR), and each district encompasses several of Mexico’s states. Because the
states have unequal populations, the population of each upper-tier district also
varies, resulting in a degree of malapportionment in the upper tier. Thus, upper
tiers do not necessarily decrease malapportionment.
Bicameral Systems
Bicameralism, like multi-tier systems, also complicates the measurement of
malapportionment. Comparative studies of bicameralism suggest that upper
chambers are significantly more malapportioned than lower chambers.
19
Upper
TABLE
2Distribution of seats in national assembly
District
12 3 4
Percentage of voters 40 30 20 10
Percentage of seats 38 27 21.5 13.5
18 Lijphart, Electoral Systems and Party Systems,p.146.
19 See, for example, Stepan, ‘Toward a New Comparative Analysis of Democracy and
Federalism’; and Lijphart, Democracies, especially chap. 10.
658 SAMUELS AND SNYDER
chambers are commonly understood to overrepresent minority groups,
especially citizens in smaller territorial units. Lower chambers, by contrast, are
seen as much less likely to overrepresent minority groups.
20
Indeed, this
interpretation of bicameralism fits the case of the United States. The Senate –
where the states are awarded the same number of seats regardless of population
–isextremely malapportioned. The House of Representatives, by contrast, has
virtually no malapportionment.
However, it would be a mistake to assume that the world’s other bicameral
systems share this design. First, there is no a priori reason to suppose that upper
chambers are more malapportioned than lower chambers: the converse might
also be true (and, as we shall see, is indeed true in some cases). Secondly,
regardless of which chamber has more malapportionment, we might expect
many lower chambers to also have high degrees of malapportionment. In India,
for example, a large number of lower chamber districts are reserved for
designated castes or tribes, and many of these districts are overrepresented.
21
And in cases such as Portugal and Croatia, lower-chamber seats are allocated
to citizens residing outside the national boundaries (i.e., citizens do not vote as
absentees, as US citizens may in Congressional elections).
22
If such extra-
territorial, ‘diaspora districts’ in the lower house have extremely small or
geographically-scattered populations yet hold guaranteed quotas of seats, they
may contribute to malapportionment in the same way as does the rule assigning
equal numbers of seats to territorial units in many upper chambers. In short, we
have compelling reasons to expect that the degree of malapportionment in lower
and upper chambers varies significantly across bicameral systems.
Two additional factors complicate the relationship between bicameralism and
malapportionment. First, one chamber may have significantly more power than
the other.
23
Such asymmetries of power can have important consequences for
how malapportionment affects legislative outcomes in bicameral systems. For
example, in cases where one chamber lacks veto powers over key items (for
example, the upper chambers in Japan and Mexico, which have historically
lacked veto powers over budgets), a high degree of malapportionment in the
weaker chamber may be inconsequential for most legislation. However, in cases
where the distribution of power between chambers is relatively symmetrical
20 Hence, one recent comparative study of federal systems (Stepan, ‘Toward a New Comparative
Analysis of Democracy and Federalism’) refers to the lower chamber as the ‘one person, one vote’
chamber, in contrast to the ‘territorial’ (i.e., upper) chamber.
21 Other countries, such as Colombia and New Zealand, reserve seats for indigenous peoples on
a non-geographic basis: indigenous groups in Colombia elect two representatives in a national
at-large district; Maori voters in New Zealand elect five representatives in a separate tier of
single-member districts. We do not include such non-geographic seats in our calculations of
malapportionment. We also exclude appointed and ex-officio members of the lower chambers from
our calculations.
22 In the Croatian case, this arrangement appears to favour a permanent, institutionally-entrenched
revanchism.
23 Lijphart, Democracies; George Tsebelis and Jeannette Money, Bicameralism (Cambridge:
Cambridge University Press, 1997).
The Value of a Vote 659
(such as Argentina and the United States), malapportionment in either chamber
is likely to have a major impact on legislative outcomes.
Secondly, even if both chambers in a bicameral system are malapportioned,
the same territorial units may not be overrepresented and underrepresented
equally in each chamber. Empirical scrutiny might even reveal cross-cutting
malapportionment: for example, the lower chamber might overrepresent urban
areas, whereas the upper chamber might overrepresent rural areas. Although it
is beyond the scope of this article, assessing whether malapportionment between
chambers is reinforcing or cross-cutting poses an important task for future
research.
MALAPPORTIONMENT IN COMPARATIVE PERSPECTIVE
Table 3 summarizes data on the degree of malapportionment in lower chambers
(MAL
LC
) for all seventy-eight countries in our sample.
24
The table also indicates
whether the country is federal, whether the lower chamber has a tier, and
whether the lower chamber employs single-member districts (SMD). As
discussed below, these variables are especially likely to be associated with the
degree of malapportionment in lower chambers. The table shows that the degree
of malapportionment varies dramatically across countries.
25
Five lower
chambers are perfectly-apportioned, and malapportionment in the remaining
cases ranges from 0.01 to 0.26, which means that between 1 per cent and 26 per
cent of the seats in these chambers are allocated in ways that violate the ‘one
person, one vote’ principle. Mean lower chamber malapportionment is 0.07,
with a standard deviation of 0.06.
Where do we find the countries with the highest levels of lower-chamber
malapportionment? The most-malapportioned countries in our sample are in
less-developed regions with many recently-established democracies: twelve of
twenty-two Latin American and Caribbean lower chambers score above the
mean, and seven of fourteen African lower chambers score above the mean. By
contrast, only six of twenty-three Western European and North American lower
chambers score above the mean, and only two of nineteen Asian, former Soviet,
and Eastern European cases score above the mean. For the advanced industrial
democracies, the mean is 0.04. Interestingly, our data suggest that malapportion-
ment in Africa plays a major role in countries with a British colonial legacy (for
example, Gambia, Ghana, Kenya, Malawi, Tanzania and Zambia).
Table 4 lists the twenty most-malapportioned lower chambers. It high-
lights the severity of malapportionment in many poor, recently-established
democracies. Of the twenty countries in Table 4, eighteen are either
recently-consolidated democracies, unconsolidated democracies, or developing
24 Table 5 below presents data on malapportionment in upper chambers.
25 In Table 3, we calculated the degree of malapportionment using data from the most recent
elections for which reliable information was available. See Appendix A on the website version of
this article for information about sources.
660 SAMUELS AND SNYDER
TABLE
3Lower-Chamber Malapportionment in Seventy-Eight
Countries
Country MAL
LC
Year Federal Tier SMD
Andorra 0.1307 1997 Yes
Argentina 0.1405 1995 Yes
Australia 0.0241 1996 Yes Yes
Austria 0.0643 1994 Yes
Barbados 0.0364 1994 Yes
Belize 0.0753 1993 Yes
Benin 0.0319 1995
Bolivia 0.1677 1997 Yes
Brazil 0.0913 1998 Yes
Burkina Faso 0.0325 1997
Canada 0.0759 1997 Yes Yes
Chile 0.1509 1997
Colombia 0.1324 1994 Yes
Costa Rica 0.0215 1994
Cyprus 0.0140 1995
Czech Republic 0.0271 1996 Yes
Denmark 0.0524 1997
Dominican Rep. 0.0793 1986 Yes
Ecuador 0.2040 1998
El Salvador 0.0713 1997 Yes
Estonia 0.0140 1995
Finland 0.0088 1991
France 0.0695 1998 Yes
Gambia 0.1395 1992 Yes
Georgia 0.0896 1995 Yes Yes
Germany 0.0344 1994 Yes Yes Yes
Ghana 0.1782 1996 Yes
Greece 0.0406 1997 Yes
Guatemala 0.0609 1990
Honduras 0.0404 1997
Hungary 0.0274 1998 Yes
Iceland 0.1684 1995
India 0.0622 1991 Yes Yes
Ireland 0.0255 1992
Israel 0.0000 1999
Italy 0.0082 1996
Jamaica 0.0755 1997 Yes
Japan 0.0462 1995 Yes Yes
Kenya 0.1946 1997 Yes
Korea 0.2075 1996 Yes
Latvia 0.0065 1995
Liechtenstein 0.0725 1997
Malawi 0.1659 1994 Yes
Mali 0.0522 1997
Malta 0.0088 1996
Mexico 0.0636 1997 Yes Yes Yes
The Value of a Vote 661
TABLE
3—continued
Country MAL
LC
Year Federal Tier SMD
Namibia 0.0000 1996
Netherlands 0.0000 1996
New Zealand 0.0163 1997 Yes Yes
Nicaragua 0.0596 1996 Yes
Norway 0.0657 1993
Panama 0.0582 1993
Paraguay 0.0405 1993
Peru 0.0000 1993
Poland 0.0174 1997 Yes
Portugal 0.0174 1995
Romania 0.0447 1996
Russia 0.0382 1995 Yes Yes Yes
Senegal 0.0361 1998 Yes
Seychelles 0.0808 1998 Yes
Sierra Leone 0.0000 1996
Slovakia 0.0131 1994
Slovenia 0.0166 1997
South Africa 0.0342 1995 Yes Yes
Spain 0.0963 1996 Yes
Sri Lanka 0.0483 1994 Yes
St Lucia 0.1622 1997 Yes
Sweden 0.0110 1998
Switzerland 0.0193 1995 Yes
Tanzania 0.2619 1995 Yes
Thailand 0.0455 1996
Turkey 0.0859 1995
UK 0.0456 1997 Yes
Ukraine 0.0129 1998 Yes Yes
Uruguay 0.0338 1992
USA 0.0144 1992 Yes Yes
Venezuela 0.0723 1998 Yes Yes Yes
Zambia 0.1725 1996 Yes
TABLE
4Lower Chamber Malapportionment, Twenty
Most-Malapportioned Cases
Rank Country MAL
LC
Rank Country MAL
LC
1 Tanzania 0.2619 11 Chile 0.1509
2 Korea 0.2075 12 Argentina 0.1405
3 Ecuador 0.2040 13 Gambia 0.1395
4 Kenya 0.1946 14 Colombia 0.1324
5 Ghana 0.1782 15 Andorra 0.1307
6 Zambia 0.1725 16 Spain 0.0963
7 Iceland 0.1684 17 Brazil 0.0913
8 Bolivia 0.1677 18 Georgia 0.0896
9 Malawi 0.1659 19 Turkey 0.0859
10 St Lucia 0.1622 20 Seychelles 0.0808
662 SAMUELS AND SNYDER
countries, whereas only two small countries (Iceland and tiny Andorra) are
Western democracies. Thus our data challenge the implicit argument in recent
studies of bicameralism and federalism
26
that malapportionment exists mainly
in upper chambers. Contrary to these claims, our data show that a high degree
of malapportionment also characterizes lower chambers in many countries –
especially those with newly-democratic regimes.
Upper Chamber Malapportionment
Malapportionment in upper chambers (MAL
UC
) ranges from 0.00 to 0.49. Mean
malapportionment in upper chambers is 0.21, with a standard deviation of 0.16.
This score is significantly different (in a two-tailed t-test, at the 0.01 level) from
the average for lower chambers, which confirms the conventional view that
malapportionment is generally greater in upper chambers. Table 5 ranks all
twenty-five upper chambers in our sample according to the degree of
malapportionment. The table also indicates whether the country is federal or not
(only Japan, Italy and Mexico employ an upper tier in their senates, and only
the Czech and Dominican Republics employ SMDs in their senates).
27
A comparison of Table 5 with Table 4 shows that the most-malapportioned
upper chamber scores almost twice as high as the most-malapportioned lower
chamber. Moreover, eleven of the upper chambers score higher than the
most-malapportioned lower chamber. Clearly, malapportionment can be
significantly more acute in upper chambers. At the same time, it bears emphasis
that eight of the upper chambers in our sample score below the mean for lower
chamber malapportionment, and four of those eight have no malapportionment
TABLE
5Malapportionment in Upper Chambers
Country MAL
UC
Federal Country MAL
UC
Federal
1 Argentina 0.4852 Yes 14 S. Africa 0.2261 Yes
2 Brazil 0.4039 Yes 15 Poland 0.2029
3 Bolivia 0.3805 16 Japan 0.1224
4 Dominican Rep. 0.3787 17 India 0.0747 Yes
5 USA 0.3642 Yes 18 Romania 0.0592
6 Switzerland 0.3448 Yes 19 Austria 0.0301 Yes
7 Russia 0.3346 Yes 20 Italy 0.0292
8 Venezuela 0.3265 Yes 21 Czech Rep. 0.0257
9 Chile 0.3106 22 Colombia 0.0000
10 Australia 0.2962 Yes 23 Paraguay 0.0000
11 Spain 0.2853 Yes 24 Uruguay 0.0000
12 Germany 0.2440 Yes 25 Netherlands 0.0000
13 Mexico 0.2300 Yes
26 E.g., Tsebelis and Money, Bicameralism,p.46.
27 Note that less-developed and newly-democratic countries are more evenly spread across the
sample of upper chambers than across the sample of lower chambers (see Table 4).
The Value of a Vote 663
at all. In short, although malapportionment in upper chambers may reach far
higher levels than in lower chambers, upper chambers are not necessarily
malapportioned.
The Correlates of Malapportionment
What explains the significant cross-national variation we observe in the degree
of malapportionment in both lower and upper chambers? To address this
question, we employ regression analysis to explore the relationship between
malapportionment and the following factors: district magnitude; district
structure; federalism; the ‘intensity’ of democracy; country size; and geographic
region. We focus on these six variables because existing work on electoral
institutions suggests they are likely to have an important impact on how shares
of population are translated into shares of legislative seats.
District Magnitude:By‘eyeballing’ the data on lower chambers, we noted that
many single-member district (SMD) systems appear to have above-average
malapportionment (see Table 3). What explains this regularity? There may be a
straightforward, mechanical explanation for this pattern: if the number of members
per district is fixed at one and subnational units (such as provinces, states, special
districts or even islands) cannot be subsumed into other districts for historical or
other reasons, then there will necessarily be some malapportionment. Yet if the
number of members per district is variable, then apportionment constraints are
likely to be less severe, and a greater possibility will exist for achieving low levels
of malapportionment.
There may also be a behavioural reason for the above-average malapportionment
seen in SMD systems: we hypothesize that SMD systems should have more
malapportionment than multi-member district (MMD) systems because, ceteris
paribus, legislators in SMD systems have higher stakes as individuals in
reapportionment decisions. Legislators in SMD systems face a far greater
probability than legislators in MMD systems that ‘their’ district will be targeted for
elimination or redesign in reapportionment processes. Thus we should expect that
legislators in SMD systems often resist reapportionment and that, consequently,
such systems will tend to reduce malapportionment more slowly than MMD
systems.
28
We operationalize this hypothesis by employing a dummy variable that
assigns SMD systems a value of ‘1’ and all other systems a value of ‘0’ (multi-tier
systems with SMD lower tiers are given a value of ‘0’).
District Structure:Dosystems with upper tiers have more or less malapportionment
than systems without tiers? Because upper tiers tend mathematically to reduce
malapportionment, this factor should be associated with lower levels of
malapportionment. We operationalize district structure as a dummy variable: all
28 District magnitude obviously offers at best a partial explanation for cross-national variation in
levels of malapportionment, because countries such as the United States and Australia have very low
malapportionment yet also employ SMDs.
664 SAMUELS AND SNYDER
‘tiered’ systems (N⫽20) are given a value of ‘1’; ‘tierless’ systems are given a value
of ‘0.’
Federalism: Are federal systems more malapportioned than unitary systems?
Because all federal systems provide some form of territorial representation
(although not necessarily according to the principle of equal representation for
unequally-populated territorial units employed in the United States), they should
have higher levels of malapportionment than unitary systems. The regression
analysis below tests whether federal and unitary systems have different degrees of
malapportionment. Federalism is coded as a dummy variable, with federal systems
having a value of ‘1’.
Democracy: Our observation that unconsolidated and recently-consolidated
democracies have especially high levels of lower chamber malapportionment raises
the question of whether the ‘intensity’ of democracy is associated with
malapportionment. Freedom House’s widely-used ranking of countries according
to civil liberties and political rights offers a helpful means for addressing this
question. The Freedom House ranking assigns countries values ranging from 1 to
7, with a score of 1 the most democratic. If a relationship exists between ‘less’
democracy and ‘more’ malapportionment, we thus expect the sign to be negative.
29
Country Size: Are big countries more malapportioned? Some of the world’s largest
countries, such as Australia, Brazil, Canada and Russia, have high levels of
malapportionment in one or both chambers. Because they usually have large
expanses of sparsely-populated territory, big countries may be prone to over-
represent underpopulated regions. To test this possibility, we hypothesize that
malapportionment should increase with country size. Size is measured in square
kilometers.
Region: As noted above, countries in some regions of the world appear to have
especially high levels of malapportionment. To test the relationship between region
and malapportionment, we use dummy variables for the following four regions:
Latin American and the Caribbean, Asia, Europe and North America, and Africa.
We exclude the European/North American dummy as a control. Consequently,
significant coefficients for any of the other three region dummy variables mean that
the countries in the region have levels of malapportionment that differ significantly
from the average level of malapportionment for the European/North American
region.
Table 6 presents regression results (using OLS) for the sample of lower
chambers.
30
The results support some of our arguments. First, only the ‘Latin America’
variable differs systematically from the mean of the excluded region dummy.
Lower chambers in Latin America are typically about 4 per cent more
malapportioned than lower chambers elsewhere. The fact that the ‘Africa’
29 We averaged Freedom House’s ‘Civil Liberties’ and ‘Political Rights’ scores to get a single
number that ranges between 1 and 7.
30 The results were generated using the White correction procedure for heteroscedasticity in STATA
version 6.0. These regressions are free of autocorrelation.
The Value of a Vote 665
TABLE
6Factors Associated with
Lower-Chamber
Malapportionment
Coefficient
Independent variables (s.e.)
Constant 0.028**
(0.013)
Size ⫺1.60E-09
(1.94E-09)
Africa 0.038
(0.020)
Asia ⫺0.001
(0.025)
Latin America 0.039**
(0.018)
Federalism ⫺0.006
(0.019)
Democracy (0.004)
(0.006)
SMD 0.039***
(0.014)
Upper Tier 0.002
(0.016)
R
2
0.25
F2.86
Degrees of Freedom 69
** ⫽p⬍0.05
*** ⫽p⬍0.01
variable is not significant may be due to the divergent effects of the French and
British colonial legacies in that region, especially the different impact on
malapportionment of the multi-member district systems characterizing the
former-French colonies, on the one hand, and the single-member district
systems characterizing the former-British colonies, on the other.
31
This
hypothesis regarding the African region is supported by the score of the ‘SMD’
variable, which shows that single-member district systems are associated with
significantly higher malapportionment (about 4 per cent on average).
None of the other variables is significantly associated with malapportionment
in lower chambers. In particular, note that the ‘upper tier’ variable is not
significant. That is, although upper tiers mathematically reduce malapportion-
ment, in reality the level of malapportionment in the lower tier in many multi-tier
31 We do not test for this difference statistically because of the relatively small number of African
cases in our sample.
666 SAMUELS AND SNYDER
TABLE
7Factors Associated with
Upper-Chamber
Malapportionment
Coefficient
Independent variables (s.e.)
Constant 0.139
(0.083)
Size 1.3E-08**
(0.000)
Africa 0.006
(0.054)
Asia 0.018
(0.083)
Latin America 0.189*
(0.095)
Federalism 0.143**
(0.069)
Democracy ⫺0.052
(0.039)
SMD 0.098
(0.134)
Upper Tier ⫺0.019
(0.053)
R
2
0.50
F3.99
Degrees of Freedom 16
*⫽p⬍0.10
** ⫽p⬍0.05
systems may be large enough to override the mitigating effects of the upper
tier. Thus the use of tiers is not in fact associated with lower malapportion-
ment.
32
Table 7 presents results (using ordinary least squares or OLS) for the sample
of upper chambers. Recall that we had already determined that, in general, upper
chambers are likely to be more malapportioned than lower chambers. We
included the same variables as in the regression for lower chambers. This
regression also provides support for some of our claims: once again, legislatures
in Latin America are significantly more malapportioned than legislatures in the
control region. In contrast to the result for lower chambers, however, the ‘SMD’
variable is not significant, whereas the ‘Federalism’ variable is significant. This
result suggests that upper chambers in federal systems are significantly more
32 We also ran regressions using the percentage of seats in the upper tier and whether the tier was
national or not. In no case was any tier variable shown to be significant. To save on degrees of freedom
in the ‘upper chamber’ regression (because another inference was more important to test, see below),
we do not show these results.
The Value of a Vote 667
malapportioned than upper chambers in non-federal systems. Indeed, by
examining Table 5, we can see that four of the ‘non-federal’ upper chambers
in the sample are perfectly-apportioned and that only three non-federal upper
chambers (all in Latin America) have greater-than-average levels of malappor-
tionment. In addition, the ‘Size’ variable is significant, indicating that
malapportionment is associated with country size, but only in upper chambers.
Unfortunately, with this relatively small sample size it is not possible to separate
the effects of federalism from country size – in fact, these two variables are
correlated, although not perfectly (0.46).
33
In sum, the regression analysis helps pinpoint several key correlates of
malapportionment. Malapportionment is generally more severe in Latin
American countries, in both upper and lower chambers. For all lower chambers
(as well as unicameral systems), the use of single-member districts correlates
strongly with higher levels of malapportionment. However, contrary to the
expectations of some scholars,
34
we found that legislative chambers with upper
tiers do not have lower-than-average malapportionment. Finally, upper
chambers do not necessarily have high levels of malapportionment: large,
federal countries are more likely to have highly malapportioned upper chambers
than are small, unitary countries with bicameral systems.
QUESTIONS FOR FUTURE RESEARCH
The findings presented above suggest several intriguing avenues for future
research on malapportionment.
Malapportionment and Strategic Politicians
One fruitful area for future research involves how malapportionment can be
employed as an independent variable, as an institutional factor that shapes
politicians’ strategies for pursuing policy agendas. For example, by influencing
the ‘costs’ of buying support in the legislature or from voters, malapportionment
may have a decisive effect on the coalition-building efforts of executives.
Overrepresented districts (i.e., districts whose share of legislative seats exceeds
their share of a country’s population) should offer more ‘political bang for the
buck’ than underrepresented districts (i.e., districts whose share of legislative
seats is less than their share of a country’s population). In malapportioned
systems, executives may thus face powerful incentives to build policy coalitions
based on the ‘cheap’ support (for example, in terms of pork per vote) of
legislators from overrepresented districts.
As noted above, recent studies of bicameral, federal systems have already
33 Size and malapportionment are correlated both for the subset of nonfederal upper chambers
(0.29) and federal upper chambers (0.34).
34 For example, Lijphart, Electoral Systems and Party Systems,p.146.
668 SAMUELS AND SNYDER
started to explore how malapportionment affects legislative processes.
35
Although our findings about the cross-national significance of malapportion-
ment provide encouraging support for these efforts, the data we analyse also
suggest the importance of incorporating unicameral and unitary systems into the
study of malapportionment.
36
In addition to probing how malapportionment
affects legislative dynamics and executive strategies in unicameral and unitary
systems, scholars should explore whether different institutional factors mitigate
or exacerbate the impact of malapportionment. Do similar levels of malappor-
tionment have stronger effects on legislative dynamics in federal or unitary
systems? The same question could be posed for bicameral and unicameral
systems.
Malapportionment and Democratization
The impact of malapportionment on the performance and fortunes of
newly-democratic countries offers another intriguing area for future research.
Our data suggest that new democracies are more likely than longstanding
democracies to experience high levels of malapportionment: fifteen of the
twenty countries with the most-malapportioned lower chambers have either
completed transitions to democracy since 1975 or are still undergoing
transitions. Thus malapportionment appears to be a key component of electoral
systems in many newly-democratic countries.
One important question involves whether malapportionment in new democ-
racies introduces a ‘conservative bias’ into the political system. In Latin
America, for example, studies suggest that malapportionment tends to favour
politically-conservative rural districts at the expense of politically-progressive
urban districts.
37
Future studies should address the issue of conservative bias by
exploring how malapportionment shapes the fortunes of both left-wing and
right-wing parties.
A related topic concerns the effects of malapportionment on subnational
politics. In new and emerging democracies, overrepresentation of rural districts
caused by malapportionment might contribute to the maintenance and even
proliferation of non-democratic enclaves at the subnational level.
38
Malappor-
35 Cf. Tsebelis and Money, Bicameralism; William B. Heller, ‘Bicameralism and Budget Deficits:
The Effect of Parliamentary Structure on Government Spending’, Legislative Studies Quarterly,22
(1997), 485–516; Gibson and Calvo, ‘Federalism and Low-Maintenance Constituencies’.
36 For example, in unitary, unicameral cases with high levels of malapportionment, such as
Tanzania, Ecuador, South Korea and Kenya, executives may have strong incentives to build policy
coalitions around overrepresented districts.
37 In this respect, malapportionment may have facilitated transitions to democracy by giving
guarantees to anti-democratic rural elites that their interests would be protected under the new
democratic regime.
38 On the issue of how national-level political and economic liberalization can contribute to the
maintenance of subnational authoritarian regimes, see Guillermo O’Donnell, ‘On the State,
Democratization, and Some Conceptual Problems (A Latin American View with Glances at Some
Post-Communist Countries)’, World Development,21(1993), 1355–70; Jonathan Fox, ‘The Difficult
The Value of a Vote 669
tionment could compel pro-democratic elites at the national level to tolerate
subnational authoritarian enclaves, because these elites may rely on overrepre-
sented, non-democratic localities to secure the national legislative majorities
they need to achieve their policy goals. Ironically, the ability of national-level
incumbents to implement and consolidate democratic reforms in a highly
malapportioned system may therefore depend on winning the overvalued
support of subnational authoritarian elites.
39
At the same time, overrepresenta-
tion of subnational authoritarian enclaves in the national legislature may
strengthen the ability of these subnational elites to fend off efforts by external
groups that seek to reform local politics. Thus malapportionment could
contribute to a process whereby democracy is simultaneously strengthened at
the centre and undermined on the periphery. How malapportionment shapes
local politics in newly-democratic regimes – especially the issue of whether
malapportionment helps protect subnational authoritarian enclaves – is an
exciting question for future research.
The impact of malapportionment on the capacities of new democratic regimes
to sustain painful, market-oriented economic policy reforms also poses
intriguing questions. In highly malapportioned systems, politicians may be able
to sustain such policies by shielding select areas from the costs of reform, or
by compensating them through targeted transfers of public resources. These
kinds of reform strategies seem especially feasible in cases where poor,
sparsely-populated rural areas command a disproportionate share of national
legislative seats. In such cases, a modest amount of well-targeted pork may go
a long way towards securing legislative majorities for reform.
40
Conversely, in
a system with low levels of malapportionment, the strategy of forging legislative
majorities by shielding or buying-off constituencies might prove so costly that
it would be fiscally incompatible with economic reform.
The Evolution of Malapportionment
This study offers a cross-sectional ‘snapshot’ of malapportionment in a large
sample of countries. Here, we have not attempted to develop a longitudinal
(F’note continued)
Transition from Clientelism to Citizenship: Lessons from Mexico’, World Politics,46(1994),151–
84; Richard Snyder, ‘After Neoliberalism: The Politics of Reregulation in Mexico’, World Politics,
51 (1999), 173–204; and Richard Snyder, ‘After the State Withdraws: Neoliberalism and Subnational
Authoritarian Regimes in Mexico’, in Wayne A. Cornelius et al., eds, Subnational Politics and
Democratization in Mexico (La Jolla, Calif.: The Center for US–Mexican Studies, University of
California, San Diego, 1999).
39 Conversely, low levels of malapportionment might work against such dependence on local
authoritarian actors. It may be that levels of malapportionment tend to be higher in new democracies
with powerful rural elites because of the ability of these elites to defend their interests during an earlier
historical era (i.e., the nineteenth century) by gaining overrepresentation in exchange for supporting
the nation-building and state-building projects of urban-based elites.
40 The case of Argentina seems to fit this pattern. See Gibson and Calvo, ‘Federalism and
Low-Maintenance Constituencies’.
670 SAMUELS AND SNYDER
analysis of how malapportionment changes over time, although we regard this
task as an important priority for future research.
41
Such a longitudinal
perspective could be achieved by compiling an index of the frequency and types
of reapportionment across cases. This perspective would help us assess whether
contemporary democracies are moving towards or away from the ‘one person,
one vote’ principle.
42
A longitudinal analysis of malapportionment would also contribute to
developing a comparative theory of the politics of reapportionment. We might
begin to build such a theory by identifying the conditions under which political
incumbents have incentives to support or oppose reapportionment. District
magnitude may have an especially powerful effect on these incentives. In
systems with single-member districts, for example, incumbents may have weak
incentives to redistrict, since reapportionment would appear to carry a strong
risk of eliminating ‘their’ districts. This hypothesis might help explain the
extremely high levels of malapportionment we observed among former British
colonies in Africa and the Caribbean, which all inherited single-member
systems from their colonial pasts.
43
Those interested in developing a theory of the politics of reapportionment
could also benefit from studying the historical origins of malapportionment. In
many contemporary democracies, the roots of malapportionment may lie in
historical processes of state-building and nation-building. Overrepresentation
of rural districts in many Latin American countries, for example, potentially
served as a tool for incorporating rural elites into nation-building projects during
the nineteenth and early twentieth century, as it did for small states during the
US constitutional convention. Because sparsely-populated rural areas would
have faced the unattractive prospect of becoming ‘permanent losers’ to urban
areas in perfectly-apportioned electoral arenas, malapportionment may have
been a concession made by urban nation-builders to rural elites. Future research
41 A few scholars have explored the evolution of malapportionment in the United States (e.g., Cox
and Katz, ‘The Reapportionment Revolution and Bias in US Congressional Elections’); Britain (cf.
Iain McLean and Roger Mortimore, ‘Apportionment and the Boundary Commission for England’,
Electoral Studies, 11(1992), 293–309); and Brazil (e.g., Nicolau, ‘As Distorc¸o˜esnaRepresentac¸a˜o
dos Estados na Caˆmara dos Deputados Brasileiros’). However, longitudinal studies of apportionment
in other countries are scarce.
42 Mexico, for example, appears to be moving towards a less malapportioned system as it slowly
democratizes (personal communication from Jeffrey Weldon, Department of Political Science,
Instituto Tecnolo´ gico Auto´nomo de Me´xico). Likewise, Japan adopted a new electoral system in
1994 that significantly reduced malapportionment – compare the figures in Hata (‘Malapportionment
of Representation in the National Diet’) with our figures, which rely on more recent data.
43 A focus on cross-national variation in who controls apportionment – for example, individual
legislators, parties, courts or an autonomous government agency – could also contribute to developing
a theory of the politics of reapportionment. See Richard Snyder and David Samuels, ‘Devaluing the
Vote: Latin America’s Unfair Elections’ (paper prepared for the Conference on Federalism,
Democracy, and Public Policy, Centro de Investigaciones y Docencia Econo´ micas, Mexico City,
Mexico, 1999).
The Value of a Vote 671
should address whether malapportionment served this historical purpose of
helping ‘bring together’ new nations.
44
The possibility that malapportionment may have played such a role in
historical processes of nation-building raises the question of whether institution-
ally-engineered inequalities of representation could have similar integrative
effects in the future. Despite scholarly warnings about its ‘pathological’ and
‘ethically unjustifiable’ nature, malapportionment might serve as an important
tool for helping contemporary democracies cope with explosive problems
ranging from ethnic conflict to urban–rural inequity.
CONCLUSION
Comparative studies of democratic regimes have largely ignored a crucial aspect
of electoral systems: malapportionment. We have addressed this shortcoming
by providing a method for measuring malapportionment in different types of
electoral systems, showing that significant degrees of malapportionment exist
across a broad range of cases, and identifying some of the correlates of
malapportionment. Our results offer a ready-to-use continuous variable for
regression analysis of a large sample of countries, and we have proposed a
number of key areas for future research.
44 On the important role that federalism played in this regard, see Juan J. Linz, ‘Democracy,
Multinationalism, and Federalism’ (paper prepared for the Conference on Federalism and Democ-
racy, Oxford University, June 1997). We borrow the notion of ‘bringing together’ from Linz. As one
reviewer observed, scholars have long noted the relative stability of many former British colonies,
and the high levels of malapportionment we see among ex-British colonies in Africa may help
explain the stability of some of these regimes. This potential connection between institutional design
at the onset of independence and subsequent regime stability merits further investigation.
671a SAMUELS AND SNYDER
APPENDIX A: DATA SOURCES
Country Source
Andorra Andorra Government website (www.cria.ed/elections)
Argentina Ernesto Cabrera, ‘Multiparty Politics in Argentina?
Electoral Rules and Changing Patterns’, Electoral Studies
15 (1997), 477–95.
Australia Australia Government website (www.acc.gov.au/elect96.
html); Stepan 1997, see fn. 3.
Austria Lijphart Elections Archive (dodgson.ucsd.edu/lij/ausdat.
html).
Barbados Political Database of the Americas (1999) Barbados:
Parliamentary Elections, 1999. Georgetown University and
the Organization of American States, in:
http://www.georgetown.edu/pdba/Elecdata/Barbados/bar-
bados 99.html.
Belize Lijphart Elections Archive (dodgson.ucsd.edu/lij/
blz93b.html).
Benin Personal communication with Professor Shaheen Mozaffar,
Department of Political Science, Bridgewater State Col-
lege.
Bolivia Bolivian Electoral Court website (http://ns.bolivian.com/
cne/).
Brazil Brasil, 1994. ‘Resultados das Eleicoes de 1994’ (computer
files). Brasilia: Tribunal Superior Eleitoral.
Burkina Faso Le Pays (no. 1395) (Monday, 19 May 1997), p. 7.
Canada Elections Canada website (www.elections.ca/election/
results/results-e.html).
Chile El Mercurio website (www.mercurio.cl/eventos/
eleccion97/portfichas.html); Georgetown Political Data-
base of the Americas website (www/georgetown.edu/
LatAmerPolitical/Elecdata/Chile/chile93.html).
Colombia Colombia, 1994. Elecciones del Congresso, 1994 (vols 1
and 2). Bogota: Repu´blica de Colombia.
Costa Rica Costa Rica, 1994. ‘Computo de Votos y Declaratorias de
Eleccion de 1994’. San Jose: Tribunal Supremo de
Elecciones.
Cyprus Lijphart Elections Archive (dodgson.ucsd.edu/lij).
Czech Rep. Czech Republic Elections Commission website
(www.volby.cz/ ASCII /volby/en/)
Denmark Danish Elections website (http://www.dknet.dk/valg94/
english.html)
Dominican Republic Juan Jaramillo, ed, (Poder Electoral y Consolidacio´n
Democratica: Estudios Sobre la Organizacio´n Electoral en
America Latina, 1989). San Jose (Costa Rica): CAPEL, and
DR Electoral Court website (http://jce.do/elecciones98/
bfinal/welcome.htm).
The Value of a Vote 671b
Country Source
Ecuador Ecuador Elections Website (mia.lac.net/opcion96/
resultados/exitdipp.htm).
El Salvador Jack Spence et al., El Salvador: Elections of the Century:
Results, Recommendations, Analysis (Cambridge, Mass.:
Hemisphere Initiatives, 1994).
Estonia Estonian Parliamentary website (http://www.rk.ee/VVK/
r95/e95index.html).
Finland Sami Borg and Risto Sinkiaho, eds, The Finnish Voter
(Tampere: Finnish Political Science Association, 1995).
France Le Monde website (www.lemonde.fr/elections).
Gambia Central Committee of the People’s Democratic Organiza-
tion for Independence and Socialism (PDOIS), ‘Analysis of
the Present Electoral System in the Gambia’, 1992).
Georgia Georgia Elections Results website (205.197.10/
ELECTIONS/respreselec.html).
Germany Personal communication with Uwe Gehring, Research
Assistant, University of Mainz (files obtained via ftp).
Ghana Electoral Commission of Ghana, ‘Presidential Results Data
Sheet’, 1996.
Greece Greek Elections Results website (www.paros.delta-inf.gr).
Guatemala Personal communication with Professor David L. Wall,
Christopher Newport University.
Honduras Honduras Election website (http:lcweb2.loc.gov/frd/cs/
honduras/hn appen.html); personal communication with
Professor Michelle Taylor-Robinson, Texas A&M
University.
Hungary Hungarian Elections Website (http://www.election.hu/
index en.htm).
Iceland Personal communication via email with Professor Gunnar
Helgi Kristinnsson, University of Iceland.
India Butler, David, Ashok Lahiri and Prannoy Roy, India
Decides: Elections 1952–1995, 3rd ed (New Delhi: Books
& Things, 1995).
Ireland Irish Elections Results website (archive/.rte.ie/election97).
Israel No data required (single-district chamber).
Italy Italian Chamber of Deputies website (www.camera.it/
elezdep/circull/home.htm); and Italian Senate website
(www.senato.it/funz/fr siste.htm).
Jamaica ‘The Observation of the 1997 Jamaican Elections’ (Atlanta,
Ga: The Carter Center, 1997).
Japan Migakawa Takayoshi, ed., Seiji Handobukku [Political
Handbook] No. 33 (Tokyo: Seiji Koho Senta, 1997).
Kenya Kenyan Elections website (http://www.kenyaelections.
com/res3/Coast.htm).
671c SAMUELS AND SNYDER
Country Source
Korea Korean Congressional Elections Oversight Committee,
‘15th National Assembly Elections’ (Seoul: Central Con-
gressional Election Authority, 1996).
Latvia Personal communication via email with Laimdota Upe-
niece, Software Developer, Information Department,
Saeima of Latvia.
Liechtenstein Liechtenstein, ‘Das Wahlsystem’ (Liechtenstein: Press and
Information Office, Principality of Liechtenstein, 1997).
Malawi Personal communication with Professor Daniel Posner,
Department of Political Science, University of California,
Los Angeles.
Mali Journal Officiel de la Re´ publique du Mali,12February
1997, p. 40; Commission Electorale Nationale Inde´pen-
dante, Mali, ‘Resultats Provisoires du 1er Tour des
Elections Legislatives du 20 Juillet 1997’.
Malta Lijphart Elections Archive (dodgson.ucsd.edu).
Mexico Instituto Federal Electoral website (www.ife.gov.mx);
Stepan 1997, see fn. 3.
Namibia No data required (single-district chamber).
Netherlands No data required (single-district chambers).
New Zealand New Zealand Government website (www.govt.nz/
elections/e-top.html).
Nicaragua Nicaragua, Elecciones 1996: Proclamaciones de Electos
(Managua: Consejo Supremo Electoral, 1996).
Norway Statistics Norway, Stortingsvalget 1993 (Oslo-
Kingsvingen: Statistics Norway, 1994).
Panama Panamanian Elections website (http://www.
elecciones99.com/infoelec/compendi.html#cargos).
Paraguay Marcial A. Riquelme, Negotiating Democratic Corridors in
Paraguay: Report of the Latin American Studies Associ-
ation to Observe the 1993 Paraguayan National Elections
(Pittsburgh: LASA, 1994).
Peru No data required (single-district chamber).
Poland Polish Elections website (http://www.pap.com.pl/
elections97/results/).
Portugal Personal communication with Domingos Magalha˜es,
Director of Services, Portuguese Electoral Information
Bureau (STAPE).
Romania Romanian Central Elections Bureau website (www.
kappa.ro).
Russia Mikhail Filippov’s website (www.caltech.edu/ ⬃filippov/
fraud/95pl.html); Stepan 1997, see fn. 3.
Senegal Personal communication with Professor Shaheen Mozaffar,
Department of Political Science, Bridgewater State
College.
Seychelles Seychelles Government website (www.seychelles.net/
misd/res1998/htm).
The Value of a Vote 671d
Country Source
Sierra Leone Stephen Riley, ‘The 1996 Presidential and Parliamentary
Elections in Sierra Leone’, Electoral Studies 15 (1997),
537–45.
Slovakia Slovakia Elections Results website (www.eunet.sk/
slovakia/slovakia/elections-94/parl-members), and per-
sonal communication with Miro Sedivy (miro.sedivy@
goecities.com).
Slovenia Slovenia Elections Results website (www.sigov.si/egi-bin/
spl/volitve/preds97/udel ve.htm?language ⫽slo).
South Africa Personal communication with Professor Andrew Reynolds,
University of Notre Dame.
Spain Ministerio de la Justicia y del Interior, ‘Elecciones a Corte
Generales, 1996’ (Computer files obtained via ftp); Stepan
1997, see fn. 3.
Sri Lanka Lijphart Elections Archive (http://dodgson.ucsd.edu/lij/
srlk.html).
St Lucia St Lucia Labor Party website (www.geocities.com/ ⬃slp/
election.html).
Sweden Sweden Elections Results website (www.math.
chalmers.se/ ⬃larsa/Val/valdata/Riks94.html).
Switzerland Swiss Elections website (http://www.admin.ch/ch/f/pore/
nrw95/broch.html); personal communication with Simon
Hug, UC San Diego.
Tanzania Personal communication with Professor Shaheen Mozaffar,
Department of Political Science, Bridgewater State Col-
lege.
Thailand Thailand, Report of the Election to the House of Represen-
tatives, 1996 (Election Division, Department of Local
Administration, Ministry of the Interior, 1996).
Turkey Ali Carkoglu and Emre Erdogan, ‘Fairness in the Appor-
tionment of Seats in the Turkish Legislature: Is There Room
for Improvement?’ New Perspectives on Turkey,19(1998),
97–124.
UK Whitacker’s Almanac for 1998 (London: The Stationery
Office).
Ukraine IFES Ukraine website (http://ifes.ipri.kiev.ua/Elections98/
index.phtml).
Uruguay Alfredo Albornoz, Elecciones (Montevideo: Republica
Oriental del Uruguay, Ca´mara de Representantes, 1992).
USA Congressional Quarterly, Congressional Districts in the
1990s (Washington, DC: CQ Press, 1993).
Venezuela Dieter Nohlen, Sistemas Electorales en Ame´rica Latina: El
Debate Sobre Reforma Electoral (Lima: Fundacio´n
Friedrich Ebert, 1993); and Venezuela, ‘Registro Electoral’
(unpublished reports, mimeos) (Caracas: Consejo Supremo
Electoral, 1993).
671e SAMUELS AND SNYDER
Country Source
Zambia Electoral Commission of Zambia, ‘Presidential and Parlia-
mentary General Elections, 1996, Provisional Results, 25
Nov. 1996’; Republic of Zambia, ‘Report of the Delimita-
tion Commission Established under Article 73 (1) of the
Constitution of Zambia to his Excellency Dr Kenneth David
Kaunda, President of the Republic of Zambia’ (n.d.).
