consequences of electoral systems in africa: a preliminary...

www.elsevier.com/locate/electstud

Electoral Studies 24 (2005) 41–64

Consequences of electoral systems inAfrica: a preliminary inquiry

Staffan I. Lindberg�

Department of Political Science, Lund University, PO Box 52, Lund 221 00, Sweden

Abstract

Based on an original data set of elections in Africa, this article addresses the orthodoxy oftheoretical hypotheses regarding the effects of electoral systems. While Africanists assume that

context make a significant difference, the empirical analysis points to very similar effects ofelectoral institutions in Africa as in established democracies; regarding reductive effects, effectson the party system, governing capacity, competition and fairness of elections. Only the logicof accountability diverges from the theories of constitutional design. In addition, the empirical

analysis seems to refute Sartori’s argument that party systems must be structured beforeelectoral systems can mature. Rather, the analysis of these emerging democracies suggests thatelectoral systems play a role in structuring party systems.

� 2004 Elsevier Ltd. All rights reserved.

Keywords: Electoral systems; Elections; Africa; Democracy; Political parties

1. Introduction

Africa’s (re)-entry into democratic politics presents a challenge to students ofAfrican politics, although systematic comparative accounts of democratizationprocesses are still rare. The contributions by Bratton and van de Walle (1997) andNohlen et al.’s (1999) voluminous reviewing of elections in Africa remain the primaryauthorities on comparative African politics. One of the understudied areas – despite itsfocal interest in comparative politics – is the study of electoral systems. A few recentcontributions deserve to be mentioned. Reynolds and Sisk’s (1998) analysis of thepotential of electoral systems for conflict management in Africa, Bogaards’ (2000)

�Corresponding author. Tel.: C46-46-222-4554; fax: C46-46-222-4006.

E-mail address: [email protected] (S.I. Lindberg).

0261-3794/$ - see front matter � 2004 Elsevier Ltd. All rights reserved.

doi:10.1016/j.electstud.2004.02.006

http://www.elsevier.com/locate/electstud

mailto:[email protected]

42 S.I. Lindberg / Electoral Studies 24 (2005) 41–64

discussion of the consequences of electoral laws, as well as Horowitz (1991) work onconstitutional recommendations for divided societies, and Foweraker and Landmann(2002) study of the effects of electoral systems on democratic performance areespecially insightful. Barkan’s (1998) study stands out as the only one of its kind onagrarian societies that has sought to develop and challenge theoretical insights in thefield of comparative politics based in Africa.

This article seeks to both synthesize earlier efforts and deepen the analysis whiletesting established hypothesis against empirical evidence gathered between 1989 and2001. From the characterization of electoral systems as portrayed in the literature, fiveexpected consequences of electoral systems follow (here presented in a simplifiedmanner leaving aside the various conditions of each hypothesis).

1. Majoritarian systems have a stronger reductive effect on the number of partiescompeting for and attaining seats in parliamentary elections, than do PR systems.

2. Majoritarian, and in particular plurality systems, produce or sustain bi-partysystems whereas PR systems are associated with multi-party systems.

3. Majoritarian systems create substantial legislative majorities providing a highergoverning capacity than do PR systems.

4. Majoritarian systems sustain lower levels of popular participation than do PRsystems.

5. Majoritarian systems provide clearer accountability for voters than do PRsystems.

African realties have sometimes challenged assumptions about the applicability oftransferring and mapping of outside models onto African politics. Yet, establishedconcepts and causal inferences generated by the study of electoral systems inestablished democracies remains the best way of gleaning useful insights into whatactually is, as well as offering understanding to deviant situations in Africa. Nohlen(1996), for example finds that electoral systems outcomes in Latin America largelyresemble those in other parts of the world. It is an important task to demonstrate thediversity of experience on the African continent as well as to confirm the relevance ofincluding African data in the comparative study of electoral systems.

The first section of this article briefly discusses the core characteristics and implica-tions of the majoritarian and the proportional electoral systems, followed by anempirical analysis based on a data set of African elections collected from 1989 to2001 to examine five classical issues: – reductive effects, effects on the party system,governing capacity, participation, and accountability. In addition, the implicationsof electoral systems on the fairness of elections on new and fragile democracies areanalyzed. The third and final section concludes on the main findings in the widercontext of comparative politics.

2. The orthodoxy of institutional design

Competitive elections are the hallmarkofmodern representativedemocracy.Whereasstudents of democracy differ on the relative importance they ascribe to elections in

43S.I. Lindberg / Electoral Studies 24 (2005) 41–64

defining the concept, no one seems to denounce them as an unimportant. Being thelargest peacetime mobilization in which a large proportion of the citizenry participate inselecting and removing political representatives, it provides a primary source ofdemocratic legitimacy (cf. Mozaffar and Schedler, 2002). Di Palma’s (1990) truism ‘todemocratize is to craft institutions’ has often been referred to, yet, institutional reformand engineering remains a neglected area in the study of democratization processes.

While it is selfevident that differences in socio-cultural and political differencesacross countries and continents must be recognized, it is equally important to payattention to issues of constitutional choice. Especially during the processes ofdemocratization since ‘‘.onemust recognize that electoral system reform is perceivedas a key, or perhaps the key, to reforming the political system’’ (Nohlen, 1996, p. 44).Electoral rules define, among other things, the electoral formula, district magnitudes,district boundaries, assembly size, suffrage rights, rules of representation, electoralcalendar, voter, party and candidate eligibility and registration, election observation,resource endowments of parties and candidates, methods of counting, tabulating andreporting of votes, election management bodies and dispute settlement authorities (cf.Mozaffar and Schedler, 2002; Sartori, 1997). What is usually referred to as ‘electoralsystems’ regards the rules concerning both voting method and the method used intranslation of votes into seats in the representative body (Sartori, 2001, p. 99).

Hence, electoral systems are a system of rules, which in essence make aninstitution (March and Olsen, 1989). Institutions delimit political behavior and areoften regarded as a powerful variable in explaining government performance.Institutions rarely, if ever, operate in a vacuum; rather, they are best conceptualizedas ‘embedded’ (Grofman et al., 1999; Przeworski and Teune, 1970). At a minimum,this is a cautionary note that electoral systems are not the only factors influencingoutcomes; path-dependencies might mitigate against expected effects of electoralsystems; and socio, political, cultural, and economic conditions may conditionoutcomes in significant ways (Nohlen, 1996, p. 45). Even in carefully designed cross-national studies, direction of causality is often difficult to establish (e.g. Fowerakerand Landmann, 2002). Even so, the extension of the law-like consequences ofelectoral systems first developed by Duverger (1954) and Downs (1957) have beentestified by the work of scholars like Bogdanor and Butler (1983), Lijphart (1984,1994, 1999), Lijphart and Waisman (1996), Mair (1990), Powell (1982, 2000), Rae(1971), Reynolds and Sisk (1998), and Sartori (1968, 1986, 1997).

Despite the almost complete consensus among scholars that institutions likeelectoral systems matter because they structure incentives, preferences and outcomes,there are very few studies on the effects of distinct choices in constitutional design innew democracies. ‘Consolidologists’ tend to agree that consolidation of democracyinvolves central political institutions alongside elite behavior and mass attitudes (e.g.Diamond, 1999; Gunther et al., 1995; Linz and Stepan, 1996; O’Donnell, 1996;Schedler, 1998, 2001; Valenzuela, 1992). Yet, few seem to ask: Does it matter whattypes of institutions are put in place? Are majoritarian, mixed or proportionalelectoral representations doing better or worse? What are the effects of the con-stitutional design on party systems, governing capacity, political competition and thequality of elections in new democracies? These are the theoretical issues this article


seeks to address by looking at empirical data from Africa. To date, no such cross-national study of Africa has been done.

An empirical analysis of the applicability of these findings to African requiresfirst, identification of the distinct characteristics of each of the electoral systems at asufficiently a high level of precision and in a way that makes theoretical sense to theirconceptualization as reciprocally exclusionary categories. Second, the proposedconsequences of each system must be put in operational terms. In other words, theempirical implications of each proposed effect must not only be falsifiable but alsomutually exclusive.

2.1. Majoritarian electoral systems

There are two main versions of the majoritarian system: single – member pluralityand absolute majority. The first is the more typical in terms of anticipated effects.Regardless, the imperative of the majoritarian vision is governing capacity, or in otherwords, the creation of stable legislative majorities. The translation of votes-to-seats istypically highly disproportional through the winner-takes-all in single-memberconstituencies. It is designated to have a strong reductive effect on the number ofparties competing for legislative seats and the number of parties in parliament. Ideally,it leads to a de facto biparty system with intense competition but clear legislativemajorities, hence high governing capacity and one party executives. Electoralcompetition will typically focus on the median voter with a tendency to excludeextremist political supporters and peripheral voting populations. (Lijphart, 1984,1999; Reynolds and Sisk, 1998, p. 23; Powell, 2000, pp. 22–23; Weaver, 2002, p. 112)

The possible dangers of majoritarian systems are predominantly in relation todivided societies where disproportionality is thought to engender alienation throughexclusion from power by minorities; a high portion of wasted votes. Hence, lowerincentives for participation, fewer women representatives (Lijphart, 1984, 1999);raise the stakes in the game by it’s winner-takes-all nature, and further clientisticvoting behavior through the close personal relationship between voter andrepresentative in single-member districts (e.g. Reynolds and Sisk, 1998, p. 24).This is in essence what is often referred to as adversarial, as opposed to consensualpolitics. The current literature has very little to say on the fairness of elections. Yet, itfollows from the winner-takes-all of majoritarian systems that the stakes are higher,hence the incentive for using irregular practices to win is stronger than inproportional systems. For political leaders in new democracies with a half-heartedcommitment to democracy at best, such incentives are likely to present a strongertemptation than to leaders in established democracies. It seems reasonable to assumethat the stronger the incentives for cheating, the stronger the guardians of fairnessand sanctions against fraud must be.

2.2. Proportional systems

Proportional representation (PR) systems differ in their degree of proportionalityin the translation from votes to seats. Generally, the higher the district magnitude


and the lower the effective threshold, the greater its proportionality. Yet, countrieswith smaller district magnitudes sometimes have compensatory seats for allocationto improve representation. A few countries in Africa also have reserved seats fordifferent minorities and for women to ensure that all portions of the population arerepresented.

The overall imperative of the proportional vision, is representative justice. It is nothowever as Sartori (2001) argues, that the PR system have no reductive effect. Theproportional design lessens the reductive effect although it is always presentsomewhat. The numbers of parties competing for votes and winning legislative seatsare typically greater than in majoritarian systems. Orthodoxy has it that there isa trade off between representative justice and governing capacity. PR systems arethought to lead to legislative-executive deadlocks in presidential systems and short-lived unstable coalition governments in parliamentary systems hence lowergoverning capacity. The level of competition can be high or low depending on thenumber and relative strength of the parties, yet, typically the relative votes – andseats shares of the parties tend to be lower than in majoritarian systems (Sartori,1997, pp. 54–55). Again the literature is almost completely silent on the issue of thepossible effects on the democratic quality of elections in new democracies. We mayinfer from the arguments by proponents of the PR vision that the effects are good.Minorities get a better representation reducing the incentives for anti-democraticbehavior. Several parties by extension groups in society, have a say in theformulation of policies, which reduces probability for selective redistributive policiesto be implemented (Weaver, 2002). In sum, PR is thought to breed peaceful conflictresolution and consensus making in politics, which should be an asset in the contextof a nascent unstable political democracy. Barkan (1998) for example, challenges thisview asserting that the more agrarian the society, the higher the geographicconcentration of the vote and the more closely will the distribution of seats undera majoritarian system be mirrored by the distribution through PR. He posits thatvoters evaluate party candidates on constituency service level rather than on nationalpolicies. Hence, party-list PR systems risk further alienation of citizens from bothdemocracy and the state.

2.3. Mixed variants

Some authors (e.g. Weaver, 2002) ascribe particular alternatives such as thealternative vote (AV) used in the Australian House of Representatives, and the singletransferable vote (STV) as mixed electoral systems. I generally prefer Sartori’s (2001,p. 99) proposal to use to the label exclusively for electoral systems where both thevoting method and the allocation of seats are in part majoritarian and in partproportional. Such systems, it has been argued, offer the best solution to satisfy thetwo main, though contrary imperatives of representative justice and governingcapacity (Dunleavy and Margetts, 1995). Empirical cases have been rare, however,although the African context now can increase that number with at least fivecountries.


3. Electoral systems in Africa: expected and empirical consequences

Africa represents a particularly interesting case of the relationship between thechoice of electoral system and its effects for a number of reasons. First,contemporary Africa presents a new empirical field for the comparative study ofconstitutional engineering. 180 competitive elections have been held 1989–2001, ofwhich 99 were parliamentary elections. Six countries conducted multiparty electionsimmediately prior to 19891 and continued in the 1990s. Including these elections, thetotal N is 116 cases of parliamentary elections.2 Every individual poll has been givenstatus as a case. Second, a wide range of electoral systems can be found in Africatoday. Different majoritarian- and PR-systems as well as mixed ones haverepresentation among the countries in Africa. Thus, the main thrust of thetheoretical issues can be addressed. An overview of the countries using each systemand some essential characteristics are found in Table 1.

Third, conventional knowledge on electoral systems is based primarily onestablished democracies. Such a context presents both voters and leaders with anempirical reality distinct from that in Africa, which is likely to influence the con-sequences of electoral systems (Nohlen, 1996; Sartori, 2001). Accordingly, one wouldexpect electoral system effects to deviate from the conventional understanding.Hence, it seems that the comparative study of electoral systems should havesomething to learn from a regional study of Africa.

1 Botswana from 1969, Mauritius from 1976, Zimbabwe from 1980, Gambia and Madagascar from

1982, and Senegal from 1983.2 The data collection was cumbersome since no source could provide reliable and at the same time

comprehensive data for all the 48 African states. Because of unreliability and unavailability of data, many

sources were consulted. Data on each single election in every country was pooled by the author in country

profile files. From that pooled file system, the data set was constructed. The data set is available at the

author’s website http://www.svet.lu.se/Staff/Personal_pages/Staffan_lindberg/Staffan_lindberg.html.

The full documentation, coder’s translation and references can be obtained from the author upon

request. The sources used to pool this information were: Nohlen, Dieter, Michael Krennerich and

Bernhard Thibaut, 1999. Elections in Africa Oxford: Oxford University Press, Keesing’s Record of World

Events sourced from http://keesings.gvpi.net/keesings/lpext.dll?f=templates&fn=main-h.htm&2.0/,

Michael Bratton and Nicholas van de Walle, Democratic Experiments in Africa: Regime Transitions in

Comparative Perspective (New York: Cambridge University Press, 1997); Michael Bratton, 1998 ‘Second

Elections in Africa’, Journal of Democracy, (July) Vol. 9(3); Reph Uwechue, ed., Africa Today, 3rd edit.,

(London: Africa Books 1996) John F. Clark, and David E. Gardiner Political Reform in Francophone

Africa, (Boulder CO: Westview Press 1997); Adebayo Olukoshi ed., The Politics of Opposition in

Contemporary Africa, (Uppsala: Nordic Africa Institute 1998). UN IRIN News Reports, various dates,

http://www.irinnews.org; Election Watch, Nos. 1–12, Journal of Democracy, various years 1990–2001;

IPU Chronicle of Parliamentary Elections, Nos. 25–33, (Geneva: IPU 1992–2000); IPU Parline, http://

www.ipu.org/parline; IFES Elections Statistics, http://www.ifes.org/eguide; Elections Around the World,

http://www.agora.stm.it/elections; NDI Elections observation reports, various issues; Independent

Election Commission; various countries, BBC News OnLine; various dates; BBC World News; various

dates.

http://www.svet.lu.se/Staff/Personal_pages/Staffen_lindberg/Staffen_lindberg.html

http://keesings.gvpi.net/keesings/lpext.dll?f=templates&fn=main-h.htm&2.0/

http://www.irinnews.org

http://www.ipu.org/parline

http://www.ipu.org/parline

http://www.ifes.org/eguide

http://www.agora.stm.it/elections


3.1. Reliability

In this kind of research, the two principal threats to reliability are biases in thesources consulted and subjectivity in the coders’ scoring. Contamination has beenminimized by the use of multiple sources of information whenever possible. For lessthan 10% of the 116 cases of elections, there was only one source. For most of theothers, three or more sources have been consulted. The net effect of the filteringshould therefore be minimized. Fewer sources are typical for elections in the categoryof pre-1990 elections, yet these elections tend also to be less problematic to code andnews-agencies archives have been very helpful in this regard. The effect of eventualerrors has also been lessened by the use of ordinal rather than binary categories foreach such indicator. Non-subjective indicators besides party composition and voterturnout are indeed rare (cf. Bollen, 1990). However, even voter turnout figures areoften loaded with measurement errors in Africa. Population figures are frequentlyrough estimates and voter’s registries, if they exist, are often highly unreliable. Theprocedure was to pool the information obtained from all of the sources beforemaking a judgment. Where sources disagreed, I was forced to use my own judgmentto discriminate sometimes after consulting additional sources. This is a practicevirtually inevitable in empirical research including research based on case studies(Bollen, 1990; Coppedge and Reinicke, 1990; Gasiororwski, 1996). In line with thenorms of social science data collection (King et al., 1994) the data set, coder’stranslations, and detailed background data files with full references to sources, havebeen made freely available for inspection.

3.2. Reductive effects

The first expected consequence regards the strength of the reductive effect.Majoritarian systems are expected to have a strong reductive effect whereas PRsystems (to varying degree) are expected to have less. We distinguish here betweenthe following systems (in rank order from strongest to least strong effect): plurality insingle-member constituencies, absolute majority in single-, or multi-memberconstituencies, mixed systems, PR with small- to medium district magnitude, andPR with large district magnitude or pure PR. The reductive effect is measured bydeducting the number of parties winning legislative seats from the total number ofparties competing in the elections (Nohlen, 1996, p. 49). That measure can bediscussed, however. First, countries may have more or less parties campaigning thanexpected for reasons other than electoral systems. Hence, the reductive effect ofa particular electoral system may vary significantly over cases. Second, and moredamaging, one of the electoral systems in particular, is highly likely to reduce notonly the number of parties gaining seats in the legislature but also the number ofparties actually running for office. That is the plurality system. By its typicalexclusionary logic, it is likely to reduce the number of parties before the actualcampaign starts in almost any context. Bearing this in mind, this measure will beapplied here simply to gauge the reductive effects and make the results compatiblewith established findings in comparative politics. A more valid measure would be to

Table 1

Electoral system, countries and key characteristics

% Seats,largest party

(geo.mean)

# Turn-overs % Turnout(geo.mean)

82 1 6070 1 34100 0 57

32 0 9261 0 5669 1 54

53 0 6699 0 7237 1 57

48 0 8579 1 8461 0 41

n.a. 0 6191 0 84n.a. 0 65

83 1 4660 1 65

56 0 63

41 0 6346 0 3671 0 50

65 1 2179 0 5531 1 50

94 0 5965 0 44

81 0 39

52 0 4662 0 6285 0 86

48

S.I.

Lindberg

/Electo

ralStudies

24(2005)41–64

Electoral system Country Nelections

Nfree

Reductiveeffect

(geo.mean)

# Effectiveparties

Plurality inSMC or MMC

Botswana 7 7 2.2 1.4210Cote d’Ivoire 3 0 7.3 2.7941Djibouti 2 1 5.0 1.0000

Ethiopia 2 2 8.5 5.7290Gambia 4 1 2.0 1.1361Ghana 3 3 2.1 2.1652

Kenya 2 0 4.9 3.2205Lesotho 2 2 4.5 1.0253Madagascar 4 2 118 4.8941

Malawi 2 2 5.9 2.6843Mauritius 7 6 13 1.4765Nigeria 3 1 1 2.2187

Swaziland 2 0 n.a. n.a.Tanzania 2 1 8 1.2025Uganda 2 2 n.a. n.a.

Zambia 3 1 3 2.8852Zimbabwe 5 2 1 1.9223

Absolute majority

in SMC or MMC

Cameroon 2 0 35 2.0924

C.A.R. 2 2 7.4 4.2640Comoros 3 0 6.6 n.a.Gabon 3 1 23 1.5773

Mali 3 1 2 1.3109Mauritania 3 1 9 1.5820RoC 2 1 21 n.a.

Sudan 2 0 6 1.0284Togo 2 0 13 1.0408

Mixed systems Burkina Faso 2 0 13 1.2264

Chad 1 1 28 2.9288Guinea 1 1 14 2.3304Seychelles 2 2 2 1.2765

PR in small MMC Benin 3 3 7.3 6.1679 25 0 66Cape Verde 3 3 1.3 2.0717 65 2 69

0 10 1.1343 89 0 80

2 4.6 3.6654 49 1 853 5.9 3.3271 38 2 373 2.9 2.3596 55 2 63

3 8.8 2.4000 70 1 48

1 6 n.a. 59 0 911 4 n.a. 80 1 911 10 1.6625 76 0 81

2 12 1.9928 53 0 743 2.6 1.6570 68 1 781 9 n.a. 33 1 50

2 8 2.1504 64 1 87

ctions only. When no such exists, figures are given for unfree elections in italics. Total means for each

nd main analysis, elections rather than countries as here, is the unit of analysis. Note also that ‘Turnovers’

ities but includes founding elections.

49

S.I.

Lindberg

/Electo

ralStudies

24(2005)41–64

Eq. Guinea 2

Guinea-Bissau 2Niger 4Sao Tome 3

Senegal 5

PR in large MMC Angola 1Burundi 1Liberia 1

Mocambique 2Namibia 3S. Leone 1

South Africa 2

Note: Figures generally represents free and fair ele

category are not reported because in the following a

report only unambiguous shifting legislative major


deduct the effective number of parliamentary parties from the effective number ofelectoral parties. In practice, this approach is impossible because data on votes perparty is too often not available. We simply have to do with what we have whilerecognizing the limitations of our measure.

The reductive effects of electoral systems in Africa, with control for free andfairness of elections are presented in Table 2.3 It seems reasonable to assume thatelections that are not free and fair will result in figures that are manufactured to someextent. Electoral fraud, voter intimidation, unfair campaigns and exclusionist policieswill necessarily affect the number of parties participating, winning seats, voterturnouts and so on. Concordant to common sense, the reductive effect for unfairelections is substantially higher on average than for elections that are free and fair.

Free and fair elections, confirms the expected relationship between absolutemajority-, mixed- and PR with large district magnitude or pure PR systems, thus themore proportional the electoral system the less reductive effect. We also find that theplurality system has a much lower reductive effect than the two-round absolutemajority system and the more proportional PR system has a greater reductive effectthat less pure PR. How do we explain this? Without a doubt, some countries havea particular number of parties for other reasons than electoral system. This isdiscussed in further detail below with regards to party systems.

Concerning the deviation by the plurality system we should recognize that thiselectoral system much more than the others, has consequences for how many partiesthat actually compete for seats. The exclusionary logic of the plurality system detersmany would-be political entrepreneurs from running on their own party platforms.Most political entrepreneurswhowish to enter the gamebut find themselves likely to bethe fourth, fifth or even sixth party, will presumably realize that the chances of winningare slim. A better strategy then is to join one of the existing parties. Especially if partiesare not programmatic but clientilist in nature which tends to be the case in newdemocracies inAfrica. In all other systems fromabsolutemajority to purePRa long listof parties can calculate to be better off on their own. In absolute majority systemspolitical entrepreneurs canwisely run in the first round to be able to sell their support toany of the major contestants in the second round for the highest possible price. Thisseems as a particularly good strategy in clientistic systems where patronage or tributedetermine the success or failure of political leaders. Similarly, actors in PR systemsmayhope for a seat in the legislative assembly only to offer their services to the highestbidder in case of an unclear legislative majority in parliament.

The theoretical expectation of reductive effects has implications for time-variationas well. In new democracies the differences between the electoral systems are likely toplay out as the political players learn to know the incentives and pay-offs. Table 3shows that all elections in countries that have performed at least three successive

3 The reader should note, however, that I use geometric-, rather than arithmetic, means in all tables.

The reason is that we find these extreme cases in all categories as the standard deviations testify to. The

geometric mean is not as sensitive to extreme figures as its arithmetic cousin is (e.g. Blume, 1974; Datton

et al., 1998).

Table

Elect

Elect Free and fair elections

Reductive

effect

Parties

particip.

Parties in

Parliament

Reductive

effect

Plura 4.3 9.9 4.2 4.6

(4.2) (32.8) (8.7) (29.1)

20 32 32 32

Abso 13.7 22.2 11.5 8.8

(12.2) (11.9) (4.0) (9.2)

13 6 6 6

Mixe 13.1 12.2 5.3 6.3

(7.1) (16.0) (3.8) (12.4)

2 4 4 4

PR s 10.1 11.5 6.7 4.3

(1.5) (14.6) (11.4) (4.8)

5 17 17 17

PR m 12.3 5.2 5.7

n.a. (4.6) (3.6) (4.3)

11 11 11

Tota 7.4 11.5 5.4 5.1

(9.0) (23.8) (8.5) (20.3)

40 70 70 70

Note er transparency. The reductive effect however, is calculated for

each nit rather than the aggregated figures above.

51

S.I.

Lindberg

/Electo

ralStudies

24(2005)41–64

2

oral system and reductive effect

oral system N Missing Unfree elections

Parties

particip.

Parties in

Parliament

lity Geo. Mean 8.7 3.4

Std.D. (5.2) (2.1)

N 57 5 20 20

lute majority Geo. Mean 20.5 4.9

Std.D. (12.7) (3.6)

N 20 1 13 13

d Geo. Mean 19.4 6.3

Std.D. (11.3) (4.2)

N 6 0 2 2

mall MMC Geo. Mean 13.4 3.1

Std.D. (0.9) (0.8)

N 22 0 5 5

ed./large MMC Geo. Mean

Std.D. n.a. n.a.

N 11 0

l Geo. Mean 12.6 3.9

Std.D. (10.2) (2.9)

N 116 6 40 40

: The #parties participating and obtaining seats in parliament are only displayed for great

election separately and the geometric means are based on an analysis with elections as u

Tab

Elec

Elec

syste

h seats Reductive effect

d 3rd 4thC 1st 2nd 3rd 4thC

Plur

SM

3 2.5 0 0 1 4.4

4 – 3 1 3 –

23 9 4 3 98 142

5 6.4 13 14 10 13.6

7 – 4 13 8 –

3 3 0 0 5 5.3

Maj

SM

5 – 12 23 31 –

.a. 8 – 2 n.a. 36 –

7 – 11 9 11 –

PR 20 – 2 13 15 –

3 – 0 2 1 –

3 – 1 5 5 –

6 23.2 11 12 8 9.9

PR 5 – 3 3 2 –

Tota

m

.5 5.8 5.3 3.2 4.7 7.5 13.5

Tota 14 13 14 13 14 13

Note the electoral cycle because of a coup or similar events.

52

S.I.

Lindberg

/Electo

ralStudies

24(2005)41–64

le 3

toral system, number of parties and reductive effect over several elections

toral

m

Country #Parties #Parties wit

1st 2nd 3rd 4thC 1st 2n

ality in

C or MMC

Botswana 4 4 4 6.9 4 4

Ghana 6 5 7 – 3 4

Madagascar 8 8 121 151 4 5

Mauritius 16 18 15 20 3 4

Zambia 6 17 15 – 2 4

Zimbabwe 4 4 8 8.8 4 4

ority in

C or MMC

Gabon 20 36 36 – 8 13

Mali 12 n.a. 44 – 10 n

Mauritania 14 12 16 – 3 3

in small MMC Benin 14 31 35 – 12 18

Cape Verde 2 5 4 – 2 3

Sao Tome 4 8 8 – 3 3

Senegal 14 14 14 33.1 3 2

in large MMC Namibia 10 8 7 – 7 5

l geometric

ean:

7.8 10.1 14.4 15.9 4.1 4

l N elections 14 13 14 13 14 13

: All these countries have held at least three successive election without a breakdown of


elections without interruption, are compared. The number of parties campaigningincreases over time in most countries across electoral systems, even if thedevelopment is generally less pronounced in PR-systems. The reductive effectincreases over time in five out of seven plurality-, and two out of three majoritariansystems but only in two out of five PR systems. The only two cases of decreasingreductive effect are found among PR systems. It seems that much of the expectedlogic of electoral systems with regards to the reductive effect over time plays out alsoin the empirical realities of Africa.

3.3. Party system effects

Second, we expect electoral systems to affect the party system indirectly via thereductive effect. We expect plurality in single-member districts to be associated withbi-party systems with an effective number of parties below 2.0. Absolute majority insingle or multi-member districts typically displays more parties but only two majorparties in the second round of voting. Hence, the effective number of parties inparliament is expected to be higher than for plurality systems but still close to 2.0.Mixed systems are expected to produce more effective number of legislative partiesthan majoritarian systems but less than variants of PR. PR with small- to mediumdistrict magnitude should be associated with above 2.5 effective number of parties inthe legislature while we expect PR with large district magnitude or pure PR to drawan average above at least 3.0.

Africa presents a challenge for those interested in party systems. All but one of thecategories of electoral systems have cases with extremely many parties running forseats of the legislature. The extreme par excellence is Madagascar operatinga plurality system with 151 registered parties running in the last election while onlynine won seats. Similarly, Cameroon using a two-round majority system had 45parties registered and seven gained legislative shares; Senegal with her mixed systemhit a record in the last election when 61 parties vied for votes and no less than 49 ofthem won at least one seat in the legislature, although forty of them are part of theloose coalition (‘Sopi’) supporting President Wade’s Senegalese Democratic Party.Senegal’s mixed system is also highly disproportional. The Sopi-coalition gained63.8% of the legislative seats with only 49.6% of the votes after a campaign that wasmarked by civil violence. Among the countries with disproportional PR systems,Benin holds the top position with 35 registered parties running for the elections in2001 and as many as twenty of those won legislative seats. Surprisingly, it is onlyamong the more proportional PR systems in Africa that we find no such extremecases. South Africa top the list with 16 registered parties and 13 in parliament.

It might seem to follow from the above discussion that the party system effects ofinstitutional design in Africa are less than concordant to that in the establisheddemocracies and Latin America. One could think that it reminds more of someEastern European countries, for example Poland where majoritarian design has notprevented a large number of parties from gaining seats in the upper chamber(Gebethner, 1996). Note that in Table 4 the difference between unfair and freeelections is replicated with regard to the number of effective parties. It is generally

Free and fair elections

Party

system

Governing

capacity

d Party’s

seats

Effective

# of parties

Winner’s

% seats

2nd Party’s

% seats

1.98 68% 11%

.4) (1.652) (20.9) (12.0)

10 32 32

2.60 52% 11%

.3) (1.896) (19.3) (6.9)

2 6 6

2.06 69% 15%

7) (0.837) (16.8) (5.8)

3 4 4

3.10 48% 22%

9) (1.570) (17.5) (8.7)

6 17 17

1.85 61% 21%

. (0.246) (13.9) (12.9)

4 11 11

2.24 60% 14%

.5) (1.412) (20.4) (11.5)

25 70 70

only for each country. For the other categories, the means

54

S.I.

Lindberg

/Electo

ralStudies

24(2005)41–64

Table 4

Electoral system, party system and governing capacity

Electoral system N Missing Unfree elections

Party

system

Governing

capacity

Effective #

of parties

Winner’s

% seats

2n

%

Plurality Geo. Mean 2.24 72% 9%

Std.D. (0.850) (19.9) (15

N 57 4 5 21 21

Absolute majority Geo. Mean 1.36% 60% 8%

Std.D. (0.443) (24.2) (13

N 20 1 5 13 13

Mixed Geo. Mean 1.23 81% 8%

Std.D. n.a. (12.2) (4.

N 6 0 1 2 2

PR small MMC Geo. Mean 1.34 85% 9%

Std.D. n.a. (9.4) (3.

N 22 0 1 5 5

PR med./

large MMC

Geo. Mean

Std.D. n.a. n.a. n.a

N 11 0

Total Geo. Mean 1.64 70% 8%

Std.D. (0.795) (20.7) (13

N 116 5 12 41 41

Note: The effective number of parties is calculated on the outcome of the last parliamentary election

are calculated on the total number of elections for each electoral system.


much higher for free and fair elections, which makes perfect sense. The exception isplurality systems. This system seems to generate, or sustain, an essentially bipartisansystem in either case. Of course, it is a strategic strength even for facade democraciesto have at least a nominal opposition party with some legislative strength for reasonsof international legitimacy.

Looking at the free and fair elections, the general pattern seems to corroborate theview that electoral systems have similar consequences for the party system regardlessof context. The more majoritarian and disproportional the electoral system, thefewer effective number of parties (given here, that the mixed systems like the ones inthe Seychelles and Guinea, are in effect highly disproportional). The puzzle is themost proportional PR systems. In countries like South Africa, Mozambique,Angola, Liberia, and Namibia the (close to) pure PR system generate, or at theminimum sustain, an essentially biparty system. This apparent paradox has beennoted by several authors (e.g. Barkan, 1998; Bogaards, 2000). Historicalcontingencies can do a long way to explain this situation. All have a history of anentrenched conflict line between two main contenders in national politics. In severalof these countries, it has led to extended civil wars between two rival armies. It is nowonder that such a trajectory leads to a electoral competition along the same linesdespite, and not thanks to, relatively pure PR systems. Nevertheless, in the longerrun a PR system might make it easier to break up these dangerous, previously highlydeadly lines of division.

3.4. Governing capacity

Plurality in single-member constituencies and majority in single-, or multi-memberconstituencies, are thought to be associated with substantial legislative majoritiesimplying a high governing capacity. Mixed systems may or may not produce clearlegislative majorities. Yet, PR with small- to medium district magnitude, and toa greater extent PR with large district magnitude or pure PR, are typically expected toconjure with unclear and/or weak legislative majorities. I suggest measuring governingcapacity directly by the strength of legislative majorities using two indicators: winningparty’s share of legislative seats and second party’s share of the same. Note that withregard to the former, I measured only the largest party’s seat shares and not winningcoalitions. The measurement of the next largest party is to indicate the strength of themajor opposition party in the legislature.

Governing capacity as discussed in the literature on institutional design in generaland on electoral systems in particular translates into legislative majority. While anabsolute majority in parliament for the governing party is then generating a securegoverning capacity, a qualified majority (required for making constitutional changesunilaterally, typically a two-thirds majority) is often conceived to err on too muchgoverning capacity. The dominant one-party system is not necessarily producing badgovernments but arguably increases the risk of negligence of minority rights, lack oftransparency and accountability, nepotism and other forms of abuse of power.

With regard to the empirical realities in Africa, the results in Table 4 once moredisplay a substantial difference between unfair and free elections. The relative shares


of legislative seats won by the largest party are substantially lower in free than inunfair elections. Electoral manipulation has many faces (Schedler, 2002) withobvious implications for the distribution of legislative power. As for the con-sequences of electoral systems in free and fair elections, it seems to be that these aremuch the same in Africa as elsewhere in the world. The more majoritarian theelectoral system, the more substantial legislative majorities. The exception again isthe relatively pure PR, which for the reasons mentioned above have biparty systemsand hence are associated with high governing capacity. However, it seems that theplurality system tends to lead to a bit of an ‘overkill’ capacity, which threatens thecompetition often necessary for the democratic values of transparency, ex postuncertainty, accountability and power of the people to decide who will representthem. We find examples of one-party dominance in several of the categories but it isamong countries with a plurality system that this seems to be a norm induced by theelectoral system. In line with the expectations based on studies of the establisheddemocracies the other category of PR system, despite the majoritarian effect of smallconstituencies, typically produce coalition or minority support for the government.This is not necessarily a problem, however. It can be argued on basis of empiricaldata that such governments perform as well, or even better, in terms of stability andpolicy outcome as do single-party majority governments. At least as long as coa-litions include no more than three to five parties (Sartori, 1997, p. 60).

3.5. Fairness of elections

A discussion specific for new democracies may offer some insight into issues ofparticipation and accountability given the frequent incidence of ‘unfree’ elections inAfrica. ‘Unfairness’ of elections, contaminates data on popular participation andaccountability to the extent of rendering analysis meaningless, that is unless we controlfor it in the analysis. That is themethodological point. Theoretically, the proponents ofthe proportional vision posit that these systems are better serving the democraticquality in plural societies, as it were, thanmajoritarian systems. Hence, the democraticquality of elections will be evaluated, using here two essential indicators: the free andfairness of elections and whether the electoral process was peaceful or not. Theidentification of the former builds primarily on international and domestic electionmonitoring assessments. The application of peacefulness used here is very strict. Onlyelections where there have been no reports of elections-related violence during thecampaign and election day are designated as peaceful.

The consequences of electoral systems on participation, accountability anddemocratic quality of elections in Africa are reported in Table 5. Generally, PRsystems have produced elections of significantly higher levels of fairness than havemajoritarian systems, with mixed variants somewhere in between. That is the generalpicture. PR systems are associated with higher shares of free and fair elections withpure (sic) PR of outstanding performance. Fairness is lowest by all means in absolutemajority systems where only 30% of all elections were deemed free and fair even bythe relatively low international standards, and of those free and fair elections only


17% were completely peaceful. Higher shares of completely peaceful processes areassociated primarily with mixed systems, and secondly with plurality systems.

Overall, the number of free and fair elections in Africa has been too low to meetthe expectations raised in the early 1990s of a new democratic era for Africanpolitics. The two most successful and oldest democracies in the region, Botswanaand Mauritius, both have plurality systems and are parliamentary democracies.There are more recent, relatively successful cases. Ghana operates a plurality system,while Mali and Gabon use the absolute majority system (Gabon changed from PR toabsolute majority system only in 1996). Benin, Cape Verde and Senegal use PR insmall MMC’s while South Africa, Namibia and Mozambique all have relatively purePR systems. Among all these recent cases all but South Africa are presidential ratherthan parliamentary regimes.

3.6. Participation in elections

The more majoritarian the system the lower the level of participation, followingcurrent wisdom. A plurality system with its exclusionary logic towards peripheralvoting subjects and typical competition for the median voter is assumed to lead tolower voter turnout. We may also assume that in new democracies parties that standlittle chance of winning (in plurality systems – small parties with medium or lowgeographical concentration of voters) may just exit. Hence we can assume that themore multipolar the competition and lower effective thresholds typical forproportional systems the higher participation both by voters and marginal parties.

First, Table 5 reveals another interesting finding. What President Mobutu andothers once did in terms of real or fake high turnouts to acclaim legitimacy is not socommon anymore. One suspects more assertive opposition forces and greaterinternational involvement in electoral administration have made it harder forincumbents to manipulate turnout figures than before. In any case, flawed electionsare no longer associated with inflated voter turnout figures in Africa. To thecontrary, unfair elections draw fewer voters than democratic ones. The peoples ofAfrica knows a fraudulent elections when they see one coming and chose abstain toa greater degree when their vote is not likely to make a difference.

In Table 5, the absolute majority system is a rather startling case of popularparticipation. A close look at the democratic quality of the elections in questionmight shed some light on this paradox. Over two-thirds of all elections in thesecountries have been non-democratic, ‘plebiscitory’ events in which the elite repro-duced itself. Consequently, few opposition parties wish to participate (in unfreeelections) and voters do not bother to go to the polling booth. When this is the ruleand the exception of a more fair election actually comes around, people are naturallystill skeptical. That would explain why figures of participation are not higher in thiscategory even for free and fair elections. Countries in this category include the well-known disgraceful cases of efforts to democratize like Cameroon, Republic ofCongo, Comoros, Mauritania, and Togo.


Table 5

Electoral system, fairness, participation and accountability

Electoral system Fairness Participation Account-ability

N (116) Peacefula Turnout Opposition

participatebLosers

acceptcTurnover

of powerd

Plurality Unfree 42% 12% 57% 46% 0% 4%

(0.150)

24 3 24 11 0 1

Free & fair 58% 39% 67% 91% 73% 18%

(0.170)

33 13 33 30 24 6

Absolute

majority

Unfree 70% 0% 41% 29% 0% 0%

(0.213)

14 0 14 4 0 0

Free & fair 30% 17% 48% 100% 67% 33%

(0.158)

6 1 6 6 4 2

Mixed Unfree 33% 0% 39% 100% 0% 0%

(0.007)

2 0 2 2 0 0

Free & fair 67% 50% 68% 75% 50% 0%

(0.200)

4 2 4 3 2 01

PR in

small MMC

Unfree 23% 0% 61% 60% 0% 0%

(0.206)

5 0 5 3 0 0

Free & fair 77% 35% 56% 100% 71% 53%

(0.177)

17 9 17 17 12 9

PR in large

MMC/pure

Free & fair 100% 27% 78% 100% 46% 36%

(0.149)

11 3 11 11 5 4

Total Unfree 40% 7% 51% 43% 0% 2%

(0.184)

46 3 46 20 0 1

Free & fair 60% 36% 64% 96% 67% 30%

(0.182)

70 25 71 67 47 21

Note: Percentages refers to within-subgroup N, except first column (‘N’) where it represents share of total

cases for each category of electoral system.


3.7. Accountability

Electoral systems are also expected to perform differently with regard toaccountability. The two main versions of the majoritarian vision supposedly presentvoters with a clear picture of responsibility, making policy accountability obvious.Hence, the expectation for a higher degree of electoral turnovers (shifting majorities)in the legislature. Mixed systems again are somewhere in between whereas in PRwith small- to medium district magnitude, and to a greater extent, in PR with largedistrict magnitude or pure PR, holding individual parties accountable for policydecisions is even harder. Shifting majorities are supposed to be less frequent anda collaborative pooling between the new coalitions from the old ruling coalition andformer opposition. To gauge these we would need data on government performanceand voters’ perceptions of that performance to use as controls. These data are notavailable. Yet, it is an improvement to our knowledge to do with what we have,rather than abstain from any inquiry at all even if it means that our results must bepreliminary. The indicator we use is legislative turnovers, or shifting majorities.Hence, the interpretation should be cautioned. The difference between PR and mixedsystems on the one hand, and majoritarian systems on the other, is so marked that itcannot be neglected because it runs contrary to expectations. Accountability issupposed to be one of the main strengths of majoritarian systems. With the measureused here it seems that plurality systems in particular perform poorly indeed. Theincidence of legislative turnover is about three times more common in PR and mixedsystems than in majoritarian systems. At least two distinct inferences are possiblefrom this. Either turnover is not at all connected to accountability, hence, we still donot know the empirical realities in Africa, or, majoritarian systems do not come out

Table 5 (continued)

Significance test

Opposition

Participate

Turnover

of Power

Free and

Fair

Loosers

Accept

Peaceful

Electoral

System

Pearson’s R 0.230 (0.001) 0.175 (0.012) 0.165 (0.018) 0.080 (0.257) 0.203 (0.004)

Chi2 value 26.195 (0.001) 22.850 (0.004) 43.736 (0.000) 19.709 (0.011) 23.583 (0.003)

df 8 8 12 8 8

a The criterion for peaceful employed here is stringent. Figures represent cases where there have been no

reports of any election-related violence during the campaign or polls. Hence, even elections with only

sporadic, minor disturbances are not included.b The criterion for participation here is stringent. The figures represent only those cases when all major

opposition groups participate. Hence, partial boycotts are treated as boycotts.c Looser acceptance is established when none of the major players challenges the results in court or

mounts other serious protests following the polls. This is a stricter application than Bratton (1998) used

since even cases where the opposition protested at first but then later decided to accept parliamentary seats

are coded as ‘not accepted’.d Turnover of power refers to electoral turnover of chief political executive in presidential elections and

changing majority in parliamentary elections. In systems with proportional representation and shifting

coalition partners, a turnover is sometimes ‘half’ in the sense that one or more parties leave the majority

and new parties enter. Such cases are not included in these figures, however.


strong on accountability in the context of socio-political realities in Africa. The pre-sent study cannot give an answer but recommends that further research is done withregard to the performance of different electoral systems on accountability in Africa.

4. Discussion

These interesting similarities and differences across countries with differenthistorical backgrounds add some new considerations. Africa provides a large numberof new cases and political outcomes in a field of study long dominated if not defined, byEuropean experiences. Changes in electoral systems alone are not likely to have exactlythe same significant effects on political behavior in Africa in established democracies.There can be no denying that a distinct colonial experience combined with extremelylow per capita income, and an agrarian existence differentiates Africa from the rest ofthe world. In addition, an eclectic ethnic population, a particular fusion of politicaland economic power, high levels of external debt and dependency, and the prevalenceof neopatrimonial politics, are some of the other characteristics that furtherdistinguishes Africa (Barkan, 2000, pp. 228–230). Yet, the introductory pages ofa recent analysis of institutional design and democratization seem to suggest thatAfrica is not altogether different. Some common denominators for the countries inLatin America and Eastern Europe are said to be:

‘‘[.] the transition is hampered in most of these countries by culturalfactors and in all of them by economic ones. Many of the democratizingsocieties have weak democratic traditions [.] and thus the legitimacy ofthe new political institutions is fragile [.] and much of society still hasa communitarian, anticapitalist mentality. Above all, economic con-ditions are not very favorable [.] most of these countries also bear theburden of large foreign debts.’’ (Lijphart and Waisman, 1996, pp. 2–3)

Some Africanists argue that majoritarian systems feed the ‘big man’ syndromeand reduce the likelihood of democracy by encouraging a zero-sum approach topolitical action. People continue to vote based on their residence rather than policypreferences. Barkan (1995) argues that the more agrarian the society, the higher thegeographic concentration of the vote. This finding led Barkan to advocate thata proportional system of representation does not really make much difference inagrarian societies and that a single-member district plurality system is equally goodin ensuring a distribution of seats in parliament that reflects the total vote.Furthermore, voters choose representatives based on how good they are as patronsof their respective community or constituency (Chabal, 1986). Based on a recent fieldexperiment, Wantchekron (2003) has been able to estimate the real effect ofpatronage platforms for voting behavior. Yet, it is not clear how much thispreference for patron over policy at the constituency level is part of a parochialpolitical culture in which dependency on some one with the right politicalconnections is prevalent (Almond and Verba, 1963). Regardless of cause, however,it provides an electoral logic that differs from other places where elections areprimarily fought over choices of policy. A recent study of parliamentary election


campaigns by Lindberg (2003) has shown that this pattern is reinforced by successivemultiparty elections, which poses several challenges to the functioning of polyarchy.The most notable effect is the proliferation of political parties and the problem ofuniting the opposition in fighting the government party in elections. Eachgeographical community is treated as an independent base, where politicianscompete for a single seat with little or no regard for what is happening in otherconstituencies or at the national level.

Instead of providing increased political stability and less conflict by generatingfewer political parties, majoritarian systems give rise to a polarized situation. Asingle party typically obtains a majority exceeding two thirds of all seats. Theopposition is often split on a large number of small parties that compete for votesand make political compromise and accommodation in the policy process difficult.The PR-systems, on the other hand, produce fewer parties while at the same timegiving the opposition a more substantial voice in the House, thus encouraging co-operation and coalition building. This finding is not surprising for those who havestudied African politics. The anomaly can be explained by the continued presence of‘big man’ politics, in which control of resources for patronage is imperative. Clientswill choose representatives that are most likely to have such control hence,a tendency to prefer an incumbent that has proved his ability to ‘deliver’. Preferencefor an alternative will be greatest in those constituencies where elected representa-tives have not lived up to client expectations. They would go for a new representativeand this person would be chosen based on his personal qualities rather than whichparty he belongs to. What effects this kind of politics will have and whether thecontinued use of a particular electoral system will, as suggested by studies from otherregions of the world, have the effect of modifying behavior and choice in predictableways, are issues that deserve more attention. This study can only provide what arebest conceived of as preliminary findings.

To the extent that the electoral system is an important variable with particulareffects in Africa, it is probably in relation to reining the detrimental effects ofneopatrimonialism. PR types of electoral system are likely to be more effective in thisregard than their majoritarian counterparts. Similarly, how boundaries of electoraldistricts are being drawn may also be significant in terms of influencing politicalbehavior among those seeking office. Given the inconclusive evidence regarding theeffects of particular types of electoral system on the number of political parties, I aminclined to recommend further studies to determine how and to what extent electoralsystem affects political outcomes in Africa.

5. Preliminary conclusions

In many respects, electoral systems have similar effects to the ones reported forthe established democracies and countries in Latin America. Overall, PR systemsare doing a better job than majoritarian systems not only in representativity,accountability and ‘democraticness’ of elections but also on governing capacity. It iswith regard to the latter that Africa distinguishes from established democracies.


Moreover, the socio-political background is different in Africa and neopatrimoni-alism remains an issue of singular importance to political development. In thisregard, majoritarian systems seem less conducive for stability and peaceful co-existence in the longer run. There are preliminary findings, however. First,multiparty elections in African countries, save a few, are recent phenomena andthe N is comparatively low to begin with. The consequences of electoral systems asdevices of constitutional engineering may not have shown yet, at least to their fulleffect. Moreover, the party systems are probably not settled yet, in the sense thatSartori (2001) frame as ‘structured’. The prevailing parochial political cultureinduces emphasis on person rather than party – hence the prevalent tendency inAfrican countries for political parties to be one-person operations. These are not‘effective’ parties as they are regarded in institutionalized democracies (cf. Laaksoand Taagepera, 1979), but they are ‘real’ parties in the sense that they are the productof rational choices by political actors. Third, the analysis presented here includes nocontrols apart from the denomination of elections as either free and fair or not.Ultimately, controls should be introduced at least for socio-economic development,types of state and constitutional regime, corruption and the extension of mass media.

It is therefore worth noting, that many of the general consequences found in moredeveloped democracies where the electoral system, as Sartori (1968, p. 273) hasargued is ‘the most manipulative instrument of politics’, are replicated also in Africaat this early stage of democratization. In particular, the main finding is that theadvocates of proportional systems for divided societies such as Lijphart andReynolds, gain support from the factual evidence reported here. The more specificfinding that systems of proportional representation are more likely than majority orplurality systems to encourage a greater number of opposition parties and rewardthem with a larger share of all seats also holds in this study of African elections. Thisfinding notwithstanding, Africa in some other respects present a challenge to theconventional wisdom about the effects of particular types of electoral systems.

Acknowledgements

The author wishes to thank Bo Bjurulf, Michael Chege, Goran Hyden, andTimothy Sisk for instructive comments on an earlier version of this article. I am alsoindebted to Winifred Pankani for excellent research assistance. This research hasbeen made possible by Sida Grant No. SWE-1999-231.

References

Almond, G.A., Verba, S., 1963. The Civic Culture. Princeton University Press, Princeton.

Barkan, J.D., 1995. Elections in agrarian societies. Journal of Democracy 6, 106–116.

Barkan, J.D., 1998. Rethinking the applicability of proportional representation for Africa. In: Sisk, T.D.,

Reynolds, A. (Eds.), Elections and Conflict Management in Africa. United States Institute of Peace

Press, Washington, DC.

Barkan, J.D., 2000. Protracted transitions among Africa’s new democracies. Democratization 7(3),

227–243.


Blume, M.E., 1974. Unbiased estimators of long run expected rates of return. Journal of the American

Statistical Association 69(347), 634–638.

Bogaards, M., 2000. Crafting competitive party systems: electoral laws and the opposition in Africa.

Democratization 7(4), 163–190.

Bogdanor, V., Butler, D. (Eds.), 1983. Democracy and Elections. Cambridge University Press, Cambridge.

Bollen, K., 1990. Political democracy: conceptual and measurement traps. Studies in Comparative

International Development 25(1), 7–25.

Bratton, M., van de Walle, N., 1997. Democratic Experiments in Africa: Regime Transitions in

a Comparative Perspective. Cambridge University Press, Cambridge.

Bratton, M., 1998. Second elections in Africa. Journal of Democracy 9(3), 51–66.

Chabal, P., 1986. Power in Africa. St. Martin’s Press, London.

Coppedge, M., Reinicke, W.H., 1990. Measuring polyarchy. Studies in Comparative International

Development 25(1), 51–73.

Datton, K., Greenies, J.S., Stewart, K.J., 1998. Incorporating a geometric mean formula into the CPI.

Monthly Labor Review 121(10), 3–7.

Di Palma, G., 1990. To Craft Democracies: An Essay in Democratic Transition. University of California

Press, Berkeley.

Diamond, L., 1999. Developing Democracy: Toward Consolidation. Johns Hopkins University Press,

Baltimore.

Downs, A., 1957. An Economic Theory of Democracy. Harper Collins, New York.

Dunleavy, P., Margetts, H., 1995. Understanding the dynamics of electoral reform. International Political

Science Review. 16(1), 9–30.

Duverger, M., 1954. Les Partis Politiques. Colin, Paris.

Foweraker, J., Landmann, T., 2002. Constitutional design and democratic performance. Democratization

9(2), 43–66.

Gasiororwski, M.J., 1996. An overview of the political regime change dataset. Comparative Political

Studies 29(August), 469–484.

Gebethner, S., 1996. Proportional representation versus majoritarian systems: free elections and political

parties in Poland, 1989–1991. In: Lijphart, A., Waisman, C.H. (Eds.), Institutional Design in New

Democracies: Eastern Europe and Latin America. Westview Press, Boulder, CO.

Grofman, B., Lee, S.C., Winckler, E.A., Woodall, B. (Eds.), 1999. Elections in Japan, Korea and Taiwan

under the Single Non-Transferable Vote: The Comparative Study of an Embedded Institution.

University of Michigan Press, Ann Arbor.

Gunther, R., Diamandouros, N.P., Puhle, H.-J., 1995. The Politics of Democratic Consolidation:

Southern Europe in a Comparative Perspective. The Johns Hopkins University Press, Baltimore and

London.

Horowitz, D.L., 1991. South Africa: Constitutional Engineering in a Divided Society. University of

California Press, Berkeley.

King, G., Keohane, R.O., Verba, S., 1994. Designing Social Inquiry: Scientific Inference in Qualitative

Research. Princeton University Press, Princeton.

Laakso, M., Taagepera, R., 1979. Effective number of parties: a measure with application to Western

Europe. Comparative Political Studies 12, 3–27.

Lijphart, A., 1984. Democracies: Patterns of Majoritarian and Consensus Government in Twenty-One

Countries. Yale University Press, New Haven.

Lijphart, A., 1994. Democracies: forms, performance and constitutional engineering. European Journal of

Political Research 25(1), 1–17.

Lijphart, A., 1999. Patterns of Democracy: Government Forms and Performance in Thirty-Six Countries.

Yale University Press, New Haven.

Lijphart, A., Waisman, C.H., 1996. Institutional design and democratization. In: Lijphart, A., Waisman,

C.H. (Eds.), Institutional Design in New Democracies: Eastern Europe and Latin America. Westview

Press, Boulder, CO.

Lindberg, S.I., 2003. It’s our time to ‘chop’: does elections in Africa feed neopatrimonialism rather than

counter-act it? Democratization 10(2), 121–140.


Linz, J., Stepan, A., 1996. The Problems of Democratic Transition and Consolidation. Southern Europe,

South America and Post-Communist Europe. The Johns Hopkins University Press, Baltimore and

London.

Mair, P. (Ed.), 1990. The West European Party System. Oxford University Press, Oxford.

March, J.G., Olsen, J.P., 1989. Rediscovering Institutions. Free Press, New York.

Mozaffar, S., Schedler, A., 2002. The comparative study of electoral governance. International Political

Science Review 23(1), 5–24.

Nohlen, D., 1996. Electoral systems and electoral reform in Latin America. In: Lijphart, A., Waisman,

C.H. (Eds.), Institutional Design in New Democracies: Eastern Europe and Latin America. Westview

Press, Boulder, CO.

Nohlen, D., Krennerich, M., Thibaut, B. (Eds.), Elections in Africa: A Data Handbook. Oxford

University Press, Oxford.

O’Donnell, G., 1996. Illusions about consolidation. Journal of Democracy 7(2), 34–51.

Powell, B.G., 1982. Contemporary Democracies: Participation, Stability and Violence. Harvard

University Press, Cambridge, MA.

Powell, B.G., 2000. Elections as Instruments of Democracy. Yale University Press, London.

Przeworski, A., Teune, H., 1970. The Logic of Comparative Social Inquiry. Wiley Interscience, New York.

Rae, D., 1971. The Political Consequences of Electoral Laws. Yale University Press, New Haven.

Reynolds, A., Sisk, T.D., 1998. Elections and electoral systems: implications for conflict management. In:

Sisk, T.D., Reynolds, A. (Eds.), Elections and Conflict Management in Africa. United States Institute

of Peace Press, Washington, DC.

Sartori, G., 1968. Political development and political engineering. In: Montgomery, J.D., Hirschman,

A.O. (Eds.), Public Policy. Cambridge University Press, Cambridge.

Sartori, G., 1986. The influence of electoral systems: faulty laws or faulty method? In: Grofman, B.,

Lijphart, A. (Eds.), Electoral Laws and Their Political Consequences. Agathon Press, New York.

Sartori, G., 1997. Comparative Constitutional Engineering. second ed. New York University Press,

New York.

Sartori, G., 2001. The party effects of electoral systems. In: Diamond, L., Gunther, R. (Eds.), Political

Parties and Democracy. Johns Hopkins University Press, Baltimore and London.

Schedler, A., 1998. What is democratic consolidation? Journal of Democracy 9(2), 91–107.

Schedler, A., 2001. Measuring democratic consolidation. Studies in Comparative International

Development 36(1), 66–92.

Schedler, A., 2002. Elections without democracy: the menu of manipulation. Journal of Democracy 13(2),

36–50.

Valenzuela, S.J., 1992. Democratic consolidation in post-transitional settings: notion, process, and

facilitating conditions. In: Mainwaring, S., O’Donnell, G., Valenzuela, S.J. (Eds.), Issues in

Democratic Consolidation: The New South American Democracies in Comparative Perspective.

University of Notre Dame Press, Notre Dame.

Wantchekron, L., 2003. Clientism and voting behaviour: a field experiment in Benin. World Politics 54,

399–422.

Weaver, K.R., 2002. Electoral rules and governability. Journal of Democracy 13(2), 111–126.

consequences of electoral systems in africa: a preliminary...

Documents